Electronic copy available at: http://ssrn.

com/abstract=1597934
Foundations and Trends
R
in
Finance
Vol. 4, Nos. 1–2 (2009) 1–184
c 2010 H. Shefrin
DOI: 10.1561/0500000030
Behavioralizing Finance
By Hersh Shefrin
Contents
1 Introduction 2
1.1 Brief History 4
1.2 Organization of Monograph: Foundations and Trends 6
1.3 Home Bias and Omissions 9
2 Survey of Surveys 10
2.1 Hirshleifer (2001): Review of Psychology
Literature and Link to Asset Pricing 13
2.2 Barberis and Thaler (2003): Review of Behavioral Asset
Pricing Literature 19
2.3 Baker et al. (2007): Survey of Corporate Finance 27
2.4 Subrahmanyam (2007): Additional Perspectives 37
2.5 Critique: Weaknesses in the Behavioral Approach 42
3 Behavioralizing Beliefs and Preferences 47
3.1 Behavioral Beliefs: Heuristics and Biases 48
3.2 Behavioral Preferences 51
3.3 Regret 59
3.4 Self-Control 59
Electronic copy available at: http://ssrn.com/abstract=1597934
4 Behavioralizing Portfolio Selection Theory 61
4.1 Preference for Positively Skewed Returns:
Full Maximization 62
4.2 Quasi-Maximization 70
5 Behavioralizing Asset Pricing Theory 82
5.1 Stochastic Discount Factor 83
5.2 How Markets Aggregate Investor Attributes 88
5.3 Aggregation and the Shape of Sentiment 91
5.4 Risk and Return 92
5.5 Long-run Fitness 100
5.6 Pricing Dynamics 102
6 Behavioralizing Corporate Finance 112
6.1 Open Questions Raised by Baker et al. 113
6.2 Managerial Psychology and Decisions 116
6.3 Social Networks 119
6.4 Entrepreneurs 122
7 Behavioralizing the Approach to Financial Market
Regulation 130
7.1 Dynamic Tug-of-War 131
7.2 Biased Cost–Benefit Assessment 140
7.3 Regulation and Financial Crises 142
7.4 Differences in Perspective 156
8 Concluding Remarks 158
References 161
Foundations and Trends
R
in
Finance
Vol. 4, Nos. 1–2 (2009) 1–184
c 2010 H. Shefrin
DOI: 10.1561/0500000030
Behavioralizing Finance
Hersh Shefrin
Mario L. Belotti Professor of Finance, Leavey School of Business, Santa
Clara University, Santa Clara, CA 95053, USA hshefrin@scu.edu
Abstract
Finance is in the midst of a paradigm shift, from a neoclassical based
framework to a psychologically based framework. Behavioral finance is
the application of psychology to financial decision making and financial
markets. Behavioralizing finance is the process of replacing neoclassi-
cal assumptions with behavioral counterparts. This monograph surveys
the literature in behavioral finance, and identifies both its strengths
and weaknesses. In doing so, it identifies possible directions for behav-
ioralizing the frameworks used to study beliefs, preferences, portfolio
selection, asset pricing, corporate finance, and financial market regu-
lation. The intent is to provide a structured approach to behavioral
finance in respect to underlying psychological concepts, formal frame-
work, testable hypotheses, and empirical findings. A key theme of this
monograph is that the future of finance will combine realistic assump-
tions from behavioral finance and rigorous analysis from neoclassical
finance.
1
Introduction
Viewed as a field, behavioral finance is the application of psychology to
financial decision making and financial markets. Viewed as a process,
behavioral finance is about the transformation of the financial paradigm
from a neoclassical based framework to a psychologically based frame-
work. I refer to this process as “behavioralizing finance.”
Neoclassical finance has both strengths and weaknesses. Among its
main strengths is its systematic, rigorous framework. Among its main
weaknesses is its reliance on the unrealistic assumption that all decision
makers are fully rational. Behavioral finance also has both strengths
and weaknesses. Among its main strengths is its use of assumptions
based on findings from the psychology literature about how people
deviate from fully rational behavior. Among its main weaknesses is the
reliance by its proponents on an ad hoc collection of models that lack
mutual consistency and a unifying structure.
In this monograph I suggest that finance is moving to a new
paradigm that will combine structural features from neoclassical
finance and realistic assumptions from behavioral finance. To develop
this position, I describe the outlines of what such a synthesis entails,
and relate it to the issues associated with structure, coherence, and
consistency.
2
3
The behavioralization of finance involves intellectual shifts by two
groups. The first shift features neoclassical economists explicitly incor-
porating psychological elements into their models. For a sense of
neoclassical resistance, see Ross (2005). The second shift features
behavioral economists developing a systematic, rigorous framework.
For example, most of the current behavioral explanations for the cross-
section of returns rely on models that feature a single risky security,
rather than the cross-section itself. In addition, one well-known behav-
ioral result might well be incorrect. The behavioral result contends that
irrational investors can survive in the long run because they take on
more risk than rational investors. However, this contention is at odds
with theorems in the general literature about long-run survivability.
Shifts necessary for the behavioralization of finance are underway.
Some neoclassical economists have begun to develop behavioral models.
Two prominent examples are Jouini and Napp (2006) and Dumas et al.
(2009). At the same time, some behavioral economists are beginning to
develop models that are as rigorous as their neoclassical counterparts.
A good example is Xiong and Yan (2009), whose formal framework
shares much in common with Dumas et al. (2009).
These shifts are taking place against a historical backdrop in which
financial economists from both camps have displayed some of the same
psychological traits that are the focus of behavioral finance. Confirma-
tion bias may well be at the top of the list, the tendency to overweight
evidence that confirms one’s views relative to evidence that discon-
firms those views. My own experience has been that both neoclassi-
cal economists and behavioral economists have been resistant to the
proposal that each needs to move to a common middle ground where
psychological assumptions and rigor come together. Take for exam-
ple, the behavioral pricing kernel framework, which is an approach
I favor. For years, neoclassical economists have tended to claim that the
behavioral pricing kernel theorems discussed in Section 5 of this mono-
graph are wrong. However, this situation is changing slowly as some of
these theorems have come to be replicated in the work of more tradi-
tional economists, such as Jouini and Napp (2006). Likewise, behavioral
economists have resisted the unified, systematic approach, arguing that
a sentiment-based pricing kernel framework is insufficiently behavioral.
4 Introduction
In the past, both neoclassical and behavioral economists have chosen to
exclude papers relating to the behavioral pricing kernel at their respec-
tive NBER meetings.
I write this survey as some neoclassical economists and behavioral
economists have begun the process of moving toward a common middle
ground. My approach to this monograph is to describe the highlights
of the behavioral finance literature, as expressed through prior surveys,
and to identify some of the weaknesses of this literature. These founda-
tions are the subject of Section 2. In the remainder of the monograph,
I have two main objectives. My first objective is to discuss works which
have emerged since the past surveys appeared, or which those surveys
overlooked for one reason or another. My second objective is to present
some ideas about trends toward a unifying framework for behavioral
finance which captures some of the rigor in neoclassical finance. The
contents of these sections represent my thinking about the nature of the
common middle ground which brings together the best of neoclassical
finance and behavioral finance.
When it comes to asset pricing, the common middle ground involves
models that identify how markets aggregate heterogeneous investors’
beliefs. In some respects, some of the models employed by neoclassi-
cal economists are structurally similar to those employed by behav-
ioral economists. Two neoclassical examples are Detemple and Murthy
(1994) and Kurz (1997). A behavioral example is Shefrin and Statman
(1994). Notably, the neoclassical assumptions involve rational behavior
by all agents. For example, Kurz describes the agents in his framework
as holding “rational beliefs.” In contrast, the behavioral assumptions
specify that the source of the heterogeneity is a series of heuristics
that give rise to biased beliefs. The modelig distance between the two
approaches is short. However, moving from the neoclassical position to
the behavioral position requires a philosophical leap.
1.1 Brief History
Behavioral finance is relatively new as a subject of study. Although
the Journal of Finance published the first formal paper in behavioral
finance in 1972 (see Slovic (1972), financial economists did not begin to
1.1 Brief History 5
apply the concepts pioneered by Slovic and other psychologists work-
ing in behavioral decision making until the early 1980s. Even then,
it took more than a decade for the behavioral approach to gain trac-
tion. The first published work in behavioral finance by economists was
Shefrin and Statman (1984). In December 1984, the late Fischer Black,
then President-elect of the American Finance Association, invited Meir
Statman and me to organize the first behavioral session at the annual
meetings. At the session Statman presented our paper on the disposi-
tion effect, Shefrin and Statman (1985), and De Bondt presented his
work with Richard Thaler on overreaction, DeBondt and Thaler (1985).
These two papers set the stage for two main streams in the behavioral
finance literature, one pertaining to irrationality in investor behavior
and the other to inefficiency in asset pricing.
Subsequently, Thaler organized a behavioral research group, first at
the Sloan Foundation, and then later at the Sage Foundation, which
ultimately led to the NBER Behavioral Finance program. In addition
to De Bondt, Statman, Thaler, and myself, the main participants at the
Sloan and Sage Foundation included Daniel Kahneman, Amos Tversky,
Paul Andreassen, Robert Shiller, Fischer Black, Richard Roll, David
Dreman, Larry Summers, and Andrei Shleifer. Thaler collected many
of the papers written by members of this group into an edited collection:
see Thaler (1993).
Despite the beginnings of a convergence that I mentioned above,
currently there is a wide spectrum of views about behavioral finance. At
one extreme you will still find the view expressed by some that behav-
ioral finance is nothing more than a collection of findings about anoma-
lies relative to the efficient market hypothesis. At the other extreme you
will find the view that the behavioral approach already constitutes a
coherent paradigm, with a systematic approach involving psychologi-
cal concepts, a formal framework, testable hypotheses, and empirical
findings. Between the extremes are views that acknowledge that behav-
ioral finance possesses some structure and can be credited with some
successes, but overall still lacks coherence and consistency.
In the last decade several surveys of behavioral finance have been
written. Hirshleifer (2008a), Barberis and Thaler (2003), Subrah-
manyam (2007), and De Bondt et al. (2008) survey the general area,
6 Introduction
while Baker et al. (2007b) survey behavioral corporate finance. There
is little point in fully replicating what can already be found in those
surveys. Instead, my intent in this monograph is to summarize the key
features of those surveys, and relate these features to the process of
behavioralizing finance.
