m

■/t^-
py-1 >X ^
-n-v

■"‘Ilf
■M 4a*-'.!- ;-
■;kv^
‘t'Tf

a-tfc-^

:r.-.
IS/'

M
Y

/■' -,, ;►(«<«%.,r>'-,.,'

>r-

RICHARD

infill
 *.'
BOSTON PUBLIC LIBRARY
’

Copley Square
On

.•'V

-»fv

i
»
AGING
and

OLD AGG
 Richard A. Posner

AGING
and

OLD AGG

THE UNIVERSITY OF CHICAGO PRESS

CHICAGO AND LONDON

Richard A. Posner is Chief Judge of the U.S. Court of
Appeals for the Seventh Circuit and senior lecturer at the
University of Chicago Law School. His many publica¬
tions include Cardozo: A Study in Reputation, from the
University of Chicago Press; Sex and Reason; and Over¬
coming Law.

The University of Chicago Press, Chicago 60637

The University of Chicago Press, Ltd., London
© 1995 by The University of Chicago
All rights reserved. Published 1995
Printed in the United States of America
04 03 02 01 00 99 98 97 96 2345
ISBN: 0-226-67566-1 (cloth)

Library of Congress Cataloging-in-Publication Data

Posner, Richard A.
Aging and old age / Richard A. Posner,
p. cm.
Includes bibliographical references and index.
1. Aged—Social conditions. 2. Aged—Legal
status, laws, etc.
3. Aging—Economic aspects. 4. Old age—Eco¬
nomic aspects.
5. Gerontology. 1. Title.
HQ1061.P67 1995
305.26—dc20 95-16462
CIP

(§)The paper used in this publication meets the mini- ■ »1Q1061

mum requirements of the American National Standard I . F’6 7
for Information Sciences—Permanence of Paper for I ^ 995
Printed Library Materials, ANSI Z39.48-1984. I
 Contents

Introduction 1

PART
ONE Aging and Old Age as Social, Biological, and Economic
Phenomena

1 What Is Aging, and Why? 17

The process of aging. Aging and evolution.

2 Old Age Past, Present, and Future 31

The history of old age. Anxiety over gerontification.
Restoring perspective.

3 A Human-Capital Model of Aging 51

The simplest life-cycle model. Posthumous utility
and the last-period problem. Relational human
capital.

4 An Economic Model of Aging with Change 66

Assumed
The knowledge shift, discount rates, and the age-
decline curve. The retirement decision. One self or
multiple selves?
VI Contents

PART
TWO The Economic Theory Elaborated and Applied

5 The Economic Psychology of the Old

Aristotle on old age. The dread of death. Physical 99
and mental decline as factors in the psychology of
the elderly. Religion, voting preferences, and
speaking.

6 Behavioral Correlates of Age 122

Driving. The involvement of the elderly in crime as
victims and as offenders. Suicide. Penny-pinching.
Sex. Work and leisure. Residence. Voting and jury
service.

7 Age, Creativity, and Output 156

Productivity and age: creativity versus leadership.
Differences in peak age between and within fields.
Changes with age in the character of creative
work.

8 Adjudication and Old Age 180

How productive are older judges? Causes of the
productivity of older judges. The effect of increased
longevity.

9 The Status of the Old and the Aging of Institutions 202

Primitive and agrarian societies. Modernity and
symbolic status. The bearing of ideology. The
institutional life cycle.

PART
THREE Normative Issues

10 Euthanasia and Geronticide 235

Definitions. An economic analysis of physician-
assisted suicide in cases of physical incapacity.
Voluntary euthanasia with implementation
deferred.

11 Social Security and Health 262

The compulsory character of social security. The
allocation of medical resources to the elderly and
Contents Vll

between elderly men and women. Social security

and redistribution. Old age and democratic theory.

12 Legal Issues of Aging and Old Age: A Sampler 298

Pension law. Elderly tortfeasors and tort victims.
Elderly offenders and prisoners. Dementia and
capacity.

13 Age Discrimination by Employers and the Issue of 319

Mandatory Retirement
The nature and consequences of age discrimination
in employment. The effects of the Age Discrimination
in Employment Act. The economics of mandatory
retirement. Mandatory retirement of judges and
professors.

Index 365
' /in.’'’'"'s ■ m-l
'’ " » , ,^ ( V •.» ^
'^ r
'J;V" " -'iV

'.'M =‘^=^_A';i)ii^l t'. jyVsw-.: .. ■•‘A'^, /•■ ■-!, ^

A>.', .’ 't\ . ■ . . ..■». . ' »■■'.>''
;!H '’,': . ‘i> ,-y\ -'■
'f, '

■<*
A .. ,
, ■"’' ■■ . . ■‘.t v;, : '

U'A .,, , ..J Vst,, ■' . ^•■■•‘Ti •^.. .■

V
r
• ■" • * • ‘ • • S*
^
\ ' X. -A f"'
■ ''- V
' • . -1» - . • A -* *lf' •
-FS. -■''»■ t
4» • i* f
< '? -..'S '■

'^Xi

'r'^ '' ,-liW ■' ' ■, ■ ■ V,;..'}":'

“^1 •. 1 .* * '•

'..' ’ '’'’'''''•^5^ ' ;il;2 ■* •

' ■■-:*•.
iVl '* ' ■■' . .■■■■- • - ‘-to.. tvr,jA.> *-.^ • ->‘i'--

/■N: .: Vf
\

.-r -
'Ml, <

''4* •?: H. '.’»> Aifi^ I tiXfi

if' '‘.7 ■^'

.?■:. * ^ ^'T.tpsarywn' ■ •, j. -^ ■ ■ '>•*.
»'■,., ' -i'' 'i"' '^ ■ S' '
,:^ . , ‘^;^,.''V-v.,.'.^7:';,,.;

i*'■■■.■:,'rS^'’' vr.“'^,i’® ■■' ^'r'SjfrjiMlifefF'lli

 Introduction

Old age, and aging more generally, raise a host of fascinating issues made
extraordinarily timely by the rapidly growing number and percentage of
elderly people in our society. This book is written in the conviction that the
economics of nonmarket behavior can enable fresh insights into a number
of issues related to aging to which economists have given little or no atten¬
tion. Examples include geronticide (euthanasia of old people), age dis¬
crimination in employment, judicial retirement, creativity and leadership
as functions of age, voter turnout and jury participation by elderly people,
negligent and criminal behavior by them, the allocation of resources to the
prevention and treatment of geriatric illness, the characteristic attitudes and
habits of the elderly, their changing social status, and the political economy
of public policies relating to the elderly. I hope to show, in fact, that eco¬
nomics can provide a unifying perspective in which to view the whole
range of social problems concerning the elderly.
My analysis is both positive and normative. That is, it is concerned
both with advancing the understanding of aging and old age and with
evaluating, or, more modestly, creating a framework for evaluating, public
policies related to age. And it is interdisciplinary. I mine disciplines ranging
from evolutionary biology and cognitive psychology to philosophy and lit¬
erature for insights into aging and old age. The idea that our young and
old selves are different persons—an idea philosophical rather than eco¬
nomic in character—and selection bias, a statistical concept, play espe¬
cially prominent roles in my analysis, as does Aristotelian psychology. But
economics wields the baton of my multidisciplinary orchestra. The basic

1
2 Introduction

hypothesis of the book is that economics can do a better job of explaining

the behavior and attitudes associated with aging, and of solving the policy
problems that aging presents, than biology, psychology, sociology, philoso¬
phy, or any other single field of natural or social science.
I have written with several audiences in mind. One is the multidisci¬
plinary gerontological community; for it, the main surprise will be the per¬
vasive relevance of economics to gerontology. A second audience consists
of economists interested in nonmarket behavior, or, more broadly, in the
application of economic theory to subjects lying outside the traditional
boundaries of economics. To this group, the surprise will be the amount of
work bearing directly or indirectly on subjects that are or ought to be
of interest to economists that has been done by the mostly nonecono¬
mist scholars of the gerontological community. A third audience consists
of those lawyers, judges, philosophers, bioethicists, social scientists, and
other professionals and scholars who are interested in specific topics in or
overlapping the study of aging and old age, such as pension regulation,
euthanasia, subsidies for health care, compulsory retirement, mental com¬
petence, favoritism and discrimination on grounds of age, and intergenera-
tional equity. All of these matters arise in legal cases. Lawyers and judges
are often surprised to discover that matters which swim into their view in
cases have been studied extensively by nonlegal disciplines. There may
even be a few people interested simply in their own aging who will find
things of value in the book.
I have tried to make the book intelligible to these different audiences.
I use few formal models, and they are very simple and fully explained and,
even so, can be skipped by the mathless reader without serious loss of un¬
derstanding. No previous acquaintance with economics is assumed, or is
necessary to follow any of the analysis.
Despite my professional affiliation, this is not primarily a book about
the laws related to the aged. But I do discuss a number of legal applications,
ranging from age discrimination and pension laws through criminal sen¬
tencing and tort liability to judicial-retirement policies. And my choice of
subject was decisively influenced by my experience as a federal judge. The
Age Discrimination in Employment Act (ADEA) and the Employee Retire¬
ment Income Security Act (ERISA) have generated many cases in my
court. And the occupation of being a federal judge, an occupation whose
incumbents have never been subject to—whom Article III of the Consti¬
tution appears to forbid subjecting to—a mandatory retirement age, is the
preeminent Nestorian profession. Many federal judges work full time, or
nearly so, well into their eighties (Justice Holmes did not retire until he was
Introduction 3

90, and there have been other nonagenerian judges as well), and octogen¬
arians have written some of the most important judicial opinions in our
history. Concerns that judges are too old have been voiced from time to
time, most famously in defense of Franklin Roosevelt’s “Court-packing
plan” (1937), which would have authorized the appointment of an addi¬
tional Supreme Court Justice for every Justice who refused to retire or re¬
sign upon reaching the age of 70, until the Court had fifteen members. The
threat to judicial independence posed by this proposal was so manifest that,
ever since, there has been little discussion of the problem of elderly judges,
even though it is well known within professional circles that some federal
judges, including Supreme Court Justices, have continued to sit long after
their judicial performance became severely compromised by age-related
disabilities.
Federal judges take for granted the existence of the federal laws regu¬
lating pensions and retirement and forbidding as “discrimination” certain
forms of differential treatment of the aged, and they also take for granted
their own peculiar tenure provisions. Judges are very good at taking things
for granted. The structure of adjudication fosters a reactive, even a passive,
attitude toward both the doctrinal content and the institutional form of the
judge’s work. The umpireal role is inherently rather passive. Someone else
stages the contest, and contest succeeds contest at a dizzying rate (espe¬
cially for an appellate judge, who may hear two or three hundred different
cases every year) and in random sequence. The judge, for whom judging is
in any event likely to be, in this country as in England, a second career,
comes in late to the play and leaves early, without ever being invited behind
the scenes. I have no quarrel with the basic structure of appellate adjudi¬
cation. But one of its consequences is that judges tend not to be well in¬
formed about the background of the cases they adjudicate or the institu¬
tional structure they inhabit. This is especially true with regard to a feature
like old age that carries a heavy emotional charge yet is not a familiar and
well-digested field of study the major findings of which are readily acces¬
sible to persons who are not experts in it. As in my book Sex and Reason
(1992), so here, one of my goals in tackling the subject is simply to inform
myself and my fellow judges about a pervasive, emotional, but little under¬
stood phenomenon (or cluster of phenomena) that provides a backdrop to
and the subject matter of several important areas of modern law, and that
in the case of old age is also a salient characteristic of an influential group
of law’s makers and appliers. It is not so taboo a subject as sex, but consid¬
erable reticence, embarrassment, and denial surrounds the public discus¬
sion of many aspects of it, notably the aspects emphasized in this book. I
4 Introduction

have tried to draw back the veil. To this end, although most of my analysis
is of a theoretical or, less grandly, a speculative cast, I have tried to back it
up with data wherever possible. There is no shortage of data on old age.
In writing about aging from an economic standpoint, I am not writing
on a clean theoretical slate, as was virtually the case when I took an eco¬
nomic approach to sex in Sex and Reason. The economic literature on old
age is vast and technically sophisticated.’ But it is focused on retirement
issues, such as the determinants of the age of retirement and the financing
(whether through private or public pensions) of consumption during retire¬
ment. It largely ignores the psychological and even the physical aspects of
aging, although, as we shall see, both aspects can be given an economic
interpretation.
Not all economists who write about aging are preoccupied with retire¬
ment. Arthur Diamond has written on scientific creativity, and on scien¬
tists’ receptiveness to new theories, as functions of age; I discuss his work
in chapter 7. And Isaac Ehrlich and Hiroyuki Chuma have written about
the efforts of elderly people to extend their life (see chapter 5). There are
other exceptions.^ But they are few. As a result, while the economic litera¬
ture is illuminating on such questions as whether the savings and consump¬
tion decisions of the old are influenced by a bequest motive and what effect
social security has on the participation of middle-aged and elderly persons
in the labor force and on the savings rate of the society as a whole, a variety
of other questions have largely been ignored. Among the ones I try to an¬
swer are. Why are old people less willing to take financial and other risks
than young people? Do organizations age in ways parallel to how persons
age? Does “blind” refereeing of articles submitted to scholarly journals

1. That literature is well reviewed in Michael D. Hurd, “Research on the Elderly: Economic
Status, Retirement, and Consumption and Saving,” 28 Journal of Economic Literature 565 (1990).
For an earlier literature review, see Robert Clark, Juanita Kreps, and Joseph Spengler, “Economics
of Aging: A Survey,” 16 Journal of Economic Literature 919 (1978). For an excellent nontechni¬
cal overview, see Victor R. Fuchs, How We Live, ch. 7 (1983).
2. Of which the most curious may be Ray C. Fair, “How Fast Do Old Men Slow Down?”
76 Review of Economics and Statistics 103 (1994), a statistical study of men’s track-and-field and
road-racing events. There is no economic model. The distinguished author believes that his finding
that in a number of racing events a man of 85 is only 49 percent slower than he was at 55 may
have policy relevance; specifically, may indicate “that societies have been too pessimistic about
losses from aging for individuals who stay healthy and fit.” Id. at 117. I do not understand the
significance of his finding for policy. But I am sure that Professor Fair will be interested to learn,
if he does not know already, that the U.S. Tennis Association has a tournament for players 85 years
and older. Dan Shaughnessy, “Seniors Serve as Inspiration,” Boston Globe, Sept. 7, 1994, p. 57.
Asked to identify the “single common denominator” that explains what “allows these gentlemen
to compete at this game after all these decades,” an 87-year-old player replied sensibly, “The
common denominator is that we’re still alive.” Id. at 59.
Introduction 5

discriminate unwarrantedly against older scholars? Why don’t old people

have high crime rates, sinee the expected cost of imprisonment is truncated
by their diminished life expectancy? (Stated differently, what is the signifi¬
cance of “last-period” analysis for the behavior of the old?) Are old people
victims of “ageism” ? Do they, on the contrary, have too much power? Why
do old people vote more than young people, and why is the gap greater in
congressional than in presidential elections? Since children cannot vote,
should elderly people be deprived of the vote in order to maintain a proper
balance between the interests of young and old people? Why do people so
often postpone the making of a will until death is imminent? Should there
be a mandatory retirement age for judges? Relatedly, how could one pos¬
sibly measure the effect of age on judicial productivity? Why are old
people admired in some societies and despised, even killed, in others? Why
do black people tend (as we shall see they do) to respect the old more than
white people do? Is the life of an older person less valuable in some mean¬
ingful sense of “value” than that of a younger one? Why does creativity
peak at different ages in different fields? Do old people really tend to be
more verbose than young ones and if so why? Why do so many sick old
people nevertheless cling to life? Should we make it easier for them to die?
Should publicly funded medical research seek to equalize the longevity of
American men and women? Should it concentrate less on diseases that kill
elderly people and more on conditions such as blindness and deafness that
reduce the quality of elderly lives? What is the effect of trying to stamp out
age discrimination on the distribution of lifetime wealth across persons
and should an old person even be considered the same person that he or she
was when young? Is saving for one’s old age morally different from saving
in order to leave bequests to one’s children or grandchildren? Many of these
may not sound like economic questions, but that is because most people
(including many economists) have too narrow a eonception of economics.
The economic concepts that I shall use to power and direct my analysis
are familiar. They are drawn from established fields of economics such as
the economics of information, the economics of health, the economics of
law, and the economics of human capital, although some I apply outside of
their usual domain (for example, by treating an individual’s young self and
his old self as distinct rational actors) or with a new twist. The book’s prin¬
cipal novelty as a contribution to economics is its attempt to shift the em¬
phasis in the economics of aging and old age from the financial, or market,
aspects of the subject to the nonfinancial, nonmarket aspects. But the shift
in emphasis can, I argue, furnish new insights even into well-studied issues
concerning retirement.
6 Introduction

Consistent with a commitment to genuine multidisciplinarity, I got the

idea of applying the economics of information to aging not from any eco¬
nomic literature but from a passage in Aristotle’s treatise on rhetoric. He is
discussing how to argue to young and old respectively and he says that this
depends on their outlooks on life, which are different. He lists many differ¬
ences but my eye was particularly caught by the following: The “lives [of
the young] are mainly spent not in memory but in expectation; for expec¬
tation refers to the future, memory to the past, and youth has a long future
before it and a short past behind it: on the first day of one’s life one has
nothing at all to remember, and can only look forward.” The old, in con¬
trast, “live by memory rather than by hope; for what is left to them of life
is but little as compared with the long past; and hope is of the future,
memory of the past.” ^ If we equate memory to knowledge and hope/expec¬
tation to imagination,"* and if we think of knowledge and imagination
(roughly what psychologists call “crystallized” and “fluid” intelligence)
as the two principal components of reason, the balance shifts with age.
Translated into economic terms, this shift will play a major role in my
analysis, whereas most of the previous economic work on aging has ab¬
stracted from the cognitive, emotional, ethical, and other differences be¬
tween young and old. Almost the only difference recognized by that work
is the difference in proximity to the end of life, because death truncates the
payback period for investments in human capital. We shall see that this
need not always be the case—that from an economic standpoint the period
over which an individual maximizes his utility need not end with his
death—and, more important, that it is not the only significant difference
between young and old. The cognitive change identified by Aristotle is also
important, as is the well-documented decline in mental and physical capa¬
bilities that is a correlate of aging and that, moreover, interacts with the
change noted by Aristotle.
It would be only a slight exaggeration to say that most economists who
write about the life cycle implicitly deny aging in the sense of a decline or
other change in relevant capacities. Their implicit conception of the func¬
tion that relates the capacity to produce or consume to age is a horizontal
line that drops abruptly to zero at death or retirement. Yet nothing in the
theory of human capital, a theory central to the economics of aging, re-

3. Aristotle, Rhetoric, bk. 2, ch. 12, in The Complete Works of Aristotle, vol. 2, pp. 2213,
2214 (Jonathan Barnes, ed., 1984) (W. Rhys Roberts, trans.). The page and column references to
the Greek text are 1389a and 1390a.
4. The word that Roberts translates first as expectation and then as hope is the same—
cAttict—and it has both meanings.
Introduction 7

quires that the physical and cognitive changes associated with aging be
ignored. Human capital (earning capacity, with “earnings” broadly defined
to include nonpecuniary as well as pecuniary returns) includes innate ca¬
pacities—such as mathematical or musical aptitude, quick reflexes, and
physical strength—as well as acquired ones, and even the acquired need
not always be the product of training. The innate capacities are subject to
biological decay, while the acquired ones include, as Aristotle recognized,
knowledge (which is also subject to decay) gained simply through living
and working, that is, through experience, which is correlated with age, as
well as knowledge acquired in schools and in training programs. One rea¬
son the innate capacities in particular have been slighted in the economic
research on aging is that economists have tended to cut off their study of
the life cycle at retirement, which normally precedes the steepest decline
in capacity. Yet the decline begins earlier, sometimes much earlier. A re¬
lated point is that retirement itself, when it is viewed functionally, may
occur long before the standard retirement age of 65 (standard, though of
course not uniform, since Bismarck’s social welfare legislation). Econo¬
mists have taken little interest in deviations from the standard life cycle.
As the point about functional versus nominal retirement suggests, al¬
though my focus is on old age this is a less confining category than one
might think. The federal age discrimination law defines the protected class
to include all employees 40 years of age or older. Many cases under the law
concern early retirement, that is, retirement before the employer’s regular
retirement age. Aging, moreover, is ordinarily a protracted and continuous,
rather than a discrete, process. In some fields, such as professional athletics,
you may be “old” at 30; and in some intellectual fields as well, such as
theoretical physics and computer software design, the career peak comes at
an early age. The age grading of occupations or activities (that is, the as¬
signing of roles on the basis of chronological age), an important feature of
primitive societies, is not limited to old people. Indeed, the study of old age
is inseparable from that of the entire life cycle. That is why the title of this
book includes “aging” as well as “old age,” although the latter is my pri¬
mary concern. Some topics of particular significance for old people, such
as the punishment of recidivists, euthanasia, the measure of damages for
loss of the nonpecuniary utility of life, and the optimal allocation of medi¬
cal resources, have application to young people as well. Similarly, the
analysis of judicial performance at advanced ages has implications for un¬
derstanding the judicial process generally, while the pattern of litigation
under the age discrimination law casts light on the determinants of plain¬
tiffs’ success in litigation generally. So the book in places spills over its
8 Introduction

banks, while in other places it does not reach them, so vast is the subject of
old age and the literature that deals with its various facets.
A thumbnail sketch of the book’s organization may help in orienting
the reader. The first four chapters set the stage. Chapter 1 presents the es¬
sential biological data concerning old age, which have tended to be over¬
looked or sugarcoated in the social scientific literature. The data show that
there really is such a thing as “normal aging”—people age; they do not
just become more prone to illness and accident—and I show that the denial
of this fact can actually create an exaggerated view of the disabilities of the
elderly. One of the questions examined in the chapter is whether there is a
genetic program for old age. The answer—“no”—plays, paradoxically, a
role in subsequent chapters in explaining in genetic terms certain features
of the experience of old age.
Chapter 2 is longer, but also sunnier. It presents additional background
data—demographic, historical, and economic, rather than biological—
concerning human aging and old age. It also offers a preliminary assess¬
ment of the widespread popular and scholarly concern that the nation is
becoming ominously gerontified. I argue that although the population is
indeed aging, this is not the disaster that alarmists descry and decry. They
have exaggerated the costs of an aging population and ignored the bene¬
fits—which is not to deny that there are costs, and that they are growing.
The next two chapters lay the principal theoretical foundations for my
analysis by developing a series of economic models of aging. Chapter 3
begins with the application of the conventional human-capital model of the
life cycle to old age. But soon I am expanding the model in several direc¬
tions, for example by considering the bearing of posthumous utility (not
limited to the utility derived from making bequests) and of what I call “re¬
lational” human capital, which includes friendship and other personal, in¬
cluding business, relationships. Chapter 4 rings further changes on the con¬
ventional model by assuming that, among other things, the passage from
youth to old age entails cognitive changes in the form of the shifting bal¬
ance between imagination and knowledge that I have already mentioned;
that the speed at which subjective time passes also changes (it speeds up);
and that physical and mental abilities decline with age. The decline is
gradual for most people until extreme old age, but can, I argue with the aid
of the concept of “excess capability,” be modeled as if there were an in¬
flection point at age 65 for many activities. I also emphasize boredom,
viewed as the obverse of habit, as an economically analyzable factor in the
decision to retire. And I introduce and defend the proposition, which plays
a big role in the book, that the difference between one’s young and one’s
Introduction 9

old self may be so profound that the two selves are more fruitfully viewed
as two persons rather than as one. Chapters 3 and 4 thus attempt to redeem
my promise to present an approach to the economics of aging that takes
aging seriously.
Chapters 5 through 9, constituting part 2 of the book, extend the ap¬
proach. In chapter 5, I use the approach to generate the “economic psy¬
chology” of old age. The term is meant to suggest the possibility of deriv¬
ing psychological traits from economic theory rather than, as is generally
done in economics, treating such traits as exogenous. I argue, for example,
that old people are (on average—always an important qualification in deal¬
ing with large and amorphous social aggregates such as “the elderly”)
worse listeners and less considerate speakers than young people. The old
invest less in the creation of human capital and therefore have less to gain
from receiving inputs of information from other people. And because they
transact less, they have less incentive to conceal egocentrism and to engage
in cooperative rather than self-aggrandizing conversation and also less to
gain from concealing traits that would reduce opportunities for advanta¬
geous transactions.^
Chapter 6 proposes solutions to a number of puzzles concerning the
behavior of the elderly with regard to driving and automobile accidents,
crime, suicide, sex, residential patterns, bequests, voting and jury service,
and (once again) retirement—such puzzles as the conjunction of very low
crime and criminal-victimization rates with high accident rates and very
high suicide rates and the conjunction of high voter turnout with low juror
turnout. The effort is to demonstrate to any doubters, first, that “old age”
really is a meaningful category for analysis, by identifying a distinctive set
of age-related behaviors, and second that economics can play its usual use¬
ful role of making sense out of seemingly arbitrary variations in social
behavior.
Chapter 7 explores the relation between age and creativity or achieve¬
ment. I emphasize the importance to the age profile of different activities
of distinguishing between lived experience and practical wisdom, on the
one hand, and book learning and abstract reasoning, on the other. I also try
to explain the difference between “creative” occupations and leadership,
as well as the cluster of age-related changes in the character as distinct from

5. Cf. M, F. K. Fisher, Sister Age 234 (1964): “I have formed a strong theory that there is
no such thing as ‘turning into’ a Nasty Old Man or Old Witch. I believe that such people, and of
course they are legion, were born nasty and witch-like, and that by the time they were about five
years old they had hidden their rotten bitchiness and lived fairly decent lives until they no longer
had to conform to rules of social behavior, and could revert to their original horrid natures.”
10 Introduction

quality of creative work that has led to the suggestion that there is an “old-
age style” or “later-life style” in the arts. Chapter 8 extends the analysis of
elderly achievement to judicial output, with special attention to appellate
judges. I use citation analysis to generate empirical evidence that judges do
in fact tend to retain their capabilities to advanced ages, though not without
measurable impairment. I explain the gradualness with which judicial pro¬
ductivity declines with age by reference to the importance of judicial ex¬
perience and of writing skills (which decline with age less than most other
cognitive skills do) to the successful performance of judicial, especially
appellate, tasks, but also by reference to the humble but recurrent factor of
selection bias. Between them, chapters 7 and 8 furnish a number of clues
to the wide variance in rates of aging across different activities.
Chapter 9 examines the even wider differences in the social status of
the aged in different societies, differences that run the gamut from compul¬
sory geronticide to ancestor worship. I lay particular stress on premodern
societies (where the variance is greatest)—trying to explain why the el¬
derly have a very high status in some of these societies and a very low one
in others—and on the transition to modernity. I also discuss the practice of
age grading and the question whether there might be fruitful analogies and
interactions (I think there are some of both) between the aging of individ¬
uals and the aging of firms, nations, and other institutions.
Parts 1 and 2 attempt to develop an economic theory of aging that will
have enough explanatory and predictive power to deserve being taken se¬
riously not only as a source of knowledge but also as a guide to reflection
on the many normative issues, both ethical and legal, that the phenomenon
of aging presents. Part 3 examines the normative issues directly. These can
be divided into macroeconomic and microeconomic issues. The former is
encapsulated by the “rip-off” question. Are the old ripping off the young?
If so, by how much, and how can it be controlled? The “rip-off” question
is analyzed in chapter 11, though the foundations for the analysis are laid,
and the answer prefigured, in chapter 2.
The second class of normative issues, the microeconomic, involves
particular markets or activities. Chapter 10 is illustrative. It is about eutha¬
nasia (“geronticide” when the persons “euthanized” are elderly). I argue
that legalizing physician-assisted suicide might lower rather than, as critics
and supporters alike believe, raise the suicide rate. I review a number of
other objections to the practice as well, such as that it will result in physi¬
cians’ rushing infirm patients to their deaths, and conclude (with an assist
from the political theory of John Stuart Mill) that physician-assisted suicide
Introduction 11

should be decriminalized in cases of terminal illness and profound physical

impairment. But I argue against allowing one’s younger self to kill one’s
older self by committing to die upon the occurrence in the future of a speci¬
fied condition, such as advanced senility.
Chapter 11 takes up a number of other ethically charged issues involv¬
ing pubUc policy toward the elderly. These are compulsory and subsidized
pensions, the subsidization of health care for the aged, the allocation of
medical research between diseases of old men and diseases of old women,
and the question whether democracy gives too much power to old people,
especially compared to children. I argue, again counterintuitively, that
women might be made better off by a reallocation of medical research from
the diseases of elderly women to the diseases of elderly men and that el¬
derly people may have the right amount of voting power after all, when
their role as “representatives” of the future elderly selves of the currently
young and middle-aged is recognized. In this chapter philosophical ques¬
tions are prominent, including the recurrent question whether and for what
purposes the young and old phases of an individual’s life should be consid¬
ered different stages of a single self, on the one hand, or different selves,
on the other. I point out that the tension between young and old selves
complicates the problem of financing the medical and home-care costs of
the elderly, costs that will grow as the number of elderly grows, and quite
possibly faster. But I conclude that there is no firm basis for believing that
elderly people either have obtained, or will obtain as their numbers in¬
crease, excessive transfer payments from the young and the middle-aged.
Chapters 12 and 13 apply the positive and normative analysis devel¬
oped in the earlier chapters to a variety of legal issues. These include issues
raised by the federal law regulating private pensions (ERISA), by tort,
criminal, and property cases involving the elderly, and by the continued
imprisonment of young criminals, under sentence of life in prison, when
they become old. My principal emphasis, however—it is the subject of
chapter 13, the longest chapter in the book—is upon age discrimination,
including the issue of mandatory retirement at fixed ages. I examine the
rationale for the Age Discrimination in Employment Act and the Act’s
probable efficacy, expense, and allocative and distributive effects; and this
leads me also to consider the pros and cons of mandatory retirement, which
the Act has now largely abolished. My conclusion is that the Act is largely
inelfectual, and to the extent effective is probably perverse in its effects on
the distribution of income and causes harm to elderly workers. It exempli¬
fies the fact, which is often overlooked, that “social as well as eco-
12 Introduction

nomic” regulation may disserve any plausible conception of the public in¬
terest once the actual presuppositions and consequences of the regulation
are understood—and may even disserve its ostensible beneficiaries.
The scope of the book is broad, both as to subject and as to method,
and thus bucks the trend (which I believe however to be on the whole a
healthy one) toward ever greater specialization in scholarly research. It pro¬
poses a new way of looking at an old (no pun intended) and variegated
subject, and the specific answers that it offers to specific questions should
be viewed as suggestive rather than definitive. Yet despite this disclaimer
of dogmatic certitude, the book will definitely not please everyone. It will
displease not only those who are offended by the application of rational-
choice theory to “noneconomic” phenomena, but also the Chicken Littles
of this world, who are clamorous about the fell consequences (as they see
them) of the rapid aging of the population, and the professional advocates
for the elderly, who exaggerate the plight of their constituency. Although I
do not join Cicero in considering old age the happiest part of a wise per¬
son’s life, and although I do not think that we have or are likely to get an
optimal set of public policies toward aging, neither do I think that the con¬
tinued aging of the population in the United States portends a national di¬
saster, requiring drastic measures such as abolishing social security or even
scaling it back greatly. The costs of an aging population have been exag¬
gerated and the benefits largely ignored. In this instance at least, as I hope
the reader will be persuaded, the facts of the matter are less alarming than
the fears about it.

In researching and writing this book, I have incurred immense debts. Ben¬
jamin Aller, Mark Fisher, Scott Gaille, Richard Hynes, Wesley Kelman,
Steven Neidhart, Andrew Trask, Clinton Uhlir, John Wright, and Douglas
Y’Barbo rendered invaluable research assistance. Christopher Hill fur¬
nished helpful research leads, and Edward Laumann, George Priest, Steven
Schlesinger, and Tom Smith provided me with valuable data. For generous
and exceedingly helpful comments on earlier drafts of one or more chapters
I am indebted to Michael Aronson, Ian Ayres, Gary Becker, Wayne Booth,
Margaret Brinig, Christine Cassel, Arthur Diamond, John Donohue, Larry
Downes, Ronald Dworkin, Frank Easterbrook, Jon Elster, Richard Epstein,
Robert Ferguson, David Friedman, Victor Fuchs, David Greenwald, John
Griffiths, Christine Jolls, William Landes, John Langbein, Edward Lazear,
Lawrence Lessig, Martha Nussbaum, Jay Olshansky, Tomas Philipson,
Charlene Posner, Eric Posner, Mark Ramseyer, Eric Rasmusen, George
Rutherglen, Cass Sunstein, John Tryneski, and Carolyn Weaver. These
Introduction 13

readers not only caught many mistakes but also opened up new vistas of
inquiry for me. My greatest debt of all is to Gary Becker, not only for his
helpful comments on the manuscript but also for a series of stimulating
discussions of its subject matter and for his foundational research on the
economics of human capital upon which this book builds.
Earlier versions of several chapters provided the text for the 1994 Tan¬
ner Lectures on Human Values at Yale University. I tried out other parts of
my argument at lectures or talks at the University of Virginia School of
Law, the Economic and Social Research Institute of Ireland, the City Front
Forum of the University of Chicago, the Law and Economics Workshop of
Harvard Law School, and an annual meeting of the American Law and
Economics Association. I thank the members of the audiences on all these
occasions for their many stimulating comments.
Since one of the points I emphasize is that the values and perspective
of elderly people may differ greatly from those of young and middle-aged
people, I should disclose that I wrote this book at the age of 55.
 -i-- vjrf

rPi • «-''
PV.--
M V J '^t.
■: . ....it Zh' ,0
■f •:■ '
■‘if* '.^^y ** ifb :’'''?4W

■'i^ ■• '•'^'<r? cic I■., ,

.-...Ktii,.-.,1^ t. _ .By ..,1, -.'r.,' ' ■ *■■'

w ■’■■ij ’ -»i: - , ^ »Tr .»•^•-.| t', .**j*«* w— >^^ir ^’'T^ ; V 'T “ »»4' • *^v •'M .' «*• *♦•-’

■ : “ i J .1 . ;, 'in ' - ._•• t- ..■, 1 '.v

'!'.; ..x^f^iii-'"' ■;
' -iiV iMt/- '; *'/ ■; ,,
i^.:
‘’^^■

M'-.S, .
■• V v' ‘ • > •« ■ • '.T ■' f «'
'*•. ' ’ * ;,rw' - k. . : ;trt,>ry’.;r^
.. . ';) • _ -.
f ■ '■'“ ■ ■; Y .'.TSJ
t '■.V ,»«, i/''.'■, t'• ,-k. .r'-' ■'
' ’■ ri' "'
■■' -'■ . 'U...l<'ScJ^s
V. tisv ‘;ti'' -5^ .- -t ,. •. i\,jXi^r:-. ■ ■■
T^Zjh
- ■■'‘’ ’•■ ' .‘ .'■^-- 0: -.r *1
■i f- U} 'r ■'y??: ^1- <V^7J7

f ,yv.j
Vcf/rV?.
!■? . .1 '' ■" • ' ■^ f I •■•'IJjM'. •1.- \,s

1 ‘W
St# ’ . \ I

nit* dV - 7'
^ j,4« «. ! • J « ..
' ■ .. .. *-
• ,l'*« * '■.••'sent |.e“ ,r.
■■ :vY‘^''^V.» MM V."." A'VJ -j ■ ■ .^, r
r*’,
,»r.

A*-,.,' %'. ■■'.* ki '

^'L, V;’JV ■ ' ‘ ^ t» ' # #. • J •»• ^' ■

w^'-' ■* .'♦♦5 ■ t, f
tn>r
'■ .»■( tv ■• : :S’^,A.-r, Va*- -V'Vk’k > V-^ "■'''■ ""sgtS
"^» >1’*, A ■ 'I'jy (#>*j!e-f».• (^i\ ^ -<. ' ■' ■.'. • tu, ^

-hK" -Y ffy '• '*5. .'. (,■

«»-'
sWv‘ v fc. - , y\.
.^.<,4
^ wn - r- .ys. ■■ ’iiJiiLr .'W

«< ■!'
{, w
 Part One

Aging and Old Age

as Social, Biological,
and Economic Phenomena
 1
What Is Aging, and Why?

The Process of Aging

As we get older, we “age.” Or do we? Some gerontologists believe that the
expression “normal aging” is an oxymoron; that “aging” does not denote
a process at all, but merely describes a medley of unhappy outcomes. As
we get older we are more susceptible to most diseases,' so maybe all it
means to “age” is to become increasingly afflicted by one or, more com¬
monly, several diseases until finally we are overpowered. On this construal
the only difference between an old person and a young one is that the for¬
mer is likelier to be sicker. If he happens not to be sicker, then he will be
identical to a young person.
One can acknowledge the fuzzy edges of the concept of “disease” ^
without finding this conception of aging remotely persuasive. For one
(little) thing, it ignores Aristotle’s point—the changing balance over the
life cycle between imagination and knowledge, a change that cannot be
described, without great semantic violence, as a form of illness. For an¬
other, the idea that “aging” denotes merely an increasing frequency of dis¬
ease ignores forms of physical and mental change that, while aptly char¬
acterized as marking a decline in capability, are again not aptly described

1. Dramatically so, beyond a point: of healthy 60-year-old men, only 30 percent can be
expected to be alive and healthy at 80. E. Jeffrey Metter et al., “How Comparable Are Healthy
60- and 80-Year-Old Men?” 47 Journal of Gerontology M73, M75 (1992). But medical science is
continuously improving these odds.
2. As famously illustrated by debates in psychiatry and law over whether homosexuality is
a “disease,” and whether having a psychopathic or sociopathic personality is a disease.

17
18 Phenomena of Aging and Old Age

as diseases. Most professional athletes in most sports, even if they escape

significant injury, are “old” by their late twenties or early thirties, but they
are not sick. Their muscles and nervous systems are not diseased in any
useful sense of the word. It is simply that their reflexes and running speed
have slowed slightly but critically.^ There are physiological causes of this
slowing, of course, and they could if one wanted be called “disease” fac¬
tors. But the “disease” of diminished athletic capabilities in one’s twenties
and thirties (or menopause in a woman’s forties or fifties) is sufficiently
different in the most socially relevant respects from such conditions as can¬
cer, coronary artery disease, stroke, and diabetes to warrant—to demand—
being called by a different name.'^
Aging is most usefully viewed as a process one element of which is an
inexorable decline across a broad range of bodily (including both physical
and mental) capabilities: call this “bodily decline.”^ Other elements of ag¬
ing, or that are correlated with it—the nonsomatic elements, examined
more closely in subsequent chapters—include the increasing proximity of
death as we get older, which affects the balance between imagination and
knowledge as intellectual resources and the incentive to invest in human
capital; the elfect of habit on adaptability to changed circumstances; the
accrual of experience in working at specific jobs, as distinct from the ac¬
crual of general life experience that Aristotle associated with aging; and
boredom as a consequence of long years of working at the same job. These
are age-correlated changes, but they are not explicitly somatic and not all
of them are declines.

3. Richard Schulz and Christine Curnow, “Peak Performance and Age among Superathletes:
Track and Field, Swimming, Baseball, Tennis, and Golf,” 43 Journal of Gerontology PI 13 (1988).
Mental performance, at least on pen-and-paper tests—an important qualification, as we shall see—
also begins to diminish perceptibly at early ages; in one study (from which unhealthy persons were
excluded), most of the decline occurred by ages 35 to 44. Kurt A. Moehle and Charles J. Long,
“Models of Aging and Neuropsychological Test Performance Decline with Aging,” 44 Journal of
Gerontology Pn6, PI77 (1989) (tab. 1).
4. For evidence that good physical health has only limited efficacy in retarding the charac¬
teristic memory loss from aging, see Douglas H. Powell (in collaboration with Dean K. Whitla),
Profiles in Cognitive Aging, ch. 5 (1994); Wojtek J. Chodzko-Zajko et al., “The Influence of Physi¬
cal Fitness on Automatic and Effortful Memory Changes in Aging,” 35 International Journal of
Aging and Human Development 265 (1992).
5. More precisely, though not necessarily more accurately, aging can be defined as “those
series of cumulative, universal, progressive, intrinsic, and deleterious functional and structural
changes that usually begin to manifest themselves at reproductive maturity and eventually culmi¬
nate in death.” Robert Arking, Biology of Aging: Observations and Principles 9 (1991). Unless,
of course, death occurs earlier for reasons unrelated to aging. For a comprehensive treatise on the
biology of aging (both human and animal), see Caleb E. Finch, Longevity, Senescence, and the
Genome (1990).
What Is Aging, and Why? 19

The physical side of bodily decline (using “physical” narrowly, in the

sense in which it is contrasted with “mental”) involves diminution in such
areas as athletic and related motor capabilities, reflexes, and muscle tone;
physical strength, energy, and stamina; acuity of vision, hearing, and other
senses; fertility and potency; scalp hair, hair color, and the smoothness of
skin; the efficiency of the immune system; height and the percentage of
weight accounted for by muscle. The mental side of the declivity includes
loss of memory (especially short-term memory), diminution in reckless
physical courage and in sexual desire, diminished willingness to take finan¬
cial risks, impairment of puzzle- and problem-solving ability, and reduced
willingness to adopt new ideas or reexamine one’s old ideas. Some of the
psychological changes may not be entirely somatic—we shall see that un¬
willingness to reexamine one’s old ideas has a counterpart in the behavior
of business firms, which do not age in a physical sense—but all have, I
believe, a somatic component. Mind and emotions, at least in a scientific
perspective, are dependent on bodily states; and the same, or at least the
same kind of, cytological and other physiological changes that produce the
symptoms of physical decline likewise produce those of mental decline.^
Although the process of aging can usefully be distinguished from the
age-related increase in susceptibility to specific diseases, that increase is a
reality which must not be ignored. Aging would have significance for is¬
sues such as the financing of medical care even if there were no normal
aging process but just an enhanced susceptibility to disease.
Resistance to the fact that there is such a thing as normal aging has
become common in our culture of heightened sensitivity. To some, age ste¬
reotyping is every bit as vicious as racial stereotyping. The concern is that
if everybody is believed to age, this might be thought to imply that every
old person is less competent intellectually than an otherwise similar young
person. That would indeed be false. Two distributions can have different
means but still overlap considerably. That is certainly the case in compar¬
ing the capabilities of young and old people; and for two reasons—that

6. For evidence and analysis of age-related physical and mental decline, see, for example,
id., ch. 5; Powell, note 4 above, at 69 (fig. 4.1); Handbook of Mental Health and Aging, chs. 6-13
(James E. Birren et al., eds., 2d ed. 1992); Timothy A. Salthouse, Theoretical Perspectives on
Cognitive Aging (1991), esp. ch. 7; Handbook of the Psychology of Aging, pt. 3 (James E. Birren
and K. Warner Schaie, eds., 3d ed. 1990); Nathan W. Shock et al.. Normal Human Aging: The
Baltimore Longitudinal Study of Aging, ch. 6 (1984); James L. Fozard et al., “Age Differences and
Changes in Reaction Time: The Baltimore Longitudinal Study of Aging,” 49 Journal of Geron¬
tology PI79 (1994); Kathryn A. Bayles and Alfred W. Kaszniak, Communication and Cognition
in Normal Aging and Dementia, ch. 5 (1987); Michaela Morgan et al., “Age-Related Motor Slow¬
ness: Simply Strategic?” 49 Journal of Gerontology Ml 33 (1994).
20 Phenomena of Aging and Old Age

people age at different rates, and that people start to age from different
levels of capability. A 75-year-old who had outstanding capabilities when
he was 30, and has aged slowly, not only may be immensely more capable
than a 75-year-old who was mediocre at 30 and has aged rapidly; he may
also—and this is what bothers people who complain about “ageism”—be
more capable than a mediocre 30-year-old.
But that there is increasing variability^ within age cohorts and overlap
between persons in different age cohorts does not refute the existence of
normal aging; it assumes it. Even cognitive as distinct from purely physical
aging—gradual until about the age of 65 and accelerating from then till
death—is normal and, pending scientific breakthroughs at present unfore¬
seen, inevitable for all of us.* Further evidence is that good physical and
even mental health appears not to retard cognitive aging significantly,^ as
one would expect it to do if such aging were simply a by-product of illness.
Nor does such aging appear to be, to a significant extent, an artifact of age-
cohort effects due to changes over time in environmental conditions such
as poverty and lack of education, or of sampling bias, or of disuse because
of lack of intellectual challenge or stimulation (“use it or lose it”).'° Far
from being “ageist,” moreover, a refusal to acknowledge normal, and in
particular normal cognitive, aging can create exaggerated doubts about the
competence of old people, doubts that the conception of normal aging can
allay. If cognitive decline is not a normal aspect of aging, but rather is
always a symptom of disease, the implication is that the vast majority of
old people are afflicted with senile dementia. Almost all elderly people
experience a cognitive decline the symptoms of which are difficult to dis¬
tinguish from the earliest manifestations of dementia; yet in most the con¬
dition does not progress to dementia." It is possible, given the steep age
gradient of dementia (of which more presently), that anyone who lived long
enough would become demented; but most people die before then.

7. Emphasized in Powell, note 4 above, esp. at pp. 12-14 and fig. 1.3; see also Dorothy
Field, K. Warner Schaie, and E. Victor Leino, “Continuity in Intellectual Functioning: The Role
of Self-Reported Health,” 3 Psychology and Aging 385, 390 (1988). Increased variability is vir¬
tually inevitable as a matter simply of mathematics. Imagine two groups that start off with a ca¬
pability of 100 and 50 respectively, and the members of each group age at different rates. After
many years, slow agers in the group that started at 100 will have a capability near 100, say 90,
while fast agers in the group that started at 50 will have a very low capability, say 10; so the spread
between the best and worst of the two groups will have widened with age.
8. See Powell, note 4 above, ch. 4, for a careful review; Salthouse, note 6 above.
9. See note 4 above.
10. See Salthouse, note 6 above, chs. 3 and 4.
11. Cf. Rajendra Jutagir, “Psychological Aspects of Aging: When Does Memory Loss Sig¬
nal Dementia?” 49 Geriatrics 45 (1994).
What Is Aging, and Why? 21

To avoid confusing normal cognitive aging with dementia requires dis¬

tinguishing carefully among the following terms: (1) Alzheimer’s disease,
or, as it nowadays is often called, SDAT (Senile Dementia of the Alzhei¬
mer’s Type), (2) dementia, (3) senile dementia, and (4) normal age-related
cognitive decline. Alzheimer’s disease, though commonly used by lay
people as a synonym for senile dementia, is actually a specific type of rap¬
idly progressive dementia that produces distinctive changes in the brain
tissues and that, though more common after age 65, can strike at earlier,
sometimes much earlier, ages; sufferers from Down’s syndrome are often
hit by Alzheimer’s disease in their teens. So renaming Alzheimer’s disease
“Senile Dementia of the Alzheimer’s Type” has been a source of confu¬
sion; indeed originally the term Alzheimer’s disease was limited to prese-
nile dementia.Some students of the disease continue to distinguish be¬
tween Alzheimer’s (presenile) and SDAT (senile), although they appear to
be a single disease which merely hits people at different ages, like many
cancers.
Dementia is the most general term for disabling mental deterioration,
and thus embraces a variety of specific disease states; and senile dementia
denotes dementia in old people. SDAT appears to account for a majority of
cases of senile dementia, perhaps as many as 80 percent, the rest being due
to such diseases or conditions as stroke, alcohol abuse, Parkinson’s disease.
Vitamin B-12 deficiency, and hydrocephalus. For my purposes the differ¬
ences between SDAT and senile dementia are unimportant. I shall use
“SDAT,” “Alzheimer’s,” and “senile dementia” interchangeably.
The number of old people afflicted with senile dementia is not known
with precision but has been responsibly estimated at 11.3 percent of the
entire 65-and-over population.'^ The percentage rises rapidly with age. In
the 65 through 74 group, it is only 3.9 percent; it rises to 16.4 percent for
persons 75 through 84 and to 47.6 percent for those 85 and older.'^ Thus,
while almost all old people suffer from some cognitive decline, especially
in fluid intelligence, which peaks earlier than crystallized,'^ only a minority
suffers from dementia, though it is a substantial minority.

12. Denis A. Evans et al., “Estimated Prevalence of Alzheimer’s Disease in the United
States,” 68 Milbank Quarterly 267 (1990).
13. Id. at 273. See also James C. Anthony and Ahmed Aboraya, “The Epidemiology of
Selected Mental Disorders in Later Life,” in Handbook of Mental Health and Aging, note 6 above,
at 27, 33.
14. Evans et al., note 12 above, at 274.
15. See Jutagir, note 11 above, at 46; Paul B. Baltes, Jacqui Smith, and Ursula M. Stau-
dinger, “Wisdom and Successful Aging,” 39 Nebraska Symposium on Aging 123, 139-143
(1992); and recall the discussion of these terms in the Introduction.
22 Phenomena of Aging and Old Age

We must be careful in interpreting these numbers. They do not differ¬

entiate between the mild early symptoms of dementia and the severe late
ones, and a further complication is that mild dementia does not always
progress to the severely demented state that is characteristic of SDAT. (This
intermediate state between normal age-related cognitive decline and pro¬
gressive dementia is called “Mild Cognitive Impairment,” or MCI.) If at¬
tention is limited to cases of severe rather than merely mild or moderate
cognitive impairment, the prevalence in the three age groups is said to fall
to 0.3 percent, 5.6 percent, and 19.6 percent, respectively.'^ But these may
be underestimates. The population sample on which they are based ex¬
cluded institutionalized persons, among whom the prevalence of severe
cognitive impairment is higher.Even taken at face value, the figures show
a very steep age gradient,'* implying that a continued rapid increase in the
size of the very oldest age group will cause an even more rapid increase in
the percentage of severely demented people in the elderly population.
To summarize the discussion thus far, age brings with it (1) increased
susceptibility to a number of diseases, (2) somatic changes that are a con¬
sequence of normal aging, and (3) nonsomatic changes that are a conse¬
quence of the same process. Somatic changes are of two kinds, (a) physical
and (b) mental. Nonsomatic changes are of three kinds: (a) the increasing
proximity of death (a purely “external” change, a change in the person’s
environment rather than in himself), which is the emphasis in the literature
of human capital; (b) the increased ratio of knowledge to imagination in
the cognitive balance—Aristotle’s point; and (c) changes due to time spent
working (experience and its baleful obverses, inflexibility, boredom, and
sometimes burnout) and therefore merely correlated with aging.
Not only are these changes correlated with each other, all being related

16. Evans et al., note 12 above, at 281. See also Fred Plum, “Dementia,” in Encyclopedia
of Neuroscience, vol. 1, p. 309 (George Adelman, ed., 1987).
17. In a study from which institutionalized persons were not excluded, the percentages of
persons suffering from severe cognitive impairment from all causes in the three age groups was
2.9, 6.8, and 15.8 percent, respectively. Anthony and Aboraya, note 13 above, at 35.
18. One study found that only 1 percent of the 65-70 population has severe dementia, while
43 percent of the 95-and-over population has it. Again both figures are underestimates because
persons having severe dementia but not institutionalized were excluded from the numerator, so
the percentages are of institutionalized demented persons in the entire (institutionalized and non-
institutionalized) elderly population. C. G. Gottfries, “Senile Dementia of the Alzheimer’s Type:
Clinical Genetic, Pathogenetic, and Treatment Aspects,” in Human Development and the Life
Course: Multidisciplinary Perspectives 31,34 (Aage B. Sprensen, Franz E. Weinert, and Lonnie R.
Sherrod, eds., 1986). For other estimates of the prevalence of senile dementia of various severities,
see Powell, note 4 above, at 140, 144-145 (tabs. 7.2, 7.4-7.5).
What Is Aging, and Why? 23

to age; some interact. Of particular significance for later chapters is the

interaction between the decline of fluid intelligence and the knowledge
shift identified by Aristotle. The combined effect is a pronounced age-
related shift from abstract to concrete reasoning,'® or in terms of another
useful Aristotelian dichotomy, from exact (logical or scientific) to practical
reasoning.^® This immediately helps us understand why, for example, ad¬
judication is a more geriatric profession than theoretical physics.
We should keep in mind that not all physical and mental changes cor¬
related with age are seriously negative and that some are even positive,
depending on circumstances. The “redistribution” of hair from the scalp to
the body, the wrinkling of the skin, and the thickening of nose and ears
have only cosmetic significance, though that is, of course, important to
many people. The pluses include escape from the diseases of the young and
slowing in the rate of growth of cancer cells. And there are changes that are
pluses for some people (perhaps for society as a whole), though not for
others, such as the reduction in sexual drive, anger, and aggressiveness that
accompanies diminished production of testosterone. Some middle-aged
and even elderly people are better-looking than they were when young.
The symptoms of old age do not all appear at the same time or progress
at the same rate. Despite my reference to “inexorable” decline, symptoms
of aging sometimes appear suddenly, as with the onset of presbyopia in
one’s forties, or of tinnitus, or of pattern baldness, and may plateau rather
than continue to grow worse. The rate of decline differs not only across
capabilities, but also, as I have pointed out, across persons,^' making the
classification of people in age groups an inescapably arbitrary method of
identifying the elderly. No one escapes the aging process, however, so that
even a “healthy” old person will be less capable along a variety of physical
and mental dimensions than an otherwise identical young person—though
the qualification “otherwise identical” is crucial. The percentage of un¬
healthy old persons is much greater than that of unhealthy young persons,
since age-related changes such as the diminished efficiency of the immune
system increase susceptibility to illness. The probability of death doubles
every eight years or so after a person reaches 30.^^ The incidence of serious

19. For evidence, see Salthouse, note 6 above, at 276—277; Steven W. Cornelius, Aging
and Everyday Cognitive Abilities,” in Aging and Cognition: Knowledge Organization and Utili¬
zation 411 (Thomas M. Hess, ed., 1990).
20. See Richard A. Posner, The Problems of Jurisprudence 71-73 (1990), and references
there.
21. For a dramatic example, see Arking, note 5 above, at 56-59 and fig. 3-8.
22. Id. at 42-43 and fig. 2-17.
24 Phenomena of Aging and Old Age

illnesses, especially of degenerative (as distinct from infectious) illnesses

such as cancer, stroke, and heart diseases, also rises at an increasing rate
with age.
Anyone who doubts that there are palpable, substantial, systematic,
universal, measurable, demoralizing, and in the present state of biological
and medical knowledge inevitable declines in physical and mental func¬
tioning even for the “normal” or “healthy” aged in this the world’s most
medically pampered society—anyone who believes that these age-related
“declines” are a product of mass delusion or of vicious, irrational preju¬
dice—is out of touch with reality.But there is scope for rational debate
over when decline sets in, how steep it is, how much variance there is
among persons within particular age groups, and the degree to which the
cognitive effects of aging may, up to a point anyway, be offset by experi¬
ence of life, including work experience, and by compensatory strategies
such as being more careful or taking more time to plan or accomplish
tasks.^'* The rate of aging, moreover, mental and especially physical, can
be, and is being, retarded by improvements in diet, by increased exercise,
and by advances in medical technology.We cannot eliminate old age, but
we can postpone it; we have postponed it. We are much less likely to think
of a healthy 60-year-old or even 70-year-old as being “old” than we were
thirty years ago. So while there are more “old” people alive today than
ever before, there are fewer than the shift in the age distribution might be
thought to imply.
And there is a danger of exaggerating the economic and social sig¬
nificance of the characteristic age-related declines in physical and mental
performance. Declines in mental functioning tend to be measured by pen-

23. For some striking evidence of age-related decline, see James N. Schubert, “Age and
Active-Passive Leadership Style,” 82 American Political Science Review 763 (1988); for a com¬
prehensive review of the evidence, see Powell, note 4 above, ch. 4; and see the other references in
note 6. This is not just a conspiracy of the young and the middle-aged against the old; some of the
most, vivid, eloquent, and arresting depictions of that decline come from elderly people themselves.
Notable examples are Simone de Beauvoir, Old Age (1972), esp. ch. 7, and B. F. Skinner, “Intel¬
lectual Self-Management in Old Age,” 38 American Psychologist 239 (1983).
24. See, for example, K. Warner Schaie and Sherry L. Willis, “Adult Personality and Psy¬
chomotor Performance: Cross-Sectional and Longitudinal Analysis,” 46 Journal of Gerontology
P275 (1991); Neil Charness and Elizabeth A. Bosman, “Expertise and Aging: Life in the Lab,” in
Aging and Cognition: Knowledge Organization and Utilization, note 19 above, at 343; James E.
Birren, Anita M. Woods, and M. Virtrue Williams, “Behavioral Slowing with Age; Causes, Or¬
ganization, and Consequences,” in Aging in the 1980s: Psychological Issues 293, 302-303
(Leonard W. Poon, ed., 1980); Daniel Coleman, “Mental Decline in Aging Need Not Be Inevi¬
table,” New York Times (national ed.), April 26, 1994, p. B5.
25. See, for example, John W. Rowe and Robert L. Kahn, “Human Aging; Usual and Suc¬
cessful,” 237 Science 143 (1987); Baltes, Smith, and Staudinger, note 15 above, at 133-134.
What Is Aging, and Why? 25

and-paper tests and other laboratory-type experimental procedures that ex¬

aggerate the decline in useful capabilities over the life span.^® A related but
more fundamental point, which I explore in chapter 4, is that the physical
and mental capabilities of the young are often in excess of the economic
and social demands placed upon them, so that up to a point—the point at
which the excess has been aged away—the aging process may not cause a
socially relevant diminution in capabilities. Another source of an exagger¬
ated impression of the effects of aging is failure to grasp a point that I shall
make in chapter 5—that elderly people rationally substitute time (which is
cheap for them) for other inputs into activity and as a result move and speak
more slowly, more hesitantly, than they are physically and mentally capable
of doing.
In view of the large preponderance of women in the elderly population,
an important question is whether aging affects men and women differently.
If so, old men and old women would be on average more different from
each other than young men and young women even after correction for
transient features of the social environment, for example the fact that to¬
day’s old women had less education relative to men than today’s young
women. This issue has been studied extensively, and as yet inconclusively.
But it appears that, if there are sex differences in the rate or character of
aging, they are small.^"^ There are, of course, more elderly women than el¬
derly men. But it appears that, in a comparison of survivors, men and
women of the same age are not at different points in the process of aging—
do not differ in “agedness”—though they do differ, on average, in their
proximity to death.

Aging and Evolution

We have not yet considered why the body (and hence mind) ages. The best
explanation is genetic.^* Maintenance of an animal’s body, like mainte-

26. Paul Verhaeghen, Alfons Marcoen, and Luc Goossens, “Facts and Fiction about
Memory Aging: A Quantitative Integration of Research Findings,” 48 Journal of Gerontology
P157 (1993). For particularly (I think excessively) far-reaching criticisms of the evidence for age-
related decline in mental ability, see'Gisela Labouvie-Vief, “Individual Time, Social Time, and
Intellectual Aging,” in Age and Life Course Transitions: An Interdisciplinary Perspective 151
(Tamara K. Hareven and Kathleen J. Adams, eds., 1982).
27. Powell, note 4 above, ch. 6. Elderly women tend, however, to be somewhat more frail
than men of the same age. Margaret J. Penning and Laurel A. Strain, “Gender Differences in
Disability, Assistance, and Subjective Well-Being in Later Life,” 49 Journal of Gerontology S202
(1994).
28. Thomas B. L. Kirkwood, “Comparative Life Spans of Species: Why Do Species Have
the Life Spans They Do?” 55 American Journal of Clinical Nutrition 119 IS (1992). Evolutionary
26 Phenomena of Aging and Old Age

nance of an automobile, is costly. The more resources that are devoted to

maintenance, the fewer that are available for the evolutionarily critical at¬
tribute of reproduction; a highly complex animal built to live a really long
time would require a very long, and hence to parents very costly, period of
gestation and infant development. So fewer of these built-to-last animals
would be produced. If accidental destruction of such an animal, which main¬
tenance would not prevent, was a significant risk, the added longevity might
not offset (whether through the provision by this exceptionally long-lived
animal of additional protection to its descendants, or by its having a longer
reproductive life) the reproductive cost to its parents. Less complex animals,
such as the turtle, that are at reduced risk of accidental destruction tend to
be long-lived because the costs of “designing” the animal for long life are
lower.But if reproductive fitness is sacrificed to durability unnecessarily,
because the durability does not translate into commensurate survival, the
parents of the more durable creature will have fewer descendants than their
less durable competitors and their line will eventually become extinct.
This analysis, which is supported by evidence that animals age, men¬
tally as well as physically, much as human beings do,^° explains why we
wear out and die, and more specifically why the death rate increases rapidly
after our prime reproductive period. But it does not provide much insight
into the psychology or behavior of old people. For, as our sketch of the
genetic theory of aging will have prepared us to see, it is unlikely that there
is a genetic program for extended human survival, although this depends
on how extended. It is relatively easy to see why in the evolutionary era—
the prehistoric era in which, through the operation of natural selection,
human beings evolved to approximately their present biological state—it

theories of aging are summarized briefly in Steven M. Albert and Maria G. Cattell, Old Age in
Global Perspective: Cross-Cultural and Cross-National Views 27-29 (1994); S. Jay Olshansky,
Bruce A. Carnes, and Christine Cassel, “The Aging of the Human Species,” Scientific American,
April 1993, pp. 46,49-50; and Arking, note 5 above, at 83-88. For a fuller summary, see Bruce A.
Carnes and S. Jay Olshansky, “Evolutionary Perspectives on Human Senescence,” 19 Population
and Development Review 793 (1994), and for an extended treatment, see Michael R. Rose, Evo¬
lutionary Biology of Aging (1991).
29. J. Whitfield Gibbons, “Life in the Slow Lane: Lugging a Shell Around Has Its Rewards,
as Turtles Have Known for Millions of Years,” Natural History, 1993, no. 2, p. 32. This is a serious
article, despite the childish title.
30. See, for example, Diana S. Woodruff-Pak, “Mammalian Models of Learning, Memory,
and Aging,” in Handbook of the Psychology of Aging, note 6 above, at 234; National Research
Council, Mammalian Models for Research on Aging 307 (1981). “A failing memory and an in¬
ability to remember new names and faces have their counterpart at the animal level.” P. L. Broad-
hurst, The Science of Animal Behaviour 110 (1963).
What Is Aging, and Why? 27

might have been adaptive for men to live for several years beyond their
prime, or women to live several years after menopause terminated their
reproductive capacity. The older man could still reproduce, and his accu¬
mulated knowledge (particularly valuable in a preliterate culture) might
compensate for physical decline in enabling him to furnish valuable protec¬
tive services to his children and grandchildren.^' The older woman could
better protect her younger children, who would not yet be fully grown, as
well as assist in the care of her grandchildren.^^ If these older (not elderly
by our standards) people were valuable to their younger relatives during the
evolutionary era, and hence valued by them—otherwise the older people
would not have had good prospects for survival—this may explain why
most people even today, even in the United States, feel some respect and
protectiveness for older people, or at least for their own elderly relatives;
these feelings may be instinctual.
The idea that nonreproducing relatives can promote inclusive fitness
(the number of copies of their genes in their descendants), and therefore
that there may be a genetic program for their survival and for their protec¬
tion by their other relatives, is no longer a novelty. It is, for example, the
key to the genetic theory of homosexuality.^^ Equation 1.1 formalizes the
idea.^'' The optimal life expectancy of individual / at a particular age (L,) is
shown as a function of i’s remaining reproductive potential (/?,) and of k’s
remaining reproductive potential (pk), where k is some relative (kin) of i,
discounted by a measure of the closeness of the kinship (rj and by the

31. David Gutmann, Reclaimed Powers: Toward a New Psychology of Men and Women in
Later Life 216 (1987).
32. Jane B. Lancaster and Barbara J. King, “An Evolutionary Perspective on Menopause,”
in In Her Prime: New Views of Middle-Aged Women 1 (Virginia Kerns and Judith K. Brown, eds.,
2d ed. 1992); Peter J. Mayer, “Evolutionary Advantages of the Menopause,” 10 Human Ecology
All (1982). Gutmann, note 31 above, at 163-173, 232, stresses the executive role of the postmen¬
opausal woman in the extended family characteristic of early families. Another way to make the
general point is that menopause is a form of birth control, which by reducing the number of a
woman’s children increases the probability that at least some will survive. Cf. Sarah Blaffer Hrdy,
“Fitness Tradeoffs in the History and Evolution of Delegated Mothering with Special Reference
to Wet-Nursing, Abandonment, and Infanticide,” 13 Ethology and Sociohiology 409 (1992).
33. For which there is growiiTg evidence, reviewed in my book Overcoming Law, ch. 26
(1995). Particularly striking is the evidence of much greater concordance for homosexuality be¬
tween identical than between fraternal twins, the former having a closer genetic relationship but
presumably no greater environmental similarity than the latter, since twins generally have the same
home environment regardless of whether they are identical or fraternal.
34. See Denys de Catanzaro, “Evolutionary Pressures and Limitations to Self-
Preservation,” in Sociobiology and Psychology: Ideas, Issues and Applications 311, 317-318
(Charles Crawford, Martin Smith, and Dennis Krebs, eds., 1987).
28 Phenomena of Aging and Old Age

benefits that z’s continued existence confers on k (bk), all summed over all
of i’s kin. So:

Li = Pi + 1. (b^rkpA- (1-1)

The equation shows that a reduction in an individual’s personal reproduc¬

tive fitness, say because of menopause, can be offset by an increase in the
reproductive fitness of kin whom the individual assists.
Although evolutionary theory may explain the survival of persons to
middle age, there probably were very few old people in the human evolu¬
tionary era. People lived in hunter-gatherer societies. Life in such societies
is physically challenging, because, among other reasons, it is nomadic—
people move around a lot. There is little surplus food with which to main¬
tain people who are not directly productive (see chapter 9) and little surplus
energy for carrying helpless old people from camp to camp. Only 8 percent
of Yanomama Indians, a primitive South American tribe, survive to the age
of 65, compared to 85 percent of modern Americans.^^ In true neolithic
cultures, as little as 2 percent of the population may have survived to the
age of 50.^^ With few people surviving to old age, there could not have
been much natural selection among old people having different qualities—
a process that would result eventually in the accentuation of those qualities
that enabled a person to have more descendants—because there could not
have been much variance. We know there were plenty of young women in
the evolutionary era, just as today, so it is plausible to imagine that selection
took place in favor of those having qualities—such as fertility and affection
for children and attractiveness (and being attracted) to males likely to be
good protectors of children—that would tend to increase the number of
their descendants. But with so few old people for selection to work on, a
comparable process is not easily envisaged for them.^^ The social and ma¬
terial progress of mankind has brought about a stage of life that the genes
have not choreographed.
Of course there is an element of circularity in arguing that the fittest

35. Albert and Cattell, note 28 above, at 31 (tab. 2.1).

36. Gy. Acsddi and J. Nemeskdri, History of Human Life Span and Mortality 188 (1970)
(tab. 58). See generally William Petersen, “A Demographer’s View of Prehistoric Demography,”
16 Current Anthropology 227, 232-234 (1975); Gottfried Kurth, “Comment,” id. at 239. For a
higher estimate, see Nancy Howell, “Toward a Uniformitarian Theory of Human Paleodemogra-
phy,” in The Demographic Evolution of Human Populations 25, 35, 38 (R. H. Ward and K. M.
Weiss, eds., 1976). These statistics are of course potentially misleading, since if the vast majority,
of persons die in infancy or childhood, a large fraction of adults might be elderly people even
though the probability (at birth) of survival to old age was very low.
37. As emphasized in Carnes and Olshansky, note 28 above, at 801-802.
What Is Aging, and Why? 29

elderly were not selected for because not enough people survived to what
we would regard as old age to enable natural selection to work on them.
If old age conferred a substantial benefit on one’s descendants, perhaps
by enabling the transmission to them of valuable information that would
increase their reproductive fitness, we might expect more old people to
survive, even if the “design” of their bodies that enabled such survival
sacrificed some reproductive fitness to their greater durability. But the in¬
formational value of older people to society may not grow much after they
are middle-aged; and it may be that only a tiny fraction of each age cohort
need survive even to middle age in order to pass on to the succeeding cohort
essential information about food, predators, and social structure.^* If so,
there would be little evolutionary value to “engineering” human beings to
survive into old age, and there would be an inevitable cost in diminished
reproductive fitness.
The underlying point is that our genetic endowment, including our bio¬
logical “clock,” is adaptive to a different environment from that of today.
Recall my earlier point about excess capability. If life was physically and
perhaps even mentally more challenging in the hunter-gatherer era than it
is in the modern era,^^ the young may have become programmed with
physical and mental capabilities that are not required for most activities of
modern life, while the old with their diminished capabilities may neverthe¬
less be capable of coping in the modern era to a degree impossible in a
hunter-gatherer society. The less that is demanded of human beings, the
more likely they are to be able to meet the demand despite diminished
capacity. Aging is not an accident of evolution, but survival to old age
may be.
This discussion illustrates how evolutionary biology can cast light on
social issues relating to old age even if survival to old age lacks survivor¬
ship properties in a Darwinian sense: even if, like birth-control pills and
sperm banks, old age as we understand it did not exist in the environment
from which we derive our genetic legacy. We shall consider in subsequent
chapters, particularly chapter 5, other examples of the paradoxical fruitful¬
ness of genetics in explaining the genetically unprogrammed stage of life
that we call old age.

38. This assumes, but plausibly, that members of the older generation cannot feasibly con¬
fine the transmission of their knowledge to their own descendants.
39. There is more scope for a variety of mental abilities in the modern world; but constant
alertness, concentration, and quickness are not conditions of survival, as they may well have been
in the conditions of extreme hardship and precariousness that characterized human life in the
evolutionary period.
30 Phenomena of Aging and Old Age

If natural selection implies that we are “designed” for limited dura¬

bility, the conquest of old age by medicine may seem a quixotic endeavor,
and the proper focus of geriatric research the alleviation of the disabilities
of old age. Biologists who accept the genetic account of aging that I have
been sketching do tend to believe that we simply are not programmed by
our genes for indefinite life.''® They may be right. There may be a biological
limit to the number of times human cells can divide and thus replace them¬
selves as they wear out. But if so, the limit is not known and, in any event,
may, for all one can know today, be extendable indefinitely by the medical
science of the future."" Just more effective control of known risk factors
that are controllable with existing techniques would increase life expec¬
tancy substantially."^ Consistent with this suggestion, recent data from
Sweden confirm the likelihood that life expectancy will continue increas¬
ing even in populations where it already is very long."^ Even without major
research breakthroughs, it is entirely possible that a life expectancy at birth
of 85 years is achievable."" That would imply a large expansion in what is
already a very large elderly population.

40. See, for example, Carnes and Olshansky, note 28 above, at 802-804. For a good account
of the debate over whether there is a biological limit to the human life span, see Marcia Barinaga,
“How Long Is the Human Life-Span?” 254 Science 936 (1991).
41. For good discussions, see Samuel H. Preston, “Demographic Change in the United
States, 1970-2050,” in Demography and Retirement: The Twenty-First Century 20, 30-37
(Anna M. Rappaport and Sylvester J. Schieber, eds., 1993); Kenneth M. Weiss, “The Biology of
Aging and the Quality of Later Life,” in Aging 2000: Our Health Care Destiny, vol. 1: Biomedical
Issues 29 (Charles M. Gaitz and T. Samorajski, eds., 1985).
42. Kenneth G. Manton, Eric Stallard, and Burton H. Singer, “Methods for Projecting the
Future Size and Health Status of the U.S. Elderly Population,” in Studies in the Economics of
Aging 41 (David A. Wise, ed., 1994).
43. James M. Vaupel and Hans Lundstrbm, “Longer Life Expectancy? Evidence from Swe¬
den of Reductions in Mortality Rates at Advanced Ages,” in id. at 79.
44. S. Jay Olshansky, Bruce A. Carnes, and Christine Cassel, “In Search of Methuselah:
Estimating the Upper Limits to Human Longevity,” 250 Science 634 (1990).
 2
Old Age Past, Present, and Future

My aim in this chapter is to introduce the reader to the basic demographic

and economic facts concerning elderly people. Many of these facts are used
in later chapters and could have been introduced for the first time in those
chapters. But it may help the nonspecialist reader’s orientation, and the
specialist’s maintenance of perspective, to have them set out in a coherent
narrative at the outset. The first part of this chapter is historical and, in a
limited way, comparative. The second and third parts focus on the current
and future United States. I argue that the aging of the population, once its
economic and not merely its financial aspects are understood and the loose
use of terms like “dependency ratio” corrected, should not be considered
a disaster, although it certainly is not an occasion for complacency.

The History of Old Age

Old people are found in all societies, but in very different proportions. (The
threshold of old age also differs across societies.) I alluded to Nestor, the
old man of the Homeric epics, in the Introduction. It is often supposed that
old people must have a special value in a preliterate society such as that
depicted in the Homeric epics—and an African motto declares, “When an
old man dies, a library burns.” ' But we shall see in chapter 9 that the reality
is more complex than this. The social status of the old has varied bewilder-
ingly across different cultures and eras, and even within them. If Nestor

1. Georges Minois, History of Old Age: From Antiquity to the Renaissance 9 (1989).

31
32 Phenomena of Aging and Old Age

cuts a great figure in the Iliad, Odysseus’s father, Laertes, cuts no figure in
the Odyssey; for that matter old Priam doesn’t cut a very impressive figure
in the Iliad.
We do not know a great deal about the age profile even of Western
populations much before the nineteenth century. Estimates such as that 10
to 15 percent of the citizen population of Sparta was 60 or older ^ are highly
tentative, to say the least.^ They are also misleading if one is trying to get a
sense of life expectancy, because a low birth rate or a high rate of death of
young men in battle or young women in childbirth can result in a high
percentage of a very small population being old even if people deteriorate
with age so rapidly that few of those spared death in childhood or early
adulthood survive to old age. Historically, falling birth rates have been
more important than increased longevity in raising the average age of a
nation’s population.
Coming to more recent times, we know that there were plenty of old
people in medieval Europe, that they were mostly men because so many
women died giving birth, and that many old men found a refuge in monas¬
tic and priestly vocations, where the age-related decline in performance as
a warrior was not disqualifying.'' We also know that ours is not the first
epoch in Western history to experience a rapid aging of the population. The
fraction of old people in the population soared in the fourteenth and fif¬
teenth centuries, because the Black Death tended to spare the old.^ The
increase spurred communal living for the elderly—the nursing home and
retirement community are not new ideas—and the concept of retirement
emerged as a way of coping with a felt surplus of old people.*’ There have
been retired people as long as there has been recorded history, and probably
longer. But the idea of stopping work before forced by ill health or decrepi¬
tude to do so—the idea of retirement as a distinct, more or less universal
phase of the life cycle—is relatively recent.

2. Ephraim David, Old Age in Sparta 9-13 (1991).

3. Tim G. Parkin, Demography and Roman Society (1992), an exemplary study, declines to
endorse any estimates of the elderly population of ancient Rome. For a useful compendium of
estimates spanning the fourteenth through eighteenth centuries, see Herbert C. Covey, “The Defi¬
nitions of the Beginning of Old Age in History,” 34 International Journal of Aging and Human
Development 325, 332 (1992).
4. Minois, note 1 above, at 179-183.
5. Id. at 210-217. In fifteenth-century Tuscany, about 15 percent of the population was over
60 (compared to 23 percent in the United States today), with males in the majority. Id. at 213.
6. Id. at 211, 246. A sudden decline in the young population would, it is true, be expected to
increase the demand for older workers, but given their limited working capacity the ratio of depen¬
dent old to productive young might still increase on balance.
Old Age Past, Present, and Future 33

Figure 2.1 Percentage of population 65 or older

After the era of the plague ended, the percentage of old people fell. In
eighteenth-century America, it appears that no more than 2 percent of the
population was 65 years old or over. As late as 1851, only 4.7 percent of
English people were 65 or older and only 0.65 percent 80 or older. The
corresponding figures for the United States were 3 percent (1870) and
0.37 percent. The difference between the American and the English figures
probably reflects the large number of young immigrants in the United
States rather than a greater mortality of old people here.^ Over the entire
period 1551 to 1901, the over-60 population of England oscillated between
5 and 10 percent.*
As recently as 1950, the percentage of people 65 and older in the
wealthy nations had not yet reached 8 percent (figure 2.1).® Since then,
rising real income (a causal factor in longevity), better health care, and a
falling birth rate have raised the percentage substantially. Between 1950

7. The statistics are from U.S. Bureau of the Census, Seventh Census of the United States:
1850 xlii (1850); U.S. Bureau of the Census, Historical Statistics of the United States: Colonial
Times to 1970 15 (1976); David Hackett Fischer, Growing Old in America 222 (1977) (tab. 1);
B. R. Mitchell, British Historical Statistics 15 -16 (1988); Mitchell, European Historical Statistics
1750-1970 37 (1976); Nathan Keyfitz and Wilhelm Flieger, World Population: An Analysis of
Vital Data 312, 479(1968).
8. E. A. Wrigley and R. S. Schofield, The Population History of England 1541-1871: A
Reconstruction 216 (1981) (fig. 7.4).
9. The data used to construct figure 2.1 and figure 2.2, as well as tables 2.1 and 2.2, are from
United Nations, Department of International Economic and Social Affairs, The Sex and Age Dis¬
tribution of Population: The 1990 Revision of the United States Global Population Estimates and
Pro/ec/mns (Population Study No. 122, 1991).
34 Phenomena of Aging and Old Age

Figure 2.2 Percentage of population 80 or older

and 1990, the percentage of the population in the developed world that was
65 years old or older rose from less than 8 percent to more than 12 percent.
It is expected to be almost 14 percent by the year 2000.
The corresponding growth in the percentage of the population that is
80 years or older, depicted in figure 2.2, has been even more dramatic—
from 1 percent to 2.6 percent. By the year 2000, it is expected to be almost
3 percent. The contrast with the less-developed (“Third-World”) countries
is striking. There the percentages of old and very old are not much higher
today than they were in Western Europe and North America in 1850.
Tables 2.1 and 2.2 present data for the United States, and for several
other countries selected for comparison. The percentage of the U.S. popu¬
lation that is 65 or older grew from 8.1 percent in 1950 to 12.6 percent in
1990 and is projected to reach 12.8 percent by the end of the century. Al¬
though high historically and compared with countries like Nigeria, these
figures are generally lower than the corresponding figures for Western
European countries and Japan, countries that have both lower rates of
immigration and lower birth rates than the United States. The disparity
is less marked with regard to the percentage of the population that is 80
or older, perhaps because of heavy expenditures on health care for aged
Americans.
The tables contain forecasts of the elderly population in the next
century because they are part of the data set from which the tables were
constructed, but such forecasts should be taken with a large grain of salt;
they may easily be overtaken by continuing rapid advances in medical sci¬
ence. Another point to bear in mind about gerontological statistics is that
they are unreliable for the very old—persons 90 and older and especially
Old Age Past, Present, and Future 35

Table 2.1 Percentage of Population 65 or Older

Country/Area 1950 1960 1970 1980 1990 2000 2010 2020

Canada 7.7 7.5 7.9 9.5 11.4 12.7 14.4 18.8

France 11.4 11.6 12.9 14.0 13.8 15.4 15.7 19.3

Germany (FRG) 9.4 10.8 13.2 15.5 15.4 17.0 20.4 22.2

Japan 4.9 5.7 7.1 9.0 11.7 15.9 19.6 23.7

Nigeria 2.4 2.3 2.4 2.5 2.5 2.6 2.8 3.2

Sweden 10.3 12.0 13.7 16.3 18.1 17.1 18.8 21.8

United Kingdom 10.7 11.7 12.9 15.1 15.4 15.2 15.7 18.2

United States 8.1 9.2 9.8 11.3 12.6 12.8 13.6 17.5

World 5.1 5.3 5.4 5.9 6.2 6.8 7.3 8.7

Less-developed 3.8 3.8 3.8 4.0 4.5 5.0 5.6 7.0

More-developed 7.6 8.5 9.6 11.5 12.1 13.7 14.8 17.4

Table 2.2 Percentage of Population 80 or Older

Country/Area 1950 1960 1970 1980 1990 2000 2010 2020

Canada 1.1 1.2 1.4 1.8 2.3 3.0 3.7 4.1

France 1.7 2.0 2.3 3.1 3.5 3.3 3.8 3.9

Germany (FRG) 1.0 1.5 1.9 2.7 3.7 3.6 4.6 5.9

Japan 0.5 0.7 0.9 1.4 2.2 3.0 4.3 6.0

Nigeria 0.2 0.2 0.2 0.2 0.2 0.3 0.3 0.4

Sweden 1.5 1.9 2.3 3.2 4.3 4.8 5.0 5.2

United Kingdom 1.5 1.9 2.2 2.8 3.4 3.6 4.0 4.2

United States 1.1 1.4 1.8 2.3 2.8 3.3 3.8 3.9

World 0.5 0.6 0.7 0.8 1.0 1.1 1.3 1.5

Less-developed 0.3 0.3 0.4 0.4 0.5 0.6 0.8 1.0

More-developed 1.0 1.2 1.5 2.1 2.6 2.8 3.5 4.1

100 and older—because some very old people exaggerate their age. There
is still prestige to being a nonagenarian and especially a centenarian. There
may be fewer than half as many centenarians in the United States as the
official estimate of 50,000.'“

10. Bert Kestenbaum, “A Description of the Extreme Aged Population Based on Improved
Medicare Enrollment Data,” 29 Demography 565 (1992), esp. p. 573. For a rich compendium and
analysis of statistics concerning the aging of the American population, see Jacob S. Siegel, A
Generation of Change: A Profile of America s Older Population, ch. 1 (1993).
36 Phenomena of Aging and Old Age

Anxiety over Gerontification

Two major concerns have been voiced about the rapidly growing number
of elderly people in wealthy countries like the United States. The first is the
heavy medical expense of older people. Although as yet there appears to
be no correlation, after other causes are corrected for, between the average
age of a nation’s population and the nation’s per capita expenditures on
health care," the combination of increasing medical costs due primarily to
advances in medical technology with a rapid increase in the number of
elderly people portends a significant increase in aggregate medical expen¬
ditures unless the demands for health care of younger people are to be
scanted. Already persons 65 and older, though less than 13 percent of the
population, account for more than a third of all expenditures on health care
in the United States.And this ignores the cost, as yet largely nonmonetary
because borne by family members in the form of personal services rather
than cash outlays, of the home care that many elderly people require. As
we shall see in chapter 11, this cost is real—it is not just a matter of chil¬
dren’s bounteous and unstinting desire to care for their aged parents.
The second concern is that the increase in nonworking life expectancy
associated with the growing fraction of elderly people implies a decline in
the ratio of productive to consuming years over the life cycle, and that the
elderly will use their political muscle to force the productive young to sup¬
port the consuming old. The decline in the ratio of producing to consuming
years over the life cycle is not an inevitable concomitant of increasing lon¬
gevity. The ratio depends among other things on the age of retirement,
which might increase proportionately to increases in longevity. But this has
not happened. On the contrary, the average age of retirement has been de¬
clining at the same time that life expectancy has been rising. Figure 2.3
compares, for the period 1950 to 2000, the average life expectancy of men
with their average retirement age." In 1950, male life expectancy was ac¬
tually below the average age of retirement, while by 1990 it exceeded the

11. Thomas E. Getzen, “Population Aging and the Growth of Health Expenditures,” 47
Journal of Gerontology S98 (1992).
12. Daniel R. Waldo et al., “Health Expenditures by Age Group, 1977 and 1987,” Health
Care Financing Review, Summer 1989, p. Ill; U.S. Senate Special Committee on Aging et al..
Aging America: Trends and Projections 133 (1991 ed.). All statistics in the rest of this chapter are
for the United States.
13. The sources for figure 2.3 are Cynthia Taeuber, Sixty-Five Plus in America 25 (U.S.
Bureau of the Census, Current Population Reports, Special Study P23-178 RV, revised May 1993)
(tab. 3-1); Murray Gendell and Jacob S. Siegel, “Trends in Retirement Age by Sex, 1950-2005,”
Monthly Labor Review, July 1992, pp. 22, 27 (1992) (tab. 4); U.S. Senate Special Committee on
Aging et al.. Aging America: Trends and Projections 25 (1987-1988 ed.).
Old Age Past, Present, and Future 37

—■— Median
retirement age

Life
expectancy
at birth

Figure 2.3 Median retirement age and life expectancy at birth, U.S. males

average age of retirement by almost a decade. A comparable graph for

women would show a greater life expectancy (as we shall see shortly) and
a higher age of retirement, but the difference between the two figures would
be approximately the same as shown in figure 2.3 for the men.
Life expectancy at birth is only a crude guide to the age distribution of
the adult population, because it is sensitive to changes in infant and child
mortality. But infant and child mortality has been low for many years, so con¬
fining attention to adult life expectancy, as in figure 2.4,does not alter the
picture presented by the previous figure. The trend depicted in figure 2.4 is,
incidentally, one of long standing. In 1840, only 50 percent of 20-year-old
Americans could expect to survive to 65 and 38.8 percent to 70. By 1910,
these figures had risen to 69.3 percent and 59.5 percent respectively.'^
Given biological constraints on longevity, one would expect efforts at
increasing it to encounter sharply diminishing returns. Yet, on the contrary,
as shown in figure 2.5, which plots the rate of change in adult life expec¬
tancy, the annual rate of increase has itself been increasing throughout most

14. The sources for figures 2.4 and 2.5 are “U.S. Longevity at a Standstill,” in Metropolitan
Life Insurance Company, Statistical Bulletin, July-Sept. 1992, pp. 2, 8; and Life Tables from the
1950, 1970, and 1990 editions of U.S. Bureau of the Census, Statistical Abstract of the United
States. A pattern similar to that in figure 2.4 is generated by data on life expectancy at age 50. See,
for example, Peter Laslett, “The Emergence of the Third Age,” 7 Ageing and Society 133, 144
(1987) (tab. 2).
15. Fischer, note 7 above, at 225 (tab. 4).
38 Phenomena of Aging and Old Age

Figure 2.4 Remaining life expectancy at age 20 for white Americans, 1910-1989

Figure 2.5 Remaining life expectancy at age 20 for white Americans, 1910-1989.
Log scale

of the century. It is no surprise, therefore, that, as shown in figure 2.6,'^ the

percentages of the population that are old and very old have accelerated
markedly in recent decades, helped by a declining birth rate. We saw in
chapter 1 that there is no firm reason to believe that increases in life expec-

16. The sources for figure 2.6 are U.S. Bureau of the Census, Historical Statistics of the
United States: Colonial Times to 1970, note 7 above, at 15, and U.S. Senate Special Committee on
Aging et al., note 12 above, at 7 (tab. 1-2).
Old Age Past, Present, and Future 39

Figure 2.6 Percentage of U.S. population 65 or older and percentage 85 or older,

1870-2050

tancy have reached or are about to reach some “natural” limit, although it
is possible that a life expectancy much above 85 years at birth for the popu¬
lation as a whole cannot be attained without major research breakthroughs.
The increase in longevity has not been equal for both sexes. Through¬
out the twentieth century female longevity has been increasing substan¬
tially faster than male (see figure 2.4) and the result is a considerable
imbalance between the number of elderly men and the number of elderly
women. In 1989 there were only 39 percent as many men as women among
Americans 85 and older, meaning that women outnumbered men by better
than 5 to 2 in this age group, while in the entire population 65 and above
women outnumbered men 3 to 2.'^
The growth in the number of older people has been paralleled by a
decline in the labor-force participation of this group because of the trend to
earlier retirement. The combination of these trends has caused the “old-age
dependency ratio,” when defined mechanically as the ratio of the number
of people 65 or older to the number of people 20 to 64 years old, to rise
from .173 in 1960 to .209 in 1990. The Social Security Administration and
the Bureau of the Census predict a further increase in the ratio to between

17. Id. at 17 (chart 1 -9). See also Leonard A. Sagan, The Health of Nations: True Causes of
Sickness and Well-Being 22 (1987) (fig. 1.3).
40 Phenomena of Aging and Old Age

.392 and .416 in 2050.'* Such futuristic demographic forecasting is unreli¬

able; and the very term “dependency ratio” is misleading because it as¬
sumes that all persons 65 and older are retired and therefore in a sense
dependent—though possibly only on their own accumulated wealth—and
that all persons 20 through 64 are working. Both assumptions are false.
Defined functionally, as the ratio of nonworkers to workers, the depen¬
dency ratio has fallen—and steeply—since the mid-1960s, mainly because
of the greatly increased participation of women in the labor force. The ratio
is expected to rise after about 2005, but not to return to the level of the mid-
1960s.'® Even redefined, the old-age dependency ratio is misleading be¬
cause “dependency” on one’s own previous earnings, reflecting a personal
decision to reallocate consumption from earlier to later years of life, is a
different kettle of fish, ethically and economically, from dependency on the
current generation of workers.
This is not to deny, of course, that the dependency ratio is influenced
by patterns of retirement. Voluntary retirement was much less common in
the United States before social security than it has become since.In 1840,
70 percent of white American males over 65 were employed, and presum¬
ably most of the others were physically incapable of working. Thirty years
later, 64 percent of American men aged 60 or over were employed, and
sixty years after that the percentage was actually a shade higher. With the
enactment of the Social Security Act in 1935, the labor-force participation
rate of elderly Americans started to plunge, and by 1980 it had fallen to
32 percent.^' In 1950, as shown in table 2.3,“ 83 percent of American men
aged 60 to 64 and a remarkable 21 percent of those 75 and older were
employed, compared to 90 percent of men 55 to 59. By 1990, when

18. Samuel H. Preston, “Demographic Change in the United States, 1970-2050,” in De¬
mography and Retirement: The Twenty-First Century 19, 23 (Anna M. Rappaport and Sylvester J.
Schieber, eds., 1993).
19. Stephen H. Sandell, “Prospects for Older Workers: The Demographic and Economic
Context,” in The Problem Isn't Age: Work and Older Americans 3, 5 (Stephen H. Sandell, ed.,
1987) (fig. 1.1).
20. Gordon F. Streib, “Discussion,” in Issues in Contemporary Retirement 27 (Rita
Ricardo-Campbell and Edward P. Lazear, eds., 1988). But it was more common than generally
believed. See chapter 9.
21. The sources of these statistics are William Graebner, A History of Retirement: The
Meaning and Function of an American Institution, 1885-1978 12 (1980); Roger L. Ransom and
Richard Sutch, “The Decline of Retirement in the Years before Social Security: U.S. Retirement
Patterns, 1870-1940,” in Issues in Contemporary Retirement, note 20 above, at 1, 13 (fig. 1.6);
Carole Haber and Brian Gratton, Old Age and the Search for Security: An American Social History
104-110(1994).
22. The source of the data in table 2.3 is Gendell and Siegel, note 13 above, at 25 (tab. 2).
The figures for the 1990s are projections—and may not be accurate, as we shall see.
Old Age Past, Present, and Future 41

Table 2.3 Labor-Force Participation Rates for Persons Aged 45 to 49 through 75 Years or
Older, by Sex, for Selected Years 1950-2000

45-49 50-54 55-59 60-64 65-69 70-74 75 and over

Men
1950 96.5 95.0 89.9 83.4 63.9 43.2 21.3
1960 96.9 94.7 91.6 81.1 46.8 31.6 17.5
1970 95.3 93.0 89.5 75.0 41.6 25.2 12.0
1980 93.2 89.2 81.7 60.8 28.5 17.9 8.8
1990 92.3 88.8 79.8 55.5 26.0 15.4 7.1
2000 91.8 89.0 79.2 54.2 27.3 15.6 7.3
Women
1950 39.9 35.7 29.7 23.8 15.5 7.9 3.2
1960 50.7 48.7 42.2 31.4 17.6 9.5 4.4
1970 55.0 53.8 49.0 36.1 17.3 9.1 3.4
1980 62.1 57.8 48.5 33.2 15.1 7.5 2.5
1990 74.8 66.9 55.3 35.5 17.0 8.2 2.7
2000 82.7 74.8 61.9 39.5 19.7 8.5 2.7

80 percent of men 55 to 59 were employed, the employment rate of the 60-

64-year-olds had fallen to 56 percent and that of the 75-and-over group to
7 percent. Of course, these declines were not because elderly people were
becoming ever less capable of working, but because they could afford to
retire voluntarily.
Among women, the declining retirement age has been offset by a
growing rate of participation in the labor force,^^ except among women 75
and older—but they represent a cohort with a history of very limited labor-
force participation. We shall see that when the sexes are combined, postre¬
tirement employment factored in, and comparison made with the recent
rather than remote past, it appears that the labor-force participation rate of
older people is actually growing rather than continuing to decline.
Even a continued decline in the labor-force participation of the elderly
would have relatively little social or economic significance, and perhaps no
significance for government policy, if average annual income in retirement
were falling as the length of retirement increased. The old would be consum¬
ing no more in the aggregate, but merely spreading their consumption over
a longer period. Retirement incomes and spending have, however, been ris-

23. See also Amanda Bennett, “More and More Women Are Staying on the Job Later in
Life Than Men,” Wall Street Journal, Sept. 1, 1994, p.Bl.
42 Phenomena of Aging and Old Age

ing, not falling, relative to incomes from work. Between 1957 and 1990 the
median incomes of persons aged 65 or older more than doubled, greatly out¬
pacing the growth in the median income of the population as a whole —and
this without regard to the value of Medicare. With these and other adjust¬
ments, it appears that the elderly are at least as well off as the nonelderly
and quite possibly better olf^^—and that they may have caught up with the
nonelderly as early as 1973.^^ “Few of the elderly are millionaires, but most
are wealthy—and becoming wealthier—at least compared with younger
households.” By 1991, only 12.4 percent of persons 65 or older were below
the poverty line, compared to 14.2 percent of all ages.^* And all this is with¬
out imputing any value to leisure, which retired people have more of than
working people, or (the same point) without imputing any cost to work.
These statistics are somewhat misleading, however. Income and lon¬
gevity are positively correlated, which means that some people survive to
old age because they are prosperous, rather than being prosperous because
society is generous to elderly people. Nevertheless, there is little doubt that
elderly people have improved their relative economic position, which
means that they would account for a growing fraction of total consumption
even if people were not living longer or retiring earlier. But they are, imply¬
ing a possibly dramatic shift in the relative income shares of productive and
nonproductive adults, especially when medical expenditures are counted as
a form of consumption. This shift, viewed in economic rather than purely
financial terms (for economics is concerned with the allocation of resources
rather than with balance sheets and income statements as such), is only
partly offset by the increase in the number of working women. When
women join the labor force, they substitute market for nonmarket work.
Nonmarket work has value. The net addition to the social product is there-

24. Taeuber, note 13 above, at 4-7 and n. 104. More dramatic still, between 1970 and 1984
the median real income of elderly (that is, over 65) households rose by 35 percent, compared to
only 1 percent for households of persons aged 25 to 64. Alan J. Auerbach and Laurence J. Kotli-
koff, “The Impact of the Demographic Transition on Capital Formation,” in Demography and
Retirement: The Twenty-First Century, note 18 above, at 163, 174.
25. Michael D. Hurd, “Research on the Elderly: Economic Status, Retirement, and Con¬
sumption and Saving,” 28 Journal of Economic Literature 565, 576-578 (1990); John R. Wolfe,
The Coming Health Crisis: Who Will Pay for Care for the Aged in the Twenty-First Century? 10
(1993).
26. See Sheldon Danziger et al., “Income Transfers and the Economic Status of the El¬
derly,” in Economic Transfers in the United States 239, 264 (Marilyn Moon, ed., 1984).
27. John C. Weicher, “Wealth and Poverty among the Elderly,” in The Care of Tomorrow’s
Elderly 11, 24 (Marion Ein Lewin and Sean Sullivan, eds., 1989); see also Pamela B. Hitschler,
“Spending by Older Consumers: 1980 and 1990 Compared,” 116 Monthly Labor Review. May
1993, p. 3.
28. Social Security Bulletin: Annual Statistics Supplement 1993 148 (tab. 3.E2).
Old Age Past, Present, and Future 43

fore less than the increase in market income, although the latter increase is
the relevant one if we are considering the impact of retirement on public
finance, which is the usual focus of discussions of the dependency ratio.

Restoring Perspective
Before becoming too alarmed by statistical indicators of incipient, if not
galloping, gerontification, the reader should ponder the following points:
1. The “real” dependency ratio, as I have already pointed out, de¬
pends not only on the fraction of the population that is old, but also on the
fraction of children and of nonworking adults, since these classes are de¬
pendent too. The growth in the number of working women has reduced the
number of nonworking adults and, through a negative effect on the birth
rate (the opportunity costs of having children are higher for women who
have good career opportunities in the job market), the number of children.
Properly measured, the dependency ratio in the United States is actually
falling. It is expected to rise again in the next century, but we should not be
too troubled by that prospect. As a nation uniquely attractive to immi¬
grants—most of whom are young adults—and still relatively uncrowded,
the United States could by liberalizing its immigration laws lower the de¬
pendency ratio pretty much at will. Even when that possibility is put to one
side (for many readers of this book will not agree that the United States is
“still relatively uncrowded”), the financial implications of the continued
aging of the population are less ominous than popularly believed. Expen¬
ditures on the old will grow, but not dramatically relative to increases in
per capita income, and will be offset in part by falling education expendi¬
tures as a result of the decline in the number of children.^®
Now it is true that this rosy forecast depends on the unrealistic assump¬
tion that health-care costs will not increase in real (that is, inflation-
adjusted) terms. But it is a legitimate assumption to employ in studying the
effect of changes in the age distribution of the population on the economy,
because the interactive effect of rising health costs with an increasing frac¬
tion of elderly people (whose per capita health costs are above average) is
relatively small. Health costs have been rising so much faster than the up¬
ward shift in the age distribution of the population that this shift is only a
minor factor in the so-called “crisis” in costs of health care.^® It may be no
factor at all. As is well known and regularly deplored, medical expenditures

29. Michael D. Hurd, “Comment,” in Studies in the Economics of Aging 33 (David A. Wise,
ed., 1994).
30. Id. at 36-37.
44 Phenomena of Aging and Old Age

are heavily concentrated in the last stages of life (the year, month, and week
before death) (see chapter 5). Suppose that all a person’s medical costs
were incurred in the last year of life—an exaggeration, of course, but an
illuminating one. Then an increase in the average life span would reduce
the average cost of medical care, by spreading the total cost per person
(assumed to be constant) over more years.^'
Even if the rosy forecast proves false, this should be known far enough
in advance to make it politically feasible to reduce the level of transfers to
the (future) aged, as we shall see in chapter 11. But as we shall also see
there, this is not a complete answer with respect to the costs of health care
and of home care. However much benefits are cut, our society is not going
to allow elderly people to be abandoned. It will defray the costs of their
medical care and basic life needs by hook or by crook; only the identity of
the payors will change as a consequence of cutting the benefits provided by
programs designed specifically for the elderly, such as Medicare and social
security. If the advance of medical science and technology continues to add
years of life without an offsetting reduction in illness and infirmity, the
aggregate cost of medical and home care for the elderly will rise even if the
unit costs of these services relative to the unit costs of other goods and
services do not.
2. Average retirement ages are misleading proxies for cessation of em¬
ployment, because many workers—in fact an estimated 25 percent—take
up other jobs, either part time or full time (mostly the former), after retiring
from their career jobs.^^ A falling average age of retirement is therefore
consistent with a stable or even increasing rate of labor participation by
older persons. In fact, contrary to the implications of table 2.3, the labor-
force participation rate of persons 65 and older has increased, though only
slightly, since 1985 (figure 2.7).” The term “retirement” is ill specified.
3. Old people are much less likely to commit crimes, a major source
of both private and social costs in this country, than young people. The old
do have higher automobile accident rates than the young, but this is an
artifact of how such rates are computed. As we shall see in chapter 6, the
actual contribution of elderly drivers to accidental injuries is small and the
private value of their driving is large.

31. See Wolfe, note 25 above, at 27.

32. Daniel A. Myers, “Work after Cessation of Career Job,” 46 Journal of Gerontology S93,
S100 (1991); Dean W. Morse, Anna B. Dutka, and Susan H. Gray, Life after Early Retirement:
The Experiences of Lower-Level Workers, ch. 3 (1983). See also Erdman B. Palmore et al.. Retire¬
ment: Causes and Consequences, ch. 7 (1985).
33. The data in figure 2.7 are from various issues of Employment and Earnings, published
by the Bureau of Labor Statistics of the Department of Labor.
Old Age Past, Present, and Future 45

Figure 2.7 Labor-force participation for U.S. population age 65 and over, by sex

4. Household production, and other forms of nonmarket production

such as volunteer services, are an important source of a nation’s wealth
when “wealth” is considered in real and not merely pecuniary terms. Re¬
tirement causes people to increase their nonmarket production, because the
opportunity cost of that production—the income they would be earning in
market production were they not retired—is now lower. The increase in
nonmarket production (excluding leisure, discussed next) does not come
close to fully offsetting the decline in their production for the market, but
there is a partial offset.^"
5. Even the increased “production” of leisure (or, equivalently, con¬
sumption of leisure activities) by old people is, to an economist, a source
of genuine wealth, at least for those old people who could get jobs in the

34. A. Regula Herzog et at, “Age Differences in Productive Activities,” 44 Journal of Ger¬
ontology S\29 (1989); Phillip B. Levine and Olivia S. Mitchell, “Expected Changes in the Work¬
force and Implications for Labor Markets,” in Demography and Retirement: The Twenty-First
Century, note 18 above, at 73, 77-78. Martha S. Hill, “Patterns of Time Use,” in Time, Goods,
and Well-Being 133, 151-153 (F. Thomas luster and Frank P. Stafford, eds., 1985) (fig. 7.5),
documents the shift in the allocation of time by the elderly from market production to personal
care (sleeping, washing, and so forth), leisure, household production, and organizational activity
(political, religious, charitable, and so forth).
46 Phenomena of Aging and Old Age

market but who derive greater utility from leisure activity, or inactivity, in
retirement. Indeed, for people who obtain a net utility from continuing to
live,^^ whether because they enjoy life or dread death, increased longevity
is a source of utility, and so in an economic sense of welfare or wealth, even
if the increased longevity must be paid for in higher costs of medical care
and in a lower production of market goods and services.
6. Leisure, moreover, is usually regarded as a superior good, in the
sense that more of it is bought as people’s incomes rise; at least this seems
a plausible explanation for the long secular decline in hours of work.^^ A
shift in time from working to leisure is a symptom not that the value of a
nation’s output is declining, but that it is growing.
7. Increases in nonworking life expectancy must be distinguished
from increases in disabled life expectancy. One can imagine a system of
health care that increased the life expectancy of retired people without
making them healthier, with the result that the percentage of retired people
who were immobilized, debilitated, demented, demoralized, or in pain
would rise because they were living longer. But the same wave of medical
technology (preventive as well as curative) that has increased adult life ex¬
pectancy has reduced the prevalence of disability and hence dependency
among the old, for example through better treatment of hip fractures, de¬
generative joints, osteoporosis, circulatory diseases, cataracts and other vi¬
sual disorders, and diabetes.^’ In addition, the average education and in¬
come of old people have been rising—and both education and income are
positively correlated with health in old age.^®

35. Which is not, as one might suppose, every living person, because of the cost of suicide—
of which more in chapter 10.
36. The reason for the hedge (“at least this seems”) is that an increase in the wage rate
increases the opportunity cost of leisure, and this substitution effect may dominate the income
effect. For evidence that the income effect dominates, producing a backward-bending supply
curve of labor as a function of wages, see, for example, B. K. Atrostic, “The Demand for Leisure
and Nonpecuniary Job Characteristics,” 72 American Economic Review 428, 435 (1982) (tab. 3);
John D. Owen, “The Demand for Leisure,” 79 Journal of Political Economy 56, 69 (1971).
37. Kenneth G. Manton, Larry S. Corder, and Eric Stallard, “Estimates of Change in
Chronic Disability and Institutional Incidence and Prevalence Rates in the U.S. Elderly Population
from the 1982, 1984, and 1989 National Long Term Care Survey,” 48 Journal of Gerontology
SI53 (1993). This study determined disability by evaluating the subjects’ ability to perform basic
activities of daily living, such as dressing oneself and going to the bathroom, and instrumental ac¬
tivities of daily living, such as preparing meals. Another careful study, however, found only modest
reductions in disability. Eileen M. Crimmins and Dominique G. Ingegneri, “Trends in Health
among the American Population,” in Demography and Retirement: The Twenty-First Century,
note 18 above, at 225, 237-238.
38. See next chapter, text and reference at note 22; also J. Paul Leigh and James F. Fries,
“Education, Gender, and the Compression of Mortality,” 39 International Journal of Aging and
Old Age Past, Present, and Future 47

But we must not push these points too hard. For example, it would be
a mistake to conclude from the positive correlation between income and
health that subsidizing pension income is not a net transfer to the old be¬
cause it is bound to be offset by reduced health-care transfer payments to
them. The subsidization of pensions also increases longevity (because there
is a positive correlation between income and longevity and not just between
income and health), and hence increases the demand for health care. And
medical advances that reduce disability in the short run can increase it in
the long run. Breaking a hip used to be the death warrant for many elderly
people. Now the hip is repaired and the patient lives to an age when he or
she is quite likely to be disabled by something else. As in this example, so
generally, the increase in the number of old people has outpaced the reduc¬
tion in the fraction disabled, so that the number of disabled has risen,
though not as fast as the number of elderly. By 1991, 4.2 percent of the
population 65 and over were in nursing homes (17.5 percent of those 85
and older),^^ while another 15 percent to 30 percent had disabilities serious
enough to require some assistance.^® These percentages will grow. At pres¬
ent, the costs of home care and nursing-home care, which are considerable,
are not being covered by Medicare but are being borne by the elderly them¬
selves and their families,'^' except that the Medicaid program defrays the
cost of nursing-home care for the indigent aged.
8. Yet it remains unclear how large the net transfer of wealth from
young to old in our society is. To the extent that elderly people are altruistic
toward young members of their families and affluent, they will compensate
the young in bequests, gifts, or other forms of transfer payment for the
forced transfers (mainly brought about by social security, including Medi¬
care) from young to old that have so increased the incomes of elderly
households; at the same time, the young, including the altruistic young.

