Introduction

Framing
effects in
decision-
making under
risk

Bundling and
mental
accounting

The Allais
Decision-Making Under Risk and Uncertainty
problem and
the sure-thing
principle

The Ellsberg
problem and
ambiguity College of Business and Economics
aversion

Probability
weighting

Discussion

1 / 53
 Introduction

Introduction

Framing • The expected utility theory, which we explored in chapter

effects in
decision- 6, is widely used. Yet there are situations in which people
making under
risk fail to conform to the predictions of the theory. In
Bundling and addition, there are situations in which it is seemingly
mental
accounting rational to violate it.
The Allais
problem and
• In this chapter we explore some such situations. We will
the sure-thing
principle also continue exploring what behavioral economists do in
The Ellsberg the face of systematic deviations from standard theory.
problem and
ambiguity • To capture the manner in which people actually make
aversion

Probability
decisions under risk, we will make more assumptions about
weighting the value function, which was introduced in chapter 3. We
Discussion will also introduce the probability-weighting function. Both
of these functions are essential parts of prospect theory,
the most prominent behavioral theory of choice under risk.
2 / 53
 Introduction

Introduction
2019-09-27
• The expected utility theory, which we explored in chapter
6, is widely used. Yet there are situations in which people
fail to conform to the predictions of the theory. In
addition, there are situations in which it is seemingly
rational to violate it.
• In this chapter we explore some such situations. We will

Introduction also continue exploring what behavioral economists do in

the face of systematic deviations from standard theory.
• To capture the manner in which people actually make
decisions under risk, we will make more assumptions about
the value function, which was introduced in chapter 3. We
will also introduce the probability-weighting function. Both
of these functions are essential parts of prospect theory,
the most prominent behavioral theory of choice under risk.

Distribute the survey handouts for section 7.2

 Asian Pandemic Problem 1

Introduction

Framing
effects in
decision-
making under
risk

Bundling and • There is an impending pandemic threatening to kill 600

mental
accounting people and you have two options:
The Allais (A) Save 200 people
problem and
the sure-thing (B) Save 600 people with 1/3 chance and save 0 people with
principle 2/3 chance.
The Ellsberg
problem and • When this problem was first presented to participants, 72
ambiguity
aversion percent chose A and 28 percent chose B.
Probability
weighting

Discussion

3 / 53
 Asian Pandemic Problem 2

Introduction

Framing
effects in
decision-
making under
risk

Bundling and • The same disease is threatening society, but you are given
mental
accounting two different options this time:
The Allais (C) Kill 400 people
problem and
the sure-thing (D) Kill 0 people with 1/3 chance and kill 600 people with 2/3
principle chance.
The Ellsberg
problem and • When this problem was first presented to participants,
ambiguity
aversion only 22 percent chose C and 78 percent chose D.
Probability
weighting

Discussion

4 / 53
 Contradiction in the Responses to the Asian
Pandemic Problems
Introduction
• There seems to be a contradiction between the outcomes
Framing
effects in of these two problems.
decision-
making under • Superficial differences aside, option A is the same as
risk
option C (both are safe), and option B is the same as
Bundling and
mental
option D (both are risky).
accounting • As long as you act in accordance with expected-utility
The Allais
problem and
theory, you will prefer the safe option (A or C) no matter
the sure-thing how it is described, or you will prefer the risky option (B
principle
or D) no matter how it is described, or you will be
The Ellsberg
problem and indifferent between the two. Your preference should not
ambiguity
aversion
depend on how the options are described.
• However, in the first case, people seem to prefer the safe
Probability
weighting choice over the risky choice (A to B) and in the second
Discussion case they prefer the risky choice over the safe choice (D to
C).
• The next two slides demonstrate graphically this
contradiction.
5 / 53
 Contradiction in the Responses to the Asian
Pandemic Problems
Introduction

Framing A risk-averse individual will chose option A (or C) regardless of

effects in
decision- the way the problem is framed.
making under
risk

Bundling and
mental
accounting

The Allais
problem and
the sure-thing
principle

The Ellsberg
problem and
ambiguity
aversion

Probability
weighting

Discussion

6 / 53
 Contradiction in the Responses to the Asian
Pandemic Problems
Introduction

Framing
A risk-prone individual will chose option B (or D) regardless of
effects in
decision-
the way the problem is framed.
making under
risk

Bundling and
mental
accounting

The Allais
problem and
the sure-thing
principle

The Ellsberg
problem and
ambiguity
aversion

Probability
weighting

Discussion

7 / 53
 Contradiction in the Responses to the Asian
Pandemic Problems
Introduction

Framing • So how do we account for the contradiction above?

effects in
decision- • The key is to notice that the behavior can be interpreted
making under
risk in terms of framing.
Bundling and
mental
• As you will recall, framing effects occur when preferences
accounting
and behavior are responsive to the manner in which the
The Allais
problem and options are described, and in particular to whether the
the sure-thing
principle options are described in terms of gains or in terms of
The Ellsberg losses.
problem and
ambiguity • Options A and B are both framed in a positive way, in
aversion

Probability
terms of the lives that might be saved; that is, they are
weighting presented in a gain frame.
Discussion
• Options C and D are both framed in a negative way, in
terms of the lives that might be lost; that is, they are
presented in a loss frame.
8 / 53
 Framing Effects in Decision-Making under Risk

Introduction
• As you may recall from chapter 3, people respond to gains
Framing
effects in differently than to losses. We captured this difference
decision-
making under using the value function.
risk • Unlike the utility function, which ranges over total
Bundling and
mental
endowments, the value function ranges over changes in
accounting the endowment.
The Allais • Since people are more sensitive to losses than gains, the
problem and
the sure-thing value function is steeper in the losses domain than in the
principle
gains domain.
The Ellsberg
problem and
• As the pandemic example demonstrates, people also have
ambiguity
aversion
different risk attitudes in the gain and loss domains. To
Probability
account for that we assume that the value function has
weighting different curvatures for losses and for gains.
Discussion • In the realm of losses, we assume that the curve is convex,
so that people are risk prone.
• in the realm of gains, the curve concave, so that people
are risk averse.
9 / 53
 Framing Effects in Decision-Making under Risk

Introduction

Framing
effects in
decision-
making under
risk

Bundling and
mental
accounting

The Allais
problem and
the sure-thing
principle

The Ellsberg
problem and
ambiguity
aversion

Probability
weighting

Discussion

10 / 53
 How Does This Help with the Pandemic Problem

Introduction

Framing
effects in
decision-
making under
risk
• The essential insight is that participants presented with
Bundling and
mental the gain frame take their reference point to be the case in
accounting
which no lives will initially be saved and the program can
The Allais
problem and save (or gain) lives.
the sure-thing
principle • Participants presented with the loss frame take their
The Ellsberg
problem and
reference point to be the case in which no lives are initially
ambiguity
aversion
lost and the program results in lost lives.
Probability • This idea is captured on the following graph.
weighting

Discussion

11 / 53
 How Does This Help with the Pandemic Problem

Introduction

Framing
effects in
decision-
making under
risk

Bundling and
mental
accounting

The Allais
problem and
the sure-thing
principle

The Ellsberg
problem and
ambiguity
aversion

Probability
weighting

Discussion

12 / 53
 Framing Effects, Another Example

Introduction

Framing
• Scenario 1: Imagine you just won $1000 betting on
effects in
decision-
horses. You are now asked to choose between
making under
risk
Option A: a 50 per cent chance of winning another $1000
and 50 per cent chance of winning nothing or
Bundling and
mental Option B: getting $500 for sure.
accounting