1.2 Organization of Monograph: Foundations and Trends
I have organized this monograph to highlight both the strengths and
weaknesses of the behavioral approach. Its strengths emanate from bas-
ing its foundations on a rich literature in psychology. Its weaknesses
emanate from an overall approach that has been piecemeal. There is no
generally accepted coherent behavioral counterparts to the workhorses
of neoclassical finance, mean-variance portfolios, asset pricing theory,
and the value maximizing agency theory approach to corporate finance.
In contrast, much of the literature in behavioral finance is scattered,
ad hoc, unsystematic, and lacking in discipline. In one sense the lack
of discipline is a positive, as behavioral finance is new, and it might
be a mistake to narrow its range of inquiry too early, thereby stifling
major new insights. However, it seems to me that many proponents of
behavioral finance have been excessively resistant to rigorous analysis,
and the time has come to move in the direction of increased rigor and
discipline.
Any discussion of paradigm transformation would be incomplete
without addressing how behavioral assumptions will impact the major
building blocks lying at the heart of the neoclassical framework. These
are: mean-variance portfolios, the efficient market hypothesis, the cap-
ital asset pricing model, the Black–Scholes option pricing formula, and
the Modigliani–Miller principle. This monograph will describe how the
substitution of neoclassical assumptions with behavioral assumptions
will impact these building blocks.
Section 2 highlights the foundations of behavioral finance by survey-
ing a sequence of surveys. The remaining five sections focus on trends
relating to a more rigorous approach to behavioral finance, or more
broadly, the behavioralization of finance.
1.2 Organization of Monograph: Foundations and Trends 7
Chapters 2–8
Section 2 is a survey of recent surveys of behavioral finance. The
section summarizes the major contributions to behavioral finance, as
seen through the eyes of some of its key contributors. This discussion
provides an overview of the field, and sets the stage for identifying both
the strengths and weaknesses of the behavioral approach. The remain-
ing sections provide an opportunity to present some thoughts on what
a systematic behavioral approach might entail.
Section 3 introduces the key psychological concepts used in
behavioral finance. These concepts divide naturally into beliefs and
preferences, both of which reflect human psychology. Among the belief-
related psychological concepts discussed in the section are unreal-
istic optimism, overconfidence, and representativeness. Among the
preference-related concepts are prospect theory, SP/A theory, regret,
and self-control.
Section 4 deals with behavioral portfolio selection. Beliefs pertain to
the manner in which investors’ judgments about risk and return reflect
bias. Behavioral beliefs involve a series of biases such as hot hand fal-
lacy, gambler’s fallacy, and home bias. Preferences pertain to the man-
ner in which investors frame and evaluate cash flow representations.
Framing effects lead many investors to adopt a piecemeal approach
to portfolio selection, in which purchasing decisions are influenced by
salience or attention, and realization decisions are influenced by a phe-
nomenon known as “the disposition effect.” The disposition effect is the
tendency to sell winners too early and ride losers too long in all months
of the year except December. Behavioral preferences also explain how
investors segment their portfolios into risk layers, and exhibit a taste for
positive skewness. The section describes a series of hypotheses about
the preference for dividend-paying stocks that stem from the behavioral
finance literature, and describes findings from tests of these hypotheses.
Section 5 describes how behavioral portfolio choices impact asset
pricing. The literature on behavioral asset pricing theory is eclectic,
and features a wide variety of models, centered on the concept of sen-
timent. Many of the models in the behavioral asset pricing literature
were developed to analyze overreaction and underreaction in financial
8 Introduction
markets. With the intent of describing a less eclectic, more system-
atic approach to asset pricing, the section focuses on a behavioral
SDF-based approach. This approach brings together the powerful asset
pricing tools favored by neoclassical asset pricing theorists and the
realistic assumptions favored by behavioral asset pricing theorists. The
behavioral SDF-based approach offers a unified treatment of behavioral
beliefs and behavioral preferences, showing how these features combine
to impact the mean-variance frontier, equity prices, the term structure
of interest rates, and option prices. In this respect, the discussion in
the section emphasizes the latter two applications.
Section 6 discusses behavioral corporate finance. Behavioral cor-
porate finance is concerned with the manner in which behavioral
beliefs, behavioral preferences, and inefficient prices impact the corpo-
rate financial decisions made by managers. The literature on behavioral
corporate finance involves the study of how psychology impacts capital
budgeting, capital structure, corporate governance, and mergers and
acquisitions. In the section I focus on three themes that strike me as
areas providing fertile ground for future work. The three areas are the
study of how specific individual traits impact corporate financial deci-
sions, the impact of group composition on collective corporate financial
decisions and corporate value, and the manner in which psychological
characteristics of entrepreneurs impact the choices they make about
risk and return.
Section 7 discusses the role which psychological factors play in the
structure of financial market regulation. The section points out the
importance of psychological forces in the shaping of the major leg-
islation underlying financial market regulation in the United States.
Examples of past regulation are merit regulation (“blue sky laws”), the
Securities Act of 1933, the Securities Exchange Act of 1934, Sarbanes–
Oxley, and regulations covering the Savings & Loan industry. Issues
which are more contemporary involve the regulation of financial deriva-
tives, investment banks, ratings agencies, hedge funds, and the credit
card industry. The section describes how capture theory, perceptions of
fairness, and shocks to the economic and financial system impact the
regulatory system, especially issues pertaining to the global financial
crisis that erupted in 2008.
1.3 Home Bias and Omissions 9
Section 8 contains some concluding remarks which highlight the
main points in this monograph.
1.3 Home Bias and Omissions
As I mentioned above, Section 2 provides a survey of recent surveys,
beginning with Hirshleifer (2008a) and moving on to Barberis and
Thaler (2003), Baker et al. (2007b), and Subrahmanyam (2007). All
of these surveys reflect the biases of their authors in terms of perspec-
tive and emphasis. Home bias is particularly evident, meaning authors
are prone to emphasize their own work. No single survey is compre-
hensive. For example, none of these surveys discusses the experimental
work on asset pricing bubbles, particularly the work of Smith: See his
Nobel address, Smith (2003).
In the interest of disclosure about home bias and omissions, I should
say at the outset that this survey will follow suit, meaning it will also
feature home bias and omissions. Surveys are part review, part syn-
thesis, and part editorial. I hold particular views about the state of
behavioral finance, its strengths, its weaknesses, and quo vadis–where
it is going? I plan to use this survey as a vehicle to communicate my
perspective about the direction in which I would like to see behavioral
finance specifically, and finance generally, go.
2
Survey of Surveys
This section describes the highlights of four surveys of behavioral
finance that have appeared in the past decade. Because there is a fair
amount of overlap in these surveys, I select portions of each to identify
key concepts and trace the different perspectives.
I begin with the survey by Hirshleifer (2008a). Hirshleifer presents
a thorough treatment of the underlying psychology upon which propo-
nents of behavioral finance draw. He also discusses how many of these
psychological concepts have been applied to asset pricing, and to a
lesser extent corporate finance. I focus on Hirshleifer’s review of the
psychology and some of the broad issues in asset pricing.
I then present the highlights of a survey by Barberis and Thaler
(2003). In discussing Barberis and Thaler’s survey, I focus on their
discussion of specific issues in asset pricing. For the discussion of cor-
porate finance, I describe the highlights of the survey written by Baker
et al. (2007b). Finally, I discuss the perspective of a more recent general
survey by Subrahmanyam (2007).
Proponents of behavioral finance draw on a rich literature in
psychology which documents phenomena that interfere with ratio-
nal beliefs and decisions. The list includes overconfidence, unrealistic
10
11
optimism, representativeness, conservatism, self-attribution bias, con-
firmation bias, anchoring, availability bias, loss aversion, aversion to
a sure loss, and probability weighting. The last three phenomena are
major features of prospect theory, a choice framework developed by
Kahneman and Tversky (1979). All other phenomena are part of an
approach known as heuristics and biases.
The volume edited by Kahneman et al. (1982) contains most of
the foundation contributions to the heuristics and biases literature.
A heuristic is a crude rule of thumb for making judgments about
probabilities, statistics, future outcomes, etc. A bias is a predisposi-
tion toward making a particular judgmental error. The heuristics and
biases approach studies the heuristics upon which people rely to form
judgments, and the associated biases in those judgments.
Glossary of Biases
Some biases are associated with specific heuristics. Examples of these
biases relate to such heuristic principles as availability, anchoring, and
representativeness.
Availability is the tendency to form judgments based on informa-
tion which is readily available, but to underweight information that is
not readily available. For example, a person might underestimate the
danger from riptides and overestimate the danger from shark attacks
because the media tends to report most shark attacks but rarely reports
less dramatic incidents involving riptides. Heuristics and biases associ-
ated with availability reflect the importance of salience and attention.
Anchoring is the tendency to formulate an estimate by using a pro-
cess that begins with an initial number (the anchor) and then making
adjustments relative to the anchor. Anchoring bias is the tendency for
the associated adjustments to be too small.
Representativeness is the tendency to rely on stereotypes to make a
judgment. For example, a person who relies on representativeness might
be especially bold in predicting that the future return of a particular
stock will be very favorable because its past long-term performance
has been very favorable. This is because they form the judgment that
favorable past performance is representative of good stocks. However,
12 Survey of Surveys
representativeness leads such predictions to be overly bold, because of
insufficient attention to factors that induce regression to the mean.
Although some biases relate directly to specific heuristics, other
biases stem from a variety of factors. For example, people tend to be
overconfident about both their abilities and their knowledge. People
who are overconfident about their abilities overestimate those abili-
ties. People who are overconfident about their knowledge tend to think
they know more than they actually do. In particular, people who are
overconfident about their knowledge tend to establish excessively nar-
row confidence intervals. Such people end up being surprised at their
mistakes more often than they had anticipated.
Other examples of biases that do not stem directly from specific
heuristics are unrealistic optimism and the illusion of control. Unre-
alistic optimism involves overestimating the probabilities of favorable
events and underestimating the probabilities of unfavorable events. The
illusion of control is overestimating the role of skill relative to luck in
the determination of outcomes.
Extrapolation bias leads people to overreact, developing unwar-
ranted forecasts that recent changes will continue into the future. Con-
firmation bias leads people to overweight information which confirms
their prior views, and to underweight information which disconfirms
those views. Conservatism is the tendency to underreact by overweight-
ing base rate information relative to new (or singular) information.