Human Development 233 (1994); Marti G. Parker, Mats Thorslund, and Olle Lundberg, “Physical
Function and Social Class among Swedish Oldest Old,” 49 Journal of Gerontology S196 (1994).
39. A1 Sirrocco, “Nursing Homes and Board and Care Homes: Data from the 1991 National
Health Provider Inventory,” Advance Data No. 244, Feb. 23, 1994, p. 4 (National Center for
Health Statistics) (tabs. 8, 9).
40. The lower figure is computed from U.S. Senate Special Committee on Aging et at,
note 12 above, at 144, and is for the period 1985-1986. The higher figure is based on an estimate
that “about one-third of the population over 65 needs some kind of assistance.” Roxanne Jamshidi
et at, “Aging in America: Limits to Life Span and Elderly Care Options,” 2 Population Research
and Policy Review 169, 173 (1992). If only about 4 percent of the elderly are in nursing homes,
the one-third estimate implies that 29 percent of the elderly population require some assistance
short of institutionalization (29 + 4 = 33).
41. Id. at 173.
48 Phenomena of Aging and Old Age

will reduce their voluntary transfers to the old.'^^ Private decisions often
offset the impact of government programs; that is one reason why so many
government programs are ineffective. Even nonaltruistic persons, more¬
over, are compelled by taxation to defray the cost of educating the young,
and this gives them a moral claim to support by the young when the latter
reach working age and the taxpayers of their parents’ generation who sup¬
ported them reach retirement age.
The widespread belief that the elderly constitute a selfish voting bloc
which is distorting the optimal functioning of democratic government is
exaggerated; subsequent chapters will explain why. Concern that the in¬
crease in the average age of the population will slow down the rate of tech¬
nological progress also is exaggerated, as we shall see. The fact, moreover,
that elderly people are a stabilizing force in politics and culture may be a
net plus from the standpoint of social welfare.
Governmentally compelled transfers from young to old differ from
other forced redistributions in the further respect that while men do not
become women (with the debatable exception of a handful of transsexuals)
or whites become blacks, most young people become old people, so in a
sense they are transferring money to themselves. But not too much weight
should be placed on this point. Some people do not survive to old age. And
people are not indifferent about the allocation of consumption across the
various stages of their life. One’s young and one’s old self can even be
viewed as separate persons (as discussed in chapter 4). The actual transfers,
moreover, are generally from today’s young to today’s old rather than from
today’s young to their own future selves. These transfers may actually re¬
duce the probability of generous transfers to today’s young when they are
old by increasing the federal government’s budget deficit and reducing the
rate of economic growth. And they may engender resentment of, and po¬
litical opposition to, future transfers to the old. The age cohort that is young
today may be a net loser over its life span.
9. Retirement ages may not continue to fall; may in fact rise. The de¬
terminants of the retirement age are complex, as we shall see in subsequent
chapters, but they clearly include the number of younger workers and the
generosity of private pension plans and of social security retirement bene¬
fits. Changes already legislated in the social security program have reduced
the expected value of retirement income,'*^ while the coming decline in the
number of young workers as the baby-boom generation ages will increase

42. See, for example, Gary S. Becker, A Treatise on the Family 275-276 (enlarged ed.
1991), and references cited there.
43. Levine and Mitchell, note 34 above, at 92-94.
Old Age Past, Present, and Future 49

the cost to employers of replacing older with younger workers."^ Since re¬
tirement benefits are reduced when retirement is taken early,retirement
will occur later the longer the worker expects to live, because he will enjoy
the higher retirement income from retiring later for a longer time."*® So in¬
creased life expectancy can be expected (other things being equal) to raise
the average retirement age.
10. To the extent that retirement is correlative with old age, it is im¬
portant to understand that “old age,” like “poverty,” is a relative term.
Different societies have dated the onset of old age at different ages, ranging
from 30 to 70."*^ The earlier dates are mostly found, as one would expect,
in societies in which life expectancy is very short and the means for cor¬
recting the characteristic physical problems of middle age (such as pres¬
byopia) limited or nonexistent. The steady increase in adult life expectancy,
and continued improvements in medicine, can be expected to result in an
upward adjustment in the perceived onset of old age, causing the percent¬
age of the old to fall. We must not suppose that the term “old age” denotes
a natural kind or that 65 or even 80 will forever be thought the onset of it.
This is the most important point of all. Every day it becomes clearer
that millions of Americans, mainly but not only of the upper middle class,
have been shifted by a combination of nutrition, exercise, and medical
technology from the ranks of the old to the ranks of the middle-aged. Forty
years ago, most 60-year-olds and all 70-year-olds were thought, by them¬
selves and others, “old.” Today a great many people retain a reasonable
simulacrum of “youth” (more precisely of middle age) until their late sev¬
enties. People do not like to be old. The shift of millions of people from
old to not-old has been a massive source of utility to the persons shifted
and to their families. Of course, the shift has costs. One is the cost in medi¬
cal care of keeping young. Another is the added burden of elder care on
young and middle-aged adults, for along with a shift from old to not-old
has come a shift from dead to old. Some of those shifted from old to not-
old find themselves “paying” for their good fortune by having to care for
a sick old parent who but for nutrition, exercise, or medical technology
would be dead—not to mention having to pay heavy taxes to support social

44. Id.
45. For example, one can begin drawing social security retirement benefits at age 62, but the
benefits are only 80 percent of what they would be if one waited to 65. A similar pattern (“back-
loading”) characterizes most pension plans. See chapter 12.
46. John R. Wolfe, “Perceived Longevity and Early Retirement,” 65 Review of Economics
and Statistics 544 (1983).
47. Covey, note 3 above. We shall see in chapter 5 that Aristotle fixed the onset of old age
for men at 50.
50 Phenomena of Aging and Old Age

security and Medicare, plus other programs that help older people. The
costs are emotional and psychological, as well as time and taxes. But of
that we hardly need reminders. We are acutely conscious of the costs of the
aging of the American population and tend to overlook the benefits. One
benefit of shifting people from the ranks of the elderly infirm to the middle
aged is to postpone the average date of retirement, which is strongly cor¬
related with health."*®
Although the picture is decidedly a complex one, full of uncertainty,
my research has persuaded me that both the net fiscal and the net full costs,
both current and future, of the aging of the American population have been
exaggerated. If true, this is very important, because efforts to solve the
“problem” of the shifting age distribution of the population are bound to
deflect resources from what may be more serious social problems.
Why there should be widespread exaggeration of the “problem” is
unclear. One possibility is that people have difficulty adjusting their think¬
ing about aging to the fact that if they are to live longer, with a larger
fraction of their life spent in retirement, and if they are to incur substantial
medical costs to achieve this additional longevity and to maintain their
health in their newly extended old age, they will have to sacrifice some of
their standard of living during their working lives, whether by working
longer or harder, saving a larger fraction of their income for their medical
and other consumption needs in retirement, or paying higher taxes.
1 do not counsel complacency about the phenomenon of aging or about
the issues of public policy that it presents. The comforting thought that the
problem of an aging population can be solved if people are brought to re¬
alize that they are going to have to spread their income over more years
may founder, as we shall see, on the fact that the young self asked to set
aside more money for his old age may not fully internalize the welfare of
his old self. But understanding the phenomenon of old age, and the issues
of social policy that it raises, is not assisted by an attitude of dread, born of
popular misconceptions about the social and demographic consequences
of the dramatic upward shift in the age distribution. I am not a Pollyanna.
I dread old age as much as the next person. I suggest merely that it may
be a more serious personal problem than it is a social, an economic, or a
political one.

48. See, for example, Herbert S. Fames and David G. Sommers, “Shunning Retirement:
Work Experience of Men in Their Seventies and Early Eighties,” 49 Journal of Gerontology S117,
SI23 (1994).
 3
A Human-Capital Model of Aging

The Simplest Life-Cycle Model

The basic model that economists have used to advance the understanding
of aging comes from what has grown to be an imposing theoretical and
empirical literature on human capital.' That model will be the starting
point—but only the starting point—for my investigation of the economics
of aging. For those readers who are not familiar with the model, I offer a
brief summary. Then I suggest ways of enriching the model to make it more
serviceable for theorizing about aging and old age. The process of enrich¬
ment is continued in the next chapter, by the end of which we shall have an
economic model flexible enough to handle a variety of psychological and
behavioral phenomena that conventional human-capital economics cannot
explain.
A number of important phenomena, however, including differences in
earnings among individuals and also over the life span of a single indi¬
vidual, can be explained by reference to differences in investment in hu¬
man capital. Economists use the term in the sense, strictly analogous to
physical capital, of an asset.that yields earnings over time rather than im¬
mediately. The earnings can be pecuniary or nonpecuniary, although the
former are usually emphasized. Formal education and on-the-job training
are examples of activities that create human capital.

1. See Gary S. Becker, Human Capital: A Theoretical and Empirical Analysis, with Special
Reference to Education (3d ed. 1993), and for a capsule summary Becker, A Treatise on the Family
26-27 (enlarged ed. 1991).

51
52 Phenomena of Aging and Old Age

Much investment in human capital is indirect. For example, even when

formal education is “free,” the individual pays for it in forgone earnings
during the period when he is in school and therefore not in the job market.
He may pay for on-the-job training, too, by receiving a lower wage than he
would if his employer were not investing in training him. He will not have
to pay for it if it is the kind of on-the-job “training” that consists simply of
experience—of learning by doing.^ Such human capital is acquired without
cost. And even an employee who acquires human capital that is costly to
produce may not have to pay for it. Much of the human capital that on-the-
job training creates, unlike that created by formal education, is usable only
in the particular employer’s employ (“specific human capital”). The em¬
ployee will be reluctant to pay for such training because, if the capital it
creates is specific, he cannot use the threat to quit his current job and take
his human capital elsewhere to extract a higher wage. At the same time, the
employer will be more willing to pay for such training than he would be in
the case of general human capital, precisely because the employee cannot
easily (through quitting or threatening to quit) appropriate the specific hu¬
man capital that the training would create. Employers are highly reluctant
to invest in the creation of general human capital because they have no
assurance of recouping their investment. This is one reason why much of
that capital is created in formal educational institutions rather than in the
workplace.
A rational person, whether employer or employee, would not incur the
costs, direct or indirect, of creating or acquiring human capital unless they
were offset by higher productivity or, in the case of the employee, by higher
earnings. Earnings must therefore contain a component that is implicitly a
repayment (with interest) of the earner’s investment in human capital. The
greater the individual’s investment in human capital, the greater must be
the present value of his anticipated earnings in order for him to recover the
investment with interest commensurate with what he could have obtained
in an alternative form of comparably risky investment. Heavy investing by
a present or future worker in his human capital (as by undergoing pro¬
tracted training in a graduate or professional school) thus creates a steep
age-earnings profile. Earnings are low or even negative during the invest¬
ment period and therefore have to be high during the recoupment period in
order to make the investment worthwhile.

2. See Kenneth J. Arrow, “The Economic Implications of Learning by Doing,” 29 Review

of Economic Studies 155 (1962); Becker, Human Capital, note 1 above, at 67-68. He will not
have to pay much, at any rate; to the extent that he works longer hours in order to obtain valuable
experience, he gives up leisure, and the value of that leisure is a cost of obtaining the experience.
A Human-Capital Model of Aging 53

Like physical capital, human capital depreciates. This is due not only
to a literal wearing out, as by memory loss or diminished dexterity—fac¬
tors not emphasized in most human-capital analysis—but also to a chang¬
ing work environment, which reduces the value of particular knowledge or
skills. It might seem that if people did not die, age, or retire, they would
replace their human capital as it depreciated, and so the portion of their
earnings that represented the recovery of their investment in human capital
would be unchanged. Not quite. If their income rose with age because of
their greater experience, so would the cost of any new investment in human
capital, insofar as the making of such an investment required them to take
time away from work (as by going back to school); the lost income would
be a cost of the schooling, just like tuition. Even so, if people were eternal
there would be no pronounced age peaks in earnings. Instead, earnings
would level off at the point at which the last dollar’s worth of investment in
human capital produced a dollar of additional earnings. Since people do
die, and before that retire, and have at least a general idea of when they will
retire, they would eventually cease reinvesting in their human capital even
if the cost of investing were invariant to age. If the minimum payback pe¬
riod for some investment in human capital is twenty years, a rational person
will not make the investment if he expects to be working for only ten more
years. So even if people did not age in the sense of becoming less adapted
to working or to learning new skills (but did die, at predictable ages not
much greater than at present), and even if because they did not age and had
a constant income they could invest in human capital at the same cost at
any point in the life cycle, we would not expect an 80-year-old to enroll in
medical school.
If because of the age-related increase in the cost of investing in human
capital or the age-related decrease in the expected return, or both, the rate
of investment in new human capital eventually declines with age below the
rate of depreciation of the worker’s existing human capital, there will be
net depreciation. This will impart a downward thrust to the component of
earnings that represents repayment of human capital and thus to total earn¬
ings. This effect may be masked by economy-wide productivity gains,
which lift earnings generally.^ But that qualification is not germane to my
concerns here and if it is ignored, along with a point that I shall discuss
the backloading of wages for incentive reasons—then the age-earnings

3. Not completely masked, however, because the gains in hourly wage rates tend to be offset
by a decline with age in the number of hours worked. See Gilbert R. Ghez and Gary S. Becker,
The Allocation of Time and Goods over the Life Cycle 85 (1975) (fig. 3.1).
54 Phenomena of Aging and Old Age

Figure 3.1 The age-earnings profile

profile (more precisely, at this stage of the analysis, the time-worked/earn¬

ings profile) will have an inverted U shape, as in figure 3.1.
Earnings (E) are plotted on the vertical axis and years worked (time,
t) on the horizontal. A simple model of their relation is

E(t) = a + b]t — bjt^, (3.1)

where E(t) is annual earnings as a function of time (years worked from first
job to retirement), a is an earnings component that is independent of in¬
vestment in human capital and is assumed to be constant over time, b^
represents an annual increase in earnings brought about by investments in
human capital, and — ^2 represents an annual reduction in earnings caused
by net depreciation of the individual’s stock of human capital. The peak
year of earnings (/*) is found by differentiating E(t) with respect to t and
setting the result equal to zero (satisfaction of the other conditions for a
maximum can be assumed). This procedure yields

implying that an individual reaches his peak year of earnings later the more
his earnings are raised by investments in human capital (^,) and the smaller
the effect of age in reducing his earnings by causing him to invest less in
replacing human capital as it depreciates ib2).‘*
Although for simplicity I have been assuming that it is the individual
himself who pays for the investment in human capital, the analysis is the

4. If, as is common in human-capital models, the dependent variable in equation 3.1 is ex¬
pressed as the natural logarithm of earnings, then b, and &2 become percentages, the first repre¬
senting the rate of return to human capital and the second the depreciation rate of human capital.
This refinement is not necessary for my purposes.
A Human-Capital Model of Aging 55

same if the employer pays. Death or retirement will truncate the em¬
ployer’s return from the investment in his employee, and so the prospect of
the employee’s death or retirement will eventually cause the employer to
stop reinvesting in the employee’s human capital to replace losses from
depreciation.
If age-related disinvestment in human capital caused an actual fall in
earnings, rather than a leveling off or a slowing down in the rate of increase,
a point would be reached where the employee would have little or no mone¬
tary incentive to remain employed; and this would then explain retirement.
As noted earlier, however, secular increases in wage levels, driven by secu¬
lar gains in productivity, counteract the effect on earnings of unreplaced
losses in human capital, so that employees often do not reach their earnings
peak until their last year of employment. (There may be additional reasons
for such a late peak, as we shall see.) If the employee has been saving a part
of his income against such time as he may be unable to work or not want
to work, his anticipated income from savings will be rising as his income
from work is leveling off, and, depending on the cost of and anticipated
income from continued working, he may decide that he is better off retir¬
ing. The cost of continued working may be high because the effects of
aging may compel the worker to exert increasing effort in order to be able
to meet the requirements of the job, and the anticipated income from his
continued working may be low for the same reason as well as because of
diminished incentive to continue investing in his human capital. Notice,
moreover, that while disinvestment in human capital influences the age of
retirement, it is also influenced by it, since the anticipated age of retirement
fixes a horizon beyond which the firm cannot expect any return on its in¬
vestment in its employee’s human capital.
The longevity of Americans can be expected to continue to rise, and
we should consider the probable effects on the formation of human capital.
Two offsetting (not necessarily perfectly offsetting) effects can be conjec¬
tured. One is a rise in the amount of investment in human capital, a rise due
to the longer potential payback period. The other is a fall in the amount of
that investment if the aging of the population increases the demand for
caretaking of elderly people*, because such caretaking is a low-skilled job.^
The second effect assumes that increased longevity will result in an in¬
creased number of helpless old people. This is a plausible assumption, as
we saw in the preceding chapter, and is reason for concern. But there is no

5. David Owen Meltzer, “Mortality Decline, the Demographic Transition, and Economic
Growth” 47, 75-77 (Ph.D. diss.. University of Chicago Dept, of Economics, Dec. 1992).
56 Phenomena of Aging and Old Age

basis for some of the most alarmist predictions about the consequences of
future increases in longevity. An example is Leon Kass’s prediction that if
average life expectancy were extended by 10 or 20 years there would be
disastrous social consequences even if the additional years involved no ag¬
ing.^ Kass believes that either the cost of supporting the lengthier retire¬
ments implied by this extension would be crushing, or, if retirement ages
were raised in order to alleviate that cost, the young would be frustrated by
the absence of promotional opportunities since old people would not be
making way for them. These concerns make no sense on Kass’s assumption
that the additional years of life would be added to the individual’s prime.
Retirement would be later because productivity would decline more slowly
with age. The resulting decline in the dependency ratio (see chapter 2)
would increase national output, creating new jobs for the young. Midlife
career changes would become more common because the return to midlife
investments in new human capital would be greater,’ and some of the mid¬
life career switchers would find themselves working for younger people in
their new jobs. Boredom (of which more in the next chapter) would be an
additional inducement to these career changes; and with these changes
there would no longer be a reason to expect retirement ages to be as low as
they are today, and there would be no blockage of the advancement of the
young. It would not even be clear that there were more old people. People
would be living longer, but more of their years would be prime rather than
elderly.
Aging as a change in the individual, a process to which I have begun
to allude, is not a part of the simplest life-cycle model employed in the
human-capital literature. In that model, the only difference among persons
of different ages is their proximity to death. Most economic analyses adopt
this model of aging, abstracting from such things as age-related declines in
flexibility, imagination, strength, or other potentially job-related capabili¬
ties. Gary Becker, for example, refers to such changes as “life-cycle ef¬
fects” that his model does not incorporate.* Jacob Mincer, in another im¬
portant study of human capital, does attempt to measure the pure effect of
aging on earnings. He finds that the effect is much smaller than that of
experience, but notes some negative effect of age after 50.^ He does not

6. Leon R. Kass, Toward a More Natural Science: Biology and Human Affairs 302-305
(1985).
7. Cf. Yoram Weiss, “Learning by Doing and Occupational Specialization,” 3 Journal of
Economic Theory 189 (1971).
8. Becker, Human Capital, note 1 above, at 86-87,92.
9. Jacob Mincer, Schooling, Experience, and Earnings 80 (1974).
A Human-Capital Model of Aging 57

consider the possibility, discussed below, that some of the older worker’s
wage is an incentive payment to discourage shirking (or to reward the
worker for not having shirked in the past) and hence overstates his current
productivity. Neither Becker nor Mincer is interested in elderly or retired
persons, or in occupations such as professional athletics or theoretical
physics in which significant aging effects are felt long before normal
retirement age.
One way to distinguish empirically between aging effects and
proximity-to-death effects would be to compare, with respect to choice of
occupation, investment, education, leisure activities, and other activities,
elderly people on the one hand with young or middle-aged people who
have truncated life expectancies but are in apparent good health, on the
other. For example, a person newly infected with the AIDS virus (HIV) has
roughly the same life expectancy as a 65-year-oldand is unlikely to have,
as yet, significant symptoms. The conventional human-capital model im¬
plies that, after correction for differences in income and for other differ¬
ences between such persons and elderly persons who have the same life
expectancy (a big difference is that the former will not have pension en¬
titlements to fall back upon), the behavior of the two groups will be similar.
It does appear to be similar, so far as investing in human capital is con¬
cerned; the truncation of the payback period causes disinvestment." And
there is a high suicide rate among HIV-infected persons (even before they
have reached the point in the progression of the disease at which they are
classified as persons with AIDS),'^ just as there is, as we shall see in
chapter 6, among elderly persons. In other respects, however, young people
who have curtailed life expectancies similar to those of the elderly but are
as yet relatively asymptomatic do not manifest the characteristic attitudes
and behaviors discussed in this book, such as increased employment of
crystallized relative to fluid intelligence, “inconsiderate” communication,
penny-pinching, an increased propensity to vote, decreased creativity, and
“behavioral slowing.”

10. And likewise a criminal defendant who has just been sentenced to death. The interval
between sentence and execution has been running about 10 years. The average interval between
becoming infected with HIV and the onset of AIDS is also about 10 years, and death usually
follows within two years. See generally Tomas J. Philipson and Richard A. Posner, Private Choices
and Public Health: The AIDS Epidemic in an Economic Perspective (1993).
11. Cf. John G. Bartlett and Ann K. Finkbeiner, The Guide to Living with HIV Infection 17,
264(1991).
12. Cesar A. Alfonso et al., “Seropositivity as a Major Risk Factor for Suicide in the General
Hospital,” 35 Psychomatics 368 (1994); James R. Rundell et al., “Risk Factors for Suicide At¬
tempts in a Human Immunodeficiency Virus Screening Program,” 33 Psychomatics 24 (1992).
58 Phenomena of Aging and Old Age

A more complicated model is necessary to do justice to the problems

specific to aging than one that treats it just as a matter of proximity to death
or retirement. The following sections add some of the essential complica¬
tions. Others are added in the next chapter, where the useful though unreal¬
istic simplifying assumption (maintained in the remainder of this chapter,
as in the human-capital literature on aging generally) that the individual
does not experience physical changes over his life span is finally dropped.

Posthumous Utility and the Last-Period Problem

Contrary to what I have been assuming so far, death need not cut off the
recovery of an investment in human capital. In the usual case, it is true, the
investment generates the funds necessary to recoup it with appropriate in¬
terest by increasing the worker’s productivity, and that productivity plunges
to zero if he dies. But some investments become embodied in forms that
survive death. Suppose an old man writes a book that he does not expect to
be published until after he dies. The time he takes away from other activi¬
ties in order to equip himself to write the book—which might require his
doing research in a field new to him, or assembling notes and diary frag¬
ments if he is planning to write an autobiography—is an investment, and it
may not be recouped until long after his death. Yet even if he cannot capi¬
talize the expected royalties by selling the copyright in the book to a pub¬
lisher for a lump sum-—and how much more good would the lump sum do
him if he is going to die soon?—the investment in writing it may still be a
profitable one from his standpoint. If he is altruistic toward members of his
family, he will derive present utility from the prospective increase in their
utility when they receive royalties after his death. Or if the prospect of
posthumous fame is a source of present utility to him, that prospect may
compensate him for the investment. And—a point combining the altruistic
and selfish dimensions of posthumous utility—the book may be designed
to enhance or even restore the author’s reputation for probity or other vir¬
tues, his “good name.’’ The prospect of improving his reputation may be a
source of direct present utility to the author and also of indirect, altruistic
utility to him because reputation is a family asset as well as an individual
asset.
Although I have cast my discussion of posthumous utility in terms of
investments in human capital, it is relevant to other costly activities under¬
taken near the end of the life cycle. A person may, out of altruistic concern
for family or comrades, or selfish concern for his own reputation, or desire
for posthumous fame or glory, sacrifice his life.
A Human-Capital Model of Aging 59

The general problem of which the effects of death’s imminence on in¬

centives are an example has been discussed in economics under the name
of the “last-period” problem. People keep promises, obey the law, avoid
shirking, and do other good things, economists assume, because the pro¬
spective gains from doing them exceed the prospective losses. But what if
they have no prospects, because they are about to die, or, more realistically,
their prospects are truncated because of the proximity of death or of some
other mode of exit (retirement, switching jobs, switching countries—what¬
ever) from the relevant arena of rewards and punishments? One possible
market response is the backloading of compensation. If an employee has
an entitlement to a generous pension that he will lose if he is found guilty
of malfeasance, he will have an incentive to behave himself even if he is in
the last period of his employment and criminal punishment for his malfea¬
sance is not a realistic prospect.'^ And likewise if he is paid currently more
than his marginal product, so that if he is fired he will not be able to get
another job that would pay as much.’'* He may be paid very well simply
because the firm has invested heavily in his specific human capital, which
by definition he cannot carry away with him if he goes to another job. His
marginal product will be high because he has so much human capital, and
his employer can therefore afford to pay him well—and has an inducement
to do so in order to prevent him from quitting, which would wipe out the
firm’s investment in his human capital.
An individual’s concern, altruistic or selfish, with the posthumous con¬
sequences of current behavior is another possible solution to the last-period
problem. The concern may be so intense that there is no last period. That is
the case for people who are convinced that there is an afterlife in which
good people are rewarded and bad ones punished. The phenomenon of
the atheist’s death-bed conversion is thus readily understandable: afterlife
benefits and costs are now so imminent that however radically they are
discounted whether for futurity or uncertainty they are likely to dominate
benefits and costs in this life, which are severely truncated by imminent

13. Gary S. Becker and George J. Stigler, “Law Enforcement, Malfeasance, and Compen¬
sation of Enforcers,” 3 Journal of Legal Studies 1,6-13 (1974). The pension is funded by the
worker’s accepting a lower wage. In effect, he posts a bond, which he pays for by accepting the
reduction in wage and which he forfeits if he misbehaves and is fired.
14. The drop in wage would thus be the penalty for his malfeasance. Edward P. Lazear,
“Why Is There Mandatory Retirement?” 87 Journal of Political Economy 1261 (1979); Lazear,
“Agency, Earnings Profiles, Productivity, and Hours Restrictions,” 71 American Economic Review
606 (1981). Empirical support for the Becker-Stigler and Lazear “bonding wage models” is found
in Laurence J. Kotlikoff, “The Relationship of Productivity to Age,” in Issues in Contemporary
Retirement 100 (Rita Ricardo-Campbell and Edward P. Lazear, eds., 1988).
60 Phenomena of Aging and Old Age

death. (This point is related to Pascal’s famous wager.) Believers in an af¬

terlife who sacrifice current consumption to improve their afterlife pros¬
pects in effect reallocate consumption from the present to a posthumous
future. The pyramids of the Pharaohs attest to people’s willingness to make
very large such reallocations.
The phenomenon is not limited to religious believers. One does not
have to believe in an afterlife to be concerned with one’s fame or reputation
after death, or with the welfare of one’s family after one’s death, or just
with leaving behind some worthwhile trace of one’s presence in the world.'^
Indeed, the nonbeliever may be more concerned with these things than the
believer, for whom worldly success, including fame as well as money, and
even happiness, may seem trivial goods.
This analysis suggest an economic reason (everyone knows the emo¬
tional reason) why many people postpone making a will until death is im¬
minent: until then the expected benefits to the intended recipients of a be¬
quest may be sufficiently remote to have little weight in the intending
donor’s utility function. The analysis also helps show why it is superficial
to urge heavy taxation of bequests on the ground that the recipients of be¬
quests have not earned them. At least if the bequest was intended, rather
than being an accident of the donor’s having died before he managed to
spend all his money, the tax, to the extent anticipated by him, is on him as
well as on the heirs because by reducing his posthumous or vicarious con¬
sumption it reduces his present utility.
The last-period problem and the suggested solutions for it have two
implications for the timing of retirement and for the balance between con¬
sumption and savings in retirement. First, the more altruistic a person is,
the later he will retire, since by working longer he can accumulate more
wealth to transfer to his family either during his life or, through bequests,
at his death.Second, when he does retire, he will save more and consume
less from his retirement income, the more altruistic he is. Since most altru¬
ism is familial, we would expect that as between two otherwise similar
retired people, the one with more children would save a larger percentage
of his retirement income. Efforts to test this hypothesis have had some suc¬
cess but not enough to command a consensus among economists.'''

15. For an interesting discussion, see Ryan J. Hulbert and Willy Lens, “Time and Self-
Identity in Later Life,” 27 International Journal of Aging and Human Development 293 (1988).
16. The choice between inter vivos (life) gifts and bequests (death gifts) will depend on
whether donor or donee is the more efficient saver, which is likely to be influenced by tax
considerations.
17. Michael D, Hurd, “Research on the Elderly: Economic Status, Retirement, and Con¬
sumption and Saving,” 28 Journal of Economic Literature 565, 617-629 (1990).
A Human-Capital Model of Aging 61

The fact that people die with any bequeathable wealth at all, rather than
holding all their wealth in the form of annuities, might seem to provide
decisive evidence of the existence and importance of a bequest motive, and
hence of altruism in old people. But there are purely selfish reasons for
wanting to have liquid wealth, for example to meet extraordinary demands
or to be able to hedge against inflation. Although in principle an annuity
can be indexed against inflation and made borrowable against for emer¬
gency needs, this requires complex contracts which a person planning for
his retirement may not fully understand or trust. A further complication is
that annuitization of part of the retiree’s wealth is not, as it might appear to
be, inconsistent with his wanting to leave a bequest. Because of uncertainty
as to when one’s benefactor will die, heirs have difficulty allocating con¬
sumption over their life cycle. This uncertainty can be eliminated by the
testator’s buying an annuity that will give him whatever level of income he
desires for his life with no residue at death, while giving the rest of his
wealth to his heirs.'*
The last-period problem may seem particularly acute in cases in which
a person works until he dies, as is true with respect to many American
judges. Yet we shall see in chapter 8 that, paradoxically, the last-period
problem disappears when one is speaking of literally the last period.

Relational Human Capital

The concept of human capital is not limited to investment in skills of a
technical, impersonal nature (such as proficiency in operating a forklift or
analyzing a balance sheet) or even skills that are used in market activities.
This is fortunate in terms of the subject of this book because the principal
activities of elderly people are not market activities. Personal relationships,
such as marriage, relations with children and other relatives, and friend¬
ship, require investments in time and effort in order to yield maximum
returns. Courtship and the early days of a friendship are examples of pe¬
riods during which individuals are incurring time and other costs to build
relationship-specific human capital, though often reaping immediate bene¬
fits as well.'^ It might seem to follow that since the returns from invest-

18. Laurence J. Kotlikoff, John B. Shoven, and Avia Spivak, “Annuity Markets, Savings,
and the Capital Stock,” in Issues in Pension Economics 211 (Zvi Bodie, John B. Shoven, and
David A. Wise, eds., 1987).
19. The principal costs often are forgone other relationships: the more time spent in courting
one person, the less time there is for other courtships. Many long courtships end in marriage for
no better reason than that the parties have run out of the time and contacts necessary to explore the
other opportunities afforded by the “marriage market.”
62 Phenomena of Aging and Old Age

ments are truncated for older people, the elderly would marry and form
friendships at a lower rate than young people even if there were no other
pertinent differences between old and young. Old people would then be
more lonely than young, because spouses and friends would be more diffi¬
cult to replace as they died. But this ignores the lower opportunity costs of
time for old people, at least when they are retired, as most are. Since retire¬
ment involves an abrupt change from full-time work to no work, the fall in
the opportunity costs of time for elderly people will be experienced as a
vertical shift from one horizontal cost distribution to another.This shift,
in conjunction with the age-related diminution in the expected returns from
new friendships, should produce a decline in the formation of new friend¬
ships as retirement approaches, a rise right after retirement, and then a fur¬
ther decline until death.
This is shown in figure 3.2. Retirement is assumed to take place at age
70. Before retirement, the benefits (b) and costs (c) (both measured in units
of utility, U) of forming a new friendship, as functions of age (a), cross at
age 60. No new friendships are formed after that until retirement, but for¬
mation continues from then until age 75, when the continued decline (due
to the decline in the number of years over which a return can be enjoyed)
in the benefits from new friendships again crosses the cost curve, which
was lowered by retirement. The benefits are likely to decline especially
steeply when the new friendship is with a contemporary, because the death
of either friend will terminate the friendship. And most friendships in the
United States, of elderly as of other people, are between contemporaries,^'
for especially in a dynamic society members of the same age cohort tend

20. In one study, women who “during most of their lives . . . had been tied to the local
community by their children, husbands, or jobs” were found to have expanded their network of
friends when old age freed them from the time demands of their previous responsibilities.
Rebecca G. Adams, “Patterns of Network Change: A Longitudinal Study of Friendships of Elderly
Women,” 27 Gerontologist 222, 226 (1987). See also Sarah H. Matthews, “Friendships in Old
Age: Biography and Circumstance,” in Later Life: The Social Psychology of Aging 233, 251
(Victor W. Marshall, ed., 1986).
21. Lois M. Tamir, Communication and the Aging Process: Interaction throughout the Life
Cycle 128-129 (1979); Arlie Russell Hochschild, The Unexpected Community 27-30 (1973);
Beth Hess, “Friendship,” in Aging and Society, vol. 3: A Sociology of Age Stratification 357
(Matilda White Riley, ed., 1972); Vivian Wood and Joan F. Robertson, “Friendship and Kinship
Interaction: Differential Effect on the Morale of the Elderly,” 40 Journal of Marriage and the
Family 367, 372 (1978). (This is an example of “age grading,” discussed in chapter 9.) Of course,
the older the sample, the younger the average age of the friends will be (as in Rebecca Gay Adams,
“Friendship and Its Role in the Lives of Elderly Women” 39, 45 [Ph.D. diss.. University of Chi¬
cago Dept, of Sociology, Aug. 1983]), since the upper tail of the age distribution is truncated by
death: if a centenarian has friends, they are unlikely to be other centenarians, because the class is
so small.
A Human-Capital Model of Aging 63

Figure 3.2 Formation of new friendships in old age

to have more shared values and experiences than members of different

cohorts.
Loneliness can be expected to accelerate in extreme old age. The loss
of friends will be increasing rapidly at the same time that the net utility
from forming new friendships is declining. The latter decline might con¬
ceivably be tempered, even reversed, by the fact that, since new friendships
are a substitute for old ones, the demand for the former should increase as
the supply of the latter decreases because of death. But an offsetting factor
would be the declining supply of new friends in the aging person’s own age
cohort, from which most new friends would be expected to come.
Empirical testing of these hypotheses is complicated by the fact that
people who survive to old age are not a random sample of their age cohort.
They are apt to be healthier than average, of course, but also more intelli¬
gent, better educated, and more affluent, since health, income, education,
and intelligence are all positively correlated.^^ Hence an 80-year-old and a
70-year-old may differ less from each other than the difference in their ages
might seem to imply. If, for example, more intelligent people are apt to be
less gregarious than less intelligent ones, the average very old person may
be less lonely in old age than the average younger old person, even if the
former has fewer friends, simply because intelligent people, who do not
have as many friends, on a\^erage live longer.^^
I said earlier that friendships between members of different age cohorts

22. See, for example, Isaac Ehrlich and Hiroyuki Chuma, “A Model of the Demand for
Longevity and the Value of Life Extension,” 98 Journal of Political Economy 761, 774-775
(1990).
23. Cf. Ethel Shanas, “The Psychology of Health,” in Ethel Shanas et al.. Old People in
Three Industrial Societies 49, 67 (1968).
 64 Phenomena of Aging and Old Age

tend to be rare. This might seem surprising. An old person will have
learned a lot from living through the different stages of the life cycle, and
his accumulation of knowledge and insight would, one might think, consti¬
tute a valuable source of useful information to a young person. The old
person has low opportunity costs of time and therefore plenty of time for
friendship with the young. Moreover—we might call this the “reverse last-
period” phenomenon—the old person is likely to be a disinterested (not
uninterested) friend. He has little to gain from refusing to share his wisdom
with the young; keeping it to himself can do him little good now that he
is no longer actively transacting. The infrequency of friendship between
members of different age cohorts thus suggests that the wisdom of the old
probably has little value to the young. I believe that this is true, but that it
is true not because the cumulative life experience of the elderly is not genu¬
ine wisdom but rather because it is not effectively communicable. To the
extent that the experience of an earlier generation can be as it were codified,
it can be communicated to the young through books or other impersonal
means. To the extent that it is uncodifiable, it often cannot be communi¬
cated at all, no matter how intimate the relationship; it depends on lived,
not read-about, experience. I return to this important distinction in later
chapters, especially chapter 7.
The concept of relational human capital that I have been discussing is
similar to the concept of “social [human] capital” introduced by the soci¬
ologist James Coleman.^^ Although Coleman stresses the role of such capi¬
tal in education and I have been discussing it in regard to one’s circle of
friends, it is also important in the workplace. Salesmen, lawyers, account¬
ants, editors at publishing houses, politicians, lobbyists, investment coun¬
selors, travel agents, and a host of other purveyors of services build up
networks of personal contacts during their careers. These networks are
sources of information, referrals, recommendations, and, when the rela¬
tions among the members of the network are ones of trust, repeat business.
Thus, like other forms of human capital, these networks, and the “good
reputation” that they both create and signify (for it is difficult to build or
maintain a network if one acquires a bad reputation as someone to transact
with), enable the individual to increase his earnings.^^ It might seem that.

24. James S. Coleman, “Social Capital in the Creation of Human Capital,” 94 American
Journal of Sociology S95 (1988). Similar points can be found in the economic literature on the
costs of transactions. See, for example, Yoram Ben-Porath, “The F-Connection: Families, Friends,
and Firms and the Organization of Exchange,” 6 Population and Development Review 1 4-12
(1980).
25. See Curtis 1000, Inc. v. Suess, 24 F.3d 941, 947 (7th Cir. 1994), for an illustration.
A Human-Capital Model of Aging 65

unlike the case of other forms of human capital, there would be little or no
depreciation with age of this “network” capital. Indeed, one might expect
it to grow automatically, continuously, monotonically, and largely cost¬
lessly, like experience, and yet be less likely than experience to generate
boredom because of diminishing returns (and because learning by doing is
in a sense a product of repetition) or to degenerate into mindless habit. In
fact, depreciation will be rapid when the members of one’s network begin
to retire. Even before then, normal job turnover will cause members of
one’s network to drop out, and replacing them may entail travel and other
effortful activity that becomes more costly with age. The benefit of rebuild¬
ing one’s network will decline, moreover, as one approaches retirement
even if the members of one’s network are not retiring, since the payback
period from networking as from other investments in human capital will be
truncated. Nevertheless, we should expect a slower age-related decline
from peak performance in activities in which network human capital is
important than in activities in which it is unimportant.
To summarize, although even the simplest human-capital model of be¬
havior over the life cycle has great explanatory and predictive power—and
is a powerful antidote to alarmist predictions about the consequences of
continued increases in longevity—it requires supplementation in order to
yield the maximum of insight into the behavior of elderly persons. Such
phenomena as striving to complete a book even though death is imminent,
deathbed conversion, bequest behavior, the postponing of the making of
one’s will until death is imminent, and the formation of new friendships in
old age are not illuminated by the simple human-capital model. But they
can be explained in economic terms once the model is enriched by bringing
in the possibility of posthumous utility and disutility, the last-period and
“reverse last-period” concepts, and the concept of relational human capi¬
tal. I have also emphasized, as I shall be doing throughout the book, the
importance of selection or retention bias in deflecting us from a true under¬
standing of aging. The elderly are not a random draw from their age cohort.
 4
An Economic Model of Aging
with Change Assumed

I come now to the most important respects in which the standard life-cycle
model used by economists must be modified for the study of aging and old
age. For thus far I have said only a little about aging as a process in which
a person changes. In the standard model, as I have emphasized, aging is
implicitly nothing more than movement toward a fixed horizon. This is an
important aspect of aging but leaves out the fact that people change as they
approach the horizon; their location on the time line changes, but not only
their location. Recognition that human capital depreciates and must there¬
fore be renewed if it is not to decline may seem to imply something akin
to the wear and tear that machinery and other forms of physical capital
undergo. If so, this would mean, contradicting my description of the stan¬
dard model, that the standard model does assume age-related change in the
individual worker. But, as I noted briefly in chapter 3, depreciation need
not imply change in the depreciating good. Just as a product may have to
be written off purely because of a change in consumer tastes or in the price
of a substitute, so the skills and know-how employed in doing a job may
obsolesce purely because of changes in the equipment with which the
worker works or in other complements to labor, unrelated to any change in
the worker’s capabilities. For example, since law changes as old statutes
are repealed and new ones enacted, and as old cases are overruled or super¬
seded by new ones, a law professor, to maintain his productivity as a
teacher, must—at any age—invest some time in relearning his fields.'

. On the depreciation of legal-knowledge capital, see William M. Landes and Richard A.

66
A Model with Change Assumed 67

And even if there is, as there usually is, human wear and tear—say,
forgetfulness or “rustiness” about rarely used but important procedures
(for example in an emergency)—requiring periodic retraining, as with air¬
line pilots, this “rusting” phenomenon need not be age-related. If a 25-
year-old and a 60-year-old require the identical refresher courses every six
months to keep up some vital but seldom-used job skill, their investment in
human capital is no more age-related than the requirement of an oil change
every six months is a function of the aging of an automobile. And rising
income may make the cost of investing in human capital rise with age,
while the approach to retirement will reduce the payback period for any
such investment and hence the incentive to invest for long-term improve¬
ment or maintenance of productivity (not my six-month example). These
things, too, are independent of any age-related decline. Yet we saw in the
first chapter that it is not just myth that people undergo inexorable and
cumulatively momentous changes, both somatic and nonsomatic, as they
age. These changes are so important that room must be found for them in
an economic model of employment and retirement.

The Knowledge Shift, Discount Rates, and the

Age-Decline Curve
Two kinds of thinking, and their relation to aging. Among the non¬
somatic changes associated with aging, the balance between memory and
imagination, or between experience and analysis, or between retrospect
and prospect, between thinking back and thinking forward, as sources of
knowledge2 for coping with the vocational and other challenges of life,
shifts with age in favor of the former term in each pair. The shift is neutral
in the sense that whether it enhances or reduces performance depends on
the character of the activity in which the individual is engaged. If the rela¬
tion of experience to imagination were merely additive, the accrual of ex¬
perience with age could only enhance performance unless, as I have yet to
consider, age brings about a deterioration in imaginative power. I am as¬
suming, with Aristotle, thaUhe relation is one of substitution, that experi¬
ence displaces imagination. This implies that imaginative power does in

Posner, “Legal Precedent: A Theoretical and Empirical Analysis,” 19 Journal of Law and Eco-
nomics 2A9 {\916).
2. I am thinking of “knowledge” mainly in the sense of knowing how to do things, rather
than in the sense of abstract knowledge. But the relevant know-how includes knowledge of how
to do abstract things, like solving complex equations.
 68 Phenomena of Aging and Old Age

fact deteriorate with age, although Aristotle himself, as we shall see in the
next chapter, apparently did not notice this implication.
In the somatic category, consisting of the complex of changes both
physical and mental that result from bodily changes that are correlated with
age, we can classify under the heading of “decline” any change that re¬
duces a person’s capability, at a given level of effort, of engaging in produc¬
tive (not necessarily market) activities. As with the knowledge shift, some
somatic changes may enhance rather than reduce capability in some activi¬
ties, for example by boiling away some of the hormone-driven emotional
intensities of young people that can interfere with the performance of cer¬
tain tasks.
An important interaction between somatic change and the knowledge
shift concerns the distinction between fluid and crystallized intelligence.
As we glimpsed in the Introduction, the first relates to problem-solving
abilities, the second to one’s basic, ingrained knowledge base—one’s com¬
petences in such things as language skills (including reading and writing),
face recognition, autobiographical history, maxims of conduct (“common
sense”), and spatial orientation.^ There is a parallel between fluid intelli¬
gence and imagination, on the one hand, and between crystallized intel¬
ligence and experience, on the other. I do not mean “imagination” in the
sense of creativity, but in the more mundane sense of making a model or
abstract representation of a problem so that it can be solved by a theoretical
procedure. Imagination in its simplest arithmetical form is illustrated by
determining the number of persons in a movie theater at a specified time
by subtracting the number of persons who have left the theater since it
opened for business that morning from the number of persons who have
entered, as distinct from going into the theater and counting the persons
there. When a young person performs a task, he is likely to rely more
heavily, relative to an older person, on his modeling abilities than on les¬
sons learned from experience, as he would be doing if he picked a solution
from a range of solutions that he had used previously in similar situations;'*

3. See, for example, D, B. Bromley, Behavioural Gerontology: Central Issues in the Psy¬
chology of Ageing 194-195 (1990); Eugene A. Lovelace, “Cognitive Aging; A Summary Over¬
view,” in Aging and Cognition: Mental Processes, Self-Awareness and Interventions 407 (Eu¬
gene A. Lovelace, ed., 1990); Donald H. Kausler, “Automaticity of Encoding and Episodic
Memory Processes, in id. at 29; Anderson D. Smith et al., “Age Differences in Memory for
Concrete and Abstract Pictures,” 45 Journal of Gerontology P205 (1990); Paul B. Baltes, Jacqui
Smith, and Ursula M. Staudinger, “Wisdom and Successful Aging,” 39 Nebraska Symposium on
Motivation 123, 128-131 (1992).
4. For evidence of substitutability (obviously not total) of accumulated knowledge for speed
in processing information, see Timothy A. Salthouse, “Speed and Knowledge as Determinants of
A Model with Change Assumed 69

he has less experience than an older person to draw on. If, as psychologists
believe with considerable evidence, fluid intelligence declines with age
much faster than crystallized intelligence does, the effect is to accentuate
what I am calling the knowledge shift. Older people will rely on experience
relative to imagination more than younger people do not only because they
have more experience to draw on but also because they are less adept at
solving problems through the exercise of what I am calling imagination.
If the knowledge accumulated by an older person is transferable to a
younger one at low cost, the combination of the knowledge shift with the
decline (even if quite gradual) of fluid intelligence may decisively favor the
young. Suppose an older person has spent years developing some algorithm
that, once discovered, can be learned in a few hours. Then a young person
may be able to acquire an important part of the older person’s knowledge
base at negligible cost, so that the only difference between the two will be
that the young person is quicker at using the algorithm to solve new prob¬
lems. It does no more good for old people to have knowledge that is avail¬
able to the young at low cost because embodied in books than for them to
have knowledge that they cannot transmit to the young at all because it is
the inarticulable embodiment of lived experience. In either case, the old
cannot appropriate the benefits of their knowledge by “selling” it to the
young.
Young people learn faster than old, in part because they have better
memories; this is an aspect of the age profile of fluid intelligence. The dis¬
parity in rate of learning is particularly marked in language acquisition. So
we would expect (other things being equal, which they often will not be)
that the average age of immigrants to countries in which a different lan¬
guage from that of the immigrant’s homeland is spoken would be lower
than that of immigrants to countries in which the same language is spoken.^
We expect immigrants to be young because if they anticipate a higher in¬
come in the new country than in their homeland they can maximize the
difference by immigrating at the earliest possible time.^ But that is a point

Adult Age Differences in Verbal Tasks,” 48 Journal of Gerontology P29 (1993); Neil Charness
and Elizabeth A. Bosman, “Expertise and Aging: Life in the Lab,” in Aging and Cognition:
Knowledge Organization and Utilization 343 (Thomas M. Hess, ed., 1990).
5. The positive effect on earnings of fluency in the language of the country of destination of
the immigrant appears to be very great. See references in Robert J. LaLonde and Robert H. Topel,
“Economic Impact of International Migration and the Economic Performance of Migrants” 75-
82 (Working Paper No. 96, Center for the Study of the Economy and the State, University of
Chicago, Aug. 1994).
6 Gary S. Becker, Human Capital: A Theoretical and Empirical Analysis, with Special
Reference to Education 87 and n. 32 (3d ed. 1993). I am speaking, as is common in economics, of
 70 Phenomena of Aging and Old Age

independent of the decline caused by aging, which I wish to stress, here the
decline in fluid intelligence.
Young people are generally considered more flexible than old, old
more rigid than young; and the preceding discussion helps explain why.
Abstract reasoning is more flexible than casuistic or other experience-
based reasoning because it is the very nature of abstraction to be applicable
to a wide variety of particulars. So quite apart from better recent memory,
though that is also a property of the young, the costs of learning new things
are lower to young than to old people. There is an argument, broadly Aris¬
totelian, that one can move from particular to particular, using reasoning
by analogy, without the interposition of a syllogism or other abstract
model. If the argument were sound, it might imply that the old were as
flexible thinkers as the young, only using different tools. But the argument
probably is unsound, as we shall see in the next chapter.
Limitations on entry into a line of endeavor, such as the limitations
imposed by mandatory apprenticeships in guilds or professions, are in part
limitations on generational competition. We can begin to see why the cur¬
rent generation might want to impose such limitations on the upcoming
generation.

Subjective time and the discount rate. Some somatic changes, though
neutral, can have important consequences for behavior. Most people report
that as they get older, time seems to go by faster.’ The young are looking
forward and, being generally optimistic (see next chapter), are looking for¬
ward with hope. They are therefore impatient; and we know that a watched
pot does not boil. The old are pessimistic, and therefore are not watching.
A related point is that a person’s life tends, as he ages, to become more

central tendencies rather than of every case. Elderly people often immigrate to be near their chil¬
dren, and some immigrants do not intend to learn the language of their new country but instead to
reside in a community in which everyone speaks their native language.
7. When one is old, “time passes quickly, as if he [the old man] were gathering speed while
coasting downhill. The year from 79 to 80 is like a week when he was a boy.” Quoted in The Art
of Growing Older: Writers on Living and Aging 50 (Wayne Booth, ed., 1992). See also Simone de
Beauvoir, Old Age 373-376 (1972); Leonard W. Doob, Patterning of Time 234-244 (1971). For
empirical evidence, .see David Licht et al., “Mediators of Estimates of Brief Time Intervals in
Elderly Domiciled Males, 21 International Journal of Aging and Human Development 211
(1985) (finding that as people age, they increasingly underestimate time intervals—an indication
that subjective units of time contract with age); Michael A. Wallach and Leonard R. Green, “On
Age and the Subjective Speed of Time, in Middle Age and Aging: A Reader in Social Psychology
481 (Bernice L. Neugarten, ed., 1968); and studies cited in Johannes J. F. Schroots and James E.
Birren, “Concepts of Time and Aging in Science,” in Handbook of the Psychology of Aging 45,
49 (James E. Birren and K. Warner Schaie, eds., 3d ed. 1990).
A Model with Change Assumed 71

routinized, habit-driven, and freer from novelty than when he was young—
these are all aspects of the knowledge shift—and so there is less to notice,
to arrest attention, and thus to interrupt the flow of subjective time.*
The effect of the acceleration of “time’s wingM chariot” is to bring
the future closer to the present. This suggests that the discount rate (the rate
at which a future value or cost is equated to a present one) declines over
the life cycle. There is both evidence that it does and an evolutionary ex¬
planation.® We know from equation 1.1 that as people age, and their repro¬
ductive potential therefore decreases, they can increase their inclusive fit¬
ness (in effect, maximize their genes in future generations) by measures
that increase the future consumption, and hence reproductive potential, of
their children or more remote descendants at the expense of their own cur¬
rent consumption. Of course, few people (and no animals) consciously en¬
deavor to increase their inclusive fitness. But during the era in which human
beings were evolving rapidly, people who carried genes that promoted in¬
clusive fitness became a larger and larger proportion of the population, and
they bequeathed their genes to their modern descendants. So we can expect
people to become more future-regarding as they age and their reproductive
potential declines. But only up to a point, since as I emphasized in the first
chapter there does not appear to be a genetic program for the behavior of
people past middle age.
There is another reason to expect elderly people to have, on average,
lower discount rates than the young. People who have low discount rates
will invest more in their long-term healthand will therefore be dispro¬
portionately represented among persons who survive into old age. Notice
that this explanation of the lower discount rates of older people, unlike the
first, assumes no age-related change in anyone s discount rate. Discount
rates are assumed constant over the life cycle but persons who happen to
have the higher rates will tend over time to be selected out, so that the
average discount rate will fall. (The importance of selection bias in ex¬
plaining the behavior of elderly people is a recurrent theme of this book.)
The lower the discount rate, other things being equal, the higher the
savings rate will be. People who have a high discount rate want to consume

8. Cf. Joseph E. McGrath and Janice R. Kelly, Time and Human Interaction: Toward a
Social Psychology of Time 15
9. On both points, see the references and discussion in Alan R. Rogers, ‘ Evolution of Time
Preference by Natural Selection,” 84 American Economic Review 460, 477 (1994).
10. For evidence, see Victor R. Fuchs, “Time Preference and Health: An Exploratory
Study,” in Economic Aspects of Health 93 (Victor R. Fuchs, ed., 1982); Isaac Ehrlich and Hiroyuki
Chuma, “A Model of the Demand for Longevity and the Value of Life Extension, 98 Journal of
Political Economy 761, 774 (1990).
72 Phenomena of Aging and Old Age

now, while those with a low discount rate are more concerned with protect¬
ing their future consumption. We might expect discount rates to rise steeply
near the anticipated end of one’s life, as there is then not much future left
to save for. But this is just the last-period problem in another guise. The
stronger the bequest motive, the less likely are the discount rates of very
old people to rise.

Capability and decline. We must take a closer look at decline caused

by aging. I am mainly interested in the effects of normal aging, rather than
of disease. Disease is, however, positively correlated with age. Strenuous,
dirty, or dangerous occupations tend to take a heavy toll of workers’ health
and fitness, inducing (or compelling) retirement at earlier ages than from
light, safe work." We say of certain jobs that they “age” the worker, but it
would be more precise to say that age and the conditions of employment
are cofactors in workers’ health and hence in their occupational longevity.
Decline of visual acuity is a normal concomitant of age even for persons
spared glaucoma, cataracts, macular degeneration, and other common dis¬
eases, which like most diseases are more common in older people. But
under the name of “presbyopia,” it is equally well described as the conse¬
quence of a degenerative eye disease. The healthy old are rarely in as
good physical or mental shape as the healthy young, but with this caveat
it is the declining work-performance capabilities of the healthy old that
interest me.
Assume for the sake of simplicity that a worker’s output is a product
purely of his innate capability and of his effort, neither influenced by ac¬
quired human capital (which I wish to ignore). This assumption will enable
us to focus first on capability viewed as a function of age. This function has
an inverted U shape. An infant has no capability for work, and neither does
a person disabled by extreme old age, which for simplicity of exposition I
shall peg at 80. Capability rises for a number of years but, since I am ab¬
stracting from acquired human capital and focusing on somatic change,
peaks quite early—for many, perhaps most, types of work, probably in the
teens. Assume for the sake of simplicity that the capability required to per¬
form the job is a constant, implying realistically that very young and very

11. See, for example, Martin Neil Baily, “Aging and the Ability to Work: Policy Issues and
Recent Trends,” in Work, Health, and Income among the Elderly 59 (Gary Burtless, ed., 1987);
Gary Burtless, “Occupational Effects on the Health and Work Capacity of Older Men,” in id. at
103; Monroe Berkowitz, “Functioning Ability and Job Performance as Workers Age,” in Berko-
witz et al.. The Older Worker 87 (1988).
A Model with Change Assumed 73

0 20 70 80

Figure 4.1 Actual and required capability as a function of age

old people will alike be incapable of performing it. We can then relate
actual and required capability as in figure 4.1.'2
Capability (c), on the vertical axis, is plotted against age (u), on the
horizontal, describing an inverted U that peaks (m) at age 20, when c = m,
and declines to zero at age 80. The horizontal line labeled r, for required
capability, cuts the curve of actual capability at two points (though so mod¬
est are the abilities required for the imaginary job pictured in the diagram
that even a young child can perform it). We are interested in the segment
of the curve to the right of its peak, the segment I have labeled d (for de¬
cline), which can be represented as

d = m — k(a), (4.1)

where k(a) relates changes in capability to age. Capability falls below the
requirements of the job when r > d, ox equivalently when r — m +
k(a) > 0. This point will be reached sooner the more demanding the re¬
quirements for the job, the more modest the worker’s capability at his peak
age, and the greater the decline in his capability as a function of in¬
creasing age.
This is not to say that the worker will quit or be fired as soon as he
reaches the age at which the requirements of his job just exceed his capa¬
bility for performing it, even if he is paid a wage just equal to the marginal
value of his output—and we saw in chapter 3 that he may be paid much

12. Cf. Joseph J. Spengler, “Introductory Comment: Work Requirements and Work Capac¬
ity,” in Juanita M. Kreps, Lifetime Allocation of Work and Income: Essays in the Economics of
Aging 3, 6-7 (1971).
 74 Phenomena of Aging and Old Age

more by the time he is nearing retirement. Output is a function of effort as

well as of capability. Up to a point, therefore, a decline in capability can be
made up by an increase in effort. But such an increase imposes additional
nonpecuniary cost on the worker, and the cost increases with age, since
beyond some point aging diminishes the capacity for effort, as well as skill.
This point implies that, independent of conventional human-capital consid¬
erations, the aging worker will reach a point at which either his marginal
product is zero, or the cost of work to him in effort exceeds the wage and
other benefits from working relative to the benefits (primarily leisure and
pension) of retirementor of switching to lighter work.'** When I say “out¬
put,” I mean implicitly output per hour or other unit of time, in other words
productivity. So the fact that by working more slowly the aging worker
might be able to produce a year’s output in two years would not show that
he was capable of meeting his employer’s work requirements.
Notice that even though the curve that describes age-related decline
falls gradually (perhaps, up to a point, imperceptibly, if either experience
or relational human capital is important in the particular job), its intersec¬
tion with the curve of required capability creates a discontinuity. To the left
of the intersection, decline is irrelevant; to the right, the worker no longer
has value to the firm. This discontinuity is not the whole story of the rela¬
tion between work and age, but it can help us understand the possible ratio¬
nality of the discontinuity involved in being employed full time one day
and no time the next day and every succeeding day.
Another point to be noted is the difference between m and r—that is,
between peak capability and required capability. As I pointed out in the
first chapter, human beings evolved in a challenging environment. The
work environment of today is for most Americans considerably less chal¬
lenging. A clerk behind the counter of a shop need use only a small fraction
of his endowment of physical and mental capabilities to perform his Job to
the boss’s satisfaction.So even though those capabilities—acute vision

13. Evidence of this is the strong negative effect of age on labor-force participation by the
elderly, after correction for education, health, pension entitlements, and other factors influencing
such participation. Giora Hanoch and Marjorie Honig, “Retirement, Wages, and Labor Supply of
the Elderly,” 1 Journal of Labor Economics 131 (1983).
14. Thomas N. Chirikos and Gilbert Nestel, “Occupational Differences in the Ability of
Men to Delay Retirement,” 26 Journal of Human Resources 1, 23 (1991); Roger L. Ransom and
Richard Sutch, “The Labor of Older Americans: Retirement of Men On and Off the Job, 1870-
1937,” 46 Journal of Economic History 1,19 (1986).
15. See, with respect to memory, Paul Verhaeghen, Alfons Marcoen, and Luc Goossens,
“Facts and Fiction about Memory Aging: A Quantitative Integration of Research Findings,” 48
Journal of Gerontology PI57 (1993), and with respect to sight, William Kosnik et al., “Visual
Changes in Daily Life throughout Adulthood,” 43 Journal of Gerontology P63 (1988). This is not
A Model with Change Assumed 75

and hearing, facility at memorization, computational speed, fleetness of

foot, and the rest—tend to erode over the course of a career,'® and even if
there are no offsetting gains from experience or maturity and no accrual of
valuable social human capital, it may be many years before the clerk’s
ability to do his job declines to a point at which he either cannot do it at all
or cannot do it without a costly (to him) increment of effort. Until that point
is reached, he may be able to compensate for diminution in occupationally
relevant capabilities with small increases in effort.
These observations suggest that aging per se need not reduce occupa¬
tionally relevant capacity, and may thus help explain why most studies do
not find age-related declines in productivity.'^ An additional explanatory
factor, one much stressed in the “optimistic” literature about older work¬
ers, is that they are less likely either to quit or to soldier on the job than
younger workers. There is a good economic reason. Unlike younger work¬
ers, they have built up a lot of firm-specific human capital for which they
are being compensated in their present wage, and it would be wiped out if
they quit. This makes them more vulnerable, and hence more loyal, though
the downside from the employer’s standpoint is that it may also make them
more susceptible to the blandishments of unions. Because of their greater
vulnerability to discharge, whether justifiable or opportunistic, older work-

to deny that there are better and worse clerks, only that the capabilities that decline steeply with
age (such as visual acuity) are not important to the performance of their jobs.
16. See, for example, Sara J. Czaja and Joseph Sharit, “Age Differences in the Performance
of Computer-Based Work,” 8 Psychology and Aging 59 (1993).
17. See, for example, Bruce J. Avolio, David A. Waldman, and Michael A. McDaniel, “Age
and Work Performance in Nonmanagerial Jobs: The Effects of Experience and Occupational
Type,” 33 Academy of Management Journal 407 (1990); Seymour Giniger, Angelo Dispenzieri,
and Joseph Eisenberg, “Age, Experience, and Performance on Speed and Skill Jobs in an Applied
Setting,” 68 Journal of Applied Psychology 469 (1983); Jack Levin and William C. Levin, Age¬
ism: Prejudice and Discrimination against the Elderly 80-81 (1980); William McNaught and
Michael C. Barth, “Are Older Workers ‘Good Buys’?—A Case Study of Days Inns of America,”
Sloan Management Review, Spring 1992, p. 53; and see generally Robert Clark, Juanita Kreps,
and Joseph Spengler, “Economics of Aging: A Survey,” 16 Journal of Economic Literature 919,
927-929 (1978). All such studies, as noted in the text below, suffer from selection bias.
Some studies do find age-related declines in productivity. Two studies of a Eortune 500 firm
find significant declines in productivity in all of the classes of worker studied (male and female
office workers, salesmen, saleswomen, and male managers), beginning in their forties. These stud¬
ies are summarized in Laurence J. Kotlikoff, “The Relationship of Productivity to Age,” in Issues
in Contemporary Retirement 100 (Rita Ricardo-Campbell and Edward P. Lazear, eds., 1988). See
also the earlier studies of factory and clerical workers summarized in Mary Jablonski, Larry Ro-
senblum, and Kent Kunze, “Productivity, Age, and Labor Composition Changes in the U.S.,”
Monthly Labor Review, Sept. 1988, pp. 34-35; Mildred Doering, Susan R. Rhodes, and Michael
Schuster, The Aging Worker: Research and Recommendations 37, 62-63, 86-88 (1983); Czaja
and Sharit, note 16 above. For an excellent review of the literature, see Berkowitz, note 11 above.
 76 Phenomena of Aging and Old Age

ers have more to gain from job security and a rigid seniority system than
younger workers do.‘*
Although the aging worker may be less likely to quit, he is more likely
to retire or to become disabled through illness. And part of his vulnerability
and hence loyalty may come from the fact that because of the decline of his
fluid intelligence it may be very difficult for him to learn a new job—and
this may hamper him in his present employment as well, if the employer’s
job requirements are changing, and may therefore make him all the more
vulnerable. So the loyalty of the older worker is not just a pure plus, but
also an effort to offset a minus. And this point completely to one side, the
studies that find that, on balance, older workers are just as productive as
younger ones have, for two related reasons, a strictly limited significance
for the evaluation of age-related decline. The first reason is selection bias.
Employers will tend to weed out unproductive older workers. The ones who
remain will be productive but their productivity will overstate that of their
age group. Second, since most elderly workers are retired, studies of age-
related declines in productivity tend in fact to compare middle-aged with
young employees, rather than old with middle-aged or young employees,
and it would be question-begging to extrapolate the findings in these stud¬
ies to the old.
The more a job calls for the exercise of physical or mental abilities that
are strongly age-related, such as physical strength and agility, or physical
or mental dexterity, the earlier the age at which the average worker will be
incapable of performing up to his employer’s expectations. So (to antici¬
pate chapter 9) we would not expect old people to command a great deal of
reverence or respect in a society such as ours. Although work is becoming
less dirty, dangerous, and strenuous with the shift in employment from
manufacturing to services, mass education enables the transfer at low cost
of much of the painfully accumulated experience of the old to the young.
And rapid social and technological change causes rapid depreciation of

18. See Richard B. Freeman and James L. Medoff, What Do Unions Do? ch. 8 (1984);
Barry T. Hirsch and John T. Addison, The Economic Analysis of Unions: New Approaches and
Evidence 58, 178 (1986); A. van de Berg and W. Groot, “Union Membership in the Netherlands;
A Cross-Sectional Analysis,” 17 Empirical Economics 537, 552 (1992) (tab. 2). In 1993, only
6.8 percent of American workers 16 to 24 years of age were represented by unions, and only
14.9 percent of those 25 to 34. The corresponding figures for workers 45 to 54 and 55 to 64 are
25.7 percent and 23 percent respectively. (For workers 65 and over the figure falls to 10 percent,
but few of these workers are in unionized occupations.) Bureau of Labor Statistics, Employment
and Earnings, Jan. 4, 1994, p. 248 (tab. 57). When other factors are corrected for, however, the
effect of age on support for union representation is seen to be weak and equivocal, perhaps because
of a tendency of unionism to flatten age-earnings profiles. Hirsch and Addison, above, at 58.
A Model with Change Assumed 77

human capital, which the old may lack the incentive to replace or, given the
decline in fluid intelligence, the ability to replace at reasonable cost. This
leaves unexplained how the old have in recent decades managed to increase
their share of the nation’s wealth relative to the young, but the answer may
lie in interactions between characteristics of the elderly and of the demo¬
cratic process that are examined in later chapters.
The shorter the distance between the worker’s peak capability and the
lesser capability, assumed constant for all ages, actually required for the
job, the earlier (holding the rate of age-related decline constant) the age at
which the worker will no longer be capable of doing the job. Occupations
that draw their members from the upper tail of the distribution of physical
or mental qualities that peak early illustrate this point, ranging from profes¬
sional athletics to particle physics. Physically strenuous occupations other
than professional athletics and a few military specialties do not draw from
the upper tail yet may require a capability close to the worker’s peak, which
implies early retirement too. Most studies, it is true, find that the minimum
capabilities required for performing most jobs in the modern American
economy do not begin to deteriorate measurably until about age 60 (and
often later),'® although productivity may decline earlier. But most jobs are
not all jobs. And the converse of the point in the first chapter about the
tendency of pen-and-pencil tests (“psychometric” tests, in the jargon of
psychology) to measure unneeded capabilities is that they may miss subtle
but occupationally relevant declines in application, focus, flexibility, or
effort.
A further reason to doubt that all or even most workers are coasting
comfortably well above the requirements of their job even if age has eroded
their capabilities to a significant extent is that promotions and demotions
can be and are used to narrow the gap between actual and required capabil¬
ity. Employers use promotions to move abler workers into more demanding
jobs—^jobs with more exacting requirements (r), hence requirements closer
to the worker’s peak-age capability (m). We might expect that as age-
related decline lowered the worker’s m, the process of promotion would be
reversed and the worker demoted to jobs with progressively lower r’s. The
effect in terms of figure 4.1 would be to impart an inverted U shape to r
between its intersections (cutting from below and then from above) with
the curve describing the worker’s changing capability over his life span.

19. See, for example, K. Warner Schaie, “The Seattle Longitudinal Study: A 21-Year Explo¬
ration of Psychometric Intelligence in Adulthood,” in Longitudinal Studies of Adult Psychological
Development M, 127—128 (K. Warner Schaie, ed., 1983).
78 Phenomena of Aging and Old Age

For reasons that are not well understood, at least by economists, and that
precede and appear to exist quite independently of the Age Discrimination
in Employment Act or the minimum-wage law, demotions are rarely used
as a means of adjustment to declining capabilities; they are considered de¬
meaning. The result is to hold r constant after the worker has passed his
peak, thus accelerating retirement. The worker would rather lose some in¬
come by retiring earlier than suffer the humiliation of a demotion.

The Retirement Decision

Assume for the sake of simplicity that the worker must decide whether to
retire at age 65 or at age 70. Economic analysis says retire now (at t = 1)
if the expected utility (t/) of retiring now exceeds that of continuing to
work and retiring later {t > 1). That is, retire now if one’s retirement in¬
come, suitably discounted to present value, will exceed the sum of one’s
working income, net of the costs of work, and of the retirement income to
which one will be entitled if one retires later, with these income figures
suitably discounted too. Consider a highly simplified model. A person con¬
sidering retiring is assumed to know with certainty that he will live to the
age of 80, and his choice is between 5 more years of working followed by
10 of retirement, or 15 years of retirement beginning now. If he postpones
retirement, he has utility from two different streams of benefits. One con¬
sists of an annual pecuniary income for 5 years from work (at wage If) plus
annual nonpecuniary income (during the same period) from work of /„.
lo will be a negative number if the person incurs net disutility (as a result
of danger, fatigue, boredom, the strain of commuting, diminished leisure,
unpleasant superiors, or whatever) from work rather than deriving net posi¬
tive utility from it as a result of the prestige, excitement, socializing, or
other psychic goods that it yields. The other stream of benefits from post¬
poning retirement begins in the sixth year and consists of annual pension
income (some fraction, \/a, of working income) plus nonpecuniary in¬
come from the substitution of leisure for work (/,). Both streams must be
discounted to present value at some interest rate (/) and then compared with
the present value of pension and leisure income if the person retires now.
The worker’s annual pension income will normally be smaller if he retires
sooner rather than later. So /3, the denominator of the fraction of working
income that will be received as pension income when retirement is taken
sooner, will exceed a; Ip/f3 < Ip/a.
Putting these points together and employing the summation sign, X,
since we are adding up annual streams of costs and benefits, we have the
A Model with Change Assumed 79

following formula: retire now only if the present value of retirement now
exceeds the present value of retirement later, or in symbols only if

(I,/P + h)
S V
(C + u ^ y
^ (I, + h) (4.2)
h (1 + i)' f= 1 (I + i)‘ ik (I + i)‘‘
The strictly financial factors in the model which influence the choice
whether to retire early, such as the structure of the pension entitlement and
the ratio of pension to working income, are the subject of a vast literature,
and are touched on in subsequent chapters. Other points need to be consid¬
ered as well, however:
1. The higher the worker’s discount rate, the likelier is retirement at the
earliest opportunity. Even if, as my model assumes, by working another
year one can earn a higher pension, the cost in hassle, fatigue, forgone
leisure, and other costs of work will be incurred in that year, while the
benefit (the higher pension) will be spread out over a more distant future,
one’s remaining life and perhaps that of one’s spouse as well, depending on
the terms of the pension contract. The higher one’s discount rate, the
smaller will be the present value of future benefits. So workers who have
low wages may retire earlier than well-paid workers simply because low
earnings and high discount rates are positively correlated; persons with
high discount rates invest less in their human capital (the returns on which
are postponed) and so have lower earnings.
2. Retirement will tend to be later for people who derive net nonpecu-
niary utility rather than disutility from working (/<, in inequality 4.2).^' The
sources of worry about the elimination of mandatory retirement for profes¬
sors (see chapter 13) are that a large fraction of a professor’s compensation

20. For a good introduction, see Joseph F. Quinn, Richard V. Burkhauser, and Daniel A.
Myers, Passing the Torch: The Influence of Economic Incentives on Work and Retirement 85-87,
199 (1990). On the economics of retirement generally, see the useful summary by Edward R La-

zear, “Retirement from the Labor Force,” in Handbook of Labor Economics, vol. 1, p. 305 (Orley
Ashenfelter and Richard Layard, eds., 1986).
21. For evidence, see Randall K. Filer and Peter A. Petri, “A Job-Characteristics Theory of
Retirement,” 70 Review of Economics and Statistics 123 (1988); Chirikos and Nestel, note 14
above; Chirikos and Nestel, “Occupation, Impaired Health, and the Functional Capacity of Men
to Continue Working,” 11 Research on Aging 174, 192, 197-198 (1989); cf. Herbert S. Parnes
and David G. Summers, “Shunning Retirement: Work Experience of Men in Their Seventies and
Early Eighties,” 49 Journal of Gerontology SI 17, S123 (1994); Martin D. Hanlon, “Age and
Commitment to Work,” 8 Research on Aging 289 (1986). F. Thomas Juster, “Preferences for Work
and Leisure,” in Time, Goods, and Well-Being 333 (F. Thomas Juster and Frank P. Stafford, eds.,
1985), presents convincing evidence that many jobs yield a high level of intrinsic satisfaction. See
also Nan L. Maxwell, “The Retirement Experience: Psychological and Financial Linkages to the
Labor Market,” 66 Social Science Quarterly 22, 30-31 (1985).
80 Phenomena of Aging and Old Age

is nonpecuniary, and some of it would be lost by retirement, and that the

cost of working as a professor is low even at advanced ages. The combined
effect is to make /„ strongly positive. At the opposite extreme, we expect
retirement to be early, other things being equal (often they are not), from
jobs that are dirty, dangerous, unhealthful, strenuous, or stressful. Since
blacks are disproportionately represented in such jobs, they can be ex¬
pected to retire earlier on average than whites do (after other differences
are corrected for) and to be happier in retirement than whites.compli¬
cating factor, however, is that job satisfaction and retirement satisfaction
are positively correlated. The reason is that persons in the managerial and
professional classes are more likely to have social skills, hobbies, and in¬
come that enable them to get the maximum benefit out of the increased
leisure that retirement gives them.^^
The concept of excess capability is pertinent to the issue of the disu¬
tility of work. Consider the simple case in which at all ages to the left of
the intersection of r and d in figure 4.1, the nonpecuniary utility of work
(prestige, sociability, etc.) just equals the disutility (stress, effort, etc.), so
that lo = 0. Consider now what happens when the individual ages beyond
that point. In order not to fall below the minimum requirements of the job,
he will have to increase his effort, and the cost to him of working will rise.
lo will turn negative and retirement will therefore become a more attractive
choice. The general point is that aging will create pressure to retire if, as is
plausible, it reduces productivity in working more than it reduces produc¬
tivity in leisure.
A neglected pair of factors in economic analysis of the cost of work,
though distinct from the also neglected factor of somatic aging, are bore¬
dom and “burnout.” They are not identical. Burnout is a reaction to stress.
There is evidence that it actually is more common among younger than

22. For evidence that blacks do tend to retire earlier than whites, at least when retirement is
functionally conceived to include early retirements due to disability, see Rose C. Gibson, “Recon¬
ceptualizing Retirement for Black Americans,” 27 Gerontologist 691 (1990), and E. Percil Stan¬
ford et al., “Early Retirement and Functional Impairment from a Multi-Ethnic Perspective,” 13
Research on Aging 5, 18 (1991). For evidence that they enjoy retirement more, see Rose C. Gib¬
son, “Aging in Black America: The Effects of an Aging Society,” in Aging in Cross-Cultural
Perspective: Africa and the Americas 105, 119, 121 (Enid Gort, ed., 1988). An additional point,
omitted in my very simple model, is that retirement will tend to be later, the longer the worker
expects to live, because the effect on total retirement income of working longer, and thus qualifying
for higher retirement benefits, will be greater. Blacks have a significantly shorter life expectancy
than whites and therefore have less to lose by earlier retirement that will reduce their annual retire¬
ment benefits.
23. See C. T. Whelan and B. J. Whelan, “The Transition to Retirement” (Economic and
Social Research Institute of Ireland, Paper No. 138, July 1988).
A Model with Change Assumed 81

among older workers.^'* But selection bias may be at work here. Workers
who are especially susceptible to stress will be selected out of stressful jobs
early; those who remain may eventually experience burnout because of the
cumulative character of stress.^^ Boredom in the sense relevant to my
analysis should also be distinguished from the kind of restlessness, also
more common among the young (and intelligent), that comes from doing
“boring” work.^^ My concern is with career boredom, the boredom people
often feel when they are doing the same thing year after year with no pros¬
pect of change. This boredom resembles, though it is not identical to, the
burnout that is caused by the accumulation of stress over a long period
of years.
As long as by gaining experience or otherwise increasing his human
capital in a nonstressful job the worker is doing different things, boredom
or burnout is unlikely even if the worker remains in the “same” job or line
of work for many years. But at some point after the worker has reached a
plateau in his career and his work becomes repetitive, boredom may begin
to grip him; and burnout as well, if it is a stressful job like being a police¬
man, or a teacher in a public school in a rough neighborhood. Boredom and
burnout are sources of nonpecuniary costs of work that although time-
related are not age-related except insofar as the length of time that one has
been in a job is positively correlated with age. The distinction between
time-related and age-related effects will become important when, in sub¬
sequent chapters, we consider the productivity of very old federal judges,
and also the prospects for raising the average retirement age. To anticipate
a bit, because judges in Anglo-American legal systems tend to be appointed
at relatively advanced ages and remain in office well into old age, they
illustrate a class of workers in which the effects of time on the job and the
effects of age should be separable. Boredom implies that within occupa¬
tions, other things being equal, workers who entered the occupation earlier
are likely to retire at a younger age, or, equivalently, that the probability
of retirement at a given age is higher the longer the worker has been in the
occupation.
Boredom illustrates a time-dependent cost in which time and cost are
positively related, while habit illustrates a time-dependent cost and benefit

24. Cynthia L. Cordes and Thomas M. Dougherty, “A Review and an Integration of Re¬
search on Job Burnout,” IS Academy of Management Review 621,633 (1993).
25. For evidence, see George J. Schunk and Harold T. Osterud, “Duration of Pediatric and
Internal Medicine Practice in Oregon,” 83 Pediatrics 428 (1989).
26. See Amos Drory, “Individual Differences in Boredom Proneness and Task Effectiveness
at Work,” 35 Personnel Psychology 141, 146 (1982).
82 Phenomena of Aging and Old Age

in which cost is negatively related to time and benefit positively related to

it.^"^ Not only is it cheaper to brush one’s teeth after brushing has become
habitual, but to stop brushing (maybe in response to convincing evidence
that it was actually bad for one’s teeth) would make one uncomfortable.
Breaking a habit, like breaking an addiction (an extreme example of habit),
causes withdrawal symptoms, though in the case of a mere habit they usu¬
ally are slight and fleeting. Working can be habit-forming. But with the
approach of old age the “habit” becomes a very expensive one, both in
greater effort and (assuming that age and time on the job are highly corre¬
lated) in boredom. Habit-formation is one way in which “learning by do¬
ing” works; tasks are performed more quickly and with less effort when
they become habitual. But the other side of the coin is that habitual activity
can become boring. So not only can learning by doing soon encounter
steeply diminishing returns (the peak of the learning curve may be quickly
reached); the returns can turn negative when habitual work becomes
drudge work.
Another significance of habit for the economics of old age is that the
work routines, methods, and practices of the older worker, having become
habitual, would be difficult for him to change even if there were no age-
related decline in fluid intelligence. Habit thus provides an additional rea¬
son for expecting old workers to be less flexible, less adaptable to changed
circumstances, less likely to learn “new tricks,” than younger ones.
3. The fact that at some point a person is likely to find himself retired
is no more surprising than that infants are not employed. The very young
and the very old have a negative marginal product of work; marginal prod¬
uct and income are generally positively correlated; and the lower a person’s
income from work, the more likely he is to choose leisure or have it chosen
for him. A more interesting question is why so many people go from full¬
time work to retirement overnight, rather than reducing their work continu¬
ously. Not all do. Partial retirement is common.^® An example discussed in
subsequent chapters is the tapered retirement offered federal judges on at¬
tractive terms ( senior status”). Norway and Sweden have government
programs in which workers approaching (Sweden) or just past (Norway)
the normal retirement age can receive a partial pension in return for work-

27. Gary S. Becker, “Habits, Addictions, and Traditions,” 45 Kyklos 327, 336 (1992); Mar¬
cel Boyer, “Rational Demand and Expenditures Pattern under Habit Formation,” 31 Journal of
Economic Theory 27 (1983).
28. Christopher J. Ruhm, “Bridge Jobs and Partial Retirement,” 8 Journal of Labor Eco¬
nomics 482, 490-493 (1990).
A Model with Change Assumed 83

ing fewer hours.^^ Still, “sudden” retirement is very common. The sudden¬
ness of the transition is understandable, however, if there are economies of
scale to leisure^® that offset any diminishing marginal utility of leisure, and
if fringe benefits and other employer labor costs do not fall as fast as the
fall in output from the employee’s switching from full-time to part-time
work. Imagine that each part-time worker required his own office, so that
two part-time workers would cost the employer the expense of two offices
whereas one full-time worker doing the work of the two part-timers would
cost the expense of only one office. Since switching to part-time employ¬
ment would therefore entail a reduction in the wage rate to compensate the
employer for the higher office expense per worker, the employee’s total
income from work, being the product of the hourly wage and the number
of hours worked, would fall faster than the reduction in working time.^' At
the same time, his nonpecuniary income from leisure would soar if he re¬
tired. Retirement would raise both the number of hours of leisure and, be¬
cause of economies of scale in leisure, the utility of each leisure hour.
4. One might think that if most work does not produce nonpecuniary
income comparable to that of leisure, the secular increase in pecuniary in¬
comes must imply a falling average age of retirement because of diminish¬
ing marginal utility of income. Not necessarily. As I pointed out in chap¬
ter 2, higher wages increase the opportunity cost of leisure at the same
time that they reduce the marginal utility of pecuniary income. The first
effect, the substitution effect, favors later retirement; the second effect, the
income effect, favors earlier retirement. The drop in the retirement age in
recent decades suggests that the income effect predominates, and there is
independent evidence for this,” although other factors have been at work
as well, as we shall see in subsequent chapters.

29. Helen Ginsburg, “Flexible and Partial Retirement for Norwegian and Swedish Work¬
ers,” Monthly Labor Review, Oct. 1985, p. 33.
30. As argued in John D. Owen, The Price of Leisure: An Economic Analysis of the Demand
for Leisure Time 72 (1969).
31. Robert L. Clark, Stephan F. Gohmann, and Daniel A. Sumner, “Wages and Hours of
Work of Elderly Men,” Atlantic Economic Journal, December 1984, p. 31, finds that for elderly
men a 10 percent increase in the number of hours worked increases their average hourly wage rate
by only 6 percent.
32. Contrary to popular belief, hourly wage rates have been rising since the 1960s. The trend
has been obscured by the fact that standard methods of computing hourly wages, for example by
dividing a weekly wage by 40, overlook a decline in the number of hours worked. F. Thomas luster
and Frank P. Stafford, “The Allocation of Time: Empirical Findings, Behavioral Models, and
Problems of Measurement,” 29 Journal of Economic Literature 471, 493-494 (1991).
33. Owen, note 30 above, at 121. See also references in chapter 2, note 36.
84 Phenomena of Aging and Old Age

One Self or Multiple Selves?

The concept of multiple selves explained. When age-related changes
in the individual, as distinct from changes in the location of an unchanging
individual on the continuum between birth and death, are brought into the
economic analysis of aging, one of the most elementary assumptions of
conventional economic analysis becomes problematic. This is the assump¬
tion that a person is a single economic decision-maker throughout his life¬
time. The idea that the individual can be modeled as a locus of competing
selves (simultaneous or successive) is not new,^'^ but it remains esoteric and
is disregarded in most economic analysis. For example, economists who
argue against awarding tort damages for nonpecuniary losses caused by
severely disabling personal injuries because the utility of wealth is likely
to be reduced in the disabled state —as shown by the fact that people
generally don’t insure against such losses—are implicitly and uncritically
adopting the standpoint of the pre-injured self, the one who makes the in¬
surance decision. The injured self may want to spend heavily to offset so
far as he is able the effects of the injury even though he has no hope of
achieving the same utility that his pre-injured “predecessor” self enjoyed.^®
The fact that the marginal utility of wealth is lower in the injured
state—the fulcrum of the economic criticism of such damages awards—is

34. Illustrative discussions from different disciplines are Thomas C. Schelling, “Self-
Command in Practice, in Policy, and in a Theory of Rational Choice,” 74 American Economic
Review 1, 6-10 (May 1984 Papers and Proceedings issue); Derek Parfit, Reasons and Persons
(1984), esp. 305-306; Allen E. Buchanan and Dan W. Brock, Deciding for Others: The Ethics of
Surrogate Decision Making, ch. 3 (1989); Elizabeth S. Scott, “Rational Decisionmaking about
Marriage and Divorce,” 76 Virginia Law Review 9, 59-62 (1990); George Ainslie, Picoeconom-
ics: The Strategic Interaction of Successive Motivational States within the Person 29 (1992). The
seminal economic paper is R. H. Strotz, “Myopia and Inconsistency in Dynamic Utility Maximi¬
zation,” 23 Review of Economic Studies 165, 173 (1955-1956). Analytically, the conflict between
successive selves is similar to the conflict between generations with regard to the rate of saving
(the higher the rate of saving, the more the current generation is conferring a benefit on future
generations at the expense of the current one), the subject of a large literature in economics. See,
for example, E. S. Phelps and R. A. Poliak, “On Second-Best National Saving and Game-
Equilibrium Growth,” 35 Review of Economic Studies 185 (1968). Debraj Ray, “Nonpatemalistic
Intergenerational Altruism, 41 Journal of Economic Theory 112 (1987).
35. See, for example, David Friedman, “What Is ‘Fair Compensation’ for Death or Injury?”
2 International Review of Law and Economics 81 (1982); Paul H. Rubin and John E. Calfee,
“Consequences of Damage Awards for Hedonic and Other Nonpecuniary Losses,” 5 Journal of
Forensic Economics 249, 251 (1992).
36. A similar argument is made (independently) in Steven P. Croley and Jon D. Hanson,
“The Nonpecuniary Costs of Accidents: Pain-and-Suffering Damages in Tort Law” 42-44 (un¬
published, Michigan Law School and Harvard Law School, Oct. 28, 1994).
A Model with Change Assumed 85

irrelevant in a Paretian analysis if the occupants of the two states are

deemed two persons rather than one, because one person is being made
better off at the expense of the other?^ But it is worse than irrelevant; it is
wrong. The marginal utility of wealth is lower to the injured self only be¬
cause his utility is being evaluated by the pre-injured self. It is the marginal
utility of the future self to the present self that the present self is calculat¬
ing. It is Idee saying that the value of an education to a child is what the
child’s parent is willing to pay for the child’s education. Unless the present
self is so altruistic toward the future self that it values the future self’s con¬
sumption as highly as its own, it will place a lower value on the utility of
the future self than the latter would. Total utility might therefore be reduced
if the tort system were altered to reduce the amount of tort damages payable
to a severely disabled person, even if the reduction in liability for such
damages brought about no change in the number of disabling accidents.
Liability insurance rates would fall, a result quite likely to be preferred by
the able-bodied self, but at the expense of utility in the disabled state and
hence to the detriment of the contingent disabled self.
Jon Elster has put the basic point well: “The absolute priority of the
present is somewhat like my absolute priority over all other persons: I am
I—while they are all ‘out there.”’Everyone knows that the farther in
advance you invite an academic to attend a conference, the likelier he is to
accept; yet when the date of the conference finally rolls around, he may
bitterly regret his acceptance. Is he being irrational? I think not. Accep¬
tance confers an immediate benefit, that of doing something that another
person (who may be in a position to reciprocate) wants, or, equivalently, of
avoiding disappointing another person. The cost is borne by a future self,
and the farther in advance the invitation is tendered, the more tenuously
connected to our present self will be the self who bears the cost. Granted,
this is not the only possible explanation for “invitation regret.” Others are
that the long lead time is a signal of just how much the invitee’s attendance
is desired, and that it makes a plausible excuse (existing commitments of
one kind or another) more difficult to concoct. Still, the multipleness of
one’s selves is one factor in the decision to accept an invitation that one
knows one will regret later. -
The principal applications of multiple-selves analysis have been to ad¬
diction, weakness of will, regret, self-deception, and, as we shall see in

37. The criticism remains valid in cases in which the tort victim is killed or is rendered
permanently unconscious.
38. Jon Elster, Ulysses and the Sirens: Studies in Rationality and Irrationality 71 (1979).
 86 Phenomena of Aging and Old Age

chapter 10, voluntary euthanasia, rather than to old age.^^ This neglect is
surprising. Aging brings about such large changes in the individual that
there may well come a point at which it is more illuminating to think of
two or more persons “time-sharing” the same identity than of one person
having different preferences, let alone one person having the same prefer¬
ences, over the entire life cycle. The tendency remarked by psychologists
and some economists to give greater weight in making intertemporal
choices to present pains and pleasures than seems rationalis entirely ra¬
tional if the present self is seen as distinct from our future or contingent
selves and naturally inclined to weight its own interests more heavily than
those of these other persons,'*' albeit persons with whom it is linked by
strong bonds of altruism based on continuity of identity. When elderly
people are asked what they would do differently if they could relive their
lives, their most emphatic answer is that they would get more education.'*^
The costs of education (primarily forgone income from working) are con¬
centrated in one’s young years; the benefits are received over many years.
So it is just the area in which one would expect the young self to under¬
spend from the standpoint of the old self.

The normative use of the multiple-self concept attacked and de¬

fended. There are objections to the use of the concept of multiple selves,
whether in the normative (“welfare economics”) or in the positive analysis
of old age. 1 begin with the normative. It can be argued that if young and
old are different selves, so are the 20-year-old self and the 40-year-old self,
or for that matter the 20-year-old self and the 21-year-old self, or the 65-
year-old self and the 66-year-old self. We could end up with as many selves
per person as there are years of life—or months of life, or perhaps hours of
life. The concept of the person, in particular of the responsible person,
would disappear. We would lose all purchase for arguing against becoming
the wards of the state for the sake of our numerous future selves, for whom
we cannot be trusted to make adequate provision.

39. I have found only two discussions, both very brief, of possible applications to aging.
Daniel Wilder, “Ought the Young Make Health Care Decisions for Their Aged Selves?” 13 Jour¬
nal of Medicine and Philosophy 57, 62-63 (1988); Michael Lockwood, “Identity Matters,” in
Medicine and Moral Reasoning 60, 64-65 (K. W. M. Fulford, Grant R. Gillett, and Janet Martin
Soskice, eds., 1994).
40. See, for example, the essays in Choice over Time (George Loewenstein and Jon Elster
eds., 1992).
41. Cf. Ian Steedman and Ulrich Krause, “Goethe’s Faust. Arrow’s Possibility Theorem and
the Individual Decision-Taker,” in The Multiple 5e//197, 204-207 (Jon Elster, ed., 1986).
42. Mary Kay DeGenova, “If You Had Your Life to Live Over Again; What Would You Do
Differently?” 34 International Journal of Aging and Human Development 135 (1992).
A Model with Change Assumed 87

This reductio ad absurdum points to real problems with using the con¬
cept of multiple selves to define the relation between the individual and the
state. But there is a big difference of degree between, on the one hand,
adjacent successive selves, so to speak, and on the other hand the young
and old self separated by many intermediate successive selves, just as there
is a big difference in degree, recognized in countless laws and social prac¬
tices, between one’s self as a child and one’s self as an adult. Once when
my mother was a vigorous woman of 65 or so she noticed a very frail old
woman in a wheelchair and said to my wife, “If I ever become like that,
shoot me.” Two decades later she had become just like that but she did not
express any desire to die; and while by that time she had become moder¬
ately demented she had not yet reached the stage at which it was no longer
possible for an observer to know whether she wanted to live or die. I do not
think that the change in her outlook was just a matter of her having ex¬
changed outside for inside knowledge. Her younger self, had it been
perched on a cloud, looking on, might well have found my mother s con¬
finement to a wheelchair, and failing mental powers, an even worse fate
than it had expected.^^ Aging changed my mother so much that she ac¬
quired a totally different outlook, becoming a stranger to her younger self.
Here is another way to see this point. Suppose you were convinced that
senile people are happy, like children, or that very elderly but not senile
people enjoy the garrulity, penny-pinching, and other characteristic but (to
the nonelderly) negative characteristics of being old, discussed in the next
chapter. In fact, amazing as it may seem to the young and the middle-
aged, elderly people often are happier than nonelderly people, as we shall
also see. Would knowing this make you more or less eager to survive to
old age? I expect it would make you less eager, by amplifying the gulf in
values and preferences between your current self and your contingent fu¬
ture elderly self.
For it is not a lack of information that drives a wedge between the
young and the old self. If it were, then as the number of very old people,
nursing homes, geriatric specialists, and so forth increased, as has been
happening, young people would find the prospect of becoming old less
depressing. They would understand better that most old people really do
want to keep on living and do actually enjoy life as distinct from merely
dreading death. No such change in the outlook of the young is discernible.
Some corroboration for this analysis is furnished by the monotonic
decline with age in the percentage of Americans who believe that it is

43. I complete the story of my mother in chapter 11.

88 Phenomena of Aging and Old Age

(%)

80 j

18-29 30-39 40-49 50-59 60-69 70-79 80-89

Age Group

Figure 4.2 Is it proper to allow incurable patients to die?

proper to allow patients with an incurable disease to die, shown in fig¬

ure 4.2.“^ Young people are much more likely to believe that it is proper,
because they discount the utility of their future contingent diseased self.
Was the preference of my mother’s younger self for not surviving into
an infirm old age more authentic than the opposite preference of her older
self? As a practical matter, of course, the younger self, controlling as it
does the body, can impose many of its preferences on the older self, while
the older self has no control over the younger. But I cannot find anything
in economic theory that tells me whether this practical control should be
elevated into a legal or moral right, entitling us for example to commit
ourselves when young to die when we reach the age of 85 rather than be
required to save for retirement. If my mother had been told what she would
become, she would have been distressed at the prospect, just as I am dis¬
tressed at the prospect, against which I cannot effectively contract (see
chapter 10), of someday becoming senile. I am not saying that total utility
is maximized by giving one’s future selves “rights” against one’s present
self. I am saying that welfare economics does not provide an answer to the
question whether future selves should be considered members of the com-

44. The data used in figure 4.2 are from the General Social Survey (GSS) conducted by the
University of Chicago’s National Opinion Research Center between 1988 and 1993. These and the
other GSS data in this book were furnished to me by the Center through the good offices of Tom
Smith.
A Model with Change Assumed 89

munity whose utility is to be taken into account, along with the utility of
the present self, by the community’s legal and ethical rules.
This is a typical shortcoming of welfare economics. Even if welfare
economics is thought to provide normative guidance in the form of a direc¬
tion to government to try to maximize the total utility or total wealth of
society, it cannot answer the question of what the boundaries of the society
are. Should fetuses be included or not? What about the unborn generally?
And therefore what about a person’s future selves? I know of no other body
of thought that offers a satisfactory answer to the question either. It is con¬
ventional enough to treat society as an aggregation of potential future per¬
sons as well as of those currently living—to suppose for example that we
living Americans have some duty to hand on a habitable planet to our suc¬
cessors—although only that tiny subset of utilitarians that believes in
maximizing total rather than average utility thinks we must weight the
utility of potential future persons equally with that of us the living. But
there one is speaking of future individuals, most of them strangers to us,
rather than of our own successive selves, of whom we might be supposed
adequate trustees—but perhaps not, if old and not old really do have dif¬
ferent values, as in the example of my mother. And while it would be odd
to weight the utility of the unborn equally with that of the living (something
not done even by opponents of abortion),'*^ most people would also think it
odd to give no weight at all to the interests of future persons in making
decisions about public expenditures on education or about the protection
of the environment. If so, this could be thought to imply some duty to our
future selves as well as to future individuals utterly distinct from ourselves;
indeed the former might seem the clearer duty. But what exactly is the
duty? The dilemma is that if the boundary of the community is drawn to
include all contingent selves, as distinct from just current selves, it is too
wide, while if it is drawn to exclude all but current selves, it is too narrow.
The concept of multiple selves will be resisted by those who believe
that a proper human life is one lived in accordance with a rational plan of
life,^^ implying continuous rather than punctuated selfhood.'*^ Yet the inte¬
gration of one’s successive selves, all entitled to respect and consideration.

45. Richard A. Posner, Sex and Reason 280-281 (1992).

46. For the philosophical version, see John Rawls, A Theory of Justice 407-416 (1971).
For the economic version, see Steven M. Goldman, “Consistent Plans,” 47 Review of Economic
Studies 533 (1980), and studies cited there.
47. “Rationality implies an impartial concern for all parts of our life.” Rawls, note 46 above,
at 293. “In the case of the individual, pure time preference is irrational: it means that he is not
viewing all moments as equally parts of one life.” Id. at 295.
90 Phenomena of Aging and Old Age

seems as challenging and worthy a project for reflective human beings as

drawing up and acting in conformity to a life plan for a single self. I grant
that the question who is to do the integrating is a difficult one. The very
question implies the existence of some master self, or even more obscurely
of bargaining among the successive selves, perhaps behind some veil of
ignorance. But the difficulties are merely concealed by ignoring the multi¬
plicity of selves. The content of a rational life plan is a function of the age
at which the plan is deemed to have been adopted. A plan that is impartial
among the stages of life will be in effect the plan that integrates our succes¬
sive selves.
We must be careful not to push the concept of multiple selves to the
point of saying that an old person may not be punished for crimes he com¬
mitted when young because they are different selves. The pragmatic reason
is that such a policy would reduce the effect of the threat of punishment in
deterring crime; and indeed one value of the concept of multiple selves lies
in redirecting analysis from ideological battles over paternalism to prag¬
matic consideration of consequences. Similarly we ought not allow people
to repudiate long-term contracts they made when they were young, for then
most mortgages would be unenforceable, and the young would suffer. But
if, therefore, old selves have duties, maybe they should have rights as well,
which the state should protect, as through social security.
Yet, contrary to the example of social security, not all the implications
of multiple-selves analysis are dirigiste. For example, the analysis high¬
lights the arbitrariness, already noted in chapter 3, of taxing bequests
heavily. A bequest reallocates consumption from one’s present self to a
vicarious future self, that of one’s children or other heirs when one dies,
and is thus no different in principle from saving for one’s old age, a “be¬
quest” by one’s younger to one’s older self. Some people are more altruistic
toward their heirs than toward their own future selves; why should they be
penalized by being taxed more heavily?

The utility of the concept of multiple selves in positive analysis. The

concept of multiple selves has the methodological advantage of enlarging
the domain of rational-choice analysis. This is important because rational
behavior is easier to model, and to make empirically testable predictions
concerning, than irrational behavior. Yet in tension with this point the most
powerful objection to using the concept of multiple selves for positive
analysis is precisely that it adds nothing useful to conventional economic
analysis. In evaluating this objection, I find it helpful to draw an analogy to
the concept of the firm. For many—until relatively recently for most—
A Model with Change Assumed 91

purposes in economics it is quite adequate to treat the firm as if it were a

single individual (a single “self”) rather than as a collection of different
individuals. This would be true for example if we wanted to predict the
effect of an increase in the cigarette excise tax on the price and output of
cigarettes. It would not be true if we wanted to predict the effect of a law
altering the composition of the board of directors of corporations that
manufacture cigarettes or placing a limit on executive salaries. To compli¬
cate analysis by departing from simple albeit unrealistic assumptions re¬
quires justification, however—for example, by showing that the more com¬
plicated analysis yields a richer set of empirically testable implications or
has greater explanatory power. As the example of the firm shows, for
some purposes the more complicated model is necessary. Let us consider
whether the aging person is another example.
We might say, returning to my mother, that all she meant by her dra¬
matic mode of expression was that she rationally forecast a very low, but
not necessarily zero or negative, net utility from life in extreme old age.
This would imply not that she wanted to make a legally enforceable suicide
contract but that she intended to minimize her investment in longevity. She
might also have intended to reduce her level of saving for old age, on the
ground that (assuming no strong bequest motive) the utility of income re¬
ceived by her in extreme old age would be very low.^* But this account is
not complete. When my mother expressed dismay at the sight of the frail
old lady in the wheelchair, she was not making a guess about the state of
mind of that woman (a complete stranger), who in fact evinced no signs of
being unhappy. My mother was evaluating the situation of the woman from
the standpoint of my mother’s current self.
A number of phenomena are illuminated by a model of the individual
as a locus of successive selfish selves: the revulsion that we feel at the idea
of enforcing against an unwilling person a contract to die if some future
condition specified in the contract comes to pass, such as becoming senile
(see chapter 10); the fact that discount rates of young and middle-aged
persons are much higher than necessary to take account of the risk of death,
and, what is closely related, the rather meager provision that most people
make for their future and other contingent selves; the age-related decline in
discount rates, which I mentioned earlier in this chapter; the different atti-

48. An important general point implied by this discussion is that increases in longevity need
not result in increases in savings and may even result in higher discount rates, if because of the
infirmities and disabilities of old age people anticipate a low level of utility from expenditures on
consumption in old age.
92 Phenomena of Aging and Old Age

tudes of old and young toward letting people with incurable diseases die
(see figure 4.2); tort law’s willingness to award substantial damages for the
cutting off of a future elderly life (see chapter 12); even the tendency of
elderly persons to pinch pennies (chapter 6). The first example, that of re¬
fusing to enforce a contract to die, may seem equivocal, as enforcement
would be refused even if the change of mind occurred within hours or days
of the signing of the contract. But all that this shows is that in some settings
we are discouraged from killing even our immediate future selves.
Let me elaborate on the point about the discount (or interest) rate.'*®
Even at a discount rate of only 2 percent, the present value of $1 to be
received in 40 years is only 45 cents. Yet a 30-year-old has a much better
than 45 percent chance of surviving to the age of 70—has in fact almost a
75 percent chance.^® At a discount rate of 4 percent, the present value of $1
to be received in 40 years is only 21 cents. The “real” (that is, setting aside
inflation) risk-free (other than the risk of mortality) interest rate demanded
by the market is probably between 2 and 4 percent.^' This implies that
people are weighting their present consumption far more heavily than their
future consumption (after adjusting for mortality), which is just what one
expects if the present self and the future self are, in some meaningful sense,
separate persons.
An alternative explanation, also consistent with the assumption of ra¬
tionality, for some of these phenomena is that the costs of imagining future
states of the world impede people in obtaining an accurate, vivid picture of
future pleasures (and pains). Rational people, being able to invest “imagi¬
nation capital” to reduce these costs, will, in order to maximize their utility,
devote more resources to enhancing the presentness of pleasurable future
states than that of painful ones.^^ A good example of a phenomenon ex¬
plained by this approach, though not one the authors discuss, is the fact that
people do not want to know the date on which they will die, even though

49. A discount rate and an interest rate are the same thing, the rate at which present and
future costs or benefits are made equivalent. One speaks of a discount rate when one is interested
in the present value of some future receipt or expenditure, and an interest rate when one is inter¬
ested in the future amount into which a present investment or obligation will grow.
50. Computed from Monthly Vital Statistics Report, Sept. 28, 1993, p. 16 (tab. 6).
51. See, for example, L. T. Evans, S. P. Keef, and J. Okunev, “Modelling Real Interest
Rates,” Journal of Banking and Finance 153, 157 (1994); Tong-sheng Sun, “Real and Nominal
Interest Rates: A Discrete-Time Model and Its Continuous-Time Limit,” 5 Review of Financial
Studies 581, 605 (1992); T. Michael Kashner, “Present-Future Gratification Tradeoffs; Does Eco¬
nomics Validate Psychometric Studies?” 11 Journal of Economic Psychology 247, 263 (1990).
52. Gary S. Becker and Casey B. Mulligan, “On the Endogenous Determination of Time
Preference” (unpublished. University of Chicago Dept, of Economics, July 20, 1994).
A Model with Change Assumed 93

the knowledge would be immensely helpful in making a variety of deci¬

sions, such as how much to invest in one’s human capital. Knowing the
date of one’s death would make one’s death more vivid, more easily en¬
visaged, creating, therefore, present disutility. In the imagination-capital
model, just as in the multiple-selves model, “excessive” discounting of
unpleasant future states, such as death or a bleak old age, is consistent with,
even implied by, rationality.
It remains to be seen which approach will prove to be more useful in
the positive analysis of old age (not the focus of the Becker-Mulligan pa¬
per). But at the very least the concept of multiple selves is a useful reminder
of the limitations of expected-utility maximizing as a normative tool. It will
play a role in part 3 of this book with respect to issues ranging from vol¬
untary euthanasia of elderly persons to tort damages for the death of such
persons. The normative use of the concept also has a positive dimension,
as I have already suggested. It helps explain society’s refusal to enforce
every irrevocable commitment that people make and its efforts to discour¬
age certain behaviors, such as drug addiction, that may seriously injure
future selves.
One final qualification: I am not contending that a person’s younger
and older selves are, in fact, different persons. There is no fact of the mat-
ter,53 just as there is no fact of the matter as to whether a firm is a single
entity or a collection of individuals. The concept of multiple selves is a
device for drawing attention to a problem with methods of analysis, eco¬
nomic and other, that assign decisive weight to the “individual’s” prefer¬
ences—for showing, for example, that economics, so often derided for tak¬
ing too “atomistic” a view of human behavior, is sometimes not atomistic
enough. For some purposes, such as evaluating a proposal to allow people
to commit themselves to die upon the coming to pass of a specified future
contingency, it is fruitful to think of the present self and the future self as
two persons. For other purposes, however, such as evaluating a proposal to
allow elderly people to repudiate obligations made when they were young,
it is more fruitful to think of the present and the future self as one person.
This dualism about selfhood is reflected in our language. Most of the time
we treat the self as a unity from birth to death, but not infrequently we
depart from the unitary model and say such things as, “I am not the man I

53. Or so I believe. For a powerful argument (not concerned with aging, however) that “per¬
sons” are (approximately) a natural kind, and not merely a social construct, see David Wiggins,
Sameness and Substance 37 and ch. 6 (1980).
94 Phenomena of Aging and Old Age

was,” or “I am in two minds about the matter,” or “X is a Jekyll and

Hyde,” or “I wasn’t myself when I did that.” When we speak thus we are
not, I contend, “merely” being metaphorical.