The Allais
• Which one do you chose?
problem and
the sure-thing
• Scenario 2: Imagine you just won $2000 betting on
principle horses. You are now asked to choose between
The Ellsberg
problem and
Option C: a 50 per cent chance of losing $1000 and 50
ambiguity per cent chance of losing nothing or
aversion
Option D: losing $500 for sure.
Probability
weighting • Which one do you chose?
Discussion
• In terms of final outcomes, (A) is obviously equivalent to
(C) and (B) to (D). Yet 84 percent of participants chose B
in the first scenario, and 69 percent chose C in the second.
13 / 53
 Framing Effects, Another Example Continued

Introduction

Framing • To make sense of the outcomes above, suppose that your

effects in
p
decision- value function v(·) is defined
p by: v(x) = x/2 for gains
making under
risk (x ≥ 0) and v(x) = −2 |x| for losses (x < 0).
Bundling and • Here is what this value function looks like :
mental
accounting

The Allais
problem and
the sure-thing
principle

The Ellsberg
problem and
ambiguity
aversion

Probability
weighting

Discussion

14 / 53
 Framing Effects, Another Example Continued

Framing effects in decision-making under risk

2019-09-27
• To make sense of the outcomes above, suppose that your
p
value function v(·) is defined
p by: v(x) = x/2 for gains
(x ≥ 0) and v(x) = −2 |x| for losses (x < 0).
• Here is what this value function looks like :

Framing Effects, Another Example Continued

The mathematica file that produced thi graph is called: PlotOfExerciseOn-

Framing
 Framing Effects, Another Example Continued

Introduction

Framing
effects in
decision-
making under • To make sense of the outcomes above, suppose that your
risk p
Bundling and
value function v(·) is defined
p by: v(x) = x/2 for gains
mental
accounting
(x ≥ 0) and v(x) = −2 |x| for losses (x < 0).
The Allais (a) Assuming that you have integrated the $1000 into your
problem and
the sure-thing
endowment, what is the value of (A)?
principle (b) Assuming that you have integrated the $1000 into your
The Ellsberg endowment, what is the value of (B)?
problem and
ambiguity (c) Assuming that you have integrated the $2000 into your
aversion endowment, what is the value of (C)?
Probability
weighting
(d) Assuming that you have integrated the $2000 into your
endowment, what is the value of (D)?
Discussion

15 / 53
 Framing Effects, Another Example Continued

Framing effects in decision-making under risk

2019-09-27 • To make sense of the outcomes above, suppose
value function v(·) is defined
p
p that your
by: v(x) = x/2 for gains
(x ≥ 0) and v(x) = −2 |x| for losses (x < 0).
(a) Assuming that you have integrated the $1000 into your

Framing Effects, Another Example Continued endowment, what is the value of (A)?
(b) Assuming that you have integrated the
endowment, what is the value of (B)?
(c) Assuming that you have integrated the
$1000 into your

$2000 into your

endowment, what is the value of (C)?
(d) Assuming that you have integrated the $2000 into your
endowment, what is the value of (D)?

ANSWERS:
p
(a) v(A) = 0.5 × 1000/2 + 0.5 × 0 = 22.36 × 0.5 = 11.18
p
(b) v(B) = 500/2 = 15.81
p
(c) v(C) = 0.5 × (−2 × | − 1000|) + 0.5 × 0 = −31.62?
p
(d) v(D) = −500 × | − 500| = −44.7
 Framing Effects Helps Explain A Range of
Phenomena
Introduction

Framing
effects in
decision-
The idea that people are risk averse in the domain of gains but
making under
risk
risk prone in the domain of losses helps explain a range of
Bundling and
phenomena.
mental
accounting • It can explain why some people are unable to stop
The Allais gambling: once they find themselves in the red, they enter
problem and
the sure-thing the domain of losses, where they are even more risk prone
principle
than they used to be.
The Ellsberg
problem and
ambiguity
• There is evidence that people betting on horses, etc., are
aversion more willing to bet on long shots at the end of the betting
Probability
weighting
day. This phenomenon is often accounted for by saying
Discussion that people who have already suffered losses are more
prone to risk-seeking behavior.

16 / 53
 Framing Effects Helps Explain A Range of
Phenomena
Introduction

Framing
effects in The idea that people are risk averse in the domain of gains but
decision-
making under
risk prone in the domain of losses helps explain a range of
risk
phenomena.
Bundling and
mental • Analogously, the idea can explain why politicians continue
accounting
to pursue failed projects and generals continue to fight
The Allais
problem and losing wars: as initial efforts fail, the responsible parties
the sure-thing
principle enter the domain of losses, in which they are willing to bet
The Ellsberg on increasingly long shots and therefore take increasingly
problem and
ambiguity desperate measures.
aversion

Probability
• Somewhat paradoxically, this analysis suggests that
weighting
people, countries, and corporations can be expected to be
Discussion
most aggressive when they are weakest – not when they
are strongest, as you might think.

17 / 53
 Framing Effects Helps Explain A Range of
Phenomena

Framing effects in decision-making under risk

2019-09-27
The idea that people are risk averse in the domain of gains but
risk prone in the domain of losses helps explain a range of
phenomena.
• Analogously, the idea can explain why politicians continue
to pursue failed projects and generals continue to fight
losing wars: as initial efforts fail, the responsible parties
Framing Effects Helps Explain A Range of enter the domain of losses, in which they are willing to bet
on increasingly long shots and therefore take increasingly
desperate measures.

Phenomena • Somewhat paradoxically, this analysis suggests that

people, countries, and corporations can be expected to be
most aggressive when they are weakest – not when they
are strongest, as you might think.

p p
A person’s value function is v(x) = x/2 for gains and v(x) = −2 |x|
for losses. The person is facing the choice between a sure $2 and a 50–50
gamble that pays $4 if she wins and $0 if she loses.
(a) Show algebraically that this person is loss averse, in the sense that
she suffers more when she loses $4 than she benefits when she
receives $4.
(b) If she takes the worst possible outcome ($0) as her reference point,
what is the value of the sure amount and the gamble? Which would
she prefer?
(c) If she takes the best possible outcome ($4) as her reference point,
what is the value of the sure amount and the gamble? Which would
she prefer?
 Framing Effects Helps Explain A Range of
Phenomena

Framing effects in decision-making under risk

2019-09-27
The idea that people are risk averse in the domain of gains but
risk prone in the domain of losses helps explain a range of
phenomena.
• Analogously, the idea can explain why politicians continue
to pursue failed projects and generals continue to fight
losing wars: as initial efforts fail, the responsible parties
Framing Effects Helps Explain A Range of enter the domain of losses, in which they are willing to bet
on increasingly long shots and therefore take increasingly
desperate measures.

Phenomena • Somewhat paradoxically, this analysis suggests that

people, countries, and corporations can be expected to be
most aggressive when they are weakest – not when they
are strongest, as you might think.