The affect heuristic refers to basing judgments on feelings rather than
on underlying fundamentals. Reliance on the affect heuristic is often
described as “gut feeling” or intuition.
Hindsight bias is the tendency to view outcomes in hindsight as
being more likely than warranted. That is, ex post the ex ante proba-
bility of the event that actually occurred is judged to be higher than
the ex ante judgment of that ex ante probability.
Categorization is the act of partitioning objects into general cat-
egories and ignoring the differences among members of the same
category. Categorization bias can be severe if the members of the same
category are different from each other in meaningful ways.
Status quo bias predisposes people to favor inaction over action, per-
haps because of regret aversion. See Samuelson and Zeckhauser (1988).
2.1 Review of Psychology Literature and Link to Asset Pricing 13
The discussion of prospect theory, which lies at the center of the
psychological-based approach to choice, is deferred until later in this
section and the section following.
2.1 Hirshleifer (2001): Review of Psychology
Literature and Link to Asset Pricing
2.1.1 Broad Issues Concerning Judgment
and Decision Bias
Hirshleifer provides a set of organizing principles to place the above
phenomena into a systematic structure. He argues that three princi-
ples provide a unified explanation for most biases. First, people rely on
heuristics because people face cognitive limitations. Because of a shared
evolutionary history, people might be predisposed to rely on the same
heuristics, and therefore be subject to the same biases. In this regard,
he mentions work on fast and frugal heuristics, as described in Todd
and Gigerenzer (2000). Second, the work of Trivers (1985, 1991) sug-
gests that people inadvertently signal their inner states to others. For
this reason, nature might have selected for traits such as overconfi-
dence, in order that people signal strong confidence to others. This
occurs because people who are well calibrated find it impossible to sig-
nal strong confidence credibly. Third, people’s judgments and decisions
are subject to their own emotions as well as to their reason.
Heuristic Simplification
Hirshleifer divides the items associated with heuristic simplification
into several subcategories. The first category relates to attention, mem-
ory, and processing. Items include salience and availability effects such
as underweighting the probability of events not explicitly available for
consideration, the use of habits as a way of economizing on thinking,
misattribution stemming from halo effects, misjudgments about truth
stemming from ease of processing, cue competition, and familiarity bias.
In this last respect, Trivers (1985) suggests that for evolutionary rea-
sons, people prefer familiar and similar individuals for reasons having
to do with genetic relatedness. Hirshleifer provides financial examples
for some of these items, such as the halo effect explaining why growth
14 Survey of Surveys
stocks tend to be overpriced, and familiarity explaining why investors
appear to exhibit a preference for local stocks.
The second subcategory for heuristic simplification pertains to nar-
row framing, mental accounting, and reference effects. Biases associated
with this subcategory largely relate to oversimplification of a deci-
sion task, perhaps because time and cognitive resources are limited.
Prospect theory, which is discussed in detail below, lies at the center
of this subcategory, with its focus on framing effects especially the iso-
lation effect, loss aversion, and reference points. The subcategory also
includes a variety of effects motivated by or associated with prospect
theory, such as the house money effect, regret, preference reversal, the
endowment effect, status quo bias, and anchoring. Hirshleifer cites a
few financial applications such as the disposition effect in connection
with mental accounting and money illusion in connection with framing.
The third category relates to representativeness, the tendency to
overrely on how representative an object is of the population from
which it is drawn. In some situations, representativeness can lead people
to underestimate unconditional population frequencies, a phenomenon
known as “base rate underweighting.” Representativeness can lead peo-
ple to overweight recent evidence (recency bias), draw inferences which
are too strong in the face of small samples (“law of small numbers”),
overpredict reversals when they know the process governing transitions
(“gambler’s fallacy”), and overpredict continuation when they infer the
process from recent observations (“hot hand fallacy”). As an applica-
tion, Hirshleifer cites the experimental work of Andreassen and Kraus
(1990), who found that in the presence of modest stock price fluctua-
tions, subjects exhibit gambler’s fallacy, but in the presence of a strong
trend exhibit hot hand fallacy.
The fourth and final subcategory relates to subtleties in the
updating of beliefs. In some types of situations people tend to pre-
dict reversals, whereas in others they predict continuation. Similarly,
representativeness suggests that people underweight base rate infor-
mation, whereas Edwards (1968) finds evidence that in some types
of situations, people underweight singular information relative to base
rate information. The challenge is to identify the relevant situations
which give rise to each. Part of the explanation might involve cognitive
2.1 Review of Psychology Literature and Link to Asset Pricing 15
processing costs and salience. This is why people might be predisposed
to underweight information that is abstract and statistical such as prob-
abilistic information and sample size, but overreact to scenarios and
concrete examples which are more easily processed. Griffin and Tver-
sky (1992) present a framework to address this issue, based on the
distinction between the “strength” of evidence (such as how extreme
or salient is the information) and the “weight” of evidence (such as how
statistically relevant is the information to prediction).
Self-deception
Overconfidence is a form of self-deception, along with cognitive disso-
nance, biased self-attribution, hindsight bias, and confirmatory bias. In
all of these cases, people feel better about themselves than they might
were they bias-free. Overconfident people take pride in feeling that
they have more ability than is actually the case and that they know
more than is actually the case. Cognitive dissonance leads people to feel
better about choices they made, even if the choices were nonoptimal.
Biased self-attribution leads people to feel that their skills were respon-
sible for their successes, and that external factors were responsible for
their failures. Hindsight bias leads people to feel that their forecasting
skills are better than is actually the case. Confirmatory bias leads peo-
ple to feel better about their views because they underweight evidence
that disconfirms their views, but overweight evidence that supports
their views. At the same time, Klayman and Ha (1987) suggest that
confirmatory bias might be an efficient mental shortcut rather than the
result of self-deception.
Emotions and Self-control
The emotion of fear exerts an important influence on the way peo-
ple make choices about risky alternatives. Many fear choices that are
ambiguous about the risk involved, possibly because they worry about
hostile manipulation of the odds they face.
Mood also appears to be an important determinant of how peo-
ple feel about taking risk. Arkes et al. (1988) found that lottery ticket
sales in Ohio increased after an Ohio State University football victory,
16 Survey of Surveys
presumably because the win elevated the moods of Ohio residents. Hir-
shleifer points out that mood exerts more of an impact on judgments
about abstract issues than concrete issues for which people possess
information.
Time Preference and Self-control
Neoclassical models tend to use exponential discounting, meaning a
constant time-invariant discount factor to reflect impatience. There is a
literature in psychology about hyperbolic discounting, where discount
factors decrease more rapidly than is the case with exponential dis-
counting. Hyperbolic discounting also gives rise to potential self-control
difficulties when it comes to immediate gratification and the delay of
unpleasantness. Applications of self-control to financial issues include
savings, liquidity premia, and the equity premium puzzle.
2.1.2 Social Interactions
By and large the literature in behavioral finance draws from the psy-
chology of individual behavior rather than social psychology. However,
there have been some applications of social psychology to issues involv-
ing both investments and corporate finance. A key issue for investments
concerns the process by which investors acquire information. In this
respect, Shiller and Pound (1989) report that interpersonal communi-
cation drew investors’ attention to the stocks they most recently had
purchased. Shiller (1999) emphasizes the role of culture in communica-
tion, stressing that because of limited attention, investors attach spe-
cial importance to ideas or facts which are reinforced by conversation
or ritual. Kuran and Sunstein (1999) suggest that such interactions
can produce “availability cascades” whereby topics under discussion
publicly generate momentum, thereby reinforcing prior discussion.
A key bias studied in social psychology is the “conformity effect,”
the tendency of people’s judgments and decisions to be skewed toward
supporting other members of one’s group. An example is Asch (1956)
who found that people will report judgments that conform to the judg-
ments of others in their group, even when they believe others to be
in error. Stasser et al. (1989) report that people tend to discuss issues
2.1 Review of Psychology Literature and Link to Asset Pricing 17
about which they have shared information rather than sharing which
they hold privately. Stasser and Titus (1985) present evidence that peo-
ple will withhold private information to avoid contradicting the views
of others in their group.
The literature in social psychology has identified other biases. For
example, because people tend to emphasize main points in conversation
and deemphasize qualifying details, listeners tend to form beliefs that
are too extreme Allport and Postman (1947), Anderson (1932). Another
example is “fundamental attribution error” where people, in the course
of determining the behavior of others, overestimate the importance
of disposition and underestimate the role of external circumstances.
Finally, in the “false consensus effect,” people come to believe that
others share their beliefs to a greater degree than is actually the case:
see Ross et al. (1977). Hirshleifer describes several applications of these
biases. For example, Welch (2000) describes the results of a survey
of financial economists in connection with the equity premium, and
reports a false consensus effect in this regard.
2.1.3 Modeling Alternatives to Expected Utility
Allais (1953) presented a set of paradoxes demonstrating that peo-
ple are unlikely to behave in accordance with expected utility theory.
Camerer (1995) describes a series of alternative frameworks, of which
the most prominent is prospect theory, (Kahneman and Tversky, 1979).
In prospect theory, the carriers of value (meaning arguments of the
utility function) are gains and losses relative to a pre-specified refer-
ence point. The value function is S-shaped so that it is concave in the
domain of gains and convex in the domain of losses. The value function
also features a kink at the origin, where it is steeper for losses than
gains. Probabilities are distorted by a probability weighting function in
which extremely low probabilities are treated as zero, moderately low
probabilities are overweighted, and moderately high probabilities are
underweighted.
2.1.4 Broad Asset Pricing Issues
Hirshleifer’s survey focuses on the impact of investor psychology on
asset pricing, relating to what he calls “the heart of the grand debate
18 Survey of Surveys
in finance spanning the last two decades.” He begins his discussion
with two presentations of asset pricing, Campbell (2000) and Cochrane
(2000) who emphasize how expected returns are related to “objective
external sources of risk.” Hirshleifer suggests that what differentiates
the behavioral approach from the neoclassical approach is its focus on
how expected returns are related to both risk and to investor misvalu-
ation. In this regard, he describes a conceptual starting point, a linear
framework like the CAPM in which expected return is a function of
two variables, one associated with objective risk and the other with a
proxy for mispricing.