This chapter has covered a lot of ground, and a brief summary may be in
order. The emphasis has been on the significance for the human-capital
model of the life cycle of assuming, realistically and indeed unavoidably,
that as people get older they change. The earliest change in adulthood is a
shift in the balance between fluid intelligence (facility at abstract reasoning
and at the acquisition of new skills and capacities) and crystallized intelli¬
gence (concrete reasoning based on one’s established knowledge base) as
the individual accrues experience and his fluid intelligence begins to de¬
cline. As a small example, this shift should lead, other things being equal,
to an even younger average age of immigrants to countries in which the
language is different from that of the immigrant’s native land.
Later—and, as we shall see in subsequent chapters, at very different
ages depending on the field of activity—the individual undergoes a more
general decline in vocationally relevant capabilities, along with neutral but
significant changes such as an acceleration of the rate at which time is felt
to pass. This change, along with the pervasive factor of selection bias, may
explain the empirical evidence that discount rates fall with age. Selection
bias is also the key to explaining—and in fact explaining away—the stud¬
ies that find no age-related decline in workers’ performance. Workers no
longer able to perform to their employers’ expectations because of age are
weeded out and so do not show up in such studies.
The significance of age-related decline for the age of retirement and
for other job-related behaviors depends critically on the distance between
peak age capability and the employer’s required capability and on the rate
of decline from the peak. Because of differences between “job” require¬
ments in the period when human beings became adapted to their environ¬
ment through the operation of natural selection, and job requirements in
the very different environment of today, peak capabilities frequently ex¬
ceed occupationally required capabilities. As a result of this phenomenon
of “excess capability,” a worker may be able to perform to his employer’s
satisfaction long after the worker has passed his peak, although the em¬
ployer may use promotions, and less frequently demotions, to exploit the

54. “Was’t Hamlet wrong’d Laertes? Never Hamlet. / If Hamlet from himself be ta’en
away, / And when he’s not himself does wrong Laertes, / Then Hamlet does it not.” Hamlet, act V
sc. ii, 11. 229-232.
A Model with Change Assumed 95

curve of the workers’ capabilities more completely. Elite occupations are

those in which the employer’s required capability is close to the worker’s
peak, and in these occupations we expect much earlier retirement, de facto
or de jure, than in other fields.
Although up to a point a worker can compensate for age-related de¬
cline by working harder, the added effort is a cost; and as the costs of work
rise, retirement becomes more attractive. The nonpecuniary utilities as well
as disutilities of work influence the age of retirement—so it tends to be
earlier in dirty or dangerous jobs—as do the nonpecuniary utilities and
disutilities of retirement itself, the worker’s discount rate, and the income
and substitution effects of rising wage rates. A neglected age-related source
of nonpecuniary disutility of work is cumulative boredom, or, what is re¬
lated but distinct, burnout due to cumulative stress. Like the acceleration in
the rate at which subjective time passes, boredom and burnout as functions
of time illustrate changes in the individual that while not aspects of decline
can profoundly affect the behavior of aging persons. Finally, economies of
scale in leisure, and the fixed component in employers’ labor costs, help
explain why retirement tends to be abrupt rather than tapered.
The change in the individual between youth and old age is so profound
that it becomes plausible to imagine the individual’s young and old selves
as different persons, “time-sharing” the same body. The concept of mul¬
tiple selves can be used to explain a number of phenomena, such as why
discount rates tend to be much higher than necessary to take account of the
risk of death, implying that the current self weights its utility much more
heavily than that of a future self; why (a related point) many people make
only meager provision for their old age; and why society refuses to enforce
every irrevocable commitment that a person might want to make. The con¬
cept can also be used—though here great caution is necessary—to justify,
nonpaternalistically, the conferral of certain rights on the future self against
the selfishness of the current self.
 'T » •’•

r
r,

y.
'.iS
J£f*i'

^.,' •, _. ■ !!i? •,'. '■'. y /.">ti|^fiSk ■ - i-.’ swci^v

1' ••■'■; ijl) >■- v.n ^5^ .'' v>' *<>< M<lrtkm<Tvj2K'Vv;>ti< ic

• , • .' vajiv • S ' ’««4-3t*'» ■

■ ''r'A^Jt 'Mii:y'yiPr/&. kM . ■. -;.
5^,;,- v^rf-i,'-- i»'»;CF[j3‘''i',-.

.upJ ^ • .'-«•■ ..li^,rt 4iv>* s f.' 1.5i*?k“ jcf: ;:4»-r»p b ^

1 'li'' ■ •'. ie,<S *>*^yj' -lu ■(

'‘S ' ^ V-itU!|^;

,,.«.,c*ws.^^>^*v|.i^_#fk*.i-i4.^,%i>*
'<wr’^f^yrJr. llj: ■'... ;>j<4 • jiit' <. * A-*. .->•> ww.) iv;sf ■
' 1M ' ' - ^ * •

' ^i0 'v»;'; J. rH^' li m ■ .*•'.' > f * ^ r - .'A*jii-Lryia;*'t->’'' -i

-4'>,' ^ ij. ..'% .9iVi . .•i^o8l■’»^^n4:^ lii:•J'*,^£^;'^f'Av4■*''J^3l&Ji>

'ipi‘aL v i^uL£j4n^vC,.^.T./ ’-> '.: * v .:-\ ^
^^.; j..iji;r#;v<t<«''i-J .>»«u .* ■' -.;ll^‘>J«'|P?,5i^^

' j?.a!ii^^' . .; ;'’W1 ;4;i0Cf -■ .’■. V'.^fi.ii'cu'«•..- ,t:^V{Wr- ,

-^•-•vU. . .•H,V4ai>*>.

i» ► oi!'' .f{J : N ;l..'wkv;-<'A'‘l '.‘ ; 1

- r I’iOvI .V—r
.1.
if V .■'' •■ s, . • ' .te
vJ^

’ ■/*•♦*>> ,. i Ky ^ •• 'i,*.,<V'«iJ
/a .’vAr" . --‘v V*-' ■• 'If. ,;>.* .■; i-1 -,•'.
^'' ’ »
■' t -. ,i.'-. r'V^ ••■ r ,v
'^’ n** ^
^ w
• »*> ■ _ i ‘i f *Tf** H:"- . '- 'Vw -*'" '. i ' i*'"'
• * « >
■•’»., '-f • ( !i»-
I ^^■ xf.

M < .V

• \‘-jfLi'j
. "».
•tV"'

-<., .>
... y w A f^x >1
 Part Two

The Economic Theory

Elaborated and Applied
 5
The Economic Psychology of the Old

Economists usually take values, preferences, and attitudes for granted and
consider how, with them as givens, a rational actor, young or old, can maxi¬
mize his utility by the choices that he makes of where to live, what occu¬
pation to follow, whom to marry, and so forth. I take a different tack here,
and consider the extent to which the psychology of the old might fruitfully
be modeled as a consequence rather than foundation of rational choice. I
do not argue that old people always or even often make conscious choices
as to whether, for example, to talk more and listen less than when they were
young. I argue only that certain choices, conscious or unconscious, appear
to be rational in the sense of utility maximizing, once the fundamental at¬
tributes of being old are understood. Many old people, of course, have se¬
rious mental problems which prevent rational choices, and my discussion
excludes such people. Nor do I believe that all aspects of the psychology
of the old can be given satisfactory economic explanations. Genetics, for
example, will play a role in my attempt to explain that psychology.
The psychology of the old has received sustained, but on the whole
rather pitiless, attention from the literary imagination,' although there are
conspicuous exceptions, of Avhich my personal favorites are the depiction

1. For anthologies and reviews of the literature on old age, see The Oxford Book of Aging:
Reflections on the Journey of Life (Thomas R. Cole and Mary G. Winkler, eds., 1994); The Art of
Growing Older: Writers on Living and Aging (Wayne Booth, ed., 1992); Simone de Beauvoir, Old
Age, ch. 3 (1972); David H. Fowler, Lois Josephs Fowler, and Lois Lamdin, “Themes of Old Age
in Preindustrial Western Literature,” in Old Age in Preindustrial Society 19 (Peter D. Stearns,
ed., 1982).

99
 100 The Theory Elaborated and Applied

of the octogenarian Jolyon Forsyte in the first volume of John Galsworthy’s

Forsyte Saga and the depiction of the narrator’s grandmother in The Re¬
membrance of Things Past. On the pitiless side we recall from As You Like
It man’s “second childishness, and mere oblivion, / Sans teeth, sans eyes,
sans taste, sans everything;’’ from Much Ado about Nothing the crack
“When the age is in, the wit is out;’’ and from Hamlet the ridicule heaped
on old age in the person of Polonius. We recall the senescent Lear—a mag¬
nificent ruin, but a ruin nevertheless—and, moving on from Shakespeare,
the Struldbruggs of Gulliver's Travels; Keats’s “few, sad, last grey hairs”
shaken by palsy; T. S. Eliot’s catalog of the gifts reserved to age, such
as the “cold friction of expiring sense” and the “rending pain of re¬
enactment / Of all that you have done, and been;” and Yeats’s fulminations
against old age (“this absurdity . . . this caricature, / Decrepit age that has
been tied to me / As to a dog’s tail”).
The Eliot quotation is from “Little Gidding,” the last of Four Quar¬
tets, a poem sequence that employs aging as a symbol of the temporal
world, contrasted by Eliot, to its disadvantage, with the timeless world. In
“East Coker,” another of the poems in Four Quartets, we read: “Do not let
me hear / Of the wisdom of old men, but rather of their folly, / Their fear
of fear and frenzy, their fear of possession, / Of belonging to another, or to
others, or to God.” The Yeats quotation is from “The Tower,” where he
describes old age as “a kind of battered kettle at the heel” and as “the
wreck of body, / Slow decay of blood, / Testy delirium / Or dull de¬
crepitude, / Or what worse evil come— / The death of friends, or death /
Of every brilliant eye / That made a catch in the breath.” Yeats was 63 in
1928, when he wrote “The Tower.” ^ In another great poem of this period,
“Among School Children,” also written when Yeats was in his early six¬
ties, he called himself “a comfortable kind of old scarecrow,” and asked
rhetorically; “What youthful mother, a shape upon her lap / Honey of gen¬
eration had betrayed . . . Would think her son, did she but see that shape /
With sixty or more winters on its head, / A compensation for the pang of
his birth, / Or the uncertainty of his setting forth?”
Even the defiant cry of Tennyson’s aged Ulysses^—

2. But Eliot s early poem of old age, “Gerontion,” published when he was in his early thir¬
ties, shows that you don’t have to be old to write perceptively about old age
3. Faintly echoing Cicero, De Senectute, ch. X, § 32. Written when Cicero was 84, De
Senectute (the full title is Cato Motor de Senectute) is one of the classics of upbeat writing on old
age. Against Cicero we might set the following laconic summary by a distinguished modern clas¬
sicist. I have sought, but have not found, aspects of old age which compensate for its ills.”
Kenneth Dover, Marginal Comment: A Memoir 243 (1994).
Economic Psychology of the Old 101

Old age hath yet his honour and his toil;

Death closes all: but something ere the end,
Some work of noble note may yet be done,
Not unbecoming men that strove with Gods.

Though much is taken, much abides; and though

We are not now that strength which in old days
Moved earth and heaven, that which we are, we are:
One equal temper of heroic hearts.
Made weak by time and fate, but strong in will
To strive, to seek, to find, and not to yield

—has a certain weary and bleak tone, scant comfort to those of us who
cannot say that we “strove with Gods.” And, as a companion piece to
“Ulysses,” Tennyson wrote “Tithonus”—a poem about the man who, like
the Cumaean Sibyl (Eliot’s and Waugh’s “handful of dust”), had the mis¬
fortune to receive the gift of immortality without an accompanying gift of
eternal youth. In the epigraph of Eliot’s great poem “The Waste Land, the
Sibyl is asked what she wants. She answers, “I want to die.”

Aristotle on Old Age

In the section of the Rhetoric from which I quoted in the Introduction,
Aristotle has a remarkable description of “the character of elderly men—
men who are past their prime.” “ Writing more than two millennia before
the term “political correctness” entered the lexicon, he minces no words.
His discussion is frankly stereotypical; as befits a treatise on rhetoric,
which must point the reader to characteristic features of its subject, features
that the speaker’s audience will recognize, its focus is on central tendencies
rather than on individual variations. It is not the complete truth about the
old, but it is, I think, like the characteristic or if you will stereotypical lit¬
erary depiction of old age, an important part of the truth. It is the part that
tends nowadays to be suppressed because we live in an era of heightened
sensitivity to any suggestion that any group defined by an unalterable char-

4. Aristotle, Rhetoric, bk. 2, ch. 12, in The Complete Works of Aristotle, vol. 2, pp. 2213-
2215 (Jonathan Barnes, ed., 1984) (W. Rhys Roberts, trans.) (the page and column references to
the Greek text are 1389a and 1390a). The Greek view of old age was not uniform, and Aristotle
cannot be deemed its authoritative expositor. Its diversity is well brought out in Bessie Ellen Rich¬
ardson, Old Age among the Ancient Greeks: The Greek Portrayal of Old Age in Literature, Art,
and Inscriptions (1933).
 102 The Theory Elaborated and Applied

acteristic such as race, sex, or age might be inferior to some other group,
even if it is inferior only in its capacity for happiness.
Despite the prestige and enormous current interest in Aristotle, it may
seem strange to use him as my authority on the psychology of aging, rather
than modern psychologists. I shall be citing some modern psychological
literature in this chapter, but I have been surprised by the relative paucity
of studies on the effects of age on personality, character, and the emotions,^
as distinct from the elTects on reasoning ability and other strictly cognitive
capacities. Remarkably, a number of the modern psychological studies of
the effect of aging on personality find no significant effect.® The studies are
not persuasive. They use categories such as “emotional stability,” “infor¬
mation seeking” (in the broadest sense), and “friendliness” that do not
correspond to areas in which age-related change is likely, such as openness
to new ideas, caution, and pessimism.^ Moreover, most of the studies are
based on self-reporting by the participants,^ and so the accuracy of the
studies depends on the participants’ having a high level of self-awareness.
Aristotle’s psychology of aging has not been superseded.
Aristotle tells us that because elderly men have lived a long time “they
have often been taken in, and often made mistakes; and life on the whole is
a bad business.” As a result, “they are sure about nothing ... They ‘think,’
but they never ‘know.’ ” Experience has made them “cynical; that is, they
tend to put the worse construction on everything,” and distrustful. They are

5. For a review of the literature, see Nathan Kogan, “Personality and Aging,” in Handbook
of the Psychology of Aging 330 (James E. Birren and K. Warner Schaie, eds., 3d ed. 1990).
6. Dorothy Field, “Continuity and Change in Personality in Old Age—Evidence from Five
Longitudinal Studies; Introduction to a Special Issue,” 46 Journal of Gerontology P271 (1991).
For illustrative such studies, see Paul T. Costa, Jr., Robert R. McCrae, and David Arenberg, “Re¬
cent Longitudinal Researeh on Personality and Aging,” in Nathan W. Shock et al.. Normal Human
Aging: The Baltimore Longitudinal Study of Aging 171 (1984); Leonard M. Giambra, Cameron J.
Camp, and Alicia Grodsky, “Curiosity and Stimulation Seeking across the Adult Life Span: Cross-
Sectional and 6- and 8-Year Longitudinal Findings,” 7 Psychology and Aging 150 (1992). For a
contrary view, see Kogan, note 5 above, at 336.
7. One relevant category that Costa, McCrae, and Arenberg, note 6 above, do discuss is
“masculinity”—and they find that it declines significantly with age. Id. at 191-192. Unfortunately
the term is not defined; but presumably it encompasses the typical, or perhaps stereotypical, mas¬
culine traits, such as physical courage. (The study was limited to men.) For other evidence that
personality changes with age when personality traits are carefully defined, see Joel Shanan, “Who
and How: Some Unanswered Questions in Adult Development,” 46 Journal of Gerontology P309
P313-P314(1991).
8. One that is not is Dorothy Field and Roger E. Millsap, “Personality in Advanced Old
Age: Continuity or Change?” 46 Journal of Gerontology P299 (1991). This study was based not
on self-reporting but on personal interviews of a sample of aging persons, conducted 14 years
apart by different interviewers, who reported their impressions. The use of different interviewers,
while perhaps unavoidable, undermines the validity of the study.
Economic Psychology of the Old 103

also “small-minded, because they have been humbled by life: their desires
are set upon nothing more exalted or unusual than what will help them to
keep alive.” This focus on keeping alive, together with bitter experience
about how hard it is to get money and how easy it is to lose it, makes the
elderly ungenerous and cowardly, “always anticipating danger.” They are
self-centered, too, guiding their lives too much by what is useful for them
rather than by what is “noble”—by, that is, “what is good absolutely.”
Stated differently, “they guide their lives by reasoning more than by char¬
acter; reasoning being directed to utility and character to excellence.”
“They are not shy, but shameless,” feeling only “contempt for what people
may think of them.” “They lack confidence in the future; partly through
experience—for most things go wrong, or anyhow turn out worse than one
expects; and partly because of their cowardice.” They are loquacious, “con¬
tinually talking of the past, because they enjoy remembering it.” “Their fits
of anger are sudden but feeble.” It is a mistake to suppose them “to have a
self-controlled character; the fact is that their passions have slackened, and
they are slaves to the love of gain.” When they feel pity, they do so “out of
weakness, imagining that anything that befalls anyone else might easily
happen to them . . . Hence they are querulous, and not disposed to jesting
or laughter.” ^
The aim of this chamber of horrors is not, however, to make us side
with youth. Youth has many redeeming features (which age appears not to),
but it is rich in its own foibles. Young people are hot-tempered and fickle,
lack self-control, are preoccupied with honor and victory, are naively opti¬
mistic. “They look at the good side rather than the bad, not having yet
witnessed many instances of wickedness. They trust others readily, because
they have not yet often been cheated. They are sanguine ... [because] they
have as yet met with few disappointments.” Their sanguine disposition
makes them easily cheated, and together with their hot tempers makes them
courageous: “the hot temper prevents fear, and the hopeful disposition cre¬
ates confidence.” And “they have exalted notions.” “They think they know
everything, and are always quite sure about it.”
We can see where this is leading: to a typically Aristotelian celebration
of the mean, that is, of men in their prime:

They have neither that excess of confidence which amounts to

rashness, nor too much timidity, but the right amount of each.
They neither trust everybody nor distrust everybody, but judge

9. Aristotle, note 4 above, at 2214-2215 (1389b-1390a in the Greek edition).

10. Id. at 2213-2214 (1389a-1389b).
 104 The Theory Elaborated and Applied

people correctly. Their lives will be guided not by the sole consid¬
eration either of what is noble or of what is useful, but by both;
neither by parsimony nor by prodigality, but by what is fit and
proper ... They will be brave as well as temperate, and temperate
as well as brave."

Aristotle concludes by observing that “the body is in its prime from thirty
to thirty-five; the mind about forty-nine.”
The essential differences that Aristotle sees between young and old
are, first, that the young are optimistic and the old pessimistic; the knowl¬
edge shift (that is, the changing balance between imagination and memory
as intellectual resources) involves an emotional and not merely a cognitive
change. Second, the old are more self-centered than the young.They are
cowardly, putting their own safety above other goods; greedier than young
people; and “shameless”—they don’t care whether people have a good
opinion of them.
What might explain these differences? (If nothing, we may be led to
wonder whether the description is adequate, or whether Aristotle isn’t just
too preoccupied with finding means between undesirable extremes.) Why
for example would the fact that the young rely more on imagination or
expectation in making judgments and the old more on experience or retro¬
spection make the young optimistic and the old pessimistic?

11. Id. at 2215 (1390a-1390b).

12. Id. at 2215 (1390b). It is not known when Aristotle wrote the Rhetoric, but it has been
surmised that he worked on it between 340 and 335 b.c. (George A. Kennedy. “The Composition
of the Rhetoric," in Aristotle, On Rhetoric: A Theory of Civic Discourse 299, 301 [George A.
Kennedy, trans., 1990])—and he was 49 in 335.
13. For corroboration, see Carol D. Ryff, “Possible Selves in Adulthood and Old Age: A
Tale of Shifting Horizons,” 6 Psychology and Aging 286, 293-294 (1991).
14. A longitudinal study of the vocabulary of a group of high-IQ lawyers (subjects of the
Terman Study of the Gifted) revealed during their seventies a shift in vocabulary from “the dyadic
activities of work and family to a slightly greater concern with soma and self—with one’s own
functioning as an aging individual.” Edwin Shneidman, “The Indian Summer of Life: A Prelimi¬
nary Study of Septuagenarians,” 44 American Psychologist 684, 692 (1989).
15. But we can at least acquit him of the charge of lacking sufficient empathy or inwardness
(not having been an old man when he wrote the Rhetoric) by recalling the elderly Montaigne’s
assessment of old age, written from the inside: “We do not so much give up our vices as change
them, and in my opinion for the worse. Besides a foolish and tottering pride, a tedious garrulity,
prickly and unsociable moods, superstition, and an absurd preoccupation with money after we
have lost the use for it, I find in old age an increase of envy, injustice, and malice. It stamps more
wrinkles on our minds than on our faces, and seldom, or very rarely, does one find souls that do
not acquire, as they age, a sour and musty smell.” “On Repentance,” in Michel de Montaigne,
Essays 235, 250 (J. M. Cohen, trans., 1958). See also Beauvoir, note 1 above, pt. 2 (1972). For a
contrasting view, see Erik H. Erikson, Joan M. Erikson, and Helen Q. Kivnick, Vital Involvement
in Old Age, pt. 2 (1986).
Economic Psychology of the Old 105

The answer may lie in the fact that people are naturally optimistic.
Charles Sanders Peirce argued that

we seem to be so constituted that in the absence of any facts to go

upon we are happy and self-satisfied; so that the effect of experi¬
ence is continually to contract our hopes and aspirations. Yet a
lifetime of the application of this corrective does not usually eradi¬
cate our sanguine disposition. Where hope is unchecked by any
experience, it is likely that our optimism is extravagant. Logicality
in regard to practical matters ... is the most useful quality an
animal can possess, and might, therefore, result from the operation
of natural selection; but outside of these it is probably of more
advantage to the animal to have his mind filled with pleasing and
encouraging visions, independently of their truth; and thus, upon
unpractical subjects, natural selection might occasion a fallacious
tendency of thought.'®
If Peirce is right that a limited tendency to view the world through rose-
tinted glasses has survival characteristics and so is plausibly a part of our
genetic endowment, we would expect the tendency to be blunted by expe¬
rience, since experience would demonstrate that our youthful optimism
was indeed excessive. Experience and age are positively correlated, so that
the effect of experience in grinding down natural but exaggerated optimism
would be to make the old more pessimistic than the young. The tendency
would be furthered by the shifting balance between anticipated gains and
anticipated losses as one ages; the older one is, the less likely is the future
balance to be positive.'^ Pessimism in turn would imply a reluctance to take
risks, financial or otherwise, because the old will have learned that it is silly
to think oneself “lucky.” This implies that in areas in which mistakes of

16. Charles S. Peirce, Essays in the Philosophy of Science 7-8 (Vincent Tomas, ed., 1957).
To similar effect, see Lionel Tiger, Optimism: The Biology of Hope (1979), esp. p. 168; Mar¬
tin E. P. Seligman, Learned Optimism 108 (1991); cf. Lauren B. Alloy and Lyn Y. Abramson,
“Depressive Realism: Four Theoretical Perspectives,” in Cognitive Processes in Depression 233,
256-257 (Lauren B. Alloy, ed., 1988). Seligman points out that the prevalence of depression
suggests that a pessimistic outlook has survival qualities also: there is clear evidence that non-
depressed people distort reality in a Self-serving direction and depressed people tend to see reality
accurately.” Seligman, above, at 111 -115. See also Alloy and Abramson, above; Mark D. Evans
and Steven D. Hollon, “Patterns of Personal and Causal Inference: Implications for the Cognitive
Theory of Depression,” in Cognitive Processes in Depression, above, at 344, 353-356. Neverthe¬
less, far more people are nondepressed-optimistic than depressed-pessimistic. For example,
“8o’percent of American men think they are in the top half of social skills.” Seligman, above,
at 109. . ^ ^
17. Paul B. Bakes, Jacqui Smith, and Ursula M. Staudinger, “Wisdom and Successful Ag¬
ing,” 39 Nebraska Symposium on Motivation 123, 145-147 (1992).
 106 The Theory Elaborated and Applied

optimism impose heavier social costs than mistakes of pessimism, respon¬

sibility should be entrusted to old rather than to young people.
A related point is that we can expect the old to be more “realistic” than
the young,'* more aware of human limitations and the operation of contin¬
gency, and therefore “wiser” in the sense that contrasts wisdom with bril¬
liance.'^ Aristotle remarks elsewhere the widespread belief (which he en¬
dorses) that “a young man of practical wisdom cannot be found. The cause
is that such wisdom is concerned not only with universals but with par¬
ticulars, which become familiar from experience, but a young man has no
experience, for it is length of time that gives experience.” The balance
between abstract and concrete reasoning, or between exact and practical
reason (the latter necessary in incompletely theorized domains of activity),
shifts with age toward the second term in each pair. Wisdom and experi¬
ence are not synonyms; one can be experienced without being wise. But
perhaps one cannot be wise without experience. Experience is necessary
not only to furnish the rich store of particulars needed for practical reason¬
ing but also to wear away the foolish optimism of the inexperienced, the
young. Hence the correlation between age and wisdom.
If wisdom grows with age, does anything decline? Put differently, can
the wise do everything the brilliant young can and then some? The answer
depends on something that is unclear in Aristotle’s account—exactly what
cognitive tools are used in practical as distinct from abstract reasoning.
There is a tradition, which builds on Aristotle’s brief discussion of rea¬
soning by analogy 2' and is particularly strong in law,^^ that practical
reason, especially when it takes the form of reasoning by analogy, enables
the reasoner to move directly from particular to particular, without the in¬
terposition of a syllogism or other device of abstract reasoning. If so,
maybe the weakening of fluid intelligence should not be expected to impair
the ability of elderly people to apply their existing knowledge to new areas.
But I think this is wrong, that an appeal to analogy is usually an appeal to
a general law that the analogy instantiates, and that the analogy cannot be

18. See Ryff, note 13 above, at 292-293.

19. Bakes, Smith, and Staudinger, note 17 above, at 134-139; cf. Maria A. Taranto, “Facets
of Wisdom: A Theoretical Synthesis,” 29 International Journal of Aging and Human Develop¬
ment 1 (1989).
20. Aristotle, Nicomachean Ethics, bk. 6, ch. 8, in The Complete Works of Aristotle, note 4
above, vol. 2, p. 1803 (1142a). But it is men in their prime, rather than old men, whom Aristotle
believes to have the practical wisdom denied the young.
21. See Aristotle, Prior Analytics, bk. 2, ch. 24, in The Complete Works of Aristotle, note 4
above, vol. 1, p. 110 (68b-69a).
22. For references and critique, .see Richard A. Posner, The Problems of Jurisprudence 86-
92(1990).
Economic Psychology of the Old 107

used to justify the law. If, for example, a property lawyer, asked what
should be the legal regime for natural gas, argues that it should be the same
as the regime governing rabbits because both types of resource can move
around “on their own,” he is proposing a general rule applicable to all such
resources, and the rule has to be defended as a sound general rule, which
cannot be done by reference to analogies.^^ So I do not think that reasoning
by analogy can bypass the fluid intelligence, enabling the old to reason as
flexibly and imaginatively as the young. (I present some evidence on this
point in chapter 7.)
I want to come back to the pessimism of the old. Since the old were
once young, their pessimism entails disillusionment, including disillusion¬
ment about schemes for human betterment, for such schemes are usually
founded on hope rather than on experience. The young may have read
about the failure of such schemes but the old have lived the failure; and in
many areas of human activity book learning is not an adequate substitute
for lived experience (see chapter 7). Being pessimistic, disillusioned, and
cynical, the old, however “wise,” become preoccupied with their own sur¬
vival and happiness, these being the only goods of certain goodness to
them. From that obsession can spring avarice and shamelessness.
A puzzle in Aristotle’s account is why people become more and more
pessimistic with age, rather than remaining on a plateau of realism reached
when they are in their prime. There is a possible economic explanation for
at least one component of the pessimistic outlook of elderly people, the
belief that things are getting worse—that they were better in the old days—
that the country is “going to Hell in a handbasket. Over a period of de¬
cades, some aspects of the social environment get worse while others get
better. It is rational that older people should be more conscious of the things
that are getting worse than of the things that are getting better, and vice
versa for young people. Many though not all improvements consist of nov¬
elties, as distinct from incremental improvements in cost or performance.
The elderly, because of the age-related decline in fluid intelligence, have
difficulty—incur large costs—in taking advantage of novelties. Thus, in¬
novations in art, fashion, or styles of living are likely to be accepted much
more readily by young than by old people;^"* the latter may even think the
“innovations” retrograde. At the same time, young people have a less acute
sense of what has been lost on the march to progress than old people do.

23. See id. at 89; also Richard A. Posner, Overcoming Law, ch. 24 (1995).
24 “As consumers, older adults have been shown to be among the last to adopt a product,
service, or idea innovation.” Mary C. Gilly and Valarie A. Zeithaml, “The Elderly Consumer and
Adoption of Technologies,” 12 Journal of Consumer Research 353 (1985), citing studies.
 108 The Theory Elaborated and Applied

The old actually experienced the good things that are no more. The young
can only read about them—and may not bother to do so, having other calls
on their time. In sum, the costs of information about the costs of progress
will be lower to the old than to the young, but the costs of information about
the benefits of progress will be higher, though I admit an exception for
cases in which the young take for granted improvements of which the old
are acutely aware, such as air conditioning and the polio vaccine.
In this analysis, the distance between the location of young and of old
on the continuum that runs between extreme optimism and extreme pessi¬
mism really is increasing in age. But it is doing so for economic reasons,
albeit ones shaped by the physiological process of aging, which affects the
costs and benefits of current and prospective experience relative to retro¬
spective experience.
We can also give an economic twist to the idea that the old are more
self-centered than the young, though we shall have to qualify the idea later.
The key is the last-period concept, discussed in chapter 3. The social vir¬
tues, including fair dealing, trustworthiness, being a good listener, gener¬
osity, and forbearance or self-control, are oriented toward transacting and,
what is closely related, toward the acquisition of new human capital that
will enable the obtaining of even more valuable transactions in the future.
Being “a good listener” illustrates both points. The good listener is polite,
thus reducing the costs to other people of transacting with him, and by
attending carefully to what other people say he increases his own stock of
useful information. In addition, by limiting his own speaking the good lis¬
tener reduces the risk that he will make “revealing” disclosures about him¬
self that may repel potential transaction partners The closer the horizon
of one’s transactional activity is, the fewer will be the benefits from adher¬
ing to virtues that increase the expected value of transacting. We might thus
expect the old to indulge in fewer regrets than the young, after correction
for the fact that the old will have a larger stock of regretted actions. The
utility of regret lies in reducing the likelihood of repeating what has turned
out to be a mistaken action. The old have more experience, but also less to
gain from learning from experience.

The Dread of Death

The most puzzling thing about the suggestion that old people are more self-
centered than the young is the inordinate fear of death that Aristotle as-

25. On the instrumental value of privacy in the sense of secrecy, see the discussion in Over¬
coming Law, note 23 above, ch. 25.
Economic Psychology of the Old 109

cribes to old people. “They love life,” he says, “all the more when their
last day has come.” How can that be, when the old have many fewer years
to lose by dying than the young and when those years may confer lim¬
ited utility at best, because of poor health? The lower utility of the com¬
ing years, moreover, is expected ex ante rather than merely experienced
ex post, and so should affect behavior; for remember that the old are
pessimistic.
The lower discount rates of the old (see chapter 4) can be only a small
part of the answer; likewise the greater excitability of the young. Death is
more imminent for the old, and the probability within a given interval
greater. Yet confront an old and a young person with the same probability
of death, and the old person may be just as fearful as the young even though
he has much less to lose from dying now.
Evolutionary biology may supply part of the explanation. All human
beings of minimal mental competence fear death. This fear, which reli¬
gious, political, and military leaders have devoted endless ingenious
thought to overcoming, is instinctual, programmed. Its contribution to in¬
clusive fitness (the survival of a person’s genes over future generations) is
obvious. A rational creature that has such a fear is more likely to survive
long enough to maximize his reproductive potential than one who lacks it.
It is true that the survival of the very old contributes little to inclusive fit¬
ness because they have little reproductive potential (none in the case of
elderly women). It may actually reduce their inclusive fitness by putting
them in competition with the younger members of their own families for
scarce supplies of food and other resources. But if, as we saw in the Intro¬
duction, selection pressures have not produced a distinctive genetic pro¬
gram for the old, there is no reason to expect them to have lost the instinc¬
tual dread of death just because it has no survival value. (If it has a negative
survival value, the genetic tendency to promote the reproductive fitness of
one’s offspring may come into play and alter the outlook of the old, as we
shall see in chapter 9.) We can thus appreciate the biological sense of Clau¬
dio’s observation that “The weariest and most loathed worldly life / That
age, ache, penury, and imprisonment / Can lay on nature is a paradise / To
what we fear of death.”
There is an economic as well as a biological reason why the old should
dread, or should behave in conformity with the hypothesis that they dread,
death as much as the young. The economic literature on damages in

26. Aristotle, note 4 above, at 2214 (1389b).

27. Measure for Measure, act III, sc. i, 11. 132-136.
 110 The Theory Elaborated and Applied

wrongful-death cases points out that for a person who derives a positive
utility from living and does not have a very powerful bequest motive, no
amount of money will compensate him for giving up his life on the spot.^*
For he will derive little or no utility from the money. By the same token,
again setting aside the case of negative utility of life or of a strong bequest
motive, a person should be willing to expend all his resources, if necessary,
on avoiding an immediate death. Those resources will have zero value to
him if he dies immediately; they have no opportunity cost.- When faced
with the prospect of imminent death, therefore, a person will behave at all
ages as if the value of his life were “infinite” to him, although what is really
going on is that the cost to him of resources expended on avoiding death is
zero. This behavior will actually be encountered more commonly among
old people than among young ones because the old are more exposed to the
risk of imminent death.
If dread of death has survival value, so presumably should happiness,
since the more that people enjoy life the more effort they will expend on
averting death. If “happiness-prone” people can be expected to live longer
on average than the miserable because they are less likely to commit sui¬
cide and more likely to take good care of themselves than the latter, they
will tend to be overrepresented among the elderly. This selection phenome¬
non might offset an age-related decline in the happiness of these survivors,
making them highly reluctant to die. In fact, as shown in figure 5.1, survey
statistics indicate that the percentage of people who are “very happy” is
greater among octogenarians than among people in their thirties or forties.^^
This is a further illustration of the pervasiveness of selection bias in ex¬
plaining the attitudes and behavior of elderly people.
Another reason why the old can be expected to display as much as or
even more aversion to death than the young is that the sacrifice of an old
person is likely to confer a smaller expected gain on other people than the
sacrifice of a young person. A warrior who sacrifices himself in combat
may, by doing the risky deed that has led to his death, have contributed to
the survival of his nation, comrades, family, and way of life, and by con¬
tributing have obtained an altruistic benefit. The risking of life by a deed
done by one too old to be effective in combat is less likely to produce a
compensating gain. It might seem that the social loss from the sacrifice of

28. See, for example, William M. Landes and Richard A. Posner, The Economic Structure
of Tort Law 187-188 (1987); Marvin Frankel and Charles M. Linke, “The Value of Life and
Hedonic Damages: Some Unresolved Issues,” 5 Journal of Forensic Economics 233, 236 (1992).
29. The source for figure 5.1 is the General Social Survey for 1988-1993. See chapter 4
note 44.
 Economic Psychology of the Old Ill

(%)

—■— Very happy

—□— Pretty happy

—♦— Not too happy

18-29 30-39 40-49 50-59 60-69 70-79 80-89

Age Group

Figure 5.1 Self-report of happiness as a function of age

the old person would also be less; the warrior who survives can fight an¬
other day. But the old may have social value other than as fighters—for
example as counselors, priests, or judges—so that their lives are not free
goods to the society, to be flung away recklessly.
Moreover, as between two persons, one old, one young, facing what
would be the same risk if they were the same age (maybe both are equidis¬
tant from a drowning child), the risk of death or serious injury is apt to be
much greater to the old person because of his physical frailty.^^ This would
tend to make the expected cost of risky activity higher to old people than
to young ones even if the old did not dread death more than the young.
And rarely is the issue life versus certain death or certain serious in¬
jury. Cowardice is unwillingness to risk one’s life in circumstances in
which honor requires assuming such a risk; it is not (in our culture anyway)
unwillingness to commit suicide. If the young are more optimistic about
the outcome of risky choices than the old, this would make the disutility of
the objectively same risk of death greater to an old person than to a
young one.

30. Lloyd Cohen, “Toward an Economic Theory of the Measurement of Damages in a

Wrongful Death Action,” 34 Emory Law Journal 295, 332 (1985); see also Erin Ann O Kara,
“Hedonic Damages for Wrongful Death: Are Tortfeasors Getting Away with Murder?” 78
Georgetown Law Journal 1687, 1717-1718 (1990). For evidence, note that the distribution by
age of death of pedestrian fatalities in automobile accidents is highly skewed toward old people.
Robert Arking, Biology of Aging: Observations and Principles 35 (1991) (fig. 2-12). See also the
discussion of driving in the next chapter.
 112 The Theory Elaborated and Applied

A further point is that as the end of life approaches, other sources of

utility besides continuing to live recede. With the marginal value of com¬
peting goods, such as sex, travel, rich food and drink, and strenuous exer¬
cise, falling, the rational old person will reallocate resources to life¬
extending investments in medical care and in safety.^' He will become more
cautious than a young person because what he gives up in alternative con¬
sumption by being cautious is worth less to him. Equivalently, he has less
to gain from taking risks than a young person would.
“Our life resembles the Sibylline Books; the less there is left of it, the
more precious it becomes.” So Goethe. There are two interpretations. One
is that the nearer the prospect of death is, the keener is the appreciation of
what death does. When death is remote, we may take life for granted, not
dwelling on its joys and rewards; and what we take for granted we tend to
value less. Second, remaining life, like other inputs into activity, may have
scarcity value. As the end of life approaches, additional months or even
weeks may have great value in enabling a person to arrange his affairs, to
make farewells, and, by thinking over his life, to achieve an aesthetically
and psychologically pleasing sense of completeness or roundedness; this is
an aspect of the Nietzschean idea (or at least an idea attributed to Nietz¬
sche) of conceiving of one’s life as a work of art.
A final point is that it is arbitrary to suppose that the only thing lost
through death is future living. There is the destruction of the complex of
memories that constitutes a person. Until senility destroys many of those
memories, the old person has a greater stock of them than a young person,
which may somewhat offset, as a source of utility, the young person’s greater
life expectancy.32 This is implicit in Aristotle’s concept of the knowledge
shift. Conversely, however, that richer stock may give the old person a
greater sense of having fulfilled his potential for living, at least if he has
time (my previous point) to arrange his memories into a pleasing pattern.
Taken together, these points show how life-endangering risk-taking by

31. Isaac Ehrlich and Hiroyuki Chuma, “A Model of the Demand for Longevity and the
Value of Life Extension,” 98 Journal of Political Economy 761, 776-777, 780-781 (1990); see
also Cohen, note 30 above, at 332. Such a reallocation is particularly likely, of course, if the life¬
extending choice is subsidized but the alternative choices are not. It is therefore not surprising that
an estimated 28 percent of Medicare expenditures are for the treatment of people who are in their
last year of life. Dennis W. Jahnigen and Robert H. Binstock, “Economic and Clinical Realities:
Health Care for Elderly People, in Too Old for Health Care? Controversies in Medicine, Law,
Economics, and Ethics 13, 29-30 (1991). This figure is for 1978; I have not found more recent
figures.
32. On the utility derived from the recollection of pleasurable or exciting experiences, see
Jon Elster and George Loewenstein, “Utility from Memory and Anticipation,” in Choice over
Time in, 229-231 (George Loewenstein and Jon Elster, eds., 1992).
Economic Psychology of the Old 113

the old might be at once less beneficial and more costly than such risk¬
taking by the young, provided that the expected utility of the remaining
years of life is not the dominant factor in the fear of death.^^ But we must
not go overboard, and forget the higher suicide rates of old than of young
people (see next chapter) or neglect afterlife concerns, which generally
carry heavier weight for the old than for the young. The influence of such
concerns on behavior in the face of danger depends, however, on the nature
of a person’s religious beliefs. If he thinks himself likely to go to Hell, this
should increase his dread of death. But if he thinks himself likely to be
punished in the afterlife for cowardice if he does not behave courageously,
this should reduce his dread of dying as a result of a courageous act. And
if he is convinced of his salvation, he may not think he is giving up much
by dying.
Even if cowardice in the particular sense of reluctance to encounter
physical danger is a characteristic of old people (not of all, of course), it is
balanced, as Aristotle neglects to point out, by a form of courage unknown
to the young—courage in facing old age without flinching. Long before I
thought of writing about old age, I had remarked that Justice Holmes
“faced the indignities and deprivations of old age . . . with great courage
and gallantry, so that his last years completed a circle with the military
heroism of his youth.” Holmes, in this as in other respects luckier than
most, had a chance to prove his courage on two fields of human endeavor.
For most of us, old age will be the only field in which we have that chance.
Clemenceau, when asked in his old age what he would do now, replied,
“I am going to live till I die.” In like vein. Holmes ended a radio address
celebrating his ninetieth birthday with the following free translation of a
line from an anonymous medieval Latin poet: “Death plucks my ears and
says. Live—I am coming.” I have mentioned the downbeat poems about
old age that Yeats wrote when he was in his early sixties. Old age continued
to be a major theme of Yeats’s poetry until his death at the age of 73, but

33. For evidence that it is not, see Edmund C. Payne et al., “To Die Young, To Die Old:
Management of Terminal Illness at Age 20 and at Age 85: Case Reports,” 8 Journal of Geriatric
Psychiatry 107 (1975).
34. The Essential Holmes: Selections from the Letters, Speeches, Judicial Opinions, and
Other Writings of Oliver Wendell Holmes, Jr. xv (Richard A. Posner, ed., 1992) (editor s introduc¬
tion). The selections in chapter 1 of The Essential Holmes (“Aging and Death”) document this
claim.
35. Quoted in The Art of Growing Older: Writers on Living and Aging, note 1 above, at 177.
36. The Essential Holmes, note 34 above, at 21. See the exchange of letters between Holmes
and Frederick Pollock, May 4 and 15, 1931, in Holmes-Pollock Letters: The Correspondence of
Mr. Justice Holmes and Sir Frederick Pollock 1874-1932, vol. 2, pp. 285-286 (Mark DeWolfe
Howe, ed., 1941).
 114 The Theory Elaborated and Applied

the tone became increasingly defiant and even triumphal. Not in the sense
of Tennyson’s Ulysses, of hoping still to accomplish against all odds some
great work, let alone a sense of peace or wisdom attained; in most of the
poems of his old age, Yeats depicts the wisdom of the old as bitter, as pure
disillusionment—indeed as the answer to the question that furnishes the
title of one of those poems, “Why Should Old Men Be Mad?” What in¬
forms Yeats’s late poems is Nietzsche’s idea that man makes his own re¬
ality. The idea recalls the philosophy of self-sufficiency of the stoics, to
whom Nietzsche, and through him Yeats, were in fact indebted.^’ Or as
Hamlet put it (and we know that Shakespeare, too, was influenced by the
stoics), “there is nothing either good or bad, but thinking makes it so.”^*
Even old age is what we make of it. In “A Prayer for Old Age” Yeats prays
“That I may seem, though I die old, / A foolish, passionate man.” And in
“The Spur,” he is unmoved that “You think it horrible that lust and rage /
Should dance attention upon my old age . . . They were not such a plague
when I was young; / What else have I [now] to spur me into song?”
We are not all Holmes, Clemenceau, or Yeats; we cannot all expect to
be stoics of either the stolid or the triumphalist variety in our old age. Es¬
pecially those of us who are men. For the “cowardice” of the old of which
Aristotle speaks may be related to the common observation that as people
age the men tend to become more like women and the women more like
men.^^ The men tend to become more nurturant and less aggressive, the
women less nurturant and more assertive. These changes in the direction of
gender convergence should not be thought of purely in terms of deteriora¬
tion. As we shall see in chapter 9, they equip the elderly to find new social
niches for themselves.
Although I do not want to denigrate the courage that so many people
display in the face of the indignities of old age and the imminence of death,
I feel bound to point out that as life draws to a close considerations of
posthumous reputation loom larger in the rational individual’s utility func¬
tion. Not only is the posthumous state closer, but alternative investments to

37. On Nietzsche’s influence on Yeats, see references in Richard A. Posner, Law and Litera¬
ture: A Misunderstood Relation 150 n. 29 (1988). On the influence of the stoics on Nietzsche, see
Martha C. Nussbaum, The Therapy of Desire: Theory and Practice in Hellenistic Ethics 4-5
(1994).
38. Hamlet, act II, sc. ii, 11. 251—252. Yeats’s late poem “Death” ends, “Man has created
death;” and another late poem, “A Dialogue of Self and Soul,” ends, “When such as I cast out
remorse / So great a sweetness flows into the breast / We must laugh and we must sing, / We are
blessed by everything, / Everything we look upon is blest.”
39. See, for example, David Gutmann, Reclaimed Powers: Toward a New Psychology of
Men and Women in Later Life (1987); also Costa, McCrae, and Arenberg, note 6 above at
191-192.
Economic Psychology of the Old 115

creating posthumous reputation are less feasible. One’s deportment in the

face of death is an important part of most people’s posthumous reputation;
their brave words, their stoical demeanor, may be long remembered. Ad¬
mirable as these things are, we should recognize that their benefit-cost ratio
rises as death approaches; that they are in a sense “selfish” behaviors.

Physical and Mental Decline as Factors

in the Psychology of the Elderly
Decline of faculties and loss of powers do not play a large role in Aristotle’s
depiction of the psychology of old age, at least explicitly. He attributes
almost all the distinctive traits of the old to what I have been calling the
knowledge shift, which he does not attribute to any deterioration in the
mental abilities of the old. His approach both explains and is explained by
his choice of age 50 as the onset of old age. By then a person has seen
enough of life to have lost his youthful optimism; and it is doubtful that the
typical 50-year-old Athenian of the fourth century b.c. was mentally or
physically decrepit. But as I have remarked previously, it is a puzzle why
if there is no age-related decline the knowledge shift should involve loss as
well as gain. The elderly would be more pessimistic, but realistically so.
They would have both imagination and knowledge, the young only imagi¬
nation. Aristotle seems to have believed that experience displaced rather
than supplemented imagination, a sound belief because there is, though not
remarked by him, an age-related decline in fluid intelligence.
Once the threshold of old age is raised to a more realistic level than
Aristotle’s, the importance of age-related decline in explaining the charac¬
teristic psychology of the elderly becomes inescapable. It helps to explain
not only their “cowardice” but also their hesitation and tentativeness. One
way to compensate for diminished physical or mental capability is by in¬
vesting more time in doing tasks. So old people walk more slowly, drive
more slowly, and make decisions more slowly. The point is not that they
are incapable of walking fast, and so on, although that is also true, at least
in the sense that the exertional (not just expected-accident) cost of speed is
higher for older people, and sometimes prohibitive. “Behavioral slowing”
over a broad range of behaviors is a marked characteristic of aging organ¬
isms, animal as well as human; not all of it is a volitional response to a
changing environment of costs and benefits."*® But the point I want to make

40. James E. Birren, Anita M. Woods, and M. Virtrue Williams, “Behavioral Slowing with
Age: Causes, Organization, and Consequences,” in Aging in the 1980s: Psychological Issues 293,
302-303 (Leonard W. Poon, ed., 1980).
116 The Theory Elaborated and Applied

is that taking more time is also a rationally chosen adaptation to diminished

capability. An old person has a greater risk of falling than a young person
does. He has poorer balance and eyesight and slower reflexes, and being
more frail is also more likely to be injured if he falls. The benefits of using
more inputs of time are therefore greater to him, so he makes a deliberate
choice to walk and drive more slowly than he is capable of doing. The cost
of time is also likely to be less to an old than to a young person, and this
will increase the tendency of the old to substitute inputs of time for other
inputs into their activities.
Age-related decline helps explain why old professionals and academ¬
ics do not keep up with the literature in their fields as assiduously as the
young do. The cost of absorbing new information is higher to the old than
to the young because of the erosion of their fluid intelligence, and the bene¬
fit is smaller because they have fewer periods remaining in which to earn a
return on any new human capital that they do acquire. The broader point is
that, because of this changing balance between cost and benefit, the old are
less receptive to new ideas than the young. This is the kernel of validity to
the cruel aphorism (a “smoking gun” of age discrimination cases) that
“you can’t teach an old dog new tricks.” The old dog is rational in not
wanting to take the time to learn new tricks, as the cost will be greater and
the benefit smaller than in the case of a young dog. An additional cost is that
of abandoning habitual behaviors and settled beliefs. The older a person is,
the more deeply entrenched and hence more costly to change are his prac¬
tices, attitudes, and responses, just as the more addicted a person is to some
activity, whether smoking cigarettes or listening to classical music, the more
he will suffer from withdrawal symptoms if he abandons the activity even if,
as in the case of music, the addiction to it is purely psychological.
This analysis of the costs and benefits to the old of novelty adds a
further dimension to the aversion of the old to risk-taking activity. Often a
proposed course of action is risky in the sense of being quite likely to fail
merely because it involves doing something new. Were it merely the repe¬
tition of an old action, the likelihood of its succeeding would be easy to
estimate because there would be a track record. The risk of failure would
be minimized. So if, because of the decline of fluid intelligence or the cost
of breaking old habits, the old have trouble absorbing new ideas, it will be
difficult for them to evaluate risky choices.
Not taking risks or doing new things imparts to a person’s conduct a
conservative, rote style. We should expect it to be a more common style
among the old than among the young. A study of military leadership found
that, after correction for other factors, older generals are less likely than
Economic Psychology of the Old 117

younger ones to adopt an offensive rather than a defensive strategy/’ The

point again is not that the old are incapable of taking risks, absorbing new
ideas, and so forth, though that is sometimes the case, but that often it is
rational for them to shun such activities as not worth the cost, which is
higher to them than to the young.
Additional considerations come into play when we consider the well-
known aversion of the old to financial risk, in the sense of variance of
expected returns about the mean. The older a person is, the larger will be
the fraction of his total wealth that is constituted by financial rather than
human capital; indeed, at retirement, labor-market human capital is in ef¬
fect written off. So variance in the income on an old person’s financial as¬
sets will impart greater variance to his total income than in the case of a
younger person. A less technical way to put this is that older people have
less capacity to bear risk than younger people because they do not have
wages with which to cover losses resulting from the taking of investment
risks. Hence the optimal investment strategy of the older person is more
conservative.^^ This is reflected in the rule of thumb that the percentage of
bonds in one’s portfolio of bonds and common stocks should be equal to
one’s age—a 20-year-old should have only 20 percent of his portfolio in
bonds, an 80-year-old 80 percent.

Religion, Voting Preferences, and Speaking

We should not be surprised to find that old people are on average some¬
what more religious than young people (in belief, not practices, since the
incapacities associated with old age may limit church attendance)As I

41. Dean Keith Simonton, “Land Battles, Generals, and Armies: Individual and Situational
Determinants of Victory and Casualties,” 38 Journal of Personality and Social Psychology 110,
115 (1980).
42. Burton G. Malkiel, A Random Walk down Wall Street, Including a Life-Cycle Guide to
Personal Investing, ch. 13 (5th ed. 1990); David P. Brown, “Multiperiod Financial Planning,” 33
Management Science 848, 859-860 (1987).
43. I keep repeating this qualification, but perhaps not frequently enough. As we shall see in
chapter 13, the essence of age discrimination is refusal to recognize the variance in the attitudes,
behaviors, and so forth of middle-aged and elderly people. Here as in most of the book, indeed as
in most of economics, I am generally concerned with the mean of the distribution rather than with
the entire distribution.
44. See, for example, Cary S. Kart, The Realities of Aging: An Introduction 344-349 (3d
ed. 1990); David O. Moberg, “Religiosity in Old Age,” in Middle Age and Aging: A Reader in
Social Psychology 497, 508 (Bernice L. Neugarten, ed., 1968); Dan Blazer and Erdman Palmore,
“Religion and Aging in a Longitudinal Panel,” 16 Gerontologist 82 (1976); John M. Finney and
Gary R. Lee, “Age Differences on Five Dimensions of Religious Involvement,” 18 Review of
Religious Research (1976); Rodney Stark, “Age and Faith: A Changing Outlook or an Old Pro-
 118 The Theory Elaborated and Applied

remarked in chapter 3, the afterlife, being more imminent for the old, has a
greater weight in their thoughts and decisions. A slightly subtler prediction
is that among the old, religiosity will be negatively related to health, be¬
cause the unhealthy old have a shorter life expectancy."*^
We might expect old people to be on average selfish, single-issue vot¬
ers. Like politicians with short terms of office, the old have truncated hori¬
zons and therefore would be irrational to take the long view. This point is
only superficially in tension with the point about religiosity, which extends
the horizons of many old people but does so in a way that affects their
perceived self-interest; it need not make them altruistic voters. A deeper
problem is that since a rational person knows that his vote will not swing
the election, why shouldn’t he vote his convictions?"*^ What is the cost?
The answer may be that most people are convinced that what is good for
themselves is good, period. This is a rational conviction to hold since they
would gain nothing from being tugged in different directions by principle
and by self-interest. They avoid the pain of cognitive dissonance, and at
trivial cost.
Familial altruism is not the answer either. The old are better off sup¬
porting than opposing policies that transfer wealth to them from the young
as a group, since they can always give some of that wealth back to their
own children and grandchildren. This is a clue to why, as we shall see in
chapter 11, it is possible to obtain widespread political support for laws
reducing social security retirement benefits in the distant future. The old
have nothing to lose from laws that will bite after they are dead, and they
may have an altruistic desire to lighten the burden of taxation on the
younger members of their own families.
The greatest objection to the idea that old people invariably are selfish
voters is that it disregards disinterest. To the extent that the aged are in the
process of disengaging from the world, they have less stake in redistributive

cess?” 29 Sociological Analysis 1 (1968). Data from the General Social Survey for 1988-1993
(see note 29 above) generally confirm these studies. The elderly respondents exhibit on average
much stronger religious affiliation and somewhat stronger feelings of “nearness to God” than
younger persons, but no greater belief in an afterlife. That may be because the afterlife is a source
of fear as well as hope—or because beliefs in such things are not in fact dominated or even heavily
influenced by wishful thinking.
45. For evidence, see Bradley C. Courtenay et al., “Religiosity and Adaptation in the Oldest-
Old,” 34 International Journal of Aging and Human Development Al, 54 (1992).
46. For that matter, why should he vote at all? This is a profound question in economics and
political science. I return to it in the next chapter.
Economic Psychology of the Old 119

policies than younger persons do. In this respect old people resemble
judges (many of them old), whom we assume to be more impartial than
other decision-makers because the rules of judicial ethics require that they
have no family or financial stake in the cases they judge.
This point requires us to qualify Aristotle’s insistence that the old are
more self-centered than the young. Even if they are, their “self” may be
less affected by various decisions and policies than the selves of younger
persons. Although the young may be less selfish than the old when the cost
of being selfless is the same, the cost of voting selflessly may be higher for
the young because they have more to gain or lose from the governmental
policies at issue in the election, having a much longer period over which
gains and losses can accrue to them.
From Nestor to Polonius, Montaigne, and beyond, a frequently ob¬
served characteristic of old people is loquacity."*^ I suggested in the Intro¬
duction that this could be explained by reference to the lesser value of
privacy, consideration, and new information to the old than to the young.
Another factor is the difficulty of interpersonal transfer of lived experience.
If the knowledge that comes through experience were easy to transfer
through books or conversation (some of it is easy to transfer that way),
there would be no socially useful age-related attribute called “experience”
or “judgment.” The young would pick up these things by reading about
them. To the extent that lived experience is imperfectly transferable (it is
literally nontransferable, of course), we should expect older people to re¬
sort to elaborate, protracted speech in an effort to overcome the obstacles
to communication.
Not all old people are loquacious; some, indeed, are more taciturn than
they were when they were young. This may reflect the effect of another
element of the psychology of the elderly, their reluctance to take risks. A
further point is that much speech is reciprocal: I tell you something in the
hope of being told something useful in return. This kind of speech ex¬
change is related to transacting and is therefore less valuable to old people.
So we might expect not that old people on average are more loquacious

47. See, for example, Cicero, note 3 above, ch. XVI, § 55. For empirical evidence, see
Dolores Gold et al., “Measurement and Correlates of Verbosity in Elderly People,” 43 Journal^
of Gerontology P27 (1988). I am not concerned here with the pathological “off-target verbosity”
that afflicts a minority of old people as a result of a specific brain dysfunction. Tannis Y. Arbuckle
and Dolores Pushkar Gold, “Aging, Inhibition, and Verbosity,” 48 Journal of Gerontology P225
(1993). Or with gossip—but it is a plausible inference that because the old have lower opportunity
costs of time they gossip more than young people.
120 The Theory Elaborated and Applied

than young people, but that there is greater dispersion along the loquacity-
taciturnity continuum among the old than among the young.
As this last example suggests, the psychology of the old is not simple;
but economics, with help from evolutionary biology and the literary ex¬
amples with which this chapter is sprinkled, can help to make it intelligible.
I have argued, for example, that if, as is plausible, most people are innately
optimistic, the knowledge shift emphasized in previous chapters—the ten¬
dency of people as they get older to substitute reasoning based on experi¬
ence for reasoning based on imagination—implies that as people age they
become more pessimistic. A supporting point, one with independent sig¬
nificance, is that the decline of fluid intelligence with age, and the effects
of habit, cause aging people to have difficulty accepting and taking full
advantage of all the improvements associated with “progress,” while at the
same time they are more conscious of all the accompanying losses than the
young because they will have experienced them and not merely read about
them. Progress implies substitution of new for old services, products, ac¬
tivities, and so forth, and thus loss as well as gain, even though the latter
predominates (this is implied by the concept of “progress”)—but it may
not predominate in the eyes of the old. Averseness to novelties gives a
conservative cast to old people, but it is rational conservatism, based on
changes in the cost of absorbing novelty, not merely a mindless stand-
pattism. It is reinforced in the financial arena by the fact that when a person
stops working, variance in the return on his financial assets imparts greater
variance to his total income because that return is now a larger part of his
total income.
The tentativeness and hesitation that are characteristic of old people
are also rational adaptations to the increased risks that they face as a con¬
sequence of age-related decline. The self-preoccupation that is also char¬
acteristic of many old people, and characteristic features of the conversa¬
tion of the old that can be summed up by saying that old people tend to be
inconsiderate conversationalists, are mutually related consequences of the
diminished benefits of transacting to the old. The other side of this particu¬
lar coin, however, is that the old may be more disinterested than the young
because, as a consequence of their truncated horizon, they have less to gain
from selfish behavior. As for the puzzling “dread of death” of old people,
this is explicable by reference to the absence of a genetic program for old
age, as a consequence of which the instinctual dread of death that promotes
inclusive fitness in the young but not the old has not been eliminated in the
old by natural selection, and to a number of strictly economic factors as
Economic Psychology of the Old 121

well. These include the fragility of the old, the paucity of uses for their
resources other than to prolong life, and the related but distinct fact that the
opportunity cost of resources devoted to averting imminent death may be
zero, since in the absence of a bequest motive the resources will be useless
to the owner if he dies.
 6
Behavioral Correlates of Age

I want to consider what economic sense can be made of typical behaviors

of the old regarding residence, driving, crime, suicide, sex, employment
and retirement, voting and jury service. Some of these topics are treated in
other chapters as well, and an area of particular interest—productivity in a
broad sense, embracing creativity, leadership, and other forms of achieve¬
ment—I do not discuss in this chapter at all but devote the whole of the
next two chapters to. My purposes in this chapter are to illustrate the utility
of economic theory in explaining a wide variety of behavioral differences
between young and old and to lay additional foundations for the analysis
of policy in later chapters.

Driving
Old people on average drive much less than young people and have a higher
accident rate than all but the very youngest drivers.' The simplest life-cycle
model, in which the only significance of aging is that it brings a person
nearer to a finite horizon, might be thought to explain the higher accident

1. Leonard Evans, “Older Driver Involvement in Fatal and Severe Traffic Crashes,” 43
Journal of Gerontology S186 (1988); Donald W, Kline et al., “Vision, Aging and Driving: The
Problems of Older Drivers,” 47 Journal of Gerontology P27, P33 (1992); Richard A. Marottoli
et al., “Driving Cessation and Changes in Mileage Driven among Elderly Individuals,” 48 Journal
of Gerontology S255, S258 (1993); Joan E. Rigdon, “Older Drivers Pose Growing Risk on Roads
as Their Numbers Rise,” Wall Street Journal (midwest ed.), Oct. 29, 1993, pp. Al, A6. Unless
otherwise indicated, the data in this chapter are limited to the United States.

122
Behavioral Correlates of Age 123

rate simply by reference to reduced longevity: the old have less to lose. But
this would be inconsistent with the psychology of aging discussed in the
preceding chapter and with the low level of criminal activity of the old (of
which more shortly). And it would leave the reduction in driving by the old
unexplained, for while it is true that retired people are not commuters,
many of the leisure activities in which people engage are driving-intensive.
I believe that the driving patterns of the old are decisively influenced
by age-related decline. The characteristic age-related erosion of visual
acuity, reflexes, and concentration has a marked effect on capability for
driving safely.^ But it would be a mistake to infer that the accident rate of
the old would rise proportionately to the decline in relevant physical and
mental skills.
An age-related decline in driving skills raises the expected cost of driv¬
ing—both the expected cost of being injured (or incurring property dam¬
age) in an accident and the expected cost of liability for injuring someone
else (or inflicting property damage). Both costs are a positive function of
the probability of an accident, and that probability is greater, other things
remaining unchanged, the older the driver. The expected cost of being in¬
jured also rises with age because old people are more fragile and therefore
more likely to be injured if they are involved in an accident.
We can expect two types of response to the higher expected accident
costs of old drivers; an activity response, and a care response.^ One way to
reduce the expected costs of accidents is by driving less—cutting out mar¬
ginal trips that are no longer cost-justified when expected accident costs
rise—or not at all, if those costs become so high that they exceed the bene¬
fit of any driving. A parallel phenomenon in the workplace—the tendency
for older workers to leave jobs in which age-related sensory and motor
decline would make them prone to accidental injury—may explain the
negative correlation between age and industrial accidents.
The other response to higher expected accident costs of driving—driv-

2. See, for example, Rudolf W. H. M. Ponds, Wiebo H. Brouwer, and Peter C. van Wolffe-
laar, “Age Differences in Divided Attention in a Simulated Driving Task,” 43 Journal of Geron¬
tology P15\ (mS).
3. A fundamental distinction in the economic analysis of torts. See William M. Landes and
Richard A. Posner, The Economic Structure of Tort Law (1987); Steven Shavell, Economic Analy¬
sis of Accident Law (1987).
4. Mildred Doering, Susan R. Rhodes, and Michael Schuster, The Aging Worker: Research
and Recommendations 79 (1983). For evidence that the decisions of elderly people regarding
whether and how much to drive are heavily influenced by expected accident costs, as my analysis
predicts, see Marottoli et al, note 1 above, esp. S258-S259.
124 The Theory Elaborated and Applied

ing more carefully—is possible for the aged driver simply by driving more
slowly. This gives him more time to focus eyes and mind and to react to a
threatening situation. Since the opportunity cost of time is lower for the
old, it is not a costly substitution, so one is not surprised that old people are
involved in fewer accidents that are due to excessive speed (by them) than
are younger people.^ I am led to predict that the accident rate of old people
will rise more slowly than the decline in their driving skills. It might seem
that if the cost of time fell faster than the decline in driving skill, the acci¬
dent rate of the old might actually decline. But this is unlikely for three
reasons. First, time is not a perfect substitute for skill; otherwise a blind
person could drive safely. Second, difference in vehicle speeds, as well as
high speed, is a risk factor in accidents; that is why high-speed highways
are often posted with minimum as well as maximum speeds. Third, driving
slowly increases the amount of time one spends in automobile travel, so if
such travel is more likely to involve an accident than the activity that would
be substituted for it, there will be a partial offset to the effect of driving
slowly in reducing the risk of accident.®
Driving more slowly is not the only care response of the old to a greater
risk of being involved in an accident. They are also less likely to drive
under the influence of alcohol’ The reader will recall from the preceding
chapter that quite apart from increased fragility, the old are likely to devote
more resources than the young to preserving their lives because the oppor¬
tunity costs of life-extending measures are lower to the old than to the
young. A pertinent example is that the costs of reducing the amount of
liquor consumed are lower to the old because their bodies are less able to
tolerate liquor anyway.
Because care and activity changes are not perfect substitutes for re¬
flexes and other driving skills adversely affected by age, the elderly have a
higher accident rate than younger drivers, and since very young drivers also
have an abnormally high accident rate the function that relates driving
safety to age has the familiar inverted U shape. A plausible interpretation
is that safety grows with experience but at a diminishing rate eventually
overcome by age-related decline.
It would be a mistake to infer from the high accident rate of old people
that their driving ought to be curtailed by more stringent licensing require¬
ments. This would overlook the value to them of their inframarginal driv-

5. Kline et al., note 1 above, at P33.

6. Cf. Landes and Posner, note 3 above, at 238 n. 17.
7. Isaac Ehrlich and Hiroyuki Chuma, “A Model of the Demand for Longevity and the Value
of Life Extension,” 98 Journal of Political Economy 161, 781 (1990) (tab. 5).
Behavioral Correlates of Age 125

ing—the driving they continue to do despite its dangerousness—and the

fact that so long as insurance companies are permitted to adjust their pre¬
miums for age, as they are, the risks posed by old drivers can be internal¬
ized without need for governmental intervention beyond the provision of a
tort law system. A high accident rate, moreover, does not necessarily imply
a high degree of risk to other users of the roads. An accident rate is a rate
per mile driven. The total number of miles driven per year by the elderly is
much smaller than in younger age groups,* both because there are fewer
elderly and because (the activity point) they drive on average less than
younger people. Hence the elderly are not nearly so big a source of auto¬
mobile accidents to others as their accident rate would suggest. The quali¬
fication is vital. Because the old are fragile, they are more likely than the
young to be killed in automobile accidents; they are also more likely to be
pedestrian fatalities.® But they are less likely than the young to inflict fa¬
talities. For example, male drivers aged 65 cause only 32.8 percent as many
pedestrian deaths as male drivers aged 40 and only 11.8 percent as
many as 20-year-old male drivers.And the rate of total vehicle crash in¬
volvements per number of licensed male drivers decreases monotonically
with age, with just a slight uptick at age 85 and above. Even in that age
group, the rate is one-fourth the teenage male crash involvement rate and
one-half the rate of male drivers aged 25 to 29."
The fact that the old are not responsible for a large number of accidents
does not prove that letting them drive is a cost-justified policy. Elderly driv¬
ers may not be a major threat to highway safety; yet if miles driven are a
good index of the benefits of driving, the low aggregate accident costs that
the elderly inflict on other users of the road would be balanced by low

8. For example, in 1990 drivers aged 75 to 79 drove 26.3 million miles, compared to
141.6 million for drivers aged 50 to 54. Ezio C. Cerrelli, “Crash Data and Rates for Age-Sex
Groups of Drivers, 1990” 10 (U.S. Dept, of Transportation, National Highway Traffic Safety
Administration Research Note, May 1992) (tab. C). For evidence that rational behavior of the
elderly with respect to safety is consistent with their having higher injury rates, see next footnote.
9. Id., p. 6; U.S. Dept, of Transportation, National Highway Traffic Safety Administration,
“Traffic Safety Facts 1992: Older Population” 3 (n.d.). See also Dawn L. Massie and Kenneth L.
Campbell, “Accident Involvement Rates by Age and Gender,” UMTRl Research Review, March-
April 1993, p. 1. In 1992, persons 70 and older, though less than 9 percent of the U.S. population,
accounted for 11.8 percent of all deaths of occupants of motor vehicles and 17.9 percent of all
pedestrian deaths. “Traffic Safety Facts 1992: Older Population,” above, at pp. 1, 3. For empirical
evidence that elderly pedestrians respond rationally to the greater risk of being injured or killed by
exercising greater care than younger, less vulnerable people, see W. Andrew Harrell, “Precaution¬
ary Street Crossing by Elderly Pedestrians,” 32 International Journal of Aging and Human De¬
velopment 65 (1991).
10. Evans, note 1 above, at S192 (tab. 1).
11. Cerrelli, note 8 above, p. 7 (fig. 1).
126 The Theory Elaborated and Applied

aggregate benefits to the elderly. But I doubt that those benefits are low.
Given the incentive of elderly drivers to reduce their driving, both for their
own protection and to minimize liability-insurance costs, it is likely that
the driving they do confers large benefits on them. Indeed, to conclude
otherwise would cast doubt on the proposition that changes in activity level
are a socially valuable method of accident prevention. Suppose that rail¬
roads are inherently much more dangerous than canals, and as a result the
pricing of transportation services to reflect accident risks as well as all
other costs causes most freight to be carried by canal. The logical inference
would be that the remaining rail transportation, that for which canal trans¬
portation is not substitutable even when accident costs are taken into
account, must confer very large benefits on shippers; else these shippers
would be induced to switch to canals too. A similar inference seems war¬
ranted in the case of elderly drivers. This is one more example of the im¬
portance of selection bias in illuminating (and, in this example, in enabling
a social valuation to be placed on) the behavior of the elderly. The elderly
who continue driving are in effect selected for the high value of their
driving.
If this analysis is correct, it would be bad policy to curtail driving by
the old by requiring the elderly driver, as a condition of retaining his or her
license, to demonstrate a level of skill equal to that of a young driver. A
less skilled elderly driver may be able to come up to the same level of care
as the young simply by driving slowly, and may well derive benefits from
driving that are large enough to offset the higher risk of an accident to
others that his driving does create. Yet it is equally a mistake to propose
measures to increase the safety of elderly drivers or pedestrians, such as
increasing the pedestrian-crossing time allowed by traffic lights, without
considering the cost of the measures to other users of streets.'^

The Involvement of the Elderly in Crime

as Victims and as Offenders
The elderly crime victim. Although the aged are overrepresented in
the statistics of traffic deaths, they are underrepresented in the statistics of
criminal victimization. Persons 65 or older were victims in only 5 percent

12. As in Russell E. Hoxie and Laurence Z. Rubenstein, “Are Older Pedestrians Allowed
Enough Time to Cross Intersections Safely?” 42 Journal of the American Geriatric Society 241
(1994).
Behavioral Correlates of Age 127

of the murders committed in 1992 and only 2 percent of all crimes com¬
mitted that year, even though they comprise more than 12 percent of the
population.'^ Because older persons are frailer than younger ones, the ex¬
pected cost to them of a criminal encounter is greater,'"' so we expect them
to take greater precautions. The cost of these precautions is, properly
speaking, a cost of crime; statistics of victimization therefore understate
those costs. But this is true for everyone, and it is not obvious that, despite
their greater frailty and hence greater vulnerability to crime, the elderly
incur higher costs (nonpecuniary as well as pecuniary, of course) of avoid¬
ing becoming victims of crime. For the cost of precautions is lower to el¬
derly than to young people. The elderly are less active and mobile anyway,
so it costs them less to avoid going out at night or living or working (they
don’t work, most of them) in dangerous areas. The large disparity between
traffic and crime fatalities to older people suggests, plausibly enough, that
it is more costly for the old to avoid automobile travel or crossing streets
than to avoid places where criminals lurk. Differences in the cost of avoid¬
ing activities are no doubt also one explanation for why the ratio of fatal
traffic accidents to fatal nontraffic accidents is much lower among elderly
than middle-aged persons.'^ It is more costly for the elderly to avoid activi¬
ties that create a danger of a fall or a burn (the most common types of
nontraffic accident to elderly people) than those that create a danger of an
automobile accident or a criminal assault. Even a person who never leaves
his residence faces a risk of falling or of scalding himself or of being in¬
jured in a fire.

13. U.S. Dept, of Justice, Bureau of Justice Statistics, “Elderly Crime Victims: National
Crime Victimization Survey” (NCJ-147002, March 1994). There is, no doubt, underreporting of
intrafamilial violence against the old, most of it committed by spouses. Karl Pillemer and David
Finkelhor, “The Prevalence of Elder Abuse: A Random Sample Survey,” 28 Gerontologist 51
(1988). But no one seems to know whether it is greater than the underreporting of intrafamilial
violence directed against younger persons, particularly child.wn. And a recent study suggests that
the prevalence of physical abuse of elderly persons has been greatly exaggerated. Beletshachew
Shiferaw et ah, “The Investigation and Outcome of Reported Cases of Elder Abuse: The Forsyth
County Aging Study,” 34 Geronm/ogwt 123 (1994).
j4 This may explain why a number of states make assault and battery a more serious crime
when the victim is elderly. Annotation, “Criminal Assault or Battery Statutes Making Attack on
Elderly Person a Special or Aggravated Offense,” 73 A.L.R.4th 1123 (1989).
15. William Wilbanks, “Trends in Violent Death among the Elderly,” 14 International
Journal of Aging and Human Development 167, 170 (1981-1982) (tab. 2).
16. See, for example, Harmeet Sjogren and Ulf Bjdrnstig, “Unintentional Injuries among
Elderly People: Incidence, Causes, Severity, and Costs,” 21 Accident Analysis and Prevention 233
(1989); Jeanne Ann Grisso et ah, “Injuries in an Elderly Inner-City Population,” 38 Journal of the
American Geriatric Society 1326 (1990).
 128 The Theory Elaborated and Applied

An alternative explanation for the low rate of criminal victimization of

the elderly is that they have an irrational fear of crime and therefore take
excessive precautions, as by becoming hermits. Although popular, this
view appears to be exaggerated.'’ Survey data do reveal, however, that el¬
derly people are more afraid to walk at night in their neighborhoods than
young people are.'*
It is noteworthy that between 1973 and 1992, a period when the overall
rate of crimes of violence was essentially constant, the rate of crimes of
violence against persons 65 or older fell by almost 50 percent, while the
rate of thefts from persons in that age group fell more slowly than the over¬
all rate of personal theft.These trends may be related to the rapidly grow¬
ing income of elderly people during this period, which both facilitated their
choice of residential and activity patterns in which the risk of violence
would be minimized and made them more attractive targets for acquisitive
crimes.

The elderly ojfender. The low rate of criminal victimization of the

elderly is matched and in fact exceeded by their remarkably low crime rate.
Of total arrests in the United States in 1992, only 0.7 percent were of per¬
sons aged 60 to 64; persons 65 and over accounted for an identical per-
centage.’° The total for both groups was thus only 1.4 percent, even though
persons aged 60 and older constitute more than 20 percent of all Americans
15 years old or older. As one would expect, the percentage of arrests of
persons 65 and over for the most serious crimes of violence—murder, forc¬
ible rape, robbery, and aggravated assault—is lower than for serious prop¬
erty crimes such as burglary, larceny-theft, motor vehicle theft, and arson.
But the difference is small—0.6 versus 0.8 percent—and there is no differ¬
ence when we drop down to the 60 through 64 year olds. The meagerness
of the difference between the violent and nonviolent crime rates of the el-

17. Randy L. LaGrange and Kenneth F. Ferraro, “The Elderly’s Fear of Crime: A Critical
Examination of the Research,” 9 Research on Aging 372 (1987),
18. According to data from the General Social Survey (see chapter 4, note 44), 60 percent
of persons in their eighties are afraid, compared to only 36.7 percent of those in their forties.
19. U.S. Dept, of Justice, Bureau of Justice Statistics, Criminal Victimization in the United
States: 1973- 92 Trends 1,13 (NCJ-147006, July 1994).
20. The statistics in this paragraph are from U.S. Dept, of Justice, Eederal Bureau of Inves¬
tigation, Uniform Crime Reports; Crime in the United States 1992 228 (1993) (tab. 38). Arrest
statistics are imperfect proxies for crimes committed, but I have found no data on the age of
persons who commit crimes but are not arrested. For an excellent review of the literature on elderly
criminality, see Kyle Kercher, “Causes and Correlates of Crime Committed by the Elderly: A
Review of the Literature,” in Critical Issues in Aging Policy: Linking Research and Values 254
(Edgar F. Borgatta and Rhonda J. V. Montgomery, eds., 1987).
Behavioral Correlates of Age 129

derly and near-elderly is surprising, especially since one might expect vio¬
lent criminals to be particularly likely to be killed before they reached the
threshold of old age.
We shall see that arrest statistics probably underestimate the amount
of criminal activity of the old. Nevertheless, that amount must be small.
Elderly criminality was found, in the only economic study of the sub¬
ject that I have discovered, to be “rational” in the sense of being responsive
to legitimate employment opportunities and other factors bearing on the
costs and benefits of alternative uses of one’s time.^' The study did not,
however, attempt to explain the striking difference between the crime rates
of old and young.
The fact that the violent and nonviolent crime rates of the elderly are
so close suggests that enfeeblement is not the principal factor in the re¬
markable law-abidingness of old people. Another reason to doubt this is
that the crime rate is already very low for persons in the 55 to 59 age group;
only 1.1 percent of total arrests are of persons in that group.^^ No doubt
enfeeblement plays some role, a controversial example being the propen¬
sity of elderly sex offenders to commit offenses against children rather than
adults.23 And there are more old women than old men, and criminality is
much less common among women than among men. But enfeeblement and
the falling sex ratio can together explain only a part of the age-related de¬
cline in criminality.
The decline challenges the simple life-cycle model of incentives and
behavior, especially in a system such as ours in which the principal method
of punishing crimes is imprisonment. Even a very long prison term is un¬
likely to be an effective deterrent to the commission of crimes by persons
unlikely to survive more than a small part of the term, especially because it
has long been the practice to release very old prisoners rather than compel
them to die in prison. And since old people are likely to seem rather “harm¬
less,” concerns with recidivism do not play a large role in the sentencing

21. Donald J. Bachand and George A. Chressanthis, “Property Crime and the Elderly Of¬
fender: A Theoretical and Empirical Analysis, 1964-1984 ” in Older Offenders: Perspectives in
Criminology and Criminal Justice 76 (Belinda McCarthy and Robert Langworthy, eds., 1988).
22. U.S. Dept, of Justice, note 20 above, at 228 (tab. 38).
23. E. A. Eattah and V. F. Sacco, Crime and Victimization of the Elderly 39-48 (1989); cf.
William Wilbanks, “Are Elderly Felons Treated More Leniently by the Criminal Justice System?”
26 IntemationalJoumal of Aging and Human Development 275, 282 (1988) (tab. 3). For evidence
that “opportunistic” pedophilia—substituting a child as the victim of a sexual offense because the
offender is too feeble to rape an adult—is rare, see A. Nicholas Groth, Men Who Rape: The
Psychology of the Offender 144-145 (1979); Marc Hillbrand, Hilliard Foster, Jr., and Michael
Hirt, “Rapists and Child Molesters: Psychometric Comparisons,” 19 Archives of Sexual Behavior
65,69(1990).
 130 The Theory Elaborated and Applied

of the old—another reason to expect their punishments to be too slight to

deter effectively.The puzzle is deepened when the concept of multiple
selves is brought into the picture. To the extent that the young self can be
regarded as a separate person from his old self, the segment of a long prison
sentence that will be served when the criminal is old is tantamount to pun¬
ishment for a different, albeit closely related, person’s crime.
In light of these points, how are we to explain the low crime rate of the
elderly? There are several explanations. To an economist, crime is a form
of work, and productivity in any line of work is a function, in part at least,
of investment in human capital. We therefore would not expect elderly
people to invest heavily in the acquisition of criminal skills; and without
those skills, the probability of apprehension and conviction is greater.
Since the expected cost of punishment is approximately the present disu¬
tility of punishment times the probability that punishment will be imposed,
a drop in the former term can be offset by an increase in the latter term.
The offset is particularly likely if criminal skills differ significantly
from those employed in lawful occupations. The decline of fluid intelli¬
gence increases the cost of learning new skills. Thus, not only are the bene¬
fits of investing in the acquisition of criminal skills truncated by the short
life expectancy of the old; the cost of making the investment is higher than
in the case of young people. Such skills are therefore unlikely to be ac¬
quired late in life, and without them the probability of apprehension and
conviction, and so the cost of becoming a criminal, will soar. If for these
reasons the demand for criminal activity by old people is slight even when
the severity of punishment is also slight, the lenient treatment of elderly
offenders by the criminal justice system may be optimal.
The severity with which elderly criminals are punished may, moreover,
be greater than an exclusive focus on the truncation of prison sentences
would suggest. The reasons are twofold, and were suggested in earlier
chapters. First, one’s time may actually become more valuable (in terms of
utility, not earnings) as one ages. Although at some point the difference

24. For evidence that elderly criminals are in fact punished lightly, see Fattah and Sacco,
note 23 above, at 72—75; Dean J. Champion, “The Severity of Sentencing: Do Federal Judges
Really Go Easier on Elderly Felons in Plea-Bargaining Negotiations Compared with Their
Younger Counterparts?” in Older Offenders: Perspectives in Criminology and Criminal Justice,
note 21 above, at 143; John H. Lindquist, O. Z. White, and Carl D. Chambers, “Elderly Felons:
Dispositions of Arrests,” in The Elderly: Victims and Deviants 161 (Carl D. Chambers et al, eds.,
1987). However, Wilbanks, note 23 above, presents evidence that elderly felons are not treated
more leniently by the California criminal justice system when other factors besides age are con¬
trolled for. On the bearing of the federal sentencing guidelines on lenient treatment of the elderly
offender, see discussion in chapter 12.
Behavioral Correlates of Age 131

between freedom and confinement becomes pretty academic—one is not

much freer in a nursing home than in a prison, and indeed the differences
between a nursing home and a prison geriatric ward may be slight—few
enfeebled elderly are capable of committing crimes.
Second, not everyone’s future is bounded by his life. If old people are
more likely than young ones to believe in an afterlife in which the good are
rewarded and the wicked punished, they are less likely to think that any
punishment visited on them for criminal acts will be truncated by the death
of the body. Likewise if they want to leave a good name. Although few of
us believe any more that the sins of the father are visited upon the children
even unto remote generations, families do have reputations, good or bad,
that affect the terms on which their members can engage in advantageous
transactions. The notion of “bringing disgrace upon one’s family” is not
entirely archaic. An old person who is altruistic toward the members of his
family who are likely to survive him will incur disutility if by committing
a crime he imposes a reputation cost on those younger members. And the
older he is, the larger will the bequest motive—including a “reputation
bequest” motive—and other forms of posthumous utility loom in his utility
function, as other sources of utility fall away.
My analysis predicts that, other things being equal, the elderly crime
rate will be greater (1) the more severely the particular type of offense is
punished (because the truncation of punishment due to the imminence of
death will be greater the longer the sentence), and (2) the less new skill is
required for its commission. So we would expect, for example, more arrests
for drunk driving (per mile driven) among old people relative to young than
for counterfeiting, because driving is not a new skill for old people. But we
would still expect fewer arrests for drunk driving among old than among
young drivers, because the expected accident cost from drunk driving is
greater for the elderly than for the young. My prediction about the age
incidence of different crimes is borne out by the statistics on arrests of the
elderly. Recall that only 0.7 percent of all arrests are of persons 65 or older.
For driving under the influence, however, the percentage rises to 1.3 (which
is still far below the percentage of elderly persons in the population), and
for forgery and counterfeiting it falls to 0.3.^^
I have been discussing the costs to the elderly of committing crimes.
Another reason for the low crime rate of the elderly, though one likely to
be operative in only a small fraction of cases, is that for crimes that yield
benefits over a long period of time (killing a hated enemy, perhaps?), the

25. Computed from U.S. Dept, of Justice, note 20 above, at 228 (tab. 38).
 132 The Theory Elaborated and Applied

benefits are greater the younger the criminal. Such a crime is, in effect, an
investment and we know that the return on investments is truncated for
elderly people because of their short life expectancy.

The career criminal. I have thus far been considering the case of
people who were law-abiding until they became old. What of career crimi¬
nals? Why do they seem to burn out when they get old?^® These are people
whose skills are specialized to criminal activity, who have already blotched
the family escutcheon, and who are less likely than the law-abiding to be
religious.^’ It might seem, given the decline with age in fluid intelligence,
that it would be as difficult to switch out of criminal activity at an advanced
age as to switch into it at an advanced age. Many criminal occupations are
strenuous and dangerous, however, and such occupations take a toll on a
person’s health, leading to earlier retirement than in softer, safer jobs; recall
the relation of stress to burnout discussed in chapter 4. A less obvious point
is that arrest and prosecution have a sorting effect. The less skillful crimi¬
nals are caught repeatedly and learn from this that they aren’t very good at
what they do; so they tend eventually to drop out, thinning the ranks of
older criminals. The more skillful avoid frequent arrest or conviction, and
hence are underrepresented in statistics of older criminals, which are based
on arrest. This point implies that the elderly are not quite so law-abiding as
one would infer from arrest statistics alone.
Possibly the most important factor in the very early “retirement” of
criminals is that every time a person is caught and convicted, the proba¬
bility that he will be caught and convicted the next time he commits a crime
is increased. He now has a criminal record, which will make the police
more likely to suspect him; and if a criminal defendant testifies in his own
defense, his prior convictions can be used to undermine the credibility of
his testimony. This prospect discourages defendants with a record of con¬
victions from testifying, and by doing so increases the probability of their
conviction (though presumably less than if they did testify), since, although
instructed otherwise, jurors frequently infer guilt from a defendant’s failure
to testify. Not only is the recidivist thus easier to catch and convict (at least
if we ignore the fact, discussed below, that he is apt to be more experienced
than a first offender); he is bound, as a recidivist, to receive a heavier sen-

26. Not all of them, of course; and as human-capital theory predicts, the “professional”
criminal who invests in his criminal skills remains a criminal longer than the amateur. Evelyn S.
Newman et al.. Elderly Criminals 8-11 (1984).
27. Lee Ellis, “Religiosity and Criminality: Evidence and Explanations surrounding Com¬
plex Relationships,” 28 Sociological Perspective 501 (1985).
Behavioral Correlates of Age 133

tence. On both counts the expected punishment cost of the already con¬
victed is higher than that of the first offender, and the cost mounts with
every successive conviction. Eventually the expected punishment cost is so
high, even with the truncation effect, that the career criminal is deterred,
“retires,” unless he has exceptional skills which enable him to avoid being
caught; on either count we expect age and known criminal activity to be
negatively correlated even for career criminals.
We might even speculate that one reason for punishing recidivists more
heavily than first-time criminals is to maintain constant deterrence in the
face of the age-related decline, due to the truncation effect, in the deterrent
effect of punishment. There is a parallel to the backloading of compensa¬
tion in order to offset the last-period problem, discussed in chapter 3. The
expected cost of imprisonment, as of being fired from one’s job, declines
as one gets closer to retirement from life and from work, respectively. The
decline can be offset, up to a point anyway (an important qualification, to
which I return in chapter 12), by lengthening the prison sentence of re¬
peat offenders, who by virtue of being “careerists” are likely to continue
committing crimes to an advanced age. A further justification for the
heavier punishment of recidivists is that it is necessary to counteract learn¬
ing by doing, that is, the reduction in the probability of being apprehended
and hence in the expected cost of punishment as a criminal gains more
experience.
This discussion of the career criminal helps to show why the reported
crime rate of the elderly is so very low. Both an aging effect and a selection
effect are at work. For those who start out straight, the benefit-cost ratio of
switching into criminal activity when old is adverse. For those who start
out crooked, few survive as criminals to old age unless they are so good
that they are never caught and convicted, in which event they do not show
up in the statistics of elderly crime.

Suicide
While crime rates decline with age, male suicide rates rise monotonically
with age after declining in-'early adulthood.^* If crime and suicide are

28. U.S. Senate Special Committee on Aging et al., Aging America: Trends and Projections
115 (1991 ed.). In 1990, for example, the suicide rate was 34.2 per 100,000 for white men aged 65
to 74, 60.2 for those aged 75 to 84, and 70.3 for those aged 85 and older, compared to only
25.3 for those aged 35 to 44. U.S. Bureau of the Census, Statistical Abstract of the United States
1993 99 (113th ed.) (ser. 137). This pattern appears to hold in most other nations as well, al¬
though in Finland, which has the highest suicide rate of any Western country, the suicide rate peaks
at ages 55-64, and in Denmark and Germany there is a dip between that and the next age
 134 The Theory Elaborated and Applied

thought to be merely different manifestations of a propensity for violence,

this divergence is puzzling. Not so if an economic approach is taken. The
utility of living, net of disutility due to suffering, bereavement, and other
losses, declines with age. A falling present value of remaining life inter¬
sects a rising curve of suffering. Gary Becker and Rebecca Kilburn empha¬
size two further points.^^ First, since perceptions of welfare are relative,
aging may cause a sharp drop in utility even for persons who are “objec¬
tively” well off, because they will compare their present state to their own
earlier state rather than just to that of other persons. Second, one compo¬
nent of expected utility for a young person—the “option” value of contin¬
ued living, a value generated by the possibility that the quality of one’s life
will improve significantly at some future time—is greatly diminished in
old age. Aging reduces both the objective likelihood of such an improve¬
ment and, because of the effect of aging in boiling away the optimism of
youth, the subjective likelihood as well. And if there is an improvement,
there will be less time to enjoy it. For all these reasons, aging reduces the
option value of living on.
Family altruism, as well as fear of death and other costs of suicide (of
which more in chapter 10), may deter the elderly person from ending his
life even if his net expected utility of continuing to live would otherwise be
negative. Or may not. The stigma of suicide is less than that of crime, and
in some circles is nil, at least when committed by an elderly person in poor
health. Hence the effect of suicide on the family’s “good name” may not
be great. In some cases family altruism will actually increase rather than
reduce the net expected benefit of suicide, by accelerating the inheritance
of the elderly person’s property. By reducing his consumption to zero,
the elderly person who, having inheritable wealth, ends his life enables
greater consumption by his heirs unless he is a much more efficient saver
than they.
Since utility is a positive function of income, and since the incomes of
the elderly have risen dramatically in recent decades, we should not be
surprised that the white male elderly suicide rate has fallen in the last half

group (65-74), although the peak rate is, as in the United States, in the 75-and-over group. See
World Health Organization, World Health Statistics Annual, various years. For an excellent dis¬
cussion of the U.S. and foreign statistics on suicide by the elderly, see John McIntosh, “Epidemi¬
ology of Suicide in the Elderly, in Suicide and the Older Adult 15 (Antoon A. Leenaars et al.,
eds., 1992).
29. Gary S. Becker and M. Rebecca Kilburn, “The Economics of Misery” (unpublished.
University of Chicago Dept, of Economics, Dec. 19, 1993).
Behavioral Correlates of Age 135

century.^® This explanation for the data is too pat, however, because the 65-
and-over suicide rate rose during the 1980s^' even though the prosperity of
elderly Americans was continuing to grow. The rise was not merely the
result of the higher number of people in the very oldest age groups, for
whom the utility of life might be especially meager; the suicide rate of
white males aged 65 to 69 rose between 1981 and 1989.^^
An alternative, noneconomic explanation for the high suicide rate of
the elderly is that depression increases the risk of suicide and is more preva¬
lent among the elderly.” But it is unclear whether depression really is more
prevalent among them” unless we give “depression” a circular definition
that makes it a mysterious something that predisposes people to suicide.
Depression is difficult to distinguish operationally from rational, clear-eyed
evaluation of declining quality and prospects of life; that is, from “depres¬
sive realism” (see chapter 5). A study that ascribes elderly suicide to de¬
pression also recommends such “treatments” as furnishing elderly people
with pets.^^ As a cat lover of the most abjectly sentimental sort, I would be
the last person in the world to question the value of “pet therapy.” But it is
more accurately if less impressively described as a method of increasing
happiness, and thus reducing the incentive to self-destruction, than as a
“treatment” for “clinical depression.”
The high rate of suicide among the elderly is especially striking be¬
cause of selection bias, the omnipresent problem in attempting to infer the

30. Patricia L. McCall, “Adolescent and Elderly White Male Suicide Trends: Evidence of
Changing Well-Being?” 46 Journal of Gerontology S43, S44 (1991) (fig. 1); Dan Blazer, “Suicide
Risk Factors in the Elderly: An Epidemiological Study, 24 Journal of Geriatric Psychiatry 175,
177 (1991); James R. Marshall, “Changes in Aged White Male Suicide: 1948-1972,” 33 Journal
of Gerontology 763 (1978).
31. Nancy J. Osgood, Suicide in Later Life: Recognizing the Warning Signs 10-13 (1992);
Mark S. Kaplan, Margaret E. Adamek, and Scott Johnson, “Trends in Firearm Suicide among
Older American Males: 1979-1988,” 34 Gerontologist 59 (\994).
32. Calculated from Bureau of the Census, Vital Statistics of the United States, 1965-1989.
33. See, for example, Kalle Achte, “Suicidal Tendencies in the Elderly,” 18 Suicide and
Life-Threatening Behavior 55, 57 (1988).
34. See James C. Anthony and Ahmed Aboraya, “The Epidemiology of Selected Mental
Disorders in Later Life,” in Handbook of Mental Health and Aging 27, 42, 46 (James E. Birren
et al., eds., 2d ed. 1992); Blazer, note 30 above, at 182-183; Gerda E. Gomez and Efrain A.
Gomez, “Depression in the Elderly,” Journal of Psychosocial Nursing, no. 5, p. 28 (1993). In
addition to the difficulty discussed in the text, many of the symptoms of depression—such as
apathy, loss of appetite, and sleeplessness—are equally symptoms of physical disorders of old age.
35. Nancy J. Osgood, Barbara A. Brant, and Aaron Lipman, Suicide among the Elderly in
Long-Term Care Facilities\\5,140(1991). On the difficulty of fitting the conventional psychiatric
criteria of depression to “depression” in elderly people, see Dan G. Blazer, “Affective Disorders in
Late Life,” in Geriatric Psychiatry 369,370-371 (Ewald W. Busse and Dan G. Blazer, eds., 1989).
136 The Theory Elaborated and Applied

(%)

18-29 30-39 40-49 50-59 60-69 70-79 80-89

Age Group

Figure 6.1 Is suicide okay if disease is incurable?

(%)

18-29 30-39 40-49 50-59 60-69 70-79 80-89

Age Group

Figure 6.2 Is suicide okay if tired of living?

effect of age on behavior from comparisons between old and young. The
most suicidal people can be expected to commit suicide at relatively young
ages, with the result that the elderly represent a population sample biased
against suicide. This may explain the puzzling fact that, as shown in fig¬
ures 6.1 and 6.2, older people are much more likely than young or middle-
Behavioral Correlates of Age 137

aged ones to believe that it is wrong to commit suicide just because one has
an incurable disease or is tired of living.^^
Selection bias has a further significance in explaining the pattern of
elderly suicides. It implies that the suicide attempts of the elderly are likely
to be more deliberated, less impulsive, and more likely to succeed than
those of the young, because those prone to commit suicide impulsively are
likely to have done so before reaching old age. (For evidence, see chap¬
ter 10.) A related implication is that elderly suicides are less likely to be
the product of clinical depression (unless such depression really is more
common among old than among young people, despite the doubts that I
have expressed), because a disproportionate number of persons suffering
from depression will have committed suicide before reaching old age.
The steep age gradient for suicide is limited to men. Not only is the
female suicide rate much lower, but it does not increase as much with age.
The ratio of male to female suicide rates is 6.10 among persons 65 and
older, compared to only 3.81 in the population as a whole.^'' Proneness to
suicide is thus an area in which there is a greater difference between men
and women in old age than at earlier ages. There is no generally accepted
explanation for either the divergence in suicide rates between men and
women or the increase in that divergence with age.

Penny-Pinching
Fortunately, unsafe driving, crime, and suicide are not the principal activi¬
ties of elderly persons. The principal activities are work, recreation, and
household production, including care of self.^® A minor but extremely cu¬
rious activity of the old is penny-pinching—as seen in coupon-shopping
and bargain-hunting generally. Young people are amazed at the lengths to
which the old will go to save a few dollars. But this behavior is rational.
The cost of time is very low for retired people; and penny-pinching, as by
careful shopping, is a time-intensive activity. Furthermore, retired people
generally have a lower income than when they were working, implying (by
the assumption of diminishing marginal utility of income) that they value
the extra dollar more in retirement than when they were working. Since the
costs of penny-pinching are lower to the old and the benefits greater, we

36. The source of the data for figures 6.1 and 6.2 is the General Social Survey. See note 18
above. The possibility that a cohort effect is at work cannot be excluded, however.
37. Robert Travis, “Suicide in Cross-Cultural Perspective,” 31 International Journal of
Comparative Sociology 237, 241, 244 (1990) (tabs. 1,2). Figures are for the United States in 1986.
38. I discussed the consumption and savings behavior of the elderly in chapter 4.
 138 The Theory Elaborated and Applied

should expect them to engage in this behavior more than young people do.
An additional reason is that penny-pinching is a game (how much money
can I save), and its entertainment value is greater to people whose alterna¬
tive entertainments have been curtailed by age. It also provides an occasion
for demonstrating the retention of mental acuity.
It could be objected that a rational person who derives more utility
from a dollar when old than when young will through saving reallocate
dollars from his young to his old self until the utility of the marginal dollar
is equalized at all ages; this will maximize his total lifetime utility. But this
assumes that the young self, who controls the allocation of income over the
life cycle (the old self cannot do anything to reduce the income or con¬
sumption of his young self), will treat the welfare of the old self equally
with his own; and it is by no means certain that he will, for the reasons
explored in the discussion of the concept of multiple selves in chapter 4.

Sex
The Kinsey reports found that sexual activity declines in men from about
the age of 30 and in women from about the age of 40.^® More recent survey
data suggest that the age of onset of the decline, at least when measured by
the percentage of persons sexually inactive for the past year, is the fifties
for men but only the twenties for women. Although the percentage of sexu¬
ally inactive women remains below 20 percent until the fifties, by the late
fifties it exceeds 40 percent, compared to less than 16 percent of men.
Many elderly men and women continue to be sexually active, some to ex¬
tremely advanced ages. In the 65-69 age group, almost 80 percent of men,
and 40 percent of women, remain sexually active, while even in the 80-84
age group more than 40 percent of men do, though fewer than 10 percent
of women.40 As part of normal aging, sexual desire declines in both sexes,

39. Alfred C. Kinsey, Warded B. Pomeroy, and Clyde E. Martin, Sexual Behavior in the
Human Male 220-221 (1948) (tab. 44 and fig. 34); Kinsey et al.. Sexual Behavior in the Human
Female 548 (1953) (tab. 153). The Kinsey samples for elderly men and, especially, elderly women
are very small, however.
40. Edward O. Laumann et al.. The Social Organization of Sexuality: Sexual Practices in
the United States 88, 90, 92-93 (1994) (tab. 3.4 and fig. 3.1). The source for the data on persons
60 and older in The Social Organization of Sexuality is the General Social Survey. See note 18
above. The data for younger persons are from the National Health and Social Life Survey con¬
ducted by the University of Chicago. Unfortunately, that survey—the most comprehensive repre¬
sentative survey of the sexual behavior of the American people that has ever been conducted_
was limited to persons between the ages of 18 and 59.
For other estimates of levels of sexual activity of elderly people, see Ananias C. Diokno,
Morton B. Brown, and A. Regula Herzog, “Sexual Function in the Elderly,” 150 Archives c>f
Behavioral Correlates of Age 139

but rarely to zero.'" The ability to have an orgasm also declines, but far
more in men than in women. Why then are old men so much more likely
than old women to be sexually active? There appear to be two reasons.
Since women cease being fertile at a much earlier age than men, one ex¬
pects on genetic grounds, and finds, that on average (the vital qualification)
elderly women are less attractive sexually to men of all ages than elderly men
are to women of all ages.'*^ And there are far more elderly women than elderly
men because of the much greater longevity of women in modern societies.
The effect of these two factors (relative attractiveness and number) is
to make the “effective sex ratio,” which is to say the ratio of sexually avail¬
able and desired males to sexually available and desired females, very low
among the elderly. Among young people a low effective sex ratio is asso¬
ciated with male promiscuity, with low levels of rape and child sexual
abuse (because willing adult females are not in short supply), with a low
female marriage rate, and with a high rate of births out of wedlock.'*^
Among the elderly a low effective sex ratio has different implications, not
only because of the effects of aging on sexual desire and performance and
on fertility, but also because the sexual opportunities of the elderly are a
function in part of choices they made when young. Because of the costs,
relative to the benefits, of forming new relationships in old age (see chap¬
ter 3), we expect (and find) that the sexual activity of the elderly is highly
correlated with marriage, especially for women.'^'* This correlation is un¬
favorable to elderly women, because so many more elderly women than
elderly men are widowed or divorced. In 1989, 78.4 percent of men aged
65 to 74 were married and living with their spouse, while only 51.4 percent
of women in that age group were. And for the 85-and-over group the dis¬
parity was much greater—48.2 percent of men versus only 9.1 percent of

Internal Medicine 197 (1990); Judy G. Bretschneider and Norma L. McCoy, “Sexual Interest and
Behavior in Healthy 80- to 102-Year-Olds,” 17 Archives of Sexual Behavior 109 (1988);
Edward M. Brecher et al.. Love, Sex, and Aging: A Consumers Union Report (1984); John W.
Lorton and Eveleen L. Lorton, Human Development through the Lifespan 497-499 (1984).
41. For good discussions, see Arshag D. Mooradian and Vicki Greiff, “Sexuality in Older
Women,” 150 Archives of Internal Medicine 1033 (1990); David L. Rowland et al.. Aging and
Sexual Function in Men,” 22 Archivis of Sexual Behavior 545 (1993).
42 See, for example, Mary B. Harris, “Growing Old Gracefully: Age Concealment and
Gender,” 49 'journal of Gerontology P149, P156 (1994). Of course, there are exceptions. See
Lois W. Banner, In Full Flower: Aging Women, Power, and Sexuality: A History (1992). I am
offering merely a statistical generalization.
43. Richard A. Posner, Sex and Reason 136-141 (1992).
44 Stephen J. Weiler, “Aging and Sexuality and the Myth of Decline,” in Aging: Stability
and Change in the Family 317 (Robert W. Fogel et al., eds., 1981); Bretschneider and McCoy,
note 40 above, at 126-127.
 140 The Theory Elaborated and Applied

women.Some gerontologists have seriously proposed that polygynous

marriage be permitted to the elderly as a way of alleviating the shortage of
men.'** There is a bit of evidence that elderly women may turn to homo¬
sexual relationships (“opportunistic homosexuality”) because of the short¬
age of men, and more evidence that they turn to masturbation.'*^ Despite
these substitutions, it is apparent that the ratio of sexual activity of men to
that of women is higher among old than among young people.'*®
The picture changes somewhat if we turn from elderly male heterosex¬
uals to elderly male homosexuals. The traditional view was that the aging
male homosexual was a pathetic individual because of a cult of youth in
the homosexual subculture.'*^ This view has been challenged by more re¬
cent scholarship.*** Yet there is persuasive evidence that homosexual men,
like heterosexual men (to whom they are similar in most respects other than
sexual orientation), place a greater premium on youth than women do.*'
One consequence is that elderly homosexual men appear to have fewer
unpaid sexual opportunities than elderly heterosexual men.*^ Another is
that a higher percentage of elderly homosexual men than elderly hetero¬
sexual men live alone.**

45. U.S. Senate Special Committee on Aging et al., note 28 above, at 184 (tab. 6-1). For
more detailed statistics, see Jacob S. Siegel, A Generation of Change: A Profile of America’s Older
Population 300-311 (1993).
46. Diana K. Harris and William E. Cole, Sociology of Aging 234-235 (1980).
47. See Brecher et al., note 40 above, at 215; Catherine G. Adams and Barbara F. Turner,
“Reported Change in Sexuality from Young Adulthood to Old Age,” 21 Journal of Sex Research
126, 133-134, 139(1985).
48. Lorton and Lorton, note 40 above, at 499; Maj-Briht Bergstrom-Walan and Helle H.
Nielsen, “Sexual Expression among 60-80 Year Old Men and Women: A Sample from Sweden,”
27 Journal of Sex Research 289, 291 (1990); see also Adams and Turner, note 47 above. This is
particularly true in the oldest age groups; recall the figures on the relative percentage of sexually
active men and women in the 80-84 age group.
49. See, for example, John H. Gagnon and William Simon, Sexual Conduct: The Social
Sources of Human Sexuality 149-151 (1973).
50. See, for example, Mary Riege Laner, “Growing Older Male: Heterosexual and Homo¬
sexual,” 18 Gerontologist 496 (1978); Heather Gray and Paula Dressel, “Alternative Interpreta¬
tions of Aging among Gay Males,” 25 Gerontologist 83 (1985).
51. See, for example. Gray and Dressel, note 50 above, at 84-85; John Alan Lee, “What
Can Homosexual Aging Studies Contribute to Theories of Aging?” 13 Journal of Homosexuality
43,62(1987).
52. Cf. Brecher et al., note 40 above, at 225; Douglas C. Kimmel, “Life-History Interviews
of Aging Gay Men,” 10 International Journal of Aging and Human Development 239, 245 (1979),
Raymond M. Berger, Gay and Gray: The Older Homosexual Man 159-160, 164, 185 (1982),
confirms that older homosexual men (over 40) have limited social contacts with younger men.
53. Compare Jean K. Quam and Gary S. Whitford, “Adaptation and Age-Related Expecta¬
tions of Older Gay and Lesbian Adults,” 32 Gerontologist 367, 370 (1992) (more than 63 percent
of male homosexuals in sample of male and female homosexuals aged 50 to 73 lived alone), and
Behavioral Correlates of Age 141

The cost of forming new relationships in old age has other implications
for sexual activity. It may be a factor in the tendency that I noted earlier for
older men to substitute child sexual abuse for other forms of sexual crime
and also in their tendency to substitute masturbation for intercourse.^'^ The
broader point is that we should expect to observe among the unmarried
elderly some tendency to substitute “spot” for “relational” contracting in
the sexual “market” because of the high cost of forming new relationships.