ANSWERS:
p p
(a) v(4) = 4/2 ≈ 1.41, v(−4) = −2 | − 4| = −8. In absolute terms,
she is suffering the $4 loss more than enjoys the $4 gain.
p
(b) Value of the sure amount is v(2) = p 2/2 = 1. Value p of the gamble
is: 0.5 × v(4) + 0.5 × v(0) = 0.5 × 4/2 + 0.5 × 0/2 ≈ 0.705.
Thus the sure $2 results in greater value than the gamble when the
gamble is presented in a gain frame.
p
(c) Value of the sure amount is v(−2) = −2 | − 2| ≈ −2.82. Value of
the gamble is: p p
0.5 × v(0) + 0.5 × v(−4) = 0.5 × 0/2 + 0.5 × −2 | − 4| = −2.
Thus the value of the gamble is greater than the value of the sure
amount when the gamble is presented in a loss fraim.
 Framing Effects Helps Explain A Range of
Phenomena

Framing effects in decision-making under risk

2019-09-27
The idea that people are risk averse in the domain of gains but
risk prone in the domain of losses helps explain a range of
phenomena.
• Analogously, the idea can explain why politicians continue
to pursue failed projects and generals continue to fight
losing wars: as initial efforts fail, the responsible parties
Framing Effects Helps Explain A Range of enter the domain of losses, in which they are willing to bet
on increasingly long shots and therefore take increasingly
desperate measures.

Phenomena • Somewhat paradoxically, this analysis suggests that

people, countries, and corporations can be expected to be
most aggressive when they are weakest – not when they
are strongest, as you might think.

Relative income It is well known that poor people, who can least afford
to play the lottery, are most likely to do so. In a 2008 study, researchers
wanted to know whether manipulating people’s perceptions of their income
can affect their demand for lottery tickets. Half of the participants were
made to feel rich by answering a question about their yearly income on
a scale from < $10k, $10k–$20k and so on, to > $60k. The other half
were made to feel poor by answering the same question on a scale from
< $100k, $100k–$200k, and so on, to $1M At the conclusion of the
study, participants who were made to feel relatively poor were more likely
to choose lottery tickets than cash as a reward for their participation. Are
these findings consistent with the analysis in this section or not?
ANSWER:The analysis in this section says that people are more risk prone
when they are in the realm of losses. Assuming people who are made to
feel poor end up feeling like they are in the realm of losses, the analysis
suggests that they should be more likely to choose lottery tickets – which
is exactly what the researchers found.
 Integrated and Segregated Payoffs

Introduction

Framing
• The shape of the value function has other interesting
effects in
decision-
implications, one being that it matters how outcomes are
making under
risk
bundled.
• Suppose, that you buy two lottery tickets at a charity
Bundling and
mental event, and that you win $25 on the first and $50 on the
accounting
second.
The Allais
problem and – You can integrate the outcomes, and tell yourself that you
the sure-thing
principle
just won $75, which in value terms would translate into
v($75).
The Ellsberg
problem and – Or you can segregate the outcomes and tell yourself that
ambiguity
aversion you first won $25 and then won $50, which in value terms
Probability would translate into v($25) + v($50).
weighting • Which would make you happier?
Discussion
• According to the classical theory, bundling should not
matter. The utility function ranges over total endowments,
and no matter how you describe the various outcomes, you
end up with an additional $75 dollars in your pocket. 18 / 53
 Effect of Integration and Segregation of Gains on
Value: Graphical Explanation
Introduction

Framing • According to prospect theory, however, bundling matters.

effects in
decision-
making under
• When the two gains are integrated, the value of winning
risk $75 can be characterized as in the figure below:
Bundling and
mental
accounting

The Allais
problem and
the sure-thing
principle

The Ellsberg
problem and
ambiguity
aversion

Probability
weighting

Discussion

19 / 53
 Effect of Integration and Segregation of Gains on
Value: Graphical Explanation
Introduction

Framing
• However, when the two gains are segregated, you have the
effects in
decision-
time to adjust your reference point before assessing the
making under
risk
value of the second gain.
Bundling and
• When the two gains are segregated, the picture will look
mental
accounting
more as in the figure below, where the green line represents
The Allais your value function relative to the new reference point.
problem and
the sure-thing
principle

The Ellsberg
problem and
ambiguity
aversion

Probability
weighting

Discussion

20 / 53
 Effect of Integration and Segregation of Gains on
Value: Sum-up
Introduction

Framing
effects in
decision-
making under
risk
• It should be clear just from looking at these two figures
Bundling and
mental that the value of a $25 gain plus the value of a $50 gain is
accounting
greater than the value of a $75 gain
The Allais
problem and ⇒ v($25) + v($50) > v($75).
the sure-thing
principle • This result follows from the value function being concave
The Ellsberg
problem and
in the domain of gains.
ambiguity
aversion • The upshot is that people value two gains more when they
Probability are segregated than when they are integrated.
weighting

Discussion

21 / 53
 Segregation of Gains Can Explain Several
Phenomena
Introduction

Framing
effects in
decision-
• The fact that gains are valued more when segregated helps
making under explain a variety of phenomena. For example,
risk
• It explains why people do not put all their Christmas
Bundling and
mental presents in one big box.
accounting
• Segregation explains why fancy meals are served up
The Allais
problem and dish-by-dish, rather than all at one time.
the sure-thing • The effect of segregating gains explains why workers
principle

The Ellsberg
receive end-of-year bonuses: receiving a $50k salary plus a
problem and $5k bonus encourages the segregation of gains in a manner
ambiguity
aversion that receiving a $55k salary does not.
Probability
• The effect explains why people on daytime television try to
weighting sell pots and pans by offering to throw in lids, knives,
Discussion cutting boards, and so on, rather than simply offering a
basket consisting of all these goods.

22 / 53
 Segregation of Gains Can Explain Several
Phenomena

Bundling and mental accounting

2019-09-27 • The fact that gains are valued more when segregated helps
explain a variety of phenomena. For example,
• It explains why people do not put all their Christmas
presents in one big box.
• Segregation explains why fancy meals are served up
dish-by-dish, rather than all at one time.

Segregation of Gains Can Explain Several • The effect of segregating gains explains why workers
receive end-of-year bonuses: receiving a $50k salary plus a
$5k bonus encourages the segregation of gains in a manner
that receiving a $55k salary does not.

Phenomena • The effect explains why people on daytime television try to

sell pots and pans by offering to throw in lids, knives,
cutting boards, and so on, rather than simply offering a
basket consisting of all these goods.