Hirshleifer makes an interesting point about the impact of irra-
tional investors on asset prices, namely that in equilibrium arbitrage
is a double-edged sword. Whereas rational investors move prices closer
to their efficient levels, irrational investors move prices away from effi-
cient levels. He asserts that in consequence, equilibrium prices reflect a
weighted average of the different investors’ beliefs. As a result, rational
investors need not eliminate mispricing through their trades.
Consider where mispricing is most likely to be severe. One possibil-
ity is where information is sparse and arrives slowly, what Hirshleifer
describes as “the dusty, idiosyncratic corners of the market.” How-
ever, because of cognitive limitations, there might not be any perfectly
rational investors in the market. As a result, systematic factors as well
as idiosyncratic factors might be persistently mispriced. Moreover, the
magnitude of mispricing might depend on the degree to which investors
not subject to bias might have detected a profitable opportunity and
are seeking to exploit it. Rational investors might have difficulty in
knowing the degree to which other like minded investors are attempt-
ing to exploit the same profit opportunity.
The neoclassical view is that wealth flows from foolish to smart
investors, so that in the long-run prices are set by rational investors.
The behavioral view, developed in a series of papers by DeLong et al.
(1990a,b, 1991) is that if foolish investors underestimate risk and take
on more than would be rational, the additional premium earned might
counter the flow of wealth to rational investors. The same argument
might serve to explain how irrational investors induce persistent cross-
section return predictability over the long-term. Again, the wealth
flow to rational investors which stem from the mistakes of irrational
2.2 Barberis and Thaler (2003): Review of Behavioral Asset Pricing Literature 19
investors is countered by the irrational investors earning a higher risk
premium from their exposure to aggregate market risk.
Hirshleifer describes challenges respectively faced by those in the
neoclassical school and those in the behavioral school. The neoclassical
challenge is to identify objective risks to explain the various puzzles and
anomalies. The behavioral challenge is identify how irrational investors
can persistently affect prices when they are losing wealth to rational
investors.
2.2 Barberis and Thaler (2003): Review of Behavioral
Asset Pricing Literature
Barberis and Thaler (2003) suggest that behavioral finance has two
main building blocks, namely psychology and the limits to arbitrage.
By psychology they mean systematic departures from rationality of
beliefs (Bayes’ rule) and rationality of preferences (expected utility
maximization). Departures from Bayes’ rule occur because of heuristics
and biases described in Hirsheleifer’s survey. In the behavioral mod-
els discussed by Barberis and Thaler, expected utility maximization is
replaced by prospect theory. Barberis and Thaler also discuss a phe-
nomenon known as ambiguity aversion, which applies to the case of
uncertainty. Limits to arbitrage refer to obstacles preventing arbitrage
by fully rational agents from eliminating price inefficiencies.
2.2.1 Aggregate Stock Market
One of the most important applications of behavioral ideas to explain-
ing aggregate stock market movements involves a triplet of puzzles
involving the equity premium, risk-free rate, market volatility, and
predictability of returns. The source of these puzzles is the inability
of a standard neoclassical representative investor model, with plausi-
ble parameters, to explain the magnitude of the historical equity pre-
mium, return volatility, and return predictability in the U.S. market.
The neoclassical model implies an equity premium of 0.1 percent, not
the historical 3.9 percent. It implies a return standard deviation of 12
percent, not the historical 18 percent. And it implies an absence of
predictability, in contrast to evidence of predictability.
20 Survey of Surveys
The main behavioral approach to explaining the equity premium
involves the work of Benartzi and Thaler (1995) and Barberis et al.
(2001a). Both approaches use a representative investor whose prefer-
ences are based on prospect theory. While neoclassical expected util-
ity theory emphasizes risk aversion in respect to total consumption or
wealth on the part of investors, prospect theory emphasizes loss aver-
sion in respect to changes in consumption or wealth. Behavioral theo-
ries of the equity premium puzzle focus on how short investment time
horizons might induce loss averse investors to behave as if they were
extremely averse to risk, thereby inducing a high equity premium in
the market. In addition, if investors are concerned they have ambigu-
ous beliefs about stock market returns, then they might demand an
ambiguity aversion premium, thereby putting upward pressure on the
equity premium.
Behavioral explanations for the volatility puzzle are based on both
beliefs and preferences. The starting point for the puzzle is the price–
dividend ratio, based on the work of LeRoy and Porter (1981) and
Shiller (1981). A quick way of introducing the main issue is through
the well-known Gordon constant expected growth formula which can
be stated as
Pt
D
t+1
=
1
r−g
, where P denotes price, D denotes divi-
dend, r denotes required return, and g denotes the expected dividend
growth rate. This formula indicates the price–dividend ratio is driven
by required return and expected dividend growth. If these are stable
over time, then the neoclassical theory predicts that the price–dividend
ratio should also be stable.
In practice, the price–dividend ratio is quite volatile, thereby
prompting the question of whether volatility in the dividend growth
rate or required return is sufficient to explain the puzzle. Although there
is little evidence to suggest that the long-run dividend growth rate is
volatile, the required return on stocks is a function of the risk-free rate,
degree of market risk, and overall risk aversion. Barberis and Thaler
point out that it is difficult to reconcile the volatility of the price–
dividend ratio in terms of volatile interest rates or volatile risk. As to
volatile risk aversion, they mention the work of Campbell and Cochrane
(1999) who develop a habit formation model in which investors become
2.2 Barberis and Thaler (2003): Review of Behavioral Asset Pricing Literature 21
more risk averse as their consumption falls toward the level to which
they have become habituated.
Barberis and Thaler argue that behavioral explanations based on
representativeness suggest that investors’ expectations about future
dividend growth and risk display excessive volatility relative to recent
events. For example, investors might view a recent spurt in dividend
growth as “representative” of the long-run sustainable future growth
rate, thereby leading subjective g to exceed objective g. A similar state-
ment might apply to perceived risk.
A behavioral explanation of the volatility puzzle based on prefer-
ences is similar in spirit to the behavioral explanation for the equity
premium puzzle. In contrast to neoclassical expected utility theory, in
which investors are assumed to be risk averse and have stable attitudes
toward risk, prospect theory emphasizes that attitude toward risk is
nonconstant and varies according to whether outcomes are registered
as gains or as losses. Barberis et al. (2001a) develop a model with a
representative investor whose preferences reflect elements of prospect
theory and hedonic editing (Thaler and Johnson, 1990). In their model,
investors become more tolerant of a loss after a rise in the market has
generated gains for investors, and more averse to a loss after a drop in
the market reduces that gain. Moreover, in line with hedonic editing, if
the drop in the market leads investors to experience losses relative to
the amount they originally invested, then they will become much more
averse to further loss.
2.2.2 Cross-section of Average Returns
Behavioral finance is most associated with the anomalies literature,
whereby patterns associated with the cross-section of average stock
returns cannot be explained in terms of the capital asset pricing model.
Among the most notable anomalies that cannot be explained on a
CAPM risk-adjusted basis are the following:
1. Size premium: stocks with low market capitalization have
subsequently earned higher returns than stocks with high
market capitalization.
22 Survey of Surveys
2. Long-term reversals: losers based on prior 36 month perfor-
mance have subsequently outperformed winners based on the
same criterion, over 60 months.
3. Scaled price-ratios: stocks associated with high ratios for
book-to-market (BM) and earnings-to-price (EP) have sub-
sequently outperformed stocks associated with lower ratios
during the next year.
4. Momentum: winners based on prior short-term month per-
formance (less than one year) have subsequently outper-
formed losers based on the same criterion during the next
year.
5. Post-earnings-announcement drift: stocks associated with
positive earnings surprises have subsequently outperformed
stocks associated with negative earnings surprises during the
next year.
6. Dividend initiations and omissions: For a year after the
announcement, stocks associated with dividend initiations
have subsequently outperformed the market, whereas stocks
associated with dividend omissions have subsequently under-
performed the market.
7. New Issues and Repurchases: Stocks of firms that have issued
new (repurchased) equity have subsequently underperformed
(outperformed) a control group of firms matched on size
and book-to-market equity over a five (four) year period.
Fama and French (1993, 1996) develop a three-factor model which
empirically captures many of the anomalies, and for which they provide
a neoclassical risk-based interpretation. The factors relate to the market
return, the difference in returns between small firms and large firms,
and the difference in returns between high and low stocks ranked by
BM.
Barberis and Thaler raise two concerns in connection with the
Fama–French risk-based explanation. First, the return to low BM
stocks tends to be below the risk-free rate. Second, neoclassical theory
measures risk by the covariance of returns with the marginal utility of
aggregate consumption growth. In this regard, the risk to a security is
2.2 Barberis and Thaler (2003): Review of Behavioral Asset Pricing Literature 23
that it will pay a low return in bad times when the marginal utility of
consumption growth is high. Therefore, a security earns a risk premium
for generating a low return in bad times when aggregate consumption
growth is low. However, there is little if any empirical support that
stocks earning above average returns perform especially poorly during
periods of low aggregate consumption growth.
There are a variety of behavioral belief-based models that have been
developed to explain cross-sectional patterns. These models emphasize
the combination of momentum in the short-term and reversals stem-
ming from overreaction in the long-term.
Barberis et al. (1998) develop a model involving a risk neutral rep-
resentative investor whose beliefs are biased by conservatism and rep-
resentativeness. Their model features a risk-free security and a risky
security. Conservatism leads the representative investor to underreact
to new information unless the new information is part of a run, in
which case representativeness leads the representative investor to over-
react to the run. When overreaction leads the risky security to become
mispriced, a subsequent correction eventually results.
Daniel et al. (1998, 2001) develop a model in which prices reflect
the beliefs of a risk-neutral investor whose biases stem from overconfi-
dence and self-attribution bias. In this behavioral combination, self-
attribution bias leads the price-setting investor to be overconfident
about information which he has generated on his own (his “private
signal”) relative to information which is publicly available. The model
produces momentum when external events confirm the price-setting
investor’s private signal, thereby increasing the investor’s overconfi-
dence. However, when the excessive weight applied to the private signal
leads the risky security to be mispriced, overreaction and a subsequent
correction eventually result.
