The Political Economy of Gender:

Explaining Cross-National Variation in the Gender Division of

Labor and the Gender Voting Gap

Abstract

Mainstream political economy has tended to treat the family as a unit when examining the
distributional consequences of labor market institutions and of public policy. In a world with
high divorce rates, we argue that this simplification is more likely to obscure than to instruct.
We find that labor market opportunities for women, which vary systematically with the position
of countries in the international division of labor and with the structure of the welfare state,
affect women’s bargaining power within the family and as a result, can explain much of the cross
country variation in the gender division of labor as well as the gender gap in political
preferences.
1. Introduction

Some of the most notable variance in income, labor market status, and the division of labor is

gender based. On average, women participate less in the labor market than men, whereas they

assume the lion’s share of unpaid work in the household. Women also tend to be less well paid

than men when they do work, and they occupy jobs with lower job security, fewer prospects of

advancement, and less responsibility. Often, these inequalities spill over into a gender gap in

political preferences and voting behavior.

Following Gary Becker’s (1964; 1971; 1981; 1985) seminal work on the family, economists

have traditionally explained the gender division of labor as the outcome of a coordination game

where a more or less complete division of labor is the efficient solution due to increasing returns

to human capital. Although the biological advantages of women specializing in household skills

are slim in a modern economy, such specialization may be reinforced by childhood socialization

in which parents rationally seek to maximize the success of children later in life. Since gender

roles are assigned before “true” preferences are observable, the coordination game is solved by

using (inherently small) gender differences as the cue.

But while the efficiency model captures some key aspects of the family as an institution, it is

incapable of accounting for the stark differences in female labor force participation across

economies at comparable levels of development, and it fails to explain why there is so much

variance in the distribution of housework between the sexes after controlling for hours spent in

paid work and earnings. Building on recent economic bargaining models of the family (Folbre

1987; Braunstein and Folbre 2001; Pollak 1999; 2003; Lundberg and Pollak 1996; 2001), we
2
argue that this division of labor puzzle can only be understood by treating marriage as an

incomplete contract that is potentially subject to termination. When this is the case, both men and

women have an incentive to cultivate their outside options by entering into paid work, and the

distribution of unpaid work is determined by bargaining where bargaining power is dependent on

political-economic factors outside the family. We make use of recent political economy

arguments (e.g. Hall and Soskice 2001, Pontusson forthcoming; author1 forthcoming, Estevez

1999) as a well as macro-sociological work on the welfare state (e.g., Esping-Andersen 1990,

1999, Orloff 1993, Huber and Stephens 2001) to tie macro-level conditions to intra-family

bargaining over the division of labor. Because some countries take advantage of opportunities in

the international economy by specializing in production where women have a comparative

disadvantage, the gender division of labor is tied to the international division of labor.

A second puzzle, the rising, but country-specific, gender gap in political preferences and voting

behavior, looms large in the political world yet has received little attention by comparativists. In

economic efficiency models of the family there is no room for men and women to favor different

public policies. Families will differ over social policies depending on their position in the age

and class structure and the like, but family members are assumed to have more or less identical

preferences. When divorce is a serious prospect, on the other hand, we have to treat family

members as individuals with distinct and potentially conflicting preferences. Basically, when

outside options are important for long-term welfare, both sexes will prefer social and economic

policies that maximize these options, even when this reduces total household income or other

measures of aggregate welfare. But whether the distributive game over outside options translates

into differences in political preferences depends critically on the structure of the economy. The

3
key, we argue, is the extent to which the political economy facilitates female labor force

participation, which in turn depends on whether production relies on specific as opposed to

general skills and the extent to which the state provides a broad range of social services.

Following a standard trade argument, whether firms make intensive use of specific skills depends

on the relative abundance of such skills, and this is in turn determined by the broader national

institutional framework -- what Hall and Soskice refer to as comparative institutional advantage.

Somewhat paradoxically, we find that it is where women are in the weakest position in the

economy that their preferences are the least distinct from men’s.

The rest of the paper is organized as follows. Section 2 lays out the standard efficiency model of

the family division of labor, and the bargaining model alternative that better fits today’s world.

Our empirical findings are that, controlling for a variety of factors, women with stronger outside

employment options are able to reduce their share of family work. Section 3 explains the cross-

country variation in gender-based political preferences. We show that the gender gap in

preferences is dependent on female labor force participation and the possibility of marriage

breakup. Section 4 concludes.

2. Explaining the gender division of labor

In Becker’s efficiency model, couples engage in a division of labor to take advantage of gains

from trade. One family member will specialize in marketable skills and paid work, while

another will specialize in household skills and unpaid work, on grounds that there are increasing

returns to human capital in both domains, and that the care of children cannot be completely

subcontracted out without loss to the children’s wellbeing. The spouse specializing in household

4
skills may enter the labor market part time if the domestic workload permits this – something

that is more likely to be the case early and late in a marriage corresponding to the years before

children are born and after they start school or leave the household.

In principle, either the husband or the wife could specialize in market or household skills, but

women will almost invariably specialize in the latter because of an initial comparative biological

advantage in caring for very young children. This advantage may only last for the first months

of a child’s life, and it is not hard to imagine distributions of preferences for type of work across

the sexes that would lead to a far more even division of family responsibilities and paid work

than we actually observe. Becker solves this puzzle in two ways. First, a small comparative

advantage is magnified by increasing returns to human capital -- people get better at tasks as they

accumulate experience. Second, parents have an incentive to prepare their children for

responsibilities they will assume later in life, and this may reinforce the gender division of labor

by instilling preferences through childhood socialization. Since children are unlikely to reveal

their true role dispositions at a very early age, parents choose to socialize their children in skills

that they are most likely to be good at, or even more reflexively, that society assigns. If females

grow up thinking it is normal for them to stay at home with the kids, this magnifies biological

differences and solves the family coordination problem later in life.

In hindsight (Becker developed his argument in the early 1960s) it is easy to ridicule Becker’s

model as an intellectual justification for the traditional male-dominated 1950s family. But it is

precisely the capacity of the model to account for the stark gender division of labor and the

differences in the socialization of girls and boys in the traditional family that makes it so

5
powerful. Still, it is clear that something fundamental has changed in a number of countries in

North America and Europe. The division of labor by sex is less pronounced, and socialization

less gendered, than a few decades ago. Families in U.S., for example, are more likely to teach

their girls to be assertive and independent than was the case some decades ago or than is still true

in countries where female labor market opportunities are more restricted (Hrdy 1999).

One factor that looms large in the explanation of these changes is the rise in divorce rates.

Whether divorce rates are a response to exogenous changes in divorce law, or whether divorce is

endogenous to a growth in labor market opportunities for women, divorce is now an accepted

part of modern life in most rich democracies. In 1950 the probability that a first marriage would

end in divorce was one in five in the U.S., and any behavior that could conceivably lead to

divorce – infidelity but also overt challenges to established gender roles – were considered taboo

by widely held religious and community norms. Today the divorce rate is one in two for first

marriages and is now considered an acceptable, even desirable, solution to marital problems.

This makes a tremendous difference to the Becker framework because spouses must now

concern themselves with what they can do to secure their welfare in the event that the marriage

breaks up. In other words, their outside options become crucial. And if we want to understand

the implications of this for the division of labor and patterns of socialization, we have to treat

family members as individuals with distinct preferences. The short hand of treating the family as

a unitary actor is less useful than it may once have been.

The most obvious potential conflict of interest concerns the division of labor -- precisely the

variable Becker’s model was designed to explain. The problem is that heavy investment in

6
household-specific skills is likely to undermine outside options. Not only will such investments

crowd out investments in marketable skills, but the value of marketable skills is likely to be

seriously reduced by longer absences from the labor market (Polachek 1975; 1978). In principle,

household skills can be “sold” on the re-marriage market, but since one critical “skill” in this

market is to produce and nurture offspring, a woman’s position in the re-marriage market will be

seriously reduced as soon as she has children in another marriage. Adding insult to injury,

another valuable commodity in the re-marriage market is youth and beauty, which also

deteriorates with time. Hence, even in the remarriage market, the only non-perishable

commodity is earnings power – and perhaps also the attractiveness that comes with education

and an active lifestyle.