Evaluation of gains Yesterday, you had a decent day: you first received a
$48 tax refund, and then an old friend repaid a $27 loan you had forgotten
p
about. Suppose that your valuep function v(·) is defined by: v(x) = x/3
forgains(x ≥ 0) and v(x) = −3 |x| for losses (x < 0)
(a) If you integrate the two gains, what is the total value?
(b) If you segregate the two gains, what is the total value?
(c) From the point of view of value, is it better to integrate or to
segregate?
ANSWER:
p
(a) v(+48 + 27) = v(+75) = 75/3 = 5.
p p
(b) v(+48) + v(+27) = 48/3 + 27/3 = 7.
(c) It is better to segregate.
 Effect of Integration and Segregation of Losses on
Value
Introduction

Framing
effects in
decision-
making under
risk • Meanwhile, people experience less dissatisfaction when
Bundling and
mental
multiple losses are integrated than when they are
accounting segregated.
The Allais
problem and • By constructing graphs like those above, you can confirm
the sure-thing
principle that, from the point of view of value, a $25 loss plus an
The Ellsberg additional $50 loss is worse than a $75 loss; that is,
problem and
ambiguity v(−25) + v(−50) < v(−75).
aversion

Probability
• This result follows from the fact that the value function is
weighting convex in the domain of losses.
Discussion

23 / 53
 Integration of Losses Can Explain Several
Phenomena
Introduction

Framing • The fact that people experience less dissatisfaction when losses
effects in
decision- are integrated helps explain a variety of phenomena. For
making under
risk
example,
Bundling and
• It explains why sellers of cars, homes, and other pricey
mental goods often try to sell expensive add-ons bundled with the
accounting

The Allais
basic package.
problem and • Similarly, the wedding industry makes good use of
the sure-thing
principle integration by adding options that individually do not seem
The Ellsberg very expensive relative to what the hosts are already
problem and
ambiguity
spending, but which jointly can cause major financial
aversion distress.
Probability • The effects of integrating losses can also explain why so
weighting
many people prefer to use credit cards rather than paying
Discussion
cash. When you use a credit card, though the monthly bill
might be alarming, it only arrives once a month, thereby
encouraging you to integrate the losses.
24 / 53
 Integration of Losses Can Explain Several
Phenomena

Bundling and mental accounting

2019-09-27
• The fact that people experience less dissatisfaction when losses
are integrated helps explain a variety of phenomena. For
example,
• It explains why sellers of cars, homes, and other pricey
goods often try to sell expensive add-ons bundled with the
basic package.
• Similarly, the wedding industry makes good use of

Integration of Losses Can Explain Several integration by adding options that individually do not seem
very expensive relative to what the hosts are already
spending, but which jointly can cause major financial
distress.

Phenomena • The effects of integrating losses can also explain why so

many people prefer to use credit cards rather than paying
cash. When you use a credit card, though the monthly bill
might be alarming, it only arrives once a month, thereby
encouraging you to integrate the losses.

The opposite arrangement Suppose that the opposite were true: when-
ever you purchase something, you have to pay cash on the spot, but your
purchases are not delivered until the end of the month in a giant box
containing everything you bought in the last four weeks.
(a) Would you make more or fewer purchases this way?
(b) Use the language of integration and segregation to explain why.
ANSWER:
(a) Chances are you would make fewer purchases.
(b) The arrangement would encourage you to segregate your losses but
integrate your gains, which would simultaneously increase the pain
of paying for your stuff and reduce the enjoyment you would derive
from it.
 Integration of Losses Can Explain Several
Phenomena
Introduction

Framing
effects in
decision-
making under
risk • The fact that people experience less dissatisfaction when losses
Bundling and are integrated helps explain a variety of phenomena. For
mental
accounting example,
The Allais • This analysis might also explain why people hold on to cars
problem and
the sure-thing even though taking cabs may be less expensive in the long
principle
run.
The Ellsberg • The analysis can also explain why people prefer to pay a
problem and
ambiguity flat monthly fee for cell phone services, internet
aversion
connections, and gym memberships: doing so permits them
Probability
weighting to integrate what would otherwise be many separate losses.
Discussion

25 / 53
 Integration of Losses Can Explain Several
Phenomena

Bundling and mental accounting

2019-09-27 • The fact that people experience less dissatisfaction when losses
are integrated helps explain a variety of phenomena. For
example,
• This analysis might also explain why people hold on to cars

Integration of Losses Can Explain Several even though taking cabs may be less expensive in the long
run.
• The analysis can also explain why people prefer to pay a
flat monthly fee for cell phone services, internet

Phenomena connections, and gym memberships: doing so permits them

to integrate what would otherwise be many separate losses.

Evaluation of losses Yesterday, you had a terrible day: you got a $144
speeding ticket on your way to the opera, and then had to pay $25 for a
ticket you thought would
p be free. Suppose that your value p function v(·) is
defined by: v(x) = x/3 forgains(x ≥ 0) and v(x) = −3 |x| for losses
(x < 0) .
(a) If you integrate the two losses, what is the total value?
(b) If you segregate the two losses, what is the total value?
(c) From the point of view of value, is it better to integrate or to
segregate?
ANSWER:
√
(a) v(−144 − 25) = v(−169) = −3 169 = −39.
√ √
(b) v(−144) + v(−25) = −3 144 + (−3) 25 = −51.
(c) It is better to integrate.
 Cancellation and Silver-lining

Introduction
• Related phenomena occur when experiencing a small loss
Framing
effects in in combination with a large gain or a large loss in
decision-
making under combination with a small gain.
risk • People gain more value when they integrate a small loss
Bundling and
mental
with a large gain. This phenomenon is referred to as
accounting cancellation.
The Allais • At the same time, people gain more value when they
problem and
the sure-thing segregate a large loss from a small gain. The small gain is
principle
often described as a silver lining.
The Ellsberg
problem and
• Example of cancellation Suppose that you win a million
ambiguity dollars, but have to pay a 10 percent tax on your winnings.
aversion
The theory suggests that you will derive more value from
Probability
weighting telling yourself that you won $900,000 than by telling
Discussion yourself that you won $1,000,000 and then lost $100,000.
• Example of a silver lining A customer may be more likely
to buy a car with a $27k price tag and a $1k cash-back
offer than to buy the very same car with a $26k price tag.
26 / 53
 Cancellation and Silver-lining

Bundling and mental accounting • Related phenomena occur when experiencing a small loss

2019-09-27 in combination with a large gain or a large loss in

combination with a small gain.
• People gain more value when they integrate a small loss
with a large gain. This phenomenon is referred to as
cancellation.
• At the same time, people gain more value when they

Cancellation and Silver-lining segregate a large loss from a small gain. The small gain is
often described as a silver lining.
• Example of cancellation Suppose that you win a million
dollars, but have to pay a 10 percent tax on your winnings.
The theory suggests that you will derive more value from
telling yourself that you won $900,000 than by telling
yourself that you won $1,000,000 and then lost $100,000.
• Example of a silver lining A customer may be more likely
to buy a car with a $27k price tag and a $1k cash-back
offer than to buy the very same car with a $26k price tag.

The pain of paying taxes The previous paragraph suggests that how you
feel about paying your taxes will depend on whether you inte- grate that
cost with the money you made or not.
(a) If you are a politician known for favoring high taxes, should you
encour- age voters to integrate or segregate? How should you shape
your message?
(b) If you are a politician known for favoring low taxes, should you
encourage voters to integrate or segregate? How should you shape
your message?
ANSEWER:
(a) You should encourage voters to integrate: “You’re still taking home
$900k!”
(b) You should encourage voters to segregate: “You made $1M! That
money is yours! The government is taking $100k of your money!”
 Cancellation and Silver-lining

Bundling and mental accounting • Related phenomena occur when experiencing a small loss

2019-09-27 in combination with a large gain or a large loss in

combination with a small gain.
• People gain more value when they integrate a small loss
with a large gain. This phenomenon is referred to as
cancellation.
• At the same time, people gain more value when they

Cancellation and Silver-lining segregate a large loss from a small gain. The small gain is
often described as a silver lining.
• Example of cancellation Suppose that you win a million
dollars, but have to pay a 10 percent tax on your winnings.
The theory suggests that you will derive more value from
telling yourself that you won $900,000 than by telling
yourself that you won $1,000,000 and then lost $100,000.
• Example of a silver lining A customer may be more likely
to buy a car with a $27k price tag and a $1k cash-back
offer than to buy the very same car with a $26k price tag.