Work and Leisure

The discussion in chapter 4 of the factors bearing on the timing of retire¬
ment was not exhaustive. Other important factors include the structure of
retirement benefits. For example, the receipt of social security retirement
benefits is, until the age of 70, contingent on the recipient’s not having
significant income from work. The effect is that of a heavy tax on the in¬
come of persons eligible for those benefits until they reach 70. Even with¬
out regard to the more favorable income-tax treatment of social security
benefits than of “earned” income, and the fact that there is no social secu¬
rity tax on them, a person who is earning $30,000 in a job when he could
retire and receive social security benefits of $20,000 is in effect paying a
two-thirds tax on his earned income. The lower the age of eligibility for
social security and the greater the benefits, the greater the incentive to retire
early. And people who retire at age 65 are unlikely to reenter the work force
five years later when they could do so without giving up any of their social
security benefits;” their work skills will have deteriorated.

Kimmel, note 52 above, at 242 (10 out of 14 male homosexuals between the ages of 55 and 81
lived alone), with the figure in the text that 78.4 percent of men aged 65 to 74 were married and
living with their spouses, implying that a maximum of 21.6 percent in this older (presumably
mostly heterosexual) group were living alone. (See also A. J. Lucco, “Planned Retirement Housing
Preferences of Older Homosexuals,” 14 Journal of Homosexuality 35, 50 [1987].) The actual
figure is undoubtedly lower than 21.6 percent, since some of these men must have been living with
someone other than a spouse. However, a smaller percentage of women in the homosexual sample
were living alone (41 percent) than the percentage of women in the 65 to 74 age group who were
not married and living with their spouse (48.6 percent). Since most people do not like to live alone,
this is consistent with other evidencetsee, for example, Posner, note 43 above, at 306-307) that
lesbians tend to be happier than male homosexuals. Another study of aged lesbians, however,
found that only 18 percent (9 out of the 50 in the sample) were at present in a “committed rela¬
tionship.” Monika Kehoe, “Lesbians over 65: A Triply Invisible Minority,” Journal of Homo¬
sexuality, May 1986, p. 139.
54. Simone de Beauvoir, Old Age 322-323 (1972); Lorton and Lorton, note 40 above, at
498-499.
55. See generally William J. Wiatrowski, “Factors Affecting Retirement Income,” Monthly
Labor Review, March 1993, p. 25.
142 The Theory Elaborated and Applied

But the effect of the social security program on the labor-force partici¬
pation of the elderly probably is swamped by the effect of private pension
plans, which generally create strong inducements to retire no later than at
65 and often substantially earlier.^^ It is true that, unlike social security
retirement benefits, private pension benefits (and for that matter most pub¬
lic pension benefits other than social security—for example military pen¬
sions and pensions of public school teachers, police officers, judges, and
other civil servants) are not conditioned on cessation of all work. They are
conditioned on cessation only of full-time work for the employer in whose
employ the benefits were earned. So we expect and find that some retirees
seek second careers. They are especially likely to do so if they retired be¬
fore they became eligible for social security retirement benefits. But these
careers are mostly in “dead-end” jobs.” The reasons should be obvious
from previous chapters. The decline in fluid intelligence with age makes it
costly for the older person to acquire new skills, and his benefits from doing
so are reduced because the payback period for any new investment that he
does make in his human capital will be truncated. (The metaphor “dead
end” is peculiarly apt.) Since dead-end jobs pay low wages, most elderly
people prefer the leisure of retirement even when their pension entitlement
does not (as it does in the case of social security) levy in effect a tax on
those wages.
This discussion has implications for the job security of the aging
worker. To the extent that his wage reflects an investment in his human
capital that is specific to his present employer, it may well exceed the wage
he could expect to obtain were he fired, since it is unlikely that he would
make a similar investment in a different firm. The difference is likely to be
especially great for the middle-aged worker, whose wage may be rising
because of experience so that he is still building human capital,^^ yet whose
experience may not be transferable to another employer because the human
capital he is building or renewing is firm-specific human capital. Until that
capital begins to depreciate more rapidly than it is being replaced, he will
be increasingly valuable to the firm. That is one reason he is able to com¬
mand a higher wage than he could hope for in another job. Another reason

56. Laurence J. Kotlikoff and David A. Wise, The Wage Carrot and the Pension Stick: Re¬
tirement Benefits and Labor Force Participation (1989). I discuss early-retirement incentives fur¬
ther in chapter 13.
57. Robert L. Kaufman and Seymour Spilerman, “The Age Structures of Occupations and
Jobs,” 87 American Journal of Sociology 827, 839 (1982).
58. Donald P. Schwab and Herbert G. Heneman III, “Effects of Age and Experience on
Productivity,” 4 Industrial Gerontology 113 (1977).
Behavioral Correlates of Age 143

is that as he gets older his ability and incentive to acquire human capital
specific to a different firm decline, and this makes him less likely to quit
and by quitting wipe out his present employer’s investment in him. Beyond
some point, however, the older he gets, the less valuable he becomes to his
present firm at the same time that he is becoming ever less employable
elsewhere, even though he may be years short of retirement. The precarious
position of the older worker makes it easy to see why unionization is gen¬
erally believed to be more attractive to older than to younger workers (see
chapter 3) and helps explain the push behind the age discrimination in em¬
ployment law, which I shall examine in detail in chapter 13.
Average age at retirement differs across occupations, and is one factor
in the different age profiles of different occupations.^® The more strenuous
or dangerous the occupation and the smaller the optimal investment in hu¬
man capital in the occupation, the earlier the age of retirement. Both points
are illustrated by military careers. Retirement is generally earlier for low-
ranking soldiers than for high-ranking officers. The former have by and
large the most strenuous and dangerous jobs, and, partly because it does
not pay to invest as much human capital in a worker who is likely to be
killed or disabled, their value ceases to grow significantly with experience
after a relatively short period of time. Senior officers, in contrast, have less
strenuous and dangerous assignments, so the age-decay curve is less steep;
and they employ more complex skills, requiring a larger investment in hu¬
man capital to bring to a peak. With military employment coming more
and more to resemble modern civilian employment (a trend aptly if exag¬
geratedly captured in the phrase “push-button warfare”), the traditional
system of retirement (half pay after twenty years of service) is increas¬
ingly being questioned.^'
The algorithm that the Social Security Administration uses to deter¬
mine whether applicants for social security disability benefits are totally
disabled conclusively presumes that an older worker with limited education
is totally disabled if he has a severe impairment, whereas if he were
younger additional evidence of disability would have to be considered or a
more severe impairment established. This makes good sense. An older
worker with limited general human capital (a limitation proxied by lack of
formal education) has especially poor employment prospects. Almost his

59. For an interesting discussion, with data, see Kaufman and Spilerman, note 57 above.
60. See Headquarters, Department of the Army, “Handbook on Retirement and Services for
Army Personnel and Their Families’ 5—1 (Pamphlet No. 600—5, Aug. 1, 1982).
61. See, for example, “Military Retirement: The Administration’s Plan and Related Propos¬
als” (American Enterprise Institute Legislative Analysis, 1980).
144 The Theory Elaborated and Applied

only human capital—specific human capital—is nonportable, and it is too

late for him to acquire new capital with a new employer.
Self-employed persons—but taxicab drivers and tailors more than law¬
yers and doctors—tend to retire later than employees.^^ A superficial expla¬
nation is that the self-employed are not “forced” to choose between full¬
time work and retirement, so they gradually reduce the amount of time they
work, thus postponing (full) retirement. But with the doubtful exception
(see chapter 13) of mandatory retirement at fixed ages, people retire be¬
cause they consider themselves better off not working, not because they are
dragged kicking and screaming from the workplace. And their decision to
retire has nothing directly to do with whether they are self-employed or
employed by someone else.
Another unconvincing explanation for later retirement by the self-
employed is that self-employment tends to attract risk-takers, and risk-
takers do not like a common form of retirement pension—the defined-
benefit plan, in which all investment risk is borne by the employer (see
chapter 12). But an employee who wants more risk in his financial assets
than a defined-benefit plan confers will usually be able to achieve his de¬
sired level of risk by borrowing more heavily than he would otherwise do
or by investing his savings in more risky vehicles than he would otherwise
find attractive.
The most convincing explanation for later retirement by self-employed
workers is that a reduction in hours worked is less costly to the worker (or
his employer) when he does not occupy expensive space, use expensive
equipment, or enjoy fringe benefits that are independent of the precise
number of hours worked. The lower these costs are, the less the hourly cost
of his work will rise as the number of hours he works falls. Consider a
worker who receives a wage of $10 an hour for a 40-hour week and uses
equipment that costs the employer $200 a week, making the hourly cost of
employing this worker $15 ($10 + $200/40). If he works only 20 hours a
week with no reduction in wage and no possibility of another worker’s
using the equipment the other 20 hours, the hourly cost of employing him
will rise to $20 ($10 + $200/20).
Self-employed professionals, such as doctors and lawyers, often have
expensive equipment the cost of which is independent of how many hours
they work. This will make their net income fall disproportionately to any
reduction in hours. In the previous example, if the self-employed profes-

62. Kaufman and Spilerman, note 57 above, at 837-838.

Behavioral Correlates of Age 145

sional’s gross income is $15 an hour (with $5 representing the cost of am¬
ortizing his equipment), that income will fall to $7.50 per hour if he reduces
his hours by half, and the result will be a 75 percent decline in his net
income (from $10 to $2.50), since his equipment cost is unaffected.
(Chapter 4 noted the positive correlation between hours worked and hourly
wage.) If the professional’s equipment, though expensive when purchased,
has little resale value and has been paid in full, the reduction in hours of
work will not be costly to him. But today most doctors and lawyers must
make substantial continuing outlays on books and equipment to remain
effective practitioners.
This analysis may explain why, even though their jobs generally are
less strenuous, self-employed professionals retire earlier on average than
self-employed tailors, barbers, and taxicab drivers, who can reduce their
hours of work without substantially reducing their hourly pay. Another
reason, however, may be the higher incomes of self-employed profes¬
sionals. And one group of professionals—^judges—constitute the most ge¬
riatric occupation in the United States; we shall seek an explanation in
chapter 8.

Residence
When there were no pensions, most retired people had no income, so they
had perforce to live with their children or other family members. This pat¬
tern of “coresidence” has changed dramatically. Most retired persons live
either by themselves or in nursing homes, though intermediate institutions
(retirement homes with nursing-home facilities available on demand) are
becoming increasingly common. The change from coresidence to indi¬
vidual residence is sometimes deplored as condemning people to a bitter
and lonely old age and depriving the young of the wisdom of the aged. But
it is important to distinguish between independent and institutional living.
The trend to the former is more convincingly understood as a voluntary
joint decision by children and parents than as the abandonment of the latter
by the former. For several reasons, the cost of coresidence has risen. With
old people living longer, the costs to children of providing them with hous¬
ing, food, companionship, and other goods and services are incurred for a
longer period. With family size shrinking, there are fewer children to share
the burden of aged parents; a fixed (and growing) cost is spread over fewer
people. And with the great expansion in women’s job opportunities, the
opportunity cost of household production, which includes taking care of
146 The Theory Elaborated and Applied

elderly relatives, a time-intensive activity, has been rising.^ At the same

time, the benefits of coresidence, both to the young and to the elderly mem¬
bers of families, have declined. Mass education, and rapid social and tech¬
nological change, have reduced the value to the young of elderly relatives’
wisdom.
Since the incomes of the elderly have been rising rapidly, it is likely
that if they wanted to live with their children they could reimburse them for
the full cost. They must not want to live with their children badly enough
to pay the price that would make the children indifferent between coresid¬
ence and independent residence. As just noted, in our rapidly changing
society the benefits of association between persons in different age groups
are reduced because of a dearth of common experiences. In addition, there
is a good deal of evidence that privacy is a superior good.^ As the incomes
of the elderly rise, they want more privacy and therefore value coresidence
less.^^ Eventually, infirmity forces many elderly people into nursing homes,
where there is little privacy. The immobility of the infirm forces them to
choose between companionship and privacy. Although the elderly value
privacy, the analysis in chapter 5 suggests that they value it less than the
young do, because the value of privacy to a person (other than a recluse) is
positively related to the person’s transactional activity.
“Older persons, both owners and renters, resist moving,” and as a re¬
sult frequently live in houses that are “too large” for their needs, or in
neighborhoods that are unsafe.^^ But this pattern makes perfectly good eco¬
nomic sense. Moving requires adaptation to new circumstances, and the
costs of such adaptation are higher for older than for younger persons. A
slightly less obvious point is that the benefits of a move tend to be lower

63. Cf. Kiyosi Hirosima, “The Living Arrangements and Familial Contacts of the Elderly in
Japan,” in The Elderly Population in Developed and Developing World 68, 75 (P. Krishnan and
K. Mahadevan, eds., 1992).
64. See my book Overcoming Law, ch, 25 (1995).
65. For empirical evidence, see Robert T. Michael, Victor R. Fuchs, and Sharon R. Scott,
“Changes in the Propensity to Live Alone: 1950-1976,” 17 Demography 39 (1980); Saul
Schwartz, Sheldon Danziger, and Eugene Smolensky, “The Choice of Living Arrangements by
the Elderly,” in Retirement and Economic Behavior 229, 243 (Henry J. Aaron and Gary Burtless,
eds., 1984); Jeffrey A. Burr and Jan E. Mutchler, “Nativity, Acculturation, and Economic Status:
Explanations of Asian American Living Arrangements in Later Life,” 48 Journal of Gerontology
S55 (1993); but see Fred C. Pampel, “Changes in the Propensity to Live Alone: Evidence from
Consecutive Cross-Sectional Surveys, 1960-1976,” 20 Demography 433 (1983), esp. p. 445. An¬
other reason for the decline in coresidence is discussed in chapter 9.
66. Siegel, note 45 above, at 574. The inertial character of the living arrangements of elderly
people is emphasized in Axel H. Borsch-Supan, “A Dynamic Analysis of Household Dissolution
and Living Arrangement Transitions by Elderly Americans,” in Issues in the Economics of Aging
89 (David A. Wise, ed., 1990).
Behavioral Correlates of Age 147

for old persons. The benefits will be received over a shorter period, and
therefore the costs of relocating are less likely to be fully amortized even if
they are no higher than for young persons. But as I have said, they are
higher.
I mentioned the loss of privacy in nursing homes. Another cost of in¬
stitutionalized living is “atmospheric.” The concentration of infirm elderly
persons, a high percentage of them senile, is depressing even when efforts
are made, as they often are, to segregate the most senile residents. It is no
surprise that income has been found to have a strong negative effect on the
probability of becoming a nursing-home resident, implying that living in a
nursing home is considered to be inferior to independent living.®^ The same
study found, after correcting for other factors, that the probability of be¬
coming a nursing-home resident is positively related to age (of course) and
negatively related not only to income but also to number of children and to
being married, and that independently of all these factors the probability
has been increasing. With longevity increasing and family size and the
marriage rate decreasing, and given the unrelated time trend just noted to¬
ward nursing-home residence—a trend that may reflect a growing disincli¬
nation of family to take care of elderly members^®—we can expect the
percentage of elderly people who live in nursing homes to continue to in¬
crease unless retirement incomes also continue to increase in real (that
is, inflation-adjusted) terms. Since nursing-home living is unpopular, we
might expect people to increase their savings for old age in order to re¬
duce the probability of such a destiny—but maybe not, as we shall note
in chapter 11, if they are not highly altruistic toward their future old,
infirm self.
In discussing the costs of nursing homes I may have seemed to over¬
look the most important cost—that of their large staffs. To include it, how¬
ever, would be implicitly to compare the cost of maintaining an elderly
person who is healthy enough to live by himself with the cost of maintain¬
ing an elderly person whose infirmities are such as to drive him into a nurs¬
ing home. If physical and mental condition is held constant, the cost of
institutionalized living is plainly less than the cost of living in a private
residence, since there are economies of scale in institutionalized living.
Indeed, were this not so, no one who was not extremely senile would
enter a nursing home, because of the loss of privacy and the atmo¬
sphere” costs. This point is important because it is the foundation of the

67. Borsch-Supan, note 66 above, at 102.

68. For the same reasons that coresidence is declining.
148 The Theory Elaborated and Applied

discussion in chapter 11 of what may be the most serious long-term prob¬

lem created by the aging of the population: the cost of home care, which is
much higher than the cost of institutional care because the economies of
scale are lost.

Voting and Jury Service

Voting in political elections is an important activity of elderly people. I
shall try to explain why, as shown in tables 6.1 through 6.4, the old have a
greater propensity to vote than the young.®®
The first two tables show that since 1986 a higher percentage of elderly
persons has been voting, despite infirmities and institutionalization, than
any other age group and that the percentage has increased since the early
1970s, while the voting percentages of all the other age groups have de¬
creased, although the changes are not great. The disparity in voting propen¬
sities of different age groups is particularly marked in off-year congres¬
sional elections (table 6.2), where we find 76.5 percent of the elderly voting
in 1990 compared to only 39.9 percent of the youngest eligible age group.
The greater propensity of elderly persons to vote persists when other demo¬
graphic variables are corrected for. A careful multivariate study concludes
that ''aging, by itself, produces not a decline but an increase in turnout.
The rate of increase in voting begins to level off at around age fifty-five but
turnout continues to rise, at an increasingly slower pace, through the
seventies.”™
What are we to make of this? Because the probability that a major
election will be decided by one vote is vanishingly close to zero, voting in
such elections is not realistically interpreted as an instrumental activity, an
investment. It is apparently a consumption activity,’' which I have com¬
pared elsewhere to applauding; people derive utility from expressing a

69. The sources of the data tor tables 6.1 through 6.4 are two papers by Jerry T. Jennings:
“Voting and Registration in the Election of November 1992” (Current Population Reports, Popu¬
lation Characteristics, Series P20-466, 1993) (tab. A, p. v, and app. A, pp. A1-A7), and “Voting
and Registration in the Election of November 1990” (Current Population Reports, Population
Characteristics, Series P20-453, 1991) (tab. B, p. 2, and app. A, pp. 77-82).
70. Raymond E. Wolfinger and Steven J. Rosenstone, Who Votes? 47 (1980) (emphasis in
original); see also G. Bingham Powell, Jr„ “American Voter Turnout in Comparative Perspective,”
in Controversies in Voting Behavior 56, 67-73 (Richard G. Niemi and Herbert F. Weisberg, eds
1993).
71. See, for example, Anthony J. Nownes, “Primaries, General Elections, and Voter Turn¬
out: A Multinomial Logit Model of the Decision to Vote,” 20 American Politics Quarterly 205
(1992).
Behavioral Correlates of Age 149

Table 6.1 Percentage of Each Age Group That Voted in Presidential Elections, 1964-1992

Age 1964 1968 1972 1976 1980 1984 1988 1992

18-24 50.9 50.4 49.6 42.2 39.9 40.8 36.2 42.8

25-44 69.0 66.6 62.7 58.7 58.7 58.4 54.0 58.3

45-64 75.9 74.9 70.8 68.7 69.3 69.8 67.9 70.0

65 + 66.3 65.8 63.5 62.2 65.1 67.7 68.8 70.1

Table 6.2 Percentage of Each Age Group That Voted in Off-Year Congressional Elections,
1966-1990

Age 1966 1970 1974 1978 1982 1986 1990

18-24 44.1 40.9 41.3 40.5 42.4 42.0 39.9

25-44 67.6 65.0 59.9 60.2 61.5 61.1 58.4

45-64 78.9 77.5 73.6 74.3 75.6 74.8 71.4

65 + 73.5 73.7 70.2 72.8 75.2 76.9 76.5

view on a performance or contest.’^ But it is not a costless consumption

activity, and herein, I believe, lies the clue to the greater propensity of el¬
derly people to vote.’3 principal cost—the time it takes to vote, and to
acquire minimum information about the candidates (akin to the time spent
in watching the performance that you are going to applaud at the end)—is
modest. But it is not negligible, especially if the candidates are obscure, as
they often are for the lesser offices. That is one reason why elections for
such offices often have very low turnouts, even though the instrumental
value of a vote is greater in a local than in a national election.
With time the principal cost of voting, we can formulate an economic
explanation for why elderly people are more likely to vote than younger
people, why the gap is greater in congressional than in presidential elections,
and why it has been growing. The opportunity cost of time is lower to elderly

72. Overcoming Law, note 64 above, ch. 3. The extensive literature on voter turnout is illus¬
trated by Steven J. Rosenstone and John Mark Hansen, Mobilization, Participation, and Democ¬
racy in America (1993); Controversies in Voting Behavior, note 70 above, pt. 1; Jan E. Leighley
and Jonathan Nagler, “Individual and Systemic Influences on Turnout: Who Votes? 1984,” 54
Journal of Politics 718 (1992); Political Participation and American Democracy (William Grotty,
ed., 1991); and references in note 75 below.
73. Wolfinger and Rosenstone, note 70 above, at 60, offer the alternative suggestion that the
greater propensity of old people to vote reflects the fact that “exposure to life in general and
politics in particular” increases people’s interest in voting.
150 The Theory Elaborated and Applied

people. Although their physical mobility may be restricted, and hence other
costs of voting besides the opportunity cost of a unit of time may be higher
to them than to the young,the political parties compete for their favor by
offering them free transportation to the polling place; and, increasingly,
polling places are being established in nursing and retirement homes.
The difference in the opportunity costs of voting of elderly compared
to young persons is greater, the less publicized and more obscure the elec¬
tion, because a greater investment in finding out about the candidates is
required. This may explain why the relative turnout of the elderly is even
greater in congressional than in presidential elections.’^
Voting is not only cheaper for older than for younger people, but also
more valuable. Not because the old have more to gain from the political
process; they may, but that is no reason for an individual old person to vote,
since his vote is not going to decide the election. Rather it is because the
old have a limited menu of activities to choose from. Not only work, but
also the more strenuous recreational activities, are largely barred to them.
This is not just another way of saying that the opportunity cost of a few
hours of their time every year or two is low. The point is rather that the
benefit derived from an activity is apt to be greater the fewer the alterna¬
tives. Voting, like penny-pinching, is a more exciting pastime to people
whose alternative pastimes are distinctly unexciting.
The greater interest that voting holds for the elderly, their low oppor¬
tunity costs of time, and a truncated horizon that tends to occlude their view
of issues not closely related to their own particular concerns (though altru¬
ism may, as we saw in chapter 3, lengthen their horizon) combine to make
the elderly a voting bloc unusually focused on a relative handful of issues.
This enhances their influence by making it difficult for politicians to win
their votes without promising to protect their interests and fulfilling their
promises. Although the percentage of the population that is elderly is not
always a good predictor of the economic welfare of the elderly, it is in the
political system of the United States.''^

74. One of these other costs is a time cost. If it takes longer for the old person to get to the
polls because of his physical infirmities, his time cost may be greater than that of the young person
even if the cost of time is less to him than to the young person per unit of time.
75. The suggestion that the costs of voting are lower in heavily publicized elections is sup¬
ported by the fact that turnout tends to be higher in such elections. John H. Aldrich, “Rational
Choice and Turnout,” 37 American Journal of Political Science 246, 266-268 (1993); Gary W.
Cox and Michael C. Munger, “Closeness, Expenditures, and Turnout in the 1982 U.S. House
Elections,” 83 American Political Science Review 217 (1989).
76. John B. Williamson and Fred C. Pampel, Old-Age Security in Comparative Perspective
116, 196, 220(1993).
Behavioral Correlates of Age 151

All this does not explain why the ratio of older persons’ propensity to
vote to younger persons’ has increased. One possibility is that the oppor¬
tunity costs of the time of older people have declined because a higher
percentage of older people are retired.
Since the percentage of the population that is 65 or older has been
growing at the same time that the percentage of that age group which votes
has been growing, the percentage of votes cast by the elderly has been
growing more rapidly than either statistic. This is shown in the next two
tables. By 1992, when 12.6 percent of the U.S. population was 65 or older,
they were casting about 50 percent more votes in the presidential election
than their percentage of the population (close to two-thirds more, in the
congressional elections of 1990). The disproportion is much smaller when
their voting percentage is compared with their percentage of the voting-age
population, as done in these tables. But the adjustment obscures the fact
that parents of minors are more or less adequate representatives of their
children’s interests, and few elderly persons have minor children. I come
back to this point in chapter 11.
Comparisons between the voting behavior of old and of young must be
interpreted cautiously because of the possibility of cohort effects. When
one compares 18-year-olds and 65-year-olds in 1992 one is not comparing
groups that differ only in age, being otherwise random draws from the same
population. They are different vintages—the class of 1974 (the year of
birth of the 18-year-olds) and the class of 1927—and have therefore had

Table 6.3 Votes Cast by Persons 65 and Over as Percentage of Total Votes, Presidential
Elections 1964-1992
(Percentage of Voting-Age Population in Parentheses)

1964 1968 1972 1976 1980 1984 1988 1992

15.4 14.9 15.8 16.8 17.7 19.4 19.0

14.9
(14.8) (15.0) (15.3) (15.7) (16.1) (16.6) (17.0)
(14.8)

Table 6.4 Votes Cast by Persons 65^ and Over as Percentage of Total Votes, Congressional
Elections 1966-1990
(Percentage of Voting-Age Population in Parentheses)

1970 1974 1978 1982 1986 1990

1966

16.6 17.0 18.5 19.1 21.1 22.0

16.0
(14.9) (15.1) (15.5) (15.9) (16.4) (16.9)
(14.8)
152 The Theory Elaborated and Applied

different experiences. This makes it difficult to assess the popular view that
older voters are more “conservative” than younger ones, an issue to which
I return in subsequent chapters.
The hypothesis that low costs of time and a paucity of other “entertain¬
ments” explain the high voter turnout of elderly persons may seem incon¬
sistent with the fact that turnout is low among the unemployed, even after
correction for other factors.^"^ The unemployed, it might seem, have plenty
of time, and a relative paucity of other distractions. But this is not true. The
term “unemployed” generally and in the relevant studies refers to people
who are looking for a job, rather than people who have dropped out of the
labor force. People looking for a job are busy—looking for a job; and are
likely to be preoccupied with that search to the exclusion of taking much
interest in the relatively remote realm of politics.’®
Jury service is another form of private participation in the political
process. Since such service is time-intensive and the elderly have lower
costs of time than younger people, we might expect them to be overrepre¬
sented on juries. It is widely believed that juries are dominated by retirees,
but on the contrary retirees appear to be underrepresented on juries. Al¬
though almost 12 percent of the U.S. adult population is 70 years old or
older, a study of federal and state juries in eight cities found that only
4 percent of jurors were that old.’^ Jury service requires greater mobility
and greater overall fitness, mental as well as physical, than voting does; so
we should not be surprised that a smaller percentage of the elderly serve on
juries than vote. Another reason is that most employers continue their em¬
ployees’ regular salary during (modest) periods of jury service, thus reduc¬
ing the opportunity costs to the young (who are more likely to be employed
than the old) of serving on juries.
Persons 60 or older accounted for 17.8 percent of the jurors in the
study, which is much closer to their percentage of the adult population
(23 percent) than in the case of the 70-and-over jurors, though still below
it. This is additional evidence that, as I have been emphasizing throughout
this book, there is indeed a pronounced age-related decline in physical and
(or) mental ability.

77. Rosenstone and Hansen, note 72 above, at 273 (tab. D-1), 282 (tab. D-5); Wolfinger and
Rosenstone, note 70 above, at 29. But see Wilma Smeenk, “Non-Voting in the Netherlands and
the United States” 10 (unpublished, Nijmegen University [Netherlands], Dept, of Sociology, n.d.),
finding that after correction for other factors unemployment does not reduce the likelihood of
one’s voting.
78. See Rosenstone and Hansen, note 72 above, at 81-82, 135.
79. Computed from Washington Project Office, “The Relationship of Juror Fees and Terms
of Service to Jury System Performance” D-1 (National Center for State Courts, March 1991).
Behavioral Correlates of Age 153

Both the preceding chapter and this one have been largely although not
entirely devoted to trying to explain puzzling phenomena concerning the
elderly, ranging from penny-pinching to loquacity. The preceding chapter
focused on paradoxes in attitude, disposition, and mind set, such as the
apparent negative relation between dread of death and expected utility. In
the present chapter the focus shifted to differences in behavior. The para¬
doxes here are numerous. Among them; high vehicular accident rates of
elderly people, both as injurers and as injured, and high suicide rates (a
form of violence in which injurer and victim coincide) coexist with low
rates of criminal conduct and victimization; the frequency of male sexual
activity relative to female sexual activity increases with age even though
male sexual performance deteriorates with age more rapidly than female
does; higher elderly incomes make the elderly less rather than more wel¬
come to live in their children’s households; and high voter turnout among
the old coexists with low juror turnout.
Age-related decline in vision, hearing, reflexes, and other neuromotor
capacities is the key to understanding the higher automobile accident rate
of older people. I argued that their driving safety declines less rapidly than
their underlying capabilities because it is possible up to a point to compen¬
sate at reasonable cost for deteriorating capabilities with greater care and
with a reduction in the amount of driving. (And this implies that such driv¬
ing as the elderly do is probably high-valued—a selection effect—albeit
dangerous.) It may be more costly to reduce “pedestrianizing” to the same
extent; this may explain why the age gradient of the pedestrian death rate
is even higher than that of the vehicle occupant’s death rate.*° And it prob¬
ably is much less costly to reduce exposure to situations in which one is
likely to become a victim of crime than situations in which one is likely to
be the victim of a traffic accident, which may explain the extremely low
rate of criminal victimization of the old. I attribute their low crime rate—
at first glance a paradox, because old age truncates the expected disutility
of a long prison sentence—not to feebleness but primarily to a low rate of
elderly entry into crime and a high rate of exit. The elderly avoid entering
the criminal “job market” from a lawful activity for the same reasons that
they avoid other career-switc'hing—the age-related cost of acquiring new
human capital—and age-related attrition is steep because of the heavy pun¬
ishment of recidivists. Selection bias is also at work. The most experienced
elderly criminals may elude arrest, and estimates of the crime rates of dif¬
ferent age groups are based primarily on statistics of arrest.

80. See note 9 above.

154 The Theory Elaborated and Applied

The aversion to risk and danger that marks elderly behavior with re¬
spect to accidents and crime is not inconsistent with the high suicide rate
of elderly men, once we realize that impulsive suicide is more common
among the young, and deliberated among the old.*' Among the young, sui¬
cide is part of a spectrum of risky behaviors. Among the old, it is a more
or less straightforward response to situations of negative expected utility,
once circular definitions of “depression” are rejected as explanations for
the high elderly suicide rate.
The key to understanding differences in the frequency of sexual behav¬
ior of elderly men and women is the concept of the effective sex ratio—the
ratio of sexually available and desired males to sexually available and de¬
sired females. The ratio drops with age both because male mortality is
higher than female and because the sexual attractiveness of men to women
declines less steeply with age than the sexual attractiveness of women to
men. I also argued that the rising cost of forming new friendships in old
age (discussed in chapter 3) will lead some unmarried elderly persons to
substitute “spot” fpr “relational” sexual contracting.
One might suppose that the secular growth in the incomes of elderly
people relative to the incomes of the young would result in an increase
rather than a decrease in “coresidence” (different generations living under
the same roof); the elderly would be in a better position to pay their chil¬
dren or other young relatives to defray the cost of the additional space and
services required. They are in a better position to pay for these things, but
they do not want them. Privacy is a superior good, so the demand for co¬
residence falls as elderly incomes rise. The cost of coresidence has been
rising in recent decades for the further reason that the opportunity cost of
female caretaking of elderly relatives has increased with the improvement
in the marketplace job opportunities of women. I also gave an economic
reason for why the tendency of many old people to remain in houses that
are “too big for them” is consistent with rationality.
The paradox that elderly people vote disproportionately to their share
of the adult population but are (contrary to popular belief) underrepre¬
sented on juries is explained by the different costs to elderly people of
these two forms of participation in civic life. Jury service requires greater
mobility and greater mental as well as physical fitness than voting, and
employers frequently pay their employees who are on jury duty, whereas
retired people on jury duty have no similar source of compensation and

81. For evidence, see Kalle Achte, “Suicidal Tendencies in the Elderly,” 18 Suicide and
Life-Threatening Behavior 55 (1988).
Behavioral Correlates of Age 155

therefore are less likely to be willing to serve.*^ The fact that the voting rate
of elderly people has been increasing may reflect the long-term decline
in the rate of participation by the elderly in the labor force. This decline
has reduced the opportunity costs of time spent both in voting itself and
in preparing for voting by obtaining information about candidates and
issues.*^
I also examined additional factors, besides those considered in earlier
chapters (especially chapter 4), bearing on the age of retirement. This en¬
abled me to offer an economic explanation for differences in the age of
retirement of different types of self-employed worker and for the peculiar
structure of military retirement, and to explore the effect of both social
security and private pensions on the age of retirement.

82. Although jury service is compulsory, people can and do get out of it either by not re¬
sponding to jury summonses (enforcement of such summonses being sporadic in most jurisdic¬
tions) or by responding to questions in jury questionnaires or in the voir dire of prospective jurors
conducted by the judge or the lawyers at the outset of trial in ways that cause them to be
disqualified.
83. This analysis implies a rising rate of jury participation by the elderly as well, but I do
not know of any time-series data on jury service by different age groups.
 7
Age, Creativity, and Output

Retirement is the most dramatic age-related change in work, but it is purely

quantitative. In this chapter I consider two other age-related changes in
work. The first is change from one kind of work to another, rather than from
work to leisure as in retirement. The second is change in the quantity or
quality of one’s work over the course of one’s working life. The second
type of change is related to retirement because it affects the value of the
worker to his employer and thus the date of retirement. But it is distinct,
not only because it can occur long before retirement but also because it can
be positive as well as negative. I shall be discussing both points—change
in the kind of work and change in the quality or quantity of work within
the same line of work—together. I shall therefore have much to say about
the different peak ages of productivity in different fields.

Productivity and Age: Creativity versus Leadership

I begin with an example of age-related change from one kind of work to
another. Academics at research universities, as they get older, often reallo¬
cate more and more of their working time from producing scholarship to
assisting in the administration of the university. Call this an age-related
substitution of “leadership” activity for “creative” activity. Because the
opposite pattern, in which an academic begins his career heavily involved
in academic administration and later reallocates time from administration
to scholarship, is almost never observed, it is unlikely that the age-related
substitution of leadership activity for creative activity is a matter of sorting.
156
Age, Creativity, and Output 157

Were it merely the case that when an academic was starting his career no
one could tell whether he was better at scholarship or at administration—
that it took several years of observation to determine which type of work
he was better at—there would be as many cases in which an academic
started in administration and later switched to scholarship as the reverse.
The key to understanding the phenomenon of age-related change from
one kind of work to another is that different kinds of work have different
age profiles of productivity. It has been suggested that the shift from schol¬
arship to administration reflects a decline in creative output caused by di¬
minishing investment in human capital over the life cycle.' But this cannot
be the entire explanation. The simple life-cycle model does not explain
why the investment profile for leadership human capital should be different
from that for creativity human capital. And the extensive sociological and
psychological literature on peak ages of creativity identifies in fields such
as mathematics and theoretical physics average peak ages that are too
low—in the thirties or even twenties—to be explained by the reduction in
investment in human capital that is induced by proximity to death.^ If we
ignore age-related decline, as the simple life-cycle model does, a 35-year-
old particle physicist would have an expected payback period of roughly

1. Arthur M. Diamond, Jr., “The Life-Cycle Research Productivity of Mathematicians and

Scientists,” 41 Journal of Gerontology 520 (1986); Sharon G. Levin and Paula E. Stephan, “Re¬
search Productivity over the Life Cycle: Evidence for Academic Scientists,” 81 American Eco¬
nomic Review 114 (1991); Daniel L. Rubenson and Mark A. Runco, “The Psychoeconomic Ap¬
proach to Creativity,” 10 New Ideas in Psychology 131 (1992). See also John M. McDowell,
“Obsolescence of Knowledge and Career Publication Profiles: Some Evidence of Differences
among Fields in Costs of Interrupted Careers,” 72 American Economic Review 752 (1982).
2. The major study of the age profile of creativity remains Harvey C. Lehman, Age and
Achievement (1953). The foremost contemporary contributor to the literature is Dean Keith Si-
monton. Among his many works, see especially Genius, Creativity, and Eeadership: Historiome-
tric Inquiries, ch. 6 (1984) (“Age and Achievement”); “Age and Outstanding Achievement: What
Do We Know after a Century of Research?” 104 Psychological Bulletin 251 (1988); Scientific
Genius: A Psychology of Science 66-68 (1988); and “Age and Creative Productivity: Nonlinear
Estimation of an Information-Processing Model,” 29 International Journal of Aging and Human
Development 23 (1989). See also Jock Abra, “Changes in Creativity with Age: Data, Explanations,
and Further Predictions,” 28 International Journal of Aging and Human Development 105 (1989);
Michael D. Mumford, Kimberly A. Olsen, and Lawrence R. James, “Age-Related Changes in the
Likelihood of Major Contributions,” 29 International Journal of Aging and Human Development
171 (1989); Simone de Beauvoir, Old Age 384-444 (1972); Wayne Dennis, “Creative Produc¬
tivity between the Ages of 20 and 80 Years,” 21 Journal of Gerontology 1 (1966), Evelyn Raskin,
“Comparison of Scientific and Literary Ability: A Biographical Study of Eminent Scientists and
Men of Letters of the Nineteenth Century,” 31 Journal of Abnormal and Social Psychology 20
(1936). Illustrative of this literature, a careful study of the productivity (as proxied by number of
publications in a refereed journal) of professors of finance found that the peak was in the late
thirties with an almost 50 percent decline by the fifties. Robert M. Soldofsky, “Age and Produc¬
tivity of University Faculties: A Case Study,” 3 Economics of Education Review 289 (1984).
158 The Theory Elaborated and Applied

40 years for any new investment that he made in his human capital. Even
at the age of 45 he would have more than a 30-year expected payback
period,^ and at ordinary discount rates there is only a small difference in
present value between a 30-year annuity and a 40-year annuity (or a 40-
year and a 50-year annuity, if we are comparing the incentives to invest of
a 25-year-old physicist and of a 35-year-old one), each yielding the same
annual income to the annuitant. At a discount rate of 10 percent, the pres¬
ent value of the 30-year annuity will be 96.4 percent as great as that of a
40-year one. Even at a discount rate of 5 percent, the present value of the
30-year annuity will be 89.6 percent of the present value of the 40-year
annuity.'*
It is true that, quite apart from any effect of aging on the ability to
develop or maintain intellectual skills, the opportunity costs of investing in
human capital will rise with hourly income, and hence (up to a point) with
age, since most investing in human capital is time-intensive. But although
a 35-year-old is therefore likely to have higher opportunity costs of invest¬
ing in human capital than a 20-year-old, early peaking implies that the op¬
timal investment in human capital in the activity in question probably is
modest. So a diminished incentive to continue investing, whether the dimi¬
nution is caused by rising opportunity costs or by the shortening of the
payback period, is unlikely to explain the decline of performance from the
peak. Physicians who specialize rarely complete their training before they
reach 30. The result is to push their peak age of earnings way back. Eields
with low age peaks are fields that, unlike medicine, require relatively little
training, with the exception of fields such as musical performance, figure
skating, and tennis, where training begins in childhood. The less human
capital (other than innate ability) required in a particular activity, and
hence the lower the cost of acquiring the necessary human capital, the less
sensitive will investment in it be to the proximity of death. An octoge¬
narian might invest in the knowledge required to play poker or bingo even
though the payback period for his investment would be short because of
his age.
If we are to explain differences in peak ages of productivity, both gen¬
erally and with specific reference to the difference in those ages between
leadership and creative work, we must, as in other chapters, move beyond
the simplest economic life-cycle model. The first thing to note is that the

3. The sum of age and remaining life expectancy increases with age, since every year lived
transforms a probability of surviving that year into a certainty.
4. The higher discount rate seems plausible, since investments in human capital are risky.
The lower discount rates used in chapter 4 were riskless rates.
Age, Creativity, and Output 159

very concept of a peak age of productivity is misleading in suggesting that

all careers have a sharp peak. There are careers with early peaks arid careers
with late ones, but also careers in which the peak, whenever attained, is
sustained without a significant decline virtually till death. Let us call these
“sustained peak” careers, as distinct from “early peak” careers and “late
peak careers.” Sustained-peak careers can in turn be divided into “early
peak, sustained” and “late peak, sustained,” thus giving us a fourfold di¬
vision: early peak, not sustained; early peak, sustained; late peak, not sus¬
tained; late peak, sustained. Examples of the first category (early peak, not
sustained) are most fields of professional athletics,^ along with mathemat¬
ics, theoretical physics,® chess,"^ heavy manual labor, and—the analysis in
chapter 6 implied—most criminal “careers.” In the case of physically de¬
manding activities, risk of injury plays a role; it is more difficult to sustain
peak performance in football than in dance.
Examples of the second category (early peak, sustained) are literature,
economics (other than the severely mathematical), musical composition
(including choreography), painting and sculpture (consider Michelangelo,
Titian, Picasso, and O’Keefe, among others), and musical performance. An
example of the third category (late peak, not sustained) is the senior man¬
agement of large firms, where the peak age will often be in the late fifties,
followed by retirement in the early sixties;* perhaps most leadership is in
this category. The fourth category (late peak, sustained) is illustrated by
judging, discussed in the next chapter. History, theology, literary criticism
and scholarship, and philosophy appear to straddle the second (early peak,

5. See, for example, Neil Charness and Elizabeth A. Bosman, “Expertise and Aging: Life
in the Lab,” in Aging and Cognition: Knowledge Organization and Utilization 343, 369-374
(Thomas M. Hess, ed., 1990). A careful study of professional baseball players finds their peak age
to be 27, with a decline in performance from the peak of 22 to 24 percent by age 30, 84 percent by
age 36, and 99.5 percent by age 40. Bill James, The BillJames Baseball Abstract 1982 196 (1982).
6. The natural sciences in general are early-peak fields. A telling statistic is the average age
at which Nobel-prize-winning scientists did the work for which they received the Nobel prize:
37.5. Only 5.6 percent of the Nobel prize winners were over the age of 50 when they did their
prize-winning work. Paula E. Stephan and Sharon G. Levin, Striking the Mother Code in Science.
The Importance of Age, Place, and Time 55 (1992) (tab. 4-2). As for math, here is what a distin¬
guished mathematician, G. H. Hardy, Had to say: “Mathematics, more than any other art or science,
is a young man’s game. To take a simple illustration at a comparatively humble level, the average
age of election to the Royal Society is lowest in mathematics.” Hardy, A Mathematician’s Apology
70-71 (1940).
7. Charness and Bosman, note 5 above, at 352-360.
8. The average chief executive officer of a major U.S. corporation is 56 or 58 and plans to
retire at 64. Sunita Wadekar Bhargava, “Portrait of a CEO: What’s the Typical Boss Like? Here
Are the Vital Statistics,” Business Week, Oct. 11, 1993, p. 64; Michael J. McCarthy, A CEO s
Life: Money, Security and Meetings,” Wall Street Journal (midwest ed.), July 7, 1987, p. 27.
160 The Theory Elaborated and Applied

sustained) and fourth (late peak, sustained) categories. The fourfold matrix
is displayed in table 7.1 with a few examples.
A more complete analysis would distinguish not only among different
fields but also among different activities within the same field. Within the
field of professional athletics, for example, performance has an early peak,
but coaching does not. There is a close parallel in academia: research gen¬
erally has an earlier peak than teaching, and the rate of decline from the
peak is also faster than for teaching. Even more refined distinctions are
possible: for example, between teaching advanced subjects and teaching
basic ones. Since the heartland of a field changes less rapidly than the fron¬
tier, teaching the heartland requires less investment in acquiring new hu¬
man capital. Older teachers are therefore more likely to teach basic than
advanced subjects. When, as is common in some fields, older and younger
scholars collaborate, the younger tend to bring mathematical and other
techniques to the collaboration, the older the insights enabled by experi¬
ence. Such a collaboration marries fluid and crystallized intelligence.
To explain early peaking with greater precision, we need to recall
figure 4.1, in which mental and physical capability to perform a task was
correlated with age, negatively so in the region to the right of the peak age
(m in that diagram), the region, that is, of the downward-sloping curve that
relates capability to age (d). Human capital in the conventional sense, that
is, leaving out innate ability, is only one factor that affects the location and
shape of that curve. This point is obvious if we consider an early-peak
activity such as professional basketball. Peak age is low in such an activity
in part because the investment in human capital (including experience, an
important source of human capital) required to reach peak proficiency ei¬
ther is small or is concentrated in one’s very early years, in part because the
activity involves abilities that age rapidly for purely biological reasons, and
in part because minimum capability is very close to peak capability. Or
consider chess. Serious players generally start to play before the age of ten.

Table 7.1 Careers with Different Age Profiles

Early Peak Late Peak

Not Sustained Mathematics, Corporate management

Basketball

Sustained Painting, Judging

Musical composition
Age, Creativity, and Output 161

capability

Figure 7.1 Activity with low peak age

Although their speed (important in tournament chess, where there are time
limits on moves) begins to decline in their twenties, at first this decline is
offset by growing experience, which even after the player reaches his peak,
usually in his early thirties, will retard the net rate of decline in perfor¬
mance until his sixties, when most players stop participating in tournament
play.^
Early peaking is depicted in figure 7.1. The proximity of r (minimum
required capability) to m (peak capability) identifies the hypothetical ac¬
tivity depicted in the figure as an elite one. Elite activities are more chal¬
lenging; that is, they draw on more of a person’s ability than routine ones
do. If an elite activity does not require a large investment in human capital
and if the biological abilities on which it draws peak at an early age, the
decline in capability can be precipitous, as in many sports. In academic
fields in which these conditions obtain, we can expect substitution into ad¬
ministration early in a scholar’s career—provided of course that adminis¬
tration does not have the same age profile. It does not. For example, Harvey
Lehman found much later age peaks in activities (including political and
judicial office) that he classified as leadership positions.'® Yet it is important
to note that a nonelite activity can have a very low age peak, too, if, even
though r lies far below m, the age-decline curve is extremely steep, as in
heavy manual labor, where the worker wears out rapidly.
Why should leadership have a later age peak than creativity? Here are
five mutually compatible possibilities. First, the ratio of fluid to crystallized
intelligence is higher in creative tasks, and we know that fluid intelligence

9. Charness and Bosman, note 5 above, at 354-360.

10. Lehman, note 2 above, at 286 (tab. 50).
162 The Theory Elaborated and Applied

ages more rapidly than crystallized intelligence. Although leadership usu¬

ally requires some problem-solving abilities, much of it (especially in the
kind of leadership dubbed charismatic by Max Weber, as distinct from
more routine management) consists of evaluating, matching, and motivat¬
ing other people. These are interpersonal skills not well correlated with
problem-solving ability. They may even in some cases be negatively cor¬
related with it. Superior intelligence can be an impediment to effective
leadership if the gap in intelligence between leader and followers is so wide
as to retard mutual understanding."
Second, effective leadership often requires having a wide network of
acquaintances—persons one can trust as aides or allies. Such a network, an
important form of relational human capital discussed in chapter 3, tends to
grow with age, and there may be very little depreciation until retirement is
imminent.
Third, experience is an important input into most forms of leadership.
Being the knowledge that accrues from living and working, as opposed to
reading or studying, experience grows with age, though in some activities
only up to a point: after a point early reached in living or working, one does
not become better at tying one’s shoe laces or tightening a bolt on an as¬
sembly line. If experience is more valuable in leadership than in creative
activity, it is easy to see why the peak age for leaders might be later than
that for scientists or poets.
Fourth, most people mature with age, though only up to a point. I am
speaking of a process that is different from the accumulation of experience;
that is emotional as much as cognitive; and that is largely complete by the
onset of middle age and may actually reverse itself in old age (“second
childhood”) even in people who do not become senile—for many elderly
people who are not senile are labile, self-centered, excessively stubborn,
and rather simple-minded. The process of maturation that I am describing
involves more than anything the shucking off of childish traits, traits such
as selfishness, stubbornness, disregard of convention, and “ask[ing] ques¬
tions that adults usually have stopped asking.” Though inimical to most

11. Dean Keith Simonton, “Land Battles, Generals, and Armies; Individual and Situational
Determinants of Victory and Casualties,” 38 Journal of Personality and Social Psychology 110,
112 (1980); Ralph M. Stogdill, Handbook of Leadership: A Survey of Theory and Research 43-
45 (1974). This is one reason why Felix Frankfurter, contrary to expectations, failed to play an
effective leadership role in the Supreme Court.
12. Floward Gardner, “The Creators’ Patterns,” in Changing the World: A Framework for
the Study of Creativity 69, 76 (David Henry Feldman, Mihaly Csikszentmihalyi, and Howard
Gardner, eds., 1994); see also Gardner, Creating Minds: An Anatomy of Creativity Seen through
the Lives of Freud, Einstein, Picasso, Stravinsky, Eliot, Graham, and Gandhi 365-366 (1993).
Age, Creativity, and Output 163

forms of leadership, these traits may actually foster creative activity in the
arts and sciences.'^
Fifth, leadership implies responsibility. An academic who is a flop as
teacher or scholar can be fired (unless he has tenure), and little harm has
been done. Even if he has tenure and cannot be fired, his ineptitude will do
relatively little harm. But a leader who is a flop may cause serious harm to
many other people; a bad leader can wreck a successful organization. It is
therefore natural to require that he demonstrate his competence in advance
by success at the next lower rung of the administrative ladder, and the ne¬
cessity of climbing the ladder makes it likely that senior leaders will be of
mature age. To put this differently, the costs of failed leadership tend to be
greater than those of failed creativity, implying the need for a more careful
screening of leaders than of creative workers. One dimension of care is
minimization of the adverse consequences of an incorrect assessment, as
by requiring that candidates for a top leadership position rotate through a
series of progressively more responsible assignments rather than being se¬
lected for the position on the basis of performance at the bottom of the
ladder. Careful screening of leaders may, as a result, require many years.
The suggested contrast between the costs of failure as a leader and as
a creative worker may be somewhat overdrawn. The uncreative researcher
may discourage promising young people from entering his field, vote
against the advancement of the able, or be occupying a place that a creative
researcher might have occupied instead. But the first two consequences
have limited significance as long as the uncreative researcher, and others
like him, are outnumbered by able colleagues. And as for the displaced
creative researcher, he probably will find a different position and make as
great or almost as great a contribution as he would have done in the posi¬
tion from which he was blocked. The failed leader does more harm than
merely blocking an opportunity for a superior leader to emerge.
There is a puzzle still. Why must some things be learned by doing
rather than by studying, given that study takes so much less time? In the
case of “physical” tasks (using the word loosely) such as riding a bicycle,
or mixed physical-mental tasks such as driving a car, or even such “purely”
mental tasks as learning to speak a foreign language, the answer is that the
task, to be done right, requires a degree of speed in responding to stimuli
that can be achieved only by habit-inducing drill. These are areas of what
philosophers call “tacit” knowledge—knowledge, largely unconscious,
inarticulate, and incommunicable, of how to do things (including how to

13. As argued by Gardner in the works cited in the preceding footnote.

164 The Theory Elaborated and Applied

produce and comprehend sentences) as distinct from knowledge of rules,

algorithms, and other propositions. Instruction may help a little in these
areas but it cannot just take the form of issuing the trainee a set of in¬
structions that he can use to accomplish the task, in the way that he could
use a set of instructions on how to assemble a bookcase to assemble the
bookcase.
Most leadership positions do require rapid decision-making. Not being
able to make up one’s mind, which is to say being slow at making decisions,
is a common cause of failure in leadership. The ability to make responsible
decisions rapidly comes, usually, only with experience. But this is not the
complete solution to the puzzle. The reason that physicists peak early is not
that they operate without a substantial knowledge base but that they can
absorb the essential parts of the base by a relatively brief course of reading,
practice, and being instructed."* Why could not a social scientist, through
careful study of the writings of political and management theorists, psy¬
chologists, politicians, and business leaders, assemble a course of readings
that would impart to students all the principles of effective leadership, so
that after a brief course of study the student would know everything that
seasoned politicians and managers knew? It is only a partial answer that
institutions differ in significant detail.'^ That Just implies having to supple¬
ment the general course of instruction in the principles of leadership with
a handbook applying them to the particular institution (university, business
firm, nation, or whatever) in which the student wanted to make his career
of leadership. Once the handbook was memorized, the fledgling leader
could make confident decisions rapidly.
Why does that seem a quixotic suggestion? Literally, of course, one
person’s experience cannot be “transferred” to another person; one’s ex¬
perience is by definition what has happened to oneself. And while even a
lifetime’s experience can be summarized or narrated in readable form, the
summary will not provide much in the way of useful guidance to a different
person who is at an earlier point in his own life cycle. Why there should be
such a tremendous loss in transmission—why some forms of knowledge

14. Cf. Harriet Zuckerman and Robert K. Merton, “Age, Aging, and Age Structure in Sci¬
ence,” in Aging and Society, vol. 3; A Sociology of Age Stratification 292, 302-306 (Matilda
White Riley, ed., 1972), noting that the various sciences differ in the degree to which they are
“codified”—where “codification refers to the consolidation of empirical knowledge into succinct
and interdependent theoretical formulations”—and that “experience should count more heavily
in the less codified fields.” Id. at 303. Law, as we shall see in the next chapter, is one of those less
codified—that is, incompletely theorized—fields.
15. This is Aristotle’s point that practical wisdom requires knowledge of particulars. See
chapter 5.
Age, Creativity, and Output 165

should depend on the knower’s own experience rather than on his knowl¬
edge of the experience of his predecessors—is a deep mystery, though one
that the field of artificial intelligence may eventually dispel (or transcend).
It is not enough to say that leadership is one of those activities about which
we do not know enough to lay down rules and are therefore left to grope
our way as best we can by employing analogies to past experiences. For
why could not all the potential analogies, or at least the most important
ones, be neatly laid out in a book for the young leader to consult when he
came upon a new problem?
I offer several suggestions, though it may seem that to the extent they
dispel the central mystery they merely bring other mysteries into view.
First, for reasons not well understood it is difficult to learn how to deal with
people effectively from books, and the essence of leadership is dealing
with people effectively. Second, effective leadership of a particular group
(nation, firm, university, whatever) often requires that the leader know a
number of specific people well, and this will take time. This is a point about
relational human capital again (see chapter 3), but here my emphasis is on
the fact that by its nature it cannot be sold or otherwise transferred. I do not
put too much emphasis on the point, however, because leaders are often
appointed from other institutions rather than from the ranks of the institu¬
tion they are to lead.
Third, leadership like warfare is often though not always a strategic
activity in the game-theoretic sense: every move invites a countermove, so
the leader who merely repeats the moves of previous leaders is easily
thwarted because predictable. Fourth, a point that carries beyond leader¬
ship, the ability to choose the apt analogy from the array in front of one
may require “judgment” based somehow on experience rather than on the
mastery of some analogy-selecting algorithm. (But there is reason to ques¬
tion this point, as we saw in chapter 5 and will see again later in this chap¬
ter.) Fifth, there may be genetic reasons, having to do with optimal family
structure, and age-grading, in the evolutionary period of human develop¬
ment (see chapter 9), why it is “natural” for younger persons to look for
leadership to older ones. I suggest one of those reasons below.
I do not want to exaggerate. There is instruction, not all of it wasted,
in techniques of management. And not all leaders have been old. Alexander
the Great and Napoleon Bonaparte are conspicuous counterexamples, but
ones easy to explain. Both were military leaders at a time when successful
military leadership required considerable daring and fitness, and both as¬
cended to power in settings in which thd usual sorting process was infea¬
sible. Alexander became King of Macedonia because he was the king’s son.
166 The Theory Elaborated and Applied

and Napoleon became a general at a time when the ordinary process for
promotion had been disrupted by the French Revolution. Both, moreover,
were charismatic leaders, although Napoleon had, and Alexander seems
rather to have lacked, considerable managerial skills as well. Successful
management normally requires patience, self-control, and willingness to
suffer fools gladly, qualities more common among middle-aged than among
young people because age tends to boil away tempestuous emotion, though
uncommon among the elderly too because such qualities require the sup¬
pression of ego.

Differences in Peak Age between and within Fields

We saw earlier that within the class of creative fields (broadly defined, to
include for example engineers as well as scientists) there are large differ¬
ences in the age profile of productivity. Some of these differences can be
explained in terms of the different rates at which fluid and crystallized in¬
telligence decline with age, or of the different levels of required investment
in human capital. Mathematics and physics (especially theoretical physics)
involve both high ratios of fluid to crystallized intelligence and low levels
of investment in human capital. The stock of knowledge they employ can
be conveyed through books and instruction economically, with relatively
little learning by doing required. As the cliche has it, each generation of
scientists can see farther because it is standing on the shoulders of giants.'^
It is no surprise that these are early-peak, not-sustained, fields. Nor is it sur¬
prising that experimental scientists have a later peak than theoretical
ones. Not only are the demands on their fluid intelligence somewhat less,
but experimentation involves team work, hence some people skills and
therefore leadership in a broad sense. Most modern economists employ
formal modeling skills that depend heavily on fluid intelligence. But expe¬
rience of social life is important too, and it accrues with age, so that econo¬
mists lacking frontier skills may still be able to do frontier work. Some
modern economists, such as Ronald Coase and Thomas Schelling, have
managed to make important contributions to economics without any use of
formal models. Economics thus appears to be an early-peak, sustained,
field.
In architecture and engineering, the visualization of objects in their
spatial relations is an important skill. This is an element of fluid intelli-

16. Michael D. Mumford, “Age and Outstanding Achievement: Lehman Revisited,” 25

Journal of Vocational Behavior 225, 231 (1984).
Age, Creativity, and Output 167

gence, and one that has been found specifically to decline with age.’’ So it
is not surprising that architecture and engineering are early-peak, not-
sustained, fields, though less so than more mathematical fields.'* Invention
(which is related to engineering) seems, surprisingly, to be a late-peak
field.'® This should allay concerns that the aging of the population threatens
to reduce the rate of inventive activity, and with it the economic growth that
is due to that activity. The threat is slight in any event, since the rate of
inventive activity depends on the number of inventors rather than on the
fraction of the population that is “age eligible” to engage in the activity,
and that number should grow as the population grows.
History and literary criticism are fields that traditionally have required
large investments in human capital. Practitioners have to master large
bodies of text or other information that cannot be reduced to a handful of
principles. And aspects of crystallized intelligence, in particular skills of
exposition, play a much larger role in these fields, relative to fluid intelli¬
gence, than is true in scientific and social scientific fields. Expository
skills—not only writing skills narrowly conceived but also the organizing
skills required in the composition of books as distinct from articles—tend
to improve with experience for a very long time, and then plateau rather
than sink.
Lehman found lyric poetry to be an early-peak field. This may seem
surprising. It is, of course, a branch of writing, although it involves less
organization, and therefore less experience, than scholarly writing, epic
poetry, or fiction (novels and short stories) because the unit of composition
is so much smaller. One explanation that has been offered for the early
peaking of poetic ability is that it is positively correlated with psychiatric
illness, which is likely to disable the poet at an early age.^" But so many
lyric poets have not suffered a creative decline with age (familiar examples
are Yeats, Stevens, Whitman, and Frost) that the field should probably be
classified with fields in which early peaks are sustained. Such a classi-

17. Timothy A. Salthouse, “Age and Experience Effects on the Interpretation of Ortho¬
graphic Drawings of Three-Dimensional Objects,” 6 Psychology and Aging 426 (1991).
18. Paul R. Sparrow and D. R. Davies, “Effects of Age, Tenure, Training, and Job Com¬
plexity on Technical Performance,” 3 Psychology and Aging 307 (1988); Dennis, note 2 above, at
2-4.
19. Id. at 3 (tab. 2). See also Naomi Stewart and William J. Sparks, “Patent Productivity of
Research Chemists as Related to Age and Experience,” Personnel and Guidance Journal, Sept.
1966, p. 28. Another study, however, finds a significant positive correlation between the percentage
of young adults in the population and the number of patents awarded. Mumford, Olsen, and James,
note 2 above.
20. Dennis, note 2 above; Raskin, note 2 above, at 29-30.
168 The Theory Elaborated and Applied

fication would be consistent with the centrality of writing skill to poetic

achievement and the dispensability of substantial life experience to the
writing of poetry, as distinct from the writing of novels, where we would
expect the peak to be reached later but, again, sustained indefinitely.
Musical creativity, both in composition and in performance, follows a
pattern similar to that of poetry—early-peak sustained. Why this should be
so in the case of musical performance is unclear. A concert pianist or vio¬
linist is engaged in an elite activity, that is, one in which required capability
is close to peak capability; and the capability in question has a physical
dimension. One might expect, therefore, that even if the age-related decline
in the requisite physical skills were slight, aging concert performers would
be quickly supplanted by equally able youngsters. Yet there are innumer¬
able examples of pianists, violinists, other soloists, and conductors who
continued to perform with distinction and undimmed popularity to a very
old age. (Think of Horowitz, Heifetz, Casals, and Toscanini.) Maybe ex¬
perience can compensate for the relatively slight physical decline. Music is
richly expressive of emotion, and a person’s emotional repertoire changes
with age.
An additional point, though relevant to the continued popularity rather
than to the performing quality of famous elderly performers, is that a per¬
son who achieves celebrity becomes an object of interest independent of
his current capability. People will go to see rather than to hear him. This
point suggests, incidentally, an explanation that does not depend on the
costs of information for the frequency with which a good reputation, per¬
sonal or institutional, persists long after the qualities that originally gave
rise to it have disappeared. A person or institution once prized for real
capacities or achievements continues to be prized as a memento. People
wanted to meet Winston Churchill even in his dotage, and they still visit
the tomb of Napoleon.
The sustained peak in philosophy may seem a mystery, since the ratio
of fluid to crystallized intelligence in the activity of philosophers is, or at
least might seem to be, very high. The peak is early, and not sustained, for
philosophers whose work borders on science or mathematics (Bertrand
Russell for example). But Wittgenstein had peaks in his twenties (when he
was doing logic) and in his fifties; and many philosophers, ranging from
Plato and Kant to Dewey and Sartre, and, among the living, Willard Quine,
Donald Davidson, John Rawls, Hilary Putnam, Stanley Cavell, Nelson
Goodman, Judith Jarvis Thomson, Bernard Williams, Richard Rorty, and a
number of others, have remained highly creative in their sixties or even
later. To complicate the picture further, a number of distinguished philos-
Age, Creativity, and Output 169

ophers have had late peaks, as distinct from early peaks sustained; ex¬
amples are Kant, Rawls, and Goodman.
I suggest two explanations for the fact that age seems to take only a
light toll of nonmathematical, nonlogician philosophers. The first is that
literary skills are far more important in philosophy than in, say, mathe¬
matics. The distinction of many philosophers, including Plato, Wittgen¬
stein, Nietzsche, James, Quine, and Rorty, is owed in no small part to those
skills. Metaphors (“language game,” “veil of ignorance,” “cash value,”
“tribunal of sense experience,” etc.) and other striking turns of phrase
(such as Kuhn’s “normal science” and “paradigm”), neologisms (“grue”),
parables (Plato’s cave, Neurath’s boat, “turtles all the way down”), dia¬
logues (Plato again), and even poetry (Lucretius, Nietzsche) have been em¬
ployed in philosophy to striking effect.
I am not comfortable with this point. Although metaphor is usually
conceived to be a verbal skill, it is similar to analogy,^' which plays an
important part in scientific discoveryand seems, therefore—this power
of discovering the similar in the dissimilar—work for the fluid intelli-
gence.23 Yet while scientific reasoning, presumably including the scientist’s
use of analogy as a tool of discovery, is an example of abstract reasoning—
reasoning that abstracts from particulars in its search for universal laws—
or at least of a form of cognition seemingly far removed from the literary,
metaphor is usually thought to illustrate, even to typify, the concreteness of
literary expression.^'^ So metaphor and analogy seem both like and unlike
each other. Maybe the explanation for the difference between the correla¬
tions of age with use of metaphor and with use of scientific analogy is that
while both devices draw on the fluid intelligence, metaphor is also a verbal
phenomenon and so draws on skills that tend not to decline with age and
may even improve with it.

21. See, for example, D. A. Boswell, “Metaphoric Processing in the Mature Years,” 22
Human Development 373, 382 (1979). In the words of a famous literary critic, “The essence of
poetry is metaphor and metaphor is finally analogical rather than logical.” Cleanth Brooks, The
Well Wrought Urn: Studies in the Structure of Poetry 248 (1947).
22. See, for example, John H. Holland et al.. Induction: Processes of Inference, Learning,
and Discovery 289-295 (1986); Mark T. Keane, Analogical Problem Solving 11-18 (1988);
Brenda E. F. Beck, “Metaphors, Cognition, and Artificial Intelligence,” in Cognition and Sym¬
bolic Structures: The Psychology of Metaphoric Transformation 9, 15 (Robert E. Haskell, ed.,
1987). Beck, indeed, treats analogy as a form of metaphor.
23. There is evidence that skill in analogical reasoning is indeed one of the intellectual ca¬
pacities that declines markedly with age. See Douglas H. Powell (in collaboration with Dean K.
Whitla), Profiles in Cognitive Aging, ch. 4 (1994).
24. See, for example, W. K. Wimsatt, Jr., The Verbal Icon: Studies in the Meaning of Poetry
79-80(1954).
170 The Theory Elaborated and Applied

A second reason for the sustained peaking of philosophers but not of

highly mathematical scientists is that philosophy is less progressive than
the scientific disciplines. Because the problems addressed by philosophers
and the analytic tools used to solve them change much less rapidly than in
the case of physics or mathematics, philosophers’ human capital depreci¬
ates less rapidly than scientists’ and therefore requires less new investment
to maintain it. The decline of their fluid intelligence is also less hampering
if they do not have to address new problems but can continue worrying the
old ones, for the ratio of fluid to crystallized intelligence employed is
higher the newer the problem being addressed is to the person addressing it.
We can model some of these points by expressing output at age a (Oa)
as a function of the peak-age output (Op), the rate at which output declines
from the peak (d),^^ and the number of years that has elapsed since the
peak age (a — p), as in

0, = 0p[(l - d)‘’-p]. (7.1)

So if, for example, the peak is reached at age 25, the individual whose
current output we are interested in measuring is now 30 (so that a — p =
5), and his output declines by 2 percent a year (making 1 — d .98), his
current output will be roughly 90 percent of his peak output.
If a = p, that is, if we are evaluating output at the peak age,
equation 7.1 reduces to = Op. If p> a, meaning that the individual has
not yet reached his peak age, then the rate of decline of output is negative,
so the equation becomes O^ = Op[(l + dy-p], or equivalently (since now
a — p < 0), OJOp = 1/(1 + d)p ~ The right side of this equation is
larger than zero and smaller than one, which shows that pre-peak output,
just like post-peak output, is below peak output.
An alternative formulation is possible, based on the equation in
chapter 3 relating earnings (or output) to time: E(t) = a + bp-b2f
(equation 3.1). If t* is the peak year, and y the number of years that the
individual is past the peak, then the amount by which his output in the peak
year will exceed his output in the current year (E) is approximated by

E = y[(2t* + y)b2-bT (7.2)

Recall that b2 is depreciation (here, age-related decline) and is invest¬

ment in human capital (for present purposes, experience). The greater the

25. A constant rate of decline is assumed for simplicity. A more realistic assumption would
be that the rate increases with age.
Age, Creativity, and Output 171

rate of decline, and the less important continued experience is, the higher
will the excess of peak over current output be. And unless experience is
terribly important, it will be higher the older the individual is.^®
Suppose that the production of the individual’s output requires three
inputs (besides time and effort, which I will ignore): problem-solving
ability, writing ability, and some “installed base” of information and tech¬
niques. Then d (in equation 7.1) or (in equation 7.2) will be an average
of the rate of decline or depreciation of these inputs, weighted by their
relative importance in the production function. Suppose the average scien¬
tist combines in fixed proportions 1 unit of problem-solving ability and
1 unit of knowledge base (with only negligible amounts of writing ability),
and that both depreciate (net of any new investment) at a high rate—
problem-solving ability because it is an aspect of fluid intelligence and the
knowledge base because of rapid advances in scientific knowledge. Sup¬
pose that the average philosopher, in contrast, combines 1 unit of problem¬
solving ability, which depreciates at the same rate as the scientist’s, with
1 unit of writing ability, which does not depreciate at all, and with 1 unit of
knowledge base which depreciates very slowly because the problems that
philosophers address are not changing and the corpus of significant philo¬
sophical literature is growing very slowly. Then the rate of decline of the
philosopher’s output as a function of age will be much lower than the scien¬
tist’s. It may even be negative if writing ability increases with age or if
experience is a large component of philosophical human capital, as it may
well be in such incompletely theorized areas (compared to logic) as politi¬
cal and moral philosophy. A related consequence if experience and writing
skill are important to the philosopher’s output will be to postpone the peak
age of creativity; and, other things being equal, this will also postpone the
age at which the ratio of current to peak output falls to the point where the
individual is no longer making significant contributions. He will remain
productive longer.
As the ratio of current to peak output falls, the opportunity set facing
the creative worker changes. Suppose that the best journals in some scien¬
tific field will not publish articles that reflect a capability of less than
90 percent of the peak capability of average practitioners, and Dr. Y is
average. When his capability falls below 90 percent of his peak, he will no

26. An admittedly odd implication of equation 7.2 is that E will be higher the older the peak
age (/*). The reason is that in the model which underlies this equation earnings increase at the
same rate with every year of experience until the peak earnings year is reached, so the longer it
takes to reach the peak the higher the peak earnings will be relative to current earnings.
172 The Theory Elaborated and Applied

longer be able to publish in such journals. He may respond by publishing

in lower-quality journals.^'^ Since there is some substitutability between
quality and quantity—though I am assuming not enough to enable Dr. Y to
continue publishing, albeit at a lower rate, in the best journals—maybe he
will write more articles for the lower-quality journals than he previously
wrote for the best ones.^* Or, borrowing a leaf from the philosopher’s book,
he may decide to concentrate his research efforts on old problems rather
than on new ones; or he may decide to become a popularizer rather than a
creator of science; or he may decide to collaborate with a younger scholar.
Whichever of these courses he chooses, he will be making less use of his
fluid intelligence.^®
Another reason for the aging scientist to adopt a conservative research
agenda is that a bold agenda might bring into question the vahdity of his
earlier work, the work on which his reputation is based.A young scientist
by challenging received wisdom threatens the reputation of other scientists,
not his own reputation, for he has none. It is otherwise for the older scien¬
tist. But I doubt that this is an important influence on the research of older
scientists. If one’s earlier research is unsound this is bound to be discovered
sooner or later, and if you are your own unmasker you will get credit for
courage as well as for intelligence. Trying to protect a reputation created
by one’s early work should be distinguished from the resistance to new
ideas that is caused by age-related decline in fluid intelligence and by dis¬
investment in human capital.
The incentive of the older scientist to concentrate on old problems is
paralleled by the incentive of the younger to concentrate on new ones. It is
not only that the younger scientist has a superior ability to deal with new
problems (or with old problems in new ways) because he has up-to-date
training and undiminished fluid intelligence. It is also that he would be at a

27. He could be thought to be “retiring” from his primary career and taking up a less de¬
manding form of work as a second career.
28. A. M. Diamond, Jr., “An Economic Model of the Life-Cycle Research Productivity of
Scientists,” 6 Scientometrics 189, 194 (1984).
29. Arthur M. Diamond, Jr., “Age and the Acceptance of Cliometrics,” 40 Journal of Eco¬
nomic History 838, 839 (1980). Diamond argues that these adaptations to aging may explain why
older scientists are slower to adopt new theories than younger scientists, although his own research
has found that the negative effect of age on the acceptance of new scientific theories has been
exaggerated. Id. at 839, 841; Arthur M. Diamond, Jr., “The Poly water Episode and the Appraisal
of Theories,” in Scrutinizing Science 181 (A. Donovan et al., eds., 1988); David L. Hull, Peter D.
Tessner, and Arthur M. Diamond, “Planck’s Principle: Do Young Scientists Accept New Scientific
Ideas with Greater Alacrity than Older Scientists?” 202 Science 111 (1978). We shall note a pos¬
sible explanation for this surprising finding in the last section of this chapter.
30. See, for example, Rubenson and Runco, note 1 above, at 141.
Age, Creativity, and Output 173

comparative disadvantage in following the groove planed by the older gen¬

eration—that is, in using established techniques to deal with familiar prob¬
lems—because he would lack the experience of the older scientist. He is
therefore led to pursue problems, or use methods, for which the older scien¬
tist’s experience is not an asset. So, even within the same field, there is a
generational division of labor; and this is why intergenerational collabora¬
tion can be fruitful.
I suggested that the aging scientist might decide to reallocate research
and writing time to administration.^’ Here I add that this tendency will be
accelerated if his accumulated experience enables him to obtain a high
salary in an administrative post. By raising his potential earnings in an
alternative activity (administration), experience increases the opportunity
cost of his remaining a researcher. Another thing that increases that cost
is—success. Successful scientists and other successful creative people are
invited, sometimes even badgered, to give prestigious lectures, accept hon¬
orary degrees, serve on boards and committees, consult, advise, write
popular papers, give memorial addresses, appear on television, write letters
of recommendation, and so forth, and to the extent that they yield to these
importunings, as most of them do (because these activities produce psychic
and sometimes pecuniary income), they have less time for research. A par¬
tially offsetting factor is that the successful, prominent person is in a better
position to obtain criticisms from other able people; indeed, his success,
his prominence, may make him a target for criticism—which he can learn
from, if he does not dismiss it as a product of envy. But that is what he is
apt to do. I am led to predict that creative people who remain obscure
throughout their lifetime will reach their creative peak later than those who
are successful in their lifetime (of course the causality could run both
ways—some people are obscure throughout their lifetime because they are
very late bloomers).
On similar grounds one might expect applied economists to peak ear¬
lier than mathematical economists. The former have better consulting op¬
portunities, which increase with age (an academic needs an academic repu¬
tation before he can “cash in” as a consultant). A countervailing
consideration, however, is that the applied economist uses more experience
and less fluid intelligence in his work than the mathematical economist,
implying a later peak and lower rate of decline. Another reason why the
rate of decline probably is lower for the applied economist is that his

31. For evidence, see Soldofsky, note 2 above, at 296; Zuckerman and Merton, note 14
above, at 318-321.
174 The Theory Elaborated and Applied

knowledge base depreciates more slowly, given the rapid advances in

mathematics. Cutting back the other way again is that the experience ac¬
quired by the applied economist may be more transferable to administrative
tasks.
A final observation is that people may tend for a variety of reasons to
exaggerate the rate at which elders’ skills erode. One reason is that the cost
of acquiring a new skill may exceed its benefit to someone who possesses
the old skill for which the new one is a replacement, even though the new
skill is an improvement over the old one and the cost of acquiring it, even
to an older economist, is not great. Suppose that mathematical economists
develop—this is happening all the time—a more compact and elegant
method of formulating some problem, yet the older method is adequate.
Young economists will be trained in the new method; older economists will
not think it worthwhile to acquire it because, the older method being ade¬
quate, the incremental benefit from the acquisition would be slight to them.
They will continue in the old way, and hence will strike the younger gen¬
eration as old-fashioned. But if the old method is indeed adequate, the
ability of the older economist to compete with younger ones may not be
adversely affected by his refusal to learn the new technique. Age-related
decline is an important theme of this book, but economic analysis can help
us identify some areas where that decline is not at work.

Changes with Age in the Character of Creative Work

We should consider not only age-related decline in the quality of creative
work but also age-related quality-independent changes in the character
of creative work. In such fields as painting and sculpture, musical compo¬
sition, and law—fields in which age-related declines in quality tend to be
slight, zero, or even negative—as well as in the careers of exceptional
poets, such as Yeats, who remained unabatedly productive until his death,
there is nevertheless change—a tendency, dubbed the Altersstil (old-age
style), toward boldness, clarity, and directness, and away from artifice.
Dean Simonton remarks the “concise directness” of the late works of the
composers in his large sample,^^ and Martin Lindauer finds support for
“those descriptions of the late-life style which emphasize a holistic per-

32. Dean Keith Simonton, “The Swan-Song Phenomenon: Last-Works Effects for 172 Clas¬
sical Composers,” 4 Psychology and Aging 42, 45 (1989); Simonton, Greatness: Who Makes
History and Why 208-209 (1994).
Age, Creativity, and Output 175

spective, a broader brush stroke, and an unconcern over detail,” while a

sixteenth-century Chinese remarked of an earlier Chinese artist, Ni Tsan,
that “in his old age he followed his own ideas, rubbed and brushed and was
like an old lion, walked alone without a single companion.” Sophocles,
Yeats, Verdi, Stravinsky, Richard Strauss, Leonardo da Vinci, Michelan¬
gelo, Titian, Bernini, Matisse, and Oliver Wendell Holmes, Jr.^^ illustrate
the tendency in their different fields.
There are plenty of skeptics about the Altersstil?^ But I take it that
there is some such tendency of the kind described above, enough to invite
an effort at explanation. Economics suggests two possibilities. First, the
diminishing value of transactions as the end of life nears reduces the cost
of outraging an audience’s expectations, as by blunt speaking. This point is
only superficially inconsistent with the remarks in chapter 5 about the ten¬
dency toward loquacity in elderly people: bluntness and loquacity are
merely different examples of speech that violates the hearer’s expectations.
Inconsiderate communication may reveal more about the speaker than the
more careful and often more reticent speech typical of younger people, who
are concerned with maximizing their transactions. Since blunt, straight-
from-the shoulder speaking or writing can be a social good, this suggests
that the problem of the “last period” (see chapter 3) can have an up side;
we shall see another example of this in the next chapter.
Second, fame may confer license. If a person has done valued work in
the past, this increases the probability that his current work is also valuable
and induces the audience to suspend its disbelief. He can therefore afford
to thumb his nose at the crowd.^^ This is merely the obverse of the “shame-

33. Martin S. Lindauer, “Creativity in Aging Artists: Contributions from the Humanities to
the Psychology of Old Age,” 5 Creativity Research Journal 211, 223 (1992); see also id. at 216.
34. Quoted in Jerome Silbergeld, “Chinese Concepts of Old Age and Their Role in Chinese
Painting, Theory, and Criticism,” Art Journal, Summer 1987, pp. 103, 105. For other examples of
the literature on the Altersstil, see Beauvoir, note 2 above, at 404-406; David Gutmann, “Age and
Leadership: Cross-Cultural Observations,” in Aging and Political Leadership 89, 93-94 (Angus
McIntyre, ed., 1988); Symposium on “Old-Age Style, ” Art Journal, Summer 1987, p. 91.
35. Another example from law is Judge Learned Hand. His biographer remarks the “ex¬
treme, stark position” that Hand took in the lectures on the Bill of Rights that he wrote and deliv¬
ered at the age of 86—“a more extreme position than he had taken earlier.” Gerald Gunther,
Learned Hand: The Man and the Judge 665 (1994). More on Hand in the next chapter.
36. See, for example, Catherine M. Soussloff, “Old Age and Old-Age Style in the ‘Lives’
of Artists: Gianlorenzo Bernini,” Art Journal, Summer 1987, pp. 115, 119; Julius S. Held, “Com¬
mentary,” Art Journal, Summer 1987, pp. 127, 129; Avis Berman, “When Artists Grow Old:
Secure in the Mastery of Their Craft, They Can’t Stop ‘Looking for a Breakthrough,’ ” Art News,
Dec. 1983, pp. 76,81.
37. See Beauvoir, note 2 above, at 488-492.
176 The Theory Elaborated and Applied

lessness” of the old, which Aristotle discussed. Peter Messeri argues in this
vein that “senior scientists are better situated than younger scientists to
withstand adverse consequences of public advocacy of unpopular posi¬
tions,” and that this factor may explain why the tendency for older scien¬
tists to resist new theories is, in fact, weak.^® And remember Kenneth
Dover’s negative verdict on old age (chapter 5)? He offered one qualifica¬
tion: “There just aren’t any [aspects of old age which compensate for its
ills]—except, maybe, a complacent indifference to fashion, because people
no longer seeking employment or promotion have less to fear.”^®
This point suggests that the use by scholarly journals of blind referee¬
ing is a mistaken policy. It may cause them to turn down unconventional
work to which they would rightly have given the benefit of the doubt had
they known that the author was not a neophyte or an eccentric."*®
Finally, as fluid intelligence declines, the cost of complexity rises,"*'
inducing the creative worker to substitute toward less complex forms of
creation and expression."*^ It might seem that if there is value in simple
works, the younger creative worker would produce them even though he
was capable of creating more complex works. But it may be difficult to

38. Peter Messeri, “Age Differences in the Reception of New Scientific Theories: The Case
of Plate Tectonics Theory,” 18 Social Studies of Science 91, 96 (1988).
39. Kenneth Dover, Marginal Comment: A Memoir 243 (1994). But why must it be
complacent?
40. I am not aware that this hypothesis has been tested in the literature on the effects of blind
refereeing. The most systematic study that has come to my attention finds very few effects of
blind (technically, “double-blind” -“single-blind” just means that the author is not told the refer¬
ee’s identity) refereeing. Rebecca M. Blank, “The Effects of Double-Blind versus Single-Blind
Reviewing: Experimental Evidence from The American Economic Review” 81 American Eco¬
nomic Review 1041 (1991).
41. “The tendency for the magnitude of age differences in cognitive performance to increase
with the complexity of the task” has been well documented. Timothy A. Salthouse, Theoretical
Perspectives on Cognitive Aging 308 (1991); see also D. B. Bromley, “Aspects of Written Lan¬
guage Production over Adult Life,” 6 Psychology and Aging 296, 306-307 (1991); Leah L. Light,
“Interactions between Memory and Language in Old Age,” in Handbook of the Psychology of
Aging 275, 285 (James E. Birren and K. Warner Schaie, eds., 1990); James E. Birren, Anita M.
Woods, and M. Virtrue Williams, “Behavioral Slowing with Age: Causes, Organization, and Con¬
sequences,” in Aging in the 1980s: Psychological Issues 293, 303 (Leonard W. Poon, ed., 1980);
Patricia K. Alpaugh and James E. Birren, “Variables Affecting Creative Contributions across the
Adult Life Span,” 20 Human Development 240 (1977).
42. An extreme example is the “unembarras.sed reductiveness” of the painting of William
de Kooning after he became senile, a style that one critic has called, apparently without irony, “the
senile sublime.” Quoted in David Rosand, “Editor’s Statement: Style and the Aging Artist,” Art
Journal, Summer 1987, pp. 91, 92. Of Matisse’s late work it has been remarked that “in his sev¬
enties and eighties, when his hands became too crippled to hold a brush, he produced wonderful
collages made of cut-out paper shapes, working in a simplified style, all excess stripped away.”
Hugo Munsterberg, “The Critic at Seventy-Five,” Art Journal, Spring 1994, p. 64.
Age, Creativity, and Output 177

make simple works that are of high quality, so that the worker economizes
by producing complex works until no longer capable of doing so.
Yeats made my essential points in his poem “The Coming of Wisdom
with Time,” which ends, “Through all the lying days of my youth / I
swayed my leaves and flowers in the sun; / Now I may wither into the
truth.” Yeats’s late poetry is candid, unadorned, bawdy, strident, violent in
diction, and often celebratory of actual violence. It goes to the heart of the
matter (as he sees it, rightly or wrongly) without any effort to conciliate
readers who might be shocked at the old man’s “lust and rage.” The ex¬
ample of Yeats shows, incidentally, that simplicity should not be equated
with popular appeal. The stripping away of artifice, the abandonment of
decorum, may shock a public accustomed to artifice and decorum.
Admittedly Yeats may be a special case among lyric poets. He was a
great poet in his youth and yet in the judgment of most critics his poetry
got continuously better until sometime in his fifties (“Easter 1916,” written
in 1916, when he was 51, surpasses any poem that he had written previ¬
ously), at which point it reached a plateau that was sustained until his death.
The character of the poetry of Yeats’s late sixties and his seventies was
different from that of his earlier poetry but not inferior. In part he main¬
tained the quality of his poetry into his old age by making old age a subject
of poetry; and while we know that young people can write feelingly and
insightfully about old age, presumably it is a subject in which old people
have a comparative advantage that can offset, up to a point anyway, an age-
related decline in ability. Another example is Oedipus at Colonus, Sopho¬
cles’ great play of old age, written shortly before his death at the age of
90.^^ Resisting the ravages of age by making age one’s subject is not an
option available in all fields. But Simonton gives the example of a chemist
who, having switched to gerontology in his nineties, published his last sci¬
entific paper at the age of 102!^*^
I have not suggested, and do not believe, that the “old-age” or “late-
life” style is characteristically superior to the artist’s or other creative work¬
er’s prime-of-life style; and this raises the question why some creative
people keep going, while others retire (E. M. Forster for example). Of

43. See generally Yeats: Last Poems: A Casebook (John Stallworthy, ed., 1968); also Doug¬
las Archibald, Yeats 233-235 (1983).
44. There is a story, though it is not well authenticated, that one of Sophocles’ sons tried to
get his father declared incompetent and that Sophocles refuted the charge by reciting Oedipus at
Colonus, which he had just written, to the court. Cicero, De Senectute, ch. VII, §§ 22-23; F. J. H.
Letters, The Life and Work of Sophocles 53-54 (1953).
45. Dean Keith Simonton, “Creativity in the Later Years: Optimistic Prospects for Achieve¬
ment,” 30 Gerontologist 626, 627 (1990).
178 The Theory Elaborated and Applied

course part of the reason is simply that people age at different rates. Yet
even in the case of someone who aged slowly, one might suppose that long
before his abilities fell below some minimal level the rising costs to him of
continued work would cross the declining benefits to him of continued
work. Creative people tend to be people who derive enormous psychic in¬
come from possessing or even from hoping to possess posthumously a
large reputation. The more one has done, the less contribution one more
work or several more works, likely to be of somewhat diminished and in
any event no greater quality than one’s previous works, will make to one’s
reputation; and there is a cost in forgone leisure, as well as, in the usual
case, a higher cost of production because of the effects of age. Hence one
expects that the creative workers who persist in working into their old age
will be those who have limited retirement income, derive psychic income,
unrelated to reputation, from working, disvalue leisure, have aged very
slowly, or have aged from a very high peak (so that their residual ability is
likely to be high).
I find the leisure disvaluers particularly interesting, and will give two
examples. The first is Justice Holmes, who in a radio address on the occa¬
sion of his ninetieth birthday said:

The riders in a race do not stop short when they reach the goal.
There is a little finishing canter before coming to a standstill.
There is time to hear the kind words of friends and say to one’s
self: “The work is done.’’
But just as one says that, the answer comes: “The race is over,
but the work never is done while the power to work remains.”
The canter that brings you to a standstill need not be only
coming to rest. It cannot be while you still live. For to live is to
function. That is all there is in living.'^^

This is the essential Calvinist outlook (all there is in living is the perfor¬
mance of duty), and hence of course not peculiar to Holmes. The passage I
have quoted from his radio address could almost have been a paraphrase of
a passage from Tennyson’s “Ulysses”: “How dull it is to pause, to make
an end, / To rust unburnished, not to shine in use! / As though to breathe
were to live.”

46. Holmes, “Radio Address (1931),” in The Essential Holmes: Selections from the Letters,
Speeches, Judicial Opinions, and Other Writings of Oliver Wendell Holmes, Jr. 20-21 (Richard
A. Posner, ed., 1992).
Age, Creativity, and Output 179

The essential findings of this chapter can be summarized very briefly.

Fields of human endeavor, ranging from poetry to burglary and from theo¬
retical physics to military generalship, can be divided into four classes de¬
fined by their peak age of productivity and by the rate at which productivity
declines from the peak. The classes are early peak, not sustained; early
peak, sustained; late peak, not sustained; and late peak, sustained. I used
the economic model developed in earlier chapters—especially chapter 4,
which emphasizes age-related decline along with familiar features of the
economic model of the life cycle—to fix the location of a number of fields
in the fourfold matrix and also to explain characteristic career switches
within fields, such as from academic research to academic administration,
and different peaks within fields (the obscure versus the prominent practi¬
tioner, and the theoretical versus the applied economist). As in the example
of the academic’s switch from a creative to a leadership job, I laid particular
stress on the difference in age profiles between leadership and creativity, a
difference that turns in part on the rather mysterious incommunicability of
life experiences. Finally, I pointed out that the economic model can be used
not only to explain differences in rates of aging across different fields but
also to provide an account—one that emphasizes both last-period “irre¬
sponsibility” (in a good sense) and decline of fluid intelligence—of the
distinctive “old-age style” found in some artists.
 8
Adjudication and Old Age

The judiciary is the nation’s premier geriatric occupation. But it must not
be considered in isolation from the rest of the legal profession. The differ¬
ent branches of the profession are interdependent. For example, precisely
because judicial creativity or achievement continues to a later age than is
the case in most academic fields, a major determinant of the peak age of
creativity of law professors can be expected to be the degree to which the
law professor’s work is closer to that of a judge, on the one hand, or to that
of a nonlegal academic, on the other. Especially but not only at elite law
schools, recent decades have seen a pronounced shift in academic scholar¬
ship from the judicial to the academic model, implying an earlier peak for
academic legal careers. Cutting the other way, however, is the fact that the
judicial model of legal scholarship was a better preparation for administra¬
tive work than the academic model is. It is remarkable how many university
presidents and provosts are former law professors, many removed from
scholarship at an early age because of the demand for their administrative
skills. Maybe the peak age of academic legal creativity is indeed falling but
the rate of decline from the peak is also falling because fewer legal aca¬
demic careers will be truncated by entry into academic administration.
The practice of law is in general a “late peak, sustained” activity, al¬
though the strenuous demands of a trial practice make litigation more of a
late-peak, not-sustained, activity, and some young lawyers have had no¬
table success in trials because jurors like young lawyers. As a consequence
of the incompletely theorized nature of law as an activity, experience plays
a larger role in successful lawyering than abstract reasoning does. The fol-

180
Adjudication and Old Age 181

lowing summation of studies that find significant age-related decline in the

power of abstract reasoning would be embraced by many an elderly lawyer
or judge as a favorable description, on the whole, of himself: “The old
person is more literal, more concrete, more concerned with tangible and
immediate impressions, less able to detach himself from the particular ex¬
ample and consider the general class or principle, less able to ignore the
individual fact in order to think in hypothetical terms.”' In addition, the
difficulty that even sophisticated corporate clients encounter in assessing
lawyers’ performance makes reputation created over many years an impor¬
tant asset. Since appellate advocacy is a more theoretical activity than trial
practice, negotiation, or counseling, we should expect (and I have ob¬
served) that appellate advocates are on average younger than other lawyers.

How Productive Are Older Judges?

Article III of the U.S. Constitution appears to forbid imposing a mandatory
retirement age on federal judges. Yet this is only an intermediate and in¬
complete reason why the average age of American judges is so much
higher than that of other professionals. The framers of the Constitution
could have imposed an age ceiling; and Congress could if it wanted lure
judges off the bench with generous early-retirement offers, as private em¬
ployers often do even when they are constrained by a law forbidding man¬
datory retirement or other “discrimination” on account of age, as we shall
see in chapter 13. Any employer can pay an employee to quit. And states
can and do fix retirement ages for their judges, but state judges are not
youngsters either.^
The remarkable thing about judges, moreover, is not that they hang on
to their jobs to such advanced ages but that they perform them creditably,
and indeed sometimes with great distinction, at advanced ages. This fact is
masked nowadays by the prevalence of judicial ghostwriting—the delega¬
tion of opinion writing to law clerks—which is enabling a small number
of senile judges, and a significant number of judges who are well past their
prime though not yet senile (merely “senescent”), to continue in office.
But long before law clerks were a significant factor in the judiciary, judges
such as Holmes, Brandeis, Hughes, and Learned Hand performed with dis-

1. D. B. Bromley, The Psychology of Human Ageing 189 (1974), quoted in Timothy A.

Salthouse, Theoretical Perspectives on Cognitive Aging 276 (1991).
2. See, for an admittedly unusual example, Roger M. Grace, “Court of Appeal Presiding
Justice Lester Wm. Roth, 96, to Retire Oct. 15,” Metropolitan News-Enterprise (Los Angeles),
Oct. 4, 1991, p. 1.
182 The Theory Elaborated and Applied

tinction into their eighties; and even after some judges started leaning
heavily on clerks, others, who did not, such as Felix Frankfurter and Henry
Friendly, turned in distinguished performances in their late seventies, and
in the case of Friendly early eighties as well. It is not unusual for lawyers
to be appointed to the federal appellate bench who are in their early sixties
and have had no previous judicial experience (a distinguished recent ex¬
ample is Guido Calabresi), yet who are realistically expected to perform
creditably in their new career for at least fifteen years. This was true even
before the era of law clerks and geriatric fitness. Holmes and Cardozo were
over 60 when appointed to the Supreme Court, though both had judicial
experience; and while Cardozo, because of ill health, lasted only a few
years on the Court, Holmes performed with great distinction for almost
30 years, though he faded some toward the end. Most lesser judges, as we
shall see, also continue to be productive well beyond the normal retirement
age. Finally, although the delegation of responsibilities to law clerks is full
of problems, it is also a method of compensating, in part anyway, for age-
related deterioration in a judge’s ability. Simonton points out that in a num¬
ber of fields, including painting and science, the elderly creative worker
can “adjust to this effect of aging [i.e., the age-related slowing in the rate
of processing information] by judicious use of assistants.”^ Why should
this not be true in judging?
The impression that the quality of judging is not highly sensitive to age
is just that, an impression, based on a handful of examples. I want to try to
be more systematic by using as a proxy for judicial quality the average
number of judicial citations to a judge’s opinions. It is a rough proxy, with
many drawbacks, but at least when it is used only to compare judges within
the same court system in the same era it has sufficient validity to be servi¬
ceable for my purposes here.'* Table 8.1 counts the average number of ci¬
tations in published decisions of the federal courts of appeals to published
cases decided in different periods by different cohorts of federal court of
appeals judges. Specifically, the population from which the cited cases are
drawn consists of the signed, published majority opinions of federal court
of appeals judges who were still sitting in 1993 but who had been appointed
between 1955 and 1984.^ As the average age of appointment was virtually

3. Dean Keith Simonton, “Creativity in the Later Years: Optimistic Prospects for Achieve¬
ment,” 30 Gerontologist 626, 627 (1990).
4. For evidence and discussion of the validity of number of citations as a tool of judicial
evaluation, see Richard A. Posner, Cardozo: A Study in Reputation, ch. 5 (1990).
5. All federal court of appeals judges appointed in this interval who were still sitting in 1993
were included in the sample. The sample was created for purposes unrelated to this book by a
Adjudication and Old Age 183

Table 8.1 Average Number of Citations to Judicial Opinions of Different Age Cohorts

Cohort
Period
of Case 1955-59 1960-64 1965-69 1970-74 1975-79 1980-84
Cited (n = 6) (n = 5) (n = 19) (n = 21) (n = 29) (n = 60)

1960-64 2.693 3.005

1965-69 4.780 4.263 4.736
1970-74 6.865 3.969 6.406 5.322
1975-79 1AA6 5.284 7.825 7.187 7.835
1980-84 6.212 3.681 7.172 7.044 7.811 6.922
1985-89 5.346 4.000 6.041 6.582 6.666 7.565

unchanged throughout this period (it varied from 50 to 54, with no trend),
the different appointment cohorts (five-year intervals between 1955 and
1984) correspond to different age groups.
I have limited the citing opinions to those issued within eight years
after the beginning of each period. So the number 2.693 in the first cell of
the table means that opinions written between 1960 and 1964 by judges
appointed between 1955 and 1959 were cited an average of 2.693 times in
opinions published between 1960 and 1968. Thus, reading across the table
provides an index of the quality of the decisions rendered by the different
age cohorts in the same period, and so should help us identify both an aging
and an experience effect. Not only identify, but (when we read down as
well as across) distinguish between. For while the oldest and also the most
experienced judges are at the left, and the youngest and least experienced
at the right, as we move down the table the most experienced judges be¬
come older. If judging is like other jobs, then at first the experienee effect
would be expected to dominate the aging effect, so that the older judges
would have more citations than the younger ones, but eventually the
aging effect would overtake the experience effect, so that the very oldest
judges, like the younger ones, would have fewer citations than the middle-
aged ones.
Neither the aging nor the experience effect is consistently observed
(the significance of the qualification implicit in “consistently” will be con-

research team at the University of Chicago Law School headed by William Landes and Lawrence
Lessig; the data used to construct tables 8.1 and 8.2 are drawn from this sample. According to data
furnished me by the Statistics Division of the Administrative Office of the United States Courts, in
1993 the average age of federal court of appeals judges was 58.9 and the average years of service
as a court of appeals judge were 9.2.
184 The Theory Elaborated and Applied

sidered shortly). For example, judges appointed between 1955 and 1959
had more experience in the period 1960-1964 than judges appointed in
that period, and were still relatively young by judicial standards (late fifties
or early sixties), yet the fledglings’ opinions were cited more often (3.005
versus 2.693) than the experienced judges’ opinions. In the last two rows
in the table, the oldest judges, although by now pretty long in the tooth,
were garnering more citations for their current opinions than the younger
but still highly experienced judges of the 1960-1964 cohort.
I am particularly interested in the aging effect. None at all is percep¬
tible until the last two periods. By the middle 1980s, the eleven judges in
the sample who had been appointed between 1955 and 1964 (first two col¬
umns) were in their eighties; and the citations to the decisions they wrote
in the period 1985-1989 are indeed fewer than the citations to the deci¬
sions written by the younger judges in that period. Still, considering the
table as a whole, one is struck by how slight the effect of aging on quality,
at least as measured by citations, appears to be.
Two qualifications are necessary. The first is that if citations per opin¬
ion are increasing over time, perhaps because a rapid expansion in the num¬
ber of decisions is increasing the “demand” for citations to earlier deci¬
sions, this might mask an aging effect. Reading the table diagonally from
left to right compares the average number of citations to the decisions of
the different cohorts at approximately the same age (for example, the first
cohort after ten years, the second after ten years, the third after ten years,
and so on). This comparison does reveal an upward trend,^ but this is a
reason only for not attaching much significance to the trend within each
column, that is, to the time trend of each age cohort’s citations. When,
instead, the comparison is across rows, that is, when what is compared are
the citations to the different age cohorts in the same period, the time trend
in citations drops out.
A more important qualification concerns the effect of aging on quan¬
tity, and indirectly on quality (depending on precisely how “quality” is
defined). This effect is explored in table 8.2, where the average annual
number of opinions written by the different age cohorts in the different
periods is compared.
The number of opinions in the first period of each age cohort is artifi¬
cially depressed, since some of the judges would have been appointed to-

6. The fall off in the last row is misleading; the count of citations ended with the end of
1993, so that many decisions rendered in the 1985 to 1989 period could not accrue a full eight
years of citations.
Adjudication and Old Age 185

Table 8.2 Average Annual Number of Judicial Opinions, by Age Cohorts

Cohort
Period

of 1955-59 1960-64 1965-69 1970-74 1975-79 1980-84

Opinion (n = 6) (n = 5) (n = 19) (n = 21) (n = 29) (n = 60)

1955-59 28.6
1960-64 34.1 16.6
1965-69 27.8 24.3 14.7
1970-74 23.2 21.8 28.1 16.4
1975-79 17.1 17.5 26.7 29.3 11.9
1980-84 15.3 17.6 24.3 28.3 33.6 20.2
1985-89 11.3 6.9 21.0 21.4 32.9 28.6

ward the end of the period and would therefore not have had an opportunity
to produce a full five years’ worth of opinions. Ignoring the first period,
therefore, and reading down the columns, we see that the production of
opinions does appear to decrease with age. Reading diagonally from left to
right, we see that this is not because the caseload per judge has been de¬
creasing. Reading diagonally compares the output of the different age co¬
horts at roughly the same age. For example, in its third period the third
age cohort wrote an average of 26.7 opinions, while the fourth cohort wrote
28.3 opinions in its third period, and the fifth 32.9 in its. This trend reflects
the fact that caseload per federal appellate Judge has been increasing over
the entire period covered by the tables.
In the case of the first and second age cohorts, a substantial decrease
in output sets in either in the third period after appointment, for the first
cohort, or in the second period, for the second cohort. In the case of the
third cohort very little decrease is observable until the fourth period, when
most of the judges would have been in their early or middle seventies.
The quantity effect of age is related to the interesting institution of
“senior status.” Federal judges are eligible either to retire at full pay, or to
take senior status (reduced workload, also at full pay), if they are at least
65 years old and the sum' of their age and their years of service is at
least 80; thus a 68-year-old could retire or take senior status with only
12 years of federal judicial service. There is no vesting, however, and there¬
fore no early retirement except for disability. A retired judge as distinct
from one who takes senior status is released from all restrictions on the
outside activities of federal judges. He can, if he wants, practice law. But
of course he can no longer exercise any federal judicial authority, and he is
186 The Theory Elaborated and Applied

not entitled, as a senior judge is, to participate in any general pay raise that
federal Judges may receive during his lifetime, as distinct from cost of liv¬
ing increases, which the retirees do receive.
Judges have a strong financial inducement to take senior status at the
earliest possible opportunity. The pay of a senior judge is deemed “un¬
earned” for purposes of social security, and therefore the senior judge does
not pay social security tax on his judicial income. Nor does that income
affect his eligibility for social security benefits (for which all judges who
take senior status are eligible, since the earliest age at which senior status
can be taken, other than in cases of total disability, is 65), as earned income
would until the judge reached 70. Yet despite this—a benefit easily worth
$100,000 to a judge eligible to take senior status at 65—only 16 percent of
the 371 current senior judges took senior status on their earliest eligibility
date.’ This is powerful evidence that federal judicial service confers sub¬
stantial nonpecuniary benefits. The average age at which judges take senior
status in recent years has varied between 67 and 69. This is substantially
higher than the average retirement age in this country—and it is not retire¬
ment. Most federal judges do not retire completely before they reach the
age of 80, and many keep going until their middle or even late eighties.
The work of a federal district judge (trial judge) is harder than that of
a federal circuit judge (appellate judge) because it requires much more time
in court. So it is not surprising that of the 21 federal judges who took senior
status on their earliest eligibility date in the period 1991 through 1993 all
but one were district judges and that the district judges had on average
fewer years of service (13.4) than the years of service of the lone circuit
judge (17.0). Obviously the circuit judge “sample” is too small to be mean¬
ingful, but if we consider all federal judges who took senior status in this
period—17 circuit judges and 82 district judges—the disparity in years
of service is even greater. The circuit judges had on average 20.2 years of
service, the district judges only 15. Yet even district judges have sometimes
remained in active status to extraordinarily advanced ages—Judge Edward
Weinfeld of the Southern District of New York, for example, until his death
at the age of 86.
Since senior status encourages judges to take a reduced caseload
at advanced ages, it helps explain the age effect on quantity shown in
table 8.2. Yet although the quantity of the older judges’ output is smaller.