Since labor market participation is essential to cultivate outside options, women have strong

reasons to resist a complete division of labor in the family. But by the same token men have an

incentive to resist taking on more domestic responsibilities. The division labor, especially the

division of household labor, therefore becomes subject to contestation. Bargaining models of

marriage capture this by assuming that compromises have to be found in a bargaining space that

is constrained by the outside options. The simplest conception of outside options is whatever

utility either party can get outside the marriage. But some models also allow for the possibility

of non-cooperative outcomes without divorce where spouses recede into separate “spheres”

characterized by more or less separate finances, partially divided living spaces, etc. (Lundberg

and Pollak 1996).

In either formulation, opportunities in the labor market shape outside options and hence the

7
marriage bargaining space. In a simple Rubinstein bargaining model with identical discount

rates and negligible first-mover advantage, the outcome of the game splits the difference between

the ideal points at the boundary of the bargaining space. So, for example, if outside options are

equally attractive to both sides, and both spouses want to maximize paid work and minimize

unpaid work (and thus invest in marketable instead of household skills), we would predict an

even distribution of household work. More realistically, since women’s outside options tend to

be inferior to men’s, women will tend to do more of the household work. But they have a strong

incentive to resist the complete division of labor that would be optimal in the Becker model.

(Braunstein and Folbre 2001; Lundberg and Pollak 1996; 2001; author2 forthcoming). This

incentive rises with the probability that a marriage will end in divorce, and the division of labor

is therefore a function of this probability (which plays no role in efficiency models).1

Since household bargaining may lead to a less complete, and hence less efficient, division of

labor it is logically conceivable that a marriage contract could compensate women for the risks

of specialization in household skills (and the associated deterioration of marketable skills) by

guaranteeing a lump sum severance payment or the sharing of future income streams (such as

alimony and child support) in the event of divorce. But to prevent problems of moral hazard,

shirking, and other well-known maladies of incomplete information such prenuptial agreements

would have to stipulate all relevant contingencies in advance – including just cause for divorce,

fair treatment of the other party in the marriage, the division of custody in the event of divorce,

and penalties for non-compliance with any stipulations in the contract. Precisely this type of

detailed marriage contracting has reached almost farcical complexity among Hollywood

1
For a formal proof of this using a standard bargaining model, see author1 and 2, 2004.
8
celebrities where the stakes and divorce rates are both very high. But few would claim that

prenuptial agreements constitute a general solution to incomplete marriage contracting. Just as

in non-standard economic contracting, comprehensive ex ante agreements are either impractical

or prohibitively expensive to write and enforce.

In this paper we focus on the bargaining process between two spouses, and leave the degree of

efficiency loss from the breakdown in specialization as an open empirical question. If, for

example, child care could be subcontracted without loss of child welfare, we might expect

couples to abandon much of the specialization of labor without efficiency loss.2 This reduces the

need to bargain over household labor. But the cost of childcare is of course very much a matter

of public policy, and preferences over these policies, which we do examine in the second part of

this paper, can be understood within the bargaining framework we use. Besides, time budget

research (which we discuss below) show that unpaid household labor continues to make up a

very substantial portion of total labor.

With incomplete contracting we expect a woman’s bargaining power within the family to be

inversely related to the labor market’s premium on specific skills. As labor economists point

out, women are generally at a disadvantage when competing for jobs with men because they are

2
At the other extreme, if the returns to specialization in family work exist but are hard to observe, or are not

sufficiently internalized by either parent, we might expect suboptimal levels of child welfare to result from a decline

in specialization. We are assuming, though perhaps without justification, that both parents have a full and equal

interest in their children’s well being, so that their bargaining over paid and unpaid work does not include the

possibility of a lower overall level of investment in their children.

9
expected to leave the labor market for purposes of child birth and rearing (Mincer 1962; 1978;

Polachek 1975; 1987). Employers will therefore be more reluctant to invest in skills of women,

and young women are likewise more reluctant to build up substantial employer-specific assets or

even invest in the education that is needed for a specific skills kind of job since these may be

forfeited with the birth of their first child (Anderson, Binder, and Krause, forthcoming).

How great the motherhood disadvantage is, however, depends on the nature of skills that

employers are seeking, as Estevez-Abe (1999) and Estevez-Abe et al. (2000) have argued. If

such skills are highly specific to firms, or even to industries, and if a substantial part of training

is paid by the employer, there is a strong disincentive to make these investments in female

employees where the average time horizon is comparatively short. This is reinforced by

women’s own decisions because they are disinclined to invest in specific skills for which they

are at a disadvantage. Even if a woman invests to acquire a specific skill, her investment will not

be protected to the same degree as a man’s. Women are therefore more likely than men to invest

in general skills and/or in skills that are less prone to deteriorate when not used for some period

of time (lower atrophy rates). This implies a heavily gendered structure of educational choices,

and it is not surprising that vocations with more general educational content and low atrophy

rates such as commerce, services and home economics are overwhelmingly female in

composition (Estevez-Abe 2002). Women facing tough labor market constraints may be better

off aiming as high as possible in the marriage market, and educating themselves in the “gracious

arts” rather than in marketable skills.

In the Becker model, the difficulty of women finding paid work does not matter for the

10
household division of labor so long as the productivity of the husband is greater. The household

division of labor is always complete. This is not necessarily the case for paid work because the

woman may have time left over to enter the labor market, especially during parts of the life cycle

where there are no dependent children. But only the bargaining model implies that the amount of

paid work, and the earnings power of the woman, will matter for household division of labor (as

long as there are any differences in productivity between the spouses). Also, in so far as skills

are specific, paid employment should benefit men more than women because men are in a better

position to accumulate specific skills.

The importance of the skill argument for understanding variation in bargaining dynamics inside

the family is reinforced by broader cross-national differences in the structure of production.

Taking advantage of the international division of labor, some countries have specialized in the

forms of production that use specific skills intensely while others have specialized in production

that uses general skills intensely (Hall and Soskice 2000, author1 forthcoming). In the latter,

women are generally better able to compete on an equal footing with men in the labor market

because investments in skills are mostly borne by workers rather than by employers (say, though

college education) and because general skills do not depend on staying with a particular

employer for a long period of time. Because firms seek to strengthen their position in the

international division of labor, they will work politically to create and reinforce institutions that

are designed either to protect specific skill investments or to encourage investment in portable

skills. Institutions that protect specific skills Protective institutions, such as high job security,

seniority pay, and generous employer-financed benefits, tend to reinforce insider-outsider

divisions, and since women are more likely to be outsiders, they are at a greater disadvantage

11
compared to more flexible labor markets where low protection encourages investment in general

skills. Furthermore, because compression of wage differentials is one way to protect investment

in specific skills, some specific skills systems are characterized by high minimum wages that

tend to push up the cost of daycare and other family-oriented services.

Because of these differences, the outside options of women in general skills systems tend to be

better than in specific skills systems, and so is their concomitant bargaining power. This implies

that, everything else being equal, female labor market participation tends to be lower in specific

skills systems, and the distribution of household work more unequal, than in general skills

systems. These effects, however, will be mediated by social and economic policies deliberately

designed to counter them, which brings us to a standard argument in the macro-sociological

literature. In particular, the extent to which the state support the ability to form independent

household, especially through publicly provided services such as daycare, and through

employment for women in these services, can compensate for the exclusion of women from good

jobs in the private labor market (Orloff 1993; Esping-Andersen 1999). The Scandinavian

countries in particular have attained high female participation rates by creating a large, and

heavily feminized, public sector.

2.1 Empirical Analysis of the division of labor.

The data for our analysis are from the 1994 International Social Survey Program, which focuses

on the family and gender relations. The data cover most established democracies, a few East

European transition economies, and one developing country (the Philippines). We focus on the

former since we have macro-level data for our institutional and labor market variables for these

12
countries. None of these data are available for the east European cases, which transitioned to

democracy a few years before the survey and were still in the early phase of privatization. The

cases included in the analysis below are Australia, Austria, Canada, Ireland, Italy, (West)

Germany, Japan, the Netherlands, New Zealand, Norway, Spain, Sweden, the UK, and the U.S.