Silver linings For this

p question, suppose that your value function
p v(·) is
defined by: v(x) = x/2 for gains(x ≥ 0) and v(x) = −2 |x| for losses
(x < 0). Last night, you lost $9 in a bet. There was a silver lining, though:
on your way home, you found $2 lying on the sidewalk.
(a) If you integrate the loss and the gain, what is the total value?
(b) If you segregate the loss and the gain, what is the total value?
(c) From the point of view of value, is it better to integrate or to
segregate?
ANSWER:
(a) v(−9 + 2) = v(−7) ≈ −5.29.
(b) v(−9) + v(2) = −5
(c) It is better to segregate.
 When Do People Integrate and When Segregate?

Introduction

Framing
effects in
decision- • When do people integrate and when do they segregate?
making under
risk • One possibility that might come to mind is that people
Bundling and
mental
bundle outcomes so as to maximize the amount of value
accounting that they experience. This is called the hedonic-editing
The Allais
problem and
hypothesis.
the sure-thing
principle
• According to this hypothesis, people will
The Ellsberg 1. segregate gains,
problem and
ambiguity
2. integrate losses,
aversion 3. integrate a small loss with a large gain,
Probability 4. segregate a small gain from a large loss
weighting

Discussion
• Unfortunately, data suggest that the hypothesis is not in
general true.

27 / 53
 Mental Accounting

Introduction

Framing
effects in • Bundling may be driven in part by mental accounting:
decision-
making under people’s tendency, in their minds, to divide money into
risk
separate categories.
Bundling and
mental • Mental accounting can be helpful in that it may stop you
accounting

The Allais
from overspending on any one category.
problem and
the sure-thing
• But mental accounting can itself cause people to
principle
overconsume or underconsume particular kinds of goods: if
The Ellsberg
problem and the mental “entertainment account” is seen as having
ambiguity
aversion money left in it, but the mental “clothing account” is seen
Probability as overdrawn, people might spend more on entertainment
weighting
even though they would maximize utility by buying clothes.
Discussion
• This kind of behavior violates fungibility: the idea that
money has no labels.

28 / 53
 Mental Accounting

Bundling and mental accounting

2019-09-27
• Bundling may be driven in part by mental accounting:
people’s tendency, in their minds, to divide money into
separate categories.
• Mental accounting can be helpful in that it may stop you
from overspending on any one category.

Mental Accounting • But mental accounting can itself cause people to

overconsume or underconsume particular kinds of goods: if
the mental “entertainment account” is seen as having
money left in it, but the mental “clothing account” is seen
as overdrawn, people might spend more on entertainment
even though they would maximize utility by buying clothes.
• This kind of behavior violates fungibility: the idea that
money has no labels.

There is a great article in Washington Post by Shankar Vidantum about

mental accounting, here is the link: http://www.washingtonpost.com/wp-
dyn/content/article/2007/05/19/AR2007051900316.html
 The Sure-Thing Principle

Introduction

Framing
effects in
decision-
making under
risk • Suppose you have to face the following option. What
Bundling and
mental
would you choose?
accounting (1a) A 90% chance of winning $1 million and 10% chance of
The Allais winning $5 million.
problem and
the sure-thing (1b) $1million for sure
principle

The Ellsberg
• Most people would chose option (1a) because it pays the
problem and
ambiguity
same amount as option (1b) with probability 90% and
aversion
there is an additional 10% chance that you may get even
Probability
weighting more.
Discussion

29 / 53
 The Sure-Thing Principle

Introduction

Framing
• The optimality of this choice can be easily seen if the problem is
effects in
decision-
presented in a decision-matrix form:
making under
risk
State 1 (S1 ) State 2 (S2 )
Bundling and
mental
P r(S1 ) = 90% P r(S2 ) = 10%
accounting
(1a) $1 million $5 million
The Allais
problem and (1b) $1 million $1 million
the sure-thing
principle
• The $1 million outcome in State 1 should not influence your
The Ellsberg
problem and decision, since it’s same under both options, it’s a sure thing.
ambiguity
aversion • The sure-thing principle says that your decisions should not be
Probability influenced by sure things. As this discussion indicates, it is
weighting
implicit in expected-utility theory.
Discussion
• Maurice Allais (a French economist and Nobel laureate)
proposed an examples that demonstrated how easily the
sure-thing principle may fail.
30 / 53
 The Allais Problem

Introduction
• Suppose that you face the following options, and that you
Framing
effects in must choose first between (1a) and (1b), and second
decision-
making under between (2a) and (2b). What would you choose?
risk
(1a) $1 million for sure
Bundling and
mental
(1b) An 89% chance of $1 million & a 10% chance of $5
accounting million & 1% of nothing.
The Allais • Most people chose (1a)
problem and
the sure-thing (2a) An 11% chance of $1 million & 89% of nothing
principle
(2b) A 10% chance of $5 million 90% of nothing
The Ellsberg
problem and • Most people chose (2b).
ambiguity
aversion • Unfortunately, this response pattern is inconsistent with
Probability expected-utility theory and the sure-thing principle.
weighting
• To see this, suppose that you spin a roulette wheel with
Discussion
100 slots: 89 black, 10 red, and 1 white. This permits us
to represent the four options in table form on the following
slide.
31 / 53
 The Allais Problem

Introduction

Framing Black (89%) Red (10%) White (1%)

effects in
decision-
making under (1a) $1 million $1 million $1 million
risk
(1b) $1 million $5 million $0
Bundling and
mental (2a) $0 $1 million $1 million
accounting
(2b) $0 $5 million $0
The Allais
problem and
the sure-thing
principle
• Under the first option, if black occurs you will get a
The Ellsberg million dollars either way, it’s a sure thing. Thus, the fact
problem and
ambiguity that you are getting $1M with 89% probability should not
aversion
affect your choice between (1a) and (1b).
Probability
weighting
• The same is true for option 2, you get the same $0 if
Discussion Black comes out.
• Thus, the column marked “Black” should not affect your
choices at all. Instead, your choices will be determined by
the other two columns.
32 / 53
 The Allais Problem

Introduction

Framing
effects in
decision-
Black (89%) Red (10%) White (1%)
making under
risk (1a) $1 million $1 million $1 million
Bundling and
mental
(1b) $1 million $5 million $0
accounting (2a) $0 $1 million $1 million
The Allais
problem and
(2b) $0 $5 million $0
the sure-thing
principle

The Ellsberg
problem and
• But once you ignore the column marked “Black,” (1a) is
ambiguity
aversion
identical to (2a) and (1b) is identical to (2b)
Probability • So if you strictly prefer (1a), you are rationally compelled
weighting

Discussion
to choose (2a); if you strictly prefer (1b), you are
rationally compelled to choose (2b).
• But in reality, people often chose (1a) and (2b)

33 / 53
 The Allais Problem More Formally

Introduction

Framing
effects in
decision-
• More formally, preferring (1a) to (1b) means:
making under
risk
u(1M ) > 0.89 × u(1M ) + 0.10 × u(5M )
Bundling and
mental
accounting ⇒u(1M ) − 0.89 × u(1M ) > 0.10 × u(5M )
The Allais ⇒0.11 × u(1M ) > 0.10 × u(5M )
problem and
the sure-thing
principle
• Preferring (2b) to (2a) means:
The Ellsberg
problem and
ambiguity
aversion
0.10 × u(5M ) > 0.11 × u(1M )
Probability
weighting • The last inequality that follows from the preference (1a)
Discussion over (1b) contradicts the inequality that follows from the
preference of (2b) over (2a).