7. The statistics in this paragraph and the next were furnished me by the Statistics Division
of the Administrative Office of the United States Courts. The figure 371 includes senior district as
well as senior circuit judges.
Adjudication and Old Age 187

the quality, at least as proxied by average citations per opinion, holds up

remarkably well compared to other occupations. Granted, most of the
judges in our sample whose performance had suffered severely because of
age would have been nudged into retirement—once a federal judge takes
senior status, his continued sitting is at the sufferance of the chief judge of
his court and the judicial council of his circuit—before the aging effect was
too conspicuous. As I pointed out in chapter 4, the failure to consider this
selection phenomenon explains (and greatly undermines) the findings in
some studies that age has no effect on workers’ quality. Nevertheless, un¬
like the handful of older workers in other fields of employment, the older
cohorts in the tables represent in the aggregate a significant fraction of the
federal appellate judiciary rather than a tiny handful of freaks who have
managed to defy the aging process. Most of the 51 judges in the sample
who had been appointed before 1975, constituting more than a quarter of
the federal appellate judiciary today, had reached normal retirement age
before or during the 1985-1989 period—yet had not retired, although
most of them were carrying a significantly reduced caseload.
Another form of selection bias is at work as well, however. People
hired for new jobs in their fifties or later are not a random draw from their
age cohort with regard to health and energy. They are likely to have aged
less than the average member of their cohort, and are therefore likely to be
productive to an older age than the average. Had Holmes been a dodderer
at 61, he would not have been appointed to the U.S. Supreme Court. The
point can be generalized: the older a class of workers is on the date of their
initial hire, the more “youthful” they are likely to be relative to their con¬
temporaries. This makes it both less surprising that elderly judges perform
as well as they do and perilous to generalize from that performance to the
vocational capabilities of the elderly in general.
The most serious objection to concluding that age takes only a slight
toll of judges lies in the problematic character of trying to distinguish be¬
tween “quality” and “quantity.” If by working more slowly an elderly au¬
tomobile worker could assemble two cars a day rather than his quota of 10,
it would be odd to commend him for the “quality” of his work even if the
cars he assembled were as sturdy as those assembled by younger workers.
His productivity would be only 20 percent as great as theirs. Elderly judges
who produce fewer opinions per year than their younger colleagues are
similarly less productive unless their opinions are better, which no one sug¬
gests is the case. Presumably these judges, despite their diminished output
(weighting quantity by quality), are not replaced because Article III of the
Constitution would make that difficult or impossible to do, because the
188 The Theory Elaborated and Applied

federal judiciary does not have to compete in the marketplace and therefore
is not constrained to maximize productivity, and because increased turn¬
over of Judges could impose social costs (of which more shortly). Still,
there are many activities in which elderly people could not duplicate the
performance of younger ones even if allowed to take more time, and we
shall have to consider why that seems not to be the case with judging, at
least within broad limits. But this does not affect my conclusion that there
is an aging effect.
Further evidence both that there is an age-related decline in judging
and that it may not set in until an unusually advanced age is presented in
table 8.3, which summarizes a citation study of Judge Learned Hand’s 38-
year career as a federal court of appeals judge, ending with his death in
1961 at the age of 89.* He took senior status in 1951, at the age of 79. The
table presents data on citations to Hand and his colleagues in five- or six-
year slices.® The order of the judges in each time slice is the order of the
total rather than average number of citations, in all years (through 1992),
to the signed, published, majority opinions that each judge wrote in the
five-year period.'® In effect I am weighting quantity (number of opinions)
by quality (number of citations per opinion). But I also give the average
number of citations, as well as a measure of the durability of a judge’s
opinions—the total number of citations to them in the most recent five-year
period for which data are available (1988 through 1992).
Hand’s distinction, at least as proxied by citation analysis, is great;
but there is a pronounced aging effect. Hand leads in total number of ci¬
tations in every period until he took senior status and in total citations
within the last five years in all but one such period. In average citations
he loses the lead three times during his active period, but in two the win¬
ning judges wrote so few opinions that their victories are not meaningful.
For reasons suggested earlier, total citations is in any event a more mean¬
ingful measure of value of output than average citations, since it weights
quality by quantity. Hand continued to perform with distinction during the
first half of his period of senior status, despite his great age. He led in
average citations, although because the number of opinions he wrote
dropped substantially he lost the lead in total citations and in citations

8. The source of table 8.3 is my essay “The Learned Hand Biography and the Question of
Judicial Greatness,” 104 Yale Law Journal 5\ I, 536-539 (1994) (tab. 1).
9. The reason for the two six-year slices is to prevent the period in which Hand was in senior
service from overlapping his period of active service.
10. The universe of citations is, as with the previous tables, limited to citations by federal
courts of appeals.
Adjudication and Old Age 189

1929 1934 1939 1944 1950 1955 1961

Figure 8.1 Total citations to different vintages of Learned Hand’s opinions

within the last five years (the latter also being a total rather than an average
number). In the last period of his career, between the ages of 84 and 89, his
annual output of opinions dropped slightly," and his average number of
citations markedly. The result was a substantial decline in total citations
(quantity weighted by quality), as shown graphically in figure 8.1. Yet the
most productive period in Hand’s entire career as a federal court of ap¬
peals judge came just before he took senior status; and he was 73 when that
last period of regular full-time service, 1945-1950, began and 78 when
it ended.
The results of the study are not altered significantly by two adjustments
routinely made to make citation counts a better proxy of judicial output—
subtracting self-citations and confining the count to citations by other cir¬
cuits than Hand’s, which were not bound as a matter of precedent to cite
Second Circuit opinions. (Hand was a judge of the Second Circuit, which
covers New York, Connecticut, and Vermont.) When the study is redone
with self-citations omitted, the only change is that Hand jumps from third
to second in average citations in 1940-1944. And when the count is rerun
to exclude citations in Second Circuit opinions—so that what is being
counted are citations by judges not bound as a matter of precedent to follow
the opinions of the judges in my sample, and therefore free to choose the
best opinions to cite—Hand does even better. Now he leads in both total

11. He wrote 84 in the last period, compared to 82 the period before, but the last period is
longer, because he was on senior status for a total of 11 years.
12. This was another “long” period, but scaling the total figures in it down to a normal five-
year period would not change the conclusion that it was his most productive period.
190 The Theory Elaborated and Applied

Table 8.3 Citations to Learned Hand and His Colleagues

1925-1929
Citations per 1988-1992
Judge Opinions Citations Opinion Citations
Hand 244 2269 9.3 46
Manton 291 1560 5.4 11
Swan 121 790 6.5 7
Hough 141 680 4.8 8
Rogers 66 660 10.0 19
A. Hand 85 417 4.9 4
Chase 17 77 4.5 3
1930-1934
Citations per 1988-1992
Judge Opinions Citations Opinion Citations
Hand 257 2300 8.9 72
Swan 245 1529 6.2 15
Manton 306 1251 4.1 9
A. Hand 216 1071 5.0 14
Chase 223 916 4.1 4
1935-1939
Citations per 1988-1992
Judge Opinions Citations Opinion Citations
Hand 244 2025 8.3 81
Swan 236 1250 5.3 18
Manton 268 1213 4.5 21
A. Hand 216 1043 4.8 21
Chase 209 830 4.0 9
Clark 21 192 9.1 4
Patterson 25 152 6.1 6
1940-1944
Citations per 1988-1992
Judge Opinions Citations Opinion Citations
Hand 215 2436 11.3 39
Clark 193 2287 11.8 49
Frank 140 1822 13.0 39
Swan 211 1430 6.8 27
A. Hand 191 1180 6.2 16
Chase 183 948 5.2 13
Patterson 33 178 5.4 3
Adjudication and Old Age 191

Table 8.3 (continued)

1945-1950

Citations per 1988-1992

Judge Opinions Citations Opinion Citations
Hand 224 3149 14.1 119
Frank 191 1624 8.5 52
Clark 198 1595 8.1 23
Swan 202 1372 6.8 16
Chase 179 1164 6.5 22
A. Hand 144 808 5.6 12
1951-1955
Citations per 1988-1992
Judge Opinions Citations Opinion Citations
Frank 147 1535 10.4 50
Clark 167 1415 8.5 27
Swan 113 1188 10.5 20
Hand 82 1049 12.8 42
A. Hand 74 731 9.9 22
Chase 102 714 7.0 22
Medina 54 500 9.3 27
Harlan 23 252 11.0 17
Waterman 3 22 13 0
Hincks 32 0 0.0 0
1956-1961
Citations per 1988-1992

Judge Opinions Citations Opinion Citations

Friendly 91 1825 20.1 107
Waterman 155 1642 10.6 61
Clark 169 1454 8.6 64
Medina 127 1283 10.1 62
Hand 84 645 7.7 41
Swan 70 562 8.0 22
Moore 68 399 5.9 23
Smith 28 356 12.7 12
Hincks 95 304 3.2 7
Frank 35 265 7.6 2
192 The Theory Elaborated and Applied

and average citations in every period—-until he took senior status. The ag¬
ing effect is not erased.

Causes of the Productivity of Older Judges

Explaining the results in the previous section is a challenge. Lehman lists
judging as a leadership job and so is not surprised to find a high average
age of judges.'^ That doesn’t seem right. Chief judges, especially the Chief
Justice of the United States, exercise some leadership functions, and the
collegial nature of appellate decision-making places some premium on in¬
terpersonal skills, but these things have figured in the reputations of only a
handful of the “great” judges. If judicial staffs, which are still small, con¬
tinue to expand, leadership may some day become an important dimension
of judicial performance—and if so that will be a boon to older judges, both
directly through the increased assistance provided by a larger staff and in¬
directly by enabling them to employ their mature leadership skills. But that
is in the future. The highly reputed judges of the past do not owe their
reputation to leadership skills, with the exception of a few Chief Justices,
such as John Marshall, William Howard Taft, Charles Evans Hughes, and
Arthur Vanderbilt (the last a chief justice of New Jersey’s supreme court).
Judges, especially in a common law system, rely heavily on reasoning
by analogy, conventionally a branch of practical reasoning, hence arguably
less dependent on fluid intelligence than exact reasoning is. But we have
considered this argument in previous chapters, and found it wanting. Rea¬
soning by analogy draws on the fluid intelligence, and declines with age.
Most people would say that judges can perform creditably at advanced
ages because to be a good judge requires good judgment, and judgment is
a function of age and experience. Judges are “wise men,” and the wise are
old, their wisdom being the product of lived experience. It is difficult to
make sense out of this claim. Judges in a mature legal system, especially
appellate judges (and all the distinguished elderly judges whom I have
mentioned, except Edward Weinfeld, were appellate judges), are not me¬
diators or conciliators. They are rule appliers and, to a limited extent, rule
makers. Why extensive lived experience, or even long experience in one or
more of the nonjudicial branches of the legal profession, should enhance
the quality of their output is therefore not obvious. But I think there is some
merit to the popular view, though not to the thinking behind it. The rules of
judicial ethics and the institutional characteristics of appellate judging limit

13. Harvey C. Lehman, Age and Achievement 286 (1953) (tab. 50).
Adjudication and Old Age 193

the information that filters up to the judges about the cases they are hearing.
In particular, the facts of a case tend for a variety of reasons to become
bleached and deformed en route to the appellate court, and lawyers are very
bad at filling in the judges with the background of practices and usages out
of which the case arises. It is therefore important for an appellate judge to
bring to his job a background of knowledge about the behaviors out of
which the cases he will be judging arise: to be, in short, “experienced.”
This point, however, is more relevant to the optimum age of appointment
than to a comparison of judges of different amounts of judicial experience.
I shall come back to this distinction.
Another important asset of a judge is disinterest; and judges are less
likely to decide cases with a view toward maximizing their future career
opportunities, and are therefore more likely to decide cases impartially, the
less of a future they have.'^ We want judging to be a terminal job rather
than a springboard to another career. This implies that judges should be
appointed at an age sufficiently advanced to make it unlikely that they will
change careers. The feasibility of such a strategy is reinforced by the
“cliff” nature of the federal system of judicial pensions. The absence of
any vesting before the earliest retirement age (65) means that as that age
approaches, the discounted cost of resignation in lost pension rights rises
steeply. The cost drops to zero or even turns negative when the judicial
pension finally vests, but the age-decline curve that I have emphasized
throughout this book implies that few 65-year-olds will seek new careers.
Granted that the practice of law is not quite a “new” career for a judge—
most judges practiced law before their appointment to the bench—its rig¬
ors and risks in comparison to senior status are sufficiently daunting that
the judge who for the sake of his pension rights sticks to judging until the
age of 65 is unlikely to retire then and seek employment elsewhere.'® The
younger the judge is when appointed, the less influenced he will be in de¬
ciding whether to remain a judge by the prospect of losing all his pension
benefits if he resigns.

14. To the extent that “book learning” is a substitute for lived experience, we can expect
judges appointed from academia to be somewhat younger than judges appointed from practice.
15. Cf. David Gutmann, “Age and Leadership: Cross-Cultural Observations,” in Aging and
Political Leadership 89, 95-96 (Angus McIntyre, ed., 1988), discussing the role of the presumed
disinterest of aged lawyers in bringing down both Senator Joseph McCarthy and President Nixon.
16. Between 1990 and 1992, 113 federal judges either took senior status or retired, but only
7 (all of them retirees, of course) went into private practice. Emily Field Van Tassel, Why Judges
Resign: Influences on Federal Judicial Service, 1789 to 1992 40 (Federal Judicial Center 1993).
The average age of the 58 federal judges who have resigned since 1950 for reasons other than bad
health or criminal conviction is only 52. Computed from id. at 126-127.
194 The Theory Elaborated and Applied

Experience and disinterest are important elements of “wisdom,” un¬

fortunately one of the vaguest concepts going.’’ Anthony Kronman calls
wisdom the defining trait of “lawyer-statesmen,” a category that in his
view includes the best judges.’® His concept of wisdom involves a combi¬
nation of cognitive and emotional skills not always easy to distinguish
(which is “empathy,” for example?). The wisdom of the judge includes
such attributes as the capacity to understand from within the plans and
projects of other people, the capacity to project in the imagination the prob¬
able consequences of alternative approaches and outcomes, habituation to
the practices and folkways of the legal profession, deliberative ability, the
capacity to avoid becoming emotionally involved with the issues or the
litigants, thereby losing perspective, and the ability to put aside one’s own
personal or career interests in the outcome of the case. In short, a mature
professional judgment is central to the concept of a wise judge, and the
intellectual and dispositional qualities that go to create such a judgment
plainly improve with age up to a point (a “wise child judge” would be a
considerable oxymoron) and then plateau until senility. The traditional as¬
sociation of age, wisdom, and judging is therefore plausible. I do not mean
to disparage brilliance in judges; but the brilliance of the great judges, like
Holmes, Hand, and Cardozo, is, as we shall see, different from the bril¬
liance of a great scientist.
A further point to consider in evaluating the elTect of age on a judge’s
quality is the backward-looking character of the judicial process. Recall
Aristotle’s time line, according to which, as we age, the focus of our think¬
ing shifts from imagining the future to recalling the past. If one desideratum
in judging is to maintain continuity with the past—to enforce settled un¬
derstandings, traditional rights, and old compacts such as the U.S. Consti¬
tution—it may be good to have as judges people whose cognitive orien¬
tation is toward the past rather than the future. The backward-looking
emphasis of adjudication is summarized in the principle of stare decisis:
judges are to adhere to precedent. We should expect therefore that the more
a judicial system emphasizes stare decisis, the older its judges will be on
average. This may be one reason why judges in Anglo-American judi¬
ciaries are older on average than Continental judges (indeed. Continental

17. Cf. Douglas H. Powell (with the collaboration of Dean K. Whitla), Profiles in Cognitive
Aging, ch. 9 (1994).
18. Anthony T. Kronman, The Lost Lawyer: Failing Ideals of the Legal Profession (1993).
The issue of wisdom pervades Kronman’s book, so I do not give specific page references. Kronman
does not associate wisdom in his sense with any particular age. He distinguishes between tyros
and seasoned professionals but does not investigate the actual age profile of professional ability.
Adjudication and Old Age 195

judges begin their judicial careers right after law school): stare decisis is a
more important principle in the Anglo-American legal system. But I do not
find this suggestion entirely convincing. The very fact that decisions are
precedents in the Anglo-American system requires the judge in that system
to look forward to the consequences of his decision in future cases. He must
look forward and backward. The causality, moreover, might run from the
age of the judges to the character of adjudication rather than from the char¬
acter of adjudication to the age of the judges. The orientation of Anglo-
American law may be backward-looking, with more emphasis on adher¬
ence to precedent than on forward-looking policy-making, in part because
law’s administrators are inclined because of their age to take a backward¬
looking approach to law.
Those who assign a balance-wheel function to the judiciary—the func¬
tion of reducing the amplitude of swings in public policy—will welcome
an aged judiciary. There is some evidence that, as popularly believed, older
people are politically more conservative than younger ones even after cor¬
rection is made for cohort effects.'® But another and probably more salient
dimension of elderly conservatism is that, consistent with characteristics of
aging emphasized in this book, in particular the age-related cost of assimi¬
lating new ideas, “older persons retain many of the political attitudes and
orientations they developed earlier in life.”^° Presumably this is true of
judges as well as of voters. So if society swings to the left of where it was
when the judges were young they will want to tug it right, and if it swings
to the right of where it was when they were young they will want to tug it
left. The effect will be to enhance political stability. Of course, just as when
comparing the effects of fixed versus floating exchange rates, we must con¬
sider long-run as well as short-run stability; they may go in opposite direc¬
tions. If judges attempt to resist the tides of social change, they may be
swept away in a flood—one interpretation of what happened when the aged
Justices of the Supreme Court challenged the New Deal. Although the ad¬
venturism of the Court during the chief justiceship of Earl Warren may
seem a dramatic counterinstance, it is arguable that the liberal justices of
the “Warren Court” were enacting the values they had acquired in their

19. Anne Foner, “The Polity,” in Aging and Society, vol. 3; A Sociology of Age Stratification
115, 132-136 (Matilda White Riley, ed., 1972).
20. John B. Williamson, Linda Evans, and Lawrence A. Powell, The Politics of Aging:
Power and Policy 106 (1982). For evidence, see Duane F. Alwin, Ronald L. Cohen, and Theo¬
dore M. Newcomb, Political Attitudes over the Life Span: The Bennington Women after Fifty
Years 90-96 (1991), and other studies cited there; also M. Kent Jennings, “Residues of a Move¬
ment: The Aging of the American Protest Generation,” 81 American Political Science Review
367 (1987).
196 The Theory Elaborated and Applied

youth—which for many of them coincided with the New Deal. The un¬
seemly haste with which German judges adopted the values of the Third
Reich may reflect the concordance between those values and the conser¬
vative values of the judges’ youth.In both cases, that of the Warren Court
and that of the Nazi judiciary, judges amplified rather than moderated a
social revolution.
A mundane but important consideration in the continued productivity
of very old judges is the bearing of boredom. Many judges in this country
are people who began a substantially new career when they were in their
fifties. If it takes (at a guess) about 30 years of working at the same job to
make boredom a substantial cost of working, we should not expect it to
push most judges into retiring at the earliest opportunity, or to degrade the
quality of their work substantially until they are well past the earliest age
at which they can retire.
My discussion of the congruence of age and adjudication may seem to
have overlooked the last-period problem. What is to keep judges on the
beam when they are in the last period of their lives? But this question re¬
veals a misunderstanding of the last-period problem (which is perhaps mis¬
named). The last period is a problem in large part because it is not really
the last period; it is simply the last period in which the individual is subject
to a particular set of incentives. The employee who soldiers on the job or
steals from the till because the only feasible sanction available to his em¬
ployer is to fire him and he is about to leave anyway has an incentive to
misbehave only because he will still be living after he leaves his job. If the
job were literally terminal, he would lack that incentive. So to the extent
that judicial employment approximates terminal employment, the judges’
incentive to misbehave will be limited. Judges will not only have nothing
to lose from deciding a case in a particular way, but nothing to gain either.
The hope is that with incentives to self-interested behavior in the usual
senseout of the picture, the judge will render a genuinely public-
interested decision. The older the judge, other things the same, the more
likely this hope is to be realized. But it is just a hope, because with self-
interest out of the picture the penalty for whimsical or arbitrary action is

21. See Ingo Muller, Hitler's Justice: The Courts of the Third Reich (Deborah Lucas Schnei¬
der, trans., 1990); Richard A. Posner, Overcoming Law. ch. 4 (1985).
22. An important qualification for some judges—those who derive satisfaction from using
their judicial office to advance a political philosophy. But below the Supreme Court level, they are
a minority. See Richard A. Posner, “What Do Judges and Justices Maximize? (The Same Thing
Everybody Else Does),” 3 Supreme Court Economic Review 1 (1994), reprinted in Overcoming
Law, note 21 above, ch. 3.
Adjudication and Old Age 197

removed along with the reward. Admittedly this is an exaggeration, since

judges may care about their posthumous reputation.
I have said that wisdom, and by implication judicial capability, do not
decline with age. But this, too, is an exaggeration, as we saw in connection
with Learned Hand’s career, and I have now to consider the issue more
carefully. Experience in the sense of first-hand engagement with the law in
operation ceases to grow the moment the judge is appointed and thereby
withdraws from that world. This implies net depreciation of his profes¬
sional human capital from that moment forward. But judicial experience
creates judicial human capital, enabling the judge to increase his quality-
weighted output of decisions and thereby offset, for a time anyway, the
inevitable age-related decline in mental energy and acuity. Judging is a
learning-by-doing sort of job. The experience of reading many briefs and
hearing many oral arguments increases the speed at which the judge can
extract the gist of a case from the briefs and arguments of the lawyers.
Speed is important in judicial work. As in certain forms of leadership (see
chapter 7), judges have to make many decisions in a limited amount of
time, and speed and confidence in judicial decision-making are functions
of experience.
In addition, the work that a judge puts into preparing to hear a case and
afterward writing the opinion in it (if he is the judge on the panel who is
assigned the case to write) teaches him something about the field or sub¬
field of law involved; so the next time he has a case in the same area he can
master it more quickly. This is important because most judges were engaged
in more specialized work before they ascended the bench. Most American
judges are generalists, most lawyers and law professors specialists. So
judges must acquire much of their judicial knowledge base as judges.
If judges acquire new human capital while judging, one might expect
the quantity and the quality of a judge’s work actually to increase as he
ages in the job, rather than just not to decrease. But the judge’s pre-judicial
experience will be depreciating even as his judicial experience is appreci¬
ating, as will his energy. His fluid intelligence will be diminishing as well.
Since such diminution is an increasing function of age, and since the cost
of acquiring judicial experience like other new human capital is higher for
older than for younger people, we might expect the quality-weighted output
of judges appointed at younger ages to increase for a time after their ap¬
pointment before leveling off and eventually declining, while the output of
work of judges appointed at more advanced ages would remain constant
until the eventual decline. But tables 8.1 and 8.2 reveal no correlation be¬
tween judicial output and judicial experience. The explanation may lie in
198 The Theory Elaborated and Applied

the peculiar terms of federal judicial employment. The reason that an

experience-related increase in an employee’s productivity is expected to
lead to an increase in his output is that his employer can reward him for the
larger output in a higher salary or fire him if he produces a smaller output.
Both carrot and stick are missing in the case of federal judges, all of whom
(of the same rank) are paid the same and can be removed only for gross
malfeasance. There is nothing to prevent a federal judge from transforming
an increase in his productivity due to increased experience into leisure. His
output remains the same, but is produced in less time, allowing him more
leisure. This interpretation of the tables is consistent with the argument I
have made elsewhere that leisure is an important element of the average
judge’s utility function.^^ The interpretation is weakened if judges would
derive additional utility from additional output. Some would. Many would
not. They derive utility from “doing their job” but not from doing more.
Another factor that may explain the low rate of decline of judicial
ability (the first was the acquisition of judicial capital)—and it is the deci¬
sive factor in the high quality of exceptional judges at advanced ages—is
the importance, to judicial distinction, of writing. The distinguishing fea¬
ture of most great judges, even those rightly pronounced “brilliant,” has
not been exceptional analytical power; it has been exceptional rhetorical
power.The volume, variety, and random sequence of cases, the commit¬
tee character of judicial decision-making, the essentially interpretive rather
than creative character of the judicial function, and the sheer age of the
judges make it unlikely that judicial decisions, even of the ablest judges,
would display a high order of intellectual creativity. Most of the ideas ex¬
pressed in judicial opinions come from outside, from lawyers, legislators,
law professors, and scholars in other fields, such as economics, political
science, and philosophy, or in the case of intellectual judges like Holmes,
Brandeis, or Frankfurter from ideas that they had developed before they
became judges. What lifts a judicial opinion out of the commonplace, apart
from the accidents of historical significance over which the judge has no
control, is the vividness, compactness, and, in short, memorableness of his
exposition. Even so famous an example of judicial “creativity” as Learned
Hand’s formula for negligence,^^ a landmark in the economic analysis of

23. See id.

24. Posner, note 4 above, at 133-137.
25. B < PL, where B is the burden (cost) of avoiding an accident, P the probability that the
accident will occur unless precaution B is taken, and L the loss (cost) that the accident will cause
if it occurs. The formula was announced in Hand’s opinion in United States v. Carroll Towing Co.,
159 F.2d 169, 173 (2d Cir. 1947).
Adjudication and Old Age 199

law, is merely a restatement in algebraic form of the conventional tort stan¬

dard of negligence; it is algebra as metaphor, not as mathematical analysis.
Writing ability, which in the best judges can fairly be termed literary,
is an aspect of crystallized intelligence, and is one of the aspects least likely
to decline with age until senility sets in. It is true that creative writers often
run out of things to say, with the result that many writing careers have an
early peak.^^ But not all by any means, as we saw in the preceding chapter;
and running out of fresh ideas is unlikely to be a problem for judges, be¬
cause they are given a fresh topic, as it were, in every case; they don’t need
to draw on an internal stock of ideas that might run down. Comparison of
the writing style (not intellectual creativity) of the same person at different
ages often reveals steady improvement to a quite advanced age, with no
decline from that late peak until shortly before death. Holmes’s most elo¬
quent opinion, his dissent in the Abrams free-speech case,^^ was written in
his late seventies, and Learned Hand wrote the “Hand formula” opinion in
his middle seventies.
Not every student of adjudication will agree with my analysis of the
elements of judicial distinction. But I do not think that my conclusion that
there is an age-related decline in judicial performance, but that it does not
set in until a relatively very advanced age, is highly sensitive to that analy¬
sis. Judge Friendly, whom I have mentioned, once listed four elements of
“outstanding [judicial] quality”; “analytical power,” “legal learning,”
“general culture,” and “the ability to write graceful and powerful En¬
glish.” The first of these—but only the first—is likely to decline with age.

The Effect of Increased Longevity

Despite the many examples of distinguished performance by old judges,
and the reasons that explain them, continuing increases in longevity may
have a negative impact on the performance of the federal judiciary if the
rules governing the retirement of federal judges remain unchanged. Even if
the age-decline curve is very gradual for judges, their performance must
eventually decline; and of course the decline is more rapid for some judges

26. Dean Keith Simonton, “Age and Creative Productivity: Nonlinear Estimation of an
Information-Processing Model,” 29 International Journal of Aging and Human Development 23
(1989), proposes a model in which the researcher begins his career with a fixed creative potential
and draws down this endowment over the course of his career. See also Simonton, Genius, Crea¬
tivity, and Leadership: Historiometric Inquiries 109-112 (1984).
27. Abrams v. United States, 250 U.S. 616, 624 (1919).
28. Henry J. Friendly, Book Review (of Learned Hand’s Court, by Marvin Schick), 86 Po¬
litical Science Quarterly 470, 471 (1971).
200 The Theory Elaborated and Applied

than for others. So if increased longevity increases the percentage of judges

who are very old, the quality of judicial output could suffer unless the effect
of increased longevity on capability were perfectly offset by a reduction in
the age-decline rate, which is unlikely. A negative effect of increased lon¬
gevity on judicial performance is especially likely if our society continues
to change very rapidly, causing the problems thrown up to the judiciary for
solution to change rapidly as well. Such change would increase the rate at
which judicial performance declines because of age. A partial offset, how¬
ever, is the increased size of judicial staffs, which, as I have noted, enables
some elderly judges to produce a larger quality-weighted output than they
could have done in the old days. A further point is that a rising average age
of retirement for judges implies, unless the average age of appointment
rises by the same amount, a reduction in judicial turnover, and this may be
a good thing if a high value is placed on legal stability.
The effect of longevity on judicial performance would have little sig¬
nificance if the average age of judicial retirement could not be expected to
rise as judges’ longevity increases. But it may rise. Judicial employment
generates substantial psychic income not offset by substantial nonpecu-
niary costs until the elderly judge becomes incapable of putting forth the
modest amount of effort required for a judicial performance of minimum
respectability. In other words, /„ in equation 4.2 is strongly positive for
most judges—and we recall that this reduces the likelihood of retirement
at the earliest possible opportunity, especially where, as in the case of the
federal judiciary, retirement is strictly voluntary. The effort costs of remain¬
ing a judge at an advanced age have fallen in recent decades because of
better health and the increase in judicial staffs. Both developments have
probably conduced to increased productivity by elderly judges. If it never¬
theless is desired that judges retire no later than they do now, the terms of
retirement can be made even more attractive than they are now. This has
been the response of the private market to the law’s curtailment of manda¬
tory retirement at fixed ages.
An alternative approach to the problem (if it is a problem) of an aging
judiciary would be to replace our system of generalist courts with one of
specialized courts. Generally, the more specialized a worker is, the less new
human capital he need acquire in order to be able to do his job creditably.
The acquisition of new human capital is costly for the old because of the
age-related decline in fluid intelligence, and a generalist judge is one who
has to keep investing in human capital as new fields of law come within his

29. See Posner, note 22 above.

Adjudication and Old Age 201

purview. But boredom is more likely to set in sooner, the more specialized
the job. So a specialized judiciary may not be the best way to accommodate
the predictable increase in judicial longevity.
As should be plain from this discussion, the age profile of judicial
achievement, and the factors that determine the profile, have implications
for the structure of the judicial system. These implications are explored
further in chapter 13. The essential points in the present chapter have been,
first, to use citation analysis to evaluate the proposition that adjudication is
an example of a late peak, sustained, activity and, second, if it is, to explain
why. The citations analysis confirmed that appellate adjudication has the
predicted properties. But it also shows that there is an aging effect; that
even so distinguished and long-lived a judge as Learned Hand eventually
experienced a pronounced decline in output as proxied by total citations,
which, properly, weight quality by quantity. As far as the causality of the
judicial age profile is concerned, we saw that the major cause appears to be
the character of legal reasoning, with its heavy emphasis on practical rea¬
son in a sense that identified it as an aspect of crystallized intelligence.
Notice that as well as using the concept of practical reason to account for
the result of the citation analysis, I have argued that that analysis supported
the proposition that practical reason is indeed the lifeblood of legal reason¬
ing, rather than—as a surprising number of judges and law professors con¬
tinue to believe—logic or some other method of abstract reasoning. Other
factors that help explain the age profile of the judicial profession include
the policy of adhering to precedent in the Anglo-American judicial system,
the (related) desire for the judiciary to act as a stabilizing rather than inno¬
vative force in social and political life, the positive correlation between old
age and disinterestedness, and the late age of judicial appointment, which
both reduces the likelihood that boredom will drive judges to retire and
selects for persons more youthful than the average of their age cohort.
 9
The Status of the Old
and the Aging of Institutions

I have been considering the old mostly from their own standpoint and that
of their employers rather than from the standpoint of society as a whole or
even of the families of aged people. But the way in which nonelderly mem¬
bers of society, including both relatives and nonrelatives, regard and treat
the elderly has great significance for both positive and normative analysis
of aging. There is an instructive analogy to the social valuation of children.
Because children are dependent on adults, the status of children—includ¬
ing how many there are and what investments are made in their human
capital—reflects the interests of the adult population. Families will be
smaller when children are costly to raise and do not provide extensive ser¬
vices to their parents, and larger when they are cheap to raise and do pro¬
vide extensive services to their parents, as in traditional agricultural soci¬
eties.' The dependent old are in a parallel position. They exist at the
sufferance of the dominant age groups, including their own adult children.
I want to consider the factors that from a rational-choice perspective
are likely to influence the treatment that the elderly will receive from soci¬
ety. Such a perspective seems, if anything, even more apt to the treatment
of the old than to the treatment of the young. People appear to be geneti¬
cally programmed to feel protective toward children; why else do we find
children, and the young of animals, “cute,” appealing, even adorable?^ We

1. On the economics of the family, see Gary S. Becker, A Treatise on the Family (enlarged
ed. 1991).
2. A point by no means inconsistent with even widespread infanticide, abortion, and neglect.
Richard A. Posner, Sex and Reason 143—144 (1992). The optimum and the maximum number of

202
Status of the Old 203

do not appear to be genetically programmed to feel as protective toward

old people in general; while most children love their parents even when the
parents are old, they do so generally with diminished intensity. This asym¬
metry makes biological sense. Inclusive fitness is unlikely to be promoted
by the devotion of huge resources to the survival of persons who, by reason
of advanced age, are not reproductively or otherwise productive, either ac¬
tually or (like children) potentially. Postreproductive individuals may con¬
tribute to the survival of their offspring and thus be productive; because of
this, respect for old people in general and for one’s parents and grandpar¬
ents in particular may be instinctual. But these feelings are less intense than
the counterpart feelings toward the young because the old make a smaller
contribution to inclusive fitness. We can therefore expect the dominant
groups in a society, because they are not being tugged as powerfully by the
genes, to be more calculating about the very old than about the very young
members of the society, even when the very old are their own parents and
grandparents. Indeed, we can expect the old themselves, altruistically in¬
clined by their genes to favor the younger members of their families, to co¬
operate to a certain extent (depending on circumstances, as we shall see) with
policies that promote the interests of the young at the expense of the old.
The status of the old, as measured by offices occupied, wealth con¬
trolled, and respect accorded, varies greatly both across societies^ and,
within societies—including, as we shall see, our own—across time. Some
ancient and primitive societies have killed their old people,'* sometimes

children are not the same, as too many children may endanger the survival of all. As I noted in
chapter 1, this is a possible explanation of menopause.
3. For a good summary, see John B. Williamson, Linda Evans, and Lawrence A. Powell,
The Politics of Aging: Power and Policy, pt. 1 (1982). Case studies are numerous. A representa¬
tive anthology is Aging in Cross-Cultural Perspective: Africa and the Americas (Enid Gort,
ed., 1988). A somewhat dated but incomparably vivid summary of the anthropological literature
on old age is Simone de Beauvoir, Old Age, ch. 2 (1972); another excellent summary is Jack
Goody, “Aging in Nonindustrial Societies,” in Handbook of Aging and the Social Sciences 117
(Robert H. Binstock and Ethel Shanas, eds., 1976); while for a wealth of data concerning the
status of the aged in primitive societies, see Leo W. Simmons, The Role of the Aged in Primitive
Society (1945).
4. Some 20 percent of primitive societies in two studies based on data in the Human Rela¬
tions Area Files. Jennie Keith, “Age in Social and Cultural Context: Anthropological Perspec¬
tives,” in Handbook of Aging and the Social Sciences 91, 92 (Robert H. Binstock and Linda K.
George, eds., 3d ed. 1990). See also Steven M. Albert and Maria G. Cattell, Old Age in Global
Perspective: Cross-Cultural and Cross-National Views 224-228 (1994). In one of the studies,
elderly were “forsaken,” “abandoned,” or “killed” in 59 percent of the societies sampled.
Anthony P. Glascock and Susan L. Feinman, “A Holocultural Analysis of Old Age,” 3 Compara¬
tive Social Research 311, 323 (1980) (tab, 7). It should be plain that geronticide is no more incon¬
sistent with an instinctual filiality than infanticide is inconsistent with an instinctual affection for
children. See note 2 above.
204 The Theory Elaborated and Applied

with the convenient rationalization that dying a natural death impairs one’s
prospects in the afterlife;^ others have neglected them; others have revered
them. Despite charges of “ageism” leveled by representatives of the elderly
and by some radical egalitarians (who see discrimination everywhere), it is
doubtful that there has ever been a society in which old people as a whole
have been as politically influential, as materially well-off, and, probably, as
happy as they are in modern American society, although they are not re¬
vered. Apart from their wealth and number, and the extraordinary medical
resources devoted to their health and longevity, old people occupy a range
of important positions in government, business, and the nonprofit sector.
We saw in the preceding chapter that judging is an old person’s profession;
and we know that the legal profession in general, and the courts in particu¬
lar, are more powerful in the United States than in any other country. How
to explain the dramatic differences in the status of the old across societies
is the main challenge of this chapter. I acknowledge that the term “status”
is vague in this setting, but it will do for now and I shall clarify it later.

Primitive and Agrarian Societies

Determinants of elderly people’s status. Since I shall be talking a lot
about “primitive” societies, and the term is likely to raise hackles, 1 should
explain what I mean and assure the reader that no pejorative or demeaning
connotation is intended. Nineteenth-century explorers, ethnographers, and
anthropologists discovered that a large number of cultures scattered around
the world were preliterate, were extremely poor by Western standards, and
were subsisting at an essentially Stone Age level of technology, with very
little contact with the inhabitants of the industrialized world. These cul¬
tures (now extensively “contaminated” by contact with technologically ad¬
vanced cultures) have been studied intensively and a good deal of data
about them have been collected in forms that enable, though with great
difficulty, the kind of comparative empirical study reported later in this
chapter. So “primitive” should be understood as a technical term, just as
when it is used to denote the style of painting of Grandma Moses.
A preliterate society is unlikely, course, to know the precise chrono¬
logical age of its elderly members. (Even today, we saw in chapter 2, the
number of the very oldest Americans is not known with precision.) How
does this cut, so far as status is concerned? Both ways. On the one hand,

5. See, for example, James George Frazer, The Dying God (part 3 of The Golden Bough)
10-14(1913):
Status of the Old 205

chronological age cannot be used as a proxy for ability to perform particu¬

lar tasks, so “mandatory retirement,” which forces the ablest old to retire
on the basis of the average capability of their age cohort, is unlikely to be
practiced. Statistical discrimination against the aged (see chapter 13) is
less feasible if there are no statistics. On the other hand, lacking a precise
knowledge of chronological ages, people may be classified as “old” on the
basis of superficial characteristics, such as the amount of grey in their hair
or the number of creases in their skin, that are only loosely related to age
in either a chronological or a functional sense. By the same token, some
young-looking people will be classified as young even though they are
functionally and chronologically old. They will benefit from the absence of
records. The effect of that absence on the position of elderly persons in a
society is thus indeterminate on the level of theory.
In an extremely poor society, the maintenance of the old, if they can
no longer maintain themselves, may be prohibitively costly. The cost of a
good or service in real as distinct from pecuniary terms is what is given up
in its consumption; that is why, as we saw in chapter 5, old people are
willing to spend so much money on modest extensions of life. What is
given up depends on scarcity. If a society is so poor that food is very scarce,
the cost of feeding an old person may be the starvation of a young one,
and in such a case the society is likely to allow the old person to starve, or
even kill him outright. The anthropological literature contains examples of
primitive societies in which elderly people acquiesce in their society’s ger-
ontocidal norms, even to the point of going to their deaths gaily.^ Is this a
product of socialization so intense that it succeeds in overcoming the in¬
stinctual dread of death? Perhaps so, but an alternative hypothesis is that
where the choice is between survival of a person who is past reproductive
age and survival of his descendants who still have reproductive potential,
the genes will incline the older family member to sacrifice himself for the
younger.
This analysis suggests that, other things being equal, the poorer a so¬
ciety is, the more likely it will be to kill or let die its oldest members. Other
things may not be equal, however. An important variable is the ratio of the
nonworking old (and any'other nonworkers, such as children, but I shall
ignore this point in order to simplify the analysis) to the working part of
the population. The lower that ratio, the less each worker will need to con¬
tribute from his output in order to support the old. Also militating against

6. See, for example, Simmons, note 3 above, at 236-238; Lucy Mair, African Societies 197
(1974).
 206 The Theory Elaborated and Applied

the neglect or mistreatment of the elderly is the effect on the last-period

problem. If a person knows that he will die as soon as he reaches the thresh¬
old of old age, this knowledge will affect his incentives while he is still
young. He will be less likely to work hard, help others, and be honest,
because the punishments and rewards for his conduct will be severely trun¬
cated. Particularly if effort, honesty, or other dimensions of conduct are
difficult to monitor, society will have a reason for preserving the lives of
old people that is similar to that underlying the “wage-bonding” theory
(discussed in chapter 3), in which compensation is backloaded as a way of
increasing the expected punishment cost of a worker’s performing his job
indolently or dishonestly.
It has long been thought that even though memory is impaired by age,
the memories of old people are potentially of great value to a preliterate
society, a society whose only “records” are the recollections of its mem¬
bers.’ To evaluate this belief, we must distinguish between the society that
is static, in the sense that technology and social practices change very
slowly in them, and the dynamic society. If it is static, as most primitive
societies are, there might seem to be little utility to a long memory: the old
would be remembering and recounting the same kind of things that the
young are experiencing in the present. Yet if the society were dynamic
(though still primitive—admittedly an unlikely combination), what the old
remembered from the old days would not be of much assistance in respond¬
ing to the challenges of the present. This point does help explain why the
old are indeed not revered in dynamic societies, such as our own. But it is
not a complete explanation. The very fact of rapid social change might
cause people to value the old as living links to a receding, perhaps nostal¬
gically regarded or even heroic, past (“the War,” “the Fifties”). However,
the old are at a disadvantage in a dynamic society for another reason be¬
sides the diminished value of their memories to the young. As a conse¬
quence of the age-related decline in fluid intelligence and the age-related
truncation of the expected returns from investing in human capital, old
people are unlikely to make a successful switch to a new type of job. If
demands change rapidly, many old people will find themselves on the shelf.

7. For an example, see Austin J. Shelton, “The Aged and Eldership among the Igbo,” in
Aging and Modernization 31, 45 (Donald O. Cowgill and Lowell D. Holmes, eds., 1972). And
recall the African motto quoted in chapter 2 about the death of an old man being equivalent to the
burning down of a library. The qualification—even if memory is impaired by age—is probably
not important. Few “old” people in primitive societies are so old as to be likely sufferers from
serious age-related cognitive impairments.
Status of the Old 207

The argument that a long memory is unlikely to be a valuable social

asset is far more persuasive with regard to a dynamic society than with
regard to a static one. For a static society is as likely as any other to be
subject to shocks that occur infrequently; the shocks indeed may be
more numerous and more severe. Famine, invasion, eclipse, plague, flood,
drought, the irruption of charismatic personalities, a run of multiple or de¬
formed births, and other alarming or challenging events may be sufficiently
rare that when they occur only the old people in the society can recall a
parallel instance. Perhaps no young person has ever seen an eclipse of the
sun but the old can reassure the young that it is not a new thing and that the
last time it happened there were no untoward consequences. Literacy, and
education more broadly, greatly reduce the value of old people’s memories
as a record of past events.*
Memory plays a further role in a static society. The immediate cause
of a society’s remaining static is that its strategy for coping with the chal¬
lenges that face it is to adhere to its existing practices—to “tradition”—
rather than to seek new solutions. The reason may be that as an aspect of a
general paucity of knowledge, problem-solving skills are not well devel¬
oped. We can observe the strategy in ourselves when we do not understand
the principles underlying an activity. In such cases, once one has hit on a
method that works, one is reluctant to deviate from it. I, for example, not
understanding the principles of computers, am very reluctant to alter my
accustomed routines even when I have no reason to believe that an alter¬
ation would cause the computer to “crash.” If the characteristic method of
“progress” in a primitive society is trial and error, followed by rigid adher¬
ence to the first successful procedure that the method yields, there will be
little scope for the employment of problem-solving skills. The characteris¬
tic intellectual attributes of youth—flexibility, imagination, problem¬
solving ability, mental quickness, openness to new ideas, and powerful
short-term memory—will confer few social benefits and therefore will not
be highly valued. They may even be perceived as dangerous.
Another reason to expect the knowledge value of the elderly to be great
in a primitive society is that, as we saw in chapter 1, life expectancies are
short in such societies. Most of the “elderly” will probably be only middle-
aged by our standards, hence only one generation removed from young

8. For examples, see Maria G. Cattell, “Knowledge and Social Change in Sarnia, Western
Kenya,” in The Elderly Population in Developed and Developing World 121, 139 (P. Krishnan
and K. Mahadevan, eds., 1992); Charles Edward Fuller, “Aging among Southern African Bantu,”
in Aging and Modernization, note 7 above, at 51, 60.
208 The Theory Elaborated and Applied

adults, to whom the “elderly” will be imparting concrete advice on the

performance of adult tasks rather than merely a generalized “wisdom” of
the aged.
We can expect preliterate societies to allocate tasks and social roles on
the basis of age (this is known as “age grading”)^ to a greater extent than
literate societies do, even when the absence of written records creates un¬
certainty about precise ages. Preliterate societies face high information
costs,'® including costs of matching workers to jobs. Everyone may know
everyone else in a small-scale society yet still have great difficulty apprais¬
ing their marginal product in alternative employments, especially when
performance in some of the employments (warfare, for example, or magic)
is intermittent or difficult to evaluate. It is natural in such a situation to use
crude proxies, such as age and parental occupation, for capacity to perform
particular jobs. So one is not surprised at the emergence of customs that
assign all the young men to be warriors, the middle-aged to be leaders, the
king’s son to be king, and the old to occupy judicial," sacral, or consultative
roles, in which wisdom and experience, plausibly if only crudely correlated
with age, dominate strength and mental quickness. In such a system the old
are not in competition with the young, and this may give them a secure
niche in the social structure. The “feminizing” of old men, remarked in
chapter 5, contributes to this reallocation of roles by reducing competition
between old and young men for the same social positions, while the “mas¬
culinizing” of aging women equips them, when age has freed them from
having to tend their own children, to exercise a leadership role in their
extended family.
The less a vulnerable segment of the population, such as the old, can
rely on legal and political rights to secure their survival, the more resources
they will devote to cultivating the goodwill of the powerful. So we can
expect parents in a primitive society to spend more time inculcating filial
piety, or a more general respect for the elderly, in their children than parents

9. See, for example, Keith, note 4 above, at 101-102; Nancy Foner, Ages in Conflict: A
Cross-Cultural Perspective on Inequality between Old and Young 17-24 (1984); Bernardo Ber-
nardi. Age Class Systems: Social Institutions and Polities Based on Age (1985).
10. As emphasized in my book The Economics of Justice, ch. 6 (1981) (“ A Theory of Primi¬
tive Society”).
11. See, for example, Walter H. Sangree, “Age and Power: Life-Course Trajectories and
Age Structuring of Power Relations in East and West Africa,” in Age Structuring in Comparative
Perspective 23, 28 (David I. Kertzer and K. Warner Schaie, eds., 1989); Albert and Cattell, note 4
above, at 70.
12. The importance of “sex-role turnover” in preserving a valuable place of elderly men
and women in primitive societies is emphasized in David Gutmann, Reclaimed Powers: Toward a
New Psychology of Men and Women in Later Life (1987).
Status of the Old 209

in a modern society bother to do.'^ This piety, this “filiality,”takes the

place of legal enforcement in guaranteeing performance of the implicit
(and sometimes quite elaborate) intergenerational contract in which parents
support children when young in exchange for support by the children in the
parents’ old age.'^ To the extent that efforts at inculcating filiality succeed,
the elderly will be more respected than they are in a society in which fewer
such efforts are made.
Odd as it may seem, an exogenous increase in longevity, other things
remaining constant, should increase the incentive of parents to inculcate
filial piety in their children, and might even increase their target number of
children, because they could anticipate a longer period of dependency on
their children.'® Adult adoption, common in societies ranging from ancient
Rome to modern Japan, is a device by which the childless obtain a surro¬
gate assurance (contractual rather than emotional) of support in old age.'^

13. Parents “may want to be taken care of when old or ill, but cannot have a contract with
their children to help out. However, they can try to shape the formation of children’s preferences
to raise their chances their children will help voluntarily.” Gary S. Becker, “Habits, Addictions,
and Traditions,” 45 Kyklos 327, 336 (1992). See also Jeffrey B. Nugent, “The Old-Age Security
Motive for Fertility,” 11 Population and Development Review 75, 78-79 (1985). Nugent points
out that a culture of arranged marriage promotes filial piety by reducing the likelihood that a child
will redirect his affections from his parents to his spouse. Id. at 91 n. 16. For some evidence that
inculcating filial piety really “works,” see Les Whitbeck, Danny R. Hoyt, and Shirley M. Huck,
“Early Family Relationships, Intergenerational Solidarity, and Support Provided to Parents by
Their Adult Children,” 49 Journal of Gerontology S85 (1994).
14. The Confucian ideal. See Benjamin I. Schwartz, The World of Thought in Ancient China
71, 100-101 (1985); “Editor’s Preface,” in The Hsiao Ching v (Paul K. T. Sih, ed., 1961); cf.
Kyu-Taik Sung, “Motivations for Parent Care: The Case of Filial Children in Korea,” 34 Inter-
nationalJoumal of Aging and Human Development 109 (1992). For a vivid but possibly unreliable
description of Japanese filiality, see Ruth Benedict, The Chrysanthemum and the Sword: Patterns
of Japanese Culture 51-52, 101-102, 121 (1946).
15. See, for example, Laura J. Zimmer, “ ‘Who Will Bury Me?’: The Plight of Childless
Elderly among the Gende,” 2 Journal of Cross-Cultural Gerontology 61 (1987), esp. pp. 76-77.
On the importance of children as a source of support for their parents in poor societies, see the
careful empirical study, Daniel C. Clay and Jane E. vander Haar, “Patterns of Intergenerational
Support and Childbearing in the Third World,” 47 Population Studies 67 (1993). For evidence that
children are indeed treated better by their parents in cultures in which the parents look to their
children for old-age support, see Margaret F. Brinig, “Finite Horizons: The American Family,” 2
International Journal of Children's Rights 293 (1994). To the extent that family altruism is “two-
sided,” that is, children are altruikic toward their parents (as I have suggested is biologically
plausible) as well as vice versa, parents will be more generous to their children because more likely
to be repaid in their old age. See Peter Rangazas, “Human Capital Investment in Wealth-
Constrained Families with Two-Sided Altruism,” 35 Economics Letters 137 (1991).
16. Isaac Ehrlich and Francis T. Lui, “Intergenerational Trade, Longevity, and Economic
Growth,” 99 Journal of Political Economy 1029, 1046 (1991). This is on the assumption that the
increase in longevity does not result merely in a higher fraction of children surviving to adulthood,
but, also or instead, in an increase in the likelihood of reaching, or the length of, old age.
17. Richard A. Posner, Sex and Reason 405 -406 (1992), and references cited there.
210 The Theory Elaborated and Applied

In nineteenth-century America, childless elderly persons would sometimes

contract for nursing and other services in exchange for a promise to will
property to the caretaker.'*
A society’s location on the static versus dynamic continuum is not a
completely exogenous factor in the status of the old. One can imagine a
society—perhaps mandarin China was it—in which the old enjoy a domi¬
nant influence because the politically dominant groups in the society want
the society to remain static. Just as the pace of social change might be
slowed by a powerful judiciary employing the characteristically backward¬
looking techniques of legal reasoning, so might it be slowed by having old
people occupy the major political and social positions. This is doubtless a
factor in the gerontocratic structure of governance of the Roman Catholic
and Mormon churches.
I have said that a low ratio of old to young helps the old by reducing
the per capita cost of supporting them. A further point is that the smaller
the proportion of the population that survives to an advanced age, the more
likely society is to attribute extraordinary powers to those who do survive.
The tougher the obstacle course the old have had to run in order to survive
into old age, moreover, the sturdier they are likely to be when they get
there; this selection effect will strengthen the impression that very old
people have exceptional powers. But it is essential to distinguish between
exogenous and endogenous influences on the ratio of old to young. The
ratio might be low because of the prevalence of diseases to which the old
were particularly vulnerable'^ or because the society had decided not to
support its old people. Obviously if the ratio were low for the latter reason
it would not be evidence that the society had a high regard for the old—
unless, perhaps, the old were willing to sacrifice themselves for the sake of
the young, as in fact they seem to be in some primitive societies.

The bearing of agriculture. We can also expect the status of the old
to be influenced by the character of the society’s economy, and, specifically,
to be higher in an agricultural society than in a hunter-gatherer society.^®
An agricultural society is likely to have, other than in times of famine

18. See, for example, Slater v. Estate of Cook, 67 N.W. 15 (Wis. 1896); Brady v. Smith, 28
N.Y. Supp. 776 (Super. Ct. 1894); Stockley v. Goodwin, 78 Ill. 127 (1875).
19. Conversely, we saw in chapter 2 that the Black Death in medieval Europe, by sparing the
old, raised the dependency ratio.
20. For evidence, see Gordon E. Finley, “Modernization and Ageing,” in The Elderly Popu¬
lation in Developed and Developing World, note 8 above, at 87. Among primitive societies, ger-
onticide is far more common in hunter-gatherer economies than in agricultural economies. Albert
and Cattell, note 4 above, at 225.
Status of the Old 211

(when we can expect the old to fare very poorly), a larger food output per
productive perst)ri' and this will reduce the cost, in foregone consumption,
of maintaining an old person.^' If the economy is sufficiently productive,
the old may have been able to save enough from current income when
young to support themselves when old, reducing to zero the cost imposed
on other people, at least if the others do not consider confiscating the sav¬
ings of the old an attractive policy, because of the last-period problem.
The part of the output of an agricultural society that is not required for
the subsistence of the producers also enables the support of specialists in
activities, such as dispute resolution, religion, and magic, that are within
the capacity of the old to perform. Since, moreover, agricultural societies
are stationary rather than nomadic, the reduced mobility of the old imposes
a lesser cost on the young; they don’t have to be carried from camp to camp.
In addition, the labor of the elderly is more valuable in such a society be¬
cause there are more light agricultural tasks (such as tending sheep) than
there are light hunting-gathering tasks, and the old can perform light but
not heavy labor. Many old people in our society love to garden, and horti¬
cultural tasks are broadly similar to agricultural ones.^^
An agricultural economy has a further significance for the status of the
old beyond generating a surplus for consumption by the elderly and provid¬
ing them opportunities for employment. An agricultural economy revolves
around a stock of durable assets, namely arable land, and the control and
devolution of that stock present difficult problems from the standpoint of
political stability and economic efficiency. The story of King Lear is in¬
structive. Because of his advancing age, Lear wishes to divest himself of
responsibility for his kingdom (“unburden’d crawl toward death,’’ as he
puts it), and he proposes therefore to divide the kingdom among his three
daughters, reserving a right of support for himself. But a division of the
kingdom is a recipe for civil war, as King Lear learns to his sorrow. (We
can see here a metaphor for the inefficiency of breaking up a large estate
into small farms.) And the right to support that he had reserved turns out to
be unenforceable.

21. Though, as Malthus argued, much of the food surplus may go to supporting a larger
population, without significantly increasing per capita wealth.
22. Even today, labor-participation rates are higher in agricultural economies because agri¬
culture provides more employment opportunities for elderly persons than most other economic
activities do. Robert L. Clark and Richard Anker, “Cross-National Analysis of Labor Force Par¬
ticipation of Older Men and Women,” 41 Economic Development and Cultural Change 489
(1993). In the United States, a disproportionate number of employed elderly men are farmers or
farm laborers. Herbert S. Fames and David G. Sommers, “Shunning Retirement: Work Experience
of Men in Their Seventies and Early Eighties,” 49 Journal of Gerontology SUl, S122 (1994).
212 The Theory Elaborated and Applied

The interest in keeping an agricultural estate intact across the genera¬

tions, or even within a generation, in a society in which contracts may be
difficult to enforce, argues for simple rules of property, such as the eldest
male owns everything, and mechanical rules of succession, such as that the
property passes intact at the owner’s death to his eldest son. (The parallel
to hereditary monarchy is obvious.) Such rules favor the old by concentrat¬
ing the ownership of property in their hands. In a society in which the
economic rent of land is a large fraction of all income, owners of property
will not experience a decline in income when they get old, unlike the situ¬
ation in a modern economy, where most people’s most valuable capital
assets are various forms of human capital, which depreciates with age. The
patriarchal principle is thus a natural though not an inevitable corollary of
an agrarian economy in which the costs of transactions are inherently high
because of the absence of a commercially sophisticated legal system and in
which, therefore, the rules governing ownership and transfers must be kept
simple.
So in a society at once agricultural and preliterate (and hence almost
certain both to be static and to lack a commercially sophisticated legal sys¬
tem), we have optimum conditions for old people to attain a high social
status. If status increases monotonically with age, we may get ancestor wor¬
ship, the extrapolation of a positive correlation between age and impor¬
tance beyond the lifetime. Ancestors are not only the very old; they are also
predecessors. Predecessors are apt to be venerated more the more static the
society; their contributions are less likely to be obsolete. So a static so¬
ciety is likely to value ancestors both directly, as specimens of the (very)
old, and indirectly, as predecessors whose contributions have not faded
with time.
But before concluding that the situation of the old in the type of pre¬
modern society that I have been describing is optimal, we must remind
ourselves not only of the poverty by our standards of everybody in such a
society, young and old, but also of the likelihood of a severe maldistribu¬
tion of wealth and status among the old. If the ownership of land is concen¬
trated, the patriarchal class will be small even relative to the small total
number of old people in a premodern society; most old people may be
condemned, therefore, to extreme penury. Indeed, the wealth and power of
the patriarchal class may engender resentment against old people in general
(this is especially likely in societies that permit polygyny, which favors
wealthy older men), and the resentment may be taken out against the weak¬
est of them.
I must defend my assumption, heretofore tacit, that there is some
Status of the Old 213

mechanism by which policies that are beneficial to a society are adopted

by it. Democratic societies have such mechanisms, though they are subject
to deformation by pressure from interest groups; so, for that matter, do
nondemocratic societies, provided they have a government. The issue of
the mechanism is more challenging when one is speaking of a prepolitical
society, a society that lacks formal institutions of government. How does
the fact that old people might have a valuable role to play in the intergen-
erational transmission of useful knowledge get translated into arrange¬
ments for supporting old people rather than letting them starve when they
get too old to pitch in? I do not have the answer. But it is enough for my
purposes to point out that prepolitical societies do have extensive, law-like,
apparently socially beneficial or functional customs. However deep the
mysteries of the provenance and persistence of such customs may be, they
are no greater with regard to the customary treatment of old people than
with regard to the other customs of such societies.

An empirical study. We can try to test the economic model of the

status of the elderly in primitive and early societies more systematically
with the aid of data on 71 primitive societies culled by Leo Simmons from
the Human Relations Area Files, a compendium of anthropological and
ethnographic materials.Simmons reported data on more than a hundred
variables, including several that measure or proxy the status of elderly men
(Simmons’s data on the status of elderly women are meager), along with
many demographic, political, and economic variables. The tables that fol¬
low attempt to group the variables in meaningful categories and to display
the relations among them.
Preliminary tables not published here reveal strong positive correla¬
tions on the one hand among hunting, gathering, and fishing, and on the
other hand between agriculture and herding, as modes of production. These
correlations enable societies to be classified according to the relative pre¬
dominance of these two polar classes of modes of production. Table 9.1
correlates the status of the old with the society’s mode of production, rang¬
ing from “AGR” (predominantly agricultural or herding, or some combi¬
nation of the two) to “HUNT” (predominantly hunting, fishing, or gather¬
ing, or, again, some combination). The middle row and middle column
represent intermediate categories. The percentages are the percentages of

23. See Simmons, note 3 above. Simmons’s bibliography lists 336 books and articles as the
sources for his data. Id. at 294-308. The empirical study whose results I summarize here was
conducted under my direction by my research assistant Mark Fisher, and he deserves the principal
credit for such merit as the study may have.
214 The Theory Elaborated and Applied

Table 9.1 Status of Elderly Men as a Function of the System

of Production

AGR Mixed A/H HUNT

High 40% 43% 38%

(8) (9) (8)

Medium 35% 33% 24%
(7) (7) (5)
Low 25% 24% 38%
(5) (5) (8)

Table 9.2 Status of Elderly Men as a Function of the Social

System

INDIV Mixed I/C COMM

High 38% 35% 50%

(6) (11) (8)
Medium 25% 39% 25%
(4) (12) (4)
Low 38% 26% 25%
(6) (8) (4)

the particular class of society (classified by mode of production) in each of

the status categories. (The numbers in parentheses are the number of soci¬
eties; they do not sum to 71 because of missing data.) As expected, the
status of the old is higher in a higher fraction of societies with much or
some agriculture (columns 1 and 2) than in societies with a strong pre¬
dominance of hunting, fishing, and (or) gathering. For example, in only
25 percent of the agricultural societies, but in 38 percent of the hunting
societies, do the elderly have a low status. But the correlations are weak.
The next two tables correlate the status of the elderly with the type of
social system—whether it emphasizes individuality (“INDIV”) or com¬
munity (“COMM”) (table 9.2), and whether market or nonmarket methods
of resource allocation (“MKT” or “NONMKT”) predominate (table 9.3).
It is not surprising that the elderly tend to have a high status in societies
that emphasize communal values, since one expects the family to be a
strong institution in such societies. It may seem surprising, however, that
they also tend to have a high status in societies that emphasize nonmarket
rather than market allocation of resources. Societies that emphasize market
Status of the Old 215

Table 9.3 Status of Elderly Men as a Function of the Eco¬

nomic System

MKT Mixed M/N NONMKT

High 25% 50% 41%

(5) (8) (9)
Medium 25% 44% 32%
(5) (7) (7)
Low 50% 6% 27%
(10) (1) (6)

Table 9.4 Mode of Production as a Function of the Social

System

INDIV Mixed I/C COMM

AGR 50% 33% 19%

(9) (11) (3)
Mixed A/H 28% 33% 38%
(5) (11) (6)
HUNT 22% 33% 44%
(4) (11) (7)

allocation might be expected to have a more secure system of property

rights, enabling people to accumulate property and retain it in their old age.
But without the free and easy transferability of property rights that is a
hallmark of a market economy, a patriarchal system of land ownership may
be, as suggested earlier, the only efficient system of property rights in an
agrarian society’s most important resource. The relative weight of these
opposed factors is an empirical issue.
A potential problem with all these correlations is that if the indepen¬
dent variables (mode of production, social system, and economic system)
are correlated with each other, the effects of each one on the status of the
aged may be difficult to separate out. For example, if agricultural societies
tend to emphasize communal values, it would be unclear whether the posi¬
tive correlation between the status of elderly men on the one hand and the
predominance of agricultural and communal features in the social land¬
scape on the other was due to agriculture or to communal values; or it might
be that the latter came from the former, and so was not an ultimate cause
of a high social status of elderly men. In fact, as shown in tables 9.4 and
216 The Theory Elaborated and Applied

Table 9.5 Mode of Production as a Function of the Economic

System

MKT Mixed M/N NONMKT

AGR 58% 30% 14%

(14) (6) (3)
Mixed A/H 25% 55% 29%

(6) (11) (6)

HUNT 17% 15% 58%
(4) (3) (12)

9.5, there are strong correlations between mode of production on the one
hand and the social and economic system of the society on the other. The
agricultural (and herding) mode of production is correlated with the pre¬
dominance of individual and market values, the hunting (and gathering and
fishing) mode with the predominance of communal and nonmarket values.
Since the agricultural mode is negatively correlated with both communal
and nonmarket values yet positively correlated with the status of elderly
men, we can infer that the likely cause of the positive correlation is not any
of the values associated with an agricultural society but the fact that such a
society generates a surplus that makes it less costly on a per capita basis to
support the aged; incurs fewer costs in transporting the elderly, because its
way of life is less nomadic; and can make more productive use of the
elderly.
It might be thought that the status of the aged would be higher in a
warrior society than in a pacific one. The aged would, it is true, have little
value in actual combat, but warfare would tend to weed out the weaker
men, so that the men who survived into old age would tend to be ones who
had been successful warriors. They would be prestigious survivors, ad¬
mired for their deeds (Othello); their memories of earlier wars would pro¬
vide a valuable stock of information for meeting fresh military challenges
(Nestor); and their experiences and maturity would make them valued as
leaders and counselors (Nestor again). Table 9.6, in which incidence of
warfare is correlated with treatment of the aged, provides some, but rather
slight, support for the hypothesis that the aged tend to be better treated in
the more warlike societies.
Not only are the correlations in table 9.6 weak, but they may reflect the
fact that there is a positive correlation between the incidence of warfare and
the presence of agriculture, as shown in table 9.7. Presumably this is be-
Status of the Old 217

Table 9.6 Status of Elderly as a Function of the Incidence of

Warfare

High Medium Low

Incidence Incidence Incidence

High status 43% 29% 46%

(13) (5) (6)
Medium status 33% 35% 23%
(10) (6) (3)
Low status 23% 35% 31%

(7) (6) (4)

Table 9.7 Incidence of Warfare as a Function of the Mode of

Production

High Medium Low

Incidence Incidence Incidence

Societies with 74% 42% 33%

agriculture (26) (8) (4)
Societies 26% 58% 66%

without agri¬ (9) (11) (8)

culture

cause there is better plunder in an agricultural society and also more surplus
to support warriors. We have already seen that the status of the aged tends
to be higher in agricultural than in nonagricultural societies. A reason in
addition to those discussed earlier may be that an agricultural society tends
to engage in warfare a lot, and warfare is a positive factor in the status of
the elderly.
Last, table 9.8 explores the relation between the status of the aged
and the prevalence of polygyny. The pattern is nonlinear. The status of
elderly men tends to be high in societies in which polygyny is either fre¬
quent or nonexistent, but low in societies in which there is some but not
much polygyny. Polygyny favors older men because it facilitates their
employing their accumulated resources to compete with young men for
young women.^'* It is therefore resented by young men. If polygyny is com-

24. Who in turn become the supports of the men in their old age. Nugent, note 13 above, at
80-81.
218 The Theory Elaborated and Applied

Table 9.8 Status of Elderly Men as a Function of the Preva¬

lence of Polygyny

Polygyny Some No
Frequent Polygyny Polygyny

High status 46% 18% 50%

(12) (3) (9)
Medium status 38% 35% 22%

(10) (6) (4)

Low status 15% 47% 28%

(4) (8) (5)

mon, the resentment is exacerbated, but the fact that it is common suggests
that older men have a commanding position in the society, implying high
status. If polygyny is nonexistent, this source of resentment is removed,
while if there is some but not much polygyny, this implies that older men
do not have the power to universalize the institution, and yet its very exis¬
tence will be an affront to young men. So the status of older men might
indeed be lowest in such societies, as table 9.8 suggests.
Because of the small size of the samples, the unreliability of much of
the data, and difficulties of classification (as in deciding whether a society
has “frequent” or merely “some” polygyny), the empirical results pre¬
sented in this chapter are merely suggestive. But they do provide some
grounds for thinking that the rational model of human behavior may have
considerable applicability to primitive societies in general and to the treat¬
ment of their aged members in particular.

Modernity and Symbolic Status

The transition to a modern mixed economy with mass education reduces
the social worth of the elderly. Mass education reduces the value of old

25. For other examples of the application of the rational-choice model to primitive societies,
see Gary S. Becker and Richard A. Posner, “Cross-Cultural Differences in Family and Sexual Life:
An Economic Analysis,” 5 Rationality and Society 421 (1993), and references cited there; Bruce
L. Benson, “Legal Evolution in Primitive Societies,” 144 Journal of Institutional and Theoretical
Economics 772 (1988); Vernon L. Smith, “The Primitive Hunter Culture, Pleistocene Extinction,
and the Rise of Agriculture,” 83 Journal of Political Economy 727 (1975). On the high status of
the elderly in static agricultural societies, see the careful empirical analysis of rural India in
Mark K. Rosenzweig, “Risk, Implicit Contracts and the Family in Rural Areas of Low-Income
Countries,” 98 Economic Journal 1148, 1168 (1988).
Status of the Old 219

people’s memories; industrial labor places physical demands on workers

that old people cannot meet; advances in medicine, nutrition, and sanitation
prolong life and by doing so increase the dependency ratio; and the greater
dynamism of the modern economy exacerbates the adverse impact on the
productivity of the old caused by the age-related decline in fluid intelli¬
gence and in investment in human capital. It is true that education and
literacy might, by enabling the young to learn what the old already know,
reduce the mutual incomprehension of the generations. But if society is
changing rapidly, the rational young may not be much interested in what
the old know; the value of that knowledge may have depreciated to nothing.
It is also true that in a postindustrial economy such as ours the physically
demanding and dangerous jobs characteristic of industrialization (such as
railroad and factory work) tend to give way to light service jobs, and the
tendency expands the opportunities of elderly workers. Many service jobs,
moreover, put a premium on relational human capital, which depreciates
less slowly than other forms of human capital. But these effects may be
offset by the increased rapidity of technological change associated with the
postindustrial economy, since it is more difficult for older workers to ac¬
quire new human capital.
Paradoxically, age grading reasserts itself in a modern society, though
it is social rather than occupational. In a patriarchal family, even in its
modified nineteenth-century form (Thomas Mann’s Buddenbrookses or
John Galsworthy’s Forsytes), there is constant intergenerational associa¬
tion, and often coresidence, within a family. This changes with moder¬
nity, as the increasing pace of cultural and technological change that is
associated with modernity drives the generations apart, making separate
communities of different age groups, particularly the old.^*’
Yet the picture of the impact of modernization on the status of the old
that I have limned so far is severely incomplete. The factors adverse to the
position of the elderly in a modern society are offset by, first and foremost,
the enormous increase in productivity that modern social and economic
arrangements, including mass education and technological innovation,
make possible. That increased productivity greatly facilitates making ade¬
quate provision for consumption by the nonproductive. For example, it
enables the individual to reallocate consumption from youth to old age
without great sacrifice, because his level of consumption is so high.
The political arrangements of modern societies tend also to work in

26. The theme of Arlie Russell Hochschild, The Unexpected Community (1973), a distin¬
guished sociological field study of old age in America. See esp. ch. 4. See also Howard P. Chuda-
coff, How Old Are You? Age Consciousness in American Culture (1989).
220 The Theory Elaborated and Applied

favor of the old. Political democracy, the characteristic if far from uni¬
versal regime of modern societies, is sometimes thought to be simply a
civilized, or an economist might say cheap, method for registering power—
civilized and cheap because it does not require that the powerful exercise
their muscle to prove they really are the powerful. This was the view of
Oliver Wendell Holmes, Jr. and James Fitzjames Stephen,^^ among others.
It is an oversimplification. Universal adult suffrage and the secret ballot
combine to confer political and therefore social and economic power on
naturally weak groups, provided they have sufficient education to be moti¬
vated to vote and sufficient agreement on objectives to constitute an effec¬
tive bloc. This describes the old in the wealthy nations of Europe and North
America today. Numerous, and adequately educated, for reasons largely
although not entirely exogenous to their electoral power, the old in these
nations have enough political power to obtain wide-ranging governmental
support and protection,^^ even though in a state of nature or under a differ¬
ent political system they would have little or no power. Yet even before the
old became a politically powerful group, increased family wealth shielded
most of them from poverty during the industrializing era of U.S. history.
“In contrast to an assumption common to both observers in the past and
present, the great majority of aged persons [in the United States] have never
been impoverished or isolated.”
We must probe more deeply the ambiguous term “status,” distinguish¬
ing in particular between pecuniary income, political power, health, and
longevity, on the one hand, and affection, respect, and veneration, on the
other. These two aspects of social status—the material and the symbolic—
may, in the case of the old, actually be negatively correlated in a modern
society.3“ The reason is related to the earlier point about the inculcation of
filial piety. The more the old are emancipated by their political power from
dependence on the young, the less they have to gain, when raising their
children, from inculcating filial piety. So the less such piety their children

27. Holmes, “The Gas-Stokers’ Strike,” 7 American Law Review 582 (1873), reprinted in
The Essential Holmes: Selections from the Letters, Speeches, Judicial Opinions, and Other Writ¬
ings of Oliver Wendell Holmes, Jr. 120 (Richard A. Posner, ed., 1992); Stephen, Liberty, Equality,
Fraternity 70 (1961 [\S73]).
28. We shall consider in chapter 11 whether they have received “too much” support and
protection in some intelligible sense.
29. Carole Haber and Brian Gratton, Old Age and the Search for Security: An American
Social History 172 (1994). Haber and Gratton’s book (especially chapter 2) provides convincing
documentation for this conclusion, to which I return in chapter 11.
30. Cf. Aaron Lipman, “Prestige of the Aged in Portugal: Realistic Appraisal and Ritualistic
Deference,” 1 Aging and Human Development 127 (1970).
Status of the Old 221

will feel. Also, it becomes harder to feel sorry for old people as their in¬
comes rise, though an offsetting consideration is that the old are less likely
to be—and to be resented as—a financial albatross to their children. With
young people better off, moreover, the old are less willing to make sacri¬
fices for them; so the young have less to be grateful for to the old.
Not only are the benefits of cultivating filial piety fewer when parents
do not expect to look to their children for support in their old age; the
benefits of having children are fewer. We can therefore expect smaller
families as filial support to aged parents declines.^' This shrinkage in¬
creases, in turn, the cost to each of the children of providing care to their
aged parents, since there are fewer of them to spread the cost among. An
equilibrium involving support for the elderly by both the state and the
family may therefore be precarious, if state support leads to smaller fami¬
lies, which in turn increases the cost to families of supporting their elderly
members. An important factor is the growing participation of women in the
labor force, which is both effect and cause of shrinking families. In most
societies, daughters and daughters-in-law are the primary family caregivers
to the elderly,^^ partly at least because their opportunity costs of time are
lower than those of men as a consequence of limited opportunities in the
market. Women’s opportunity costs of family caregiving rise as the demand
for their services in the market rises. This increased demand also raises the
opportunity costs of having children, reinforcing the trend to smaller fami¬
lies which in turn, as I have noted, raises the cost to each child of providing
care for his or her elderly parents.
Large families are generally considered “warmer” and closer-knit
than small ones, and this should benefit the elderly members. But the
greater warmth of the large family may be a consequence in part of the fact
that less is demanded of each child—family obligations, while possibly
more extensive, are also more widely shared. And large family size may be
the product in part of a high value that is placed on close intrafamilial
relations in the environment in which the family finds itself. The causality

31. For evidence, see, for example, Alice Mannings, “Intergenerational Interdependence; A
Cross-Cultural Study of the Care of Elderly Parents,” in Heterogeneity in Cross-Cultural Psy¬
chology 561, 572 (Daphne M. Keats, Donald Munro, and Leon Mann, eds., 1988).
32. See, for example, Rhonda J. V. Montgomery and Yoshinoro Kamo, “Parent Care by Sons
and Daughters,” in Aging Parents and Aging Children 213, 216-217 (Jay A. Mancini, ed., 1989);
Hal L. Kendig and Don T. Rowland, “Family Support of the Australian Aged: A Comparison with
the United States,” 23 Gerontologist 643, 647 (1983).
33. For evidence that these factors erode the sense of filial obligation, see Nancy J. Finley,
M. Diane Roberts, and Benjamin F. Banahan, III, “Motivators and Inhibitors of Attitudes of Filial
Obligation toward Aging Parents,” 28 Gerontologist 73, 74, 77 (1988).
222 The Theory Elaborated and Applied

thus may run from warmth to large families rather than from large families
to warmth.
The undependability of filial obligation as a protection for the elderly
is evidenced by the fact that most states have laws requiring children to
support (if they can) their destitute parents.^"* These laws date back to the
sixteenth century in England, when the system of poor support was rudi¬
mentary and it was natural to require families to take care of their own
members to the extent possible. The laws have ceased to be of any impor¬
tance in this country, given social security and welfare; we shall consider
in chapter 11 the justice of allowing families to shift some of the burden of
supporting their elderly members to the taxpaying public. If my analysis of
filiality is correct, laws requiring children to support their parents would,
but for the existence of a generous social security program, be more need¬
ful today than they were in the sixteenth century.
In addition to being liked or at least reverenced less, both generally and
by their own younger relatives, the old in our society, being numerous, are
no longer such objects of fascination as they once were.^^ Octogenarians
used to be prized for their rarity; someone who lived so long seemed spe¬
cially blessed. Now that they are a dime a dozen they have ceased to fasci¬
nate. Selection bias is relevant here, as I have already suggested. The
tougher the obstacle course, the tougher the winners. This, along with scar¬
city, may explain the evidence that, on average, blacks respect old people
more than whites do.^^ Because blacks receive on average poorer medical
care than whites (and possibly for other reasons as well), a substantially
smaller fraction of blacks than of whites survive to old age.^'^ Their old
people thus are more rare than in the case of whites, and maybe tougher as
well,3* and therefore more impressive. The point has nothing to do with

34. Marvin B. Sussman, “Law and Legal Systems,” in Family and Support Systems across
the Life Span 11, 26-28 (Suzanne K. Steinmetz, ed., 1988).
35. The scarcity of the old as a factor contributing to the honor in which they are held is
stressed in David Hacked Fischer, Growing Old in America 29, 33 (1977).
36. See, for example, Finley, Roberts, and Banahan, note 33 above, at 77; Elizabeth Mutran,
“Intergenerational Family Support among Blacks and Whites: Response to Culture or to Socio¬
economic Differences,” 40 Journal of Gerontology 382, 388 (1985); cf. Amasa B. Ford et al.,
“Race-Related Differences among Elderly Urban Residents: A Cohort Study, 1975-1984,” 45
Journal of Gerontology SI 69 (1990); Colleen L. Johnson and Barbara M. Barer, “Families
and Networks among Older Inner-City Blacks,” 30 Gerontologist 726 (1990).
37. Of the adult (> 18) white population, 18.0 percent is over the age of 65 and 1.8 percent
over the age of 85; the corresponding percentages for blacks are 12.1 percent and 1.1 percent. U.S.
Bureau of the Census, Current Population Reports, ser. P-25, p. 2 (1991) (tab. 1). See also Jacque-
lyne Johnson Jackson, Minorities and Aging, ch. 4 (1980).
38. Donald S. Shepard and Richard J. Zeckhauser, “The Choice of Health Policies with
Heterogeneous Populations,” in Economic Aspects of Health 255, 308 (Victor R. Fuchs, ed..
Status of the Old 223

race as such. Edmund Wilson remarked that the “consecrated authoritative

role” that the nation assigned to Justice Holmes was due in part to “the
prestige of longevity when the ancient has retained his faculties.”
In a society as dynamic as ours, workers are probably becoming obso¬
lete at younger and younger ages; it is a factor in the falling age of retire¬
ment. And the political power of the old—a factor in their high material
status in our society—is a source of resentment to the young. The resent¬
ment is exacerbated by the fact that the greatly improved market opportu¬
nities of women have, as I have mentioned, increased their opportunity
costs of caring for the elderly members of their families. And the more the
rate of aging seems under conscious human control by virtue of advances
in medical understanding, the more the infirmities of old age seem almost
culpable, rather than inevitable. We are increasingly apt to think that a de¬
crepit old person is such because he failed to follow the advice of doctors
and nutritionists concerning a healthful style of living. The diminished
symbolic or prestige status of the old is thus the price they have paid for
their improved material status.
I have deliberately not emphasized the growing specialization of eco¬
nomic activities in modern economies as a factor in this diminution. It
is true that the more work is specialized, the fewer are the opportunities
for the old to make a productive contribution in a new field after altered
conditions of demand and supply have ejected them from their former
specialty. But a countervailing consideration is that, as I noted in the
preceding chapter, specialist work is easier for old people to perform
productively than generalist work. A generalist must be adept at adapt¬
ing to changed conditions; a specialist need only continue planing his
accustomed groove—though he may become bored more quickly than the
generalist.
What has been the impact of modernity on the distribution of income
or wealth within the ranks of the old, as distinguished from between old
and young? As I have already suggested, one would expect that in an
agrarian society, such as that of colonial America, some elderly men, es¬
pecially landowners, would be wealthy and powerful but many others

1982). Elderly blacks actually have a longer life expectancy, and possibly better health, than el¬
derly whites. Rose C. Gibson, “The Age-by-Race Gap in Health and Mortality in the Older Popu¬
lation: A Social Science Research Agenda,” 34 Gerontologist 454 (1994); Bert Kestenbaum, “A
Description of the Extreme Aged Population Based on Improved Medicare Enrollment Data,” 29
Demography 565,572 (1992); Ford et al., note 36 above, at S167-S168. This supports the “tough¬
ness” hypothesis.
39. “Justice Oliver Wendell Holmes,” in Edmund Wilson, Patriotic Gore: Studies in the
Literature of the American Civil War 743, 795 (1962).
224 The Theory Elaborated and Applied

would not be.'^° Even though the average income of the elderly relative to
that of middle-aged and younger people has increased since colonial times,
the position of the wealthiest old relative to the wealthiest middle-aged has
probably deteriorated. This change is due in part to estate taxation, which
creates incentives for wealthy people to distribute a large portion of their
estate before they die, and in part to the declining importance of land—a
form of wealth that does not depreciate with the age of the owner—relative
to human capital, a form of wealth that does decline with age. As wealth
shifts from land and other forms of physical capital to human capital, more¬
over, the ability of elderly parents to extract services from their children by
threatening to withhold bequests decreases;'*' human capital is not trans¬
ferred by bequest. Societies differ, incidentally, in the extent to which they
allow the use of threats to disinherit. In England and the United States, for
example, people are free to disinherit their heirs; in France, they are not.
We can expect that the more influential the elderly are in a society, the more
reluctant the society will be to limit disinheritance, a tool by which elderly
people can extract services from the young.

The Bearing of Ideology

I have emphasized the role of economic factors in influencing both the
material and the symbolic status of the old, but other factors may also be
important. For example, the difference between the ancient Greek and the
early Christian view of old age seems to reflect religious rather than eco¬
nomic factors, though underlying the religious differences may have been
an intensely practical difference between a warrior and a civilian outlook,
a difference that made the Greeks set a higher value on physique. On the
whole the Greeks were not so committed to mind-body or soul-body dual¬
ism as the Christians. They did believe that the spirit survived the death of
the body. But especially though not only in Homer and the tragedians, it
was a weak, pitiable, inglorious spirit, fit only for Hades. The Christian
conception of the soul (greatly influenced, to be sure, by a Greek—Plato)
endowed it with much greater dignity. Consistent with this difference,
Greek art celebrates the beauty of the body, and Christian art (until the

40. For evidence, see Fischer, note 35 above, ch. 1, esp. pp. 58-66.
41. Paul FI. Rubin, James B. Kau, and Edward F. Meeker, “Forms of Wealth and Parent-
Offspring Conflict,” 2 Journal of Social and Biological Structures 53 (1979). Not only can a
person threaten to cut off his heirs; he can, as noted earlier, make a legally enforceable promise of
a bequest to another person in exchange for a legally enforceable promise of support in his old
age. See note 18 above and accompanying text.
Status of the Old 225

Renaissance) apologizes for the body—depicting it as etiolated beneath

copious garments calculated to conceal its shape—and for bodily func¬
tions, such as eating and sex, notwithstanding the doctrine of bodily resur¬
rection. The more the physical aspect of man is prized, the greater the con¬
tempt and revulsion that old age is likely to produce, in the same way that
gourmets are more likely to be distressed by poor food than people for
whom the only function of eating is sustenance. The more the body is dis-
prized, the greater the reverence for old age, seen as the transitional period
in which the stripping away of the accidental and even shameful attributes
that human beings share with animals prepares the soul to meet God.
But we should not exaggerate the importance of religious factors in the
changing status of the old. Although the Christian Middle Ages provided a
dignified retirement for some old people—those received into monaster¬
ies—the lot of the run-of-the-mill old, especially those who lived in cities
or did not have living children (and there were many, owing to high death
rates among the young), was not a happy one. Society was not prepared to
devote substantial resources to their survival. A straw in the wind was the
charivari—the practice, widely condoned, by which young bachelors ha¬
rassed widowers who remarried, protesting against the tendency of wealthy
older men to monopolize the marriageable women.^^
Darwinism may have had a greater impact than Christianity on the
status of elderly people. Before Darwin, it was common to think that the
world was regressing rather than progressing. The present was thought to
mark a decline from a Golden Age located in the dim past; and nonhuman
primates, rather than being thought of as man’s predecessors, were thought
to be degenerate versions of man. In effect the outlook was that of the
typical elderly person—the world is going to the dogs (see chapter 5). A
society in which such an outlook was common would provide a congenial
environment for the elderly, while a society such as ours in which progress
is anticipated by most people will be more congenial to the young because
they are more forward-looking and optimistic.
David Fischer, in his comprehensive study of changing attitudes in
America toward old age, identifies the period 1770-1820 as a crucial pe¬
riod of change.^^ Before then, Americans venerated the old, even to the

42. See, for example, Natalie Zemon Davis, “The Reasons of Misrule: Youth Groups and
Charivaris in Sixteenth-Century France,” 50 Past and Present 41 (1971). The high death rate of
women in childbirth during the Middle Ages enabled some men to practice a form of serial po-
lygyny despite the prohibition against divorce.
43. Fischer, note 35 above, ch. 2. His thesis has not gone unchallenged. For a summary of
the criticisms, see Haber and Gratton, note 29 above, at 5-8.
226 The Theory Elaborated and Applied

point that American men exaggerated their age and dressed to look older
than they were. Afterward, youth was celebrated, people tried to look
younger rather than older than their chronological age, mandatory retire¬
ment was imposed on (nonfederal) judges, and the valence of words refer¬
ring to old people, such as “gaffer,” changed from honorific to pejorative.
During this period the nation was changing in ways consistent with the
economic explanation of attitudes toward age. The nation was becoming
less agrarian and more industrial and old people were becoming more plen¬
tiful. But Fischer argues that the reversal of attitudes toward old age was
too abrupt to be explained by these very gradual economic and demo¬
graphic trends. He attributes the reversal to the libertarian and egalitarian
ideas of the American and French Revolutions.'*^ Americans, he thinks, be¬
came restive with traditional hierarchies, including that of age. Consistent
with his conjecture, it has been argued that the extreme emphasis on filial
piety in Confucian China was motivated in part by hope that habits of
deference and subordination inculcated in the family would radiate into
people’s political attitudes and behavior.'*^
Even today, and even if attention is confined to the wealthy nations,
the status of elderly people differs across nations; the analysis in this
chapter has pointed to possible explanatory factors. Might it be possible
to develop an empirical test of the relative status, or influence, of the el¬
derly in different nations? Perhaps so. We might expect that, other things
being equal, “young selfish” societies, in which the preferences of old
selves are given little weight, would have higher social discount rates
than societies in which old selves are given equal or greater weight. The
latter societies, to the extent that they were guided by a consistent con¬
ception of the public interest rather than tugged hither and yon by com¬
peting interest groups, would presumably use lower social discount rates in
evaluating projects having deferred payoffs, for example certain kinds of
environmental projects. A potentially important qualification, however, is
that the truncated horizon of “old selfish” societies may cause them to
discount very drastically the benefits of projects that will not come “on
line” until the current young are old, since by then the current old will be
dead. So we might expect that “old selfish” societies would have lower
social discount rates than “young selfish ones” but that the gap would
narrow (and the lines might even cross) the more distant the future being
discounted. *

44. Fischer, note 35 above, at 108-112.

45. See Schwartz, note 14 above, at 100-101.
Status of the Old 227

The Institutional Life Cycle

Fischer’s analysis relates attitudes toward aging to institutional age.
Generalizing, we might conjecture that a “young” nation, perhaps any
“young” institution, will orient itself toward the values of youth rather than
of age and therefore that the age of an institution and the age of its leaders
will be positively correlated. It is a plausible conjecture. Revolutionary po¬
litical leaders, for example, generally are much younger than other political
leaders.'^® Most of our “founding fathers” were young men, though the ap¬
pellation suggests a persisting respect for mature wisdom. Other examples
of the correlation between individual and institutional age are the generally
youthful leadership of young companies in young industries, such as the
computer-software industry, and the generally old leadership of declining
institutions, such as American labor unions. Established religious sects,
ranging from Orthodox Judaism to the Roman Catholic Church, tend to
have old leaders, new sects young ones.
A young in the sense of a new organization entails risk-taking, because
the mortality rate of young organizations is high,'*’' much like that of infants
before modern medicine. Since older people have more trouble than
younger people in finding new jobs, the risk of having to search for a new
job because one’s present employer has folded is more costly to older than
to younger employees. So we can expect new firms to attract more young
workers than old ones. And for two other reasons as well. A new organi¬
zation is apt to require new skills and ideas, which tend to be properties of
the young. And a new organization if successful will be growing'**—for it
is unlikely to have reached its mature size at the date of initial entry—and
a growing organization is likely to have a younger age distribution than a
static or declining one. For it will be hiring more new employees; and most
hires are of young people, the older having firm-specific human capital that
would be wiped out by their switching to new jobs. So if young organiza¬
tions tend to be growing organizations, and old ones tend to be static be-

46. Dean Keith Simonton, Genius, Creativity, and Leadership: Historiometric Inquiries
102-103(1984).
47. Boyan Jovanovic, “Selection and the Evolution of Industry,” 50 Econometrica 649
(1982); Howard Aldrich and Ellen R. Auster, “Even Dwarfs Started Small: Liabilities of Age and
Size and Their Strategic Implications,” 8 Research in Organizational Behavior 165, 177 (1986)
(tab. 1); Michael T. Hannan and John Freeman, “The Ecology of Organizational Mortality: Ameri¬
can Labor Unions, 1836-1985,” 94 American Journal of Sociology 25, 32-33, 42 (1988).
48. For evidence that young firms grow faster than old ones, see David S. Evans, “Tests of
Alternative Theories of Firm Growth,” 95 Journal of Political Economy 657 (1987). See also
Jovanovic, note 47 above.
 228 The Theory Elaborated and Applied

cause they have grown to a size where further growth would encounter
diseconomies of scale, young organizations will have a higher percentage
of young employees independently of the different qualities or attitudes of
young and old workers. The causality can run in both directions, however:
a firm may be growing because it has a youthful age distribution, or static
because it does not.
The analogy between the individual and the institutional life cycle
must not be pressed too hard. The life-cycle theory of the firm, proposed
by Michael Spence and other economists, predicts that young firms will
seek to maximize revenue and capacity in an effort to obtain permanent
advantages of cost or demand over competitors and to deter new entrants
That theory has no fruitful applications to aging that I can see. And “learn¬
ing by doing” has different implications for the individual than for the col¬
lective; can explain, for example, the secular improvement in athletic re¬
cords without implying that the individual athlete will continuously break
his own records.50 Furthermore, although the “birth,” “growing,” and
“prime” stages in the growth of a living person have close counterparts in
organizations, the life-cycle analogy fails completely when we consider the
“elderly” organization. There is no reason in principle why organizations
cannot be for all practical purposes eternal; consider the Roman Catholic
Church, now almost two millennia old. It may be that as an organization
grows larger it becomes less adaptable to a changing environment be¬
cause the chain of communication that connects stimulus with response is
longer.^' Yet this negative effect of size and hence of age may be offset by
the greater hardiness of a large organization, due in part to its superior
ability to diversify, compared to a small one. In any event, my interest here
is in the pure effect of age on survival rather than in the effect that is due to
the fact that age is positively correlated with size; and it appears that, alto¬
gether unlike the human situation, that effect is positive. Old firms, unlike

49. See, for example, Joseph H. Anthony and K. Ramesh, “Association between Accounting
Performance Measures and Stock Prices: A Test of the Life Cycle Hypothesis,” 15 Journal of
Accounting and Economics 203 (1992), and studies cited there. On life-cycle theories of organi¬
zations generally, see, for example, Douglas D. Baker and John B. Cullen, “Administrative Reor¬
ganization and Configurational Context: The Contingent Effects of Age, Size, and Change in
Size,” 36 Academy of Management Journal 1251 (1993); Herbert Kaufman, Time, Chance, and
Organizations: Natural Selection in a Perilous Environment (1985), esp. ch. 4.
50. See William Fellner, “Specific Interpretations of Learning by Doing,” 1 Journal of Eco¬
nomic Theory 119 (1969).
51. Michael T. Hannan and John Freeman, “Structural Inertia and Organizational Change,”
49 American Sociological Journal 149, 163 (1984); Aldrich and Auster, note 47 above, at 169.
See generally Jitendra V. Singh and Charles J. Lumsden, “Theory and Research in Organizational
Ecology,” \ fy Annual Review of Sociology 161, 168-169, 180-182(1990).
Status of the Old 229

old human beings, have a higher likelihood of continued survival than

young ones.^2 This is why they are more attractive to older workers.
We should distinguish between workers (up to middle management)
and leaders. Although old firms in decline (hence not hiring any or many
new workers) are likely to have old workers, their leaders will sometimes
be young. It may be that the decline of the firm can be reversed only by
turning the firm in a completely new direction; and if, thus, the firms needs
a “revolution,” it will need revolutionary leaders, and they tend to be
young, as we have seen. Also, if a firm seems likely to fail anyway, the
expected cost of bad leadership may be much less than the expected benefit
of good leadership. In that event, recruiting an older leader, one who has
survived a patient screening of future leaders by being rotated through a
variety of gradually more responsible jobs in the corporate hierarchy, may
not be the best strategy. This screening method is designed more to weed
out persons who might fail in top leadership and by failing wreak heavy
damage on a stable and successful enterprise than to identify the person
most likely to rise to a daunting challenge requiring bold, original thought.
If individuals and institutions age, so can whole fields of scholarship
or creativity. Kuhn’s distinction between revolutionary and normal sci¬
ence —the first involving a paradigm shift, the second working within a
paradigm—is germane. We might say, translating Kuhn into age-speak,
that a field is new, or young, or renewed, or rejuvenated, when a new para¬
digm (Copernican, Newtonian, or Einsteinian cosmology, for example)
emerges; is middle-aged when scientists work incrementally within an un¬
challenged paradigm; and is old when the accumulating anomalies of the
current paradigm make it ripe to be overthrown. We would expect a science
in its revolutionary phase or phases to be attractive to the young, because
older scientists will incur higher costs of developing and adapting to a new
paradigm, and normal science to be attractive to older scientists, whose
accumulated knowledge will give them an edge in dealing with problems
within its domain.
Admittedly, the causality runs in both directions; a science that hap¬
pens to attract young persons is more likely to experience a paradigm shift.
But not every science, or other field of endeavor, is at all times ripe for a

52. Michael T. Hannan and John Freeman, Organizational Ecology, ch. 10 (1989); David S.
Evans, “The Relationship between Firm Growth, Size, and Age: Estimates for 100 Manufacturing
Industries,” 35 Journal of Industrial Economics 567 (1987). It is true that infants in most societies
have had high death rates, but the comparison in the cited study is not infant versus mature, but
younger versus older, firms.
53. See Thomas S. Kuhn, The Structure of Scientific Revolutions (2d ed. 1970).
230 The Theory Elaborated and Applied

paradigm shift. Constraints deriving from the relation between existing re¬
sources (money, techniques, and so forth) and the natural or social phe¬
nomena that the field seeks to explain often make paradigm shifts infea¬
sible in particular fields at particular times. We should expect such fields to
be less attractive to the young at such times, especially to the adventurous
and creative young. These fields, if they want to maintain their staffing, will
have to accept less creative applicants.

My emphasis in this chapter has been on two senses of social status, the
material and the honorific. We have seen, with reference to the elderly, that
the two senses need not coincide. In primitive societies they tend to coin¬
cide, but the resulting status of the elderly can be very high or very low, or
anywhere in between. The association of reverence for the elderly with
primitive man is spurious; many primitive societies kill, abuse, or seriously
neglect their elderly. I presented some evidence that a key variable in the
power and respect accorded elderly people in premodern societies is the
degree to which the society is agrarian, because the value of the old is, for
a variety of reasons, greater in agrarian societies than in other societies at
the same level of development.
In modern societies, notably that of the United States, the material and
honorific status of the elderly tend to diverge. The material status of elderly
Americans is without historical precedent and is high relatively as well as
absolutely, in part because of the tendency of democratic politics to am¬
plify the power of “naturally” weak groups. But the honorific status of
elderly Americans is lower today than it was when the nation was founded.
Plausible explanatory factors are numerous. One is the increased rate of
social and technical change, which because of the characteristic resistance
of the elderly to novelty has widened the gap between the generations. Oth¬
ers are the diminishing incentive of parents to instill filial piety in their
children (because the parents’ welfare in old age is no longer crucially de¬
pendent on their families), the greatly increased number and proportion of
elderly persons in the population, the decline of coresidence, and the
shrinking size of families. That shrinkage, in conjunction with the in¬
creased number of elderly persons and the increased market opportunities
of women, has increased the cost of personal caregiving to old by young
family members, making the maintenance of close relations with elderly
family members increasingly irksome. Cutting the other way, however, is
the greater wealth of today’s old people, which makes them less of a finan¬
cial burden to the young and increases the probability of a bequest or other
gifts to the young.
Status of the Old 231

I took a brief look at institutional aging. Although old age has no clear
counterpart in the life cycle of institutions, youth does. Moreover, there
are economic reasons to believe that the age of institutions is positively
correlated with the age of their staffs, with the causality running in both
directions.
 *
,

.■«!■.;■■■ •: 'iri
t -A
1
"' '^' * '.* • - ^ *• J' • !•’
--if? ■.«> :.l.,
;'v<, ^ , .,.-, v. *,!.■;.-

•rAii-'^. JJl- V' "'' - ■' ^••.‘ '‘ r'' '* V,';4'v’i

'"f/t-' 1^;* ^ - u:“,': V > ,:V^bl^'rx 'fiyif. ‘I-' \>f

tT/kA£4i'..S*v '■* .. '• ■ -■ - -L. . ■■ v:
•'■ '^V ‘v-a;-
r .^ri
-‘.•V ■■' u -■ - V '-^■{. >v;'.!.;'^'*fr!*. tvr.'/ ^•' -■,- '.('X'.-i' ^'' ■ '
*L- ^ ' '■ ' ^-'‘’V ,v

vi "■ «-* j, -;■ ‘^.1; f? •' :< * •''.■>•'*■

V V
»-* ‘i* '■■ -.1 :>■• V -f '.’■' •'**'i
f- * - ‘ *•*,
;.-.;^<su: ,,'<■. v'',i> t' '-^ a. <c;;.'
■■.'■Sit, ^-^...'r v>'V J j ..-•ri<4;’v T^vi.tnH; ■r’ ! ^"

* Jj^jki^r»V<p.s; /■ ■•■■.'. ' fi'.■

O -A. ' .ji- ; V -AVjB .4,-\U ■. Xf -Jf .J !' , .

. .. ■ ■ ‘ ,;• -• "•■ U.^^^ M.

I . - .
*v't! • i<i-! •' . '»-r
i <-7
K->*
# w •■

, o ^^■,
,T

'. 'H -V
■Tv ,.1^1
'i'i* *;-rl" c>i..' _ •< V J
v;.' ,

4:r,;3Vir-.11 . ,*;'
V.. ilf Aj tf^ . ,■ 4-.mV fex
. >_ '■•"^ ■ ::j'-
, 'Mt^y . . ^ . 4*' '« * *^* 'i ■-*

* •-. . t * . ■'.<' 1!“

V* . ■' -. ^.fX * .1 *. - ' .•■ I -
.'ANI .^ •-- - .- '■ ■ ,!■■ ;
a?-- < •■ •-. i
•. . :!. . .
• V"* 7 Vi
>-4 ’T ''"V'-'Hi, ,<» \t*'"■ -j ^
f V \f • A.»v., 'jaLA.,*'. - .V,, Iir il
.^T*'
tf* '. -V fit
ns
, fii
..-‘V, ' ■vX
« •r
■<1' ■ ;-v
•* iK. , .i a:- VT?

/
Part Three

Normative Issues
 *■
•i

■ •

II . r'. --. /BMP-

- Sm

■• ; ..' ^f''^ • • >♦ v-i#V

■ 5v!

c ■;

■^-i- 1 ^
i~A
»
-♦ ’ ’■ #'»

t
.'i 3.1
^ #■
. ’ ' -v
^ V.

^ *’■

S
J ■■
 10
Euthanasia and Geronticide

The previous chapters have elaborated an explanatory and predictive theory

of old age as a social phenomenon, a theory that I consider to have social
scientific value in its own right but that I shall now employ to evaluate
present and proposed policies concerning old people. This chapter and the
next analyze policies in which moral and political concerns predominate
over legal ones, although the line is blurred; assisted suicide, for example,
is a crime in many states—with no exception for physician-assisted sui¬
cide, the focus of this chapter.

Definitions
Physician-assisted suicide is an aspect of a larger issue, that of euthanasia.
Immediately one is plunged into a terminological thicket. Euthanasia can
be voluntary, a form of suicide in a broad sense, or involuntary—the sort
of thing the Nazis practiced, as have a number of primitive societies, as we
saw in the preceding chapter. I am not interested in involuntary euthanasia,
other than where an individual is in a vegetative state or otherwise inca¬
pable of giving consent to die. That is an important case for the issue of
geronticide, the euthanasia of old people, since many very elderly individ¬
uals are severely demented; and I shall touch upon it. But mostly I shall be
using the term “euthanasia” as shorthand for “voluntary euthanasia” and
interchangeably with “physician-assisted suicide.” In so doing, I shall be
ignoring not only the distinction between voluntary and involuntary eutha¬
nasia but also the distinction often made between euthanasia in a narrow

235
 236 Normative Issues

sense as a physician’s administering drugs intended to kill the patient and

physician-assisted suicide narrowly defined as a physician’s helping the
patient to kill himself. I shall treat both as simply different modalities of
physician-assisted suicide (or euthanasia, or voluntary euthanasia).
A further complication is that euthanasia as I am defining it is merely
a subset of medical events (or nonevents) that have the effect of bringing
on death earlier than is medically inevitable. The other subsets are with¬
holding or withdrawing medical treatments that are considered useless
because they cannot prolong conscious life significantly; administering
painkillers likely to shorten the patient’s life; and acceding to a patient’s
refusal to accept further medical treatment or to take food or water.' The
entire set has been dubbed “MDEL” (medical decisions at end of life).
Even in the Netherlands, where euthanasia or, equivalently in my though
not in the Dutch terminology, physician-assisted suicide is not punish¬
able if proper guidelines are followed, it accounts for only a small fraction
of the total number of deaths due to MDEL.^ I have not found reliable
estimates for the United States. One might expect MDELs other than
physician-assisted suicide to be more common in this country (a sub¬
stitution effect)—or to be less common because the hostility to physician-
assisted suicide may reflect a desire to prolong life at all costs. We shall see
that MDELs are being facilitated in this country by the recognition of the
living will and the durable health-care power of attorney.
The issue of physician-assisted suicide transcends the elderly, but con¬
cerns them more than most simply because they are much more likely to
be terminally or otherwise terribly ill than younger people.^ That is not to
say that the rate of physician-assisted suicide is higher among the elderly
when—the essential qualification—state of health is held constant. It may
well be lower.^ There are several reasons. One is that with many persons of
very advanced age there is uncertainty about their ability to give a valid
consent. And they will often be so frail that the dosage of painkillers re-

1. For an excellent discussion, see G. K. Kimsma and E. van Leeuwen, “Dutch Euthanasia:
Background, Practice, and Present Justifications,” 2 Cambridge Quarterly of Healthcare Ethics
19(1993).
2. Id. at 27-28.
3. Hence the term “geronticide,” as in Stephen G. Post, “Infanticide and Geronticide,” 10
Ageing and Society 317 (1990). And recall from chapter 6 that the suicide rate is highest among
elderly persons. In the Netherlands, 38 percent of MDELs involve persons aged 65 to 79. Kimsma
and van Leeuwen, note 1 above, at 27.1 do not have a separate figure for euthanasia.
4. Some evidence for this conjecture is that euthanasia is said to be rare after the age of 75
and especially after 85. Gerrit van der Wal and Robert J. M. Dillmann, “Euthanasia in the Neth¬
erlands,” 308 British MedicalJoumal 1346, 1347 (1994).
Euthanasia and Geronticide lyj

quired to subdue their pain (most cases of euthanasia in the Netherlands

involve cancer patients) will kill them. Because the very elderly have
reached a “natural” age of death, moreover, physicians may not think it
necessary to report death as being due to euthanasia. Selection bias is also
at work; the elderly population is likely to contain a disproportionate num¬
ber of people who have a very strong will to live. A related point is that for
people so old that they must have adjusted to a greatly reduced utility of
living, the further drop entailed by the prospect of entering the terminal
stage of life may not register with as much vividness as the same prospect
would for a young person.
I shall distinguish between suicides in which the intention is formed
and executed at more or less the same time and suicides in which the exe¬
cution is substantially deferred (A decides at time t that he wants his life to
end at time t + k, where k might be many years), and within the first cate¬
gory between suicides in which there is assistance from another person and
those in which there is no assistance. I focus on the assisted suicide because
if a person who wants to end his life can do so without the assistance of
another person, the right to assist in the suicide without incurring criminal
liability has limited practical importance, though not none, as we shall see.
For a reason to be explained, I exclude assisted suicide where the assistance
is rendered by someone other than a physician.
By referendum in November 1994, Oregon became the first American
state to authorize physician-assisted suicide. (The law was supposed to go
into effect on January 1, 1995, but at this writing has been delayed by a
court challenge.) Subject to elaborate safeguards, physicians will be autho¬
rized to prescribe “suicide pills” to patients expected to live no more than
six months. The six-month limitation is problematic, not only because es¬
timates of how long a dying person has to live are fraught with error, but
also because some of the strongest cases of rational suicide involve people
who face an indefinite lifetime of paralysis, severe pain, or other terrible
disability. Even so, experience with Oregon’s new law may eventually pro¬
vide decisive evidence concerning the merits of physician-assisted suicide.
In the meantime, as debate intensifies in other states, economic analysis
has a significant and unrecognized contribution to make.