Hong and Stein (1999) present a model featuring the interaction
between “newswatchers” who base their trades on privately observed
fundamentals and “momentum traders” who base their trades on the
most recent price change. Privately observed “news” diffuses slowly
into prices, at which point momentum traders amplify the effect. The
resulting momentum and subsequent positive feedback effect operate
like a mechanical thermostat which does not click off when the correct
24 Survey of Surveys
temperature is reached. The resulting overreaction eventually leads to a
correction. Hong and Stein do not emphasize particular psychological
biases in their approach, although the traits of “newswatchers” are
similar to those of the price-setting agents in the model developed by
Daniel et al. Perhaps, momentum traders suffer from confirmation bias
and the opposite of aversion to ambiguity, to the extent that they
rely solely on technical analysis and fail to generate information about
fundamentals.
Chopra et al. (1992) and La Porta et al. (1997) find that long-
term reversals associated with past winners and losers as well (low
BM stocks) appear to occur at the time of earnings are announced.
Barberis and Thaler suggest that these findings reflect investors receiv-
ing news that their past forecasts, driven by representativeness, were
extreme. Hong et al. (2000) find that momentum is more pronounced
in the stocks of small cap firms and firms with low analyst coverage.
They suggest that this finding supports a hypothesis stemming from
the Hong–Stein model.
A series of papers discuss the implications associated with heteroge-
neous beliefs in combination with restrictions on short sales. See Miller
(1977), Diether et al. (2004), Chen et al. (2002), and Jones and Lamont
(2002). Short-sales restrictions create an asymmetry in which the beliefs
of optimistic investors carry more weight in price formation than the
beliefs of pessimistic investors. In this regard, the evidence indicates
that increased breadth of ownership or higher costs of shorting will
lead to upward bias in price and subsequent lower returns. Harrison
and Kreps (1978) and Scheinkman and Xiong (2003) point out that in
the presence of short-sale constraints, investors who believe a security is
overvalued might choose to buy the security in anticipation of selling it
later to investors who are especially optimistic. Hong and Stein (2003)
suggest that pessimistic investors who cannot short an overvalued stock
will buy the stock after its price declines to fundamental value, but not
if its price increases. This leads to an asymmetric volatility response
leading to skewness in the return distribution.
The model developed by Barberis and Huang (2001) explains the
value premium associated with BM by suggesting that loss aversion
and narrow framing lead investors to act as if they are more averse to
2.2 Barberis and Thaler (2003): Review of Behavioral Asset Pricing Literature 25
risk after a stock has declined in price. The dynamics are similar to the
model developed by Barberis et al. (2001a).
Although the behavioral models described above were developed
to explain the cross-section of returns, it is worth noting that they all
feature a single risky asset rather than multiple assets whose returns are
jointly determined. This means that the explanations associated with
these models, are at best suggestive of the role played by underlying
psychological phenomena.
A similar statement holds in respect to explanations of the closed-
end puzzle (Lee et al., 1991). The puzzle concerns the existence
of a closed-end fund discount (or premium, when positive) relative
to net asset value. The behavioral approach to the puzzle is based
on the notion that investor sentiment varies over time, fluctuating
between degrees of optimism and pessimism. Pessimistic sentiment pro-
duces closed-end fund discounts whereas optimistic sentiment produces
closed-end fund premiums. The sentiment-based explanation suggests
that closed-end fund prices will covary in response to changing sen-
timent. In equilibrium, rational investors will demand a positive risk
premium as compensation for bearing the component of risk associated
with changes in investor sentiment as well as the component associated
with changing fundamentals.
One of the most important aspects of the behavioral approach to
risk premiums is the idea that the premium can be decomposed into a
fundamental component and a systematic sentiment component, where
sentiment is understood to mean excessive optimism or pessimism.
In this respect, being “systematic” means that sentiment should also
explain the premiums associated with small cap stocks, high BM stocks,
and so on. Barberis and Thaler point out that the discount attached to
closed-end funds as a class and premium attached to small cap stocks
covary. They also review work by Fama and French (1993, 1995) show-
ing the existence of common factors in earnings and returns associated
with small cap stocks. In this regard, they emphasize that while such
common factors exist, they are weaker for earnings than for returns,
thereby suggesting that other issues are at work. Barberis and Shleifer
(2003) suggest that investors are prone to “categorization” and treat
certain members of certain groups of stocks (such as small cap stocks)
26 Survey of Surveys
as being more similar than fundamentals warrant. As a result, cate-
gorization produces common factors in returns to stocks in the same
group. A pertinent example of a group is the S&P 500. See Barberis
et al. (2001b) who document the impact of a stock’s beta when the
stock is first included in the index.
2.2.3 Investor Behavior
Barberis and Thaler discuss how psychology influences the behavior of
individual investors, an issue they describe as being less controversial
than issues involving the determinants of market prices. They introduce
the discussion by mentioning the trend toward increased participation
in equities and the increased emphasis on defined contribution plans in
retirement planning.
Individual investors do not choose mean-variance portfolios.
Instead, their portfolios tend to be insufficiently diversified. Benartzi
and Thaler (2001) suggest that ambiguity aversion and the desire for
familiarity lead investors to exhibit “home bias,” the tendency to over-
weight securities with which investors are familiar. French and Poterba
(1991) document the high concentrations of domestic equity in the
equity portion of investors’ portfolios in the USA, Japan, and the
United Kingdom Grinblatt and Keloharju (2001a) find that Finnish
investors concentrate on the stocks of firms that are located close to
them geographically. Huberman (2001) documents a regional bias in
the holding of stocks of regional telephone companies. Benartzi (2001)
reports a strong bias toward own company stock in 401(k) plans.
Investors might also choose poorly diversified portfolios because they
use crude heuristics, such as the “1/n” heuristic which involves divid-
ing assets equally across choices. See Benartzi and Thaler (2001) who
report evidence of instances in which the relative number of equity
funds appears to drive the portion of participants’ allocations toward
equity.
Many investors trade excessively and unprofitably. Barber and
Odean (2000) find that for individual investors, risk-adjusted portfolio
returns are negatively correlated with the frequency of trading. They
attribute this result to a combination of overconfidence and transaction
2.3 Baker et al. (2007): Survey of Corporate Finance 27
costs. That is, overconfident investors mistakenly believe that their
skills are sufficiently high to compensate for the costs of trading.
In terms of the selling decision, investors are prone to pay more tax
than necessary because they sell winners too quickly and hold losers
too long. Shefrin and Statman (1985) call this behavior “the disposition
effect” and attribute it to the following series of psychological phenom-
ena: the reference point property, mental accounting, regret aversion,
and self-control. A key feature of the disposition effect for stocks is that
the original purchase price serves as the reference point. Using discount
brokerage data, Odean (1998a) provides strong evidence of the dispo-
sition effect. Genesove and Mayer (2001) present evidence from the
housing market that homeowners are reluctant to sell their homes at
prices below original purchase price. As study by Coval and Shumway
(2005) demonstrates that the disposition effect is not confined to indi-
vidual investors. Professional traders in the Treasury Bond futures pit
at the Chicago Board of Trade take greater intraday risks in the after-
noon if they have incurred losses in the morning. Barberis and Thaler
do touch on issues of pricing in connection with the disposition effect.
They describe work by Grinblatt and Han (2004) which indicates that
the disposition effect might underlie momentum. Momentum results
after gains when investors who are eager to take profits after good
news retard the adjustment of price to fundamental value. Momen-
tum results after losses when investors who are eager to hold their
stocks after bad news retard the adjustment of price to fundamental
value.
In terms of the buying decision, Odean (1999) and Barber and
Odean (2008) find that investors are prone to buy stocks to which their
attention has recently been drawn. Operationally, attention for stocks
is based on one or more of the following traits: abnormally high trading
volume, extreme returns, and stocks with news announcements.
2.3 Baker et al. (2007): Survey of Corporate Finance
Baker et al. (2007b) provide a two-pronged approach in surveying the
literature on behavioral corporate finance. The first prong pertains
to the impact of irrational investor behavior on rational corporate
28 Survey of Surveys
managers, while the second prong pertains to the manner in which
irrational decisions by corporate managers impact corporate value.
2.3.1 The Irrational Investors Approach
There are several reasons to believe that some corporate managers
might have the ability to identify mispricing. First, they have superior
information about their own firms. In this respect there is evidence
that corporate managers earn abnormal returns on their own trades.
See Muelbroek (1992), Seyhun (1992), and Jenter (2005). Second, they
are less constrained by the limits of arbitrage than equally smart money
managers. Third, managers might have access to efficient heuristics: see
Baker and Stein (2004).
Consider how rational managers might function in a world in which
irrational investors cause inefficiencies in market prices. Baker et al.
suggest that these managers balance three objectives: (1) maximizing
fundamental value; (2) undertaking actions to boost the market value
of their firms above fundamental value (catering); and (3) undertaking
financing decisions to exploit temporary mispricings of the securities
issued by their firms (market timing).
Theory
Baker et al. develop a model to capture the criterion function a ratio-
nal manager uses to balance these objectives. For this purpose, think
about a neoclassical firm whose investment and financing strategies are
determined independently. Define e as the deviation in the firm’s book
value of equity from its optimal neoclassical level. Let K be the amount
of capital invested, f(K, e) be a function specifying the present value
of future cash flows inclusive of financing effects, and λ be a number
between 0 and 1 measuring the degree to which the manager’s horizon
is long-term. The product eδ represents the value transfer to original
shareholders from exploiting the mispricing. The criterion function has
the form:
λ[f(K; ) − K + eδ(·)] + (1 − λ)δ(·). (2.1)
2.3 Baker et al. (2007): Survey of Corporate Finance 29
The first-order conditions for this problem are as follows:
f
K
= 1 −

e +
(1 − λ)
λ

δ
K
, (2.2)
−f
e
= δ(·) +

e +
(1 − λ)
λ

δ
e
. (2.3)
In the neoclassical benchmark for Equation (2.1) managers have a
long-term horizon (λ = 1) and there is no mispricing (δ = 0). In this
case, Equation (2.2) states that f
K
= 1 (meaning the net present value
of investment is zero at the margin) and e = f
e
(0) = 0.
Consider how choices of K and e are impacted by λ < 1 and
δ = 0. Begin with Equation (2.2). Suppose that δ
K
> 0, meaning higher
K exacerbates overpricing. In this case, Equation (2.2) implies that
f
K
< 1, so that managers cater to investor sentiment by increasing K
relative to the neoclassical case.