One case, Spain, is missing so many of the key independent variables that it had to be excluded

from the individual-level analysis (but we kept it in the macro-level analysis).

We focus exclusively on married and cohabitating couples. The dependent variable is an index

constructed from a battery of four questions asking who in the household, the man or the woman,

performs a variety of household tasks. For example, one question reads: In your household who

does the laundry, the washing and ironing? 1. Always the woman, 2. usually the woman, 3.

about equal or both, 4. usually the man, 5. always the man? The other three questions ask who

cares for sick family members, who shops for groceries, and who decides what is for dinner. One

additional question asks who does repair work around the house. But as pointed out by

Hochschild (1989), such work is infrequent and often has a leisure or hobby component. This is

confirmed by a principal factor analysis performed on all five items, which identifies two

dimensions: one where only the first four items have high, and about equally large, factor

loadings, and one where only the repair item has a (moderately) high loading.3

3
The complete factor loadings are as follows:

Factor 1 Factor 2
Laundry 0.51 -0.09
Caring for sick 0.59 0.08
Shopping 0.67 0.01
Dinner 0.66 -0.05
Repairs 0.15 0.22
13
Based on these results, we created a simple additive division of household labor index based on

the first four items, where higher values mean that more of the work is performed by the woman.

Since most household labor is done by the woman, one can loosely think of higher values as

indicating more inequality in the division of labor. The variable ranges from 1 to 5, with 3 being

an even sharing of work. The mean for the variable is 3.97, which is equivalent to an average

response to each question of “usually the woman”. None of the reported results below change

substantively if we instead use an index based on all five items.

The fact that child care is left out of these questions undoubtedly leads to a substantial

understatement of the woman’s share of work. Research on family work based on time diaries,

which do include a category for child care, show that children of all ages increase women’s

overall unpaid work time three to four times more than they increase men’s (Bittman, England,

Folbre, Matheson 2001).4 But we expect, at least, that the male-female division of childcare

responsibilities will parallel the way they divide other family tasks.

While it is not possible to know with precision how the survey-generated index (without

childcare time) maps on to actual hours of work done, we can get a good sense of this by

comparing the index to the results of international time budget research. According to one

authoritative study, women on average perform more than two thirds of total household work

(Gershuny 2000). This study also shows that the average adult spends 230 minutes per day on

4
Time diaries, which ask respondents to keep track of how they allocate time during the day, are preferable to less
complete surveys of this sort. Unfortunately, they are only available for a few countries.

14
domestic work, equivalent to 460 minutes, or almost 8 hours, for a household with two adults. If

the answer “always the (wo)man” means that the (wo)man literally does all the work, the index’s

range of 4 units is equivalent to 460 minutes, or about 115 minutes per unit (or 14 hours per

week) assuming equidistant spacing between the different values. One standard deviation on the

index is .67 or about 77 minutes of work (9 hours per week).

For paid work we use two variables that ask about the employment status of the respondent and

of the spouse. It is coded 1 for those who are full-time employed, 0.5 for part-time employed,

0.25 for less than part-time employed, and 0 for those who consider themselves homemakers or

who are retired. Unemployed and students are ignored. The variables are coded for men and

women separately. They are included as independent variables in the regressions of unpaid work.

In addition, we use five sets of independent variables to explain the individual-level variance in

the division of labor. One is the (pre-tax) wage income of the husband and wife, measured

separately. Since we only have information about the income of the respondent the earnings of

the spouse are inferred from information about household income. To do this we have to assume

that all income is wage income and that husband and wife are the only wage earners in the

household. Since there are non-wage sources of income, and sometimes more than two adult

wage earners, this would suggest that income estimates based on the difference between family

income and respondent=s income exceed the latter on average. In fact, inferred incomes of

spouses are slightly lower than respondent incomes, but generally very similar (within 90 percent

of the respondent=s income). This suggests that the inferred number is a fairly good proxy for

15
the spouse’s income.5

We measure the probability of divorce in two different ways. There is no direct measure of the

probability of divorce at the individual level, so we use past divorce as a (very imperfect) proxy

since we know that the aggregate likelihood of divorce is higher for those who are previously

divorced. The variable is coded 1 if one of the spouses is previously divorced, 0 otherwise. At

the aggregate level we simply use national divorce rates as a proxy for the society-wide

probability of divorce.

Marketable skills are at least partly a (negative) function of time spent on household labor. We

capture this logic using a battery of questions about past family-related labor market absences.

Specifically, the questions inquire about time taken off during four different phases of child

rearing: i) before the birth of the first child, ii) before the youngest child entered school, iii) after

the youngest child entered school, and iv) after the children have left home. The variable takes

on the value 1 when the wo(man) did not work during any of these periods, and the value 0 when

the wo(man) worked full time during all four periods (part-time work is coded .5). This coding

follows Librizzi (2002).

Another family-related variable measures the number of dependents. It is calculated by

combining information about the number of household members with information about whether

the family is headed by one or two adults. In most cases it refers to the number of children,

5
It does at any rate not systematically bias the estimates of male and female income since the respondents
were roughly equally divided between men and women.

16
although it will also capture older generations of family members living in the household. In

either case, this variable is a proxy for the demand for household labor, and it will tend to raise

the share of household labor assumed by the spouse specializing in such labor – i.e., usually the

woman.

The final variable of theoretical interest is age. Although information about age is only available

for the respondent, the respondent’s age is highly correlated with the age of the spouse and thus

serves as a proxy for both. Age does not play any role in efficiency models, except in so far as it

affects labor force participation or is associated with having dependent family members, and we

control for these variables directly. By contrast, age plays a role in bargaining models because it

differentially affects the position of men and women in the re-marriage market. As suggested

above there are two reasons. First, the value of household specific skills deteriorates with age

because they are so closely related to the bearing and rearing of children. Second, age itself

tends to be a liability in the re-marriage market.

The final individual level control is the gender of the respondent because there may be a

tendency for people to exaggerate how much work they do in order to look better in the eyes of

the interviewer. The measurement of macro-level variables, including relative skill abundance

induced by international competition, is described below.

2.2 Empirical Results

The results reported in Table 1 have been pooled across data from our 12 cases, using country

dummies to capture national differences in the division of household labor. The evidence in this

17
table is thus purely for individual-level differences within countries. To explore aggregate-level

effects, we will subsequently drop the dummies in favor of national-level variables for skill

specificity, public sector size, and divorce rates.

Table 1: Individual-level determinants of the gender division of labor

Unpaid work Paid work

(female share) Women Men

Divorce -0.087*** -0.083*** 0.060*** 0.027**

(0.023) (0.027) (0.017) (0.011)
Past absence from paid work 0.155*** 0.175*** -0.555*** -0.021
(0.032) (0.044) (0.023) (0.015)
Male labor force participation 0.253*** 0.225*** - -
(0.025) (0.036)
Female labor force participation -0.173*** -0.130*** - -
(0.017) (0.026)
Male income (log) - 0.033** - -
(0.017)
Female income (log) - -0.065*** - -
(0.014)
Number of dependents 0.026*** 0.035*** -0.012*** 0.028***
(0.007) (0.009) (0.005) (0.003)
Age 0.006*** 0.006*** -0.007*** -0.019***
(0.001) (0.001) (0.001) (0.000)
Gender of respondent 0.207*** 0.204*** -0.023*** -0.049***
(0.015) (0.019) 0.011 (0.007)
Adj. R-squared .191 .178 .239 .464
N 5942 3718 6523 6688
No of countries 12 12 12 12
Note: All models include a full set of country dummies (not shown).
Key: *** p<.01; ** p<.05

The first two columns are for household work, the latter two for paid work. Because there is a lot

of missing data for income, the first column excludes these variables. But it makes little

18
difference to the results. As expected, the probability of divorce decreases the female share of

unpaid work and increases their share of paid work. Divorce also makes men somewhat more

prone to be in paid work, but the effect for women is twice as high. Women in families with one

previously divorced spouse also spend about 1 hour less a week on household work.