An Economic Analysis of Physician-Assisted Suicide

in Cases of Physical Incapacity
Benefits and costs. I have narrowed my focus to physician-assisted
suicide in cases of severely disabling and debilitating, usually though not
 238 Normative Issues

always terminal, illness.^ These are the cases in which the patient is likely
to lack the capacity to commit suicide on his own, at least without experi¬
encing prohibitive pain or fear; the cases therefore in which the demand for
physician-assisted suicide is greatest; the cases of most importance to el¬
derly people; and the only cases in which physicians are likely to be willing
to assist people to commit suicide. I set to one side religious objections to
suicide, not because I consider them unanswerable,* but because they be¬
long to the domain of individual choice rather than to that of social policy.
My conception of the appropriate scope of legal regulation is that of
John Stuart Mill: the only voluntary activities of competent adults with
which government can properly interfere are those that impose tangible
harms, as distinct from causing merely disapproval or even revulsion. The
fact that X's committing suicide with the assistance of Y is contrary to Z’s
religious beliefs is therefore not a good reason for having a law against
assisting suicide.
The main nonreligious objection to generally making suicide easier
than it is, whether by permitting the sale of suicide pills and suicide kits or
just by authorizing physicians to assist in the suicide of persons who are
dying or hideously impaired, is that many suicides are impulsive, the prod¬
uct of a bout of depression, intense grief or shame, bad news that may be
wrong (as in Romeo and Juliet), or other transient causes that, ex ante, the
affected individual might want to prevent from affecting him. Efforts to
discourage such suicides, as by making them more costly by punishing
people who assist in them, can be loosely analogized to the prohibition of
extortion (as in “your money or your life”), in which a class of transactions
yielding a short-term gain (when you buy your life by giving the robber
your money) is denied legal sanction because the vast majority of people
would consider themselves better off if the occasion for such a transaction
never arose. A prohibition against assisting suicide cannot be persuasively
defended on this ground in cases in which the person who wants to end his
life is incapable of doing so. The condition that makes it infeasible for the
individual to take his own life furnishes a rational motivation for suicide.^

5. Almo.st three-quarters of euthanasia cases in the Netherlands involve cancer, and in

83 percent the patient was estimated to have less than a month to live. Van der Wal and Dillmann,
note 4 above, at 1347. Nevertheless, as I point out in the text, it is doubtful that imminent death
should be a precondition to allowing physician-assisted suicide.
6. See David Hume, “Of Suicide,” in Hume, Essays: Moral, Political, and Literary 577
(Eugene F. Miller, ed., rev. ed. 1987). For philosophical arguments pro and con, see Suicide: Right
or Wrong? (John Donnelly, ed., 1990).
7. “The Elder Pliny . .. regarded suicide as the greatest gift given to man amid life’s suffer¬
ings.” Miriam Griffin, “Philosophy, Cato, and Roman Suicide; 2,” 33 Greece and Rome 192, 193
Euthanasia and Geronticide 239

A recent judicial decision invalidated, as an arbitrary deprivation of the

liberty protected by the due process clause of the Fourteenth Amendment,
a state statute criminalizing physician-assisted suicide.® Setting to one side
the question of the legal merits of the decision (which has since been re¬
versed), a reader cannot fail to be moved by the court’s harrowing descrip¬
tion of the situations of the three terminally ill plaintiffs (two elderly). Con¬
trary to widespread belief, in our society dying people usually experience
significant pain or other unpleasant symptoms;® the “peaceful” death cele¬
brated in Victorian fiction continues to be rare, or at least is not to be
counted on. It is easy to see that an individual who is soon to die anyway
and anticipates extraordinary pain or suffering in the interval that remains
may have a negative expected utility of living.'® We need only recall Kent’s
comment when signs of life are noted in the dying Lear: “Vex not his ghost:
O! let him pass; he hates him / That would upon the rack of this tough
world / Stretch him out longer.” “
A right to seek assistance in committing suicide has value to the holder
even if he never exercises it. The right of suicide is an option,and options
have value independent of the value of exercising them, just as insurance
has value for people who never have occasion to file a claim with an in¬
surer. Knowing that if life becomes unbearable one can end it creates peace

(1986). “Every day, rational people all over the world plead to be allowed to die.” Ronald Dwor-
kin, Life’s Dominion: An Argument about Abortion, Euthanasia, and Individual Freedom 179
(1993). I suggested in chapter 6 that elderly suicide is less likely to be impulsive than youthful
suicide; more on this later.
8. Compassion in Dying v. Washington, 850 F. Supp. 1454 (W.D. Wash. 1994), rev’d, 49
F.3d 586 (9th Cir. 1995). The U. S. Supreme Court had earlier held that a person has a constitu¬
tional right to refuse medical treatment even though death will result. Cruzan v. Director, Missouri
Dept, of Health, 497 U.S. 261, 278-279 (1990).
9. Robert Kastenbaum and Claude Normand, “Deathbed Scenes as Imagined by the Young
and Experienced by the Old,” 14 Death Studies 201, 212 (1990). Physicians know this best—
and it is well known that many of them keep “stashes” of lethal drugs on hand so that they can
kill themselves if they find themselves in the terminal stage of an illness. In a recent study in
which terminally ill patients were allowed to refuse to eat or drink, and as a result died of a
combination of dehydration and starvation, only 13 percent were judged to have experienced dis¬
comfort during their dying period. Efforts, apparently mostly successful, were made to relieve dry
mouth, thirst, and other symptoms of these modes of death. Robert M. McCann, William J. Hall,
and Annmarie Groth-Juncker, “Comfort Care for Terminally Ill Patients: The Appropriate Use
of Nutrition and Hydration,” 272 JAMA (Journal of the American Medical Association) 1263,
1265 (1994).
10. “It will generally be found that, as soon as the terrors of life reach the point at which
they outweigh the terrors of death, a man will put an end to his life. Schopenhauer, On Suicide,
in Essays of Arthur Schopenhauer 399, 403 (T. Bailey Saunders, trans., 1902).
11. King Lear, act V, sc. iii, 11. 314-316.
12. An argument that goes back at least to Seneca. For a modern version, see C. G. Prado,
The Last Choice: Preemptive Suicide in Advanced Age, ch. 7 (1990).
240 Normative Issues

of mind and so makes life more bearable. This is important, in any cost-
benefit analysis of permitting physician-assisted suicide, as a reminder that
the benefits of euthanasia are not limited to the relatively small number of
people who actually undergo it. The fact that the benefits are not limited to
those people is an offset to the concern that the costs will not be so limited
either—that authorizing assisted suicide, in however circumscribed a set
of cases, will inevitably encourage other suicides, and perhaps, by making
life seem cheaper, murders as well. Later I shall give reasons and data that
suggest that this concern is in any event exaggerated.
While on the subject of the third-party effects of allowing physician-
assisted suicide, I should address the argument that even if an individual
really and truly wants to die, his family may not want him to die and there¬
fore his death will impose a cost on uncompensated third parties. I do not
think this argument can survive a careful consideration of the relations of
altruism that connect the members of a loving family. In deciding whether
he wants to die, an individual will consider the effect of the decision on the
members of his family; and in deciding whether and in what spirit to accept
that decision, the family members will consider the cost to him if he is
forced to prolong his life. The decision he makes is therefore likely to
maximize the utility of the family as a whole.
It has been argued that “preemptive suicide on grounds of age actually
amounts to a kind of perverse faith that we can predict our own future, that
we can know what sources of unexpected meaning life has in store for
us.” We cannot know for certain. But we can have a pretty good idea;
human choices, including the irreversible ones, are made on the basis of
probabilities, not certainties. We shall see that the presence of uncertainty
is actually an argument/or a right of physician-assisted suicide.
It has been argued that since most elderly people who commit suicide
“have emotional or psychological illnesses,” their decision to commit sui¬
cide is irrational and should not be respected.'^ The principal illness men¬
tioned is depression. Anyone who decides to kill himself must find his life
depressing, and, with “suicidal ideation” and the like used to diagnose de¬
pression, it is apparent that one would have to assume that suicide is irra¬
tional in order to be justified in declaring a suicide irrational because the
person who committed suicide was depressed.The argument is circular.

13. Harry R. Moody, “ ‘Rational Suicide’ on Grounds of Old Age?” 24 Journal of Geriatric
Psychiatry 261,274 (1991).
14. Thomas J. Marzen, ‘“Out, Out Brief Candle’: Constitutionally Prescribed Suicide for
the Terminally Ill,” 21 Hastings Constitutional Law Quarterly 799, 811-812 (1994).
15. Recall the discussion of the suicide-depression circle in chapter 6.
Euthanasia and Geronticide 241

Another common argument against allowing physicians to assist in the

suicide of a patient, one that is also made against the right of abortion, is
that it is bad for society if physicians are used to kill as well as to save; by
blurring their mission, it may make them less committed to healing. Yet if
their healing efforts sometimes, perhaps often, place people in a situation
of such ghastly pain or incapacity that they are desperately eager to be
dead, physicians may become ambivalent about healing. It is also argued—
by opponents of capital punishment as well—that any policy which facili¬
tates the ending of human life as a deliberate choice undermines respect for
human life. The argument is especially weak in the case of capital punish¬
ment, when it is confined to murderers and can therefore be defended as
showing respect for the lives of the victims. We shall see in a moment that
a “life-saving” rationale may also be available to defend euthanasia, im¬
probable as that may seem. But in addition the argument that euthanasia is
inconsistent with a proper sense of the dignity of human life overlooks the
relation of dignity to quality. Respect for human life must have something
to do with perceptions of the value, not wholly metaphysical, of that life.
The spectacle of nursing homes crowded with frail and demented old
people, or of hospital wards crowded with dying people so heavily sedated
as to be barely sentient or so twisted with pain as to be barely recognizable,
might be thought rather to undermine than to enhance a sense of the pre¬
ciousness of life. The better the quality of lives, the greater the perceived
value of preserving them. Doctors and nurses who talk about “watering the
vegetables” on their rounds have not been made sensitive, by their expo¬
sure to the practical consequences of sacrificing quality of life, to the desire
to prolong life regardless.
Voluntary euthanasia has been practiced openly in the Netherlands
since the early 1970s,'^ yet the Dutch have not become more violent or
callous than other Europeans, let alone Americans. The Dutch murder rate
is only one-tenth that of the United States; more to the point, it is well
below the average of the European Union.Later we shall see that the

16. The extensive literature on euthanasia in the Netherlands is illustrated by Kimsma and
van Leeuwen, note 1 above; van der Wal and Dillmann, note 4 above; John Griffiths, Recent
Developments in the Netherlands concerning Euthanasia and Other Medical Behavior That Short¬
ens Life,” 1 Medical Law International 347 (1995); G. van der Wal et al., “Euthanasia and As¬
sisted Suicide, 1, How Often Is It Practised by Family Doctors in the Netherlands?” 9 Family
Practice 130 (1992); Paul J. van der Maas et al., “Euthanasia and Other Medical Decisions con¬
cerning the End of Life,” 338 Lancet 669 (1991); M. A. M. de Wachter, “Active Euthanasia in the
Netherlands,” 262 JAMA {Journal of the American Medical Association) 3316 (1989).
17. United Nations Development Programme, Human Development Report 1994 186 (1994)
(tab. 30). The statistics are for murders by males only, but most murderers are male.
242 Normative Issues

practice of euthanasia in the Netherlands appears not to have increased the

suicide rate either.
Carlos Gomez argues on the basis of 26 case studies of euthanasia in
the Netherlands'^ that there are insufficient controls over the practice to
ensure that it is always voluntary. Only one of his case studies (one of three
that he describes as “even more troubling” than the other 23) provides
even a modicum of support for his thesis: a young woman dying of leuke¬
mia may not have been told that there were less painful alternative treat¬
ments to chemotherapy.Then again she may have been told—Gomez
doesn’t know. During a one-year remission from the disease, she and her
husband had spoken with their family doctor many times about euthana¬
sia,^' but he may not have been conversant with the full range of alternative
therapies.
Gomez’s fear of doctors’ rushing patients to their deathhas not been
substantiated^^ and does not appear realistic. Such behavior would go
against the grain of the medical profession, which strongly favors treat¬
ment, however unlikely of success. It might also be contrary to the profes¬
sion’s financial self-interest, although this depends on the method of fi¬
nancing medical services. If doctors are paid for services rendered—the
payment method that prevails in the United States—the incentive is to give
patients too much rather than too little treatment. (This point implies that
physicians should be forbidden to specialize in assisting suicide, as that
would realign their financial incentives. Apparently no such specialty has
emerged in the Netherlands.) Yet even in the United States, many patients
are not treated on a fee-for-services basis. This is true not only of people
enrolled in health maintenance organizations (HMOs) but also of other
people, veterans for example, who receive medical care from salaried phy¬
sicians. Here the financial incentive is to avoid expensive end-of-life treat¬
ments, for which euthanasia might be a cheap alternative.
Also pertinent in evaluating the danger of a rush to death is the hospice
movement, which is hostile to euthanasia. It holds that the terminal phase

18. Carlos F. Gomez, Regulating Death: Euthanasia and the Case of the Netherlands 64-
89(1991).
19. Id. at 111.
20. Id. at 112.
21. Id. at 79.
22. As in Waugh’s burlesque of euthanasia. Evelyn Waugh, Love Among the Ruins: A Ro¬
mance of the Near Future, ch. 2 (1953).
23. Another case study of euthanasia in the Netherlands found no serious abuses. G. van
der Wal et al.. Euthanasia and Assisted Suicide, 2, Do Dutch Family Doctors Act Prudently?”
9 Family Practice 135 (1992).
Euthanasia and Geronticide 243

of a person’s life can be made bearable and therefore need not be shortened
by suicide. A hospice offers an alternative to a dying person who considers
suicide, and hence it offers competition to physicians who provide assis¬
tance in committing suicide.
The danger of the abuses that Gomez fears can be minimized by rela¬
tively simple regulations, such as a requirement that the patient’s consent
to euthanasia be witnessed or in writing, that the physician performing eu¬
thanasia report any case in which he performs it to a hospital committee,
and that before performing it he consult with a duly certified specialist in
the ethics of dealing with dying patients.^"* The feasibility of such regula¬
tions, and the culture of the medical profession, are reasons why allowing
physician-assisted suicide does not as a matter of logic entail allowing non¬
physicians to assist in suicides. Fear that many so-called “mercy” killings
are nothing of the kind, that they lack the consent whether explicit or rea¬
sonably implied of the person killed, is weakly grounded when the person
doing the killing is a physician not related to or otherwise personally in¬
volved with the person killed.
More questionable than anything recounted by Gomez is a case in
which a Dutch doctor was acquitted of a criminal charge for assisting in
the suicide of a middle-aged woman who was neither physically nor men¬
tally ill but who was determined to die and had attempted to commit suicide
before, and who the doctor was persuaded would try again and eventually
succeed.^^ The case goes beyond any measure legalizing euthanasia that I
would be inclined to support. It was not a case in which the person request¬
ing assistance had a terminal illness or other progressively disabling con¬
dition that might incapacitate her from taking her own life when her suffer¬
ing became unbearable. But I do not believe on the basis of a single Dutch
case that by authorizing physician-assisted suicide in cases of physical in¬
capacity as I have defined it the United States would be taking an irrevers¬
ible step toward unregulated assisted suicide.

Fewer and later suicides? So far I have suggested merely that the op¬
ponents of physician-assisted suicide underestimate the benefits and exag¬
gerate the costs. I have taken for granted that one consequence is that there

24. See Franklin G. Miller et al., “Regulating Physician-Assisted Death,” 331 New England
Journal of Medicine 119 (1994).
25. Office of Public Prosecutions v. Chabot, translated and analyzed in John Griffiths, “As¬
sisted Suicide in the Netherlands: The Chabot Case,” 58 Modern Law Review 232 (1995). The
Dutch Supeme Court reversed the acquittal, but only because there had not been an independent
examination of the woman by another physician; and the court waived punishment.
244 Normative Issues

will be more suicides and a net loss of years of life. The bearing of this on
a Millian analysis of the right to physician-assisted suicide is unclear. If the
number of suicides rose as a consequence of legalizing physician-assisted
suicide (in appropriate cases, subject to appropriate safeguards), this might
indicate nothing more than that many people place a negative value on
extending the period in which they are dying. An alternative possibility,
however, which makes investigating the impact of allowing physician-
assisted suicide on the suicide rate worth pursuing even for a Millian, is
that a rise in the rate might indicate—though it would not prove—that
people were indeed being rushed to their deaths by selfish relatives and
callous physicians. So it becomes relevant to point out, as I shall, that per¬
mitting physician-assisted suicide limited to what I am calling cases of
physical incapacity might actually reduce the number of suicides and post¬
pone the suicides that occur. Gomez argues in effect that there will be more
deaths (and they will come sooner) than those attributable to a genuine and
fully informed choice by the patient to die. The argument I shall be explor¬
ing is that there will be fewer deaths (and later) than under a regime in
which physician-assisted suicide is unlawful.
Of course, if this is correct, aggregate medical costs might rise, since
seriously ill people who decide not to end their life immediately are bound
to incur substantial medical costs as a consequence of their decision. To the
extent that these costs are borne by third parties rather than by the person
making the decision to live—and this is a pervasive feature of our health¬
care system—it becomes difficult to say whether allowing physician-
assisted suicide would be socially cost-justified. Mill’s approach enables us
to exclude (as a strictly economic or utilitarian analysis would not) the
disutility that third parties experience merely as a consequence of abhor¬
ring suicide. But it does not entitle us to disregard the tangible costs borne
by people who through their taxes, health-insurance premiums, or doctors’
bills are forced to pay other people’s medical expenses. I shall not attempt
to estimate those costs; and to that extent the analysis presented in this
chapter must be considered tentative.
Suppose an individual learns that he has a progressive disease that will
reduce him to a state in which he would consider himself better off dead
than alive because of acute suffering unredeemed by any hope of recovery
or improvement or by the diminished utility from living in this state. He
realizes, let us further assume, that at some point the progress of the disease
will incapacitate him from committing suicide. This may be one reason
why elderly suicide attempters tend to use more lethal methods, such as
Euthanasia and Geronticide 245

firearms instead of drugs, than younger ones, and have a higher success
rate.2® The elderly person fears that if his attempt fails, he may be incapable
of repeating it; the cost of failure is greater to him. An alternative explana¬
tion is that elderly suicides are more deliberated for the reasons discussed
in chapter 6, and the deliberative as distinct from impulsive attempted sui¬
cide is more likely to choose an effective means. Fair enough; but by im¬
plying that elderly suicides are more likely to be rationally considered than
the suicides of younger persons, the point provides additional support for a
right of physician-assisted suicide.
To make the case more realistic, assume that our hypothetical sufferer
is not certain that the disease will progress to a point where he will prefer
to be dead, though he is certain that if it does progress to that point he will
be incapable of killing himself without assistance. The possibility that he
will recover after all, at least recover sufficiently to be glad that he is still
alive, or at the very least that he will live longer than he expected and in
circumstances less oppressive than he anticipated—the possibility, in short,
of a mistake about the future course of his disease—is omnipresent in sui¬
cide situations and is one of the objections that I listed earlier to making
suicide easy. A surprising number of people have had the experience of
being misinformed that they had a terminal illness.^’
We need to compare alternative regimes for our hypothetical case. In
the first, physician-assisted suicide is forbidden. So when the individual
first learns his probable fate he must choose between two courses of action:
one in which he commits suicide now, at a cost (in dread of death, pain,
moral compunctions, whatever) of c; the other in which he postpones the
decision to a time when, if he still wants to commit suicide, he will be
unable to do so. The question is which course will confer greater utility on
him. If he commits suicide now, he will have utility of — c. He will expe¬
rience neither positive nor expected utility from living, because he will be
dead, but he will incur the cost of getting from the state of being alive to

26. John L. McIntosh and John F. Santos, “Methods of Suicide by Age: Sex and Race Dif¬
ferences among the Young and Old,” 22 International Journal of Aging and Human Develop¬
ment 123 (1986); Ellen Mellick, Kathleen C. Buckwalter, and Jacqueline M. Stolley, “Suicide
among Elderly White Men: Development of a Profile,” Journal of Psychosocial Nursing, no. 2,
1992, p. 29.
27. One of my grandfathers was told in his forties by a reputable medical specialist that he
had a fatal kidney disease but could eke out another year or two of life if he gave up meat. He did
not give up meat and he died at the age of 85 of an unrelated ailment. This was a long time ago,
but the problem persists. Like other professionals, doctors sometimes speak with greater confi¬
dence than the facts warrant.
 246 Normative Issues

the state of being dead. If he decides not to commit suicide now, he avoids
incurring c and obtains whatever utility, positive or negative, continued life
confers upon him. Because of uncertainty, that utility is an expected utility;
it is equal to the weighted average of his negative utility in the doomed
state, — Ud —the disutility that he will incur if it turns out that he really
does have a terminal or otherwise horribly painful or disabling illness—
and his positive utility in the healthy or at least relatively healthy state that
he will be in if he recovers to the point of wanting to live after all: .
Each expected utility must be weighted by the probability (p or 1 —
p) that the individual will in fact find himself in the doomed or in the
healthy state. He reasonably expects the former, but not with certainty: that
is, 1 > p > fl — p). The sum of these utilities is p{~ U^) + (\ — p)Uh,
and must be compared with the utility of committing suicide (— c). I as¬
sume that Ud > c, an important assumption that will be relaxed later.^*
With these assumptions, our hypothetical individual will commit
suicide if

pUd > (I -p)U, + c (10.1)

—in words, if the expected utility of death now, which is to say the disu¬
tility averted by death now, exceeds the expected utility of life plus the cost
of suicide. The loss of that expected utility, and the cost of suicide, are the
costs that he incurs by committing suicide now.^^ If c = 0, he will commit
suicide if pUd > (\ —p)Uh, that is, if the expected (that is, probability-
weighted) disutility of living in what I am calling the doomed state exceeds
the expected utility of living in the healthy state.
Since the doomed state is more likely—that is, p > (1 -p)—the an¬
ticipated disutility of that state need not be so great as the anticipated utility
of the saved (healthy) state for suicide to be a rational decision; indeed, if
p is high enough, the disutility of living in the doomed state could be con¬
siderably smaller than the utility of living in the saved state without making
the decision to commit suicide an irrational one. This also depends on the
size of c, however. If the cost of committing suicide is great enough, an
individual will refrain from committing suicide even if he would consider
himself much better off dead than alive. So c is a type of transaction cost,
a one-way ticket to oblivion.

28. It is because c usually is high and because, even so, often Uj> c that suicide can be and
has been regarded both as courageous and as cowardly. For an interesting discussion, see Miriam
Griffin, “Roman Suicide,” in Medicine and Moral Reasoning 106, 122-123 (K. W. M, Fulford,
Grant R. Gillett, and Janet Martin Soskice, eds., 1994).
29. I ignore, as inessential to the analysis, the discounting of these future values to present
values.
Euthanasia and Geronticide 247

Contrast the situation in which the individual has a choice between

committing suicide now, again at cost c, and committing it later, at the
same cost, with a physician’s assistance. It is a real choice because, by vir¬
tue of the possibility of assistance (assumed to have been legalized), the
individual can postpone the decision to commit suicide.^® If we assume for
simplicity that the unbearable suffering that gives rise to — Uj, the dis¬
utility of the doomed state, will begin at some future time when the indi¬
vidual will know for certain that he will not recover into the relatively
healthy state Uh, the assumption that suicide is possible later at a cost of c
implies the substitution of c for in inequality 10.1. That is, as soon as
the doomed state sets in, the individual (with assistance, for I am assuming
that the onset of the doomed state will incapacitate the individual from
committing suicide without assistance) will substitute for it a lesser dis¬
utility, the cost of committing suicide.
With this substitution into 10.1, our hypothetical individual will com¬
mit suicide now, rather than postpone the decision, only if

pc > (\ — p)Uh + c (10.2)

or equivalently if - > c—which makes clear that he will not commit

suicide now, since both Un and c are positive. Even if discounting to present
value is ignored, the cost of committing suicide with probability 1 must
exceed that cost when multiplied by a probability of less than one and off¬
set by some expectation of entering a state in which continued life will
yield net utility. Indeed the cost of suicide now must exceed the expected
cost of suicide later (since c is assumed constant and exceeds pc), even if
the expected utility from living is ignored.
The analysis implies that if physician-assisted suicide in cases of
physical incapacity is permitted, the number of suicides in the class
of cases that I have modeled will be reduced by 1 - p, the percentage of
cases in which the individual contemplating suicide is mistaken about the
future course of his disease or its effect on his desire to live. Moreover, in
the fraction of cases in which suicide does occur (p), it will occur later than
if physician-assisted suicide were prevented. Weeks, months, or even years
of life will be gained, and with it net utility.
The intuition behind these results is straightforward. If the only choice
is suicide now and suffering later, individuals will frequently choose sui-

30. According to the physician in one of Carlos Gomez’s case studies, “the availability of
euthanasia gave the woman [his patient] enough assurance to at least try one round of chemo¬
therapy.” Gomez, note 18 above, at 111.
248 Normative Issues

cide now. If the choice is suicide now or suicide at no greater cost later,
they will choose suicide later because there is always a chance that they are
mistaken in believing that continued life will impose unbearable suffering
or incapacity on them. They would give up that chance by committing sui¬
cide now. The possibility of physician-assisted suicide enables them to wait
until they have more information before deciding whether to live or die.
Another way to put this is that the availability of physician-assisted suicide
increases the option value of continued living. We noted in chapter 6 that
the diminution in that value with age is one of the factors that contributes
to the high suicide rate of elderly people.
The general point—that the availability of a service can reduce rather
than, as one might expect, increase the utilization of the service—is neither
inconsistent with assuming rational behavior by persons facing horrific
choices nor limited to suicide. Suppose that you get a sharp pain in your
abdomen on Friday afternoon. If your physician’s office is closed on week¬
ends, you may rush to the office on Friday, lest your condition worsen dur¬
ing the weekend. But if the office is open on weekends you may decide to
wait and see whether the pain gets better or worse. In most cases it will get
better, so there will be fewer total visits, in the class of cases represented
by the example, if the physician is more available.
The effect of physician-assisted suicide in reducing the number of sui¬
cides will be amplified if, as is plausible, physician-assisted suicide is less
costly to a person contemplating suicide than unassisted suicide would be
rather than, as I have been assuming, just as costly. The difference in cost
will increase his incentive to wait because physician-assisted suicide is not
permissible in my analysis until the patient has become incapable of taking
his own life. Paradoxically, then, cheaper suicide may result in less suicide.
But this depends on the assumption in the model that c < Uj, the cost of
suicide is less than the utility of the dying state. Suppose that unassisted
suicide (cj is so costly (in search for the requisite means, in pain, in fear
of failure, and in the expected consequences of failure) that many a person
who anticipates with certainty a life of utter misery will nevertheless not
attempt to commit suicide unless he can have the assistance of a physician.
(That is, c„ > Uj.) Then if physician-assisted suicide (cj at a sufficiently
lower cost that is available when the period of misery begins, he
will terminate his life then if, and only if, physician-assisted suicide is per¬
mitted. That the cost of suicide is an important factor in the suicide rate,
especially for older people, is shown by the fact that the suicide rate of
elderly English people fell when cooking gas was detoxified; putting one’s
Euthanasia and Geronticide 249

head in a gas oven had been a favorite (and very easy) method of suicide,
especially for middle-aged and elderly persons.^'
Physician-assisted suicide could also increase rather than reduce the
number of suicides if people systematically underestimate either the proba¬
bility or the severity of the doomed outcome and as a result do not commit
suicide when they first learn their probable fate. When they wise up it is
too late if physician-assisted suicide is not permitted. Notice however that
if the analysis in chapter 5 is correct, this problem of foolish optimism is
less likely to be acute with old than with young people.
Even in the case where > Uj > c^, it is possible that allowing
physician-assisted suicide would, as before, lower rather than raise the sui¬
cide rate. With physician-assisted suicide cheaper than unassisted, persons
contemplating suicide will tend to choose physician-assisted over unas¬
sisted. This implies that before committing suicide they will consult with a
physician. The delay required by such a consultation will reduce the num¬
ber of impulsive suicides; others will be avoided by the physician’s identi¬
fying a treatable mental illness. The frequently remarked difficulty of di¬
agnosing suicidal tendencies in elderly patients is reduced when patients
have an incentive to disclose those tendencies because they are seeking
help in killing themselves. Physician-assisted suicide thus lowers the cost
not only of suicide but also of interventions that can avoid suicide. This
effect is not limited to cases of physical incapacity, but is not, I think, large
enough to justify a broader right to physician-assisted suicide. I base this
judgment not only on the Chabot case,^^ which may of course be unrepre¬
sentative, but also on the fact that a general right of physician-assisted sui¬
cide could (though needn’t, as I have just suggested) reduce the cost of
impulsive as well as of deliberated suicides. I pointed out earlier that people
may not want to make it easier for themselves to commit suicide
impulsively.
It may be objected that my entire analysis violates the economist’s Law

31. Dan G. Blazer, “The Epidemiology of Psychiatric Disorders in Late Life,” in Geriatric
Psychiatry 235, 251 (Ewald W. Buse and Dan G. Blazer, eds., 1989); see also George Winokur
and Donald W. Black, “Suicide—What Can Be Done?” 327 New England Journal of Medicine
490 (1992); Bijou Yang and David Lester, “The Effect of Gun Availability on Suicide Rates,”
19 Atlantic Economic Journal 74 (1991).
32. See Carmelita R. Tobias, Raymond Pary, and Steven Lippmann, “Preventing Suicide in
Older People,” 45 American Family Physician 1707 (1992); Yeates Conwell and Eric D. Caine,
“Rational Suicide and the Right to Die: Reality and Myth,” 325 New England Journal of Medicine
1100, 1101-1102(1991).
33. See note 25 above and accompanying text.
250 Normative Issues

of Demand; that lowering the price of a good or service—here, suicide—

must increase rather than reduce the demand for it. This is not the correct
way to frame the issue. We have two goods, not one: unassisted suicide,
and physician-assisted suicide. They are substitutes, so lowering the price
of the second (by legalizing it) will reduce the demand for the first, and
nothing in economics teaches that this reduction must be fully offset by the
increased demand for the second good. A razor blade that retains its sharp¬
ness for ten shaves is a substitute for one that retains it for only one shave,
but if the former takes over the market the total number of razor blades
produced and sold will decline even if the longer-lasting blade is no more
expensive than the other blade.
Although I have been stressing the physician’s role in reducing the cost
of suicide, he also has an important role to play in reducing the benefits, in
particular by administering effective painkillers, which by reducing re¬
duce the likelihood of suicide. The two roles merge when, as is common,
the physician administers painkillers in potentially lethal doses. In such a
case Ua is eliminated either by killing the patient’s pain or by killing the
patient.
Notice, finally, that if my analysis is correct, physicians in a fee-
for-service health-care system, the dominant system in the United States,
should support a change in the laws to authorize physician-assisted suicide.
Such a change would increase the demand for physicians’ services both
directly, and, if it is true that the laws would reduce the suicide rate of sick
and elderly people and postpone the suicides of such people that do occur,
indirectly by prolonging the life of people who have very grave diseases
requiring protracted and expensive medical attention.

Evidence. The question whether allowing physician-assisted suicide

in cases of physical incapacity would increase or reduce the suicide rate
can be studied empirically. Table 10.1 regresses state suicide rates in the
United States on state per capita income, the percentage of the state’s popu¬
lation that is black (blacks have much lower suicide rates than whites), and
a dummy variable that takes a value of 1 if a state has a law criminalizing
physician-assisted suicide and 0 otherwise.^"*

34. More than half the states have such laws. Data on suicide rates and per capita income
are from the Statistical Abstract of the United States for 1993. Data on race are from Kathleen
O’Leary-Morgan et al., 1991 State Rankings: A Statistical View of the 50 United States (1991).
Data on assisted-suicide laws are from Julia Pugliese, “Don’t Ask—Don’t Tell; The Secret Practice
of Physician Assisted Suicide,’’ 44 Hastings Law Journal 1291, 1295 n. 20 (1993).
Euthanasia and Geronticide 251

Table 10.1 Regression of Suicide Rate on Assisted-Suicide

Law and Other Variables (t-statistics in parentheses)

Per Capita Percentage Assisted-

Income Black Suicide Law R2

- .0005 -.1287 -.7601 .31

(-3.388) (-2.999) (-0.951)

The coefficients of the income and percentage-hlack variables are

negative and highly significant statistically, and these two variables explain
a good deal of the variance across states in the suicide rate. The coefficient
of the law variable is also negative, implying that states that forbid physi¬
cian-assisted suicide do have lower suicide rates than states that permit it.
But it is not statistically significant, though perhaps only because most su¬
icides are not committed by terminally ill or otherwise desperately ill
people and thus do not come within the scope of the hypothesis that I am
trying to test. Although these results do not suggest that repealing an as-
sisted-suicide law is a sound method of reducing a state’s suicide rate, they
cast at least some doiibt on the hypothesis, which I have been questioning
despite its intuitive appeal, that making suicide easier is likely to lead to
more suicides.
But I stress “some” doubt. If assisted-suicide laws are rarely enforced
against physicians, this would suggest that such laws probably have very
little deterrent effect. (The alternative hypothesis, that there is little enforce¬
ment because of perfect compliance, is hardly credible; there is consider¬
able, and cumulatively persuasive, anecdotal and survey evidence that
physician-assisted suicide is not rare in the United States.) Or if, though
such laws are commonly enforced in states that have them, states that do
not have them punish physician-assisted suicide as ordinary homicide,^® the
absence of a special law would not be expected to make much difference,
unless juries were willing to convict physicians of assisted suicide but not
of homicide.
In fact it appears that neither type of law is used with any frequency
against physicians who assist their patients to commit suicide. I have found

35. See id. at 1305-1306.

36. See David R. Schanker, “Of Suicide Machines, Euthanasia Legislation, and the Health
Care Crisis,” 68 Indiana Law Journal 977, 985-992 (1993).
252 Normative Issues

only three published judicial opinions involving such conduct since 1950,
all involving Dr. Jack Kevorkian —surely a special case in view of his
decision to conduct his activities in the open, and indeed in the glare of
publicity. I have found only four American cases, other than those involv¬
ing Kevorkian, in which a physician was prosecuted for assisting a patient
to commit suicide.^* No doubt there are more, but the total must be very
slight in relation to the number of physician-assisted suicides. This makes
it unlikely that a law criminalizing physician-assisted suicide would actu¬
ally increase the suicide rate, although it helps explain the absence of a
statistically significant effect of such laws. And yet the existence of an
unenforced prohibition, whether its source is a special statute on assisted
suicide or the general law of homicide, could increase the suicide rate by
retarding the development of legal and ethical norms that regulate, and by
regulating limit, the practice.
Another bit of evidence concerning the effect on the suicide rate of
laws relating to physician-assisted suicide is presented in figure 10.1,
which graphs the trend in the suicide rate of elderly males (75 years old
and older), relative to that of all males, in the Netherlands and other north¬
ern European countries. That rate was very high in the Netherlands before
euthanasia became common in the early 1970s and has fallen since, both
absolutely and relatively to the other countries in the sample.^®
Deaths caused by euthanasia, however, including physician-assisted
suicides, are not counted as suicides in the Dutch statistics. One article, it
is true, refers to incurably ill elderly patients who “request euthanasia or
physician-assisted suicide,” in a context suggesting that if their requests
are granted they are counted as suicides.'*® But this is wildly implausible
(and in fact incorrect), since the estimated number of annual deaths due to

37. People v. Kevorkian, 527 N.W.2d 714 (Mich. 1994); People v. Kevorkian, 517 N.W.2d
293 (Mich. Ct. App. 1994); Bobbins v. Attorney General, 518 N.W.2d 487 (Mich. Ct. App. 1994).
38. See H. Tristam Englehardt, Jr., and Michelle Malloy, “Suicide and Assisting Suicide: A
Critique of Legal Sanctions,” 36 Southwestern Law Journal 1003, 1029 (1982); Michael Winerip,
“Prosecutor Ponders Mercy for a Mercy-Killing Doctor,” New York Times (national ed.), Nov. 25,
1986, p. B4; Lawrence K. Altman, “Jury Declines to Indict a Doctor Who Said He Aided in a
Suicide,” New York Times (national ed.), July 27, 1991, p. 1. Only one of the four cases resulted
in a conviction; the defendant pleaded guilty to manslaughter. Assisted suicide is rarely prosecuted
even when the person rendering assistance is not a physician. Catherine D. Shaffer, Note, “Crimi¬
nal Liability for Assisting Suicide,” 86 Columbia Law Review 348, 369-371 (1986).
39. The source of the data for figure 10.1 is World Health Organization, World Health Sta¬
tistics Annual, various years. The pattern with respect to female suicide is similar though with an
uptick for Dutch suicides in the most recent period.
40. A. J. F. M. Kerkhof et al., “The Prevention of Suicide among Older People in the Neth¬
erlands: Interventions in Community Mental Health Care,” 12 Crisis 59, 63 (1991).
Euthanasia and Geronticide 253

—■— Netherlands

Sweden
—♦— England
and Wales
Federal Republic
of Germany

Figure 10.1 Suicide rate of elderly males as a multiple of the total male suicide rate,
1965-1990

euthanasia (2,700, of which 400 are physician-assisted suicides) exceeds

the total number of reported Dutch suicides, which is less than 2,000.'*' The
Dutch count as “suicide” for statistical reporting purposes only a fraction
of deliberate efforts by persons to bring about their immediate death. Lack¬
ing as we do a time series for euthanasia, we cannot infer from figure 10.1
that the total number of elderly suicides in the broadest sense has fallen in
the Netherlands since euthanasia became common. It is possible that, con¬
sistent with my analysis, what has happened is a substitution of euthanasia
for conventional suicide.
It has been argued, we saw, that allowing euthanasia would encourage
suicide (as well as murder) at all ages by undermining the sanctity of life.
If so, we might expect the Dutch suicide rate to have risen since the early
1970s. It did rise by 25 percent between 1974 and 1988, but the average
increase in the other member nations of the European community was
36.25 percent; the Dutch increase was the fourth lowest out of 12.“*^

41. See id. at 59; Kimsma and van Leeuwen, note 1 above, at 27-28; Netherlands Central
Bureau of Statistics, Statistical Yearbook 1993 of the Netherlands 418 (1993) (tab. 43). A phone
call to the Netherlands Central Bureau of Statistics confirmed that no cases of euthanasia are
included in the Dutch suicide statistics.
42. Computed from Colin Pritchard, “Is There a Link between Suicide in Young Men and
Employment? A Comparison of the UK with Other European Community Countries,” 160 British
Journal of Psychiatry 750, 753 (1992) (tab. 3). The table from which I have derived these data lists
Scotland and Northern Ireland separately from England and Wales. I have included the figure for
Scotland, but not for Northern Ireland—a distinct outlier with its 206 percent increase in the sui¬
cide rate.
254 Normative Issues

Voluntary Euthanasia with Implementation Deferred

I turn now to the case in which there is a nontrivial interval between the
decision to die and the carrying out of the decision. I shall consider two
versions of this case. The first is where A, having acquainted himself with
the facts about old age, decides that the physical decrepitude of that state is
such that he would greatly prefer not to enter it; but fearing that he will
have different preferences when he reaches that age, he wants to commit
himself now to die at age 75, which he regards as the threshold of too old
age. In the second case, B is anxious not about old age as such but about
senility, which he considers a living death. He knows that if he becomes
senile it may be too late for him to terminate his life voluntarily, so like A
he wants somehow to commit to die if and when he becomes senile, at
whatever age. fi’s anxiety about becoming senile cannot be considered neu¬
rotic or irrational. Senile dementia afflicts a substantial fraction of old
people, as we saw in chapter 1, causing grievous and degrading cognitive
impairment. The risk of becoming severely demented, especially for people
in their eighties or nineties, is great enough to be a source of understand¬
able dread to many aging people.

The case of physical decrepitude. The economic argument for giving

A what he wants is that we permit people to make irrevocable commitments
about their future—to foreclose any realistic prospect of becoming a doctor
by going to law school instead of to medical school, or, coming closer to
home, to impair one’s longevity by adopting an unsafe or unhealthy mode
of life. Suicide can be regarded in that light. But there is a counterargument
when the decision to commit suicide is made many years before the in¬
tended execution of the decision. For it can be argued, as we saw in chap¬
ter 4, that the self at time t and the self at time r + k are actually two
persons. A, and A,+*, at least when k is a substantial number. What is a
person? By hypothesis A, and have different preferences concerning
the fundamental issue of life versus death. The younger self has of course
a degree of control over the older, and the older has no control at all over
the younger, simply because time runs forwards but not backwards. It may
be impossible as a practical matter to make A, a fiduciary of A,+i,^^ But it
does not follow that the law should affirmatively assist in the younger self’s
destructive designs against the older self, as by enforcing a contract be-

43. With a limited exception discussed below; and notice the analogy to cases in which
a pregnant woman is punished for not taking adequate care of herself and thus endangering the
fetus.
Euthanasia and Geronticide 255

tween A, and some third party to kill A at time t + k. For on what ground
shall the younger self be adjudged more authentic than the older self? The
problem of picking the authentic self is a standard one in multiple-selves
analysis."*"* Perhaps the game theorists can help us, by modeling the out¬
come of bargaining between one’s current and one’s future self. Perhaps;
but at present no satisfactory solution has been suggested to the problem of
arbitrating the conflict among successive selves. I find it odd therefore that
Prado should think it an argument in favor of “preemptive” suicide that in
old age we may cease “to be the persons we are” and “adjust to what even
a few months before we would have rejected as intolerable.”"*^
This analysis might seem to support a much stronger position, that
suicide should always be prevented if prevention is feasible, since when A,
kills himself he is also killing A,+* who may have positive utility from
living. Yet one can easily imagine a case in which A, kills A as it were
impartially, because even though he knows that his future self will derive a
positive utility from living (that is, t/A,+^ > 0), the sum of present and
future utilities is negative (— UA, -I- LA,+* < 0); the disutility to A, of
living (maybe his life is blighted by fear that he will end up in a nursing
home,"*^ which he may consider the equivalent of a concentration camp)
exceeds the utility to A,+* of living. In such a case, suicide will be utility
maximizing even if the present self weights the future self’s utility equally
with its own.
So multiple-selves analysis need not condemn all suicide—at least if
we are utilitarians, a big “if” for many people. Yet even if we are utilitar¬
ians, willing to trade off one “person’s” life against another’s, the implica¬
tions of the analysis for the permissible scope of governmental interference
with individual choices are disquieting to anyone who believes in liberty.
For example, if our future self has a moral claim as great as a fetus (another
potential person) even if not so great as our present self, the argument for
forbidding a pregnant woman to smoke becomes an argument for forbid¬
ding anyone but a dying person to smoke.
The argument for forbidding suicide commitments cannot be dis¬
missed out of hand as paternalistic even by those who reject all paternalis-

44. See, for example, Thomas C. Schelling, Choice and Consequence: Perspectives of an
Errant Economist 67-68, 98, 152-156 (1984).
45. Prado, note 12 above, at 119; see also id. at 124-125.
46. Although, contrary to the popular impression, most nursing-home stays even for elderly
people are relatively short, whether they end in death or in return to the noninstitutionalized com¬
munity, “nearly 25 percent of all women who enter a nursing home will spend three years there.”
Andrew Dick, Alan M. Garber, and Thomas A. MaCurdy, “Forecasting Nursing Home Utilization
of Elderly Americans,” in Studies in the Economics of Aging .365, 392 (David A. Wise, ed., 1994).
256 Normative Issues

tic grounds for interference with choices made by competent adults;the

person is not choosing for himself if his future self is a different person
from his present self. Nevertheless the pragmatic objections to the argu¬
ment are similar to the pragmatic objections to paternalistic arguments for
government interference with personal liberty. In particular, even if the
younger self is not a perfect agent of the older self, how likely is it that the
state will be a better agent, or more precisely will better balance the com¬
peting claims of the two selves? In just the same way, even though parents
are not perfect agents of their children, we assume that except in the ex¬
treme cases which we call by such names as neglect and abuse they are apt
to be better agents than the state. We observe that people do make provision
for their old age and also for their other contingent selves—they buy dis¬
ability insurance, for example, rather than simply writing off their possible
disabled self that may come into being after an accident. They are not
wholly neglectful of their future selves. The question is whether they are
sufficiently neglectful to warrant government intervention, with all its
costs.

The case of senility. B's case, in which the younger self wants to kill
the older only if the older becomes severely demented, differs from A’s
because there is a question whether B,+k is a person. If he is not, the ques¬
tion whether he is a separate person, entitled to some protection against
B,, does not arise. For there is a distinction between identity and person-
hood.'^* B,+* has the same name and other indicia of identity as B,, but if
personhood requires some degree of mentation"*^ and not merely a func¬
tioning brain stem, may not be a person and therefore may not be
entitled to any protection.
I am not comfortable with this argument.^*' Dan Brock, who is, ac¬
knowledges that his “view of personhood implies that infanticide need not

47. On the difference between paternalistic arguments and arguments based on the concept
of multiple selves, see discussion in next chapter of the compulsory character of social security.
48. See, for example, Dan W. Brock, Life and Death: Philosophical Essays in Biomedical
Ethics, ch. 12(1993).
49. Signally including “autobiographical memory” in the elementary sense of knowing who
one is and was. See D. B. Bromley, Behavioural Gerontology: Central Issues in the Psychology of
Ageing 231 (1990). This, the reader will recall from chapter 5, is part of the reason why an old
person may be eager to postpone his death even if only for a short time.
50. Dworkin, note 7 above, at 232, expresses ambivalence about it in the case where it ap¬
pears that the demented individual is happy despite his state. A further concern, which I shall not
discuss, is the use of guardianship and procedures for involuntary civil commitment to confine to
institutions or otherwise impair the autonomy of elderly persons who are not in fact incompetent,
but merely an inconvenience to their relatives. See George J. Alexander, “Age and the Law,” in
Perspectives on Aging: Exploding the Myths 45, 54-66 (Priscilla W. Johnston, ed., 1981).
Euthanasia and Geronticide 257

wrong a newborn infant and that infants lack any serious moral right not to
be killed.”^' Brock’s speculations have carried him beyond the gravita¬
tional field of American morality. The immorality of infanticide is not up
for reconsideration, and it is equally unthinkable that a suicide contract
would be enforced and a person dragged to his death against his will be¬
cause he had signed a contract and was now deemed incompetent to repu¬
diate it, or even that it would be enforced by the forfeiture of a bond or by
some other monetary sanction. We happen to have unshakable moral intui¬
tions concerning the wrongfulness of infanticide and of enforcing suicide
contracts. These intuitions precede and inform, rather than following and
being informed by, philosophical analyses of personhood. If the case for
allowing a person to arrange in advance for his death should he some day
become senile stands or falls on whether infanticide is just, or whether a
monkey or a computer should be deemed more of a person than a severely
demented or profoundly retarded human being, we shall make no progress
in dealing with the senile case, for we shall be up against immovable intui¬
tions concerning the priority of human beings over animals and machines.^^
I do not mean to suggest either that all of our unshakable moral intuitions
are universaF^ or even that they are permanent within our society. Many of
them seem in fact rather local and fluid. But they are not changeable by
reasons, in part because they are not founded on reasons. Because of our
genetic programming, and because of the material conditions of our society
discussed in the preceding chapter, we have more regard for the lives of
infants than for the lives of the senile. That is a good enough ground, even
though it is not a “rational” ground, for decoupling the issues, although
not to the extent of enforcing suicide contracts.
All this talk of unshakable moral intuition may seem inconsistent with
my commitment announced at the outset of this chapter to John Stuart
Mill’s theory of limited government. There is no inconsistency, once the
concept of multiple selves enables us to see that a decision by the current
occupant of one’s body to kill a future occupant is “other regarding” in
Mill’s sense, that is, it imposes a cost on one who has not consented to bear
it. The problem is not serious in the case of physical incapacity, with which
I began. Remember that the vast majority of Dutch euthanasia cases in¬
volve persons with less than a month to live. They are not making decisions

51. Brock, note 48 above, at 385 n. 14.

52. On the priority of intuition over analysis in moral judgments, see my book The Problems
of Jurisprudence 16-11, 339-340 (1990).
53. There are societies (premodern Japan is an example) in which old men are revered and
infanticide tolerated. Cf. chapter 9.
258 Normative Issues

for a future self so remote, so likely to have different values and prefer¬
ences, that it can plausibly be regarded as a different person.
At the level of nonphilosophical practice, the issues of infanticide and
geronticide have been decoupled to a limited extent. Although contracts of
assisted suicide are unenforceable, people have a limited power to bring
about, through the devices of the living will and, especially, the durable
power of attorney for health care, a state of affairs in which they are un¬
likely to survive for long in a severely demented state.But the emphasis
belongs on the word “limited”—and not only because it appears that, as
recently as 1989, less than a quarter of the elderly population had living
wills and an even smaller percentage had granted durable powers of attor¬
ney for health care.^^ (Undoubtedly the percentage has increased, but I have
no current figures. Nor do I know what percentage of the people who need
living wills the most have them, dying people for example; perhaps it is
large.) The more fundamental problem is that the living will is designed for
the case in which “death is imminent except for death delaying proce¬
dures” (I am quoting from the form approved in Illinois), and this point
will not be reached until very late in the progression of the dementia. The
power of attorney is broader, authorizing the holder of the power (again I
quote from the Illinois form) “to make any decision you could make to
obtain or terminate any type of health care, including withdrawal of food
and water.” But like the living will, the power of attorney is revocable as
long as the grantor of the power remains competent, and it may be difficult
to determine when that point has been reached in the progress of the de¬
mentia. And until the demented patient has entered the vegetative state, the
holder of the power will be reluctant to authorize measures that amount to
inflicting death by starvation or dehydration, although it appears that these
deaths can be made relatively painless.
A number of states now have “surrogate decision-making statutes,”
whereby medical decisions, including cessation of treatment, can be made
by a surrogate (normally a close relative) of a patient who is comatose or
otherwise incapable of making decisions and does not have a living will or

54. For background on living wills and durable powers of attorney, see Barry R. Furrow
et al., Bioethics: Health Care Law and Ethics 263-279 (1991); Peter J. Strauss, Robert Wolf, and
Dana Shilling, Aging and the Law, ch. 22 (1990), and Supplement thereto, ch. 10 (1991). And for
a model statute that reflects the latest thinking on the subject, .see National Conference of Com¬
missioners on Uniform State Laws, “Uniform Health-Care Decisions Act” (1993).
55. Wayne Moore, “Improving the Delivery of Legal Services for the Elderly: A Compre¬
hensive Approach,” 41 Emory Law Journal 805, 812-813 (1992).
56. See McCann, Hall, and Groth-Juncker, note 9 above.
Euthanasia and Geronticide 259

a health-care power of attorney.The combination of living wills, health¬

care powers of attorney, and surrogacy statutes has reduced the problems
involved in prolonging the life of comatose patients to manageable pro¬
portions, and I shall not discuss it further. The problem with using these
devices to terminate the life of a severely demented person is that such
a person is not comatose and usually does not appear to be suffering
unbearably.
Notice that by facilitating MDELs other than euthanasia—which
as we saw at the beginning of this chapter are substitutes for physician-
assisted suicide—the living will, the durable health-care power of attorney,
and surrogate decision-making could all be considered substitutes for,
or way stations toward, the legitimization of physician-assisted suicide.
When, pursuant say to a living will, a physician facilitates a patient’s vol¬
untary death of dehydration by giving him salves to relieve the discomfort
of cracked lips incident to dehydration, it is difficult to distinguish what the
physician is doing from assisting in a suicide. Yet the Dutch experience
suggests that MDELs, other than physician-assisted suicide itself, are not a
perfect substitute for physician-assisted suicide. If they were, there would
be no reason to authorize physician-assisted suicide.
Although enforcing a contract of suicide against a person who has
changed his mind about being killed is patently inconsistent with the moral
feelings of our society and not required by Millian liberalism, I feel bound
to point out that the refusal to enforce such contracts may increase the sui¬
cide rate and, what I have explained is qualitatively similar, reduce the av¬
erage age of suicide. People in the early stages of senile dementia often
both know that they have the disease and know that it will get worse. These
people may have several pretty good years before their dementia progresses
to the point at which, ex ante, they would consider themselves better off
dead. Unable to “schedule” death to occur at that cross-over point, how¬
ever, and fearful that when the point is reached they will lack the will or
the means to kill themselves, they may decide to kill themselves earlier,^*
thus losing valuable years of life in order to prevent a more than offsetting
loss of expected utility. Those years would be saved if they could make an

57. Strauss, Wolf, and Shilling, note 54 above, at 607-614; Supplement, note 54 above, at
234-238. See generally James Lindgren, “Death by Default,” Law and Contemporary Problems,

Summer 1993, p. 185.

58. An actual case is discussed in Christine K. Cassel and Diane E. Meier, Morals and
Moralism in the Debate over Euthanasia and Assisted Suicide,” 323 New England Journal of

Medicine 750 (1990).

260 Normative Issues

enforceable agreement to be killed painlessly when a responsible judgment

is made that their quality of life has fallen below the level at which they
would (if capable of making a rational judgment) want to continue living.

The main conclusions of this chapter can be summarized briefly. In cases

of terminally ill, pain-wracked, or severely impaired people who are or
anticipate shortly becoming physically incapable of committing suicide.
Mill’s theory of the proper limits of government suggests, although it
does not prove, that a right of physician-assisted suicide should be recog¬
nized (subject to appropriate safeguards, and to details of design and
implementation that I have not discussed) and that therefore the laws for¬
bidding the practice should be repealed. The fear that under such a regime
physicians will hustle their patients to a premature and undesired death
seems greatly exaggerated; indeed, the suicide rate might actually fall if
physician-assisted suicide were permitted in the subset of cases that I have
described. 1 do not believe, however, that a broader right of physician-
assisted suicide, for example a right to make an enforceable contract to die
when one becomes senile, can be justified on Millian grounds once the
concept of multiple selves is brought into the analysis, as it should be.
I want to close with one of those details of implementation.That is the
political level at which reform should be implemented. I believe that it
should be at the state rather than at the federal level. Today more than ever
the United States is a morally heterogeneous country, and some of the
moral boundaries approximate the boundaries of states. A midwestern or
far-western state inhabited mainly by persons of northern European origin
may occupy a different moral universe from that of most of our southern
states, at least with regard to a number of issues heavily freighted with
religious and emotional meaning, including issues of life and death.I can¬
not think of a compelling reason why, whether through congressional ac¬
tion or judicial interpretation of the provisions of the Constitution that are
applicable to the states, the views of a national majority (or the political or
intellectual elites from which federal courts, especially, often take their

59. Including such minute but important details as whether physician-assisted suicide
would, if made legal, still count as “suicide” within the meaning of life insurance policies that
contain an exception (usually limited, however, to the first two years in which the policy is in
effect) for self-inflicted death. Cf. Robert J. Kovacs, “Insurance Issues in Physician-Assisted Sui¬
cide,” New Jersey Lawyer, Sept. 26, 1994, p. 15. A number of living-will statutes provide that
death which results from the honoring of the decedent’s living will is not a suicide. Legal Counsel
for the Elderly, Decision-Making, Incapacity, and the Elderly 24 (1987).
60. Notice that Dworkin bracketed abortion and euthanasia in his book Life's Dominion,
note 7 above.
Euthanasia and Geronticide 261

cues) concerning such a matter as physician-assisted suicide should be im¬

posed on the entire nation. As I mentioned at the outset of this chapter,
Oregon recently became the first state to authorize a form of physician-
assisted suicide; a bill is pending in the Connecticut legislature that would
do the same thing; legislative activity in other states can be anticipated;
in still other cases the cause is and will long remain hopeless. This pat¬
tern, this ferment and variety, is as it should be. Each state should be
allowed to decide, and premature nationalization (illustrated by Roe v.
Wade, which interrupted and preempted a liberalizing tide of state abortion
laws) avoided.

61. 1994 Conn. Sen. Bill No. 361.

 11
Social Security and Health

The question whether to permit people in general and the elderly ailing in
particular to end their lives draws on ethical and economic arguments that
turn out to be helpful in resolving other policy issues concerning the el¬
derly, as I hope to show in this chapter. The most important of these poli¬
cies, and the ones on which I shall focus, are the social security program of
retirement benefits and the public provision of health care (comprising ex¬
penditures on medical research as well as on treatment) for the elderly.

The Compulsory Character of Social Security

I begin with an issue curiously linked through the concept of multiple
selves with the subject of the preceding chapter. That concept illuminates a
feature of the social security retirement program—namely, the fact that it
is compulsory—that only seems remote from euthanasia. If you work for
an employer covered by social security (which is now virtually every em¬
ployer, including the self-employer), you are forced to contribute to the
social security program; you cannot make a side deal with your employer
whereby in exchange for a higher wage you agree that neither you nor he
will make any contribution to the program and in consequence you will
have no entitlement to social security benefits when you reach retirement
age. Nowadays the main practical reason for the compulsory character of
social security is to make sure that there is money in the social security till
to honor someone else’s social security entitlement. But the original ratio¬
nale was to force people to save for their old age; and even those critics of

262
Social Security and Health 263

the social security system who would prefer to substitute an entirely private
system believe (most of them anyway) that people should be compelled to
enroll in a private pension system and thus to make provision for their old
age. And likewise with the Medicare component of social security: most
critics of Medicare would support an alternative system under which
people would be forced to buy private medical insurance for their old age.
The reasons most commonly offered for the compulsory character of
the social security program are two. The first is paternalism: people are
short-sighted and therefore cannot be trusted to make arrangements for the
distant future. This would be a good reason if people typically underesti¬
mated their life expectancy and therefore the amount of money they should
be setting aside for consumption (including consumption of medical ser¬
vices) in old age. But the evidence is to the contrary.' The second reason is
Realpolitik: society won’t in fact let people starve to death, or die because
they cannot pay for essential medical treatments, so the nonsavers would
be free riders.^ This reason has some support in the fact that persons who
have not contributed to social security are entitled to a modest government
pension anyway and that the indigent elderly received medical care on a
charity basis even before there was a Medicaid or Medicare program.
The concept of multiple selves enables a different approach to be
taken. A at working age, especially at young working age, is a different
person from A at retirement age. A„ should not be allowed to condemn A^
to penury by refusing to make any provision for the support of A,, who
will be unable to support himself. A„, on this view is a kind of trustee of A,
the body that A„ and A, successively inhabit. A compulsory pension sys¬
tem, like a prohibition against enforceable contracts of assisted suicide,
imposes a limited fiduciary duty on the young self. Lawyers will perceive
an analogy to the duty that a life tenant owes a remainderman not to waste
the assets of the property in which they have their successive interests, as
by cutting trees before the trees have reached maturity. The analogy is help¬
ful in showing that the multiple-selves argument for compulsory social se¬
curity, unlike the paternalistic argument, does not depend on any notion
that people are ignorant or that government knows best.
Lfnfortunately, as we have already seen, there is no good method of
determining the relative weight to give the preferences of the different
selves. The clearest limit on the claims of the old is the case where the

1. See Daniel S. Hamermesh, “Expectations, Life Expectancy, and Economic Behavior,”

100 Quarterly Journal of Economics 389 (1985).
2. See, for example, Laurence J. Kotlikoff, Avia Spivak, and Lawrence H. Summers, The
Adequacy of Savings,” 72 American Economic Review 1056 (1982).
264 Normative Issues

young self wants to sacrifice those claims on behalf of other selves and so
is not being selfish. Recalling the trade-off identified by evolutionary biol¬
ogy between the longevity and the reproductive fitness of an organism, we
can see that the old self should not be heard to complain if the young self
decides to divert resources from the old self to the production of children;
that is not a selfish expenditure by the young. But beyond that, perhaps we
can do no better than to allow the political process to arbitrate the compet¬
ing claims of successive selves, as discussed in the last section of this chap¬
ter. Contrary to general fears, that process is unlikely to subordinate the
interests of the young to the interests of the old completely.

The Allocation of Medical Resources to the Elderly

and between Elderly Men and Women
A question even more closely related to that of geronticide is whether so¬
ciety has any obligation to furnish medical care (other than care designed
for the alleviation of pain) to the severely demented. This question blends
into the broader one of how much and in what form to devote resources to
the medical needs of elderly persons—persons who, to put it bluntly, will
soon be dead anyway—and into the still broader question, analyzed exten¬
sively but inconclusively in the biomedical-ethics literature under the rubric
of “triage,”^ of the allocation of scarce medical resources among compet¬
ing demanders. I believe that economics can reduce the indeterminacy of
that analysis, both generally and with particular reference to the question
of medical care for the elderly.

Private expenditures. We should distinguish between allowing an in¬

dividual or his family to spend his own or its own resources on medical
care, a negative liberty, and forcing the taxpayer to pay for it, a positive
one. Even the negative liberty is not completely unproblematic. The private
health expenditures of the old, that is, expenditures over and above Medi-

3. See, for example, Robert T. Francoeur, Biomedical Ethics: A Guide to Decision Making,
ch. 5 (1983); Robert W. Derlet and Denyse A. Nishio, “Refusing Care to Patients Who Present to
an Emergency Department,” 19 Annals of Emergency Medicine 262 (1990); and, for the fullest
discussions, John F. Kilner, Who Lives? Who Dies? Ethical Criteria in Patient Selection (1990),
and Gerald R. Winslow, Triage and Justice {\9S2). The term “triage” (the French word for “sort¬
ing”) used to be employed much more narrowly, to denote methods for maximizing the value of
emergency medical treatment of battlefield casualties, as by dividing the wounded into those likely
to survive without immediate treatment, those likely to die even if they received immediate treat¬
ment, and those likely to survive if but only if they received such treatment, and treating only the
third group, as distinct from treating everyone on a first-come, first-served basis.
Social Security and Health 265

care (which does not reimburse the total expenses of Medicare patients)
and other public programs, are large enough to affect the costs of medical
care to other people. Persons 65 and older, although less than 13 percent of
the American population, account, as noted in chapter 2, for roughly one-
third of all expenditures on health. Two-thirds of the elderly’s tab is picked
up by the government through Medicare and other public programs,"* but
this means that more than 11 percent of the nation’s total expenditures on
health are private expenditures by the elderly. If a service is provided under
conditions of increasing average cost, an increase in demand will raise the
market price, and everyone who shops in the market will pay more than he
did before. I do not know whether this is true of medical care, especially in
the long run, when supply is more elastic. But if it is true, the large private
demand by the elderly for health care has probably driven up the prices
paid by other consumers of medical services. Such an effect on other con¬
sumers, however, would be a purely pecuniary externality; that is, it would
be completely offset by the increased revenue of the sellers of medical ser¬
vices. The distribution of income across persons or groups might be af¬
fected, but the total wealth of society would be unchanged. Whether eco¬
nomic equality would be promoted is anybody’s guess. Not all providers of
medical services are wealthy, by any means; indeed most health-care work¬
ers are rather poorly paid.
Public expenditures. The analysis would be the same if, instead of
paying as they went, the old bought medical insurance policies when young
that guaranteed the payment of their medical expenses when they grew old.
(This is easier said than done, since future medical costs cannot be forecast
reliably decades in advance. But that is a separate problem, though one to
which I return.) One might even suppose the analysis unchanged if the
expenditures of the old on health care are subsidized, as of course they are,
primarily by the federal taxpayer through the Medicare program.^ It is true
once again that to the extent that essential inputs into health-care services
are in irremediably short supply, an expansion of the health-care sector will
result in higher prices. But we have just seen that the same thing would
happen, under the identical assumption of a long-run rising average cost of
health care, if the old decided to spend more of their own money on health

4. U.S. Senate Special Committee on Aging et al., Aging America: Trends and Projections
133 (1991 ed.).
5. To the tune of $102 billion in 1991; $126 billion if other federal programs that provide
subsidized medical care to elderly persons are included, primarily the portion of Medicaid that
goes to pay the medical expenses of the elderly poor. Id. at 239 (tab. 8-1).
 266 Normative Issues

care. The difference between the two cases is that a public subsidy of medi¬
cal care reduces the total value of the economy’s output of goods and ser¬
vices, by inducing the old to substitute medical care for things they would
value more highly if medical services were priced to them at their market
value. If the total annual expenditures on Medicare were simply given to
the Medicare-eligible population in cash,® the old would be unlikely to
spend all their new wealth on medical care. An elderly person who now
receives $20,000 in social security retirement benefits a year would not, in
all likelihood, if given another $10,000 a year to spend as he wants, use it
all to buy health insurance, even if he could not count on public or chari¬
table assistance if he got ill and his insurance benefits ran out. Elderly
people sometimes elect, or are nudged by their physicians or by hospitals
to elect, marginal medical procedures because the private cost (what they
pay) is so much lower than the social cost (what society pays). Medicare
has thus brought about a misallocation of resources from the standpoint of
economic efficiency.
Is it a serious misallocation? I am not so sure. Although Medicare is
too generous in the economic sense of giving the old more medical care
than they would pay for if they had an undistorted choice among competing
goods and services, no one seems to know how serious the distortion is.^
And Medicare is not too generous—Ronald Dworkin to the contrary not¬
withstanding*—merely because old people receive more medical care than
they would when young sign a contract to receive in a lifetime health-
insurance policy. Dworkin is no doubt correct that most young people
would not buy a policy that required heavy premiums to defray the ex¬
pected cost of dramatic though usually futile medical interventions in the
last few weeks or months of life. But his argument is nevertheless vulner¬
able not only to the obvious objections that implementation would require
far more information than the government could plausibly be expected to
obtain and that no one knows how to price lifetime medical insurance, but
also to the subtler objection that to allow the young to make life and death

6. Implying a roughly 50 percent increase in the size of social security pensions See id at
239 (tab. 8-1).

7. See generally Jerry L. Mashaw and Theodore R. Marmor, “Conceptualizing, Estimating,

and Reforming Fraud, Waste, and Abuse in Healthcare Spending,” 11 Yale Journal on Regulation
455 (1994).

8. Dworkin, “Will Clinton’s Plan Be Fair?” Aew J?evt(?w£>/Soo/tj, Jan. 13, 1994, p. 20.
For a similar argument, see Dan W. Brock, Life and Death: Philosophical Essays in Biomedical
Ethics 358—360 (1993) ( prudential allocator” approach). Survey evidence, reviewed in James
Lindgren, “Death by Default,” Law and Contemporary Problems, Summer 1993, p. 185, supports
Dworkin’s premise that people in some .sense don’t “want” expensive end-of-life treatments.
Social Security and Health 267

decisions for the old is to give one person, the younger self, undue con¬
trol over a resource (a body) shared with another, the same individual’s
older self.^
This is a common oversight in the philosophical analysis of aging—
surprisingly so, since “deconstructing” the self is the sort of thing philos¬
ophers like to do. Norman Daniels, discussing the issue of justice between
generations, argues that because youth and age are merely different stages
of a single life, treating them differently “generates no inequality at all...
From the perspective of stable institutions operating over time, unequal
treatment of people by age is a kind of budgeting within a life.”Yet we
saw in earlier chapters that it is entirely rational for old people to spend
heavily on medical care to extend their lives; they do not have good alter¬
native uses of their resources. The young self may scant his future self’s
interest in extending life not because the young self is short-sighted or lacks
self-control but simply because it has different preferences.
The point is also overlooked by even the best health economists. Sher-
win Rosen points out that “focus group explorations of hypothetical life-
path experiences showed graphically that people in their twenties have little
or no interest in their health prospects for their seventies or even their fif¬
ties. A different picture emerges from the responses of people in their fifties
or sixties.”" But Rosen does not think that this presents any normative
problem concerning the behavior of the young, provided they have accurate
information.'^
Dworkin proposes to limit the amount of medical care available to the
elderly to what the young self would, under conditions of complete infor¬
mation and a just distribution of wealth, be willing to pay for. Not only is
the proposal rendered problematic by multiple-selves analysis and the
other objections I mentioned; but it is morally unacceptable for much
the same reason as Brock’s somewhat similar proposal for dealing with the

9. He alludes to this problem in his book Life’s Dominion: An Argument about Abortion,
Euthanasia, and Individual Freedom 257 n. 12 (1993), which refers the reader to a longer, unpub¬
lished study. But both the note and the study are confined to the special issue (which I discussed
in the preceding chapter) whether we should be able to make decisions binding our future senile
self.
10. Norman Daniels, “Justice and Transfer between Generations,” in Workers versus Pen¬
sioners: Intergenerational Justice in an Ageing World 57, 61,63 (Paul Johnson, Christoph Conrad,
and David Thompson, eds., 1989). See also Margaret P. Battin, “Age Rationing and the Just Dis¬
tribution of Health Care: Is There a Duty to Die?” 97 Ethics 3\1 (1987).
11. Sherwin Rosen, “The Quantity and Quality of Life: A Conceptual Framework,” in Valu¬
ing Health for Policy: An Economic Approach 221, 247 (George Tolley, Donald Kenkel, and Rob¬

ert Fabian, eds., 1994).

12. Id. at 244, 246.
268 Normative Issues

severely demented, which I discussed in the preceding chapter. Let me turn

personal again. In 1987, at the age of 87, my mother broke her hip. When
this happened she was already a frail and shrunken old lady with the char¬
acteristic symptoms of senile dementia of moderate severity: she suffered
from occasional disorientation and confusion, and her short-term memory
was gone, causing her conversation to be extremely repetitive. But her es¬
sential rationality was intact, and she was living at home with my father.
Before the medical advances of recent decades, death would have been the
sequel to my mother’s breaking her hip, and she would have been thought
someone who had reached the natural terminus of a long life. Instead the
hip was skilfully repaired, and she lived on. She never walked again, how¬
ever, and had to be placed in a nursing home. Her mental condition contin¬
ued to deteriorate. Eventually she lost the use of her hands and her power
of speech, and ceased to recognize anyone. She was finally carried away
by pneumonia at the age of 90, a shadow, totally bereft of reason (“Without
the which we are pictures, or mere beasts”).'^
In 1992, my father-in-law, aged 75 and in ostensibly good health, was
diagnosed by doctors at a distinguished medical center as having an un-
treatable cancer and given three months to live. He accepted the diagnosis
and went home. Three months later he began to develop some respiratory
distress. A radiologist told him that unless he underwent chemotherapy he
would be dead in two weeks. The radiologist did not tell him what would
happen if he did undergo chemotherapy. My father-in-law decided in favor
of the chemotherapy and was dead after three ghastly weeks. His medical
expenses (all picked up by Medicare) for those last three weeks came to
$30,000.
The expenditures incurred in prolonging my mother’s life by three
years and my father-in-law’s by one week were, in my opinion, wasted. My
mother’s and my father-in-law’s younger selves would surely have agreed
with me on this point, and might even have been willing to sign a contract
to that effect. But I cannot think of any feasible method of enforcing such
a contract except by concealing treatment options from elderly people,
which I do not think Dworkin would approve of doing. You just do not tell
people in our society that, although the technology exists to prolong their
lives, a judgment has been made that the costs exceed the benefits and
therefore the technology will be withheld from them. You do not tell them
this, pace Dworkin, even if you are in a position to remind them that they
agreed when young not to insist on expensive end-of-life treatments. My

13. Hamlet, act IV, sc. v, 1. 86.

Social Security and Health 269

father-in-law’s radiologist can be faulted for not having advised my father-

in-law of the likely futility and ugly side-effects of the course of chemo¬
therapy being offered, but I do not think the advice would have changed
my father-in-law’s decision. Americans are fighters, not fatalists.
Of course there is a point beyond which a society will refuse to keep
old people alive. That point can be reached quite early in a very poor soci¬
ety, as we saw in chapter 9. But in a society as wealthy as ours, the elderly
would have to be consuming a far larger share of the medical budget than
they are doing even today for society to decide to close the spigot. My
father-in-law was well to do and could have paid for the chemotherapy out
of his own assets; and in the case of my mother, the major expense of her
protracted old age was the expense of the nursing home, which Medicare
did not reimburse. It is right to force elderly people to “spend down” before
they qualify for public assistance. But it would not be consistent with con¬
temporary morality to take the next step, as Dworkin would do, and deny
all public assistance (and private as well?) that goes beyond what a repre¬
sentative younger self would have agreed to set aside for the medical ex¬
penses of his elderly self.
The young self will not be permitted fully to control the old one. And
because there is no satisfactory or acceptable analytical procedure for bal¬
ancing the claims of the young and of the old self, the current old, in effect
as proxies for the future old selves of the current young, struggle with the
current young in the political marketplace for the allocation of consump¬
tion over the life cycle. The increasing productivity of medical expendi¬
tures in extending life and in improving the health of the old is increasing
the intergenerational tension. The Medicare hospital trust fund is expected
to be depleted by the year 2001,'^ at which point the social security tax will
have to be raised, benefits cut, or other steps to restore balance taken. The
old self has an argument (if only it could make it!) that the ever-increasing
productivity of expenditures on medical care warrants a reallocation of re¬
sources from the young self; that the marginal dollar will purchase more
utility for the old self than it would for the young. The young self, however,
may disagree.

Research priorities. The issue of subsidizing health care for the el¬
derly population arises in still another form, that of federal support of
medical research on diseases such as heart disease and cancer that afflict
old people disproportionately. Not all serious diseases do. Asthma and mi-

14. Robert Pear, “Benefit Funds May Run Out of Cash Soon, Reports Warn,” New York
Times (national ed.), April 12, 1994, p. A12.
270 Normative Issues

graine, for example, as well as certain cancers—and of course AIDS—are

more common among young than among old people, whereas heart condi¬
tions are 10 times more frequent among men 65-74 years old than among
all men under 45 and 15 times more frequent among men 75 and older than
among the under-45 set.'^ The U.S. Public Health Service spends much of
its annual research budget on diseases that afflict old people disproportion¬
ately (with most of the rest going to AIDS).'*’ The marginal benefit of these
expenditures may be small, quite apart from “value-of-life” considera¬
tions. Old people are highly vulnerable to a large number of lethal or inca¬
pacitating diseases. In effect these diseases compete to kill or grievously
impair the old. Curing one or two such diseases will eliminate competition
for the other diseases, enabling the latter to do more harm than would have
been the case had the competitors remained in the field. Curing heart dis¬
ease saves the patient for cancer, and curing both would save him for ne¬
phritis, blindness, or Alzheimer’s. The “benefit” that these diseases derive
from medical research that cures their rivals is enhanced by the fact that
improvements in medical technology benefit persons of weak constitu¬
tion disproportionately, thus providing easier targets for the lying-in-wait
diseases.'^
This point undermines the kind of naive cost-benefit analysis in which,
for example, the benefits of measures to reduce the number of accidental
injuries to elderly people are assumed to equal the medical costs of treating
those injuries.'* The assumption disregards the fact that avoiding such an
injury sets up the elderly person for another problem that will require medi¬
cal treatment that could cost as much or even more. The cost of that treat¬
ment should be added to the cost of treating the accidental injury in decid¬
ing whether the latter treatment is cost-justified.
This is not to suggest that the benefits of measures to improve the
health of elderly people are zero. Although longevity is positively related
to income and the income of the old has soared in recent times, the increase
in longevity (and the reduction in the fraction of the aged that is disabled)

15. See U.S. Bureau of the Census, Statistical Abstract of the United States 1993 135
(113th ed.)(ser. 206).
16. William Winkenwerder, Austin R. Kessler, and Rhonda M. Stolec, “Federal Spending
for Illness Caused by the Human Immunodeficiency Virus,” 320 New England Journal of Medi¬
cine 1598, 1602 (1989) (tab. 4).
17. See Donald S. Shepard and Richard J. Zeckhauser, “The Choice of Health Policies with
Heterogeneous Populations,” in Economic Aspects of Health 255, 268 (Victor R. Fuchs, ed.,
1992). The age-race “crossover” in longevity, discussed in chapter 9, illustrates this point.
18. As in Amy B. Bernstein and Claudia L. Schur, “Expenditures for Unintentional Injuries
among the Elderly,” 2 Journal of Aging and Health 157 (1990).
Social Security and Health 271

has far exceeded what is plausible to assign to the increase in income.'®

“Saving” an old person with heart disease for cancer will still give him
additional years of life, in all likelihood; and we know that increases in
longevity confer substantial private benefits on the elderly, especially if I
am correct that the length of remaining life is not an important variable in
the demand for life. We also know that concerns that an increase even in
healthy longevity will have disastrous social consequences by raising the
dependency ratio probably are exaggerated.
Yet it does not follow that just because a young person and an old
person may feel an equal dread at the prospect of imminent death, a cure
that saves many years of life confers no greater utility than one that saves
only a few. This is a case in which both ex ante utility and ex post utility
are relevant to the choice of policy. Where the former seems clearly pref¬
erable is in cases in which choices sensible when made turn out badly: one
takes a fair gamble, but loses and now wants one’s stake returned. To use
assessments of ex post utility to invalidate ex ante choices would greatly
reduce the scope of free choice. In the long run, ex post as well as ex ante
utility—utility, period—would be diminished. But if young and old dread
death equally, the only basis for choosing between them, if a choice must
be made, may be the difference in ex post utility—that the young person
will live longer if he is saved.
Even so, it is merely likely, not certain, that medical research that pri¬
marily benefits the old is a poorer investment from the social standpoint
than medical research that primarily benefits the middle-aged.^° Consider
two alternative medical investments, costing the same and having the same
probability of success. (These qualifications are vital and should be borne
in mind throughout the discussion. The net benefits of a research program
depend not only on the benefits if the program is successful, including spill¬
overs to research on other diseases, but also on the cost of the program and
the probability of success.) One investment will extend the life of an 80-
year-old to 85 and the other the life of a 60-year-old to 80. In strictly finan¬
cial terms, the former is quite likely to be the better (which is not to say a
good) investment. For it will add “only” 5 years of old age, while the latter

19. For evidence that improved treatment of the characteristic diseases of elderly people is
responsible for much of the increase in longevity in recent decades, see George C. Myers and
Kenneth G. Manton, “The Rate of Population Aging: New Views of Epidemiologic Transitions,”
in Aging: The Universal Human Experience 263 (George L. Maddox and E. W. Busse, eds., 1987).
20. Cf. Tomas J. Philipson and Richard A. Posner, Private Choices and Public Health: An
Economic Perspective on the AIDS Epidemic 120-125 (1993), and references cited there, discuss¬
ing the possible fiscal savings from a disease that kills young people.
272 Normative Issues

investment will add 5 years of productive life and (assuming retirement at

age 65) 15 years of old age during which the individual will be receiving
social security retirement benefits and incurring heavy, and heavily subsi¬
dized, health costs.^' Those costs may be especially heavy because the in¬
dividual is constitutionally weak, which is why he would have died at the
age of 60 had it not been for the new medical technology.
But costs to the public fisc do not exhaust the considerations relevant
to evaluating a public investment. Success in fighting the disease that kills
the younger individual is, as we just saw, likely to create greater nonfinan-
cial utility than a similar success with a disease of the old. This is not only
because more years of life will be saved and most people derive utility from
living, and not only because healthy younger years may confer more utility
than sickly older ones. It is also because the older a person is, the fewer
surviving family members he is likely to have and the less they are likely
to grieve at his death.^^ (The costs of death are not borne by the dying
person alone.) I conclude that failing to give priority to life-threatening
diseases of the young would signify an inefficient allocation of resources
to medical research unless the old were being short-changed in other gov¬
ernment services that their taxes support, which is unlikely, or unless medi¬
cal research on the diseases of the elderly would produce much greater
savings of life relative to expenditures.
These are important qualifications and another is that efforts to cure
diseases that are greatly feared because they cause premature death will
willy-nilly prolong the lives of old people. Heart disease and cancer are
principal examples, except to the extent that research can fruitfully be sepa¬
rated between forms of these diseases that affect the young more and those
that affect the old more. The incidence of a number of cancers differs dra¬
matically by age group. For example, the incidence of Hodgkin’s disease
persons is higher in the 20 to 24 age group than in the 85 and over group
(4.8 versus 3.8 per 100,000), while conversely the incidence of prostate

21. Thomas Schelling, “Value of Life,” in The New Palgrave Social Economics 269, 270
(John Eatwel, Murray Milgate, and Peter Newman, eds., 1989). In this vein, it has been argued
that “smokers ‘save’ the Social Security system hundreds of billions of dollars.” John B. Shoven,
Jeffrey O. Sundberg, and John P. Bunker, “The Social Security Cost of Smoking,” in The Econom¬
ics of Aging 231, 244 (David A. Wise, ed., 1989); see also W. G. Manning et al., “The Taxes of
Sin: Do Smokers and Drinkers Pay Their Way,” 261 JAMA (Journal of the American Medical
Association) 1604 (1989).
22. Weighting number of family members by closeness of kinship. The longer a person lives,
the more likely he is to have surviving grandchildren and other remote descendants, but they are
much less likely to grieve deeply for him than a spouse, a parent, or a young child. Another factor
in the comparison, to which I return shortly, is the quality of additional years of life to young and
to old people respectively.
Social Security and Health 273

cancer is 0.0 per 100,000 in the younger age group and 328.8 per 100,000
in the older.^^
Still another complication is that young people are better able than old
ones to avoid disease by making changes in their style of living, as by
giving up cigarettes, losing weight, reducing the amount of fat in their diet,
or moderating their intake of alcohol; most elderly people have already
made these changes. The effect of better treatment for diseases of the young
may be to induce the young to relapse into unhealthful habits, since those
habits cost less the lower the expected disease costs to which the habits
give rise. Such relapses, possibly illustrated by the recent increase in obe¬
sity in the American population,^'' reduce the effectiveness of medical re¬
search on diseases of the young, although they confer net utility on the
young.
Another implication of the analysis in this section is that, other things
being equal, from a strictly financial standpoint more resources should be
devoted to the prevention and treatment of mental illnesses than of physical
illnesses. The former are more likely to increase disability without short¬
ening life, so curing them is more likely to remove a disability than to
lengthen life. Enabling a person to become a productive worker, without
lengthening his life, lowers the ratio of dependent to productive persons. It
might therefore be in the interest of the old as a group to support some
reallocation of research and treatment funds to mental illnesses.

The question of gender. A neglected and sensitive issue—neglected

perhaps because it is sensitive—is the allocation of public funds between
research on diseases of old men and research on diseases of old women.
We saw in chapter 2 that the life expectancy of women in the United States
greatly exceeds that of men and that the difference translates into a decided
preponderance of women in the older age groups. Justice between the sexes
might seem to require, therefore, that medical research on the diseases of
the old be tilted in favor of diseases, such as prostate cancer and coronary
artery disease, that kill old men but not (or, as in the case of coronary artery
disease, not as soon) old women.^^ Feminists might object that women

23. U.S. Dept, of Health and Human Services, National Cancer Institute, “Annual Cancer
Statistics Review, Including Cancer Trends: 1950-1985” III.B.30 (Jan. 1988).
24. Marian Burros, “Despite Awareness of Risks, More in U.S. Are Getting Fat,” New York
Times (national ed.), July 17, 1994, p. 1.
25. In 1990, cancers of the genital organs, which include the prostate, killed 358.5 out of
100,000 men in the 75-84 age group but only 95.3 out of 100,000 women, a difference of almost
4 to 1. U.S. Bureau of the Census, note 15 above, at 97 (ser. 133). Heart disease of all types killed,
in this age group, 2,968.2 men out of every 100,000 but only 1,893.8 women. Id. at 96 (ser. 132).
 274 Normative Issues

should not be penalized for their “natural” advantage in longevity over

men. But such an objection would rest on a biological essentialism rejected
by feminists in most other policy settings, such as maternity leave, preg¬
nancy benefits, and abortion rights. Nor is it much of an argument that the
greater longevity of women merely compensates them for the higher death
rates of women before medicine eliminated most of the risk of dying in
childbirth. Women who died long ago are not compensated by the longer
life expectancy of modern women.
Here are better, though not necessarily decisive, arguments against try¬
ing to equalize the life expectancies of men and women. First, government
should not buy into the feminist rejection of biological essentialism and try
to offset natural differences between the sexes—especially since govern¬
ment has not as yet done much to offset the natural disadvantages of
women. (The abolition of the military draft eliminated a traditional advan¬
tage of women.) Second, women can take steps to reduce their own life
expectancy and increase that of their husbands. Third, women’s natural ad¬
vantage in longevity may be, in part anyway, illusory; as the occupational
profiles of men and women converge, so may their mortality statistics.
Robert Arking reports an estimate “that about 18% of the sex differential
in total mortality may be due to . . . sex-specific hormonal effects on the
cardiovascular system, and thus may well represent an intrinsic and perhaps
unchangeable risk to males,” while “at least 55% ... can be attributed to
destructive behaviors,”^® such as smoking, which are influenced by occu¬
pation and other social factors. Even if women do not become as self¬
destructive as men as the occupational profiles of the two sexes converge,
so that a “voluntary” difference in longevity persists, to pour research
funds into diseases avoidable by behavioral changes will reduce the incen¬
tive to make those changes and so may have only a minimal effect on the
difference.
Fourth, to the extent that men are inherently more vulnerable than
women, expenditures on fighting the diseases of men may have a lower
payoff in years of life saved because “competition” between diseases to
kill men is more intense. A hundred million dollars spent to develop a cure
for some disease of women might add a month to female longevity, yet the
same expenditure to develop a cure for a disease of men that had the same
prevalence might add only three weeks to male longevity.

26. Robert Arking, Biology of Aging: Obserx'ations and Principles 223 (1991); see also
William R. Hazzard, “Why Do Women Live Longer Than Men? Biologic Dilferences That Influ¬
ence Longevity,” 85 Postgraduate Medicine 271 (1989).
Social Security and Health 275

Table 11.1 Government Expenditures per Death for Various

Diseases

Disease Expenditures per Death

AIDS $79,000

Cervical cancer 7,300

Diabetes 6,300

Kidney disease 5,100

Breast cancer 2,800

Heart disease 1,100

Prostate cancer 800

Lung cancer 600

Stroke 600

So there may be substantial unexploited gains from expenditures on

research into the diseases of women. This would be especially likely if at
present the research budget were heavily tilted in favor of men, but
table 11.1 suggests that it is not.^^ Notice the much higher ratio of expen¬
ditures to deaths for breast and cervical cancer (female) compared to heart
disease, lung cancer, and prostate cancer (primarily or exclusively male). It
is true that AIDS, a disease predominantly of men, swamps all the other
diseases, so far as generosity of government spending is concerned. But
AIDS is not a disease of elderly men. Of all AIDS cases reported in the
United States through June 1994, only 2 percent were of men aged 60-64
at age of diagnosis and only 1 percent were of men 65 or older at age of
diagnosis.^*
Yet even if old men are not being favored by the research establish¬
ment, the fact remains that a dollar spent on fighting women’s diseases is
likely to buy more longevity than a dollar spent on fighting men’s diseases.
This may seem a conclusive economic argument against reallocating medi¬
cal expenditures from old women to old men, by establishing that the mar¬
ginal benefit of curing women’s disease must exceed that of curing men s
diseases. It establishes no such thing. Utility and longevity are related, but
they are not interchangeable. A simple form of the relation is given by

27. The source for the data in table 11.1 is “Sickness and Politics: The Influence of Disease
Advocates: Cancer vs. AIDS,” Cancer Weekly, June 8, 1992.
28. U.S. Dept, of Health and Human Services, Centers for Disease Control and Prevention,
HIV/AIDS Surveillance Report, Mid-Year Edition, 1994, p. 13 (tab. 8). The corresponding percen¬
tages for women are 1 (60-64) and 2 (65 and over), but fewer than 20 percent of all cases in the
60-and-over age group are women. Id. at 13 (tab. 8).
276 Normative Issues

U = N-Y-V(R). (11.1)

U is the utility of spending one more dollar for medical research on a dis¬
ease that is likely to shorten the life of either men or women. N is the
number of either elderly men or elderly women and Y is the average num¬
ber of years of life, per individual of the given sex, that the additional dollar
spent on medical research saves. V is the value to the average individual of
each unit of Y and is a function of R( - NJN^), the ratio of elderly men to
elderly women.
In this equation, a shorter extension of life (F) for men, and even a
smaller number of men {N), can be offset by a higher value per unit of
additional life (V). Given the imbalance between the number of elderly
men and the number of elderly women, an extra month of life for an elderly
woman is not necessarily worth as much (in a utilitarian, not a financial,
sense) as an extra month of life for an elderly man. That is, may be
lower than V„, offsetting the effect of YJs being higher than Y„. Notice
the dual significance of N, the number of men (or women). The more
women there are, the more, other things being equal, research expenditures
that extend the lives of women will increase total utility. But the more
women there are relative to men (in other words, the smaller R is), the
likelier is the value of extending the life of an elderly man by a given
amount to exceed the value of extending the life of an elderly woman by
the same amount (that is, dVJR)/^R > 0, dVJR)/dR < 0), since a scarcity
of elderly men increases women’s demand for longer male life.
The trade-off can be made more transparent by breaking equation 11.1
into two equations, one for men and one for women, as in

U^ = Y^-VJR), (11.2)

= N^-Y„-VJR), (11.3)

and then substituting NJN„ for R. It is easily shown that will exceed
—the expenditure on the men’s disease will create the greater utility—if

Y^/Y^ ■ VJN„,/NJ/VJN„,IN„) ■ NJN^. >1. (11.4)

I have assumed that the first term is smaller than 1—that a dollar expended
on fighting women’s diseases will extend life by more than a dollar ex¬
pended on fighting men’s diseases. And the third term also—there are more
old women than old men. But what about the second term? The lower the
ratio of old men to old women, the likelier is the second term, which mea¬
sures the relative value of extending elderly male versus elderly female life
by a given amount, to exceed 1. It may do so by enough to make the in-
Social Security and Health 277

equality hold. Of course, to the extent that men benefit from a surplus of
women, an increase in the ratio of men to women reduces the value of
increased male longevity to men; but I assume that this offset is slight.
The argument that I have sketched may sound, if not sexist, at least
adverse to women’s interests. Not so. Women as a group might benefit from
policies that promoted greater equality in the number of elderly men and
women—for example policies that added a year to female longevity but
two years to male longevity—because it would give elderly women a
greater prospect of male companionship, something many of them greatly
value,^^ though radical feminists believe them mistaken to do so. We know
from chapter 6 that the continuation of sexual activity on the part of elderly
women is heavily dependent on marital status.^® And—though this will
change as women work more and accrue greater pension rights—elderly
women who are married are far better off financially than ones who live
alone.^' A much higher fraction of men than of women aged 65 and over
are married, and the disparity grows with age. Male-female differences in
widowhood are particularly striking. In the 65-69 age group, only 7.3 per¬
cent of men, but 33.8 percent of women, are widowed; in the 80-84 age
group, the figures are 27.3 and 72.3 percent.^^ ^hese disparities would be
smaller if men lived as long as women.
In the language of economics, male longevity is complementary to fe¬
male. Hence an increase in male longevity will benefit women, who may
therefore be willing to pay ex ante by a reallocation of medical resources
from the prevention and cure of women’s diseases to the prevention and
cure of men’s diseases.
This suggestion will seem less outre if one understands its relation to
the question whether to shift some medical resources from the diseases of
older people to the diseases of younger people. If adding a year of life to a
65-year-old would confer greater utility than adding a year of life to a 75-
year-old, then adding a year of life to an old man is likely to confer greater
utility than adding a year of life to an old woman, quite apart from the
imbalance in numbers, simply because the average old man is younger than

29. For evidence, see Jane Jraupmann, Elaine Eckels, and Elaine Hatfield, “Intimacy in
Older Women’s Lives,” 22 Gerontologist 493 (1982).
30. For other evidence of the importance of marriage to many older women, see id.; and
Timothy H. Brubaker, “Families in Later Life: A Burgeoning Research Area,” 52 Journal of
Marriage and the Family 959, 965-966 (1990).
31. F. N. Schwenk, “Women 65 Years or Older: A Comparison of Economic Well-Being by
Living Arrangement,” 4 Family Economics Review, no. 3, 1991, p. 2.
32. Jacob S. Siegel, A Generation of Change: A Profile of America’s Older Population 364-
365 (1993) (tab. 6A.1).
278 Normative Issues

the average old woman. Notice that this analysis also implies that it might
be a good idea to shift medical resources from diseases that particularly
endanger whites to those that particularly endanger blacks, given the sub¬
stantially shorter life expectancy of blacks than of whites. Yet the case for
shifting medical resources from women to men is actually stronger than
that for shifting such resources either from whites to blacks or from older
to younger persons regardless of sex. Once again, the reason is selection
bias. A 75-year-old with one year to live may be no more unhealthy than a
65-year-old with one year to live, since the older person, simply by virtue
of having survived longer, is likely to have aged more slowly. But in the
case of reallocation between the sexes, keeping the (weaker) male alive
another year benefits not only him but also his spouse, by postponing her
widowhood; and the only cost is that her life as a widow will end (in her
death) slightly sooner than if the resources devoted to keeping her husband
alive another year were devoted to her instead.
I hope it is clear that I am not advocating a reallocation of medical
resources from women’s to men’s diseases. Whether that would be a good
thing to do from an economic standpoint requires, at a minimum, plugging
numbers into the variables in my equations, something I have not attempted
to do. I merely suggest that it is a matter that deserves to be studied. It
should not be dismissed out of hand, as one might be tempted to do.

Quality and quantity of life. Age is an issue not only in fiscal deci¬
sions concerning health care and medical research, but also in decisions
regarding medical treatment. When medical resources are short, as in the
classic triage situation, and price is not used to clear the market, should age
be a criterion of the decision whom to treat? It frequently is used as a cri¬
terion, to the disadvantage of the elderly. English doctors will not provide
dialysis to “elderly” sufferers (>55) from kidney disease, and American
doctors use age as a criterion for determining admission to the last open
bed in intensive-care units and for eligibility for a heart transplant.^^ The
use of this criterion is often defensible on strictly “medical” grounds—the
elderly patient may be much less likely to survive or otherwise benefit from
the procedure than the younger competitor. But not always. What to do?
If it is a true triage situation, so that the question is not how many
resources shall go to this group or to that group but who shall die, the use
of an age criterion should be relatively uncontroversial. Giving the treat-

33. Kilner, note 3 above, at 77-78. For criticism, see Nancy S. Jecker and Lawrence J.
Schneiderman, “Is Dying Young Worse Than Dying Old?” 70 Gerontologist 66, 70 (1994).
Social Security and Health 279

ment preference to the young will maximize the quantity of life saved, and
even more the quality-weighted quantity, since the utility of life tends to
decrease with age. Dread of death is on both sides of the balance, so drops
out, making a comparison of utilities appropriate.
But we must be careful not to endorse, uncritically, resource decisions
that create triage situations and by doing so make the use of an age crite¬
rion deceptively reasonable. If the number of intensive-care beds is chosen
so that managers of intensive-care units are often forced to choose between
giving the last empty bed to a young person or to an old person, we should
consider whether the nation has sufficient intensive-care capacity. Reduc¬
ing capacity in order to force the “easy” choice between saving a young
life and an old life is akin to not informing elderly persons about treatment
options.
So the use of age criteria to allocate medical resources is tricky. But
some of the arguments against it are poor. An example is the argument that
it is inconsistent with punishing murderers of elderly people as heavily as
murderers of young people. No social purpose would be served by encour¬
aging the murder of elderly people by punishing such murder more lightly.
The objective of most criminal statutes is to punish the criminal as heavily
as is consistent with maintaining marginal deterrence and economizing on
expenditures on the criminal-justice system.^"* Neither of these constraints
points to a punishment “discount” for murdering the elderly. It is true that
“mercy killing,” mainly of elderly people, is usually punished more lightly
than other murders. Many mercy killings are akin or even equivalent to
assisted suicide, which ought probably to carry a different moral charge
from murder, although the discussion in the preceding chapter cannot be
considered decisive on the question; the discussion was confined to
/7/zy5ic/an-assisted suicide. But even people adamantly opposed to legaliz¬
ing any form of mercy killing do not usually favor punishing the mercy
killer as severely as other deliberate killers. No more should they insist that
age be ignored when a choice has to be made between saving the life of a
young person and saving the life of an old one.
I have touched on the distinction between longevity and quality of life,
or between quantity and quality of life, and I want to examine it a little
more closely. We have seen that the expected-utility perspective has limi¬
tations when the issue is understanding the behavior of old people or evalu-

34. Richard A. Posner, “An Economic Theory of the Criminal Law,” 85 Columbia Law
Review 1193 (1985).
280 Normative Issues

ating normative questions concerning them. But it is very helpful in fram¬

ing the issue of quality versus quantity of life.^^ In a rough but serviceable
way (ignoring complications like discounting) we can say that people want
to maximize the product of quantity times quality of life. So 10 years of
life each of which would confer 100 utiles would yield a lower expected
utility than 8 years of life each of which would confer 150 utiles, and there¬
fore the shorter life expectancy will be preferred. I argued that women
might actually prefer a slightly shorter life expectancy if the consequence
were to increase the utility of their lives when old by making it more likely
that they would have male companionship. Their expected utility might be
greater. This is not the only thing that society should consider in making
decisions on the allocation of medical resources, but it is one thing.
And it has implications for the allocation of medical resources between
research on lethal and on nonlethal diseases, say between research on heart
disease and research on deafness. One thing or rather pair of things that
greatly reduces the utility of elderly life is failing eyesight and hearing,
which particularly in tandem make a person feel cut off from life and
greatly curtail the range of his or her activities. Yet blindness and deafness
have only a slight effect on life expectancy. They reduce the quality rather
than the quantity of life. But once it is recognized that expected utility is a
product of both quality and quantity, it is no longer obvious that the balance
of research on the diseases of the elderly should be heavily skewed, or
skewed at all, in favor of the life-threatening diseases. A 2 percent reduc¬
tion in the prevalence of blindness among elderly people might contribute
more to the expected utility of elderly life than a 2 percent increase in
elderly life expectancy. Advances in cataract surgery must have contributed
greatly to the quality of life of the elderly.
Unfortunately, nothing in the analysis so far supplies a formula for
determining how much of society’s wealth to devote to the health of elderly
people. To the extent that people finance their own consumption in old age,
including consumption of medical resources, out of their own pocket (and
without the help of any tax subsidies—that is, without dipping into the
taxpayer’s pocket indirectly as well as directly), by accepting a lower stan¬
dard of living in their youth and middle age, there is, as I hope I have
shown, very little basis for begrudging them a medically affluent old age.
The difficult question is how large a claim on public resources they should

35. The expected-utility approach to normative issues in health economics is illustrated by

George Tolley et al., “The Use of Health Values in Policy,” in Valuing Health for Policy; An
Economic Approach, note 11 above, at 345. This essay includes an interesting discussion of the
benefits of curing senile dementia. Id. at 368-375.
Social Security and Health 281

have. Multiple-selves analysis—that normative mischief maker—merely

helps to show how difficult the question is. It does not point to a solution.
But we may be able to get some help in answering the question by inquiring
how much the elderly are taking (through tax and expenditure policies, and
in other ways) from the other age groups today and what political checks
exist on their taking more tomorrow. These issues will preoccupy us in the
remainder of the chapter.

Social Security and Redistribution

If for some purposes an individual is a different person when young and
when old, for other purposes he is the same person. Most of us do not want
to kill our future old self or even take steps that will impair its longevity,
not because we have a sense of responsibility to our body’s cotenant but
because we consider that future old person to be ourself. The issue of mul¬
tiple selves arises only in those exceptional cases in which the younger self
decides as it were to abandon the older one. It appears to be largely myth
that before social security “the vast majority of Americans lived only for
the economic moment.” David Fischer, whose assessment I have just
quoted, supports it with the statement that “in 1910, when the first serious
attempt was made to determine the economic status of elderly people in
Massachusetts, the findings were grim. Nearly one out of four [of persons
65 or older] were on the dole.” Yet his own figures show that the vast
majority of these “dole” people were receiving Civil War pensions, and the
ratio of the size of those pensions to average wage rates was only slightly
lower than the ratio of social security retirement benefits to average wage
rates today.^* Only 7.6 percent of the people in the survey on which Fischer
relied were receiving charity or poor relief.^^ All the others (apart from a
tiny number in prison) either were classified as “not dependent” or were
receiving federal pensions."*®

36. David Hackett Fischer, Growing Old in America 164 (1977).

37. Id. at 161.
38. Judith Treas, “The Historical Decline in Late-Life Labor Force Participation in the
United States,” in Age, Health, and Employment 158, 164-165 (James E. Birren, Pauline K. Rob¬
inson, and Judy E. Livingston, eds., 1986).
39. Computed from Fischer, note 36 above, at 161 n. 5.
40. For other evidence that long before social security most elderly Americans were ade¬
quately even if not generously provided for, see Carole Haber and Brian Gratton, Old Age and the
Search for Security: An American Social History (1994), esp. ch. 2; Carolyn L. Weaver, On the
Lack of a Political Market for Compulsory Old-Age Insurance prior to the Great Depression;
Insights from Economic Theories of Government,” 20 Explorations in Economic History 294,
302-316(1983).
282 Normative Issues

In the normal case, which is that of people sufficiently regardful of

their future selves to make at least minimally adequate provision for their
old age, it is problematic to speak of a redistribution of wealth from young
to old or the subsidizing of the old by the young. Even so, it might seem
that laws compelling pension contributions and otherwise granting entitle¬
ments to the old might interfere with the preferred pattern of consumption
over the life course, even for people who want to have a long and healthy
old age. For they may not want to allocate as large a share of their lifetime
consumption to their old age as the laws try to make them do. If forced by
law to do so, once they get old they will (unless they have a strong bequest
motive) live high on the hog, as they cannot redistribute wealth to their
younger selves. Yet even when old they may regret that they could not have
consumed a larger fraction of their lifetime income when they were young.
This scenario is implausible, it is true, so long as the compulsory bene¬
fits and the taxes required to defray them are modest in relation to the cov¬
ered individual’s income. For then he can achieve his desired allocation of
consumption over the life course by altering the time profile of his earnings
stream, taking less in the form of retirement benefits and more in current
income, or by borrowing more when he is young, enabling him to consume
more now and less later (when he has to repay his loans). These possibili¬
ties may seem to confine the effects of social security to those young selves
who care nothing about their old selves—who would like to let them
starve. Not so. The program has potentially significant redistributive effects
unrelated to the issue of multiple selves; it redistributes wealth among dif¬
ferent individuals. Two types of redistributive effect should be distin¬
guished: intragenerational, which is to say the effect on the relative in¬
comes of persons born at the same time, and intergenerational—the effect
on the relative incomes of persons born in different generations.'^' Because
social security benefits do not rise with income as fast as the social security
tax does, the social security program redistributes income from more to
less affluent participants of the same age cohort. This effect is reduced but
not eliminated by the fact that income and longevity are positively corre¬
lated, so that the more affluent tend to collect social security benefits for a
longer period."^
The more dramatic redistributive effect, however, is intergenerational.
The creation of the social security program in 1935 conferred a windfall on

41. For a careful analysis and a review of the literature, see Nancy Wolff, Income Redistri¬
bution and the Social Security Program (1987).
42. Id. at 124-125.
Social Security and Health 283

those persons who were eligible for immediate retirement benefits under it,
since they had paid no social security tax. (A similar windfall was con¬
ferred on the elderly thirty years later by the creation of the Medicare pro¬
gram.) Their social security benefits were a straight transfer from persons
of working age who were paying the tax, though of course some of these
taxpayers were sufficiently near retirement age to be able to anticipate a net
benefit from social security, paid for by still younger taxpayers. As time
passed, lengthening the period during which social security recipients had
been employed and paid taxes, the intergenerational transfer diminished,
and it has been estimated that persons born in 1960 or later have (unless
they are very poor) negative expected benefits from social security.'^^
The redistributive effects of social security are the consequence of its
“pay as you go” character, and so would be eliminated if, as in a defined-
contribution plan, a person was entitled to social security benefits equal to
the annuity that he could purchase with his social security tax payments,
no more, no less. A system in which the taxpayer supports the retiree rather
than the retiree supporting himself out of his own deferral of consumption
invites each generation of old people to use their concentrated political
might to plunder the young. Such a system is also highly sensitive to shifts
in the ratio of the nonworking to the working population and therefore
to birth and immigration rates, retirement ages, the female labor-force
participation rate, the unemployment rate, and the longevity of old
people—all factors that are both difficult to predict and difficult to influ¬
ence by changes in social security policy. The rise in the dependency ratio
expected early in the twenty-first century (see chapter 2) may require a
steep increase in social security taxes in order to maintain the value of
social security benefits. Consider a simple model of a pay-as-you-go system
in which social security retirement income (r) is generated by a tax at a
constant rate t on the wages (w) of workers, where W is the number of
workers and R the number of retired persons. Then

rR = twW, (11-5)

or equivalently,

t = (R/W)(r/w). (11.6)

43. Michael J. Boskin et al., “Social Security: A Financial Appraisal across and within Gen¬
erations,” 40 National Tax Journal 19, 23 (1987). On “generational accounting” generally, see
Alan J. Auerbach, Jagadeesh Gokhale, and Laurence J. Kotlikoff, “Generational Accounting: A
Meaningful Way to Evaluate Fiscal Policy,” Journal of Economic Perspectives, Winter 1994,
p. 73; and for criticism, see Robert Haveman, “Should Generational Accounts Replace Public
Budgets and Deficits?” in id. at 95.
284 Normative Issues

If R/W, the dependency ratio, rises by 20 percent, the social security tax
rate would have to rise by 20 percent too in order to maintain the existing
ratio of retirement to working income.
The effect of an increase in the dependency ratio on the total cost of a
retirement system that, like ours, has a medical-care component (Medicare)
is magnified by the increased medical expense associated with increased
longevity. Annual days of hospital care per person rise from 2.1 for persons
aged 65 to 74 to 4.0 for those 75 and above.'” An increase in longevity thus
increases not only the number of years in which a retired person receives
benefits but also the annual level of those benefits, when health benefits are
included in the retirement package along with a pension. It is no surprise
that several foreign countries, including Chile and Singapore, are moving
toward a system in which each age cohort finances its own retirement.'*^
Yet in tension with his recent criticism of the pay-as-you-go system,"**
Gary Becker has argued that the system can be viewed as a fair contract
between young and old. Adults of working age pay taxes to support the
public school system, in effect lending the young money with which to
purchase human capital. The young repay the loan when they become
adults of working age and the generation that paid for their public school
education reaches retirement age, by defraying, through social security
taxes, some of the living expenses and medical expenses of that genera¬
tion."*^ This is an aspect of the earlier point that since our old selves have
no vote until we become old, the current old act as their proxies. The point
is overlooked in the standard analyses of the redistributive effects of social
security because those analyses consider only social security taxes, not
school taxes, in figuring the extent to which recipients of social security
benefits paid for them when employed.
Another argument for pay as you go is that, contrary to appearances,

44. U.S. Bureau of the Census, note 15 above, at 125 (sen 185). The figures are for 1991.
45. Gary S. Becker and Isaac Ehrlich, “Social Security: Foreign Lessons,” Wall Street Jour¬
nal (midwest ed.), March 30, 1994, p. A18.
46. See id.
47. Gary S. Becker, A Treatise on the Family 369-374 (enlarged ed. 1991). The principle
of our pay-as-you-go public pension scheme must be distinguished from its actual design, in which
practical and political considerations have combined to produce many anomalies. See, for ex¬
ample, Jonathan Barry Forman, “Promoting Fairness in the Social Security Retirement Program:
Partial Integration and a Credit for Dual-Earner Couples,” 45 Tax Lawyer 915, 926-948 (1992).
And Becker does not explain the political mechanism by which a fair contract is struck between
different generations; school taxes, on the one hand, and social security taxes and benefits, on the
other, are set by different branches of government—by state and local legislatures, and by Con¬
gress, respectively. It is therefore unclear how a fair contract of the sort suggested by Becker might
have emerged from the political process.
Social Security and Health 285

the social security system is conferring benefits on the current generation

of working age, as well as on the recipients of social security. In the ab¬
sence of social security, the burden of maintaining the old would fall to a
great extent on the younger members of their families. Much of the politi¬
cal momentum for social security came from the added burden on the
young and middle-aged when the Depression threw a number of elderly
people out of work and many others lost their savings.The problem
would be less acute today because private pensions are a larger source of
wealth. Yet millions of elderly people still depend on their social security
retirement benefits for a decent standard of living, and many of them could
not or would not save enough for their old age in the absence of social
security.
Even with social security, a considerable burden of elder care falls on
adult children (especially women).This may be due in part to the fact that
Medicare does not defray the cost of the nonmedical assistance required by
many elderly people, and in part to the failure of the private insurance mar¬
ket to offer nursing-home insurance or other long-term elder-care insurance
on attractive terms.^° That failure may be due to radical uncertainty about
the future costs of elder care, which are a function of longevity as well as
of changes in the price of such care. But a more interesting suggestion is
that the demand for such insurance is weak because it primarily protects
bequests, since elderly people can, by spending down their assets, qualify
for free nursing-home care from the state.^' A related possibility, suggested
by multiple-selves analysis, is that people may decline to buy such insur¬
ance because they do not want to transfer income to their future old, sick,
quite possibly demented self. This would be consistent with wanting to
protect one’s consumption in one’s healthy old age, since long-term care is
for the unhealthy. It is not a complete answer. People who are not particu-

48. Brian Gratton, “The Creation of Retirement: Families, Individuals, and the Social Se¬
curity Movement,” in Societal Impact on Aging: Historical Perspectives 45, 61-68 (K. Warner
Schaie and W. Andrew Achenbaum, eds., 1993).
49. Roxanne Jamshidi et al., “Aging in America; Limits to Life Span and Elderly Care
Options,” 2 Population Research any Policy Review 169, 174-176(1992); Barbara Adolf, “How
to Minimize Disruption Caused by Employees Taking Care of Elderly Relative,” 3 Journal of
Compensation and Benefits 291 (1988).
50. See Jane G. Gavelle and Jack Taylor, “Financing Long-Term Care for the Elderly,” 42
National Tax Journal 219 (1989); Joseph P. Newhouse, “Comment on Predicting Nursing Home
Utilization among the High-Risk Elderly,” in Issues in the Economics of Aging 200, 202-203
(David A. Wise, ed., 1990).
51. Mark V. Pauly, “The Rational Nonpurchase of Long-Term-Care Insurance,” 98 Journal
of Political Economy 153 (1990).
286 Normative Issues

larly altruistic toward their old selves may be altruistic toward their chil¬
dren and may fear that the latter will incur expenses to care for them. These
people may not be worried about protecting bequests—they may not an¬
ticipate having any money to leave their children—yet may not wish to
impose a kind of “negative bequest” in the form of a burden of support on
their children.
Of course to speak of the “burden” of obligations assumed volun¬
tarily, however reluctantly, is to depart from the usual conventions of eco¬
nomic analysis. The vestigial laws mentioned in chapter 9 requiring people
to support their destitute parents are not a significant factor compelling
Americans to take care of their helpless or impoverished parents, so if they
do so they must “want” to. But while many adult children are sufficiently
altruistic toward their aged parents to be willing to incur substantial costs
in money, time, irritation, distress, and even revulsion rather than neglect
or abandon them, they would greatly prefer to shift the burden of caring for
their parents, or at least a part of the burden, onto other shoulders. Elder
care thus differs from types of consumption from which people derive plea¬
sure. There is plenty of evidence that a sense of filial obligation, like other
moral sentiments, can be a source of cost to the person feeling it.^^ From
the standpoint of distributive equity, therefore, social security can be de¬
fended (how strongly I shall not consider) as spreading at least the mone¬
tary component of the burden of elder care among the entire current
generation. The greatest benefit is to people in small families; an only child
faces a greater burden of supporting his parents in their old age than a child
who has many siblings with whom to share it. But as it is large families
that derive the greater benefit from school taxes because they have more
children to educate and school taxes do not vary with the number of the
taxpayer’s children,” it is at least rough justice that small families should
derive the greater benefit from social security.
If we think of the “voluntary” care of the elderly by their families as a

52. See, for example, Nancy J. Finley, M. Diane Roberts, and Benjamin F. Banahan, III,
“Motivators and Inhibitors of Attitudes of Filial Obligation toward Aging Parents,” 28 Gerontolo¬
gist 73 (1988). Even in Japan, famous for filial piety (see chapter 9), the fall in the size of families,
which increases the burden on the younger members of caring for the elderly members, has
contributed to a large expansion in public provision of services to elderly persons. Daisaku Maeda,
“Family Care of Impaired Elderly in Japan,” in Aging: The Universal Human Experience 493,
497 (George L. Maddox and E. W. Busse, eds., 1987).
53. At least if we ignore the taxes those children will pay when they grow up and go to work.
David Friedman, “Laissez-Faire in Population; The Least Bad Solution” 13 (Population Council
1972).
Social Security and Health 287

form of “tax” that transfers wealth from young to old, it becomes relevant
to note that the tax is growing. It is growing both because the ratio of el¬
derly people who require care to the number of young and middle-aged
people is growing and because the market opportunity costs of women’s
time are growing.
Even if the intergenerational-contract aspect of social security is set to
one side, the size of the transfer from young to old that is brought about by
government programs such as social security is difficult to estimate. Fed¬
eral expenditures benefiting the elderly amounted to $387 billion in 1991,^"^
but this is not a net figure. Some fraction of the social security benefits
received by the elderly merely offset the contributions they made when
they were working and paying social security tax, and can thus be viewed
as an annuity for which they paid, rather than a transfer to them. The net
transfer varies enormously across persons, depending on income, family
situation, the length of time that the person worked, and (since both social
security taxes and social security retirement benefits are constantly chang¬
ing) the person’s age cohort.^^ In addition, the elderly pay many billions of
dollars in income and other taxes.^*^ Although they also receive the benefit
of the governmental services that those taxes defray, if they receive fewer
such services than other age groups (as is plainly the case with school
taxes) their taxes would be offsetting some of the social security benefits
that they receive. A further complication is that by encouraging people to
retire earlier than they otherwise would, social security reduces the amount
of income tax paid by elderly people. There is also the complication intro¬
duced by the growing burden of the family elder-care “tax.”
The unknown adjustment required by the intergenerational-contract
aspect of social security and the adjustment required by its annuity aspect
are not additive. If the intergenerational-contract approach is sound, the

54. U.S. Senate Special Committee on Aging et al., note 4 above, at 239 (tab. 8-1).
55. It is estimated that for the cohort that retired in the early 1970s the transfer compo¬
nent in their social security retirement benefits ranged from about 77 to 89 percent. Wolff,
note 41 above, at 43; Richard V. Burkhauser and Jennifer L. Warlock, “Disentangling the An¬
nuity from the Redistribution Aspects of Social Security in the United States,” 27 Review of
Income and Wealth 401, 407 (198 L). For the cohort retiring in the middle 1980s it has been esti¬
mated that the transfer component fell to about 67 percent. Michael J. Boskin et al., note 43
above, at 20.
56. Barry Windheim and Charles Crossed, “Salaries and Wages Reported on Income Tax
Returns, by Marital Status and Age, 1983,” Statistics of Income Bulletin, Winter 1987-1988,
pp. 65, 73-75 (tabs. 6-9); Sheldon Danziger et al, “Income Transfers and the Economic Status
of the Elderly,” in Economic Transfers in the United States 239, 256 (Marilyn Moon, ed., 1984).
I have not found more recent data.
288 Normative Issues

social security taxes paid by workers are not contributions to their own
pensions but reimbursements of the cost borne by their parents’ generation
in educating them.