Next consider Equation (2.3). Consider the case in which δ > 0
(overpricing) combined with δ
e
< 0. The inequality means that when
the firm issues more equity, the resulting selling pressure leads the
degree of overpricing to decline. In this case, Equation (2.3) is analo-
gous to the microeconomic problem faced by a monopolist who chooses
quantity to equate marginal revenue and marginal cost. In this analogy,
e corresponds to quantity and δ corresponds to price. The right-hand
side of Equation (2.3) corresponds to the two terms making up marginal
revenue, while the left-hand side corresponds to marginal cost. In this
respect, when e > 0, the firm takes on too much equity relative to debt,
thereby losing tax shield and related benefits, in which case f
e
< 0.
Equation (2.3) indicates that the firm chooses e > 0, up to the point in
the monopoly analogy whereby marginal revenue equals marginal cost.
There is also the case δ < 0 (underpricing). A firm that is finan-
cially constrained would find itself unable or unwilling to raise external
financing, so that e might be negative with f
e
> 0. For managers with
long enough horizons (λ sufficiently close to 1), Equation (2.2) would
imply low investment as f
K
> 1. Such firms are reluctant to issue under-
priced equity in order to finance investment.
30 Survey of Surveys
Empirical Studies Involving Real Investment
Most research about real investment and mispricing focuses on the case
of overpricing. Baker et al. review some of the early literature on this
topic, beginning with Keynes’ discussion about “animal spirits” dur-
ing the 1920s, and moving on to other bubble episodes involving elec-
tronics (1959–1962), growth stocks (1967–1968), and dot-com stocks
(late 1990s). They point out that the first modern empirical studies
were inconclusive about the impact of mispricing on the magnitude of
investment. These studies sought to control for fundamental value and
searched for a residual effect on investment from stock prices.
The authors of later studies searched for evidence that proxies
for mispricing impact investment. Chirinko and Schaller (2001, 2004),
Panageas (2003), Polk and Sapienza (2004), and Gilchrist et al. (2004)
find such evidence. Polk and Sapienza also show that this impact
is greater for shorter managerial horizons, as happens when share
turnover is higher. Following Stein (1996), Baker et al. (2003) study
whether investment will be most sensitive to mispricing for equity-
dependent firms, the case discussed above where e < 0.
Baker et al. point out that open questions remain about the mag-
nitude of the mispricing-investment effect and the associated efficiency
implications. In this last respect, Polk and Sapienza (2004) find that in
the cross-section, firms that invest more earn lower subsequent returns.
Titman et al. (2004) find the same pattern, and Lamont (2000) reports
something similar in a time series analysis. Nevertheless, Baker et al.
do not view these results as strong evidence of inefficiency in an ex ante
sense, and claim not to be convinced that the pattern stems from cater-
ing.
Managers might use mergers and acquisitions to cater to investor
sentiment, perhaps by using the overvalued stock of their own firms
to acquire less overvalued targets, or by engaging in a combination
for which sentiment is positive, see Shleifer and Vishny (2003). This
theory offers some insight into the time-series relationship between
merger volume and stock prices. It also explains why cash acquirers
subsequently outperform stock acquirers (who earn negative long-run
returns), see Loughran and Vijh (1997) and Rau and Vermaelen (1998).
2.3 Baker et al. (2007): Survey of Corporate Finance 31
Evidence of a positive correlation between market-level mispricing and
merger volume can be found in Dong et al. (2003), and Ang and Cheng
(2009). These studies report that undervalued targets are more likely
to be hostile to acquisition, and that overpriced acquirers tend to pay
higher takeover premiums. Moreover, subsequent returns to mergers
conducted during high-valuation periods tend to be poor (Bouwman
et al., 2009).
Baker et al. pose an interesting question about mispricing-driven
mergers: why would an overvalued firm use a merger instead of issuing
more equity? They suggest a framing-based explanation that a merger
might be more effective at hiding the underlying misvaluation. Baker
et al. (2004) also propose that investors might respond more passively
to accepting the acquirer’s shares as part of the merger (a form of status
quo bias), but would not actively purchase those shares when issued
directly.
An issue related to mergers and acquisition is the diversification
discount. Although there is little in the behavioral finance literature
dealing with this issue, Baker et al. suggest that the 1960’s conglomer-
ate wave reflected a desire by managers to cater to investors’ preference
for conglomerates. They discuss the work of Klein (2001) who finds a
diversification premium of 36 percent in the period 1966–1968, followed
by a decline between July 1968 and June 1970, as reported in Raven-
scraft and Scherer (1987).
Empirical Studies Involving Financial Policy
As was mentioned in the theoretical section above, mispricing can lead
a firm’s managers to engage in market timing, for example by issuing
new shares when equity is overpriced and by repurchasing shares when
equity is underpriced. Survey evidence presented by Graham and Har-
vey (2001) indicates that financial managers view such market timing
activities as very important. Jenter (2005) presents related evidence
that managers conduct secondary equity offerings (SEOs) while simul-
taneously engaging in insider selling.
There is considerable empirical evidence suggesting market tim-
ing in respect to equity. Loughran et al. (1994) report a high positive
32 Survey of Surveys
correlation between volume of initial public offerings (IPOs) and inter-
national stock market valuations. As for seasoned equity offerings
(SEOs), Marsh (1982) reports that equity issuance increases after
recent stock market appreciation. Loughran and Ritter (1995) provide
evidence suggesting the presence of mispricing. Relative to a size-
matched firm, after five years the average IPO underperforms its non-
IPO peer by 30 percent. After reviewing the relevant literature, Baker
et al. conclude that on average U.S. firms issuing equity subsequently
underperform the market by between 20 and 40 percent over five years.
At the same time, they also point out that the equity market timing
issue is controversial, and is still a subject of considerable debate.
In terms of share repurchases, Brav et al. (2005) report survey
evidence indicating that the great majority of firms agree that they
repurchase shares when they view their shares as underpriced in the
market. Indeed, Ikenberry et al. (1995) find that the average repur-
chaser outperformed firms matched by size and book-to-market equity
by 12 percent over the subsequent four years. See also Ikenberry et al.
(2000). Dichev (2004) reports that firms who engaged in market timing
reduced their cost of capital by 1.3 percent a year.
Moving from equity to debt, the survey evidence presented by Gra-
ham and Harvey (2001) does indicate that firms issue debt when they
feel that interest rates are low. Moreover, firms tend to issue short-term
debt when the yield curve is steep, and when waiting for long-term
rates to fall. Marsh (1982) reports that in his sample of U.K. firms,
interest rates impacted the debt–equity choice. Using a sample of U.S.
firms, Guedes and Opler (1996) report that firms do indeed issue short
maturity debt when the yield curve is steeply sloped, and longer matu-
rity debt when the yield curve is flatter or negatively sloped. In this
respect, empirically subsequent bond returns are positively correlated
with the slope of the yield curve, thereby supporting the market timing
decisions managers make about debt issuance. Moreover, the Graham–
Harvey survey evidence indicates that 44 percent of firms engage in
cross-border market timing to exploit differential interest rates. In this
respect, Froot and Dabora (1999) present evidence of cross-border mis-
pricing, while Henderson et al. (2009) report that subsequent to periods
of high foreign issues, returns tend to be low.
2.3 Baker et al. (2007): Survey of Corporate Finance 33
Interestingly, Richardson and Sloan (2003) find that equity returns
tend to be low after debt issuance, suggesting that firms might be tim-
ing their idiosyncratic credit quality. That said, there are other explana-
tions for the pattern. For example, as equity valuation is a component of
the cost of debt, equity overvaluation might make debt cheap. There-
fore, firms issue debt because it is cheap, and the overvalued equity
subsequently corrects itself.
If managers engage in market timing, and are also subject to sta-
tus quo bias, then firms which issue equity when they believe their
stocks are overpriced will tend to feature low leverage. Baker and Wur-
gler (2002) provide such a market timing-based explanation of capital
structure. Their analysis employs a weighted measure of market-to-
book equity, where the weights are based on the degree of external
financing. A high value of this ratio suggests that the firm raised the
bulk of its external equity or debt when its market-to-book ratio was
high. According to the theory, in the cross-section the weighted average
will be negatively related to debt-to-assets. Despite the evidence sup-
porting the prediction, Baker et al. point out that the issue is still under
debate. One issue is that the relationship Baker and Wurgler identify
might stem from growth opportunities rather than market timing. See
Hovakimian (2006).
Other Corporate Decisions
Besides investment and financial policy, behavioral phenomena play a
role in other corporate decisions such as dividends, choice of firm name,
earnings management, and executive compensation.
The major issue in respect to dividends is whether firms cater
to investor sentiment by initiating dividends when sentiment is pos-
itive and omitting dividends when sentiment is negative. Baker et al.
(2003) measure a “dividend premium” as the difference between aver-
age market-to-book equity between payers and nonpayers. They claim
to find evidence of catering by noting that initiations increase when
the dividend premium is high, and omissions increase when the divi-
dend premium is negative. Baker and Wurgler (2004a,b) suggest that
catering might explain the result described in Fama and French (2001)
34 Survey of Surveys
about the sharp decline between 1978 and 1999 of the number of firms
that pay dividends. However, they also note that the catering theory of
dividends is limited in that it only explains initiations and omissions,
not the magnitude of dividends. In addition, catering as defined above,
does not address other psychological issues associated with dividends
such as self-control, regret aversion, and hedonic editing (Shefrin and
Statman, 1984).
Cooper et al. (2001) presented evidence that during the Internet
bubble, many small firms changed their name to include a “dotcom”
reference. They noted that consistent with a catering motive, such name
changes resulted in an average announcement effect of 74 percent.
A survey conducted by Graham et al. (2004) found that financial
managers believe that investors care more about earnings per share
(EPS) than cash flows, and for this reason are willing to sacrifice net
present value for higher EPS. For this reason, earnings manipulation
is a form of catering. Teoh et al. (1998a,b) report that firms who most
aggressively manage earnings are associated with the highest underper-
formance in connection with equity issuance. Bergstresser et al. (2004)
found that as CEO compensation becomes more sensitive to share price,
earnings management increases. Sloan (1996) finds that firms associ-
ated with high accruals tend to feature low subsequent returns.
2.3.2 The Irrational Managers Approach
Consider the opposite side of the coin, where market prices are effi-
cient, but corporate managers are subject to psychological pitfalls. The
most common pitfalls in the behavioral corporate finance literature are
unrealistic optimism and overconfidence. Here unrealistic optimism is
typically interpreted as overestimating the first moment of cash flows
or returns, and overconfidence is interpreted as underestimating the
second moment (risk as measured by the standard deviation).