Past absence from the labor market to care for children has a strong negative effect on women’s

labor force participation (column 3); but it has virtually no effect on men’s participation. A

woman who has taken off the maximum amount of time for child rearing is predicted not to work

at all whereas someone who has not taken time off is predicted to work at least part-time. Very

few men exit the labor market to care for children (less than 16 percent), and when they do it

tends to be for very brief periods (less than 2 percent have taken full-time leaves). This neither

appears to affect their subsequent participation in paid work, nor to increase their share of

household work. Indeed, men seem to take off work only to the extent that it does not hurt their

career. By contrast, women who have sacrificed work for family in the past end up with a greater

share of the household workload. In other words, taking care of children is a principal

(proximate) cause of the gender division of labor – a finding that is supported by the effects of

the “number of dependents” variable.

Unsurprisingly, labor force participation reduces the share of household work for both spouses,

but more so for men. As we noted above, this is consistent with an interpretation where men are

better able to take advantage of opportunities to acquire specific skills in the labor market.

Interestingly enough, however, this effect is partly (but only partly) offset by a smaller effect of

income for men. In other worlds, women get a greater reduction in household work for every

dollar they earn than men. Why this is so is unclear, but it appears to be at least partly accounted
19
for by a declining marginal effect of earnings on bargaining power (with men earning more than

women on average).

Another consistent result is that age increases the share of work performed by the woman. The

only possible explanation for this effect in an efficiency model is that age is correlated with labor

market participation or the scope of domestic work. Yet, the effect of age is stronger when we

include controls for labor market participation and the number of dependents. Specifically, if we

compare a newly wed couple at age 20 to a married couple at age 40, and controlling for

everything else, the woman in the latter will work about 14 additional minutes a day.6

As noted, this effect of age is consistent with a bargaining perspective because age differentially

affects men and women on the re-marriage market. Yet, it is also consistent with a generational

hypothesis that younger generations have more equitable work norms. The data does not allow

us to distinguish between these interpretations. But if norms have changed over time the next

question is the cause of this change, and the bargaining model in fact has something to say about

that. When outside options are important, and they have become more important over time in

line with the rise in divorce rates, there is reason to expect that parents will raise their daughters

to have more similar tastes for paid work as their sons. This makes daughters less willing to

assume all domestic duties as adults. We consider this a fruitful area for future research.

6
20 years in age is equal to .12 units on the dependent variable, and a unit is equivalent to about 115
minutes of work.

20
To gauge the effect of aggregate-level variables, we substitute the country dummies for measures

of national divorce rates, national abundance of specific skills, and size of the public service

sector. Division of household labor, if the model is right, should be affected by the interaction of

skill specificity (which disadvantages women) and the size of the public sector (which

compensate for such disadvantages). As we move from general to specific skill countries we

would expect the division of household labor to become increasingly inegalitarian, except where

the state steps in to provide jobs in the public sector. This is what is being tested in Table 2.

Public sector size is not included as a component variable because it turns out to have practically

no effect on either dependent variable, yet produces serious problems of collinearity (95 percent

of the variance in the interaction term is explained by its components). With that qualification,

all the relationships turn out as expected. Divorce at the macro-level plays the same role as at the

micro-level, raising women’s paid employment and reducing their share of unpaid work. The

effect is quite strong, increasing female labor force participation by one half of a standard

deviation, which, if we use actual participation data is equivalent to about 5 percent of all

working age women. Like before, labor force participation reduces the share of household work,

but more so for men than for women – partly offset by a stronger effect of income for women.

21
Table 2: Aggregate-level determinants of the gender division of labor

Unpaid work Paid work

(female share) Women Men

Divorce (micro-level variable) -0.098*** 0.045*** 0.021*

(0.027) (0.017) (0.011)
Divorce (macro-level variable) -0.004*** 0.004*** 0.001***
(0.001) (0.001) (0.000)
Past absence from paid work 0.148*** -0.577*** -0.028*
(0.045) (0.023) (0.015)
Male labor force participation 0.244*** - -
(0.035)
Female labor force participation -0.109*** - -
(0.026)
Male income (log) 0.034*** - -
(0.012)
Female income (log) -0.073*** - -
(0.011)
Skill specificity 0.242*** -0.155*** 0.057**
(0.066) (0.034) (0.023)
Public sector * Skill specificity -0.401*** 0.248*** -0.163***
(0.106) (0.056) (0.038)
Number of dependents 0.040*** -0.007 0.033***
(0.009) (0.005) (0.003)
Age 0.008*** -0.007*** -0.019***
(0.001) (0.001) (0.000)
Gender of respondent 0.199*** -0.016 -0.048***
(0.019) (0.011) (0.007)
Adj. R-squared .143 .221 .452
N 3718 6523 6688
No of countries 12 12 12

Key: *** p<.01; ** p<.05

The key new finding in Table 2 concerns the effect of skill specificity. The emphasis on specific

as opposed to general skills in national training systems is measured by an index, which is equal

22
to the mean, after standardization, of vocational training intensity and firm tenure rates.7 We

assume that skills tend to be more specific the more training systems emphasize vocational

training and the longer workers stay with particular firms (see Estevez et al., 2000 and author1,

forthcoming). In turn, relative abundance of specific skills give firms a comparative advantage

that they will exploit in a manner that is disadvantageous to women. The results show that skill

specificity unambiguously hurts the labor market participation of women when the public service

sector is small. The opposite is the case for men who appear to thrive when the production

system relies heavily on specific skills (although the effect is smaller).

For example, in an economy with the smallest public sector in the sample (Japan at 8.6 percent

of GDP), going from an economy with little emphasis on specific skills to one with a high

emphasis, is predicted to reduce female employment by about one third of a standard deviation

on the dependent variable (about 3 percentage of the female working age population), and to

raise unpaid work by between one third and one half of a standard deviation (about 4 hours per

week).

Crucially, however, the relationship between skills and the division of labor is attenuated by a

larger public sector, measured here as public consumption as a percent of GDP (standardized to

vary between 0 and 1 for ease of interpretation). This is illustrated in Figure 1. The relationship

between specificity of skills and a less favorable division of labor for women disappears when

government spending is about one standard deviation above the mean. Clearly, while specific

7
Vocational training intensity is the share of an age cohort in either secondary or post-secondary
(ISCED5) vocational training. Source: UNESCO (1999).Tenure rates are the median length of enterprise tenure in
years, 1995 (Norwegian figure refers to 1991). Sources: OECD Employment Outlook, 1997, Table 5.5. For Norway:

23
skills undermine private sector employment, the government can eliminate the effect, and

potentially reverse it, by increasing the public service sector and the number of women employed

in it.

Figure 1: Skills, public sector, and the division of household labor

4.6 Smallest public sector:

US, Ireland, Japan Japan
3rd largest public sector:
Australia, New Zealand,
4.4 Spain, Italy
Spain
Division of household labor

Italy
2nd largest public sector:
4.2 Ireland UK, Austria, Germany
Austria

4
Australia Germany
Netherlands
New Zealand UK

3.8 Largest public sector:

Canada Canada, Norway, Sweden
US NorwaySweden

3.6
0 0.2 0.4 0.6 0.8 1
Specific skill emphasis

3. Gender policy preferences

A key issue for any theory of gender preferences over public policy is the question of whether

OECD Employment Outlook, 1993, table 4.1.

24
the family can be treated as a unitary actor or as separate individuals. In Becker’s efficiency

model the interests of family members are fully aligned and so, one should expect, are their

political preferences and attitudes. Political conflict over family and other social policies will

tend be along lines other than gender. For example, families with incomes below the mean, by

the standard Meltzer-Richard argument, would have an incentive to favor policies that are

redistributive (Meltzer and Richard 1981).

The possibility of a gender gap in political preferences emerges when marriage contracting is

incomplete and termination of the contract is an ever present possibility. In this case spouses

will have conflicting preferences over who receives family benefits, and they will differ over any

policies that affect their outside options. This is so not merely, or even primarily, because they

could be forced one day to take the outside option, but also, as we have argued, because outside

options affect the current bargaining power inside the family.