Old Age and Democratic Theory

So we do not know how large the net transfer from young to old is, but it
probably is positive rather than negative and it may be very large.^^ There
would be little reason for concern if it reflected merely a spontaneous out¬
pouring of altruism or affection by the working part of the population for
the old. But this is implausible, for reasons explored in chapter 9. There is
altruism toward people afflicted by conditions such as poverty and ill health
which may be correlated (nowadays mostly the latter) with old age, but the
natural response would be to try to alleviate those conditions rather than to
scatter bounty over the old as a class. The present scale of transfers to
elderly, whether just or unjust as a matter of moral or economic theory,^*
would be incomprehensible were it not for their political power. That power
is an accident of the fact that, as explained in chapter 6, the low cost of
time of the elderly, the homogeneity of their interests, and the disfranchise¬
ment of children makes the elderly a highly effective political pressure
group on behalf of policies that favor them.
Nothing in political or moral theory implies that a group which votes
more heavily than its percentage of the population because it has low costs
of time, and votes disproportionately effectively because of the focus and
homogeneity of its interests, is morally entitled to the political power that
these attributes confer. To the extent that the old can be trusted to vote
disinterestedly, their disproportionate power would not be very troubling.
But they cannot be trusted so, because they have a strong interest in policies
that redistribute wealth toward them. This interest is not much mitigated by
the altruistic feelings of old people toward their children and grandchildren.
The altruistic elderly can offset public transfers to them by making gifts or
bequests if they want, and this course is preferable to them to losing the
transfers, because it maximizes the power of the old. Although elderly
people may be more disinterested voters than younger ones because the
expected gains and losses to them from particular policies are truncated by

57. See Danziger et al., note 56 above, at 264.

58. Becker, note 47 above, at 370-373, presents some evidence that the elderly are not being
over-repaid for the “loan” they made to the young to finance the latter’s education. Nor is it certain
that the elderly are receiving more than our future elderly selves would legitimately demand if they
had a voice in current policy deliberations.
Social Security and Health 289

the prospect of death, this is probably not a very important factor either.
Most elderly people have a significant remaining life expectancy, and those
near death are unlikely to vote.
I do not mean to suggest that age is the only factor that influences a
vote by an older person; for example, educated old people may vote for
public school bond issues because they value education, even though they
do not have children of school age.^^ There are nevertheless grounds for
concern that the elderly have disproportionate voting power, especially
in contrast to the disenfranchised at the other end of the age con¬
tinuum—children. A Utopian solution would be to give each child a vote,
half of which would be cast by each parent. It would not be an unworkable
or even a particularly cumbersome solution. The danger of fraud would
presumably be no greater than in the case of the dependent deduction from
income tax, and it would probably be less since the gains from fraud would
be smaller. This child-weighted voting scheme would go some way toward
redressing a political balance that appears to be skewed in favor of the
interests of those who are nearing the end of their life and against the inter¬
ests of those who have the greatest life expectancy.
I said “appears” to be skewed; for the danger of excessive redistribu¬
tion to the old as a consequence of their disproportionate voting power may
be overstated, quite apart from the fact that some elderly people “undo”
public wealth transfers to them by gifts and bequests to their children. The
pressure for redistributing wealth from young (or middle-aged) to old is
self-limiting. As the percentage of old people increases, causing the depen¬
dency ratio to rise, the burden on each young earner of redistribution to the
old rises even if the average benefits received by the old do not. The in¬
creased individual burden should stiffen the young’s opposition to any fur¬
ther increases in benefits to the old. Even if the old are a better organized
and more effective political pressure group than persons of working age, at
some point the cost to the latter of redistributing wealth to the former will
reach a level at which organization for effective political resistance be¬
comes feasible. That level will be reached sooner, the faster the dependency
ratio rises. There is thus a natural, though not necessarily an optimal, po¬
litical equilibrium in the struggle between young and old selves.
This point is somewhat blunted, however, by the phenomenon of
“sunk” social security taxes.^^^ Even if a person has not yet retired, and

59. For evidence, see James W. Button and Walter A. Rosenbaum, “Seeing Gray; School
Bond Issues and the Aging in Florida,” 11 Research on Aging 158 (1989).
60. Boskin et al, note 43 above, at 30-31.
290 Normative Issues

considers the social security program excessively favorable to the elderly,

in evaluating the expected benefits to himself from the program he will
ignore the social security taxes that he has already paid; they are sunk costs,
unrecoverable. If he is nearing retirement, most of the expense of social
security to him will lie in the past and all the benefits in the future, so it
may well be in his interest to resist any curtailment of benefits rather than
to support such curtailment coupled with lower taxes. There is a natural
alliance between the elderly and the almost-elderly, magnifying the politi¬
cal power of the former beyond even what is implied by their formidable
numbers and voter turnout.
The effect of the self-interest of the young in checking the size of gov¬
ernmental wealth transfers to the old might be expected to be especially
great in a society such as ours in which old people are not revered. Yet this
point seems hard to square with the fact that Japan, in which old people are
still revered (if less so than formerly), has a much higher savings rate than
the United States. One might expect Japanese to save less, confident that in
their old age they could “free ride” on their children.^' But maybe I have
the causation backwards. Maybe the high social status of elderly Japanese
results from their having saved heavily during their working years, thus
increasing their wealth in old age and becoming respected because of that
wealth, not because of their age.
A second reason not to fear excessive redistribution to the old too
much is that it can be checked at less political cost than many other redis¬
tributions extracted by interest groups. Because the approximate number
of old people can be predicted some years in advance (although the rapidity
of medical progress is making such predictions increasingly uncertain) and
because relatively few old people will be around for more than another
decade or two,**^ it is possible to exploit the discounting of future costs to
present value by passing laws now that will limit transfers to the old in the
future. It is also possible to exploit the inertia of the political process (laws
are difficult to repeal even if the political balance has shifted against them)
in order to provide some security against the repeal of such laws when the
future becomes the present. Consider the law enacted in 1983 raising the

61. Edward P. Lazear, “Some Thoughts on Savings,” in Studies in the Economics of Aging
143, 162 (David A. Wise, ed., 1994).
62. In a sense they are “transients,” whom we expect to have less political weight than
“permanent” members of the polity. But of course the contrast is exaggerated. No one lives for¬
ever; and the nearly old will tend to identify with the old, thus increasing the number of (function¬
ally) old and their staying power.
Social Security and Health 291

age of entitlement for full social security retirement benefits from 65 to 67

in the year 2023“ (and also increasing the discount for claiming social
security benefits at age 62 from 20 percent to 30 percent). The law cannot
hurt any current recipients of social security retirement benefits, because
all these people have qualified already. The people hurt are those who are
now working who will have to work longer before they receive benefits.
They incur a loss measured by the present value of the future lost bene¬
fits—a modest amount because of discounting to present value.“ And this
small loss is offset by the gain from not having to pay the higher social
security taxes that would be necessary to fund larger benefits in the future,
or more precisely from the part of the gain that will be received by the
children and other family members of the current old people. Not too much
should be made of that gain, however. There is always the possibility that
in the future as in the present the elderly will have sufficient political power
to make the young pay for higher social security retirement benefits, and
the further possibility that raising the retirement age will result in more
applications for social security disability benefits,“ a program that in effect
accelerates the payment of social security retirement benefits to any worker
who can establish that he is totally disabled.
When the gains^ and losses that a proposed policy will visit on politi¬
cally powerful groups are small (here because of discounting to present
value), considerations of the public interest, which is to say of the costs and
benefits of policies to diffuse groups such as consumers or taxpayers, are
apt to play a larger, and often a decisive, role in the shaping of policy. It
would be difficult otherwise to understand how laws punishing crime get
passed and enforced, since the beneficiaries (besides lawyers and law en¬
forcers) are a diffuse group—the entire law-abiding public.
If increases in longevity are accompanied by improvements in the
health of the aged, as they appear to be, one might think that raising the age
of retirement would be an automatic and costless response to the fact that

63. Actually the age is to be increased by a month every year beginning in 2000.
64. At a discount rate of 3 percent, the present value of $1 to be received in 40 years
(2023 - 1983) is 31 cents. It is not clear that the benefits to future generations from more efficient
policies should be discounted similarly—or at all. The social discount rate may be zero, although
I doubt it (see next chapter). But if people’s political behavior is assumed to be guided mainly by
private rather than social benefits and costs, the question of the correct social discount rate is
distinct from the political feasibility of legislation that reduces the entitlements of future
generations.
65. Thomas N. Chirikos and Gilbert Nestel, “Occupational Differences in the Ability of
Men to Delay Retirement,” 26 Journal of Human Resources 1 (1991).
292 Normative Issues

the age of onset of old age was rising. I think this is largely true (see
chapter 2), although it may seem to overlook the factor of boredom, em¬
phasized in previous chapters. If after some point the disutility of work
increases as a function of time worked rather than of age, lengthening
people’s working life will impose disutility on them even if, because of
better health, their capabilities for work and their wages do not decline
before the new, higher age of retirement. But this assumes that the only
effect of increased healthy longevity would be to lengthen people’s careers.
There would be other effects. Graduate education might become more at¬
tractive, because it postpones entry into the workforce. Careers such as the
military that are geared to facilitating the worker’s taking up a second ca¬
reer at the end of his period of service would also become more attractive.
Most important, career-switching would be facilitated because the in¬
creased length of one’s remaining working life would increase the return to
mid-life investments in acquiring new human capital.
As is well known, social security benefits are overindexed for inflation.
If this feature were corrected, benefits would (without new legislation)
grow more slowly than average wages, because the benefits are not indexed
for real (that is, inflation-adjusted) wage increases.
Still another consideration that argues against becoming too worried
about the voting power of the elderly is the contribution which that power
makes to political stability. Recall the point from chapter 8 that elderly
voters tend to be conservative in the sense of resistant to new fashions in
public policy. Resistance to new ideas, a characteristic by-product of the
age-related decline in fluid intelligence, is no virtue in scientific fields,
where objective procedures exist for winnowing good ideas from bad.
There are no such procedures in politics (or at least they operate very
slowly), and as a result bad new ideas—as fascism and communism once
were—can quickly become deeply entrenched. A large elderly voting
population, by retarding the adoption of new political ideas, can reduce the
inherent riskiness of a political system. Another and in politics an apt term
for being receptive to new ideas is “impressionable.” We might not want
the electorate to be dominated by impressionable voters, and we might con¬
sider the transfers that the elderly receive from the young as, in part, com¬
pensation for the contribution that elderly voters make to political stability,
from which the young benefit as well.
An offsetting consideration, however, is that elderly voters may be par¬
ticularly susceptible to radical ideas when packaged as revivals of old
ideas—which they may be. Elderly voters, it appears, disproportionately
Social Security and Health 293

supported Hitler in the elections that paved the way for his ascension to
power.^®
A final argument against the idea that the elderly have a dangerous
excess of voting power brings us back once again to the concept of multiple
selves. Every young and middle-aged person has a future self of which he
may not be—probably is not—a completely trustworthy fiduciary, just as
parents are not entirely trustworthy fiduciaries of their children because
their interests are not identical. As a practical matter, the current elderly
represent not only their own interests but also those of the future elderly
selves of people who are still young. The argument that I made earlier for
giving parents extra votes in order to increase the weight of children’s in¬
terests in the political process can be made for tolerating tlfe “excess” vot¬
ing power of the elderly as well, in order to assure that the process will give
adequate weight to the interests of future elderly selves.
Balancing all the comforting facts marshaled in this section, however,
are two counterweights. One has to do with truncated horizons; the other
with free riding. I said that the current old can be thought of as proxies for
the future old (the future selves of the current young and middle-aged). But
this is not likely to work for projects whose payoffs are long deferred, for
example research on a cure for cancer not expected to be found until the
year 2020. For by then the current old will be dead. The persons who are
most interested in such a project are the future old people—who are the
current young.
Second, while it appears to be politically feasible both to curtail Medi¬
care and social security infuturo, and (so far) to refuse public subsidization
of home care for elderly people who need assistance to cope with the chal¬
lenges of daily living but not medical assistance, it is not politically feasible
either to deny the elderly medical care (as we saw in discussing Dworkin’s
proposal) or to abandon them should they need home care. The burden of
the former will fall on Medicaid when Medicare and private insurance runs
out and the burden of the latter on family, friends, and social workers.*^^
These will be costly burdens for which society (viewed as the aggregate of
its members) will pay in one way or another unless people are forced to
save more for their old age. The burdens will grow in real terms as the
fraction of elderly people grows and as more and more elderly people live

66. See Richard F. Hamilton, Who Voted for Hitler? 61-62, 512 n. 46 (1982).
67. I emphasize home care rather than institutional care because the former is greatly pre¬
ferred by elderly persons for reasons explained in chapter 6.
294 Normative Issues

to the really advanced ages at which heavy medical and home-care costs
are unavoidable, even if—what at the moment seems unlikely—the rela¬
tive cost of medical and home care does not continue to grow. It is not yet
apparent what public or private means exist for preventing the elderly from
shifting these burdens to taxpayers and family members. They cannot be
counted on to make adequate provision through private savings, because of
the conflict between young and old selves that I have stressed throughout
the book.
On this view—which incidentally provides another reason for real¬
locating expenditures on medical research from fatal to merely disabling
physical and mental conditions of the elderly—curtailing social security.
Medicare, Medicaid, or other public benefits is not much more than chang¬
ing the bookkeeping, or more precisely changing the identity and relative
shares of taxpayers and other bearers of the costs of the elderly. This is
an exaggeration, of course. As I pointed out in discussing Medicare, a
change in the structure of public programs can create incentives to econo¬
mize. The elderly would, I argued, be better off if they could commute
the expected value of their Medicare entitlement to cash to spend on
anything they liked; and the result would be a reduction in the nation’s
overall bill for health. And private medical charity would no doubt be more
skimpy than public. Yet even the abolition of all public programs for the
support of the elderly might not result in a dramatic reduction in the
medical and other consumption costs of the growing phalanx of elderly
Americans.
But as always we must be careful to maintain perspective, or, stated
differently, to reckon in all benefits as well as all costs. Middle-aged people
who incur heavy expenses for the home care of their aged parents will (they
or their future selves!) some day be old themselves and want such care.
And those heavy expenses are in part the unavoidable by-product of ad¬
vances in medical and scientific knowledge that are extending healthy
middle age, postponing old age, and by this combination generating enor¬
mous utility to which the burden of elder care is probably a relatively minor
offset. A greater one may be the emotional toll that unhealthy old age takes
of the elderly and their families.

The focus of this chapter has been on intergenerational wealth transfers,

nonpecuniary as well as pecuniary—and conjectured as well as proven.
The social security retirement program. Medicare, other special public
programs for the aged that 1 have not even discussed, publicly financed
medical research on diseases that disproportionately afflict elderly people.
Social Security and Health 295

and treatment decisions that avoid the use of age as a criterion for allocat¬
ing scarce medical resources bring about in the aggregate a large gross
redistribution of wealth from young to old; but the net redistribution is
another matter. The net intergenerational redistributive component of the
largest program of public transfers to the old—the social security retire¬
ment program—may be small, or conceivably if improbably even negative,
when the annuity value of the social security taxes paid by the recipients of
social security retirement benefits during their working lives is subtracted
from the benefits; when due regard is given to the fact that the generation
that is receiving these benefits paid in the form of school taxes for a large
part of the taxpaying generation’s human capital that in turn generates the
tax revenues that defray the cost of social security retirement benefits; and
when the effect of social security in reducing the caretaking burden on the
children of elderly parents and grandparents is taken into account as well.
Some of the arbitrary-seeming features of our public old-age programs,
moreover, appear to somewhat better advantage in a careful analysis. The
compulsory feature of the social security program, for example, can be
defended by reference to the concept of multiple selves. Not through short¬
sightedness, but through sheer indifference, the young self may abandon
his future old self. Compelled savings for retirement, like refusals to en¬
force suicide pacts and efforts to discourage harmful addictions, may re¬
flect society’s unwillingness to treat the current self as the “owner,” for all
purposes, of the body of which he is the temporary tenant. Multiple-selves
analysis also undermines the suggestion that Medicare is overly generous
because most young people would if given a choice be unwilling to make
“lavish” provision for their medical needs in old age. This is just another
example of the propensity of the young self to weight his own utility much
more heavily than that of his future old self. On strictly fiscal grounds, it is
even possible to defend the heavy allocation of medical resources to the
diseases of the old, because saving young people for old age may well add
more to net expected lifetime costs of health care than moderately prolong¬
ing the lives of people who are already old would do. On broader grounds
of utility, however, it would undoubtedly be better to concentrate resources
more heavily on the diseases of the young.
I argued that the allocation of public resources in the medical sphere
should probably be shifted somewhat in favor of mental illnesses, because
here a cure or substantial improvement is less likely merely to “save” the
patient for another disease. Most controversially, I stressed the possible
benefits of a shift in the allocation of public resources for medical research
in favor of men’s diseases over women’s. The enormous and growing im-
296 Normative Issues

balance between the number of elderly men and the number of elderly
women—“in favor of” the women—so reduces the marital and sexual op¬
portunities of elderly women that it is plausible (though no stronger term can
be used, in light of the qualifications that I suggested) that, ex ante, women
would be made better olf by a reallocation of expenditures to medical re¬
search designed to increase the longevity of men faster than that of women,
at least for a time. The general point which this suggestion illustrates is that
quality as well as quantity of life must be taken into account in determining
what allocation of medical resources will maximize expected utility.
There are plenty of objections to details, some of them very big details,
of the nation’s current old-age programs;®® and it would be reckless to sug¬
gest that the programs approach optimality or to deny the possibility that
they involve a large and perhaps unconscionable transfer of wealth from
young to old. The dramatic imbalance in voting power between young and
old—with children having no votes at all, and the elderly voting dispropor¬
tionately even to their share of the adult population—fuels such a concern.
But subtle factors that limit the dangers created by elderly voting power are
easily overlooked. Among other things: The “conservatism” in the sense
not of a political slant but of resistance to novelty that is characteristic of
the elderly voter is a force ordinarily though not invariably for healthy po¬
litical stability. The larger the transfer of wealth from young to old, the
easier it is to organize the young in effective political opposition to further
such transfers. The ability to forecast (more or less)®^ the number of elderly
at specific periods in the future, in conjunction with legislative inertia,
which makes it difficult to repeal legislation once it is adopted, facilitates
legislative curtailment of future benefits for the elderly; the costs of such
legislation will be borne by a future generation of elderly, while the benefits
in lower taxes will begin to accrue immediately. Even the “excessive” vot¬
ing power of the old may be justified, by the role of the current old as
proxies for our otherwise unrepresented future elderly selves.
I admitted that the analysis might be too static. The burdens of medical
care and nonmedical home care will grow as the population of elderly
grows, and quite possibly faster.’^ Someone will have to pay, since it is

68. Raymond G. Batina, “On the Time Consistency of the Government’s Social Security
Benefit Policy,” 29 Journal of Monetary Economics 475 (1992).
69. The less is emphasized in Kenneth G. Manton, Eric Stallard, and Burton H. Singer,
“Methods for Projecting the Future Size and Health Status of the U.S. Elderly Population,” in
Studies in the Economics of Aging, note 61 above, at 41.
70. Recall, however, the caveat in chapter 2 against exaggerating the impact of rising costs
of health care on the aggregate social costs of the elderly population.
Social Security and Health 297

highly unlikely that the elderly will actually be denied the care they need.
That “someone” may not be the elderly themselves when they are young,
given the conflict between young and old selves. There is a real policy
dilemma here, and no feasible solution handy, although a greater emphasis
on quality-improving as distinct from longevity-enhancing medical re¬
search offers the possibility of at least a partial solution. Even disregarding
that possibility, I suggested that the heavy costs of elder care are offset by
the benefits, to those who bear those costs, of modern medicine and science
in postponing (while elongating) old age. Healthy middle age has been
extended; one cost is the burden that the middle aged must shoulder for the
expenses of the old, for old age has inevitably been extended as well.
 12
Legal Issues of Aging and Old Age:
A Sampler

Although the Social Security Act is the most important law relating to
old age, the actual administration of the Act has generated relatively
few issues of application and interpretation whose intelligent resolution
requires the sort of understanding of aging and old age that I have tried
to develop in this book. Yet there are many such issues under other laws.
The most interesting concern the law’s effort to eliminate discrimination
on grounds of age in employment. To that subject I devote the next
chapter, while in this one I discuss several other subjects involving legal
regulation that bears on the elderly: the federal law governing pensions
(other than those provided through the social security system); the tort
rules applicable to elderly injurers and victims; the punishment of elderly
offenders and the continued imprisonment of elderly people who were con¬
victed when young of crimes for which thesentence was life in prison with¬
out possibility of parole; and the bearing of senile dementia on legal capac¬
ity and responsibility. The laws relating to assisted suicide, and to living
wills and related methods of regulating MDELs (medical decisions at end
of life), were discussed in chapter 10.
The sample of legal issues relating to the elderly that I discuss in this
chapter is a small one. Other interesting issues include the shielding of the
assets of elderly people in bankruptcy, the legal rights of grandparents, the
obtaining of informed consent to medical experimentation on demented
patients, and the rights of persons confined to nursing homes against their

298
Legal Issues: A Sampler 299

will yet without either a formal commitment order or the employment of

physical force to confine them.'

Pension Law
In 1974 Congress passed the Employees Retirement Income Security Act
(ERISA).2 This complex statute is a major source of litigation in the federal
courts, although much of that litigation concerns issues unrelated to the
subject of this book. For, despite its title, the statute regulates employer
health and welfare benefits as well as pensions and provides a federal forum
for litigating contractual disputes over the terms of particular pension or
other covered plans. The statute’s principal significance in relation to age
is its impact on the employer’s choice between a defined-benefit and a
defined-contribution pension plan and its regulation of the former type.^
A defined-benefit plan promises to pay the employee a fixed annual
pension based on the employee’s wage in his last year or last few years of
work and on his years of service. The higher his terminal salary and the
longer he has worked for this employer, the larger his pension entitlement
will be. For example, the plan might entitle him to an annuity income equal
to 1 percent of his terminal salary times his number of years of service, so
that if he retired after 30 years with the company he would be entitled to

1. See, regarding these issues, Daniel L. Skoler, “The Elderly and Bankruptcy Relief: Prob¬
lems, Protections, and Realities,” 6 Bankruptcy Developments Journal 121 (1989); Madeline
Marzano-Lesnevich, “Grandparents’ Rights,” New Jersey Lawyer, Jan./Feb. 1991, p. 46; Rob¬
ert L. Schwartz, “Informed Consent to Participation in Medical Research Employing Elderly
Human Subjects,” 1 Journal of Contemporary Health Law and Policy 115 (1985); Carthrael Ra-
zin. Comment, “ ‘Nowhere to Go and Chose to Stay’; Using the Tort of False Imprisonment to
Redress Involuntary Confinement of the Elderly in Nursing Homes and Hospitals,” 137 University
of Pennsylvania Law Review 903 (1989). On the legal problems of elderly people generally, see
Peter J. Strauss, Robert Wolf, and Dana Shilling, Aging and the Law (1990); Joan M. Krauskopf
et al., Elderlaw: Advocacy for the Aging (2d ed. 1993) (2 vols.).
2. 29U.S.C. §§ 1001 etseq. For helpful economic analyses of the pension aspects of ERISA,
see Jeremy I. Bulow, Myron S. Scholes, and Peter Menell, “Economic Implications of ERISA,”
in Financial Aspects of the United States Pension System 37 (Zvi Bodie and John B. Shoven, eds.,
1983); Laurence J. Kotlikoff and David A. Wise, “Pension Backloading, Wage Taxes, and Work
Disincentives,” in Tax Policy and the Economy, vol. 2, p. 161 (Lawrence H. Summers, ed., 1988).
3. On the difference between the two types of plan, see Zvi Bodie, Alan J. Marcus, and
Robert C. Merton, “Defined Benefit versus Defined Contribution Plans: What Are the Real Trade-
Offs?” in Pensions in the US. Economy 139 (Zvi Bodie, John B. Shoven, and David A. Wise,
eds., 1988). Although defined-contribution plans are much more numerous, defined-benefit plans
hold in the aggregate more assets. Pension and Welfare Benefits Administration, Abstract of 1990
Form 5500 Annual Reports,” Private Pension Plan Bulletin, Summer 1993. But see note 10 below
and the text accompanying it.
300 Normative Issues

receive an annual pension equal to 30 percent of his final year’s wage. Thus,
in a defined-benefit plan the employee’s entitlement is independent of the
amount of the employer’s contributions to the plan or of the plan’s invest¬
ment performance.
In a typical defined-contribution plan, contributions are made by both
employer and employee to a separate account for each covered employee."*
When the employee retires, the amount of money in his account, which
will be the sum of his and his employer’s contributions plus any interest or
other return earned on those contributions from investing them, is used to
buy an annuity that will give him and his spouse an assured income for the
rest of their lives. (In some plans, the employee gets a lump sum rather than
an annuity.) So while in a defined-benefit plan the investment risk is borne
by the employer or by the plan (or the plan’s insurer), in a defined-
contribution plan it is borne by the employee, though he should be able to
reduce that risk by investing the contributions in a diversified portfolio of
securities.
The allocation of the risk of inflation in the years after retirement is
reversed in the two types of plan. Benefits in a defined-pension plan are
stated in nominal terms (so many dollars a year until the employee dies),
whereas benefits under a defined-contribution plan are paid to the em¬
ployee upon retirement and can then be invested in vehicles that will give
him at least some protection against inflation. I have not seen an explana¬
tion of why the benefits in defined-benefit plans are not indexed against
inflation.
A defined-benefit plan has the greater effect on the age profile of the
employer’s workforce. Within limits fixed by the federal pension and age-
discrimination laws, employers can adopt benefit formulas that make it
highly disadvantageous for employees to quit before—or fail to quit at—a
particular age, such as 55, 60, or 65. A defined-contribution plan is not as
well adapted to manipulating the term of employment, since the amount of
the employee’s retirement benefits is not within the control of the employer,
although it is influenced of course by the amounts contributed.
Because benefits under a defined-benefit plan are tied to the employ¬
ee’s final salary, such a plan provides greater although not complete as¬
surance of stable (though not necessarily full) replacement of wage

4. In some defined-contribution plans, only the employee contributes. But this is a detail
without economic significance, since, as we shall see in the next chapter, workers pay for benefits
by accepting lower wages, even if the employer is the nominal payor.
Legal Issues: A Sampler 301

income; the employee’s accustomed standard of living will be more or less

preserved in retirement. Benefits under a defined-contribution plan depend
strictly on the amount contributed and on the investment performance of
the contributions.
If the employer is the better risk-bearer, the defined-benefit form will
generally be more attractive to employees than the defined-contribution
form, although differences in vesting rules and the omnipresent risk of infla¬
tion complicate the trade-offs. The defined-benefit form also facilitates op¬
timal investments in human capital, as we shall see, and confers tax advan¬
tages on the employer when the plan is overfunded, that is, when its assets
exceed its pension liabilities. The income on those assets will inure in part
to the employer because it will not all be needed to defray the employer s
pension liabilities, and that income will accumulate tax free because the
plan is a tax-exempt entity. So there is actually a tax incentive to overfund.
What about an incentive to underfund? Before the enactment of
ERISA, pension liabilities in a defined-benefit plan could be and often were
liabilities of the pension plan alone rather than of the employer as well. If
the plan was underfunded because contributions had been inadequate, in¬
vestment performance disappointing, or actuarial projections inaccurate,
the loss fell on the employees; in this way some of the investment risk was
shifted back to them. If the plan was overfunded, the employer’s sharehold¬
ers, being the residual claimants to the plan’s funds after all entitlements of
employees were honored, reaped the gain.
Depending on the rules of vesting and of crediting years of service
adopted by the particular plan, a worker who left before retirement age
might find himself with a pension benefit worth much less than his contri¬
butions and perhaps worth nothing at all. So he had a strong incentive to
remain with the same company until he reached retirement age. This incen¬
tive both reduced the mobility of labor and increased the employer’s power
to expropriate employees’ firm-specific human capital by implicitly threat¬
ening to fire them before their pension rights vested if they insisted on a
salary commensurate with their value to the company. Indeed, quite apart
from human-capital considerations, one could imagine an employer reduc¬
ing the employee’s wage, to a point at which the wage and the pension
benefit together would just exceed the employee’s wage in his next best
job. The year before the employee retired and became eligible for the pen¬
sion, the wage could be zero—or even negative; the employee would pay
to be allowed to work long enough to become entitled to his pension.
Before concluding that pension practices were exploitative before
302 Normative Issues

ERISA, we must remind ourselves that the terms of retirement, including

pension rights, are a matter of negotiation between employer and prospec¬
tive employee, not a unilateral imposition. Even if, as was and is common,
the employer refused to negotiate separately with each employee but of¬
fered terms of employment on a take it or leave it basis, and even if the
employees were not represented by a union (with which the employer could
not lawfully refuse to negotiate), competition among employers would give
prospective employees a choice between different wage-benefit packages.
The packages offered by some employers would emphasize good retire¬
ment or other benefits at the cost of lower wages, while those offered by
other employers would emphasize high wages at the expense of less gen¬
erous or secure retirement or other benefits. Employees would tend to be
sorted to employers according to the individual employee’s preferences re¬
garding risk and the allocation of consumption over the life cycle.
Even incomplete vesting was not a scam, or an impediment to optimal
mobility of labor. By making pension benefits contingent on the employ¬
ee’s remaining with the firm and performing satisfactorily, incomplete
vesting facilitated the recovery by employers of their investment in their
employees’ firm-specific human capital. It also solved the last-period prob¬
lem: not only with the stick (the threat of discharge before pension rights
vested) but also with the carrot, since pension benefits in a defined-benefit
plan are heavily influenced by the employee’s wage in his last years of
employment. The incentive of employers to abuse the power that incom¬
plete vesting conferred on them to renege on their unwritten contract to
deal fairly with their employees could be expected to be held in check by
the employer’s concern with preserving a reputation for fair dealing (if he
lost that, he would have to pay new employees higher wages) and by the
bargaining power that the possession of firm-specific human capital confers
on a worker. If the worker quits in anger or disgust, or is fired to eliminate
his pension benefits, the firm must invest in training a green employee to
replace him.^
These market checks on the exploitation of employees by employers
did not work perfectly; very little does. In particular they could not be ex-

5. Cf. Donald P. Schwab and Herbert H. Heneman III, “Effects of Age and Experience on
Productivity,” 4 Industrial Gerontology 113 (1977). For an excellent review of the modern eco¬
nomic theory of the pension contract, .see Richard A. Ippolito, “The Implicit Pension Contract:
Developments and New Directions,” 22 Journal of Human Resources 441 (1987); for fuller treat¬
ment, illustrating what has become a vast literature, see the essays in Pensions, Labor, and Indi¬
vidual Choice (David A. Wise, ed., 1985). The modern theory views the pension contract as a
device for imparting proper incentives to workers; the older theory saw it merely as a form of
savings stimulated by favorable tax treatment.
Legal Issues: A Sampler 303

pected to work well when an employer was in its last period.*^ But there is
no persuasive evidence that pension abuses were so widespread as to justify
the imposition of a complex scheme of federal regulation. On the contrary,
careful empirical research has shown that before ERISA opportunistic dis¬
charges of workers covered by a pension plan were rare and that ERISA
has had no detectable impact on discharges of covered workers.^
ERISA’s principal thrust, so far as pertains to retirement (recall that
ERISA also regulates medical and other employee-benefit plans), is to re¬
quire that defined-benefit plans vest the employee’s pension rights after he
has been a participant in the plan for five years and to back up this require¬
ment by limiting pension “backloading.” The term refers to the practice of
making pension benefits, even if vested, so age-dependent that, until retire¬
ment age is reached, the value of the employee’s pension right is very small.
Other pension-related objects of the law were to create a federal judicial
remedy for breach of the fiduciary obligations of pension plans to the par¬
ticipants in and beneficiaries of the plans, to reduce underfunding of pen¬
sion plans, and, through the Pension Benefit Guaranty Corporation, an
agency created by the Act, to guarantee pension entitlements against the
consequences of underfunding. These last two objects work at cross¬
purposes, since the existence of guaranteed pension benefits undermines
the market checks on underfunding. Workers and their unions will press
less hard for adequate funding of their pension benefits if they know there
is a government backstop. The analogy to federal deposit insurance, which
increases the incentive of financial institutions to make risky loans, is plain.
Apparently the checking provisions of ERISA are inadequate. There are
fears of an eventual pension crisis to rival the savings and loan debacle of
the 1980s.*
ERISA’s effect on pension backloading has been small; defined-benefit
plans remain heavily backloaded.® The Act has, however, made defined-

6. See, for example, Daniel Fischel and John H. Langbein, “ERISA’s Fundamental Contra¬
diction: The Exclusive Benefit Rule,” 55 University of Chicago Law Review 1105, 1132 (1988).
7. Christopher Cornwell, Stuart Dorsey, and Nasser Mehrzad, “Opportunistic Behavior by
Eirms in Implicit Pension Contracts,” 26 Journal of Human Resources 704 (1991); Richard Ippo-
lito, “A Study of the Regulatory Impact of ERISA,” 31 Journal of Law and Economics 85, 91-
102(1988).
8. See Carolyn L. Weaver, “Government Guarantees of Private Pension Benefits: Current
Problems and Market-Based Solutions” (unpublished, American Enterprise Institute, Aug. 1994,
forthcoming in Public Policy toward Pensions [John B. Shoven and Sylvester J. Schieber, eds.].
Twentieth Century Eund).
9. Laurence J. Kotlikoff and David A. Wise, The Wage Carrot and the Pension Stick: Re¬
tirement Benefits and Labor Force Participation (1989); cf. Bodie, Marcus, and Merton, note 3
above, at 143.
304 Normative Issues

benefit plans more costly for employers (per dollar of benefits) and more
secure for employees. But with costs and benefits both greater, it is difficult
to assess the Act’s net effect on the system of private pensions. It may have
been small. Interest rates have generally been higher since 1974 than before
then, so that (depending on corporate tax rates) the tax benefits of over-
funding have increased. The Act does not forbid overfunding, although it
makes it slightly more difficult for the employer to recapture the over-
funded portion of the pension plan’s assets for its shareholders. The effect
of interest rates in magnifying the tax benefits of overfunding have tended
to offset the Act’s effect in making defined-benefit plans less attractive to
employers. It is true that since the late 1970s the tide has been running in
favor of defined-contribution plans (especially 401(k) plans, which allow
employees to make tax-deferred contributions to retirement accounts).'®
But whether ERISA has been a factor in this trend is unclear.
By limiting incomplete vesting, the Act has tended to reduce the con¬
trol of employers over their older employees, since the pension benefits of
such employees are securely vested. Such a loss of control would be ex¬
pected to lead employers to invest less in the firm-specific human capital
of their employees, and—because employers would have a smaller invest¬
ment in them to protect and the employees would have less incentive to
perform well (not being faced with a substantial loss of pension benefits if
they were fired)—to resort more frequently to an explicit or implicit threat
of discharge in order to maintain discipline. The extent of these effects is
not known.
The biggest objection to ERISA is the absence of persuasive theoreti¬
cal or empirical grounds for thinking that the market in private pension
entitlements was not working adequately before 1974. Most of the abuses
discussed in the legislative history concerned multiemployer pension plans
administered by unions—hardly a representative case of the operation
of unregulated labor markets. Yet if there were no social security law,
so that all pensions were private, but it was desired on multiple-selves or
other grounds to compel young people to put aside something for their
old age, it would be necessary to set minimum pension levels, limit under-
funding, forbid assigning or borrowing against pension entitlements, and
perhaps adopt other regulations as well in order to prevent employers and
employees from striking side deals that would trade meaningful pension

10. John R. Woods, “Pension Coverage among Private Wage and Salary Workers: Prelimi¬
nary Findings from the 1988 Survey of Employee Benefits,” Social Security Bulletin, Oct. 1989,
Legal Issues: A Sampler 305

rights for higher wages and thereby empower the young employee to con¬
demn his future old self to starvation, private charity, or the dole. If social
security benefits are considered adequate provision for our future selves,
it is unclear why we need ERISA. But if social security were replaced
by a law that “simply” required people to save for their old age, some¬
thing like ERISA would be necessary to make the law more than an empty
gesture.
Unless the law did no more than require employees to make a specified
level of contributions to defined-contribution plans. But that is a big “un¬
less,” as it would greatly curtail the use of defined-benefit plans by making
defined-contribution plans compulsory. If defined-benefit plans have any
efficiency justifications (and we have suggested several), this government-
ally compelled substitution would be a source of social costs.

Elderly Tortfeasors and Tort Victims

The cluster of issues to which I turn now, while somewhat esoteric, pro¬
vides good illustrations of the value of economic analysis for the under¬
standing and appraisal of legal rules and practices. The first issue is
whether tort law ought to hold old people (whether as injurers or as vic¬
tims) to a lower standard of care than young people, on the theory that the
old are incapable of coming up to the same standard. Tort law does this
with regard to children and the blind. Holding children and blind people
liable for injuries that the adult and sighted could have avoided at reason¬
able cost but that children and the blind could not avoid because they have
a much higher cost of care would not reduce the number of accidents
caused by these classes; it would merely shift the cost of those accidents
from victim to injurer.” I exaggerate slightly. As we saw in chapter 6 in
connection with the driving behavior of old people, accidents can be
avoided not only by greater care but also by avoidance or curtailment of
the activity that generates accidents as an unintended by-product. The old
can drive less, and will do so if the expected cost of driving includes an
expected cost of liability to pay damages to accident victims. In just the
same way, the blind can go out less and parents can keep their children
under greater restraint.
The difference between old people on the one hand and blind people
and children on the other, and the reason I conjecture why the law has not

11. William M. Landes and Richard A. Posner, The Economic Structure of Tort Law 123-
131 (1987).
306 Normative Issues

fixed a more “realistic” standard of care for the former group as it has for
the latter ones, is that the cost to the old of avoiding inflicting accidental
injuries is lower than that of the child or the blind person. The old driver
can substitute time for the sharper vision and quicker reflexes of a younger
driver. And because most old people are retired and therefore are not com¬
muters, the value of driving to them is diminished, which reduces the cost
to them of avoiding accidents by reducing the amount of driving they do
and so making up for any irremediable deficit in care by a change in ac¬
tivity level. If the standard of tort liability for the old were fixed with ref¬
erence to the average physical capabilities of old people, the incentive of
the old to take feasible steps along both the care and the activity dimen¬
sions to avoid causing accidents would be reduced with no corresponding
social gain.
That the value of driving to the average elderly person is less than that
to the average young person is a good argument for holding the elderly to
the same standard of care as the young, but is not a good argument for
holding them to a higher standard. As I emphasized in chapter 6, tort lia¬
bility as well as concern with self-protection has, together with the low
average value to the elderly of driving, already curtailed the amount of
driving by the elderly; the amount that remains is likely to confer substan¬
tial benefits on them.
Old people are frequent victims of accidents, including fatal accidents.
In 1990, 26,213 Americans 65 years old and older were killed in accidents,
of which almost 30 percent were automobile accidents.'^ Some of these
accidents were tortious, which raises the question how to estimate the value
of an elderly life for purposes of tort compensation. Estimation of the
purely financial loss is no more difficult than in the case of the death of a
young person—in fact easier, when a forecast of earnings is required, be¬
cause the period for which the forecast must be made will be shorter. But
what about the nonpecuniary value of life to a person who does not have
many years to live in any event, and those years perhaps of limited utility
to him because of anticipated poor health? It might seem intuitively obvi¬
ous that the value of life would be much less for the average old person
than for the average young one, but the discussion in chapter 5 of elderly
persons’ “dread of death” should make us skeptical.
William Landes and I have suggested that the proper way to value life
for purposes of assessing tort damages is to determine from studies of seat-

12. U.S. Bureau of the Census, Statistical Abstract of the United States 199393 (113th ed.)
(ser. 128); National Safety Council, Accident Facts 12 (1993 ed.).
Legal Issues: A Sampler 307

belt use and other measures that people take to protect themselves from
injury how much people are willing to pay to avoid the risk of accident that
the defendant created by his tortious conduct, and to divide that amount by
the risk.'^ So if the defendant had created a .000001 risk of death and the
victim would have demanded $1 for bearing such a risk, the amount of
damages that should be awarded if, the risk having materialized, the victim
was killed would be $1 million. This is the correct measure of damages
because it confronts potential injurers with the expected cost of their dan¬
gerous conduct ($1). If they spend less than this on accident avoidance,
they are negligent in both the economic and the closely related legal sense.
Although many studies seek to infer the “value of life” (perhaps mis¬
named) by the method that Landes and I believe could be used to compute
tort damages in death cases,I am not aware of any that investigates the
effect of age on the amount of care demanded of potential injurers. But if
the analysis in chapter 5 is correct, old people should be expected to de¬
mand as much or almost as much care as young ones, despite their trun¬
cated life expectancy, and possibly more; and this implies, by the method
proposed by Landes and me, that they should be entitled to damages for
loss of the nonpecuniary value of life at least as great as young victims of
wrongful death are entitled to. Corroboration for the suggestion that the old
demand as much care from potential injurers as the young do—or even
more—is provided by evidence that persons aged 65 and over are more
likely to be regular seatbelt users than younger people.Admittedly, this
does not prove that the old will pay as much as, let alone more than, the
young for the same reduction in the risk of injury and death. Their cost of
time is lower, though this is a trivial consideration when one is speaking of
fastening a seatbelt; more important, their risk of being injured or killed in
an automobile accident is much greater. If the risk were twice as great, we
would expect them to demand the same level of care as young people even
if the utility to them of their remaining life were only half as great as that
of the young.
Nevertheless it is apparent that elderly people do not consider their

13. Landes and Posner, note 11 above, at 187-189; see also Erin Ann O’Hara, “Hedonic
Damages for Wrongful Death; Are Tortfeasors Getting Away with Murder?” 78 Georgetown Law
yoMt-na/1687, 1697-1700(1990).
14. See references in Marvin Frankel and Charles M. Linke, “The Value of Life and He¬
donic Damages: Some Unresolved Issues,” 5 Journal of Forensic Economics 233, 237-243
(1992), and the very interesting discussion in Sherwin Rosen, The Value of Changes in Life
Expectancy,” 1 Journal of Risk and Uncertainty 2S5 (1988).
15. Isaac Ehrlich and Hiroyuki Chuma, “A Model of the Demand for Longevity and the
Value of Life Extension,” 98 Journal of Political Economy 761,781 (1990) (tab. 5).
308 Normative Issues

lives to be of negligible value merely because just a few years remain to

them. So we would expect the tort damages awarded in cases in which
elderly persons had been killed to be substantial, at least in those jurisdic¬
tions that do not confine recovery in wrongful death cases to purely pecu¬
niary losses sustained by survivors. The qualification is important. Origi¬
nally those were the only losses recoverable in a wrongful-death suit. The
common law did not award damages in death cases; and the earliest
wrongful-death statutes, which altered the common law rule, merely pro¬
vided a remedy for the decedent’s dependents, who had been deprived of
his support. This is still the regime in some states. Strictly interpreted, it
limits damages to the pecuniary loss suffered by the survivors, and this will
ordinarily be slight when the decedent is retired. (It may even be negative:
death accelerates bequests.) Damages for nonpecuniary losses are, how¬
ever, increasingly allowed in suits for wrongful death, under one rubric or
another, such as loss of companionship.'® And a number of states now try
to compensate the loss to the decedent himself, rather than just the loss to
survivors. For example, a Missouri statute allows the estate of the resident
of a nursing home to maintain a wrongful death suit against the home, and
the reason behind the statute is, precisely, the unlikelihood that a resident
of a nursing home would be providing support or other services to any¬
one.'^ Yet only one state, Connecticut, plus the federal courts in civil rights
tort suits, goes all the way and allows damages for loss of the enjoyment
(utility) of life'*—what are called “hedonic” damages—although juries
often award damages for this loss sub rosa, as by exaggerating the dece¬
dent’s lost earnings or his conscious pain and suffering before death.
Jurisdictions that do allow the recovery of damages for nonpecuniary
losses in death cases also allow the recovery of damages for pecuniary
losses in such cases, so we would expect total damages in wrongful-death
cases involving old victims to be on average lower than in cases involving
young ones. More precisely, we would expect that the ratio of damages in
wrongful-death cases involving younger victims to damages in wrongful-

16. See Dan B. Dobbs, Law of Remedies: Damages—Equity-Restitution, vol. 2, § 8.3(5)

(2d ed. 1993). The point was made with express reference to elderly plaintiffs in Borer v. American
Airlines, Inc., 563 P.2d 858 (Cal. 1977).
17. See Stiffelman v. Abrams, 655 S.W.2d 522 (Mo. 1983).
18. O’Hara, note 13 above, at 1692 n. 26; see also Dobbs, note 16 above, § 8.3(5), p. 443;
Andrew Jay McClurg, “It’s a Wonderful Life: The Case for Hedonic Damages in Wrongful Death
Cases,” 66 Notre Dame Law Review 57, 62-66, 90-97 (1990), esp. 65 n. 33; cf. Annotation,
“Excessiveness or Adequacy of Damages Awarded for Personal Injuries Resulting in Death of
Retired Persons,” 48 A.L.R.4th 229 (1986).
Legal Issues: A Sampler 309

death cases involving older victims will be greater than 1 but smaller than
the ratio of the purely financial losses of the two groups.
Consistent with this expectation, in a sample of verdicts in wrongful
death cases rendered in 1992 and 1993 the average verdict in 73 cases in
which the victim was between 65 and 85 years old was $1.2 million, which
was 73 percent of the $1.7 million average for victims aged 25 to 45.And
in a study of tort compensation in fatal airplane accidents, the ratio of com¬
pensation received when the victim was in the 40 through 49 age group (the
peak age group for receipt of compensation) to the compensation received
when the victim was 70 or older was 7.17, whereas the ratio of the purely
financial loss (present value of anticipated future earnings and retirement
income) of the victims in the two age groups was 8.49.^° The difference in
result between the two samples is striking. The airplane-accident sample is
older, and the much higher ratio of young decedents’ to old decedents’
damages than in the more recent (1992-1993) sample may indicate that the
severe limitations that most jurisdictions traditionally placed on damages
for nonpecuniary losses in wrongful death cases are waning.
The largest fatal-accident sample I have found comprises 224 verdicts
in wrongful death cases, rendered by juries in the courts of Cook County,
Illinois, between 1959 and 1979.^' The peak age is 46 through 50; the
17 plaintiffs in that group obtained an average verdict of $294,682. The
24 plaintiffs in the highest age group (61 through 70) obtained an average
verdict of $145,861, which was 49.5 percent of the average verdict for
members of the younger group. Despite the age of the sample, the results
are much closer to those of the most recent sample. Within the 61-70
group, however, the average verdict falls rapidly with age, being only
$40,498 for decedents aged 66 through 70; but the size of this subsample
is very small (10).
One would expect the ratio of verdicts for elderly tort victims relative
to the verdicts for nonelderly victims to be higher in nondeath cases, since
the same accident is expected to inflict a more severe injury on the victim
the older he is. This is confirmed by Cook County data. For verdicts in all

19. The sample was obtained from the LRP-JV computerized database of jury verdicts,
which is available on West Publishing Company’s “Westlaw system.
20. Computed from Elizabeth M. King and James P. Smith, Economic Loss and Compen¬
sation in Aviation Accidents 35, 48 (RAND Institute for Civil Justice R-3551-ICJ 1988) (tabs. 4.4,
5.7).
21. The sample was obtained by Professor George L. Priest of the Yale Law School m con¬
nection with his ongoing study of tort litigation in Cook County.
310 Normative Issues

such cases brought by victims of automobile accidents (including accidents

to pedestrians), the peak age was again 46 through 50 (n = 349), and the
average verdict was $43,034, compared to $28,159 for the 61 through 90
age group (n = 383). The average verdict of the elderly victims in the
nondeath cases was thus 65.4 percent of that of the nonelderly victims,
compared to only 49.5 percent in the death cases. It may seem odd that
the average verdict of the elderly victims does not exceed that of the non¬
elderly, if the average injury is more severe. But, precisely because the
expected injury is more severe to the elderly, we can expect them to take
more efforts to reduce that severity, as by fastening their seatbelts more—
which we know they do. And damages in nondeath as in death cases in¬
clude lost earnings, and these are higher for people who are working than
for retired people—indeed, retirement income is not diminished at all by a
disabling accident. The younger the comparison group, the more likely are
the damages of elderly victims of nonfatal accidents to equal or even ex¬
ceed those of young victims, since younger victims are likely both to have
lower earnings and to be injured less severely. Consistent with this sugges¬
tion, the average damages of the 61-to-90 group ($28,159) were 136.8 per¬
cent of the average damages of the 26-to-40 group ($20,590; n = 1150).
Studies of the value of life bear on age in the following respect: they
show that in deciding what premium to demand for dangerous work, young
workers discount the cost of future injury or death at a high rate.^^ It could
be argued that they should be held to this decision if they are later killed or
injured. But this would ignore the multiple-selves problem. The fact that
the young self has relatively little consideration for the old should not au¬
tomatically determine the rights of the old. The behavior of elderly people
with regard to matters of personal safety suggests that they do not value
their lives at the discounted value set on those lives by their young selves.
For similar reasons explained in chapter 4, the fact that young persons
might be unwilling to buy insurance that would provide for their medical
and other needs should they be disabled by an accident in their old age is
not a persuasive argument against awarding full tort damages in cases of
such disability.

Elderly Offenders and Prisoners

Chapter 6 identified as a potential problem of deterrence the truncation of
punishment for the old offender when punishment takes the form of im-

22. W. Kip Viscusi, “The Value of Risks to Life and Health,” 31 Journal of Economic
Literature 1912, 1921 (1993).
Legal Issues: A Sampler 311

prisonment. This problem can be alleviated although not solved by placing

greater emphasis on fines as a method of punishment of the old. This is not
done at present. The federal sentencing guidelines do state that “age may
be a reason to impose a sentence below the applicable guideline range
when the defendant is elderly and infirm and where a form of punishment
such as home confinement might be equally efficient as and less costly than
incarceration.”^^ But there is no reference to fines. The only point that
seems to have occurred to the authors of the guidelines with respect to
elderly offenders is that their feebleness makes them less of an escape risk,
so a laxer form of incarceration than confinement to a prison may be effec¬
tive (and will cost less). It is not a good point. To the extent that their
physical or mental infirmities “imprison” elderly offenders in their home
anyway, home confinement will have little or no effect as punishment; the
increment of confinement will be too small, and possibly nil. To solve this
problem by imposing harsher conditions of confinement on the elderly and
infirm offender would be barbarous. Imprisonment seems not to be the an¬
swer for elderly offenders; more consideration should be given to the use
of fines.
The obvious objection is that many offenders, old as well as young, are
insolvent. Yet this may not be quite so serious an obstacle to the greater use
of fines as a method of deterring elderly offenders as it would be to deter¬
ring young ones. The old offender is likely to have some social security or
other retirement income, so that a fine payable in installments, for the rest
of his life if necessary, can bring about a dramatic reduction in his standard
of living. Should he have a big lump of property (and this is more likely in
the case of an old offender than in the case of a young one), chances are
that he ascribes great utility to that lump—maybe he has a strong bequest
motive—so again he can be hurt badly by a heavy fine. Indeed, he can be
hurt more in this case than in the first, since, in the case of elderly persons,
installment fines suffer from the same defect as imprisonment: truncation
due to the abbreviated life expectancy of the elderly. But unless truncation
is a more serious problem with fines (and why should it be?), they are
superior to imprisonment because cheaper to administer.

23. United States Sentencing Commission, Guidelines Manual § 5H1.1, p. 303 (Nov.
1994). The Violent Crime Control and Law Enforcement Act of 1994, § 70002, 18 U.S.C.
§ 3582(c)(1)(A), authorizes the release at age 70 of a prisoner who has served at least 30 years in
prison pursuant to a federal sentence of mandatory life imprisonment for having committed three
or more violent felonies, provided the defendant is not a present danger to the safety of any other
person or the community. This is a step in the right direction, but applies to only a tiny subset of
criminal defendants.
 312 Normative Issues

The most serious problem of criminal punishment of the old comes

from the increasing frequency of life sentences, mainly of young offenders,
who are numerous, without possibility of parole. The statute that ordained
the federal sentencing guidelines also abolished parole for offenders sen¬
tenced under the guidelines. The states are moving in the same direction.
There is a danger, though it seems not yet to have materialized, at least in
the federal prison system, that the abolition of parole will result in a sub¬
stantial increase in the number and percentage of elderly prisoners. Accord¬
ing to statistics furnished me by the U.S. Bureau of Prisons, as of March 17,
1994, 5 percent of federal prison inmates were aged 51 - 55, 3 percent were
56-60, 2 percent were 61-65, and only 1 percent (compared to almost
13 percent of the U.S. population as a whole) were over 65. Even though it
is widely believed—plausibly, in view of the trend to longer sentences and
to abolishing parole—that the number of elderly prisoners is soaring,^"*
the total percentage of federal prisoners over 50 years old in 1994—
11 percent—was essentially unchanged from five years earlier, when
12 percent of prisoners had been 50 years old or older.^^ But it is too soon
to measure the impact of recent changes in punishment practices. Young
offenders harshly sentenced within the last decade and ineligible for parole
will not be elderly for several more decades.
We have time to consider carefully, therefore, whether we want to have
substantial numbers of geriatric prisoners. I believe we should not want
that. Since old people, even if they committed serious crimes in their youth,
generally are harmless now that they are old, the incapacitative function of
criminal punishment is not served by retaining them in prison. Since most
violent crimes are committed by relatively young people, by the time the
prisoner reaches old age the chances are that his crimes have been largely
forgotten, reducing retributive pressures for continued punishment, unless
his crimes were of a particularly heinous character—and, fortunately, such
crimes are still relatively rare. And the deterrent effect of such punishment
is apt to be negligible, because of the discounting of future costs to present
value, or more dramatically the tendency of one’s young self to write off
one’s old self. Consider a 20-year-old murderer sentenced to life impris¬
onment without possibility of parole. Suppose his life expectancy at age 20
is 55 years. Consider the incremental deterrence from imprisoning him af-

24. Gary Marx, “Some Would Free Inmates Held in Chains of Age,” Chicago Tribune,
June 13, 1994, p. 1.
25. Peter C. Kratcoski and George A. Pownall, “Federal Bureau of Prisons Programming
for Older Inmates,” Federal Probation: A Journal of Correctional Philosophy and Practice June
1989, pp. 28, 30.
Legal Issues: A Sampler 313

ter he reaches the age of 70. At a discount rate of 10 percent, a cost of $1

to he paid 50 years hence is equal to only $0.0085 today, which is to say
less than 10; for the entire period between ages 50 and 55 it is much less
than a nickel. This is monetary discounting, but (as multiple-selves analysis
implies) nonmonetary cost and benefits are also discounted by most people.
Criminals probably have on average very high discount rates. One reason
people become criminals is that their opportunities for lawful earnings are
meager, in part presumably because they have not invested heavily in their
human capital; and we know that the higher one’s discount rate is, the less
one is likely to invest in one’s human capital, since the costs are incurred
in the present but the benefits are obtained in the future. Not too much
weight should be placed on this consideration, however. An alternative rea¬
son for not investing heavily in one’s human capital is that the expected
return may be small, perhaps because one is not sufficiently intelligent to
benefit significantly from education or on-the-job training or, as a conse¬
quence of poverty, discrimination, or other factors, one does not have ac¬
cess to education or training. But if, despite this qualification, criminals do
tend to have high discount rates, the prospect of having to spend additional
years in prison when old is unlikely to have a significant deterrent effect on
the young criminal.
So the social benefits, in terms of any of the influential theories of the
purpose of criminal punishment, of keeping young offenders in prison
when they become elderly probably are slight, except in the case of particu¬
larly atrocious offenses; and here capital punishment is a possibly attractive
alternative, as I shall note. The discounted present cost of imprisoning the
youthful offender when, many years later, he is elderly may also seem
small. But it may not be. The discount rate applicable to public expendi¬
tures (here, expenditures on building and operating prisons) is lower than
the private discount rate of the average criminal. Indeed, it may be zero.^*
The argument that it is low or even zero draws support from the idea
of multiple selves. Discounting the costs to be borne by future genera¬
tions in effect treats the members of those generations as future stages
in the life of the present generation rather than as different persons. If we
are wrong to discriminate dgainst our future selves, even more clearly are
we wrong to discriminate against future persons whom we have no argu¬
able right to control.

26. See, for example, Derek Parfit, Reasons and Persons 480-486 (1984); Tyler Cowan and
Derek Parfit, “Against the Social Discount Rate,” in Justice between Age Groups and Generations
144 (Peter Laslett and James S. Fishkin, eds., 1992).
314 Normative Issues

Granted, this analysis has more force with regard to discounting utility
than with regard to discounting wealth. If real (inflation-adjusted) incomes
are rising over time in dollar terms and the marginal utility of income di¬
minishes as income rises, a dollar of future income will confer less utility
on the future recipient of that dollar than a dollar today would confer on a
present recipient, assuming the two individuals have similar values and pre¬
ferences.^’ But I do not require a social discount rate of zero to make my
point, which is simply that the costs of retaining the young offender in
prison when he is old are likely to swamp the benefits.
The higher medical expenses of elderly persons, however, do not figure
in my analysis. Not only are there potentially offsetting cost savings be¬
cause an old prisoner creates a less acute problem of security than a young
one, but if the old prisoner is released he will incur the same medical ex¬
penses and they will be paid for by the government too, through the Medi¬
caid or Medicare programs. True, he may be healthier in prison than out,
and so incur lower medical costs if he is not released, but by the same token
he is likely to live longer in prison, and therefore incur greater lifetime
medical expenses.
The older the prisoner was when he committed the crime, the less
weight discounting will have in determining the optimal length of impris¬
onment. So it would not be wise to have a rule automatically releasing all
offenders when they reach retirement age, as it were, even if it is certain
that they will commit no further crimes. Otherwise there would be no ef¬
fective punishment for crimes committed by persons on the verge of retire¬
ment. It does seem to me, however, that all prison sentences should be
capped at 40 or 50 years, or at least that parole should be retained for pri¬
soners of advanced age.’^ For crimes so heinous that such sentences seem
too short, capital punishment may be preferable to adding on meaningless
but not costless increments of imprisonment.
Comparison between this section of the chapter and the previous one
indicates that the tort system has a better (though not perfect) grip on the
special problems presented by old age than the system of criminal justice
does. This is not surprising. Other studies have found evidence, and offered
theoretical reasons, for the proposition that common law (that is, judge-
made) systems of regulation are more responsive to economic considera-

27. Gordon Tullock, “The Social Rate of Discount and the Optimal Rate of Investment:
Comment,” 78 Quarterly Journal of Economics 331 (1964); William J. Baumol, “On the Social
Rate of Discount,” 58 American Economic Review 788, 800-801 (1968).
28. Even without parole, the governor, or in the case of federal crimes the President, can use
the power of clemency to comihute the elderly prisoner’s sentence.
Legal Issues: A Sampler 315

tions than legislative systems are;^® and legislation looms larger in the
modern criminal law than it does in tort law. In the next section, we con¬
sider some additional evidence concerning the common law’s grasp of the
problems of old age.

Dementia and Capacity

The bearing of senile dementia on legal capacity—such as the capacity to
make a will or a contract, or to testify as a witness—and legal responsi¬
bility, for example for criminal acts, presents fascinating issues of both an
analytical and a practical character. We know that senile dementia, as dis¬
tinct from the mild cognitive impairment that is an especially common but
relatively minor affliction of old age,^° is progressive. So it would be absurd
to declare a person incompetent to change his will, sign a lease, testify in
court, or be prosecuted for a criminal act, merely upon proof that he had
displayed the earliest symptoms of senile dementia, such as occasional
disorientation and loss of short-term memory. It would be equally absurd
to ascribe legal capacity or responsibility to a person so demented that he
did not recognize the members of his immediate family. But where in the
continuum between these extremes should the line be drawn? The increas¬
ing prevalence of senile dementia makes this an urgent question.^'
Paradoxically, the progressive character of the disease enables the
question to be avoided in some cases. For it implies that an individual who
has been determined to be mentally competent (despite his dementia) at
time t must have been competent at all earlier times.^^ So if a criminal
defendant is determined to be competent to stand trial, this implies that he
was competent at t — n to commit the crime for which he is to be tried,
assuming that commission of the particular crime did not require greater
mental capacity than is required to follow events at trial and confer with
counsel.
The last qualification invites attention to a critical point: the type and
amount of mentation required for an act vary significantly among the many

29. See Richard A. Posner, Economic Analysis of Law (4th ed. 1992), esp. pt. 2.
30. See chapter 1. The distinction has long been recognized by the law. See, for example. In
re Will of Wicker, 112N.W.2d 137, 140-141 (Wis. 1961).
31. See Edward Felsenthal, “Judges Find Themselves Acting as Doctors in Alzheimer’s
Cases,” Wall Street Journal (midwest ed.). May 20, 1994, p. Bl.
32. Senile people often flicker in and out from day to day, so it is important that t be a
sufficiently extended period to enable a reliable determination of the individual’s average mental
capacity.
33. See, for example. United States v. Rainone, 32 F.3d 1203, 1208 (7th Cir. 1994).
316 Normative Issues

different acts to which the law attaches legal significance. This is a general
point about challenges to mental competence. It is clearest when the issue
is responsibility rather than capacity. The law attaches sanctions, criminal
or otherwise, to conduct in the hope of discouraging it; that at least is a
major goal of sanctioning. From the standpoint of deterrence the issue is
not the defendant’s sanity or intelligence, but his deterrability. One ob¬
serves that many people who are retarded or psychotic nevertheless re¬
spond to incentives—for example, buy less of a good when the price
rises, and avoid obvious dangers—sufficiently to be able to live on their
own, rather than having to be institutionalized. Why should they not also
be influenced by threat of punishment or (less probably) civil liability for
transgressions known by them to be such? There rightly is no generalized
defense of insanity or mental defect to legal responsibility. To avoid legal
responsibility a defendant must show that his mental condition made it im¬
possible for him to understand or respond to the law’s “signals” concerning
the consequences of particular conduct.
The same thing is true with respect to senile dementia. I think the court
was right, therefore, in In re Estate of Peterson^'^ to hold that the fact that a
man whose will was challenged was 83 years old, blind and (at least to
some degree) senile, was unable to feed or clothe himself, and had “hallu¬
cinations of little horses flying around the room and wrinkling their noses
at him, and his chasing them away by spitting at them,” did not necessarily
show that he was incompetent to destroy his old will and make a new one.^^
There was testimony that he had had lucid conversations with others con¬
cerning the will. For example, he had told one person that the beneficiary
under the old will had received enough from him in gifts—a perfectly
sound reason for making a new will with a new beneficiary. It would be
different if he had thought that the beneficiary of his old will was one of
the little horses or was responsible for their harassment of him.
Another way to understand the law’s refusal to attach controlling sig¬
nificance to an individual’s being afflicted with senile dementia is by noting
that the effect of dementia on an individual’s capacity to make a rational
Judgment is a function not only of the stage of the disease but of his mental

34. 360 P.2d 259, 267 (Nev. 1961). For similar cases, see In re Will of Wicker, note 30
above; Wright v. Kenney, 746 S.W.2d 626 (Mo. Ct. App. 1988). For a case finding that the testator’s
senility had progressed to a point at which he was incompetent to make a will, see Creason v.
Creason, 392 S.W.2d 69 (Ky. 1965).
35. Similar cases with a similar result, although involving deeds rather than a will, are
O’Brien v. Belsma, 816 P.2d 665 (Ore. Ct. App. 1991); Feiden v. Feiden, 542 N.Y.S.2d 860 (App.
Div. 1989), and Weir v. Ciao, 528 A.2d 616 (Pa. Super. Ct. 1987). All three cases note the “lucid
interval” phenomenon. See note 32 above.
Legal Issues: A Sampler 317

capacity when the disease began. If A had an IQ of 100 on the eve of exhib¬
iting the earliest symptoms of senile dementia, and by the time he made his
new will his IQ had declined to 80, he would be at the same mental level
as B, a young person, mildly retarded, who has an IQ of 80 (90 percent of
the population has a higher IQ). So the question would be: are persons
having IQs of 80 competent to make wills? (Yes.)
This approach is a little simplistic, I admit, because the mental conse¬
quences of senile dementia are not identical to those of all other forms of
mental impairment, even if a “bottom line” IQ assessment would score
them the same. In particular settings the loss of short-term memory that is
a conspicuous symptom of senile dementia may bear more heavily on A’s
mental capacity than on, say, 5’s inability to read complex documents. Or
more lightly. The court in State v. Manocchio^^ was wrong to think that
cross-examination designed to bring out a witness’s loss of short-term
memory would tend to show that his memory of events fifteen years earlier
must be equally or more impaired. But the principle is sound: it is the in¬
dividual’s mental capacity, rather than whether the limitations of that ca¬
pacity are due to something classified as a disease, in which the law takes
an interest.^’
I am led by this analysis to question the reasoning of Davis v.
A woman sued her adult daughter for the wrongful death of the woman’s
husband (the daughter’s father). The father was senile; “his condition
vacillated between irrational behavior and drug induced quietude and his
memory was almost entirely gone.” He moved in with his daughter and
her husband, who gave him their bedroom. The bedroom contained a chest
of drawers, in one of which there was a loaded pistol that the daughter had
forgotten to remove when her father moved in. One morning the father
opened the drawer, took out the pistol, and shot himself. The court held that
it was foreseeable to the daughter “that an ambulatory senile patient sub¬
ject to spells of ‘wildness’ ” might come upon the loaded pistol and, having
found it, would “mindlessly use it to work injury on himself or another.” ‘‘o
But why assume that the father had acted “mindlessly?” Senile people
often are aware that they are senile—or in any event that something is
terribly wrong with them—and, depending on the individual’s values and

36. 523 A.2d 872 (R.I. 1987).

37. This point is clearly recognized in the cases. See, for example, Dulnikowski v. Stan-
ziano, 172 A.2d 182, 183-184 (Pa. Super. Ct. 1961).
38. 206 S.E.2d 655 (Ga. Ct. App. 1974).
39. Id. at 656.
40. Id. at 657.
318 Normative Issues

circumstances, this may provide an entirely rational motive for suicide. It

is unclear whether that is what happened in Davis. It is not even certain
that the father intended to shoot himself. But it was a mistake to assume, as
the court appears to have done, that everything a demented person does is
irrational. In a will contest, courts examine what might be termed loosely
the “objective” rationality of the testator for corroboration or refutation of
an inference of competence. Recall that in the Peterson case the beneficiary
of the revoked will had been provided for, and this provided a rational
motive for the revocation of the old will and the making of the new one.
The same approach could have been used, or at least considered, in the
Davis case. But the continued stigmatization of suicide makes judges re¬
luctant to acknowledge that it can be a rational act.
 13
Age Discrimination by Employers and the
Issue of Mandatory Retirement

Even before the enactment of ERISA made it more likely that an employer
would resort to the threat of discharge (a threat that to be credible would
have to be carried out from time to time) in order to discipline its employ¬
ees, Congress had made it more difficult to fire older employees by enact¬
ing the Age Employment in Discrimination Act in 1967.' The Act, as sub¬
sequently amended, forbids employers to discriminate on grounds of age
against any employee aged 40 or over. Originally the protected class was
40 to 65, so mandatory retirement at age 65 was permitted. The lid was
raised to 70 in 1978 and removed altogether in 1986. Mandatory retirement
at any age, along with any other measures retail or wholesale by which an
employer treats an employee worse because of age, is, with a few excep¬
tions, now forbidden. I argue that the age discrimination law is largely in¬
effectual but that to the extent it is effective it has a perverse impact both
on the welfare of the elderly and on the equality of income and wealth
across the entire population. The age discrimination law is at once ineffi¬
cient, regressive, and harmful to the elderly.

1. 29 U.S.C. §§ 623 et seq. For contrasting evaluations, both emphasizing the economics
of the statute, see Richard A. Epstein, Forbidden Grounds: The Case against Employment Dis¬
crimination Laws, ch. 21 (1992), and Stewart J. Schwab, “Life-Cycle Justice: Accommodating
Just Cause and Employment at Will,” 92 Michigan Law Review 8 (1993). Many states have their
own laws forbidding age discrimination in employment, and there are other federal statutes forbid¬
ding age discrimination, such as the Age Discrimination Act, 42 U.S.C. §§ 6101 et seq., but I shall
ignore these other laws.

319
320 Normative Issues

The Nature and Consequences

of Age Discrimination in Employment
Animus discrimination. The justification offered for the law was that
people over 40 are subject to a form of prejudice, “ageism,” that is analo¬
gous to racism and sexism. After putting to one side the use of the word
as a synonym for anything that disadvantages an older worker^ (so presby¬
opia would be “ageist”), we can posit two kinds of ageism, only one plau¬
sible. The implausible is a systematic undervaluation, motivated by igno¬
rance, viciousness, or irrationality, of the value of older people in the work
place. This is sometimes referred to as “animus discrimination.” I do not
deny that there is resentment and disdain of older people in our society
(see chapter 9), or widespread misunderstandings, some disadvantageous
to the old. I have given a number of examples; recall from chapter 7, for
example, that elderly people may seem old-fashioned because they “cling”
to “outmoded” methods, yet the outmoded methods may be “clung to”
only because the incremental benefit of the latest method is slight.
But the present chapter is about the work place. Even apart from com¬
petitive pressures for rational behavior, which are considerable in private
markets, the people who make employment policies for corporate and other
employers and most of those who carry out those policies by making deci¬
sions about hiring or firing specific workers are at least 40 years old and
often much older. It is as if the vast majority of persons who established
employment policies and who made employment decisions were black,
federal legislation mandated huge transfer payments from whites to blacks,
and blacks occupied most high political offices in the nation. It would be
mad in those circumstances to think the nation needed a law that would
protect blacks from discrimination in employment. Employers—who have
a direct financial stake in correctly evaluating the abilities of their employ¬
ees and who for the most part are not young themselves—are unlikely to
harbor either serious misconceptions about the vocational capacities of the
old (so it is odd that employment should be the main area in which age
discrimination is forbidden) or a generalized antipathy toward old people.
To put the point differently, the kind of “we-they” thinking that fosters
racial, ethnic, and sexual discrimination is unlikely to play a large role in
the treatment of the elderly worker.^ Not because a young person will (in

2. The sense in which it is used in William Graebner, A History of Retirement: The Mean¬
ing and Function of an American Institution, 1885-1978, ch. 2 (1980).
3. Cf. John Hart Ely, Democracy and Distrust: A Theory of Judicial Review (1980).
Age Discrimination by Employers 321

all likelihood) someday be old; to put too much weight on the continuity of
personal identity would slight the multiple-selves issue. But because the
people who do the hiring and firing are generally as old as the people they
hire and fire and are therefore unlikely to mistake those people’s vocational
abilities. One should not be surprised at how slight and equivocal the evi¬
dence that employers misconceive the ability of older workers is. Such
workers do have trouble finding new jobs at high wages. But this is because
the wages in their old jobs will have reflected firm-specific human capital
that disappeared when they left and that they cannot readily replace be¬
cause of the cost of learning new skills, and also because the proximity of
these workers to (voluntary) retirement reduces the expected return from
investing in learning new skills.'^
One study found that “nearly 90 percent of [elderly] job losers’ wage
reductions are explained by the nontransferability of the workers’ firm-
specific skills and knowledge or seniority.’’^ The authors ascribed the re¬
maining 10 percent to age discrimination, but they had no basis for this
ascription. The 10 percent, as they acknowledged, was merely “a residual
remaining after accounting for other factors” —and among the factors not
accounted for was a possible age-related decline in capability. As the study
was of the wages in new jobs of elderly workers who had lost their previous
jobs, the possibility that the workers sampled were underperformers was
indeed a significant one. The very next essay in the collection, a study of
young and old workers employed by the same firm, finds that the entire
difference in wages between the two groups is due to differences in invest¬
ment in human capital.^ And as I pointed out in chapter 4, empirical find¬
ings that workers 65 or older perform their jobs as well as younger workers
in the same enterprise are vitiated by selection bias: demonstrably unsatis¬
factory older workers will have been fired or nudged into retirement. The
fact that some elderly people are able to perform to an employer’s satisfac¬
tion is consistent with many not being able, in which event we would ex¬
pect the average wages of elderly workers to be lower for reasons unrelated
to discrimination.

4. Dian E. Herz and Philip L.-Rones, “Institutional Barriers to Employment of Older Work¬
ers,” 112 Monthly Labor Review, April 1989, pp. 14,20.
5. David Shapiro and Steven H. Sandell, “The Reduced Pay of Older Job Losers: Age
Discrimination and Other Explanations,” in The Problem Isn ’t Age: Work and Older Americans
37, 47 (Steven H. Sandell, ed., 1987).
6. Id. at 48.
7. Paul Andrisani and Thomas Daymont, “Age Changes in Productivity and Earnings
among Managers and Professionals,” in The Problem Isn’t Age: Work and Older Americans, note
5 above, at 52.
322 Normative Issues

One might think that if substandard elderly workers are weeded out,
the average wages of the elderly employed would be no lower than those
of the nonelderly employed, unless there were discrimination. But some of
those weeded out of their current employment because they no longer per¬
form to their employers’ satisfaction will not leave the labor force; instead
they will find lower-paying jobs, commensurate with their diminished ca¬
pabilities, and their wages will depress the average.
The very idea of “animus” age discrimination rests on its own miscon¬
ceptions—for example that employers insisted on mandatory retirement at
fixed ages because they underestimated the capabilities of older people.*
As we shall see, that was not the reason.

Statistical discrimination. The form of ageism (if it should be called

that) that is more plausible and better substantiated than animus discrimi¬
nation against the old consists of attributing to all people of a particular
age the characteristics of the average person of that age. It is an example of
what economists call statistical discrimination and noneconomists “stereo¬
typing”: the failure or refusal, normally motivated by the costs of infor¬
mation, to distinguish a particular member of a group from the average
member. Age, like sex, is one of the first facts that we notice about a person
and use to “place” him or her. We do this because we operate with a strong,
though often an unconscious, presumption, echoing the rigid age grading
that structures activities and occupations in many primitive societies, that
particular attitudes, behaviors, and positions in life go with particular ages.
“We judge one another with a notion of what status goes with what age:
he’s old to be a student, young to be a professor, old to marry, young to
retire. Some people sometimes are ‘off time’ but most people most of the
time ‘act their age.’
The presumption that age matters in these ways is rational. Otherwise
this book would have no subject; any talk of “65-year-olds” or “octogen¬
arians” would be as irrelevant to public policy as talk about the attitudes
and behaviors of people with green eyes or chestnut hair. But there is a
great deal of variance in the capacities, behaviors, and attitudes of persons

8. Erdman B. Palmore, Ageism: Negative and Positive 5 (1990).

9. Arlie Russell Hochschild, The Unexpected Community 21 (1973). On the salience of
perceived age in making judgments about a person’s traits, see Robert Bornstein, “The Number,
Identity, Meaning and Salience of Ascriptive Attributes in Adult Person Perception,” 23 Interna¬
tional Journal of Aging and Human Development 127 (1986). And for a brief but serviceable
summary of theories of statistical discrimination, see Paula England, “Neoclassical Economists’
Theories of Discrimination,” in Equal Employment Opportunity: Labor Market Discrimination
and Public Policy 59, 60-63 (Paul Burstein, ed., 1994).
Age Discrimination by Employers 323

in particular age groups and, partly as a result, great overlap between the
capacities of persons in different age groups. People age at different rates
and from different levels of capacity. So if age is used as a proxy for attri¬
butes desired or disliked by an employer, some people who are entirely
competent to perform to the employer’s specifications will not be hired, or
will be fired or forced to retire to make away for young people who actually
are less able.
This phenomenon does not, however, make age discrimination in em¬
ployment inefficient any more than the substitution in some other field of
activity of a rule (for example, do not drive faster than 65 miles per hour)
for a standard (do not drive too fast for conditions) need be inefficient. A
rule is simpler to administer than a standard and therefore cheaper, and the
cost savings may exceed the loss from disregarding circumstances that may
make the rule disserve the purposes behind it in a particular case. Rules
have higher error costs but lower administrative costs, standards lower error
costs but higher administrative costs, and the relative size of the two types
of cost will determine the efficient choice between the alternative meth¬
ods of regulation in particular settings. Statistical discrimination is an ex¬
ample of rule-based behavior, and since it is a method of economizing on
information costs we can expect it to be more common in settings where
those costs are high. One is not surprised therefore that age grading (like
literalism, another example of rule-based behavior designed to economize
on information costs) is more common in primitive than in advanced soci¬
eties.'® Yet even in advanced societies rules are frequently more efficient
than standards; so mandatory retirement, and other employment classifica¬
tions based on age, cannot be condemned out of hand as archaic. Few of us
would be comfortable if airline pilots or military officers could not be
forced to retire at any age without proof of individual unfitness."
Age grading illustrates how statistical discrimination can sometimes
operate in favor of, rather than against, a particular group, here by ascrip¬
tion of the maturity, wisdom, and disinterest possessed by some old people
to all or most of them. Another circumstance that has favored the old is that
few people understand selection bias. People generalize from the impres¬
sive performance of octogenarian judges that octogenarians have unsus¬
pected capabilities; but the advanced age at which most judges are ap-

10. Richard A. Posner, The Economics of Justice 169-170(1981).

11. The age discrimination law permits mandatory retirement of airline pilots at age 60, and
the pilots’ union objects. For evidence of relevant age-related decline, see Joy L. Taylor et al., “The
Effects of Information Load and Speech Rate on Younger and Older Aircraft Pilots’ Ability to Exe¬
cute Simulated Air-Traffic Controller Instructions,” 49 Journal of Gerontology'P\9\ (1994).
324 Normative Issues

pointed operates to draw judges from an unrepresentative segment of the

aging population. If the elderly benefit from statistical discrimination as
well as being hurt by it, maybe they would enjoy an undue advantage over
other groups if the law succeeded in eradicating statistical discrimination
against, as distinct from statistical discrimination in favor of, the elderly. I
would not put too much weight on this factor, however. For reasons stated
earlier, I would expect employers to have a generally clear-headed notion
of the characteristics of the average worker in the different age groups and
not be fooled by selection bias.
Mandatory retirement—a blanket rule against retaining a worker who
has reached a specified age, regardless of the particular worker’s actual
productivity—has three supports besides the general benefits of a rule.
First, knowing far in advance the age at which one will retire facilitates an
individual’s financial and retirement planning. A person could always de¬
cide he was going to retire at some particular age, yet he might fear that he
might change his mind—the multiple-selves problem, once again.
Second, because full social security benefits are available at age 65 and
are sharply reduced until 70 if the recipient continues working after reach¬
ing 65, there are powerful financial advantages to retiring at 65. If, there¬
fore, few workers would want to continue working after that point, the
benefits from individualized assessment of their fitness to do so will be
small, yet there are apt to be significant fixed costs of establishing and
operating the requisite machinery of assessment. This point suggests that
the abolition of mandatory retirement is unlikely to have a big effect on the
labor-force participation of the elderly, and we shall encounter evidence of
this later.
Third, if, as is plausible, a significant decline in a worker’s perfor¬
mance is probable within a few years after he reaches 65, the benefits
from individualized assessment will be reduced further because they will
be realized for only a short period. The costs of such assessment will
rise, moreover, because the employer will have to monitor the performance
of workers who have reached the stage of life at which a decline in job
performance is highly probable more carefully than the performance of
younger workers.
Conceivably the reaction against mandatory retirement, and the con¬
cern with age discrimination generally, may reflect the fact that statistical
discrimination, being a function of information costs, probably is nega¬
tively correlated with education and IQ, since educated and intelligent
people can absorb and use information more easily than other people. The
“rigid” or “authoritarian” personality that psychologists associate with
Age Discrimination by Employers 325

discriminatory attitudes’^ can be given an economic meaning: people of

lower intelligence or less education employ cruder screening devices, such
as stereotyping. As information costs fall on average in a society, statistical
discrimination increasingly becomes the domain of the uneducated and the
unintelligent, so class prejudice may incline the society’s elite to disparage
or even forbid the practice.
I wonder how apt this point is to age discrimination in employment,
though. The difference between an employee and an independent contrac¬
tor has a bearing on this question. The difference is this: the employee does
not sell his output to his principal, as the independent contractor does, but
instead is paid for his time. Usually this is because the worker’s output is dif¬
ficult to value precisely, which may be because it is team output rather than
individual output. The difficulty of valuation implies that assessment of the
employee’s contribution to the firm will be probabilistic rather than certain.
The employer will be trying to infer that contribution from characteristics of
the worker and of his performance. One characteristic is the worker’s age.
We know that age is often correlated with performance; and with age being
directly observable and performance not, it may be entirely rational for even
the most intelligent employer to use the former as a proxy for the latter.
Whether this procedure should be viewed as a form of “discrimina¬
tion” in an invidious sense may be doubted, unless it is somehow unfair to
judge a person as a member of a group rather than as an individual. Is it
unfair? We do it all the time, and could hardly act otherwise. Lacking un¬
mediated access to the “inner man” or complete knowledge of his life his¬
tory, we relentlessly “type” people and base our judgments on this typing.
If, as in the example of team output in a business enterprise, individual
evaluation would cost a lot, one needs an argument for forcing other people
to shoulder the cost. If employers are forbidden to use efficient methods of
evaluation, their labor costs will rise, and it is now generally accepted that
increases in payroll taxes or other labor costs are borne largely by the work¬
ers themselves, in the form of reduced wages or benefits.'^ The increase in
cost operates as a tax, and the incidence of a tax does not depend on which
side of the market (here, employer or employee) the tax is assessed on.'^’ If

12. See, with specific reference to “ageism,” Palmore, note 8 above, at 53-54.
13. See Jonathan Gruber and Alan B. Krueger, “The Incidence of Mandated Employer-
Provided Insurance; Lessons from Workers’ Compensation Insurance,” 5 Tax Policy and the
Economy 111 (1991), and studies cited in id. at 117-118.
14. See, for example, Joseph A. Pechman, Federal Tax Policy 223-224 (5th ed. 1987);
Laurence Kotlikoff and Lawrence Summers, “Tax Incidence,” in Handbook of Public Economics
1043, 1047 (Alan Auerbach and Martin Feldstein, eds., 1987).
326 Normative Issues

employers are forced by law to keep on inefficient elderly workers, workers

as a whole, few of whom either are wealthy or are guilty of “ageism,” will
in effect be taxed for the benefit of these elderly workers—yet the elderly,
prosperous recipients of substantial public largesse, are implausible candi¬
dates for the status of an oppressed class (see chapter 2).
Anyway the law cannot, merely by outlawing a particular form (or
many particular forms) of discrimination, force employers to judge every
worker as an individual. The costs of information are too high. They may
be particularly high in the case of elderly workers. As we saw in the first
chapter, variability in performance in an age cohort tends to grow as the
cohort ages. The greater the variability in a population of workers, the
longer the employer will have to search in order to find workers suitable
to his needs, unless he relies on some simple proxy or rule of thumb. The
costs of individualized assessment will be high. Increased variability may
also make the value of such assessments greater; but remember that the
expected return from selecting the very best worker (if he is old) will
be truncated, because an older worker is unlikely to continue working
for long.
Deprived of the age proxy, some employers will use other proxies for
ability or performance, such as test results, thus “discriminating” against
workers whose performance those proxies underpredict. Airlines, for ex¬
ample, if forbidden to impose mandatory retirement on their pilots, might
raise their standards of physical fitness, with the result that some perfectly
competent young pilots might be forced out. Some employers denied the
use of the age proxy will throw up their hands and, unable to distinguish
between good and bad workers of the same age, treat both groups indis¬
criminately, with the result that bad workers will benefit at the expense of
good ones. This is just another form of statistical discrimination. The vic¬
tims of the two different forms of statistical discrimination—the victims
of lumping together people of disparate abilities though the same age (the
form of statistical discrimination that the law encourages), or of treating
separately people who have the same abilities but are of different age (the
form of statistical discrimination that the law forbids)—will be different.
But there will still be victims. And the victims of the age discrimination
law, as distinct from the victims of what the law calls age discrimination,
may be people more marginal, more necessitous, than the average elderly
worker. Some evidence for this conjecture is that, as we shall see, most
plaintiffs in age discrimination cases are not “workers” at all, but manag¬
ers, professionals, and executives.
This discussion does not settle the question whether age discrimination
Age Discrimination by Employers 327

is efficient, but it bears on it. As with other forms of discrimination, statis¬

tical discrimination against the aged worker may impose an external cost—
that is, a cost to nonparties to the transaction between a given employer and
a given worker or applicant for work. If the exceptional aged—those young
in mind, body, and spirit—cannot cash in their exceptionality in the em¬
ployment market because very few employers will look behind chronologi¬
cal age in making employment decisions, they will have a suboptimal in¬
centive to invest in their human capital because the payback period will be
artificially truncated. A middle-aged professional who rationally believes
that he has and will retain youthful energy and intellectual flexibility will
nevertheless forgo making an investment in human capital that would not
be completely amortized until he was 70 years old, if he thinks that he will
be forced to retire at age 65 or denied a promotion merely because of the
average characteristics of his age cohort. The resulting underinvestment in
human capital is the joint product of individual decisions by a multitude of
employers no one of whom would be better off incurring substantial costs
to identify the handful of exceptional elderly workers. It is not a complete
answer that an individual worker might, by accepting a lower wage, induce
the skeptical employer to make an exception in his favor. The prospect of
having to accept such a wage cut would reduce the expected return to
the exceptional employee’s investment in his human capital and thus the
amount that he would be willing to invest. Also, as I shall point out, the
age discrimination law may as a practical matter rule out such transactions.
The argument that statistical discrimination is inefficient in the case of
age is unpersuasive. When the costs of making individualized assessments
of employees’ performance are prohibitive, as they often are, prohibiting
the use of the age proxy will lead to the substitution of other proxies. The
problem of underinvestment will be shifted, not solved. Whoever is “un¬
fairly” disadvantaged by the new proxy, in the sense that it does not mea¬
sure his abilities accurately (perhaps he does not do well on pen-and-pencil
tests because of deficiencies in his formal education, but is an excellent
worker nevertheless), will lack the incentive to make the optimal invest¬
ment in his human capital. And if the new proxies are less efficient than
age—as they probably will be, because otherwise they would in all like¬
lihood have been adopted without government prodding—wages will fall
because employers’ labor costs will be higher, and with lower wages there
will be less incentive for workers to invest in their human capital.
Only if other proxies—other handles for statistical discrimination—
are, unlike age, more costly to the employer than individualized assessment
will forbidding the use of the age proxy respond to the problem of under-
328 Normative Issues

investment in human capital by the exceptional elderly worker, by inducing

individualized assessment. This is unlikely, though no stronger statement
is possible on the basis of existing knowledge. And one does wonder how
serious the problem of underinvestment in human capital by the excep¬
tional aged ever was. Regardless of federal regulation, employers would be
unlikely to have the identical policies about retirement. There never was a
time when all employers had a policy of mandatory retirement at fixed
ages. Competent workers wanting to retire late or not at all—always a mi¬
nority of all workers—would tend either to be sorted to employers having
compatible retirement policies or to become self-employed.

The Effects of the Age Discrimination in Employment Act

The previous section questioned the need for a law against discrimination
on grounds of age. But we have the law, and we must now consider more
carefully its probable effects both on elderly workers and on the rest of
society. The first thing to note is the misfit between the scope of the Act
and the concerns of the elderly. The prohibition against mandatory retire¬
ment is clearly related to those concerns, since mandatory retirement be¬
fore 65 was rare. But the Act’s general prohibition against age discrimina¬
tion kicks in when a worker turns 40. I have done a study of court cases
under the Act (see discussion below), and only 10 percent of the plaintilfs
in my sample of cases, including those plaintiffs who challenged manda¬
tory retirement, are 65 or older—a smaller percentage than the percentage
of elderly people in the U.S. population as a whole. The main reason is
plain enough; most people who are 65 or older are voluntarily retired, so
are not protected by the Age Discrimination in Employment Act. Yet the
Act was “sold” by means of emotional rhetoric concerning the plight of
the elderly, in 1967 still viewed as a disadvantaged segment of American
society, even though the Act seems to have been designed and to be admin¬
istered in the interest primarily of nonelderly workers. It is unlikely that an
age discrimination statute so configured would benefit the elderly much,
and we are about to see that it may harm them.
Of course, since an elderly person’s income is apt to depend signifi¬
cantly on his income when he was in his prime working years, a statute that
increased the incomes of workers in those years could be thought to be
benefiting the elderly. But even if we disregard the multiple-selves problem
(the benefited elderly self may not be the same person as the younger self
who receives benefits under the age discrimination law), we shall see that.
Age Discrimination by Employers 329

ex ante, the beneficiaries of the law probably bear the costs of it as well,
and therefore do not, on average anyway, obtain a net benefit from it.

Hiring cases. We know that, wholly apart from any laws, employers
are reluctant to hire older workers. The cost of training an older worker is
higher than that of training a younger one because of the age-related de¬
cline of fluid intelligence, while the expected return to the investment in
training is lower because the older worker has a shorter working life expec¬
tancy.'^ The age discrimination law adds to the costs of employing older
workers, and hence to the reluctance of employers to employ them, by giv¬
ing them more legal rights against their employer than younger workers
have. By thus reducing the hiring prospects of older people, the Act per¬
versely impairs the incentive of the exceptional old to invest in their human
capital, by reducing the expected return to such an investment.
The Act does forbid age discrimination in hiring as well as in firing,
demotions, wages, and so forth. But it is largely ineffective against hiring
discrimination because of the extreme difficulty of proving substantial
damages in such cases. Damages are not the only relief available in a suit
under the Act; injunctive relief is also possible. But the disappointed appli¬
cant is unlikely to be satisfied with an order requiring the employer to hire
him; he would be entering upon the employment under most inauspicious
circumstances.
The reason that substantial damages are difficult to prove in hiring
cases is that ordinarily the plaintiff-applicant, if hired, would have received
a wage only slightly higher than his reservation wage; for if the job that he
applied for pays much more than his present job, he will have great diffi¬
culty persuading a jury that he was the best-qualified applicant. (If his next
best wage was $20,000, it is hardly likely that, if only he had been younger,
the defendant would have offered him $100,000.) The monetary stakes in
a discharge case will often be much greater. If the discharged employee’s
wage contained a return for firm-specific human capital, that wage will be
higher, maybe much higher, than he could get elsewhere—especially if he
is too close to retirement age, or too inflexible with regard to learning new
skills, for a new employer to think it worthwhile to invest in new specific

15. For empirical evidence of employers’ reluctance to hire older workers, see Robert M.
Hutchens, “Do Job Opportunities Decline with Age?” 42 Industrial and Labor Relations Review
89 (1988). Although health costs are also higher for older workers, it is not a violation of the age
discrimination law for the employer to take these costs into account in designing a wage-benefits
package for his employees.
 330 Normative Issues

capital for him. The difference between what the old worker was paid be¬
fore he was fired and the much lower wage that is the best he can hope for
in a new job provides the measure of his compensatory damages.
So it is no surprise that a large sample of litigated age discrimination
cases contained no hiring cases; more than two-thirds of the cases involved
termination (discharge or involuntary retirement), with most of the others
involving promotion or demotion.'^ A more recent study, of age discrimi¬
nation complaints lodged with the Equal Employment Opportunity Com¬
mission, finds that 87.9 percent of the complaints in which no other form
of discrimination was alleged besides age discrimination involved termi¬
nation, only 8.6 percent hiring.’’
My own study is of all court cases under the Age Discrimination in
Employment Act in which a final decision was rendered between Jan¬
uary 1, 1993, and June 30, 1994, on other than procedural grounds and was
reported in Westlaw, the West Publishing Company’s computerized data¬
base of judicial decisions. Table 13.1 summarizes the results, broken down
by form of age discrimination alleged and also by the outcome of the
litigation.
Limited as it is to reported court cases, the sample is not random and
cannot be assumed to be representative. In just the single year ending Sep¬
tember 30, 1993, the EEOC received 19,884 complaints of age discrimi¬
nation.'* Only a small and nonrandom fraction of the complaints filed with
the EEOC end up in court,'® and only a fraction of that fraction is decided
on the merits rather than being settled or being disposed of on a procedural
ground, and is reported in Westlaw. There are no data on the outcomes of
the other proceedings.

16. Michael Schuster and Christopher S. Miller, “An Empirical Assessment of the Age
Discrimination in Employment Act,” 38 Industrial and Labor Relations Review 64, 71 (1984)
(tab. 3); see also Michael Schuster, Joan A. Kaspin, and Christopher S. Miller, “The Age Discrimi¬
nation in Employment Act: An Evaluation of Federal and State Enforcement, Employer Compli¬
ance and Employee Characteristics: A Final Report to the NRTA-AARP Andrus Foundation” iv
(unpublished. School of Management, Syracuse University, June 30, 1987).
17. George Rutherglen, “From Race to Age: The Expanding Scope of Employment Dis¬
crimination Law,” tab. 5 (University of Virginia Law School, May 1994; forthcoming in Journal
of Legal Studies, June 1995). A person who believes that he has a claim under the Age Discrimi¬
nation in Employment Act may complain to the EEOC, which is empowered to sue on the person’s
behalf. If, as is usually the case, the EEOC decides not to sue, the person may then sue on his own.
Or he may bypass the EEOC and sue without having filed a complaint with the agency, provided
only that he notifies it of his intent to sue.
18. EEOC News Release, Jan. 12, 1994, p. 4 (tab. 3).
19. But remember that some court cases do not originate in complaints to the EEOC; filing
a complaint with the EEOC is not a prerequisite to suit.
Age Discrimination by Employers 331

Table 13.1 Reported ADEA Decisions by Outcome and by Type of Discrimination, 1993

Plaintiff Defendant % Plaintiff

Type Won Won Won % Type

Discharge 27 127 17.5 35.8

Constructive discharge' 2 26 7.1 6.5
RIF2 4 122 3.2 29.3

Mandatory retirement 6 4 60.0 2.3

Promoted/Demoted^ 1 27 3.6 6.5

Hiring 2 43 4.4 10.5

Other/Mixed 7 32 17.9 9.1

Total 49 381 11.4 100.0

' Employee quit because employer made the working conditions unbearable for him.
^Reduction in force, discussed later in this chapter.
^Employee complains about failure to be promoted, or about being demoted.

My study is consistent with the earlier studies in finding that hiring

cases are relatively rare—only 10.5 percent of the total cases in the sample.
The vast majority involved a termination of one kind or another. Further¬
more, the plaintiff won only two of the hiring cases—a winning percentage
of only 4.4 percent. In general, plaintiffs did very poorly, winning only
11.4 percent of the cases they brought. Such success as they did have (out¬
side of the other or mixed category) was, with the exception of the two
hiring cases, confined to cases that involved a termination. No plaintiff
complaining about a promotion or a demotion obtained a money judgment.
A low winning percentage for plaintilfs in a class of cases is not (pro¬
vided it exceeds zero!) conclusive evidence that these cases are “losers”
for plaintiffs to bring. Two points are important. The first is that a low
winning percentage could be an effect of high damages awards. The higher
the award if the plaintilf wins, the likelier he is to sue even if the probability
of winning is small. It is just like a lottery: the bigger the pot, the longer
the odds that the organizers of the lottery can set and still sell tickets. The
second point is that a winning plaintiff in an age discrimination suit is en¬
titled to reimbursement of his attorney’s fees by the defendant on top of
any damages awarded, while a losing plaintiff can be ordered to pay the
defendant’s attorney’s fees only if the suit was frivolous.
Neither point is compelling in light of the results of my study. In the
29 cases in which the plaintiff obtained damages in lieu of or in addition to
equitable relief, the average damages award was $257,546. This is an un-
 332 Normative Issues

impressive figure when one considers not only that the risk of winning
nothing is very great—so that when averaged together with the cases in my
sample in which the defendant won, the total damages awarded come to
only $29,360 per case, a modest expected gain for a federal case litigated
all the way to final judgment—but also that cases involving large stakes
are likely to be overrepresented in a sample of cases litigated to judgment.
For, in general, the greater the stakes in a case, the more likely the case is
to be litigated rather than to settle.^®
If courts awarded a winning plaintiff attorney’s fees large enough to
compensate the plaintiff’s lawyer, ex ante, no matter how small the prob¬
ability of his winning, then, unless lawyers were risk averse and could
not assemble a large enough pool of cases to eliminate the risk of losing a
particular case in the pool, every case in which the probability of the plain¬
tiff’s winning exceeded zero would be brought. But usually the courts
award the winning plaintiff the attorney’s fee he actually incurred, with
no multiplier to reflect the risk of loss. I do not have complete figures for
the attorney’s fees awarded in the cases in my sample, but in the 10 cases
in which the award of attorney’s fees was disclosed, the average award
($97,449) was 37.8 percent of the damages award. Projected to the entire
sample, this would raise the expected judgment (damages plus attorney’s
fees) from $27,466 to $40,469, which is still a modest amount for a federal
case litigated to judgment. When one considers that more than 100 million
people are employed in this country, a large percentage of them over 40,
the total number of suits that left a trace in Westlaw, and even the 20,000
complaints lodged with the EEOC in a somewhat shorter period, are mea¬
ger. The outcomes of the cases in my sample suggest why.
Damages in a hiring case are expected to be even lower on average
than in a termination case. The award in the sole hiring case in my sample
in which damages were awarded was $63,000, far below the average of
cases in which damages were awarded, though the case was remanded for
a determination whether the age discrimination had been willful, in which
event the plaintilf would be entitled to a doubling of his damages. Suppose
he gets his double damages, for a total of $126,000; and suppose that the
same amount is awarded in the other hiring case the plaintilf won, in which
the damages award is undisclosed. When these amounts are averaged over
the 45 hiring cases in the sample, 43 of which the defendant won, the ex¬
pected damages (ex attorneys’ fees) in this class of case is only $5,689. So
it is no surprise that so few hiring cases are brought. My guess is that they

20. Richard A. Posner, Economic Analysis of Law 556 (4th ed. 1992).
Age Discrimination by Employers 333

are brought mainly by inexperienced lawyers. Experienced lawyers rarely

litigate federal cases in which both the probability of winning and the judg¬
ment if the case is won are low; the expected cost of the suit is likely to
swamp the expected benefit. The causality could run in the opposite direc¬
tion; hiring plaintiffs lose because they are represented by inexperienced
lawyers. But this is unlikely. If experienced lawyers can win a particular
type of case, they will be drawn to it.
The high ratio of termination to hiring cases is not a peculiarity of age
discrimination cases.^' The essential feature that distinguishes the damages
potential of a hiring case from that of a firing case—absence versus pres¬
ence of significant firm-specific human capital—is unrelated to the nature
of the discrimination charged, except that the older the plaintiff is, the more
such capital he will have accrued, which will magnify his claim for dam¬
ages in a firing case. A point made even by scholars who support the laws
against employment discrimination is that the more likely a member of a
protected group is to bring a termination suit and the less likely he or she
is to bring a hiring suit, the greater will be the disincentive of employers to
hire the members of such groups.