2.3.2.1 Theory
The theoretical framework for the case of irrational managers operat-
ing within efficient markets is similar in structure to the model used to
study the case of rational managers operating within irrational markets.
2.3 Baker et al. (2007): Survey of Corporate Finance 35
A new variable γ is added to the model to represent the combination of
managerial excessive optimism and overconfidence. An unbiased man-
ager is associated with γ = 0. The criterion function has the form:
(1 + γ)f(K; ) − K − eγf(K; ). (2.4)
The first-order conditions for this problem are as follows:
f
K
=
1
1 + (1 − e)γ
, (2.5)
(1 + γ)f
e
= γ(f(K; ) + ef
e
(K; )). (2.6)
In the neoclassical benchmark for Equation (2.4) γ = 0. This results
in the usual first-order conditions f
K
= 1 and f
e
= 0. When γ > 0 man-
agers overinvest, and balance a shift away from the neoclassical capital
structure against perceived losses from market timing. Suppose that
e > 0. This happens when the firm is cash poor and chooses to issue
more equity than is associated with the neoclassical case. Equation (6.1)
implies that the degree of overinvestment falls, thereby leading to cash
flow sensitivity. If there is an upper bound on leverage (f
e
> 0), then
a financial pecking order results, as managers never sell equity they
perceive to be underpriced unless it is absolutely necessary in order to
fund investment.
A collection of papers in the behavioral literature points out that
unrealistic optimism and overconfidence can produce benefits in the
presence of agency-related managerial traits that lead to underinvest-
ment, such as risk aversion or debt overhang. (Gervais et al., 2003; Goel
and Thakor, 2008; Hackbarth, 2009).
2.3.2.2 Empirical Issues
A key empirical task associated with the rational managers/inefficient
market approach involves identifying proxies for mispricing. Baker et al.
discuss a variety of possible measures, such as market-to-book as an
ex ante measure and abnormal subsequent realized returns as an ex post
measure.
By the same token, a key empirical task associated with the irra-
tional managers/efficient market approach involves identifying proxies
36 Survey of Surveys
for excessive optimism and overconfidence. This is especially impor-
tant in that the behavioral structure represented by Equation (2.4)
is formally similar to an agency structure, with γ representing pri-
vate managerial benefits and eγf(K; ) being replaced by a function
c(e) corresponding to the agency cost of raising external equity. This
means that overinvestment, underinvestment, cash flow sensitivity, and
pecking order financing have both agency cost explanations and behav-
ioral explanations. For this reason, identifying proxies which distinguish
between the two possibilities is a critical task.
Malmendier and Tate (2005) suggest a longholder proxy for exces-
sive optimism and overconfidence. The longholder proxy identifies chief
executive officers (CEOs) by their propensity to hold in-the-money
executive options on their own firm’s stock until expiration. The point
of the proxy is that because CEOs portfolios are overly concentrated
in the value of their own firms, rational CEOs would exercise in-the-
money options long before expiration, whereas excessively optimistic,
overconfident CEOs would wait the expiration year to exercise. Mal-
mendier and Tate find that cash flow sensitivity is pronounced among
firms whose CEOs are longholders, and especially high when the firm
is equity dependent.
There is ample evidence suggesting that excessive optimism and
overconfidence lead to overinvestment. Merrow et al. (1981) and Stat-
man and Tyebjee (1985) present evidence of excessive optimism in
project cost forecasts. Cooper et al. (1988) report that 68 percent of
entrepreneurs believe that their own startups are more likely to succeed
than comparable firms. Moreover, only 5 percent believe the odds to
be worse, and one-third perceive the risk of failure to be close to zero.
Indeed, half of startups do not survive more than three years (Scarpetta
et al., 2002). Given the undiversified nature of entrepreneurial invest-
ment, Moskowitz and Vissing-Jorgensen (2002) suggest that returns to
private equity seem too low, a point to which I return in Section 6.
Landier and Thesmar (2009) survey French entrepreneurs. They
find that 56 percent initially estimate that there will be “development”
in the near future and only 6 percent expect difficulties. However,
three years into their ventures, only 38 percent expect development
and 17 percent expect difficulties. To counter excessive optimism
2.4 Subrahmanyam (2007): Additional Perspectives 37
and overconfidence, Landier and Thesmar discuss contracts which
transfer control from entrepreneurs to investors when the former per-
sist in following their initial plans, despite signals suggesting changes.
They also suggest that because excessively optimistic, overconfident
entrepreneurs underestimate short-term risks, they choose short-term
financing over long-term financing, thereby losing control of their enter-
prises to investors more frequently than they planned.
Malmendier and Tate (2008) apply their longholder analysis to
mergers and acquisitions. Roll (1986) suggested that hubris leads man-
agers of acquiring firms to overpay for targets, thereby suffering from
“the winner’s curse.” As it happens, longholder CEOs complete more
mergers than longholders, especially diversifying mergers. In addition,
longholder CEOs of the least equity dependent firms are more apt to
engage in acquisitions, perhaps because they are least concerned about
financing the acquisition with equity they believe to be underpriced.
Baker et al. also discuss behavioral managerial patterns dis-
tinct from excessive optimism and overconfidence. For example, the
Graham–Harvey survey evidence indicates that many financial man-
agers rely on heuristics such as the payback rule and use firm-wide
discount rates to compute hurdle rates instead of risk-specific hurdle
rates. More than half of firms use target debt–equity ratios, less than
10 percent set strict targets, and target debt–equity ratios tend to be
based on book values instead of market values as called for in the
neoclassical framework.
Managers also exhibit reference point effects. They are prone to
“throw good money after bad:” (Statman and Sepe, 1989; Guedj and
Scharfstein, 2004). Loughran and Ritter (2002) apply a portion of the
framework presented by Shefrin and Statman (1984) to suggest that
initial underpricing for IPOs is pronounced when the offer price lies
above the upper end of the file range, thereby generating a perceived
gain relative to the file range reference point. The argument is that this
gain makes owner/managers more willing to accept underpricing.
2.4 Subrahmanyam (2007): Additional Perspectives
Subrahmanyam (2007) begins by identifying what he calls the central
paradigms of modern finance, “all derived from investor rationality.”
38 Survey of Surveys
The paradigms are: (i) portfolio allocation based on expected return
and risk, (ii) risk-based asset pricing models such as the CAPM and
other similar frameworks, (iii) the Miller–Modigliani theorem and its
augmentation by the theory of agency, and (iv) the pricing of contingent
claims. His survey then focuses on how behavioral ideas have impacted
these paradigms. To minimize discussion which overlaps the preceding
three surveys, attention will focus on works that appeared subsequent
to the earlier surveys, or ideas not covered.
2.4.1 Cross-Section of Stock Returns
The debate about whether risk or psychology explains the cross-section
continues. Part of this debate focuses on the Fama and French (1993)
three-factor model and a critique by Daniel and Titman (1997) centered
on whether it is firm characteristics or factor loadings that explain
returns. More recently, the debate has become more nuanced, focus-
ing on whether factors proxy for both fundamentally based risk and
sentiment. In this respect, Daniel and Titman (2006) suggest that the
book-to-market effect is driven by overreaction to that part of the book-
to-market ratio not related to accounting fundamentals. Moreover, they
argue that the component of this ratio which relates to fundamentals
does not actually forecast returns, bringing into question the “distress-
risk” explanation.
Behavioral analyses of the cross-section focus on momentum in the
short-run followed by reversals in the long-run. Hong et al. (2007)
explain momentum, asset bubbles, and other phenomena using a model
in which investors use overly simplified models to evaluate stocks. This
type of model provides insight into the implications of investors relying
on heuristics rather than full optimizing models. Kausar and Taffler
(2006) provide evidence which supports the approach advanced by
Daniel et al. (1998). They report that stocks initially exhibit momen-
tum in response to an announcement that the firm is in distress (a
going-concern audit report), but later exhibit reversals. Doukas and
Petmezas (2005) find support for the self-attribution hypothesis, one
of the elements of the Daniel et al.’s approach. However, they find
the phenomenon in the market for corporate control, where managers
2.4 Subrahmanyam (2007): Additional Perspectives 39
appear to earn successfully smaller returns in each successive acquisi-
tion because they become increasingly overconfident with each success-
ful acquisition. At the same time, there is mixed evidence about the
regime switching hypotheses advanced by Barberis et al. (1998). Chan
et al. (2003) find no support suggesting extrapolation after a sequence
of news events within returns data. However, Frieder (2004) does find
supporting evidence in order flow data around earnings announcements.
Doukas and McKnight (2005) find that the results reported by Hong
et al. (2000) hold in Europe, thereby providing out-of-sample confir-
mation for the Hong and Stein (1999) hypotheses.
Subrahmanyam points out that it is difficult for the neoclassical
approach to justify the magnitude of Sharpe ratios observed in prac-
tice. Based on the Euler equation from a neoclassical representative
investor model, a high Sharpe ratio implies highly variable marginal
utility across states, for which there is little if any empirical support.
(Hansen and Jagannathan, 1991). For theory to support a positive pre-
mium in respect to size and book-to-market equity, the returns of small
and high book-to-market stocks would need to covary negatively with
marginal utility. Given that marginal utility is inversely related to con-
sumption growth, this implies that the returns associated with small
cap firms and stocks with high book-to-market equity would need to
be particularly high in good times, and vice versa. However, Lakon-
ishok et al. (1994) find no evidence that this is the case. In a related
vein, Haugen and Baker (1996) find that the strongest determinants
of expected returns are past returns, trading volume and accounting
ratios such as return on equity and price/earnings. Notably, they find
no evidence that risk measures such as systematic or total volatility are
material for the cross-section of equity returns.
The Barberis–Thaler survey discusses belief-based models with
institutional friction. Subrahmanyam’s treatment of this issue mentions
the work of Baker and Stein (2004). Baker and Stein argue that opti-
mistic investors generate volume, producing a negative relation between
returns and past volume. Notably, the optimism is reversed in subse-
quent periods. However, because of short-selling constraints, pessimism
is not adequately reflected in stock prices.