Starting in the late 1970s in the U.S. and Scandinavia, and some years thereafter in many other

European countries, women have in fact begun moving out of sync with their husbands in their

voting behavior, often voting to the left of men in aggregate. Women tend to support activist

government across a range of economic policies (Shapiro and Mahajan 1986; Ladd 1997;

Greenberg 2000; Alvarez and McCaffery 2000). This move is striking, because in what

Inglehart and Norris term the “traditional gender gap,” women typically voted to the right of men

in these countries, perhaps because their greater longevity put them in greater numbers in the

most conservative age bracket; and perhaps because of their social role as protector of family

values and perhaps resulting tendency to be more religious (Inglehart and Norris 1999 and 2002;

25
Studlar, McAllister, and Hayes 1998).

There are several competing explanations for “the modern gender gap,” where women’s

preferences and voting patterns appear to be moving to the left. Some scholars argue that women

are more altruistic than men, and they therefore favor more welfare spending (Conover 1988;

Welch and Hibbing 1992; Gidengil 1995). But this argument is a static one that fails to explain

the change in voting behavior over recent decades. Other scholars have pointed out that women

are more likely than men to be economically vulnerable (Tedin 1994; Sears and Citrin 1982).

But survey research suggests that women throughout the wage distribution are more likely to

vote left than their male counterparts (Goertzel 1983; Carroll 1988; Inglehart and Norris 1999;

2002). Edlund and Pande (2001) get around this problem by arguing that high divorce rates

leave women at a higher risk of income loss than before, and that they are therefore voting for

more redistribution even before they receive it, as a sort of insurance. Using variation in divorce

rates across U.S. states and some European countries, they derive a measure of “divorce risk”

and find that it corresponds with the likelihood of women voting farther left than their socio-

economic status warrants.

If Edlund and Pande are right that women vote left as insurance against post-divorce poverty,

we would expect that women staying out of the work force are at the greatest risk and hence

most likely to vote left. What the data suggest instead, however, is that women in the work force

are more likely to vote left than housewives (Greenberg 2000). This would seem to damage the

insurance argument, since these women have already reduced their economic exposure to the

26
possibility of divorce.8

The data suggest to us an alternative explanation based on household bargaining: working

women gain bargaining power at home from the partial socialization of family work such as

child care and elderly care, and these are precisely the sorts of policies that parties on the left are

more likely to espouse. The logic is that with some of her family burden lifted by the public

purse, a woman is better able to invest in her marketable skills. By raising her level of economic

independence closer to her husband’s, a wife reduces her stake in keeping the relationship going

closer to his level. We should observe more equal shares of family work in the household, not

only because the government is undertaking part of it, but also because a woman is less willing

to give up increasing amounts of her time to keep the marriage from dissolving.

As soon as outside options matter for bargaining power, men and women will differ over policies

that affect these options. The most obvious matter of disagreement, perhaps, is over publicly

subsidized daycare. Since women are much more likely to end up as primary care givers, their

welfare is disproportionately affected by the availability of high quality, low-cost daycare. Men

may prefer to spare the public purse and hence their tax bill if their wives are default child care

givers. This logic also applies to public care for the elderly and the sick because it helps women

escape some of their traditional duties and thereby permit more time to be spent in paid

employment. In addition, as we have stressed throughout, the welfare state is an important source

of employment for women precisely because so many of the jobs replace caring functions that

8
One could argue, of course, that there is a selection effect here: only the women who feel at the
greatest risk will seek outside employment; and that their resulting outside remuneration only
partially offsets their perceived risk.
27
are otherwise provided “for free” in the family.

Men and women are also likely to disagree about a range of social transfer programs. First,

research based on the Luxembourg Income Study has shown that transfers typically result in a

reduction of inequality (Bradley 2002), and since women are on average paid less than men they

tend to have a stronger interest in redistributive spending (assuming again that they care about

their outside options).9 Social transfers are also important in so far as they allow women

temporary career interruptions without significant loss of income. As Estevez-Abe (1999) has

argued, social protection for women involves two factors that are not equally relevant for men:

(i) protection against dismissal during and after pregnancy, such as maternity, parental and

family leave policies; and (ii) income maintenance during leaves and guarantees of reinstatement

to the same job at the same wage level upon return to work.

Support for redistribution and social insurance obviously does not come from women alone. In

the classic Meltzer-Richard setup, any person with an income below the mean will prefer at least

some redistributive spending. And when an insurance motive is added to the model, those

exposed to greater risks will also demand more spending. One key source of such risks is the

transferability of workers’ skills. The harder it is to transport skills from one job to another, the

9
The gap in income between men and women may be due to “statistical discrimination” where
employers pay women less than men because women are on average more likely than men to
leave the labor market for purposes of child birth and rearing, and because women are more
likely than men to trade off working time for time on domestic duties, including care for sick
family members (see Daly 1994; Rubery, Fagan and Maier 1996). There is a vicious circle here
because the less well paid women are the weaker their bargaining position in the family and the
more time they will spend on household duties to the detriment of their attractiveness to
employers.

28
greater the importance of income protection though social insurance programs (guaranteed health

care, pensions, unemployment benefits, job security, etc). This argument implies that position in

the international division of labor matters for the size and structure of the welfare state – but

through a mechanism that is not identified in existing arguments about the relationship between

the international economy and social protection (see Garrett 1998; Rodrik 1998).

This mechanism also has implications for understanding gender preferences for social protection.

As suggested above, the reason is that women will find it harder to enter into occupations that

require extensive specific skills (Estevez-Abe 1999). For a woman to invest in specific skills she

has to be assured that potential career interruptions, if temporary, will not lead to dismissal or

reduce her wage level in the long run. A high probability of dismissal reduces the incentives to

acquire firm-specific skills. Likewise, a high probability of reduction in wages after becoming a

mother reduces the incentives to invest in either firm-specific or industry-specific skills (Estevez-

Abe 2002).

The key implication of our argument is that women at any given level of income and skill

specificity will prefer higher social protection than men. This gender gap is magnified by an

indirect effect through income since women earn less than men. On the other hand, the effect is

reduced to the extent that women invest more in general than in specific skills. An interesting

corollary of this argument is that women should be more supportive than men of public

investment in general education. Inexpensive access to good formal education presumably

benefits women disproportionately because they have a comparative advantage in general skills.

We leave this as a hypothesis for future research.

29
An important qualification to our argument is that the gap in gender preferences depends on the

extent to which women participate in the labor market, as well as on the probability of divorce.

Because non-working women’s welfare depend more on the income of men than is the case for

working women’s, they have a stronger incentive to support policies that raise the take-home pay

of males. Non-working women will still care about their outside options, as argued above, but

policies that reduce the relative wage of men also reduce the income of families where the

woman does not work. What makes divorce important is that it raises the salience of outside

options, and hence sharpens differences in policy preferences between the sexes. Since both

female labor force participation rates and divorce rates vary across countries, the macro-level

implication is that the gender gap will vary accordingly.

Using the logic developed in this section we can revisit some claims that are sometimes made

about the gender and political preferences. Orloff (1993; 1999) and O’Connor et al. (1999), for

example, strongly suggest that women are most disadvantaged in countries, such as those in

southern Europe and East Asia, where female labor force participation rates are low,

stratification on the labor market high, and the distribution of domestic work very unequal. If

access to paid work and the ability to form autonomous households are fundamental interests of

women, as Orloff and others argue, one would expect gender conflicts to be most intense in these

countries. Yet, as we will see below these are countries in which the policy preferences of men

and women appear the most similar, and where there does not appear to be a strong gender gap

in electoral politics. An explanation for this puzzle is that the family as an institution is heavily

protected through labor market conditions, and reinforced by legislation and norms against

30
divorce. The likelihood of a first marriage ending in divorce in Italy is less than one in 10—even

lower than the 1950s US. As we have argued, if divorce is a highly unlikely prospect, men and

women are much less likely to adopt conflicting policy preferences.

Another recent controversy surrounds the role of the public-private sector division in

Scandinavia. According to some, this division--which concerns issues of public sector size,

relative pay, and public sector job protection--has emerged as a salient cleavage in electoral

politics. The high gender segregation in the public sector also helps explain a widening gender

gap. Paul Pierson points out that since men in the private sector tend to be married to women in

the public sector, there is no compelling reason that spouses should quibble over issues of

relative pay (Pierson 2000, 807). At the end of the day, the income of both spouses simply adds

to family income. But this logic only applies when husband and wife have few reasons to

concern themselves with outside options. And since pay in the public sector is financed by taxing

the private sector, policies affecting relative pay are a perfect example of an area where gender

conflict is likely to be intense.