Firing (and other discharge) cases. I have said that only if the em¬
ployee is let go is an age discrimination case likely to get as far as an award
of damages or any other remedy. And yet the merits of awarding substantial
damages even in such cases could be questioned, on the following ground
which has not, however, so far as I am aware, ever been advanced by an
employer, although it is firmly rooted in the basic human-capital model.
We recall from chapter 3 that part of a wage that reflects the greater value
to the firm of an employee who has firm-specific human capital may be
generated by the employer’s contribution to that capital. The employee
himself pays for the employer’s investment in the employee’s general hu¬
man capital—skills that he can employ elsewhere in the economy—by ac¬
cepting a lower wage. The employer would have no protection against the
employee’s using that capital to obtain a higher wage from another em¬
ployer, and he therefore will not pay for it. But the employer will pay for a
share, possibly a very large share, of the employee’s firm-specific capital,
because the employee by definition cannot obtain a return on that capital
from another employer. The greater the share of this capital that the em¬
ployer pays for, the lower will be the turnover of employees (for they will

21. John J. Donohue III and Peter Siegelman, “The Changing Nature of Employment Dis¬
crimination Litigation,” 43 Stanford Law Review 983, 1015 (1991).
22. Id. at 1024.
 334 Normative Issues

have a higher salary) and hence the likelier will the employer be to reap the
benefit of its investment.^^
This analysis argues for allowing the employer to deduct from dam¬
ages in an age discrimination firing case that portion of the difference be¬
tween the wage he paid the employee, and the lower wage that the em¬
ployee would command in his best alternative employment, that represents
a return on specific human capital for which the employer paid. The em¬
ployer loses this investment in the employee’s human capital by losing the
employee, and should not have to pay twice—first by swallowing the loss
of the investment, and then by in effect “buying” the investment back by
having to pay damages measured by its value.
The argument is sound, but it is too exotic, and the computations re¬
quired by it are too difficult, to be likely to commend itself to the courts.
Its real significance to an evaluation of the effects of the Age Discrimina¬
tion in Employment Act lies in directing attention to the fact that employers
have their own incentives, unrelated to law, to avoid firing competent em¬
ployees of any age, even if replacements are available. The employer has
invested in the employee, and if the employee is still productive the em¬
ployer is continuing to earn a return on the investment.
The analysis to this point suggests that insofar as the age discrimi¬
nation law forbids discrimination against individual employees, as dis¬
tinct from discrimination against age-defined classes of employees (mainly
through mandatory retirement at fixed ages, about which more later), it
may, like ERISA, have little effect. The abuse against which it is directed,
the arbitrary treatment of older workers, would be rare, at least in private
markets, even without the law. It would be rare because, as with ERISA,
employers have market incentives to avoid the abuse. But the qualification
“at least in private markets” is significant. The age discrimination law also
applies to public employment and to employment by colleges, universities,
foundations, and other not-for-profit employers. Public and not-for-profit
employers can be expected to discriminate more than private for-profit em¬
ployers, for two reasons. They face fewer market pressures to minimize
their labor costs; and the constraint on their obtaining profits gives them an
incentive to substitute nonpecuniary for pecuniary income, and one form
of nonpecuniary income is avoiding undesired personal associations.^-^ Of

23. Gary S. Becker, Human Capital: A Theoretical and Empirical Analysis, with Specific
Reference to Education 33-49 (3d ed. 1993).
24. See Armen A. Alchian and Reuben A. Kessel, “Competition, Monopoly, and the Pursuit
of Money,” in Aspects of Labor Economics 157 (National Bureau of Economic Research 1962);
Posner, note 20 above, at 350, 652-653.
Age Discrimination by Employers 335

the 256 cases in my sample for which the necessary information is avail¬
able, 23 percent were brought against government employers and 7.8 per¬
cent against nonprofit employers; these percentages greatly exceed the
percentages of the labor force for which these two classes of employer
account.
So: age discrimination against individual workers in whom the em¬
ployer has invested significant firm-specific human capital is unlikely, at
least by private employers, while age discrimination against other workers,
and against older persons seeking jobs, is unlikely to be rectified by the age
discrimination law because plaintiffs drawn from these categories are un¬
likely to have a substantial financial stake in suing.
Further evidence of the likely inefficacy of the law is the very poor win
rate of plaintiffs in age discrimination cases, as shown in table 13.1. It is a
surprising statistic because, unlike the situation in other employment dis¬
crimination cases until the 1991 amendments to the civil rights laws, plain¬
tiffs in age discrimination cases are entitled to a jury if their case gets as far
as trial. But few cases do. In my sample, while plaintiffs won 47.7 percent
of the 20 percent of the cases that went to verdict, 74.7 percent of the cases
were disposed of by summary judgment, meaning that there was no triable
issue; and plaintiffs won only 1.6 percent of those cases.^^

Why do most age discrimination plaintiffs do so poorly? When the

Age Employment in Discrimination Act was enacted back in 1967, many
employers were practicing age discrimination (primarily of the statistical
sort), and doing so openly. It took some time for the message that age dis¬
crimination was now an unlawful practice which if continued must be con¬
cealed to filter down to the corporate personnel who make the actual em¬
ployment decisions. They continued for some years blithely to generate
“smoking gun” evidence of age discrimination. By now, however, employ¬
ers have largely succeeded in purging such slogans as “you can’t teach an
old dog new tricks” from the vocabulary of their supervisory and person¬
nel staffs.^’ Some evidence that age discrimination cases are indeed in¬
creasingly difficult for plaintiffs to win comes from comparing the winning

25. The sum of 20 percent and 74.7 percent is only 94.7 percent; the remaining cases were
disposed of by other forms of judgment.
26. Or “old dogs won’t hunt,” held evidence of age discrimination in Siegel v. Alpha Wire
Corp., 894 F.2d 50, 55 (3d Cir. 1990).
27. “It is important to sensitize all managers to the fact that any type of age reference, even
in informal conversations, may have a negative impact on the organization's position [in an age
discrimination suit].” Robert A. Snyder and Billie Brandon, “Riding the Third Wave: Staying on
Top of ADEA Complaints,” Personnel Administrator, Feb. 1983, pp. 41,45 (emphasis in original).
336 Normative Issues

percentage of plaintiffs in my sample (11.4 percent) with the much higher

percentage in the Schuster sample (32 percent), which was drawn from
cases decided between 1968 (the first year after the enactment of the age
discrimination law) and 19863* But the difference may reflect, in part any¬
way, differences in the design of the studies.
In the absence of smoking-gun evidence of age discrimination, now
difficult to come by, a plaintiff must as a practical matter show that an
equally competent but younger employee was treated better. Such proof is
difficult because of the intangible elements in evaluating a worker’s perfor¬
mance other than in the simplest jobs—and the simplest jobs do not gen¬
erate the plausibly high damages claims that repay the costs and uncertain¬
ties of litigation. The simplest jobs require little human capital, whether
general or specific, and (partly for that reason) pay low wages.
Moreover, a firm that wants to get rid of an older employee can often
do so with near impunity by cashiering a younger employee at the same
time. One hears rumors that this is a common practice. It may be feasible
because there is high turnover among young employees anyway and the
firm may not yet have invested much in the young employee’s firm-specific
human capital (a principal reason why turnover of young employees is
high) and so has little to lose from firing him, though concern with reputa¬
tion must inhibit this Machiavellian strategy to some and perhaps to a great
extent. The “RIF” (reduction in force) is a related strategy; victims of
RIFs who complain on age discrimination grounds do very poorly—in
table 13.1, even more poorly than victims of alleged age discrimination in
hiring. (More on RIFs shortly.)
Another thing that makes it hard for the employee to win an age dis¬
crimination case is that older employees tend to be more costly to a firm
than younger ones, by virtue of receiving a larger package of wages and
benefits. The more costly they are, the more difficult it is to ascribe their
discharge to their age, as distinct from their expense. The older employee
may be more productive by reason of his greater experience, or he may be

28. See references in note 16 above. It is interesting to note that the winning rate of the
plaintiffs in my sample is only about half that of plaintiffs in all employment discrimination cases
(that is, including racial, ethnic, and sexual discrimination cases along with age discrimination
cases). See John J. Donohue III and Peter Siegelman, “Law and Macroeconomics: Employment
Discrimination Litigation over the Business Cycle,” 66 Southern California Law Review 709, 756
(1993) (fig. 4); Theodore Eisenberg, “Litigation Models and Trial Outcomes in Civil Rights and
Prisoner Cases,” 77 Georgetown Law Journal 1567, 1578 (1989). But these studies, like the
Schuster studies of age discrimination, deal with an earlier period (1977 to 1988) than my study,
so are not strictly comparable.
Age Discrimination by Employers 337

paid a higher wage either to discourage shirking in his last period of em¬
ployment or as a reward (akin to a pension in the contract theory of pen¬
sions, examined in chapter 12) for not having shirked previously. If he is
more productive, then he is not in fact more costly to the firm than a
younger, less well paid, but also less productive worker. And if he is being
paid a so-called “efficiency” wage either to discourage shirking or to repay
the “bond” that he posted as a young employee by accepting a lower salary
in exchange for an implicit promise of compensation later if he behaved,
he is merely receiving the benefit of his bargain. But a court is not apt to
tumble to the reason why the older employee is not really being “over¬
paid” and to see therefore that the employer is reneging on an implicit
contract. All the court can see is that the employer had a reason unrelated
to age for firing the older worker—he was more expensive.
Courts could try to deal with this problem by treating “discrimination”
based on salary as a form of age discrimination, since age and salary tend
to be positively correlated. But maybe because this would make it difficult
for firms to take rational steps to reduce their costs when they find, for
whatever reason, that they are paying wages in excess of the market, the
Supreme Court has rejected this approach.^^
My analysis suggests that age as such is unlikely to be a good predictor
of the likelihood of the plaintiff’s winning an age discrimination suit. The
older the employee, the easier it will be for the employer to make a plau¬
sible case that the employee was fired because he was failing or too expen¬
sive, and not because of his age as such. The length of time the plaintiff
was employed by the defendant is likely to be a better predictor of the
likelihood of the plaintiff’s winning. It is a proxy for the amount of specific
human capital invested in him, hence the amount of his damages, hence the
likelihood of his having sufficiently large expected damages to be able to
attract a competent lawyer to represent him. It is true that the larger the
expected gain from suit to the plaintiff, the larger the expected loss to
the defendant, who therefore can be expected to defend more vigorously
the higher the stakes. But there is a double asymmetry. First, there is some
threshold of expected gains from suit below which a potential plaintiff can¬
not make a credible threat to sue. Second and more interesting, the defen¬
dant normally has more to lose in an age discrimination case than the plain-

29. Hazen Paper Co. v. Biggins, 113 S. Ct. 1701 (1993). That is the interpretation placed on
Hazen by such cases as Anderson v. Baxter Healthcare Corp., 13 F.3d 1120, 1125-26 (7th Cir.
1994) (see also Hamilton v. Grocers Supply Co., 986 F.2d 97 [5th Cir. 1993]), although Schwab,
note 1 above, at 45 and n. 148, is not sure it is the correct interpretation.
338 Normative Issues

tiff has to win, because a victory for the plaintiff will encourage other suits
against the defendant.^” This asymmetry is greater, the smaller the mone¬
tary stakes, since those stakes are symmetrical. So it will be harder for
plaintiffs to win small cases than big ones. In a small case the plaintiff will
find it impossible to hire an excellent lawyer because the expected gains
from suit are so slight, while the defendant will be willing to pay to hire
such a lawyer because it will fear the effect of losing the suit on the number
of future claims against it.
The asymmetrical relation of employee and employer in the small case
may justify the “one-way” attorney’s fee shifting that is the norm in age
discrimination as in other discrimination cases. (That is, the winning plain¬
tiff normally obtains an award of attorney’s fees, while the winning defen¬
dant normally does not.) But in view of judicial reluctance to award fee
multipliers, it is unlikely that one-way attorney’s fee shifting eliminates the
asymmetry.
Mention of age-related decline makes it timely to note that the inter¬
action between the Age Discrimination in Employment Act and the Ameri¬
cans with Disabilities Act (ADA) has yet to be explored by the courts, so
recent is the latter Act; it was enacted in 1990, and the employment provi¬
sions did not become fully effective until July 1994. The legislative history
is pretty emphatic that old age is not in itself to be deemed a disability.^^
Yet the characteristic age-related ailments and deficits, such as frailty, mild
cognitive impairment, lack of energy and strength, and failing hearing and
vision may be.^^ In that event employers may be forced to adjust the de¬
mands they make on older workers in order to make it easier for the older
worker to remain employed, just as employers are required to do for work¬
ers suffering from more conventional disabilities such as paralysis or blind¬
ness. For the Americans with Disabilities Act does not merely forbid dis¬
crimination against disabled workers who can perform to the employer’s
normal expectations (hence who are not really, or at least relevantly, dis¬
abled). It also requires the employer to make “reasonable accommoda¬
tions” to the worker’s disability.^^ In the case of a worker hobbled by age
this conceivably might require the employer to offer him lighter work.

30. As noted in Rutherglen, note 17 above, at 27-28.

31. 42U.S.C. §§ 12101 etseq.
32. S. Rep. No. 116, 101st Cong., 1st Sess. 22 (1989).
33. “Disability” is broadly defined to include “a physical or mental impairment that sub¬
stantially limits one or more of the major life activities of [the disabled] individual ” 42 U S C
§ 12102(2)(A).
34. 42U.S.C. § 12112(b)(5)(A).
Age Discrimination by Employers 339

shorter hours, or lower output quotas, since the Act provides that “ ‘reason¬
able accommodation’ may include .. . job restructuring.”^^ The ADA may
succeed in helping older workers where the ADEA has been ineffective.
But it is too early to tell.

Early-retirement offers. At least one important aspect of the Age Dis¬

crimination in Employment Act has, it might seem, surely been effective
without much litigation; the prohibition of mandatory retirement at fixed
ages. The existence of a policy of mandatory retirement is not concealable,
so it is doubtful that much litigation has been required to extirpate the prac¬
tice. It is noteworthy that there were only 10 mandatory-retirement cases in
my sample and that plaintiffs won 6—a much higher win rate than in any
other category of case. Here, surely, the Act has been efficacious.
Not necessarily. Employers can, without violating the law and often
without incurring heavy other costs, manipulate the age distribution of their
employees through offers of early-retirement benefits.^^ The reasons these
offers are not very expensive are twofold. First, the right to take early re¬
tirement on terms sufficiently advantageous to make the exercise of the
right attractive is a form of employee compensation no different from any
other fringe benefit. Its value is uncertain but so is that of health or life
insurance, the value of which to an individual worker depends on his indi¬
vidual health and longevity, which cannot be known with anything ap¬
proaching certainty in advance. The more munificent the early-retirement
offer, therefore, the less the employer need pay in wages and other benefits.
The creation of a generous early-retirement program funded (in equi¬
librium) by a reduction in the wage level would, it is true, make the em-

35. 42 U.S.C. § 12111(9)(B). See generally Daniel B. Frier, Comment, “Age Discrimination
and the ADA: How the ADA May Be Used to Arm Older Americans against Age Discrimination
by Employers Who Would Otherwise Escape Liability under the ADEA,” 66 Temple Law Review
173 (1993).
36. Michael C. Harper, “Age-Based Exit Incentives, Coercion, and the Prospective Waiver
of ADEA Rights: The Failure of the Older Workers Benefit Protection Act,” 79 Virginia Law
Review 1271, 1278-1279 (1993). For evidence that such offers are effective in bringing down the
average age of the work force, see Laurence J. Kotlikoff and David A. Wise, The Wage Carrot and
the Pension Stick: Retirement Benefits and Labor Force Participation (1989); James H. Stock and
David A. Wise, “Pensions, the Option Value of Work, and Retirement,” 58 Econometrica 1151
(1990); Rebecca A. Luzadis and Olivia S. Mitchell, “Explaining Pension Dynamics,” 26 Journal
of Human Resources 679 (1991); Robert M. Lumsdaine, James H. Stock, and David A. Wise,
“Pension Plan Provisions and Retirement: Men and Women, Medicare, and Models,” in Studies
in the Economics of Aging 183 (David A. Wise, ed., 1994). And for specific evidence that employ¬
ers use early-retirement offers to “get around” legal restrictions on mandatory retirement, see
Edward P. Lazear, “Pensions as Severance Pay,” in Financial Aspects of the United States Pension
System 57, 82-84 (Zvi Bodie and John B. Shoven, eds., 1983).
340 Normative Issues

ployer more attractive to workers who set a high value on leisure relative to
pecuniary income,whereas the employer might prefer workers with a
stronger work ethic. The more employers who adopt such programs, how¬
ever, the less will be the effect on the composition of any given employer’s
work force.
The second factor holding down the net cost to the employer of offer¬
ing early retirement is that the risk to an employee of turning down even a
rather chintzy such offer may be so great that offers of early retirement need
not be princely to induce widespread acceptance. Unless the employee can
prove that the employer’s package of retirement and other benefits is not a
bona fide benefits plan, but is instead designed to evade the statute’s pro¬
hibition against age discrimination—and that is not an easy thing to
prove—the employer can penalize the employee for refusing an offer of
early retirement by offering lower benefits to employees who retire later.^*
The making of an offer of early retirement, moreover, does not commit the
employer to retaining until normal retirement age an employee who turns
down the offer. Herein lies the greatest risk to the employee of turning the
offer down. An offer of early retirement usually reflects a desire by the
employer to reduce the number or average age of its employees, and if not
many employees take up the offer, and even if many do, the employer may
resort to other measures for achieving the desired size and composition of
his work force. The employee knows this, and knows therefore that if he
turns down an offer of early retirement today, he may be fired or laid off
tomorrow. And he knows that this may happen in circumstances in which
it will be impossible for him either to prove age discrimination and thus
obtain compensation, because the employer can demonstrate the business
necessity for his reduction in force, or to find alternative employment at an
equivalent wage, especially if his current wage includes a return to firm-
specific human capital. I have mentioned the plight of the elderly job¬
seeker before and now I add that there is empirical evidence that older
workers remain unemployed longer than younger ones in the wake of a
plant closing or move, even after correction for the fact that they have a

37. For empirical evidence, see Olivia S. Mitchell and Gary S. Fields, “The Economics of
Retirement Behavior,” 2 Journal of Labor Economics 84, 103 (1984).
38. Public Employees Retirement System v. Betts, 492 U.S. 158 (1989). The provision of the
age discrimination law on which Betts was based, which exempted bona fide benefits plans, was
modified by the Older Workers Benefit Protection Act, Pub. L. No. 101-433, 104 Stat. 978 (1990)
(codified at 29 U.S.C. § 621). But employers retain considerable latitude in withholding bene¬
fits to workers who turn down offers of early retirement. Harper, note 36 above, at 1309-1321.
Age Discrimination by Employers 341

higher reservation wage as a result of having pension income to fall

back on.^®
Of course the early retiree may also confront an inhospitable job mar¬
ket. Early “retirement” is a bit of a misnomer. The early-retirement benefits
may be too meager to finance a comfortable retirement. But the employee
who declines the offer and is then discharged not only lacks the cushion of
early-retirement benefits, which might at the least have financed a more
leisurely and productive job search; he also bears the stigma of involuntary
termination.
The risk to employees of turning down an offer of early retirement is
brought out in the following bit of lawyers’ advice to employers:

Most companies conduct RIFs [reductions in force] in two phases.

The first phase is generally a voluntary program in which the com¬
pany offers incentives to induce early retirement or other voluntary
separation by those employees who are not yet eligible for normal
retirement. The second phase is the involuntary termination plan,
focusing upon position elimination, job performance, or some mix
of the two. Management should form oversight committees to en¬
sure that each phase will comply with company policy and appli¬
cable law."^®

So anyone who doesn’t take the hint and retire early becomes a candidate
for phase 2—involuntary termination. And, should this happen, he cannot
count on having a good claim of age discrimination, let alone getting a
good job with another employer. Recall from table 13.1 that only 2 out of
the 84 RIF plaintiffs in the sample won.'*' If the possibility of such an out¬
come is set to one aside, as being negligible, then, in the notation of in¬
equality 4.2, an employee who declines the offer of early retirement may
have to discount Ip, the annual pecuniary income from continuing to work
rather than taking early retirement, by a probability substantially less than
1 that he will actually receive that income, either from his present employer
or from some future employer, in any years between now and the normal

39. Douglas A. Love and William D. Torrence, “The Impact of Worker Age on Unemploy¬
ment and Earnings after Plant Closings,” 44 Journal of Gerontology S\90 (1989).
40. Michael R. Zeller and Michael F. Mooney, “Legally Reducing Work Forces in a Reces¬
sionary Economy,” Human Resources Professional, Spring 1992, pp. 14, 15.
41. For other evidence that victims of RIFs are rarely successful in age discrimination suits,
see Christopher S. Miller, Joan A. Kaspin, and Michael H. Schuster, “The Impact of Performance
Appraisal Methods on Discrimination in Employment Act Cases,” 43 Personnel Psychology 555,
568(1990).
342 Normative Issues

retirement age. And we know that the lowerIp is (or more realistically kip,
where k is the probability of remaining employed and 0 < k < 1), the
greater the incentive to take early retirement.
The analysis suggests that it might actually pay an employer to engage
in outright age discrimination from time to time in order to increase the
incentive of older employees to elect early retirement. The demonstrated
likelihood of such discrimination would increase the number of employees
who accepted the offer of early retirement, or would enable the employer
to achieve his target number of acceptances with a less attractive early-
retirement offer, or would do both. The rationally calculating employer
would trade off this benefit of discrimination against the cost in damages
and other expenses of violating the Age Discrimination in Employment
Act. We have seen that these apparently are small. But of course they may
be small because employers do not follow this strategy, and they may not
because of the costs in the form of a bad reputation that an employer who
treats his workers with such calculated ruthlessness would incur. He might
have to pay for ruthlessness with higher wages.
Even if the employer is law-abiding, and even if he does not announce
a “phase 2” (involuntary termination), the employee who turns down an
offer of early retirement takes a considerable risk. For if not enough em¬
ployees accept the offer, the employer may decide to institute a reduction
in force, though it was not his original intention to do so. And even if the
RIF hits older and younger employees indiscriminately, some of the em¬
ployees who refused offers of early retirement will be among those riffed,
and they will be worse off than they would have been had they accepted the
offer and worse off than younger workers, who will, on average, have better
employment opportunities elsewhere. Knowing this, each older offeree has
an increased incentive to accept the offer. The strategy of making an early
retirement offer followed by a RIF resembles two-tiered tender offers in the
corporate merger market, where shareholders of the target firm compete for
the generous front-end offer so as not to be stuck with the less generous
back-end offer.
Offers of early retirement are not a panacea from the standpoint of the
employer concerned about the age distribution of his employees. To the
extent that the statutory abolition of mandatory retirement was not antici¬
pated, employers were unable to “charge” workers (in the form of lower
wages or other benefits) for the value of the option of taking early retire¬
ment. Also, the employer cannot be certain how many or which workers
will accept the offer of early retirement. If he underestimates the accep¬
tance rate, or if key workers elect early retirement, he may find himself
Age Discrimination by Employers 343

having to incur the costs of rehiring workers to whom he has just paid early-
retirement benefits.'*^ The “which” problem should be separated from the
“how many” problem. Bountiful offers of early retirement might lead to
an undesired reduction in the quality of the employer’s work force. But this
is improbable. It is true that one group of workers likely to jump at such
offers consists of those who have excellent alternative job prospects; any
income from a new job would be in addition to their retirement benefits
from the old one. These will tend to be the best workers—the most adapt¬
able and energetic. The employer can try to hire them back—but his effort
to do so will signal to them their value to him, and they will demand a high
wage. However, this group of workers will be balanced by three other
groups. The first consists of those who are particularly afraid that if they
refuse the offer of early-retirement benefits they will be terminated shortly
anyway, without the benefits; and they will tend to be the poorer workers.
The second group consists of those who attach a high value to leisure; they
will tend to be the less committed and enthusiastic workers. The third
group consists of those elderly workers who for reasons that may be un¬
related to any particular fears of being fired or any unusual demand for
leisure find early retirement a preferable alternative to continuing to work.
While these workers may be competent, we must remember that, by hy¬
pothesis, the employer instituted the early-retirement program in an effort
to reduce the average age of his work force, so he derives a benefit from
inducing a more or less random selection of aging workers (random with
regard to their competence) to leave. In the employer’s eyes, at least—and
I have suggested that there is no reason to suppose them misperceiving—
the result is to increase the average cost-adjusted quality of his work force.
Early-retirement offers appear to be common in many situations in
which the employer is not interested in reducing the average age of his
work force. If it is often efficient wholly apart from any desire to get around
the age discrimination law, this is some indication that the costs of adapting
it for that purpose are not likely to be terribly high.
The alternative to offering early retirement as a method of dealing with
the problem of a superannuated work force is to identify underperforming

42. Frank E. Kuzmits and Lyle Sussman, “Early Retirement or Forced Resignation: Policy
Issues for Downsizing Human Resources,” S.A.M. Advanced Management Journal, Winter 1988,
pp. 28, 31; Mark S. Dichter and Mark A. Trank, “Learning to Manage Reductions-in-Force,”
Management Review, March 1991, pp. 40, 41; William H. Honan, “New Law against Age Bias on
Campus Clogs Academic Pipeline, Critics Say,” New York Times, June 15, 1994, p. B6; see also
Joseph W. Ambash and Thomas Z. Reicher, “Proper Planning Key to Avoiding Discrimination
Suits,” Pension World, May 1991, PP- 14, 15.
344 Normative Issues

workers of any age and fire them. The drawback to this alternative is not
only the risk of an age discrimination suit, but also the cost of making
individual performance evaluations. All the employer may know is that his
work force is “too old”; he may not be able, at tolerable cost, to determine
(at least in advance) which of the older workers is not carrying his load.
This, along with the maintenance of harmonious labor relations, may be
a more important consideration in the substitution of the carrot of early-
retirement offers for the stick of termination than the risk of an age dis¬
crimination suit, since the expected cost of such a suit to the employer may
well be small.

What kind of worker brings an age discrimination suit? As a judge I

have been struck by the number of cases in which a salesman contends that
he was fired because of his age. This impression is confirmed by my study.
Of 388 cases for which the information is available, 38 (9.8 percent) were
brought by salesmen, excluding retail sales clerks.^^ At first glance the large
number of such cases is surprising. Salesmen are paid largely on a piece-
rate basis, so that if their productivity diminishes, whether because of age
or for any other reason, their compensation and hence cost to the employer
automatically falls.'^ But, to recur to an earlier point, if there are fixed costs
of employment, as there are, a reduction in wages will not necessarily com¬
pensate the employer for the reduction in the worker’s output, just as in the
parallel case of part-time work; so also if the salesman has an exclusive
territory, so that if he falls down on the job his employer’s sales will
diminish.
The frequency of cases involving salesmen may be due to two factors.
First, selling is more strenuous than most other white-collar work,"*^ and
while this will induce earlier retirement, it will also induce more discharges
of workers who have not yet reached retirement age. Second, salesmen
often have a large amount of firm-specific human capital (much of it rela¬
tional human capital, as discussed in chapter 3—specifically, a network of
valuable customer contacts). This makes it less likely that they will be fired

43. Graebner, note 2 above, at 44-49, has a fascinating discussion of a wave of hostility to
old salesmen in the period 1900 to 1924, when youthful physical energy was believed to be the
key to successful salesmanship.
44. Laurence J. Kotlikoff and Jagadeesh Gokhale, “Estimating a Firm’s Age-Productivity
Profile Using the Present Value of Workers’ Earnings,” 107 Quarterly Journal of Economics 1215
1236 (1992).
45. For evidence, see Pauline K. Robinson, “Age, Health, and Job Performance,” in Age,
Health, and Employment 63, 69 (James E. Birren, Pauline K. Robinson, and Judy E. Livingston,
eds., 1986).
Age Discrimination by Employers 345

Table 13.2 Distribution of ADEA Plaintiffs, Compared with Work Force

as a Whole

% of Sample % of Work Force

Clerical 22.6 21.7

Manual 16.7 48.6
Professional/Managerial 50.9 23.4
Sales 9.8 6.3

but if they are fired it creates the prospect of very sizable damages, since
their alternative wage is likely to be far below what they were receiving
when they were fired. The second-largest award of damages in the sample
was in fact obtained by a salesman.'^^ In contrast, employees whose princi¬
pal human capital is general rather than specific, so that their alternative
wage is close to the wage they were receiving when fired, or whose wages
are in any event too small for the prospect of an award of lost wages
to warrant the bother and expense of a lawsuit, will rarely show up as
plaintiffs.
Consistent with this conjecture, most age discrimination suits are
brought by professional or managerial employees, who have high salaries;
and most plaintiffs are in their fifties, and so have accumulated a lot of
specific capital and therefore have a substantial damages claim.'*’ In my
study, 234 of 388 cases were brought by professional, managerial, or sales
employees—more than 60 percent. This figure is greatly disproportionate
to the share of these classes of workers in the working population as a
whole, as shown in table 13.2.**

46. By a curious coincidence, this case came up to my court on appeal, and it fell to me to
write the opinion affirming the judgment in the salesman’s favor. EEOC v. G-K-G, Inc., 39 F.3d
740 (7th Cir. 1994). I was not aware at the time that the case was in my sample. I should point out
that many salesmen have, by virtue of their network of eustomer contacts, market-specific capital,
that is, capital they can carry with them to competitors of their present employer, who will often
try to prevent this by requiring his salesmen to sign covenants not to compete, as in the Suess case
cited in chapter 3. See Paul H. Rubin and Peter Shedd, “Human Capital and Covenants Not to
Compete,” 10 Journal of Legal Studies 93 (1981).
47. On both points, see Schuster and Miller, note 16 above, at 68 (tab. 1); Schuster, Kaspin,
and Miller, note 16 above, at iii (59.3 percent of cases filed by managerial and professional em¬
ployees). “The ADEA has become a grievance mechanism primarily utilized by white males and
white-collar workers.” Christopher S. Miller et al., “State Enforcement of Age Discrimination in
Employment Legislation” 18 (unpublished, Syracuse University, School of Management, n.d.).
48. The differences in percentages between the case sample and the population as a whole
are statistically significant at the conventional 5 percent level except with respect to the clerical
workers.
346 Normative Issues

The largest number of plaintiffs in the sample were in the 55 to 59 age

group and the second largest in the 50 to 54 and 60 to 64 groups; only
24.3 percent were younger than 50. Young people would not have accu¬
mulated as much firm-specific human capital and therefore would have
lower damages. Elderly people (only 8.6 percent of the plaintiffs were 65
or older) would be either retired, and therefore no longer protected by a
statute protecting employees (unless they had been involuntarily retired),
or close to retirement, in which event their expected damages would be low,
just as in the case of young workers.
As other studies have shown, blacks and women are underrepresented
as plaintiffs in age discrimination cases."*® One explanation is that blacks
and women don’t “need” to file an age discrimination suit as much as white
males, because they have other civil rights statutes to base a suit on. This is
not a plausible explanation, as it is normally advantageous to sue on as
many tenable claims as one has. A more likely explanation is that blacks
and women tend to have smaller investments in human capital, including
firm-specific human capital, than white males.^° The smaller that invest¬
ment, the less the expected gain from bringing an age discrimination suit.
Given who the plaintiffs are, the Age Employment in Discrimination
Act cannot realistically be characterized as progressive legislation. To the
extent that the employer must factor into his labor costs the expected costs
of damages Judgments or settlements along with all the other costs of com¬
plying with (or violating) the Act, he will pay lower wages. He will try to
lower the wages of those classes of workers most likely to bring and win
age discrimination suits, but a perfect match-up cannot be expected, and
this means that the costs of the Act will be borne in part at least by average
workers. The Act’s effect is thus to redistribute wealth from younger to
older workers, or, after it has been in effect for many years, from the same
workers’ youth to their middle age; and, to a lesser extent, from average
workers to members of the professional and managerial class (including
commissioned salesmen), who account for most age discrimination cases.
These wealth effects are a clue to how statutes like the Age Discrimi¬
nation in Employment Act and ERISA get passed in the first place even

49. See, for example, note 47 above; Rutherglen, note 17 above, at 21-22.1 do not have the
racial identity of the plaintiffs in my sample, but I do have their sex: 24.9 percent were women,
which is only slightly more than half their percentage of the total U.S. work force.
50. For evidence regarding women, see Elizabeth Becker and Cotton M. Lindsay, “Sex Dif¬
ferences in Tenure Profiles: ElYects of Shared Firm-Specific Investment,” 12 Journal of Labor
Economics 98, 107-108 (1994); Elisabeth M. Landes, “Sex-Differences in Wages and Employ¬
ment: A Test of the Specific Capital Hypothesis,” 15 Economic Inquiry 523 (1977). I do not have
evidence regarding firm-specific human capital of blacks.
Age Discrimination by Employers 347

though they are likely to have few effects in the long run other than to raise
labor costs somewhat. They have substantial one-time effects, on the cur¬
rent generation of employers and employees. ERISA transferred substan¬
tial expected wealth from the federal taxpayer to the participants in under¬
funded pension plans, and the Age Discrimination in Employment Act
increased the job security of a number of workers (notably professors, as
we shall see), subject to the qualifications discussed earlier. After employ¬
ers have had time to adjust to the new regime and the current generation of
employees has given way to the next, most of the costs imposed by the laws
will be shifted back to employees, so that workers as a whole, including the
members of the current generation as they age, will on average derive no
benefit from the laws.
Of course to the extent that multiple-selves analysis is valid, a redistri¬
bution of wealth from one’s young self to one’s old self is not the same
thing as moving wealth from one’s left trouser pocket to one’s right pocket.
Nevertheless the welfare of the two selves is closely linked, in much the
same way that the welfare of a man and of a woman is closely linked,
through joint consumption, in marriage. As a result, efforts to prefer old
over young, like efforts through sex discrimination law to prefer female
over male, are unlikely to have substantial net redistributive effects. If sex-
discrimination law increases employers’ labor costs, resulting in lower
wages to male employees, the wives of those male employees will suffer.^'
Similarly, laws that redistribute wealth from young to old harm the old to
the extent that the old self internalizes the welfare of the young self. Also,
the redistribution in question here is from young to middle aged rather than
young to old, and the multiple-selves perspective is less consequential in
the former case.

The effects on the economy summarized. I have emphasized the ex¬

tent to which the goals, at least the ostensible goals, of the Age Discrimi¬
nation in -Employment Act appear to have been subverted as a result of
rational profit-maximizing conduct by employers. But it would be a mis¬
take to conclude that the Act’s elfect on the economy has been negligible
(although probably it has been small), even if costs of litigation and of legal
counseling are ignored. The transition costs, both the costs of gearing up
for compliance with the new law and the costs of revising the terms of

51. Richard A. Posner, “An Economic Analysis of Sex Discrimination Laws,” 56 Univer¬
sity of Chicago Law Review 1311 (1989). There is an offset; husbands of wives who are benefited
by laws against sex discrimination benefit also. But the offset is only partial, because fewer women
than men work and working women are less likely to be married than working men.
348 Normative Issues

employment in order to reestablish the employer’s desired age profile,

should not be ignored just because they are transitional.^^ They can, how¬
ever, be minimized by advance notice of the law. Congress wisely gave
universities eight years’ notice of the abolition of mandatory retirement at
fixed ages, though the universities were, as we shall see, insufficiently wise
to take advantage of this breathing space.
It is possible that the Act, rather than raising the average retirement
age—a stated objective, motivated by the hope of reducing the cost of the
social security program—has, by encouraging offers of early retirement
some of which are accepted by employees whom the employer does not
want to lose, reduced the average retirement age, inducing an inefficient
substitution of leisure for work. Although evidence presented in the next
section of this chapter casts doubt on that particular suggestion, the Act has
undoubtedly caused other labor-market distortions, as we have seen, in¬
cluding discouraging the hiring of elderly workers. Another way in which
the Act may have hurt elderly workers is by discouraging contracts in
which a worker agrees to work for a reduced wage, because of his dimin¬
ished capacity, in lieu of being discharged. Such contracts would not be
common even if there were no law against age discrimination, because of
the fixed costs of employment and other considerations that we encoun¬
tered in the discussion of retirement in chapters 3 and 6. Still, there might
be some. The age discrimination law does not forbid such contracts, but it
makes them unattractive to employers. It is much easier to escape liability
by discharging a worker whose productivity has diminished because of his
age than by attempting to justify paying a lower wage to the elder of two
workers who have the same job.
The law causes another distortion. In chapter 9,1 pointed out that new
firms are likely to have a younger age distribution than old ones, because
they do more hiring and most hires are of young workers; older workers
tend to be locked into their existing employments by their firm-specific
human capital and also to be less adaptable to new situations. This sug¬
gests that the age discrimination law discriminates in favor of new firms
because a smaller proportion of their employees are in the protected
class (which, remember, begins at age 40) and because, as I have pointed
out, the law is far more effective against discrimination in firing than in
hiring. A possibly offsetting factor, however, is economies of scale in

52. Louis Kaplow, though highly critical of transitional relief (including delayed imple¬
mentation), acknowledges that it may be efficient when adjustment costs are high. Kaplow,
“An Economic Analysis of Legal Transitions,” 99 Harvard Law Review 509, 591 n. 251, 592
n. 254(1986).
Age Discrimination by Employers 349

compliance with complex laws. These economies would tend to disfavor

new firms relative to old ones because the former would be on average
smaller.

The Economics of Mandatory Retirement

I want to examine more closely the effects of the 1986 amendment to the
Age Discrimination in Employment Act that abolished mandatory retire¬
ment at any age in most occupations. A mundane but important point is
that most workers want to retire when they reach what has been the normal
retirement age. A few must want to stay on; otherwise firms would never
have imposed mandatory retirement. But many who do want to stay on can
negotiate mutually satisfactory terms with employers for doing so.” And if
not, they can, as we have seen, be gently pried out, even after mandatory
retirement has been abolished, by means of offers of early retirement on
advantageous terms. It has been estimated that prior to the 1978 amend¬
ment to the Age Discrimination in Employment Act that raised the mini¬
mum mandatory retirement age from 65 to 70, only 5 to 10 percent of
retired workers had been retired involuntarily.^"* This may be an overesti¬
mate. For the year following the passage of the amendment saw no inter¬
ruption in the steady downward trend in the percentage of persons 65 and
older employed.^^ In addition to offering the carrot of early retirement, em¬
ployers can wield a stick: they can require their workers to work harder, in
the expectation that the disutility of working harder will fall disproportion¬
ately on the older workers and induce them to retire “voluntarily.”” It
would be very difficult as a practical matter for a worker to challenge such
a tactic successfully under the age discrimination law.
So it would not be surprising if the 1986 amendment had had no effect
except to confer windfalls on the relative handful of workers sufficiently
likely to hang on to induce a sweetening of early-retirement offers. But of

53. See generally Robert L. Kaufman and Seymour Spilerman, “The Age Structures of Oc¬
cupations and Jobs,” 87 American Journal of Sociology 827 (1982).
54. Philip L. Rones, “The Retirement Decision; A Question of Opportunity?” Monthly
Labor Review, Nov. 1980, pp. 14, 15; see also Joseph F. Quinn, Richard V. Burkhauser, and
Daniel A. Myers, Passing the Torch: The Influence of Economic Incentives on Work and Retire¬
ment 85-&1, 199(1990).
55. Rones, note 54 above, at 15-16. A subsequent study confirms that the 1978 amendment,
which raised the minimum mandatory retirement age to 70, had no significant effect on the rate of
participation in the labor force of persons 65 and over. Edward F. Lawlor, “The Impact of Age
Discrimination Legislation on the Labor Force Participation of Aged Men: A Time-Series Analy¬
sis,” 10 Evaluation Review 794 (1986).
56. Rones, note 54 above, at 15.
350 Normative Issues

this we cannot be certain. The downward trend in the labor-force partici¬

pation rate of elderly men did finally bottom out, in 1985, and it has risen
moderately since (see figure 2.7). The rise has been greater for women, but
this means little; one would expect the labor-force participation rate of el¬
derly women to be growing without regard to any legal changes, simply
as a function of the greatly increased participation of women in the labor
force in recent decades. But it is conceivable that the rise in the labor-
participation rate of elderly men has been due, at least in part, to the aboli¬
tion of mandatory retirement, although further study would be necessary in
order to separate out other possible causal factors.
The 1986 amendment may have had a perverse effect on economic
equality. We know from chapter 3 that education steepens the age-income
profile—lower wages during the schooling period being made up by higher
wages later—so earnings peak later for the educated worker and he is
therefore more likely to want to work to an advanced age than a less edu¬
cated worker. This implies that the abolition of mandatory retirement is
likely to benefit mainly the better-educated, who are also likely to be the
abler workers and therefore have higher incomes beyond what is necessary
merely to compensate them for their greater investment in human capital.
This is further evidence that the Age Discrimination in Employment Act is
regressive.^’ But it is regressive only in the short run, neutral in the long,
since in the long run the highly educated will pay for the opportunity of
working longer than their employer would like by accepting lower wages
in their earlier years.
Let us take a closer look at the history and rationale of mandatory
retirement. It did not become common until after World War II.^® Its advent
coincided with the provision of pension benefits on a systematic basis by
large firms to all their long-time employees. The combination of a pension
plan with mandatory retirement protects the worker against a penurious old
age and the employer against having to pay a wage to a worker who is no
longer productive yet may actually be receiving a wage premium in order
to counteract his incentive to slack off in his last period. Even if the worker
is not receiving an incentive wage premium, and even if there is no contrac¬
tual impediment to a reduction in his wage,^'^ age-related decline in capa-

57. See Edward P. Lazear, “Why Is There Mandatory Retirement?” 87 Journal of Political
Economy 1261, 1281-1283 (1979).
58. James H. Schulz, The Economics of Aging, ch. 3 (5th ed. 1992).
59. And there may be, especially if the worker is covered by a collective bargaining agree¬
ment. Unions want employers to have as little discretion as possible to discharge workers. So it is
no surprise that before mandatory retirement was abolished, unionization and mandatory retire-
Age Discrimination by Employers 351

bility may have reduced his productivity to the point where the employer
would be better off replacing him. The employer could discharge the
worker or force him to retire. But in the absence of mandatory retirement
at a fixed age the employer would have to make a costly and (to the worker)
humiliating determination that the worker had ceased to be sufficiently pro¬
ductive to justify retention. Because of its relation to incentive wage pre¬
mia, union protections, and concern with avoiding stigma, mandatory re¬
tirement at fixed ages, rather than being a symptom of the exploitation of
older workers or the operation of a mindless ageism, is correlated with
employment terms and practices that favor older workers.^“
Mandatory retirement may also reduce an employer’s agency costs (the
costs of aligning employees’ incentives with those of the employer). When
involuntary retirement is at the discretion of the employer, employees who
want to work past the normal retirement age will have an incentive to forge
alliances with their supervisors that may undermine the loyalty of both sets
of employees, the supervised and the supervisory, to the employer. This con¬
cern may help explain why the federal civil service was one of the earliest
institutions to adopt mandatory retirement. The goal was to sever the “per¬
sonal ties and informal bonds” that continued to characterize relationships
within the civil service long after the formal abolition of the spoils system.^'

Mandatory Retirement of Judges and Professors

As a former professor and present federal judge, I should perhaps apologize
for devoting a separate section of this chapter to the issue of mandatory
retirement for members of these two famously tenured professions. But I
do not think that it is self-interest alone that makes me think the issue a
fascinating one.
Consider to begin with the highly unusual retirement system for fed¬
eral judges, a system that has never included mandatory retirement. Judg¬
ing is light work that confers substantial nonpecuniary income; this makes
lo in inequality 4.2 strongly positive for most judges, discouraging retire¬
ment. And because the removal of a federal judge is a cumbersome under-

ment were strongly positively correlated. Duane E. Leigh, “Why Is There Mandatory Retirement?
An Empirical Reexamination,” 19 Journal of Human Resources 512, 525 (1984).
60. As not only economists realize. See Carole Haber and Brian Gratton, Old Age and the
Search for Security: An American Social History 108-109 (1994). But see Martin Lyon Levine,
Age Discrimination and the Mandatory Retirement Controversy (1988), a powerfully argued law¬
yer’s brief against mandatory retirement, questioning the economic rationale for the practice.
61. Graebner, note 2 above, at 87.
352 Normative Issues

taking,it might seem all but impossible to get rid of a federal judge inca¬
pacitated by old age. In a sense it is. And yet the problem of age-induced
incapacity in the federal judiciary has not been a serious one. To begin
with, because judging is light work, senility is virtually the only condition
short of death that disables a judge from performing at a satisfactory al¬
though not necessarily distinguished level. Stated otherwise, the curve of
age-related judicial decline is relatively flat, reducing the social gains from
involuntary retirement. Second, through the institution of senior status ex¬
plained in chapter 8, federal judges can retain full salary (and actually in¬
crease their net after-tax income) while carrying a lighter caseload. It is a
deal that few federal judges can resist. And here is the kicker: once a judge
takes senior status, he serves at the pleasure of the judicial council of his
circuit, which can reduce or eliminate the senior judge’s caseload (of course
without affecting his salary) if it decides that he is incapable of performing
at a minimum level of competence. This is one reason why not all judges
take senior status as soon as they can. But most do sooner or later. As a
result the absence of mandatory retirement has not reduced the quality of
federal adjudication substantially and conceivably may even have in¬
creased it, both by making employment as a federal judge a more attractive
career for lawyers who anticipate a vigorous and productive old age and by
eliminating the last-period problem that would arise if judges looked for¬
ward to postjudicial employment. Although like other part-time work se¬
nior status makes it more difficult for the employer to recover fixed labor
costs such as office rent (senior judges do not give up or share offices),
those costs are relatively low; most judges do not yet work with expensive
machinery. A final consideration is that judges who have served for ten
years but have not yet reached the minimum retirement age of 65 are nev¬
ertheless entitled to retire at full pay if they become totally disabled. This

62. Article III of the Constitution entitles judges to remain in office “during good behavior.”
If this means as long as they don’t commit high crimes or misdemeanors, which are the stated
grounds for impeachment, then it is not clear that failing powers alone, not being an ethical flaw,
are a valid ground for removal even by impeachment, let alone by any means short of (and hence
cheaper than) impeachment. See Melissa H. Maxman, “In Defense of the Constitution’s Judicial
Impeachment Standard,” 86 Michigan Law Review 420 (1987). Congress has sought to skirt the
problem by enacting legislation which empowers the judicial council of each of the federal circuits
(the governing committee of judges of the circuit) to certify a judge as disabled and terminate the
assignment of any further cases to him. 28 U.S.C. § 372(c). Since the judge would not be removed
from office, this procedure may not violate Article III, although the issue has never been defini¬
tively resolved. Informal pressures have usually sufficed to bring about the retirement of a judge
physically or mentally disabled because of extreme old age or otherwise. Charles Gardner Geyh,
“Informal Methods of Judicial Discipline,” 142 University of Pennsylvania Law Review 243,
284-285 (1993).
Age Discrimination by Employers 353

fat carrot has made it largely unnecessary to wield the stick of compelled
retirement for disability.
I am not prepared to argue that the senior-judge system is in fact su¬
perior to mandatory judicial retirement at a fixed age. We know from chap¬
ter 8 that senior judges have a lower output than the regular judges. Since
they are paid the same, it follows that they are less productive unless the
average quality of their opinions is higher, which it is not. But the diminu¬
tion in productivity has to be compared with the costs, in a reduction in the
quality of judicial candidates and in last-period problems, of imposing
mandatory retirement on judges. Such a comparison (using citation counts
to proxy quality of judicial output) should be possible, since the state judi¬
ciaries have different tenure provisions, but has not to my knowledge been
undertaken.
Federal judges generally do not view the institution of senior status as
I am doing, that is, as an alternative to mandatory retirement. The prevail¬
ing view in the federal judiciary is that senior judges are working for “free”
(apart from staff and other office costs) since they could retire at full pay.
But nothing in the Constitution requires that judges be allowed to retire
at full pay. The generous retirement provision, like a generous early-
retirement offer, is a method of inducing retirement in a system in which
mandatory retirement is unavailable.^ Its generosity cannot be taken as
given when the issue (however academic, given Article III of the Constitu¬
tion) is whether it would be better to institute mandatory retirement.
The problem discussed earlier of the possible effects of generous re¬
tirement offers on the quality of a work force might seem especially acute
with respect to judges. The ablest might be expected to retire as soon as
possible in order to pursue lucrative opportunities in the private practice of
law, while the worst would have no incentive to retire because they would
not have those opportunities, and they could not as a practical matter be
forced to retire. The critical difference is that federal judicial retirement is
not early retirement. The earliest possible age of retirement is 65, and at

63. See preceding footnote.

64. The fact that the generosity of the judicial retirement system is functional does not mean
that it is generous because it is functional. Judges are a type of civil servant, and federal civil
servants have proved to be an effective interest group, among other things in extracting unusually
generous retirement benefits. Ronald N. Johnson and Gary D. Libecap, The Federal Civil Service
System and the Problem of Bureaucracy: The Economics and Politics of Institutional Change 89-
91, 112, ch. 6 (1994). An additional point is that because public discontent with the remuneration
of public employees focuses on salaries—the costs of information make it difficult for the voting
public to cost out an entire compensation package—legislatures tend to bias the compensation of
these employees toward nonsalary benefits and perquisites, including pensions.
 354 Normative Issues

that age few judges have a stomach for the pressures and risks of practice
when they can retain and exercise their judicial office at full pay yet with
a substantially reduced workload. The qualification (“when they can re¬
tain ...”) is important. Even though few judges will take up the practice of
law at age 65 if they can continue judging on a part-time basis with no
reduction of salary, remove that condition and the number would be
greater, and perhaps large.
Analysis of mandatory retirement for academics is broadly similar to
that of judges, though there are some important differences.^^ Academic
work is light and confers nonpecuniary benefits which normally exceed any
nonpecuniary costs of the work. The combination implies that many aca¬
demics, like many judges, would prefer not to retire at the usual retirement
age. This is especially likely at the elite research universities, where teach¬
ing loads are light. A multivariate study of the age at which academics
retire found that “tenured faculty members in the arts and sciences retire
later when their jobs consist in large part of research, when their teaching
loads are lighter, and when they teach good students” —in other words,
when the nonpecuniary benefits of work are large and the nonpecuniary
costs small. The better the university, the more likely these conditions are
to be satisfied. So it is generally agreed that the abolition of mandatory
retirement will prove to be significant mainly for the relative handful of
elite research universities.^''
The authors of the study were surprised to find that the average age of
retirement was unrelated to whether the college or university had a man¬
datory retirement age.®* They should not have been surprised. The leitmotif
of this chapter is that an employer can use offers of early retirement to bring
about its desired age distribution of employees. Mandatory retirement is
merely one instrument. Yet every private university (as distinct from public
university or public or private college) in the authors’ sample had manda¬
tory retirement. This is not surprising either. Since the net nonpecuniary
benefits of working past the “normal” retirement age tend to be greater in
those institutions (most elite research universities are private, and none,

65. See Epstein, note 1 above, at 459-473, for a highly critical analysis of the abolition of
mandatory retirement for academics; also Honan, note 42 above. For a more favorable view, see
National Re,search Council, Ending Mandatory Retirement for Tenured Faculty: The Conse¬
quences for Higher Education (1991).
66. Albert Rees and Sharon R Smith, Faculty Retirement in the Arts and Sciences 23 (1991).
67. See, for example. National Research Council, note 65 above, at 38.
68. See Reese and Smith, note 66 above, at 22-23.
Age Discrimination by Employers 355

obviously, are colleges), the cost of offers sufficiently lavish to induce most
older faculty to retire early would be greater too.
Of course there would be no reason to try to induce older faculty to
retire unless academic performance tends to decline with age. Nor would
either mandatory retirement or early-retirement incentives be important if
it were feasible to fire declining faculty members for cause. These points
turn out to be related.
Although tenured academics can be removed for cause more easily
than federal judges can be, this is an empty observation. Elite universities
are not much worried about downright incompetents. Distinguished aca¬
demics are not immune from aging; their capabilities, like those of other
people, tend to decline with age. But since distinguished academics are
declining from a high level, they will ordinarily retain enough ability into
their seventies and even eighties to perform passably. In the notation of
earlier chapters, the higher m, the performance peak, is, the longer it will
take, other things being equal, for performance to intersect r, the minimum
capability for the job in question. The university wants its older faculty
members to retire only so that it can hire younger persons who would be
even better because they would be at or approaching their peak years, rather
than past their peak. Not being as good an employee as a potential replace¬
ment would be is not the usual understanding of “cause” for involuntary
termination; to demonstrate “cause” the university must demonstrate in¬
competence, immorality, or insubordination.®® Tenure protection is rarely
(though not never) necessary to prevent the replacement of a better by a
worse employee. Apart from its rare invocation by the politically unpopu¬
lar, tenure protection for academics serves primarily as protection precisely
against being fired when their value in the academic market falls below that
of the young academics clamoring to take their place.
So the power reserved in every tenure contract to fire for cause is not
likely to be of much more help in removing older faculty than impeachment
is in removing older judges. And because fluid intelligence is a larger
component of research capability than of adjudicative capability, the age-
related curve of decline of academics is steeper than that of judges, at least
at the elite universities, where research is emphasized more than teaching.
Also (and again especially at research universities) the relation between
older and younger academics is more competitive than that between older

69. See, for example, Drans v. Providence College, 383 A.2d 1033, 1039 (R.I. 1978); Rob¬
ert Charles Ludolph, “Termination of Faculty Tenure Rights Due to Financial Exigency and Pro¬
gram Discontinuance,” 63 University of Detroit Law Review 609 (1986).
356 Normative Issues

and younger judges; older academics may hang on in order to block the
appointment or advancement of youthful challengers. This is an argument
against delegating the power of faculty appointments to faculty. We can
expect a reduction in that delegation as one response to the abolition of
mandatory retirement.
We need to know how steep the age-related decline of faculty perfor¬
mance is, for that is the key to how costly the abolition of mandatory aca¬
demic retirement is likely to prove. The profile of course differs by field,
paralleling the differences in age profiles of creativity that I discussed in
chapter 7. The multivariate study mentioned earlier confirmed that there is
an age-related decline in academic research,^° but noted that since the de¬
cline begins at age 35 and is gradual, the decline between ages 65 and 70
is slight. The authors found this reassuring. They should not have. A slight
decline in performance between 65 and 70 is consistent with a steep decline
between 71 and 75, or 71 and 80. The study sample did not include enough
professors over 70 to venture estimates of the performance of this group. It
would be question-begging to extrapolate from the performance of younger
faculty to that of older faculty.
Like other employers, universities can obtain or at least approximate
their preferred age profile of employees by offering early retirement on
attractive terms. The judiciary lacks this flexibility under existing law. Se¬
nior status and the rules governing social security taxes and entitlements
provide some inducement to judicial retirement; universities could offer
their faculty members more. The long-run cost to universities, as to other
employers, need not be high, since the prospect of early retirement on gen¬
erous terms will reduce the wage demands of young faculty. The universi¬
ties had eight years in which to prepare for the abolition of mandatory
retirement and could during that time have reduced wages or benefits (or,
for tenured faculty, the rate of increase in wages or benefits) to fund the
option of continued employment that the law would require them to offer
their faculty members beginning in 1994. I am not aware that any univer¬
sities did this, even though young faculty members are (and presumably
were then) well aware that the law gave them a valuable option,^' and might
be expected to realize they might have to pay for it.

70. Reese and Smith, note 66 above, at 70 (fig. 4-3). To similar effect, see Alan E. Bayer
and Jeffrey E. Dutton, “Career Age and Research-Professional Activities of Academic Scientists:
Tests of Alternative Nonlinear Models and Some Implications for Higher Education Faculty Poli¬
cies,” 49 Journal of Higher Education 259 (1977).
71. Honan, note 42 above, quotes a 33-year-old assistant professor of history as saying “I’m
glad about the change in the law, because I myself may not wish to retire when I’m 70.”
Age Discrimination by Employers 357

Because of lack of preparation, universities may, like other employers

forced by the Age Discrimination in Employment Act to abandon manda¬
tory retirement at fixed ages, experience a one-time loss measured by the
early-retirement benefits that they are paying to faculty members whose
wage-and-benefit package did not, when negotiated, reflect the expected
value of such benefits. This loss may be greater than that incurred by most
other employers affected by the Act’s successive waves of lifting (and fi¬
nally removing altogether) the minimum mandatory retirement age, since
tenure makes it difficult for a university to hold an implied threat of termi¬
nation over the heads of its faculty employees. One effect of the abolition
of mandatory retirement for professors may be, therefore, a movement to
abolish academic tenure in order to reduce the cost of early-retirement of¬
fers—and also to counteract the potentially serious effect of the abolition
of mandatory retirement on the quality of faculty at elite universities. Re¬
call that early-retirement offers are likely to be snapped up by both the best
and the worst workers—but by the latter only because there is an implied
threat of termination if they refuse the offer, and that threat is weak in the
case of tenured professors. Only the best professors have strong incentives
to take early retirement. The best older professors tend to be those who
remain active in research and may therefore welcome retirement as an op¬
portunity to allocate even more time to research; who are not jealous of the
young and therefore do not fear displacement by them; who are more likely
to be scrupulous and self-aware and therefore to worry about overstay¬
ing their welcome; and who have attractive opportunities for employment
elsewhere. The worst older professors do not engage in research any more;
they may not enjoy teaching, but they are not likely to have heavy teach¬
ing responsibilities, because students will avoid their classes; and they
may enjoy faculty intrigue. They may be jealous and fearful of the com¬
petition of the young and insensitive to their own decline. And they will
have no good opportunities for employment elsewhere. Hence the aboli¬
tion of mandatory retirement for academics may, though probably only at
elite universities, reduce the average quality of the faculty unless tenure is
abolished.
The prohibition against mandatory retirement of federal judges poses
a less serious social problem. This is not only because most judges who
decide not to retire are capable of creditable performance of their duties at
what in other occupations would be considered bizarrely advanced ages,
and not only because mandatory retirement would have adverse effects on
the selection and incentives of judges, but also because there is no danger
of old judges’ blocking the advancement of the young. Judges do not ap-
358 Normative Issues

point, and rarely even influence significantly the appointment of, their
successors.

It is time to wind up. There are polar positions on the situation of elderly
people in the United States of today. One is that they are—or would be,
were it not for extensive government transfer and regulatory programs—
an oppressed class no different from blacks or women or homosexuals; are
or would be, that is, victims of a pervasive “ageism” just as vicious and
irrational as racism, sexism, or homophobia are thought to be; victims of
false stereotypes concerning the mental and physical capacities of elderly
people that make them despised by the young and by themselves and that
exclude them from the workplace. The opposite position is that the elderly
are pampered parasites, denying the reality of aging, selfishly employing
their disproportionate political power to siphon the wealth of the country
into the support of their ever more clamorous demands for generous pen¬
sions and extravagant medical care, dooming the country to gerontocratic
stagnation and mediocrity.The discussion in the chapters in the closing
part of this book, and in the earlier parts as well, suggests that neither ex¬
treme position is tenable. The evidence that there really is a process called
aging that takes its toll of everyone, albeit at different rates, generating
palpable and often occupationally relevant physical and mental differences
between older and younger persons, is more compelling than any evidence
thus far advanced to demonstrate occupationally relevant differences in the
fundamental capacities of men and women, whites and blacks, or persons
who differ in their sexual orientation. And while, as I have continually
emphasized, the young self and the old self may for some purposes be
different persons, they are connected by bonds of altruism and personal
identity far closer than those that link the dominant and subordinate partic¬
ipants in other relations argued to be discriminatory. So it is implausible
(and for the further reason that there appears to be a genetic basis for the
respect and affection that most people accord to the elderly members of
their families) either to cast young and old in an “us-them” opposition that
might explain discrimination against elderly people, or to infer discrimi¬
nation from policies such as mandatory retirement that treat elderly people
differently and superficially—but only superficially—less favorably than

72. For responsible statements of the respective polar positions, compare Howard Eglit,
“Health Care Allocation for the Elderly: Age Discrimination by Another Name?” 26 Houston
Law Review 813 (1989), with Jan Ellen Rein, “Preserving Dignity and Self-Determination of the
Elderly in the Face of Competing Interests and Grim Alternatives: A Proposal for Statutory Re¬
focus and Reform,” 60 George Washington Law Review 1818 (1992).
Age Discrimination by Employers 359

young people. In many fields of endeavor, especially highly creative fields,

where fluid intelligence is more important to success than crystallized in¬
telligence, the declension of ability with age is very steep. It is sentimental
to pretend otherwise, or to treat rare exceptions as stating the rule.
Many of the prevalent misunderstandings about aging and old age re¬
sult from simple methodological errors, above all a failure to correct for
selection or retention bias. Studies that show that elderly workers are as
good as younger ones, thus making the frequent preference of employers
for young over old workers seem irrational—“ageist”—overlook the fact
that the employer will have gotten rid of those elderly employees who
could not perform up to snuff. The ones whom the employer retains will
therefore not be representative of the average abilities of their age cohort.
The broader point is that failure to correct for selection bias can cause ob¬
servers to exaggerate the capabilities of the old, and thus see discrimination
where there is none, by mistaking the exceptional members of an age co¬
hort for its average members.
So they are not a victim class, our old people; and we saw that they
were not victims even before the modem era of age-friendly social welfare
legislation that began with the enactment of the Social Security Act in
1935. Practices that appear to discriminate against the elderly, such as man¬
datory retirement at fixed ages or late vesting of pensions—practices that
the law has taken upon itself to change, though apparently with little ef¬
fect—have strong efficiency justifications and little or no tincture of injus¬
tice. Today at any rate, in this country at any rate, the elderly are on average
highly prosperous. Which is not to deny the existence of false, denigrating
stereotypes about the elderly, such as that they are sexless, or that they are
a terror on the roads. I have been at pains throughout this book to dispel
such stereotypes. Yet the more myths I puncture, the more I may seem to
lend credence to the claim that the elderly are indeed victims if not of dis¬
criminatory practices then at least of discriminatory attitudes, much like
Jews or Asians in this country. But misunderstandings about groups to
which one does not belong and with which one may not have much face-
to-face contact are pervasive, and are an unmysterious product of the fact
that the costs of information are positive and often high. It is difficult to see
how or where current misunderstandings about old people are hurting
them. Employers have a real stake in understanding the capabilities of em¬
ployees and potential employees, and there is no convincing evidence that
before the flood of protective legislation employers systematically under¬
valued older workers or exploited or otherwise mistreated them, for ex¬
ample with regard to the late vesting of pension rights; I do not of course
360 Normative Issues

deny the existence of isolated errors and abuses. A neglected point is that
costs of information about people may lead to over- as well as under¬
estimation of qualities. It is possible, for genetic reasons discussed in
chapter 9, that people exaggerate the wisdom and other good qualities con¬
ventionally associated with elderly people, and that they ascribe the char¬
acteristic good qualities of the elderly—in a reversal of statistical discrimi¬
nation and through a neglect of selection bias—to elderly people who do
not possess them.’^
Yet it is probably true that old people in the United States of the present
day do not command the respect and affection they once did. The fact that
they are materially and in point of health better off on average than they
ever were has its underside: they are less appealing objects of charity and
solicitude. Stated otherwise and more positively, loss of popularity is the
price that elderly Americans pay, probably willingly in most cases, for the
dramatic increase in their prosperity and political influence. Demographic
changes—a falling birth rate and a falling death rate, the latter due to the
higher incomes of the old but above all to the advance of medical tech¬
nology—have greatly increased both the relative and the absolute size of
the elderly population. They are less scarce, so less valued. Most important
perhaps are social changes, including mass education and the increasing
rapidity of social, economic, and technological change—the increasing dy¬
namism of American society—that have reduced the social value of the
memories, wisdom, and experience of the elderly.
But we must bear steadily in mind that if not as valued as they once
were, the elderly in America are a lot wealthier. When they were less
wealthy, they had more respect; as they became wealthier, respect for them
declined. The keel remains even. The elderly in America are not and never
were a vulnerable, pariah class; but neither were or are they a class of par¬
asites. They do vote more and thus have more political power than other
age groups, yet it has proved feasible to curtail transfers to the elderly,
provided that a long transition period is allowed. There is a natural though
not necessarily an optimal equilibrium between the demands of the old for
support and the willingness of the young to supply that demand. The hun¬
dreds of billions of dollars in transfers to the old that are carried on the
books of the federal government are an imposing but misleading emblem
of generational redistribution; they are not net figures. In part they replace

73. For evidence of favorable stereotypes of old people in our society, see Mary Lee Hum-
mert et al., “Stereotypes of the Elderly Held by Young, Middle-Aged, and Elderly Adults,” 49
Journal of Gerontology P240 (1994).
Age Discrimination by Employers 361

transfers that young people would make voluntarily to their old selves
through greater savings for retirement. In part they compensate the old for
the expense of educating their children’s generation and commute into cash
what would otherwise be a heavy burden borne by adult children of tending
their aged parents. In part they reflect an irresolvable tension between the
claims of the young self and the old self within each of us. In part they
simply reflect a falling birth rate and a falling death rate, the combined
effect of which has been to shift the center of gravity of the dependent
population from the dependent young to the dependent old without actually
increasing net dependency, though the maintenance of an even keel is due
in part to the increased labor-force participation rate of women, which has
both increased the number of job holders and, indirectly, reduced the num¬
ber of children.
The huge expenditures on medical care for elderly people, and for
medical research that is increasing the longevity of the elderly, are not
wasted merely because the elderly have, at best, relatively few years of
remaining life. They dread death as much as the young. The subsidization
of those expenditures through the Medicare program can be criticized,
however, as giving the elderly more medical care than they “really” want;
a subsidy earmarked for a particular service, as distinct from a cash sub¬
sidy, which the recipient can use for anything he wants, is a recipe for
overuse. As this example shows, even if the direst predictions about the
effects of our old-age laws on the economy are unfounded, as I believe they
are, all is not right with our policies toward the aged. Among many ex¬
amples, the federal pension law (ERISA) appears to have been a response
to largely nonexistent problems and its costs of administration, though
modest in the overall scheme of things, appear to buy no benefits. The Age
Discrimination in Employment Act is a particularly misbegotten venture in
tilting at the windmills of ageism, and not only because most elderly people
are retired and therefore are outside the scope of the Act. The Act may have
few long-term effects of any sort. Its commands are readily avoidable,
probably at modest long-run cost, by offers of early retirement (which are
self-financing in the sense that, like other fringe benefits, they are ulti¬
mately paid for by the workers themselves in the form of lower wages); by
avoidance, at what appears to be only a slight legal risk, of hiring elderly
workers; and by matching (as through a carefully designed reduction in
force) the termination of unwanted older workers with the termination of
some young ones in whose firm-specific human capital the employer has
not yet invested heavily. This means that the Act does little good for the
aged (and some harm to them) and little harm to the rest of the population.
 362 Normative Issues

except that its abolition of mandatory retirement probably has hurt the elite
universities. Yet this particular wound is in part a self-inflicted one because
the universities failed to take advantage of the opportunity that a generous
transition period gave them to shift the costs back to the faculty. The beauty
of early-retirement offers, from the employer’s standpoint, as a method of
circumventing the age discrimination law is that, in the long run, he can
probably make the offerees pay for their own early-retirement benefits,
when they are young.
I have said that to describe the elderly of our society either as victims
or as exploiters is inaccurate. The multiple-selves perspective that I have so
stressed invites consideration of a third possibility: that the prolongation of
life has placed the unity of the self under increasing pressure. The years
that are added at the end of life are years of diminished capability, and
generally of impaired health and diminished utility as well. Young people
know this, and may not want to make generous provision for those future
years, just as they may not want to make generous provision for the contin¬
gency of becoming severely disabled, an analogous case. But when they
become old, their point of view changes. They want to live (not all of them,
but most). Not only or even mainly because of their political power, society
will not abandon them, even if the costs of caring for them and attending
to their medical needs are very high. The financing of those costs presents
a policy dilemma, since the young will resist being forced to pay either in
taxes or in compelled savings for the level of care that their old selves will
demand or that the current old are demanding. Multiple-selves analysis
does not point the way to a solution to this dilemma; on the contrary, it
shows why a solution will be difficult to devise.
The diminished utility of an ever more protracted old age is a concern
in its own right. All the hard knocks that utilitarianism has taken from phi¬
losophers have not much diminished the importance that the average per¬
son, and indeed the above-average and even the philosophical person, at¬
taches to happiness. If old age were happier, the tension between the young
self and the old self would be less, so it would be easier to finance old age.
It is possible, though not certain, that we can make old age happier, by
some reallocation of medical resources from the diseases of old women to
the diseases of old men, and from lethal to nonlethal diseases of the elderly.
We are in danger of losing sight of the fact that the costs associated
with adding years to the end of life—“low-value” and high-medical-cost
years for most people—are a by-product of improvements in income, pre¬
ventive medicine, and medical treatments that have shifted the boundary
between middle age and old age, in effect extending the former. Old age
Age Discrimination by Employers 363

begins later and lasts longer; the “begins later” means that low-value, high-
medical-cost years are being transformed into high-value, low-medical-
cost years at the same time that years of the first sort are being tacked on at
the end of life. When all costs and benefits are reckoned in, there is no solid
basis for concluding that the aging of the population has been or in the
foreseeable future will be a source of net diminution in the overall welfare
of the American people.
So I end on an optimistic note. But social prophecy has not been the
principal goal of the book. I have mainly tried to formulate an economic
theory of human aging and old age, building upon the seminal contribu¬
tions of the human-capital economists but drawing on other fields both in¬
side and outside of economics, ranging from imaginative literature and phi¬
losophy both ancient and modern to cognitive psychology, medicine, and
evolutionary biology. This theory employs a broader conception of aging
than the one that human-capital economists employ. It is a conception that
emphasizes the process of aging, that provides a simple but realistic frame¬
work for understanding a variety of age-related attitudes and behaviors, and
that enlarges the domain of rational choice to include the choices that we
make for and in old age, including, for some of us, the choice of when to
die. Such at least has been my ambition. I acknowledge the possibility that
the vastness of the subject, the rapidity with which the relevant social, eco¬
nomic, and medical context is changing, and my own limitations of knowl¬
edge, technique, and time have caused my reach to exceed my grasp. If so,
take this book not as arrival but as embarkation, or as a glimpse of the
promise of multidisciplinary inquiry guided by economics, or even merely
as an invitation to meet the challenges of modernity—of which our rapidly
aging population is one—with better social science.
 V.:
, i- f - '■ • V^'
■, If
’ '■ ■ ^ .,, ^ ,„ ... ■.-. , I •
f^Vi
» .■ y
^ .. , ^ ■■ ^/■'V^,

'/ . •; '*< «> VfjfJil.;; V'*' ■i^.,.-,af j^;^^.i^;y •,

i^-; \Y5«?'>\-:,. /'i . v- '->.»i,-;.7vi,

A''^^;^v|jV-:!<(, •;^•■ /'^■ • ;^;v < i**’

1
-.« ' ■'•■^ ^.%vx^y''i ■ J
,
: -/cr^-. .'.t •>
. V '^ : 'vs;, ft; : -l"'•':r■.w •-•■.■,»*;'• f
■■ I'jro t . i.4(w;^ ■ • • '■i, -
■ [■>'"

L’ - -i ' • .. , -;: v/USldtf:55-;i • • V... ■.;, -* -

luf /_-<iAi. ^ ih .. • 4 »V * •* « ’■- .-f^* . » — . *1 ■ '*M

’ . • • ’m

>. t.. V 'fS . . w . - 1. .

^A^f>v--_._^t. -j. ;\v .; ^I^ifv »)*’- .|?jtf \;^,'..^-li,0l,i ^1. .1 '. 1-
1''^^*.^.^ 1 j .*' rt'cfn«< l/i’ ■ if.'''^,V^
■ ■'•• ■
■•J I'i-t'; !b ■»,;;' \ }(•, 1': ■■,■' I' ' ,-.
' - v..,. :'H',-f*( V ♦'» ‘
.1,. ■•. tT
-ar. >,
^ iy. -i ■ \u \ 'v/*;‘:2^*'" ■::^'
‘iji ^ .1^^.'. j.vt . o'* ^/ “v■'^ ■■ IT''» '‘'^' •14
''. .J, i> !J-.- ^ ■■-■ O
.- i.ivj ■
'J„7j: V.,
V *»

■ *J J ^ J;i
V' I
,,,v-
*t'^ ^ ^ >'t.<'N*.» *
.,

.\ V'.
_ .J '-**^'' \4-, .r,-: wj(:l of h^i , 1 -f'. •’
!iVf’ f • '■ *'^^^i'•"'•.*k^ V - ' ^. J^f^v^;'; • , !, •, ‘ ;vi.j^, ..
\t ^ ’ '■•’? .i’ V'^-jT.ifisv’- ,,. ct ,• g<:: ,ri.

.. ■ *>■ tUwm." ,«iirf^ ■ s ■■'

.1 /;WHfc>.' ':i,h.Vl^ . '.l- i'- ..- 'Pr ..-..A'' ..K* • '. .'ililtw
 Index

Accidents and accident avoidance, 111 n, vate employers, 334-335, 351; hir¬
115-116, 127; accident rate of el¬ ing cases, 329-333; incidence of
derly drivers, 122-127; care versus costs of compliance with, 346-347;
activity-level responses, 123-124, percentage of elderly plaintiffs, 328;
126, 306; seatbelt use by elderly, politics of, 346-347; redistributive
307. See also Driving, Tort law effects of, 346-347, 350; reduction
Addiction, 116. See also Habit in force (RIF) cases, 331, 336, 341-
Adjudication. See Judges 342; “smoking-gun” evidence,
Adoption, adult, 209 335-336; who sues, by age and oc¬
Age discrimination, 7, 319-357; “ani¬ cupation, 345-346
mus,” 320-322; statistical, 322- Age grading, 7, 62n; as method of sort¬
328, 359-360. See also Age Dis¬ ing workers to jobs, 208; in modern
crimination in Employment Act, society, 219, 322-324
Social status of elderly Age-cohort effects, 136n, 151
Age Discrimination in Employment Act, Age-earnings profile, 52-55
319-357, 361; abolition of manda¬ Ageism, 20, 204, 320, 351, 359. See also
tory retirement by, 319, 349-357; Age discrimination. Social status of
age versus salary discrimination, elderly
336-337; aggregate economic ef¬ Agency costs, 351. See also Efficiency
fects of, 347-349; and Americans wage
with Disabilities Act, 338-339; Aging, of animals, 26; defined, 18; evo¬
asymmetric stakes of parties, 337- lutionary theories of, 25-30, 74-
338; awards of attorneys’ fees un¬ 75, 202-203; male versus female,
der, 331-332, 338; damages in 274; normal, 8, 17-24; of organiza¬
cases under, 329-335; effects of, tions, 227-229. See also Psychol¬
328-357; differential effects of on ogy of aging
large versus small firms, 348-349; Agricultural societies, status of elderly
discrimination by public versus pri¬ in, 210-218

365
366 Index

AIDS, 57, 270, 275 “Blind” refereeing, 176

Airline pilots, age of retirement of, 323, Boredom, 8, 18, 22, 56, 80-82, 292; ju¬
326 dicial, 196,201
Altersstil, 174-177 Brock, Dan W., 256-257, 266n, 267-
Altruism, 47-61 passim, 90, 118, 131, 268, 285-286
134, 203, 205, 210, 286, 288; two- Burnout, 80-81; of criminals, 132
sided, 209n
Alzheimer’s disease, 21. See also Senile Calabresi, Guido, 182
dementia Capability, excess, 8, 25, 29, 73-75, 80,
Americans with Disabilities Act (ADA), 94-95
338-339 Capital punishment, 314
Analogy. See Reasoning by analogy Cardozo, Benjamin N., 181-182
Ancestor worship, 212 Centenarians, number of, 35
Ancient societies, old people in, 31-32. Charivari, 225
See also Aristotle; Social status of Chess, age profile of creativity in, 159-
elderly, in ancient Greece and Rome 161
Annuity, 61, 287, 300 Children, as caretakers of elderly, see
Apprenticeship, 70 Family, care of aged persons by
Architecture, age profile of creativity in, family members; legal obligation of
166-167 to support elderly parents, 222; vir¬
Aristotle, 115, 164n, 176; on old age, 6- tual representation of, by parents,
7, 101 -109, 119. See also Knowl¬ 151; lack of voting power of, 289.
edge shift See also Altruism
Article III of U.S. Constitution. See China, 209n, 210, 226
Constitution Christianity, effect of on status of elderly,
Athletics, 4n, 7, 18 -19, 77, 159-160,228 224-225. See also Religion
Chuma, Hiroyuki, 4, 11 In
Becker, Gary S., 13, 56-57, 59n, 92-93, Cicero, 12, lOOn, 119n
134,284, 288n Citation analysis of judges, 182-192, 354
Behavioral slowing versus adaptive slow¬ Clemenceau, Georges, 113
ing, 115-116 Coase, Ronald, 166
Bequest motive. See Altruism, Bequests Cognitive dissonance, 118
Bequests, 60-61, 131, 224; negative, Cohen, Lloyd, 1 lOn, 11 In
286; effect of suicide in accelerat¬ Cohort effects. See Age-cohort effects
ing, 134; taxation of, 90, 224 Coleman, James, 64
Biology, essentialism, 274; sex differ¬ Communication. See Conversation
ences in biology of aging, 274. See Compassion in Dying v. Washington, 239
also Aging; Genetics, and old age Compensation, backloading of, 59, 133;
Biomedical ethics, 264 backloading of judicial, 193; of civil
Bismarck, Otto, 7 servants, 353; basic economics of,
Blacks, as plaintiffs in cases under Age 59; role of pensions in, 301-303;
Discrimination in Employment Act, piece-rate system, 344; of salesmen,
346; discrimination against, 320; 344; in wrongful-death cases, 109-
life expectancy of, 222; medical re¬ 110. See also Human capital.
search on diseases of, 278; respect Pensions
of for elders, 222-223; retirement Complementarity of male and female
of, 80 longevity, 277
Index 367

Confucian attitude toward elders, 209n, Depression, 105n, 135, 137, 240; depres¬
226 sive realism, 105n
Conservatism of elderly, 116, 120, 151, Diamond, Arthur M., Jr., 4, 172n
195-196 Disability, 46-47; age as, 338-339;
Constitution, provision for judicial tenure judicial, 352
in Article III of, 181,351-353 Discount rates, 70-72, 91-92, 158, 291;
Consumer behavior of elderly, 107n of criminals, 313-314; evolutionary
Contracts, cost of enforcing, 212; to sup¬ theory of, 71; bearing of on decision
port elderly persons, 209-210; of to retire, 79; in young selfish versus
suicide, 254-261 old selfish societies, 226; social ver¬
Conversation, considerate versus incon¬ sus private, 291, 313 - 314
siderate, 108; inconsiderate, 9, 57, Discrimination, 333; nonage cases, 336n;
120, 175 by public versus private employers,
Coresidence (of family members of dif¬ 334-335, 351; on grounds of sex,
ferent generations), 145-146, 147n, 347; statistical, 205, 322-328; un¬
219 wanted association, 334. See also
Crime, career criminals, 132-133; el¬ Age discrimination. Age Discrimi¬
derly as victims of, 126-128, 279; nation in Employment Act, Ageism
elderly offenders, 128-133; fines Disease, defined, 17-18
versus prison as modes of punish¬ Disinterest, 118-120, 193-194
ment of elderly, 311; and mental Distribution of income and wealth. See
competence, 315-316; number of Income, of elderly; Social security,
elderly prisoners, 312-315; punish¬ redistributive effect of
ment of elderly offenders, 90, 311- Dover, Kenneth, lOOn, 176
314; punishment of murderers of el¬ Down’s (or Down) syndrome, 21
derly, 279 Driving, 122-126. See also Accidents
Cruzan v. Director, Missouri Dept, of and accident avoidance
Health, 239n Dworkin, Ronald, 266-268
Custom, 212
Economics, age profile of creativity in,
Damages, hedonic, 308. See also under 159, 166; age profile of productivity
Age Discrimination in Employment in mathematical versus applied,
Act, Tort law 173-174
Daniels, Norman, 267 Education, 52-53; elderly people’s pre¬
de Kooning, William, 176n ferences concerning, 86; effect of on
De Senectute. See Cicero status of elderly, 218-219. See also
Death, dread of, 108-113, 205, 271. See Literacy
also Value of life EEOCv. G-K-G, Inc., 345n
Deathbed conversions. See Utility, Efficiency wage, 59, 73-74, 337; rela¬
posthumous tion of to mandatory retirement,
Defined-benefit pension plans. See under 350-351
Pensions Effort, economics of, 74, 80, 95
Dementia. See Senile dementia Ehrlich, Isaac, 4, 11 In
Democracy, 220. See also Social secu¬ Elder abuse, 127n
rity, politics of; Voting Elder care, 293-294; to what extent truly
Dependency ratio, 39-40, 43, 56, 206, “voluntary,” 286-287. See also
210,219,283-284 Home care of aged
368 Index

Eliot, T. S., 100-101 Friendly, Henry J., 182, 199

Elster, Jon, 85 Friendship, 61-65
Employee Retirement Income Security
Act (ERISA). See ERISA Gardner, Howard, 162n, 163n
Engineering, age profile of creativity in, Gender, as issue in medical research,
166-167 273-278; “sex-role turnover”
Epstein, Richard A., 354n (male-female convergence with
Equal Employment Opportunity Com¬ age), 114, 208; and suicide rate,
mission (EEOC), 330 137. See also Masculinity, Sex,
ERISA, 299-305, 319, 361; main provi¬ Women
sions and consequences of, 303 - Genetics, and discount rates, 71; and
304; politics of, 346-347 dread of death, 109; and the family,
Ethics. See Biomedical ethics 202-203; and old age, 8, 25-30,
Euthanasia, 235-261, 279; definitions 109, 139, 165, 202-203, 205, 264;
of, 235-236; voluntary versus in¬ and optimism, 105; and elderly
voluntary, 235 sexuality, 139
Evolution. See Aging, evolutionary theo¬ Geronticide, 203-205, 236n. See also
ries of; Genetics Euthanasia
Experience, 106, 162, 164-165; impor¬ Gomez, Carlos F., 242-243, 247n
tance of to judges, 183-184, 193-
194; effect of judicial, 197; lived, Habit, 81-82, 116
64, 69-70, 107; effect of on opti¬ Hand, Learned, 175n, 181, 188-192,
mism, 104-106. See also Intelli¬ 197-199, 201
gence, crystallized; Learning by do¬ Hand Formula, 198-199
ing; Wisdom of elderly Hardy, G. H., 159n
Externality, pecuniary, 265 Hazen Paper Co. v. Biggins, 337n
Health care, status of benefits for workers
Fair, Ray C., 4n under Age Discrimination in Em¬
Fame, 175. See also Reputation ployment Act, 329n; costs of, 36,
Family, care of aged persons by family 43-44; durable power of attorney
members, 145-147, 202-203, for, 258-259. See also Medical in¬
208-209, 220-222, 285-287; eco¬ surance, Medical research. Medicare
nomic theory of, 202, 208-210, History, age profile of creativity in, 159,
221; evolutionary theory of, 202- 167
203, 272; determinants of size of, Hitler, support for among elderly, 292-
221-222, 286; relation of size of to 293
elder care, 286; suicide as utility- Hochschild, Arlie Russell, 62n, 322n
maximizing decision of, 240; and Holmes, Oliver Wendell, Jr., 113, 175,
physician-assisted suicide, 260-261 178, 181-182, 187, 198-199, 220,
Feminism, 274, 277 223
Filial obligation. See Filiality Home care of aged, 47, 147, 285-287,
Filiality, 203, 208-209, 220-222, 286. 293-294
See also Family; Genetics, and old Homosexuality, 27, 139-140
age Hospice movement, 243
Fischer, David Hacked, 225-227, 281 Household production. See Nonmarket
Fisher, M. F. K., 9n output
Frankfurter, Felix, 162n, 182, 198 Human capital, amount required in differ-
Index 369

ent fields, 166; basic theory of, 6-7, Intergenerational contracts, 209-210
22, 51-58, 350; of blacks and Intergenerational equity, 284-294 passim
women, 346; in creative versus ad¬ Intergenerational transfers, 47-48, 211 -
ministrative jobs, 157-158; of 212, 282-294 passim. See also
criminals, 313; in criminal occupa¬ Bequests
tions, 130, 132n; defined, 7, 51; de¬ Intragenerational transfers, 282
preciation of, 53-55, 66-67, 77- Investment strategy of elderly, 117
78, 142, 170-171,212; effect of IQ, 317, 324-325
discrimination on investment in,
327-328; general, 52, 333, 345; ju¬ Japan, 257n, 286n, 290
dicial, 197; market-specific, 345n; Judges, 119; age at appointment, 193;
purchased by school taxes, 284; re¬ age of in Anglo-American versus
lational, 8, 61-65, 165, 219, 344; Continental judicial systems, 194;
social, 64; specific, 52, 75, 142- citation analysis of, 182-192; el¬
143, 227, 301-302, 304, 321, 329- derly, 2-3, 180-201, 324-325,
330, 333-336, 345; of uneducated 351- 358, see also Supreme Court;
workers, 143; and worker turnover, in primitive societies, 208; qualities
336; of young versus older workers, of best, 198-199; retirement of,
336,346 181-201 passim, 351-358; senile,
352- 353; senior status, 185-189,
Imagination, 68-69; “imagination capi¬ 193, 351-354; trial versus appel¬
tal,” 92-93. See also Intelligence, late, 186, 192-193
fluid Jury service by elderly, 152, 154n, 155n
Immigration, 43; age of, 69-70 Justice between generations. See Mul¬
Inclusive fitness, defined, 27 tiple selves
Income, of elderly, 41-42, 281-294 pas¬
sim; and health, 46-47; and longev¬ Kaplow, Louis, 348
ity, 42, 47. See also Social status of Kass, Leon, 56
elderly. Wealth Kilburn, M. Rebecca, 134
Infanticide, 202n, 203, 256-258 King Lear, 211, 239
Inflation, 292, 300 Kinsey, Alfred C., 138
Information, costs of, 208, 326, 359. See Knowledge, tacit, 163
also Discrimination, statistical; Ex¬ Knowledge shift, 6, 17, 18, 22-23, 67-
perience; Psychology of aging 71, 104, 112, 194. See also Aris-
Inheritance. See Bequests totle. Memory, Senile dementia
Innovation, age profile of, 48, 167 Kronman, Anthony T., 194
Intelligence, crystallized, 6, 21, 57, 67- Kuhn, Thomas S., 229-230
70, 160-162, 166-167; fluid, 6,21,
23, 57, 67-70, 76-77, 106-107, Labor costs, incidence of, 325-326
116, 142, 160-172, passim, 176, Labor force participation of elderly, 39-
192, 197, 200, 206, 219, 292, 355; 41,44,349-350. a/jo Retirement
ratio of fluid to crystallized in phi¬ Landes, William M., 1 lOn, 183n, 306-307
losophy, 168; ratio of fluid to crys¬ Last-period problem, 59, 61, 72, 108,
tallized in philosophy versus in sci¬ 175, 206, 302-303, 337, 352; in ad¬
ence, 170-171 judication, 196-197; reverse last-
Interest rates, 92. See also Discount rates period problem, 64
Intergenerational competition, 70 Law, practice of, 180-181
370 Index

Law clerks, 181-182, 192, 200 Loquacity, 119-120, 175. See a/50 Con¬
Law professors, age profile of, 180 versation, inconsiderate
Lawyers, age profile of appellate versus
trial lawyers, 181; elderly, 193n Management, age profile of, 159-160,
Lazear, Edward R, 59n 166. See also Leadership
Leadership, age profile of, 156-166, Mandatory retirement, 319, 324-325,
229; judicial, 192; revolutionary, 328, 331; in academia, 351-358;
227; by women, 208 economics of, 350-351; in govern¬
Learning by doing, 52, 82, 163 ment, 351; history of, 350; effect of
Legal profession. See Judges, Law, Law abolition on labor force participa¬
professors. Lawyers tion of older workers, 349-350
Lehman, Harvey C., 157n, 161, 167, 192 Market versus nonmarket allocation of
Leisure, 45-46, 52n, 339-340, 343; dis- resources, 214-216
valuers of, 178, 340; economies of Marriage, 139, 277; arranged, 209n; mar¬
scale in, 83; judicial, 198; as supe¬ riage market, 6In; remarriage by el¬
rior good, 46 derly widowers, 225
Lesbians, 140n Marzen, Thomas J., 240
Lessig, Lawrence, 183n Masculinity, as affected by aging, 102n.
Levine, Martin Lyon, 35In See also Sex, sex-role turnover
Life, quality versus quantity, 279-280 Masturbation, 140-141
Life cycle, of organizations, 227-229; of Mathematics, age profile of creativity
sciences, 229-230 in, 157, 159-160, 166-167,
Life expectancy, accuracy of forecasts of 173-174
own, 263. See also Longevity Matisse, Henri, 176n
Life insurance, effect on of physician- Maturity, 162; of judges, 194
assisted suicide, 260n MCI (Mild Cognitive Impairment), 22,
Literacy, 219; and status of elderly, 31, 351
204-208 MDEL (Medical Decisions at End of
Literature, age profile of creativity in, Life), 236, 259
159, 167-168; literary depiction of Medicaid, 47, 265, 293-294, 314
old age, 99-101 Medical care, demand of elderly for, 112.
Living will, 258-259, 260n See also Death, dread of
Logic. See Reasoning, abstract Medical ethics, and assisted suicide, 241.
Loneliness, 63 See also Biomedical ethics
Longevity, 45-47; biological limits of, Medical insurance, 265-266. See also
30; effect on of efforts to inculcate Medicare
filial piety, 209; in ancient Greece Medical profession, payment method as
and Rome, 32; effect of on human bearing on physician-assisted sui¬
capital, 55-56; effect on of income, cide, 242, 250. See also Health care;
42, 47, 50, 270-271, 282, 293- Suicide, physician-assisted
294; effect of increasing on judicial Medical research, 269-280; on lethal
performance, 199-200; male versus versus nonlethal diseases, 280; on
female, 39, 273-281; in Middle male versus female diseases, 273-
Ages, 32; occupational, 72; in prim¬ 278
itive societies, 28-29, 207-208; Medicare, 42, 44, 47, 49, 264-269, 284-
trends in, 33-39; in United States, 285, 293-294, 314; expenditures in
33-39; and utility, 275-278 last year of life, 112n
Index 371

Memory, 6, 18-19, 67, 69, 74, 112, Old age, not a natural kind, 49
206-207 Older Workers Benefit Protection Act,
Menopause, evolutionary theory of, 27- 340n
28, 202n Optimism, 70, 104-108, 134; evolution¬
Mental competence, and deterrability, ary biology of, 105
315; to execute legally binding Oregon, physician-assisted-suicide law,
documents, 315-318; to stand trial, 237,261
315. See also Alzheimer’s Disease, Organization theory, 227-229
Senile dementia Organizations, age distribution of em¬
Mental illness, 273. See also Depression ployees in, 227-229, 348-349
Mercy killing, 243, 279
Messeri, Peter, 176 Paternalism, 90, 255-256, 263-264
Metaphor, 169 Patriarchy, 212, 215, 219
Military leadership, 116-117 Pedestrian fatalities, 11 In, 125, 153
Military retirement, 143 Pedophilia, 129, 141
Mill, John Stuart, 238, 244, 257, 259-260 Peirce, Charles Sanders, 105
Mincer, Jacob, 56-57 Penny-pinching, 137-138
Montaigne, Michel de, 104n Pension Benefit Guaranty Corporation, 303
Moody, Harry R., 240 Pensions, 59, 141-142, 282-283;
Morality, intuitive versus rational, 257 alleged abuses before ERISA,
Mulligan, Casey B., 92-93 301-304; backloading of, 301 -
Multiple selves, 8-9, 50, 84-94, 130, 304; defined-benefit versus defined-
138, 254-256, 266-269, 281-282, contribution pension plans, 144,
293, 304-313 passim, 321, 347; 299-305; incentive theory of,
and social security, 90, 263-264 302; of federal judges, 185-186,
Musical composition and performance, 193; over- and underfunding of,
174-175; age profile of creativity 301, 303-304; tax treatment of,
in, 159, 168 301, 304. See also ERISA, Social
security
Natural selection, 25-30 Pessimism, 70, 104-108
Netherlands, euthanasia in, 236-249 pas¬ Pet therapy, 135
sim, 252-259 passim Philosophy, age profile of creativity in,
Networks and networking. See Human 159, 168-171
capital, relational Physics, age profile of creativity in, 156-
Nietzsche, Friederich, 111, 114, 169 159,164, 166
Nobel Prize-winners, age at which did Pilots, airline, age of retirement of, 323,
work for which prize was awarded, 326
159n Poetry, age profile of creativity in, 167-
Nonmarket output, 42, 45. See also 168
Leisure Politics. See Democracy; Social security,
Nursing homes, 146-147, 150, 255, politics of
298-299; nursing-home insurance, Polygamy. See Polygyny
285; population of, 47 Polygyny, 139,212,217-218; serial, 225n
Postindustrial economy, status of elderly
Oedipus at Colonus, 177 in, 219
Office of Public Prosecutions v. Chabot, Poverty, elderly. See Income, of elderly;
243, 249 Social status of elderly
372 Index

Power of attorney, durable, for health Reasoning by analogy, 70, 106-107,

care, 258-259 165, 169, 192
Practical reason, 23, 106, 164n, 201 Recidivism, 132-133
Prado, G. R., 239n, 255 Redistribution. See Social security, redis¬
Precedent, decision in accordance with. tributive effect of
See Stare decisis Reduction in force (RIF), 331, 336, 341 -
Primitive societies, age grading in, 322- 342
323; definition of “primitive,” 204; Refereeing, blind, 176
old people in, 28-29, 31-32, 203- Regret, 108
210; static versus dynamic, 206- Religion, 113, 132, 224-225; belief in
207, 212 afterlife, 59-60, 117-118, 131,
Prisons. See under Crime 205; and gerontocracy, 210, 227;
Privacy, 108n, 146-147; as superior and health, 118; longevity of Roman
good, 146 Catholic Church, 228; and suicide,
Problem-solving, 207. See also Intelli¬ 238
gence, fluid Reputation, 58, 60,64, 114-115, 131,
Productivity, academic, 156-174 passim; 302; “celebrity” phenomenon, 168;
of aged workers, 74-78, 80, 130, importance of in law, 181; scientific,
156-174, 336-337; in creative oc¬ 172
cupations, 156-174 passim; of fi¬ Research. See Science, Universities
nance professors, 157n; judicial, Residence of elderly, 145-148
187, 188, 198, 200; in leadership Retirement, 4, 7, 74; age of, 36-37, 48-
and administration, 156-166; effect 50, 55-56, 141-145, 291-292,
of on status of elderly, 219; effect on see also Retirement, early (below);
of success, 173 how age of affected by choice be¬
Progress, 107-108, 120 tween defined-benefit and defined-
Promotion and demotion, 77-78 contribution pension plan, 300;
Property rights, 211 - 212, 215 early, 181, 339-343, 349, 353-357;
Psychology of aging, 19, 22, 24-25, 70- economic model of retirement deci¬
72, 80-82, 99-121, 137-138, 162. sion, 78-83, 341-342; formation
See also Aristotle, on old age of friendships in, 62-63; of judges,
Psychometric testing, 18n, 24-25, 77 181-201 passim, 351-358; manda¬
Public Employees Retirement System v. tory, see Mandatory retirement;
Betts, 340n military, 143; partial, 44, 82-83;
post-retirement careers, 142;
Quality of life, 241, 219-ISO by self-employed, 143-145; by
self-employed professionals, 144-
Race. See Blacks 145; trend to earlier, 32, 39-41
Rational choice, applicability of concept Risk, allocation of by retirement plans,
to primitive societies, 218n; nature 301; attitude of elderly toward,
of, 99; role of in treatment of elderly 116-117, 144; aversion to, 117. See
by young, 202-203 also Death, dread of
Rawls, John, 89n, 168-169 Romeo and Juliet, 238
Reasoning, abstract, 70, 169, 181, 192, Rosen, Sherwin, 267
201; abstract versus concrete, 106.
See also Knowledge shift. Practical Salesmen, 64, 344-345; and age dis¬
reason crimination, 344-345
Index 373

Savings, 294; by elderly, 60-61, 290; Social discount rate. See under Discount
rate, 84n, 9In, 147 rates
Schelling, Thomas, 166 Social security, 40, 44, 47-49, 118, 141,
Schopenhauer, Arthur, 239n 186, 222, 262-264, 266, 281-294,
Science, acceptance of new theories 304-305; annuity component of,
by elderly, 172n; age profile of 287; as fair contract between old
creativity in, 156-174 passim; and young, 284-294 passim; dis¬
optimal research strategy of eld¬ ability benefits, 143, 291; effect of
erly scientist, 172; optimal re¬ on retirement age, 40, 324; free¬
search strategy of young scientist, rider rationale, 263; multiple-selves
172-173; revolutionary versus rationale, 90, 263-264; paternalistic
normal, 229-230; theoretical rationale, 263; pay as you go char¬
versus experimental, 166. acter, 282-283; politics of, 285,
See also Physics 288-294; redistributive effect of,
SDAT (Senile Dementia of the Alzhei¬ 282- 294; social security tax, 269,
mer’s Type), 21-22. See also 283- 284, 289-290. See also
Alzheimer’s Disease Medicare
Selection bias, 63, 76, 81, 126, 187, 210, Social security disability benefits. See
222-223, 321, 323-324, 360; in under Social security
crime statistics, 132-133; in dis¬ Social status of elderly, 77, 203-226; in
count rates, 71; and decision for eu¬ agricultural societies, 210-218; in
thanasia, 237; and happiness of old ancient Greece and Rome, 31-32,
versus young, 110-111; in suicide lOln, 224; material versus symbolic
rates, 135-137 or honorific, 220, 223, 230; in
Self. See Multiple selves Middle Ages, 32, 225; in primitive
Self-centeredness of elderly, 108, 119 societies, 203-210, 213-218; in the
Selfishness. See Self-centeredness of United States, 204, 218-224, 226,
elderly 281; in warrior societies, 216-218
Seligman, Martin E. R, 105n Sophocles, 175, 177
Senile dementia, 20-22, 280n; incidence Specialization, effect of on productivity
of, 21-22; and legal capacity, 315- of older workers, 223; Judicial,
318. See also Alzheimer’s disease. 200-201; of legal profession, 197
Senility Spence, Michael, 228
Senility, 87, 147, 254-261 passim. See Stare decisis, 194-195
also Senile dementia Statistical bias, 323-324
Senior status (of federal judges), 185- Status. See Social status of elderly
189,193 Stereotyping, 19-20. See Discrimina¬
Sex, 19, 23; elderly sex offenders, 129, tion, statistical
141; sex-role turnover, 114, 208; Stigler, George J., 59n
sexual behavior of elderly, 138- Stoics, 114
141,277 Style, old-age or later-life, 174-177
Sex ratio of elderly, 39, 129, 273, 276- Success, effect of on productivity, 173
277; “effective,” 139 Suicide, 57, 153; commitments to or con¬
Shamelessness, 103-104, 175-176 tracts of, 254-261; constitutional
Simmons, Leo W., 203n, 213 issue, 239; cost of, 246-250; by de¬
Simonton, Dean Keith, 157n, 174, 177, hydration and starvation, 236, 239n,
182, 199n 258; by elderly, 133-137, 244-245,
374 Index

Suicide (continued), 248-249; by el¬ Universities, 180, 334-335; retirement

derly, men versus women, 137; by in, 351 -358. See also Tenure,
elderly, trends in, 134-135; by el¬ academic
derly, U.S. versus foreign statistics, Utilitarianism, 89, 243
133n; by elderly demented, 317- Utility, diminishing marginal utility of
318; with execution deferred, 254- income or wealth, 83-85, 137-138;
261; and Law of Demand, 249- ex ante versus ex post, 109, 271;
250; effect of laws against assisted, expected, 109, 112-113, 279-
250-252; by physicians, 239n; 280; and longevity, 275-276; non¬
physician-assisted, 235-261 pas¬ pecuniary utility of work, 79-80;
sim; rates, 250-253; rate in Nether¬ posthumous, 58, 114-115, 131,
lands, 252-253; rational, 238-239; 197. See also Suicide, rational-
rational-choice model, 243-250, choice model of
255; religious objections to, 238;
right to as option, 239-240. See Value of life, 109-110,306-310
also Euthanasia Verbosity, off-target, 119n. See also Con¬
Supreme Court, “Court-packing plan,” 3 versation, inconsiderate; Loquacity
Surrogate-decision-making statutes, Voting, child-weighted, 289; as con¬
258-259 sumption activity, 148-149; by el¬
derly, 118-119, 148-152, 288-294
Taciturnity, 119-120 passim; by unemployed, 151-152;
Taxation. See Bequests, taxation of virtual representation of children by
Teaching, age profile of creativity in, parents, 151, 289; voter turnout,
159 148-152
Technological progress, effect on of
aging population, 48, 167 Wage-bonding theory, 206. See also
Tennyson, Alfred Lord, 100-101 Compensation, backloading of; Effi¬
Tenure, academic, 79-80, 355; Judicial, ciency wage
2-3, 351-353 Wages, average of elderly, 321-322. See
Third World, longevity in, 34-35 also Compensation, economics of;
Time, cost of to elderly people, 116, 124, Human capital; Productivity of el¬
137, 149-150; subjective, 70-72 derly workers; Wage-bonding theory
Time preference. See Discount rates Waugh, Evelyn, 242n
Tort law, award of damages in death Wealth, distribution of, 212, 223-224.
cases, 306-310; award of damages See also Income, of elderly; Social
for nonpecuniary losses, 84-85; status of elderly
standard of care of elderly, 305-306 Weinfeld, Edward, 186, 192
Tradition, 207 Welfare economics, 86-89
Transaction costs, 212 Widowerhood, 225
Triage, 264, 278-279 Widowhood, 139, 277-278
Trial and error, 207 Wills, 60, 315-318; living, 258-259,
260n
“Ulysses” (Tennyson), 100-101, 113, Wisdom of elderly, 65, 106, 146, 208; of
178 elderly judges, 192, 194
Unions, appeal of to elderly workers, 75- Wittgenstein, Ludwig, 168-169
76, 143; elderly leadership of, 227; Women, as caretakers of the elderly,
and mandatory retirement, 350n 145-146, 221, 223; as plaintiffs in
Index 375

cases under Age Discrimination in Writing skills, and age, 167-169, 171,
Employment Act, 347; participation 199; and importance of injudicial
of in labor force, 40-41, 145, 221, process, 198
351; rate of aging compared to that Wrongful-death suits, 308-310; verdicts
of men, 25; research on diseases in, 309-310
of, 273-278; age of retirement
of, 37; sexual activity of elderly, Yeats, WB., 100, 113-114, 167, 174-
138-141 175,177
Work, cost of. See Effort Youth, Aristotle’s view of, 103-104, 106
 3 9999 03730 329 2

No longer cne picperty of

PutoJIe Library.
■■'C ' '! *• '• 1 li.'-; . ~ •
 Boston Public Library

COPLEY SQUARE
GENERAL LIBRARY

The Date Due Card in the pocket in¬

dicates the date on or before which
this book should be returned to the
Library,
Please do not remove cards from this
pocket.
Economics / Sociology

R ichard A. Posner is one of the most interesting figures in American

social thought. . . . Mr. Posner’s aim. . . is to demonstrate the
power of the ‘human capital’ approach put forward by Gary S. Becker
and other economists.”
—Joseph Adelson, The New York Times Book Review

“Posner puts his robust common sense in the service of demystifica- ^

tion. . . . Posner’s case against the prejudices of our times is established 1
with spirited arguments and a fusillade of figures. . . . [H]is clarity,
range of reference, eye for telling detail, bold expansion and occasion- ]
al quaint digressions make his book a pleasure to read.” ^
—Roy Porter, Times Literary Supplement

“Posner is at home with and draws from a wide multidisciplinary litera¬

ture. There is hardly a page on which some useful insight cannot be
found.”
—Michael Uhlmann, Washington Post Book World \

“Richard Posner. . . is one of today’s most important heirs of the \

Enlightenment. . . . [T]he principal contribution of this book is to sug- *
gest that the economic model must take into account man’s biological !
nature. . . . By writing about subjects of universal concern. Judge \
Posner has done more than anyone to publicize the biological and eco- j
nomic model of man that is the roadmap for the social sciences in the ^
coming century.”
—John O. McGinnis, Wall Street Journal j

Richard A. Posner is Chief Judge of the U.S. Court of Appeals for the
Seventh Circuit and senior lecturer at the University of Chicago Law
School.

The University of Chicago Press

Cover Illustration: Leonardo da Vinci, ISBN Q-BEh-b75bfl-fl

“Old Man and Youth,” red chalk, about 1495. 900 0
Printed in U.S.A.

780226 675688