34 / 53
 The Alais Problem and the Sure-thing Principle:
An Exercise
Introduction

Framing
effects in
decision- Suppose that you face the options in Table below and that you
making under
risk must choose first between (1a) and (1b), and second between
Bundling and (2a) and (2b). What choice pattern is ruled out by the
mental
accounting sure-thing principle?
The Allais
problem and
the sure-thing
principle
X Y Z
The Ellsberg
problem and
(1a) 80 100 40
ambiguity
aversion
(1b) 40 100 80
Probability
(2a) 40 0 80
weighting (2b) 80 0 40
Discussion

35 / 53
 The Alais Problem and the Sure-thing Principle:
An Exercise

The Allais problem and the sure-thing principle

2019-09-27 Suppose that you face the options in Table below and that you
must choose first between (1a) and (1b), and second between
(2a) and (2b). What choice pattern is ruled out by the
sure-thing principle?

The Alais Problem and the Sure-thing (1a)

X
80
Y
100
Z
40
(1b) 40 100 80

Principle: An Exercise (2a)

(2b)
40
80
0
0
80
40

ANSWER:
The choice pattern (1a) and (2a) is excluded, as is the choice pattern (1b)
and (2b).
 The Alais Problem and the Certainty Effect

Introduction

Framing
effects in
decision-
making under
risk
• One way to describe the Allais paradox is to say that
Bundling and
mental people overweight outcomes that are certain. This
accounting
tendency has been called the certainty effect.
The Allais
problem and
the sure-thing
• The certainty effect might result from regret aversion:
principle whenever you forego a certain option for a risky one, there
The Ellsberg
problem and
is some chance that you will experience regret.
ambiguity
aversion • Thus, a desire to minimize anticipated regret would lead
Probability to the rejection of the option that is not certain.
weighting

Discussion

36 / 53
 The Certainty Effect, an Example

Introduction

Framing
effects in
decision-
1. Which of the following options do you prefer:
making under (A) a sure (100% probability) win of $30;
risk

Bundling and
(B) an 80% chance to win $45 (and 20% chance to win $0)?
mental
accounting • 78% of respondents favored A over B.
The Allais 2. Which of the following options do you prefer:
problem and
the sure-thing (C) a 25% chance to win $30 (and 75% chance to win $0);
principle
(D) a 20% chance to win $45 (and 80% chance to win $0)?
The Ellsberg
problem and
ambiguity
• 58% favored D over C.
aversion
3. Show that it is a violation of expected-utility theory to
Probability
weighting choose (A) over (B) and (D) over (C). (Hint: Notice that
Discussion (C) and (D) can be obtained from (A) and (B) by dividing
the probabilities by four).

37 / 53
 The Certainty Effect, an Example

The Allais problem and the sure-thing principle

2019-09-27 1. Which of the following options do you prefer:
(A) a sure (100% probability) win of $30;
(B) an 80% chance to win $45 (and 20% chance to win $0)?
• 78% of respondents favored A over B.
2. Which of the following options do you prefer:

The Certainty Effect, an Example (C) a 25% chance to win $30 (and 75% chance to win $0);
(D) a 20% chance to win $45 (and 80% chance to win $0)?
• 58% favored D over C.
3. Show that it is a violation of expected-utility theory to
choose (A) over (B) and (D) over (C). (Hint: Notice that
(C) and (D) can be obtained from (A) and (B) by dividing
the probabilities by four).

• Just a general comment regarding peoples’ preferences manifest in

the choice problem above: a reduction from 100 percent to 25
percent makes a bigger difference to people than a reduction from
80 percent to 20 percent.
• Answer to question 3:
A strict preference for A over B entails that EU (A) > EU (B),
which means that:
1 × u(30) > 0.80 × u(45).
Divide each side by four, and you get
0.25 × u(30) > 0.20 × u(45)
A strict preference for D over C entails that EU (D) > EU (C),
which means that
0.20 × u(45) > 0.25 × u(30).
But these are inconsistent.
 The Ellsberg Paradox

Introduction
• The following decision problem is referred to as the Ellsberg
Framing
effects in problem (paradox) after Daniel Ellsberg, a US military analyst
decision-
making under otherwise famous for releasing the so-called Pentagon Papers.
risk • You are picking a ball from an urn with 90 balls in it. 30 are red
Bundling and
mental
and the remaining 60 are a mix of black and yellow (ranging
accounting from all black to all yellow):
The Allais
problem and Red Black Yellow
the sure-thing
principle Number of balls in urn 30 60
The Ellsberg
problem and • You are given the following 2 bets:
ambiguity
aversion
Bet 1
(I) $100 if the ball is red
Probability
weighting (II) $100 if the ball is black
Discussion Bet 2
(III) $100 if the ball is red or yellow
(IV) $100 if the ball is black or yellow
• When presented with these two bets, many people chose (I)
rather than (II) and (IV) rather than (III). 38 / 53
 The Ellsberg Paradox Continued

Introduction
• However, the choice of (I) from the first pair of options and (IV)
Framing
effects in from the second pair violates the sure-thing principle.
decision-
making under
• The decision matrix below makes it clear:
risk
Red (R) Black (B) Yellow (Y)
Bundling and
mental
accounting I 100 0 0
The Allais II 0 100 0
problem and
the sure-thing
III 100 0 100
principle IV 0 100 100
The Ellsberg
problem and
• Because the outcomes in the ”Yellow” column are the same
ambiguity
aversion
under each bet and due to the sure-thing principle, your choices
should not depend on what happens when you draw a yellow
Probability
weighting ball. However, your choices must depend on what is going on in
Discussion the ”Red” and ”Black” columns.
• However, (I) and (III) are identical, as are (II) and (IV). Hence,
unless you are indifferent, you must either choose (I) and (III) or
(II) and (IV).
39 / 53
 The Ellsberg Paradox More Formally

Introduction
• There is another way of showing how the choice pattern (I) and
Framing
effects in (IV) is inconsistent with expected-utility theory.
decision-
making under
• A strict preference for (I) over (II) entails that
risk EU (I) > EU (II), which means that:
Bundling and
mental P r(R)×u(100) + P r(B) × u(0) + P r(Y ) × u(0)
accounting

The Allais
> P r(R) × u(0) + P r(B) × u(100) + P r(Y ) × u(0)
problem and
the sure-thing • A strict preference for (IV) over (III) entails that
principle
EU (IV ) > EU (III), which means that:
The Ellsberg
problem and P r(R)×u(0) + P r(B) × u(100) + P r(Y ) × u(100)
ambiguity
aversion > P r(R) × u(100) + P r(B) × u(0) + P r(Y ) × u(100)
Probability
weighting • For simplicity assume that u(0) = 0 and u(100) = 1. If so, then
Discussion the two expressions above imply:
P r(R) > P r(B)
P r(B) > P r(R)
which is impossible. 40 / 53
 How Do We Explain the Ellsberg Paradox?