3.1 Empirical Analysis of Gender Preferences

To test our hypotheses we turn to the 1996 International Social Survey Program on the role of

government. The data contain a number of questions about government spending and social

policy as well as information on the key independent variables. We have more or less complete

data for 11 advanced democracies at the individual level (Australia, Britain, Canada, France,

Germany, Ireland, Italy, Norway, New Zealand, Sweden, and United States) and for 13 (adding

Japan and Spain) at the macro level.

31
We use three dependent variables. The first is an additive index intended to tap preferences for

social transfers and based on three questions about whether the respondent want more or less

government spending on a) unemployment benefits, b) health care, and c) pensions. The variable

ranges from 1 to 5, where 5 means a strong preference for more spending. The second variable is

an additive index based on three questions about the role of the government in providing

employment: a) should the government finance projects to create new jobs?, b) should the

government reduce the working week to produce more jobs?, and c) should the government be

responsible for providing jobs for all who wants to work? The variable ranges from 1 to 5 where

5 indicates a strong preference for job creation. The final variable is based on a single question

that asks whether the respondent wants the government to spend more or less money on public

education. Like the other two variables, this one varies between 1 and 5 with the latter indicating

strong support for additional spending.

The key independent variables are gender (0=male; 1=female), income (logged), skill specificity,

and labor market participation (as defined in the previous section). We also control for public

sector employment, but the results for this variable are presented separately because data are not

available for the US. Skill specificity is based on the ISCO job classifications, the respondent’s

formal education, and a question about the difficulty the respondent would face if having to find

a similar job. Higher values mean that the respondent’s specific skills are extensive relative to his

or her general skills. The measurement strategy is described in detail in author1 (2001). The

regressions also include controls for age and unemployment, and as in the previous section we

use pooled regressions with a full set of country dummies (not shown).

32
3.2 Results

Model 1 in Table 3 shows the results of a simple linear model predicting support for social

transfers. All variables have the expected signs and, with the exception of labor force

participation, all effects are statistically significant at a .01 level or better. Most importantly,

women are more supportive than men of social transfers (i.e., want to increase them more, or

decrease them less). On average, the effect is equivalent to 20 percent of a standard deviation on

the dependent variable. Unfortunately, it is very hard to give substantive meaning to this number

because we do not know how any particular level of agreement with any of the questions map

onto actual monetary commitments. We do know, however, that gender is one of the best

predictors of preferences and that there is a statistically significant gender gap in each one of our

11 countries.

In model 2 we have included an interaction term for gender and labor force participation. Doing

so creates problems of collinearity, with 83 percent in the variance in the gender variable being

explained by the other variables. Still, the difference between men and women who work is

statistically significant, and the substantive results make perfect sense in terms of the theoretical

argument. Thus, when women are not working (the value on the labor force participation

variable is zero), their predicted preferences are no different than men’s. When a woman is

working full time, by contrast, the difference between men and women is .164, which is greater

than the direct effect in the non-interactive model. In other words, women vote together with

men, and presumably in the collective interest of the family, when they are not working. But

33
 Table 3. Social preferences and gender

Public employment Social transfers Education

1 2 3 4 5 6 7 8 9
Female 0.158** 0.052 0.138** 0.116** 0.048 0.100** 0.107** 0.088* 0.091**
(0.019) (0.041) (0.020) (0.016) (0.035) (0.017) (0.020) (0.043) (0.021)
LF participation 0.052 -0.025 0.028 0.057* 0.007 0.058* 0.016 0.002 0.030
(0.032) (0.042) (0.034) (0.027) 0.035 (0.028) (0.034) (0.044) (0.035)
Female*LF part - 0.137** - - 0.088* - - 0.025 -
(0.047) 0.040 (0.049)
Income (log) -0.003** -0.003** -0.003** -0.002** -0.002** -0.002** -0.000 -0.000 -0.000
(0.000) (0.000) (0.000) (0.000) 0.000 (0.000) (0.000) (0.000) (0.000)
Skill specificity 0.115** 0.115** 0.126** 0.128** 0.129** 0.123** -0.001** -0.001 0.001
(0.012) (0.012) (0.012) (0.010) 0.010 (0.010) (0.012) (0.012) (0.013)
Age -0.001* -0.002** -0.001 0.003** 0.003** 0.004** -0.004 -0.004** -0.003**
(0.001) (0.001) (0.001) (0.001) 0.001 (0.001) (0.001) (0.001) (0.001)
Unemployment 0.177** 0.152** 0.184** 0.102* 0.086 0.099* 0.044 0.040 0.056
(0.055) (0.056) (0.057) (0.047 0.047 (0.047) (0.057) (0.058) (0.059)
Public sector - - 0.198** - - 0.044** - 0.098**
(0.020) (0.017) (0.021)
Adj. R-sq .188 .189 .191 .187 .197 .211 .068 .068 .069
N 7933 7933 7129 8140 8042 7311 8260 8260 7417

Key: ** p<.01; * p<.05

Note: Regressions included a full set of country dummies

when women enter the labor market, they express social policy preferences that are clearly

distinct from men’s. This is also true for social transfers (models 5 and 8), and the magnitudes of

the effects are very similar (but again with the qualification that it is difficult to give substantive

meaning to the numbers). The results are not sensitive to control for public sector employment

(columns 3, 6, and 9), but public sector employment not surprisingly raises the support for

spending.

The other two relationships that emerge from the regressions on social policy preferences are the
34
negative effect of income, as predicted by the Meltzer-Richard model, and the positive effect of

skill specificity, as predicted by the asset model. These variables, however, do not appear to

matter much in explaining support for educational spending (model 5 and 6) where age and

gender are the most important predictors. Again, an explanation for the gender effect that is

consistent with our argument is that women have a stronger incentive than men to invest in

general skills, hence also a greater interest in public subsidization of education. The alternative,

or perhaps complementary, explanation is that women are biologically forced to invest more in

their children, and manifest this by spending more resources on them (Hrdy 1999; Schultz

2000).10

In Table 4 we examine two predicted indirect effects of gender. Model 1 uses income as the

dependent variable and shows that women, not surprisingly, earn less than men. On average they

make 652 dollars less than men (in 1996), and this figure roughly doubles if we also take into

account that women participate less in the labor market. Since lower income translates into

greater support for social spending, the effect of gender is obviously magnified by income.

However, about half of this effect is canceled out by another indirect effect: the greater

propensity of women to invest in general skills (model 2). Both effects are, of course, entirely

consistent with the theory.

10
The most likely explanation for the effect of age is that the older you are the less you will
benefit from spending on education.

35
Table 4. Skill specificity, income, and gender

Income Skill specificity

Female -23.59 -0.287
(0.95) (0.016)
LF participation 56.53 -0.217
(1.62) (0.025)
Skill level 16.71 -
(0.47)
Age 0.67 0.004
(0.04) (0.001)
Unemployment 11.16 0.193
(3.32) (0.048)
Adj. R-sq .470 .435
N 6830 9610
Note: Regressions included a full set of country dummies

Exploring the macro level implications of our argument is hampered by the limited number of

cases (13), as well as by high colinearity between divorce and labor force participation rates

(.90). However, the relationship between divorce rates and the size of the gap in gender

preferences (measured as an average across issues) is indeed positive, with a correlation

coefficient of .6. The same is true for labor force participation and the gender gap (r=.5). More

specifically one cluster of countries, consisting of Ireland, Italy, Japan and Spain have relatively

low divorce and female participation rates, with small differences in gender preferences, whereas

another, consisting of Canada and the Scandinavian countries, have high divorce and

participation rates, with relatively large gender differences in preferences. We know from the

analysis in the previous section that female labor force participation is explained by the

interaction of countries’ position in the international division of labor (skill specificity) and the

size of the public sector, while the probability of divorce affects the importance women assign to

public policies that strengthen their outside options. That divorce, labor force participation, and

the gender gap are related is therefore not surprising.

36
4. Conclusions

Explaining cross national variation in income inequality has been one of the greatest

preoccupations of modern political economy. But much of this analysis masks, we have argued,

inequality within the very unit of analysis that is typically taken for granted: the family. When

we abandon the traditional assumption of the family as a welfare maximizing unit, we confront

the reality of strategic interaction between spouses. Because a spouse might favor his or her

share of family welfare even at some expense of the total family welfare, it is important to

disaggregate the family to understand the effects of economic institutions and the public policies

that govern them.

In this paper we examined two areas where the assumption of family as a unit can lead us astray:

the household division of labor, and political preferences. First, we join a growing chorus of

social scientists challenging the idea that the household division of labor reflects an efficient

allocation of family resources (Lundberg and Pollak 1996, 2001; Braunstein and Folbre 2001).

A husband may resist his wife’s outside employment, even if it could increase total family

income (or more broadly, family utility including children’s well being), because her

accumulation of market skills and experience broadens her exit options to the marriage. By

ramping up her bargaining power within the marriage, this greater economic independence can

result in the husband contributing more and receiving less in the way of unpaid work in the

home.

Economists have already noted, and we have found as well, that female labor force participation

37
and higher female income do in fact shift the burden of household work a bit farther onto men’s

shoulders. But economists do not relate this effect to differences in national institutions, and the

way these institutions are embedded in the international economy. In particular, we have argued

that institutions that put a premium on the accumulation of specific skills hurt women’s ability to

gain equality in household work. Because women in specific skills economies typically bear a

bigger penalty for career interruptions such as for child rearing, they face more limited work

opportunities and may invest less in their market-relevant education as a result. This, in turn,

weakens their bargaining power at home, and they get stuck sweeping floors more of the time

than their counterparts in economies that specialize in general skills. This occurs unless, as we

have argued, the state steps in and adopt policies to deliberately counter the disadvantages of

women in specific skills countries. This is where the gender gap in preferences enter the story.

Given the overall trend towards more women in the workforce, we are not surprised to find that

women as a group seem to be moving to the left politically. Although women actually vote as a

group to the left of men only in a few countries (most dramatically in the US), time series

analysis of women’s political preferences show a systematic move leftwards in rich democracies.

Given that women were starting from a position to the right of men, and given that the

percentage of women fully integrated into the labor force still lags in many countries, it may take

some years before women are actually positioned to the left of men across the board.

In terms of social policy preferences, however, women have already arrived. Our data show

unambiguously that women desire more spending on transfers, public employment creation, and

education. The reason, we argue, is that partial socialization of family work, even at the cost of

38
higher taxes from the private sector, increases a woman’s ability to work outside the home and

thereby increases her exit options and her household bargaining position. Public investment in

formal education also benefits women disproportionately because they have a comparative

advantage in general skills.

Political economists have largely ignored the politics of gender and the family, but we hope this

paper demonstrates the potential for comparative and international political economy to answer

the numerous remaining questions in this rich and largely underexplored area of inquiry. A

particularly promising area of inquiry, we believe, is changing gender roles in childhood

socialization. Political economists rarely venture into this area, yet their models potentially have

much to contribute.

39
References

Allen, Douglas. 1998. No-fault Divorce in Canada: Its Cause and Effect. Journal of Economic
Behavior and Organization, 37: 129-49.

Alvarez, R. Michael, and Edward McCaffery. 2000. Is There a Gender Gap in Fiscal Political
Preferences? Paper presented at the 2000 Annual Meetings of the American Political Science
Association.

Anderson, Deborah, Melissa Binder, and Kate Krause. Forthcoming. The Motherhood Wage
Penalty Revisited: Experience, Heterogeneity, Work Effort, and Work Schedule Flexibility.
Industrial and Labor Relations Review.

Andersson, Gunnar, Guiping Liu, and Andres Vikat. 2003. Impact of Spouses’ Relative Income
on Divorce Risk: A Register-Based Study of First Marriages in Sweden. Working Paper, Max
Planck Institute for Demographic Research.

Becker, Gary. 1964. Human Capital. New York: Columbia University Press.

Becker, Gary. 1965. A Theory of the Allocation of Time. Economic Journal 75(299): 493-517.

Becker, Gary. 1971. The Economics of Discrimination. Chicago: University of Chicago Press.

Becker, Gary. 1981. A Treatise on the Family. Cambridge: Harvard University Press.

Becker, Gary. 1985. Human Capital, Effort, and the Sexual Division of Labor. Journal of Labor
Economics 3(1/2): S33-S58.

Becker, Gary, Elisabeth Landes, and Robert Michael. 1977. An Economic Analysis of Marital
Instability. Journal of Political Economy. 85/6: 1141-87.

Becker, Gary, and K. Murphy. 1988. The Family and the State. Journal of Law and Economics
31: 1-18.

Bentzen, Jan, and Valdemar Smith. 2000. An Empirical Analysis of the Effect of Labour
Market Characteristics on Marital Dissolution Rates. Working Paper, Aarhus School of Business.

Bittman, Michael, and Paula England, Nancy Folbre, and George Matheson. 2001. When
Gender Trumps Money: Bargaining and Time in Household Work. Paper presented at the
Population Association of America conference, March 2001.

Blau, Francine and Kahn, Lawrence. 1992. The Gender Earnings Gap: Learning from
International Comparisons. American Economic Review 80(2): 533-39.

Box-Steffensmeier, Janet, Suzanna DeBoef, and Tse-Min Lin. 1997. Macroideology,

40
Macropartisanship, and the Gender Gap. Paper presented for presentation at the 1997 Annual
Meetings of the American Political Science Association, Washington D.C., August 1997.

Bradley, David, Evelyne Huber, Stephanie Moller, François Nielsen, and John
Stephens. Forthcoming. Distribution and Redistribution in Post-Industrial
Democracies. World Politics.

Bremmer, Dale, and Randall Kesselring. 1999. The Relationship Between Female Labor Force
Participation and Divorce: A Test Using Aggregate Data. Paper presented at the International
Atlantic Economic Conference, Montreal, Canada, October 1999.

Bremmer, Dale, and Randy Kesselring. 2003. Divorce and Female Labor Force Participation:
Evidence from Time Series Data and Cointegration. Working Paper, Rose-Hulman Institute of
Technology.

Burns, Nancy, and P. Schumaker. 1988. Gender Cleavages and the Resolution of Local Policy
Issues. American Journal of Political Science. 32/4: 1070-95.

Carroll, Susan. 1988. Women’s Autonomy and the Gender Gap: 1980 and 1982. in Carol
Mueller, ed., The Politics of the Gender Gap: The Social Constructionof Political Influence.
Sage Press. 236-257.

Chiappori, Pierre-Andre. 2002. Marriage Market, Divorce Legislation, and Household Labor
Supply. Journal of Political Economy 110/1: 37ff.

Conover, Pamela Johnston. 1988. Feminists and the Gender Gap. Journal of Politics 50/4: 985-
1010.

Conover, Pamela Johnston and Virginia Sapiro. 1993. Gender, Feminist Consciousness, and
War. American Journal of Political Science 37: 1079-99.

Cook, Elizabeth Adell, and Clyde Wilcox. 1991. Feminism and the Gender Gap: A Second
Look. The Journal of Politics. 53/4: 1111-1122.

Edlund, Lena, and Rohini Pande. 2001. Why Have Women Become Left-Wing? The Political
Gender Gap and the Decline in Marriage. Working Paper, Columbia University Department of
Economics.

Esping-Andersen, Gosta. 1990. The Three Worlds of Welfare Capitalism. Polity Press.

Esping-Andersen, Gosta. 1999. Social Foundations of Postindustrial Economies. Oxford

University Press.

Estevez-Abe, Margarita. 1999. Comparative Political Economy of Female Labor Participation.

paper prepared for presentation at the annual meetings of the American Political Science
Association, Atlanta, September 2-5.

41
Estevez-Abe, Margarita. 2002. Gendering Varieties of Capitalism. Paper prepared for a
conference on The Political Economy of Family and Work, Yale University, July.

Estevez-Abe, Margarita, Torben Iversen, and David Soskice. 2000. Social Protection and the
Formation of Skills: A Reinterpretation of the Welfare State. In Peter Hall and David Soskice,
eds., Varieties of Capitalism: The Institutional Foundations of Comparative Advantage. Oxford
University Press.