Introduction
• The two rejected options – (II) and (III) – have something in
Framing
effects in common, namely, that the exact probability of winning is
decision-
making under unclear.
risk • We say that these probabilities are ambiguous.
Bundling and
mental
• By contrast, the favored options – (I) and (IV) – are not
accounting associated with ambiguous probabilities.
The Allais • The observed choices seem to reflect an unwillingness to take
problem and
the sure-thing on gambles with ambiguous probabilities. We refer to this
principle
phenomenon as ambiguity aversion.
The Ellsberg
problem and
• Some people have a greater tolerance for ambiguity than others,
ambiguity but any aversion to ambiguity is a violation of expected-utility
aversion
theory.
Probability
weighting • Insofar as people are in fact ambiguity averse, expected-utility
Discussion theory fails to capture their behavior. And insofar as it is
rationally permissible to take ambiguity into account when
making decisions, expected-utility theory does not capture the
manner in which people should make decisions.
41 / 53
 Can Someone Be Ambiguity Prone?

Introduction

Framing
effects in
decision-
making under
• There is no principled reason why people cannot be
risk ambiguity prone rather than ambiguity averse.
Bundling and
mental • In fact, evidence suggests that people’s behavior in the
accounting
face of ambiguous probabilities depends on the context.
The Allais
problem and
the sure-thing
• According to the competence hypothesis, for example,
principle people are less averse to ambiguity in contexts where they
The Ellsberg
problem and
consider themselves particularly knowledgeable. Thus, a
ambiguity
aversion
football fan may be ambiguity averse in the Ellsberg case
Probability (where outcomes are completely random) but ambiguity
weighting
prone when predicting the outcomes of football games
Discussion
(where he or she feels like an expert).

42 / 53
 Implications of the Ellsberg Paradox

Introduction

Framing
effects in • The Ellsberg paradox and ambiguity aversion have
decision-
making under potentially vast implications.
risk
• What they suggest is that people do not in general assign
Bundling and
mental
accounting
probabilities satisfying the axioms of the probability
The Allais
calculus to events with ambiguous probabilities.
problem and
the sure-thing
• And in the real world, ambiguous probabilities are
principle
common. While the probability of bankruptcies, oil spills,
The Ellsberg
problem and and nuclear meltdowns can be estimated, outside games of
ambiguity
aversion chance, some ambiguity almost always remains.
Probability • Thus, it is highly likely that people’s choices do reflect the
weighting

Discussion
fact that people are ambiguity averse – or prone, as the
case may be. And perhaps choices should reflect the
ambiguity of the probabilities too.

43 / 53
 Probability Weighting

Introduction

Framing
effects in
decision-
making under
risk • There are widely observed behavioral patterns that cannot
Bundling and
mental
be accommodated within neither the expected utility
accounting framework nor the S-shaped value function.
The Allais
problem and • Consider the fact that some people simultaneously gamble
the sure-thing
principle and purchase insurance. This is paradoxical from the point
The Ellsberg of view of expected-utility theory because if they are risk
problem and
ambiguity averse they should buy insurance but not gamble. If
aversion
people are risk prone, on the other hand, they should
Probability
weighting gamble but not buy insurance.
Discussion

44 / 53
 Probability Weighting

Introduction

Framing
effects in
decision- • This fact is also paradoxical from the the point of view of the
making under
risk value function theory. The fact that people are willing to accept
Bundling and a gamble in which they may win a large sum of money suggests
mental that they are risk prone in the domain of gains, while the fact
accounting

The Allais
that they are willing to reject a gamble in which they may lose
problem and their house suggests that they are risk averse in the domain of
the sure-thing
principle losses. This would entail that their value function is convex in
The Ellsberg the domain of gains and concave in the domain of losses, which
problem and
ambiguity
is the very opposite of what we have assumed to date.
aversion • Although it is theoretically possible that people have inverted
Probability
weighting
S-shaped value functions, but it would contradict other
Discussion
circumstantial evidence that people are often risk prone in the
loss domain and risk averse in the gain domain.

45 / 53
 Probability Weighting

Probability weighting
2019-09-27 • This fact is also paradoxical from the the point of view of the
value function theory. The fact that people are willing to accept
a gamble in which they may win a large sum of money suggests
that they are risk prone in the domain of gains, while the fact
that they are willing to reject a gamble in which they may lose
their house suggests that they are risk averse in the domain of

Probability Weighting losses. This would entail that their value function is convex in
the domain of gains and concave in the domain of losses, which
is the very opposite of what we have assumed to date.
• Although it is theoretically possible that people have inverted
S-shaped value functions, but it would contradict other
circumstantial evidence that people are often risk prone in the
loss domain and risk averse in the gain domain.

The only way to accommodate this behavior pattern within the loss-
aversion framework above is to assume that people take the state when
they win the grand prize as their reference point when gambling, and the
state in which they lose things as their reference point when buying in-
surance. This seems artificial, however, in light of the other evidence that
people otherwise frequently take their endowment as their reference point.
 Probability Weighting

Introduction

Framing
effects in
decision-
making under
risk
• One way to understand the observed behavior pattern
Bundling and
is to think of those who gamble as well as those who
mental buy insurance as prone to paying too much attention
accounting
to unlikely events.
The Allais
problem and • The more weight you put on the probability of winning
the sure-thing
principle the lottery, the more likely you will be to gamble. And
The Ellsberg the more weight you put on the probability of losing
problem and
ambiguity
your house, car, life, and limb, the more likely you will
aversion be to purchase insurance. This insight suggests a more
Probability systematic approach to explaining how people can
weighting
simultaneously buy lottery tickets and insurance.
Discussion

46 / 53
 Probability Weighting

Introduction
• Prospect theory incorporates this kind of behavior by utilizing
Framing
effects in the notions of probability weighting and probability
decision-
making under weighting function π(·).
risk

Bundling and Definition

mental
accounting Given a decision problem with n states of the world {S1 , S2 , . . . , Sn },
The Allais the value (or weighted value) V (Ai ) of an act Ai is given by
problem and
the sure-thing
principle V (Ai ) =π (P r(S1 )) × v(Ci1 )
The Ellsberg
problem and + π (P r(S2 )) × v(Ci2 )
ambiguity
aversion + ...
Probability + π (P r(Sn )) × v(Cin )
weighting

Discussion
where Cij , (j = 1, 2, . . . , n) is the outcome of act i under j th state of
the world.
• Optimal act maximizes weighted value.
47 / 53
 Probability Weighting

Introduction
• As the graph of the probability-weighting function
Framing
effects in below indicates, π(·) assigns weights, from zero to one
decision-
making under inclusive, to probabilities. It is assumed that π(0) = 0
risk and that π(1) = 1.
Bundling and • For low probabilities, it is assumed that π(x) > x, and
mental
accounting for moderate and high probabilities, that π(x) < x.
The Allais This mathematical representation mimics the human
problem and
the sure-thing tendency of overestimating low probabilities.
principle

The Ellsberg
problem and
ambiguity
aversion

Probability
weighting

Discussion

48 / 53
 Probability Weighting: an Example

Introduction

Framing
• You are facing the following investment payoff matrix:
effects in
decision- Economy
making under
risk Growing Stable Declining
Bundling and
mental Bonds $40 $45 $5
accounting
Stocks $70 $30 -$13
The Allais
problem and Mutual Funds $53 $45 -$5
the sure-thing
principle
• The probability of the economy being in growing, stable,
The Ellsberg
problem and and declining states are 0.36, 0.60, and 0.04 respectively.
ambiguity
p
aversion • Your value function v(·) is defined by: v(x) = x/2 for
p
Probability gains(x ≥ 0) and v(x) = −2 |x| for losses (x < 0).
weighting
Moreover, your initial reference wealth is $0.
Discussion
• Finally, the probability-weighting function
√ is given by the
p
following expression: π(p) = √ √
( p + 1 − p)2
• Which investment is optimal? 49 / 53
 Probability Weighting: an Example

Probability weighting • You are facing the following investment payoff matrix:

2019-09-27 Growing
Economy
Stable Declining
Bonds $40 $45 $5
Stocks $70 $30 -$13
Mutual Funds $53 $45 -$5

Probability Weighting: an Example • The probability of the economy being in growing, stable,
and declining states are 0.36, 0.60, and 0.04 respectively.
p
• Your value function v(·) is defined by: v(x) = x/2 for
p
gains(x ≥ 0) and v(x) = −2 |x| for losses (x < 0).
Moreover, your initial reference wealth is $0.
• Finally, the probability-weighting function
√ is given by the
p
following expression: π(p) = √ √
( p + 1 − p)2
• Which investment is optimal?

All computations are done in ProbWeightingFunctionExample.xls (also can

use ProbabilityWeightingFunction mathematica file for graphical demon-
stration of the probability-weighting function).
• The first step is to compute weighted probabilities:
π(P r(Growing)) = π(0.36) = 0.31,
π(P r(Stable)) = π(0.60) = 0.39, and
π(P r(Declining)) = π(0.04) = 0.14
• Second, obtain values of the payoffs:

Economy
Growing Stable Declining
Bonds 4.47 4.74 1.58
Stocks 5.92 3.87 -7.21
Mutual Funds 5.15 4.74 -4.47
 Probability Weighting: an Example

Probability weighting • You are facing the following investment payoff matrix:

2019-09-27 Growing
Economy
Stable Declining
Bonds $40 $45 $5
Stocks $70 $30 -$13
Mutual Funds $53 $45 -$5

Probability Weighting: an Example • The probability of the economy being in growing, stable,
and declining states are 0.36, 0.60, and 0.04 respectively.
p
• Your value function v(·) is defined by: v(x) = x/2 for
p
gains(x ≥ 0) and v(x) = −2 |x| for losses (x < 0).
Moreover, your initial reference wealth is $0.
• Finally, the probability-weighting function
√ is given by the
p
following expression: π(p) = √ √
( p + 1 − p)2
• Which investment is optimal?

• Third, obtain weighted values.

Bonds: 0.31 × 4.47 + 0.39 × 4.74 + 0.14 × 1.58 = 3.45

Stocks: 0.31 × 5.92 + 0.39 × 3.87 − 0.14 × 7.21 = 2.29
Mutual Funds: 0.31 × 5.15 + 0.39 × 4.74 − 0.14 × 4.47 = 2.79

• Obviously, in the given example, investing in bonds would be the

best option as it results in greatest weighted value.
 Probability Weighting

Introduction

Framing
effects in
• The probability-weighting function can help resolve the
decision-
making under
paradox that some people simultaneously buy lottery
risk tickets and insurance policies. This helps explain why
Bundling and people fear airplane crashes, terrorist attacks, and
mental
accounting
many other unlikely things so much.
The Allais • Probability-weighting also explains why people
problem and
the sure-thing purchase extended warranties on equipment such as
principle computers, in spite of the fact that simple
The Ellsberg
problem and
expected-value calculations suggest that for most
ambiguity people extended warranties are not a very good deal.
aversion
• The probability-weighting function can also account
Probability
weighting for the certainty effect: the tendency to overweight
Discussion outcomes that are certain. Events that are not certain
(even when their probability is very high) will be
underweighted relative to events that are certain.

50 / 53
 Probability Weighting

Probability weighting
2019-09-27
• The probability-weighting function can help resolve the
paradox that some people simultaneously buy lottery
tickets and insurance policies. This helps explain why
people fear airplane crashes, terrorist attacks, and
many other unlikely things so much.
• Probability-weighting also explains why people

Probability Weighting purchase extended warranties on equipment such as

computers, in spite of the fact that simple
expected-value calculations suggest that for most
people extended warranties are not a very good deal.
• The probability-weighting function can also account
for the certainty effect: the tendency to overweight
outcomes that are certain. Events that are not certain
(even when their probability is very high) will be
underweighted relative to events that are certain.

A good example regarding the certainty effect: Russian roulette Suppose that
you are forced to play Russian roulette, but that you have the option to pay to
remove one bullet from the loaded gun before pulling the trigger. Would you
pay more to reduce the number of bullets in the cylinder from four to three or
from one to zero? According to Kahneman and Tversky, if you are like most
people, you would pay more to reduce the number from one to zero than from
four to three. Why? Reducing the number of bullets from four to three would
reduce the probability of dying from 4/6 to 3/6. In this range, the probability-
weighting function is relatively flat, meaning fairly unresponsive to changes in
the underlying probability. Reducing the number of bullets from one to zero
would reduce the probability of dying from 1/6 to 0. Here, there is a jump from
π(1/6) > 1/6 to π(0) = 0 . Thus, the value to you of reducing the number of
bullets from one to zero exceeds that of reducing it from four to three.
 Probability Weighting

Introduction

Framing
effects in
decision- • The upshot is that people’s behavior in the face of risk
making under
risk depends not just on whether outcomes are construed
Bundling and
as gains or as losses relative to some reference point,
mental
accounting
but on whether or not the relevant probabilities are
low.
The Allais
problem and • In the domain of losses, people tend to be risk prone,
the sure-thing
principle except for gambles involving a low-probability event of
The Ellsberg significant (negative) value, in which case they may be
problem and
ambiguity risk averse.
aversion
• In the domain of gains, people tend to be risk averse,
Probability
weighting except for gambles involving a low-probability event of
Discussion significant (positive) value, in which case they may be
risk prone.

51 / 53
 Probability Weighting

Introduction

Framing
effects in
decision-
making under
risk

Bundling and
mental Table: Risk attitudes according to prospect theory
accounting

The Allais
problem and
Domain
the sure-thing
Probability
principle Losses Gains
The Ellsberg
problem and Low Risk averse Risk prone
ambiguity
aversion Moderate or High Risk prone Risk averse
Probability
weighting

Discussion

52 / 53
 Discussion

Introduction
• In this module we outlined some instances when people’s actual
Framing
effects in choices diverge from the predictions of the expected-utility
decision-
making under theory. Though the divergences are not universal, they are
risk substantial, systematic, and predictable, and they can have real,
Bundling and
mental
and sometimes adverse, effects on people’s decision-making.
accounting • We have discussed situations where people’s firmly held
The Allais intuitions about the rational course of action differ from the
problem and
the sure-thing recommendations of expected-utility theory, as in the presence
principle of ambiguous probabilities. This raises deep issues about the
The Ellsberg nature of rationality.
problem and
ambiguity • We have also explored more theoretical tools developed by
aversion
behavioral economists to capture the manner in which people
Probability
weighting actually make decisions. Among other things, we studied other
Discussion
components of prospect theory, including the S-shaped value
function and the probability-weighting function.
• One conclusion we arrived on was that, under certain
conditions, a person’s risk preferences can be reversed simply by
changing the frame of the relevant options. 53 / 53