Gainous, Jason. 2002. Is There a “Woman’s Perspective?” An Exploration of Gender

Differences Along Republican and Conservative Lines. Paper presented at the annual meeting of
the Southwestern Political Science Association, New Orleans, March 2002.

Garrett, Geoffrey. 1995. Capital Mobility, Trade and the Domestic Politics of Economic Policy.
International Organization, 49, 4, 657-87.

Glendon, Mary Ann. 1987. Abortion and Divorce in Western Law. Harvard University Press.

Goertzel, Ted George. 1983. The Gender Gap: Sex, Family Income, and Political Opinions in
the Early 1980s. Journal of Political and Military Sociology. 11/Fall: 209-222.

Goode, William. 1993. World Changes in Divorce Patterns. Yale University Press.

Greene, William and Aline Quester. 1982. Divorce Risk and Wives’ Labor Supply Behavior.
Social Science Quarterly 63/1: 16-27.

Greenberg, Anna. 2000. Deconstructing the Gender Gap. Working Paper, John F. Kennedy
School of Government, Harvard University.

Hansen, Jorgen; and Roger Wahlberg. 2000. Occupational Gender Composition and Wages in
Sweden. IZA Institute for the Study of Labor, Discussion Paper No. 217.

Hrdy, Sarah Blaffer. 1999. Mother Nature: A History of Mothers, Infants, and Natural
Selection. Pantheon Books.

Huber, Evelyne, Leonard Ray, and John Stephens. 1999. The Welfare State in Hard Times. In
Kitschelt et al., Continuity and Change in Contemporary Capitalism. Cambridge University
Press.

Inglehart, Ronald, and Pippa Norris. 1999. The Developmental Theory of the Gender Gap:
Women and Men’s Voting Behavior in Global Perspective. Working Paper, John F. Kennedy
School of Government, Harvard University.

Inglehart, Ronald and Pippa Norris. 2002. The Gender Gap in Voting and Public Opinion.
Chapter 4 in Rising Tide, book ms.

42
Johnson, W. R., and J. Skinner. 1986. Labor Supply and Marital Separation. American
Economic Review 76: 455-69.

Kitschelt, Herbert, Peter Lange, Gary Marks, and John Stephens, eds. 1999. Continuity and
Change in Contemporary Capitalism. Cambridge University Press..

Librizzi, Christopher J. 2003. Bargaining and the Division of Household Labor: A Cross-
National Analysis of Individual- and National-Level Factors. Honors Thesis, Department of
Government, Harvard College.

Lombardo, Karen. 1999. Women’s Rising Market Opportunities and Increased Labor Force
Participation. Economic Inquiry. 37/2: 195-212.

Mancinelli, I., L. Ceciarelli, A. Comparalli, S. Lazanio, P. Girardi, R. Tatarelli. 2001. Suicide in

Adolescents in Italy. Canadian Journal of Psychiatry. November.

Mansbridge, Jane. 1985. Myth and Reality: The ERA and the Gender Gap in the 1980 Election.
Public Opinion Quarterly. 49/2: 164-178.

Meltzer, Allan H. and Scott. F. Richards. 1981. A Rational Theory of the Size of Government.
Journal of Political Economy 89 (5): 914-27.

Mincer, Jacob. 1958. Investment in Human Capital and Personal Income Distribution. Journal
of Political Economy. 66(4): 281-302.

Mincer, Jacob. 1966. Labor Force Participation and Unemployment: A Review of Recent
Evidence. In Prosperity and Unemployment, ed. Robert A Gordon and Margaret S. Gordon.
New York: Wiley.

Mincer, Jacob. 1978. Family Migration Decisions. Journal of Political Economy. 86(5): 749-
773.

Mincer, Jacob. 1985. Intercountry Comparisons of Labor Force Trends and Related
Developments: An Overview. Journal of Labor Economics. 3/1: S1-32.

Nock, Steven and Margaret Brinig. 2002. Weak Men and Disorderly Women Divorce and the
Division of Labor. In Anthony Dnes and Robert Rowthorn, eds. Marriage and Divorce: A Law
and Economics Approach. Cambridge University Press.

Orloff, Ann Shola. 1993. Gender and the Social Rights of Citizenship: The Comparative
Analysis of Gender Relations and Welfare States. American Sociological Review, 58, 303-328.

O’Connor, Julia, Ann Orloff, and Sheila Shaver. 1999. States, Markets, Families: Gender,
Liberalism and Social Policy in Australia, Canada, Great Britain, and the United States.
Cambridge University Press.

43
Orloff, Ann Shola. 1999. Motherhood, Work, and Welfare: Gender Ideologies and State Social
Provision in Australia, Britain, Canada and the United States. in George Steinmetz, ed.,
State/Culture. Cornell University Press.

Pierson, Paul. 2000. Three worlds of welfare state research. Comparative Political Studies 33 (6-
7), 791-821.

Polachek, Solomon. 1975. Differences in Expected Post-School Investment as a Determinant of

Market Wage Differentials. International Economic Review 16(2): 451-470.

Polachek, Solomon. 1978. Simultaneous Equations Models of Sex Discrimination. In Income

Inequality, ed. John R. Moroney. Lexington: Lexington Books.

Polachek, Solomon. 1981. Occupational Self-Selection: A Human Capital Approach to Sex

Differences in the Occupational Structure. Review of Economics and Statistics 63(1): 60-69.

Polachek, Solomon. 1985. Occupational Segregation: A Defense of Human Capital Predictions.

Journal of Human Resources. 20(3): 437-40.

Pontusson. Forthcoming. Social Europe versus Liberal America. Cornell University Press.

Posner, Richard. 1992. Sex and Reason. Harvard University Press.

Rodrik, Dani. 1998. Why Do More Open Economies Have Larger Governments. Journal of
Political Economy 106 (5)(October).

Schlozman, Kay Lehman, Nancy Burns, Sidney Verba, and Jesse Donahue. 1995. Gender and
Citizen Participation: Is There a Different Voice? American Journal of Political Science. 39:
267-93.

Sears, David, and J. Citrin. 1982. Tax Revolt: Something for Nothing in California. Cambridge:
Harvard University Press.

Shapiro, David and Lois Shaw. 1985. Growth in Supply Force Attachment of Married Women:
Accounting for Changes in the 1970s. Southern Economic Journal. 6/3: 307-29.

Schultz, Paul. 2000. Using Community Data to Improve Analysis of How Child Well-being is
Produced. In Arland Thornton, ed., The Well Being of Children and Families: Research and
Data Needs. University of Michigan Press.

Smith, Ian. 2002. European Divorce Laws, Divorce Rates, and Their Consequences. In Anthony
Dnes and Robert Rowthorn, eds., The Law and Economics of Marriage and Divorce. Cambridge
University Press.

Studlar, Donley, Ian McAllister, and Bernadette Hayes. 1998. Explaining the Gender Gap in
Voting: A Cross-National Analysis. Social Science Quarterly. 79/4: 779-798.

44
Sullivan, Oriel, and Jonathan Gerhuny. 2001. Cross-National Changes in Time-Use: Some
Sociological (Hi)stories Re-Examined. British Journal of Sociology. 42(2): 331-47.

Tsubouchi, Yoshihiro, and Reiko Tsubouchi. 1970. Rikon: Hikaku shakaigakuteki kenkyu
[Divorce: A Comparative Sociological Study]. Tokyo: Sobunsha.

Waite, Linda J. and Mark Neilsen. 1999. The Decision to Allocate Time Between Market and
Non-Market Activities. Paper presented at the Workshop on Measurement of and Research on
Time Use, National Research Council, May 1999.

Welch, Susan and John Hibbing. 1992. Financial Conditions, Gender, and Voting in American
National Elections. Journal of Politics 54: 197-213.

Wolfers, Justin. 2001. Comment on Edlund and Pande. Wallis Conference on Political
Economy. September 29, 2001.

Wolfers, Justin. 2002. Did Unilateral Divorce Laws Raise Divorce Rates? A Reconciliation and
New Results. Working paper, Stanford Graduate School of Business.

Zanatta, L. 1997. Le Nuove Famiglie. Bologna: Il Mulino.

45