40 Survey of Surveys
In cross-sectional regressions, stocks of small firms which have low
analyst following appear to feature greater mispricing in terms of
return predictability from book-to-market and momentum. (Zhang,
2006). There is evidence that mispricing is most severe for stocks where
institutional ownership is lowest (Nagel, 2005) who suggests that insti-
tutional ownership is a proxy for short-selling constraints, in that short-
selling is typically cheaper for institutions.
Some findings appear to lack mutual consistency. Subrahmanyam
mentions a finding by Hvidkjaer (2006) whereby on net, small investors
buy loser momentum stocks and subsequently become net sellers in
these stocks. This suggests that they may create momentum by under-
reacting to negative information. Yet, Hvidkjaer (2005) reports a nega-
tive relationship between trade imbalances of small investors and future
stock returns in the cross-section. This result is consistent with an
overreaction effect by small investors, followed by the reversal of their
sentiment, thereby generating stock return predictability.
Subrahmanyam discusses several papers concerning one of the cen-
tral claims made in the behavioral finance literature, namely that
investors who are imperfectly irrational are able to persist in the long
run. The specific papers he mentions in this regard are DeLong et al.
(1991), which is the seminal work mentioned earlier on this point, Kyle
and Wang (1997), and Hirshleifer et al. (2006).
2.4.2 Other Psychological Influences on Returns
As was mentioned in Hirshleifer’s survey, Griffin and Tversky (1992)
provide a psychological theory based on weight and strength explain-
ing why some situations feature underweighting of base rates, whereas
other situations feature overweighting of base rates. Sorescu and Sub-
rahmanyam (2006) test the argument that investors overreact to the
strength of a signal and underreact to its weight. They proxy weight by
analyst experience and strength by the number of categories spanned
by analyst revisions. Doing so, they find some support for the Griffin
and Tversky hypothesis.
Hirshleifer’s survey mentions the role of emotions, especially mood,
on returns. For example, consider the possibility that mood is positively
2.4 Subrahmanyam (2007): Additional Perspectives 41
correlated with exposure to sunlight. A study by Saunders (1993) doc-
uments that stock returns on the New York Stock Exchange tend to be
positive on sunny days but mediocre on cloudy days. Extending this
study to a number of international markets, Hirshleifer and Shumway
(2003) confirm this finding. At the same time, Goetzmann and Zhu
(2005) find little evidence that the return pattern stems from individ-
ual investors’ trading patterns.
A related approach focuses on seasonal affective disorder (SAD).
Kamstra et al. (2000) focus on returns for weekends involving the switch
from daylight savings time to standard time. They find these returns
to be very negative, and suggest that the cause is induced depression
of investors subject to SAD. In a different type of mood study, Edmans
et al. (2007) study how stock market returns in a country are impacted
by the outcomes of sporting events involving that country. They find
that returns are positive when the outcome is positive and vice versa.
Although Subrahmanyam does not mention it explicitly, mood has also
become a specific object of study in the neurofinance literature: see
Knutson and Kuhnen (2009).
In discussing the disposition effect, Subrahmanyam describes a
comprehensive study of trading activity by Grinblatt and Keloharju
(2001b). They confirm the disposition effect in Finnish data. Finnish
data is exceptionally rich, in that it is comprehensive on an account
and trade-by-trade basis. Kaustia (2004) also provides insight into the
disposition effect by focusing on IPO markets. The idea is that because
the offer price is a common purchase price, the disposition effect will be
clearly identifiable. Kaustia finds that volume is lower when the stock
price is below the offer price. He also finds that there is a sharp upsurge
in volume after the price rises above the offer price for the first time.
As to diversification, Goetzmann and Kumar (2008) find that
the portfolios held by individual investors who are young and less
wealthy are less diversified than older, more wealthy investors. In sup-
port of the familiarity hypothesis, Frieder and Subrahmanyam (2005)
present evidence that individual investors prefer stocks with high brand
recognition.
Poteshman and Serbin (2003) present evidence that option investors
do not optimally exercise options. For example, they exercise options
42 Survey of Surveys
when they should instead sell them. Both Stein (1989) and Potesh-
man (2001) report that options implied volatility appears to reflect
short-term overreaction. Evidence presented by Bakshi et al. (2000)
suggests that traders often cause movements in option prices to be
inconsistent with comparative statics based on traditional rationality
assumptions.
2.5 Critique: Weaknesses in the Behavioral Approach
Behavioral finance has both strengths and weaknesses. The prior sur-
veys do a good job of articulating its strengths, but not as a good job in
articulating its weaknesses. Indeed, only Baker et al. strive to present
an approach which is balanced by focusing on weaknesses as well as
strengths. This issue is important, in that some of the weaknesses in
the behavioral approach are foundational in nature, and need to be
addressed for the behavioralizing of finance to be successful. In this
section, I provide a brief description of what I see as key weaknesses
in the behavioral approach. Addressing these issues underlies much of
the remainder of this monograph.
2.5.1 Preferences
We do not yet have a coherent, systematic behavioral approach to pref-
erences. Discussions about preferences in the behavioral finance liter-
ature are overwhelmingly dominated by prospect theory. To be sure,
prospect theory offers important insights, and captures a wide variety of
behavioral patterns observed in experimental settings. However, there
is a difference between describing the empirical patterns and modeling
those patterns. The literature in behavioral finance has a tendency to
treat the patterns and the theory as one and the same, when in fact
they are not.
Unfortunately, for all its richness, prospect theory is incomplete
and its formal structure possesses some serious flaws. It is incomplete
in respect to framing. The shape of the functional form used to capture
probability weighting is sensitive to the choice of parameter value. In
particular the shape is at odds with the principles underlying the theory
(Ingersoll, 2008). The convex portion of its value function gives rise to
2.5 Critique: Weaknesses in the Behavioral Approach 43
boundary solutions that are highly unrealistic (Shefrin, 2008a). This
feature presents difficulties in respect to the character of equilibrium
(De Giorgi et al., 2009).
In introducing the disposition effect, Shefrin and Statman (1985)
begin with elements of prospect theory. In this regard, Statman and
I suggest that original purchase price serves as a likely reference point,
concavity of the value function in the domain of gains suggests a
tendency to realize winners, and convexity of the value function in
the domain of losses suggests a tendency to hold losers. However, we
were quick to indicate that prospect theory cannot provide a com-
plete explanation for the disposition effect. We proposed that a series
of additional psychological phenomena underlies the behavior. These
phenomena included a realization utility mental accounting compo-
nent, regret, and self-control. Our discussion of self-control indicated
that some people could learn to reduce, if not eliminate, the tendency
to exhibit the disposition effect.
My sense is that finance academics have been very slow to pick
up on the nuances in Shefrin and Statman (1985). Only recently have
there been a wave of papers addressing the inability of prospect theory
to explain the disposition effect (Hens and Vlcek, 2005; Barberis and
Xiong, 2009; Kaustia, 2009). In addition, Feng and Seasholes (2005)
present evidence that people can learn to reduce their vulnerability to
the disposition effect, which is consistent with the self-control discussion
in Shefrin and Statman (1985).
Applications of prospect theory in the behavioral finance literature
have also been piecemeal. For example, Barberis et al. (2001a) empha-
size hedonic editing, but use a piecewise linear value function that does
not conform to the specification in Tversky and Kahneman (1992).
They also use probabilities rather than weights, as called for by Tver-
sky and Kahneman (1992). In contrast, Barberis and Huang (2008)
uses nonlinear probability weighting.
2.5.2 Cross-Section
Most of the behavioral asset pricing models to explain cross-sectional
patterns do not actually model the cross-section. Instead they rely on
44 Survey of Surveys
models that feature, but a single risky security, and focus on overreac-
tion and underreaction.
Taken together, the contributions to this literature are something
of a behavioral hodge podge. Barberis et al. (1998) develop a model
in which a representative investor makes rational use of Bayes’ rule,
but relative to a flawed regime switching model. Daniel et al. (1998)
develop a model based on time varying overconfidence and biased
self-attribution. Hong and Stein (1999) develop a model based on
differential heuristics. Underlying all these models are a variety of inter-
esting stories. However, taken together they do not provide for a unified,
coherent, systematic behavioral approach to asset pricing.
In addition, reverse engineering has characterized the behavioral
approach to underreaction, in that behavioral models have been con-
structed to explain the observed patterns rather than predict them. An
exception is DeBondt and Thaler (1985) which formulated its predic-
tion of overreaction based on Kahneman and Tversky (1973). However,
the findings about underreaction took behaviorists by surprise.
2.5.3 Long-Run Dynamics
There is reason to believe that a key behavioral contention is in error.
This is the claim that in the presence of rational investors, irrational
investors will survive in the long-term because they take on excessive
risk. The most influential paper in this regard is DeLong et al. (1991),
see Subrahmanyam (2007). As it happens, the results in the latter paper
are inconsistent with the general literature on long-run survivability
(Blume and Easley, 2008). This is problematic. Rigor matters.
2.5.4 Contingent Claims
The contingent claims approach has provided finance with a frame-
work to unify a series of different models of asset pricing and portfolio
choice. An excellent treatment of this approach is Cochrane (2005),
who develops asset pricing theory around the concept of a stochas-
tic discount factor (SDF). Subrahmanyam (2007) mentions contingent
claims as one of the four central “paradigms” of finance. This mention
2.5 Critique: Weaknesses in the Behavioral Approach 45
occurs in his introduction; however, he does not return to the topic in
the body of his survey.
None of the other surveys accord the contingent claims approach
much emphasis. I regard this deficiency as significant. A major result
in neoclassical asset pricing is that in the absence of full arbitrage,
all securities can be priced in terms of a single stochastic discount
factor (SDF). Yet, in Barberis et al. (2001a), a risky security is priced
relative to one SDF, and the default-free security is priced relative
to another SDF. This dual structure implies the existence of arbitrage
profits. However, Barberis et al. (2001a) do not discuss the issue, either
to explain it or to provide some form of justification, behavioral, or
otherwise.
It seems to me that behavioralizing the contingent claims approach
lies at the heart of behavioralizing finance. In Section 5, I discuss the
nature of a behavioral contingent claims approach. This discussion will
address two particular topics, sentiment, and the character of a behav-
ioral representative investor.
2.5.4.1 Sentiment
Sentiment is one of the central concepts of behavioral asset pricing. Yet,
absent from much of the behavioral asset pricing literature is a coherent
definition of sentiment. Instead, sentiment seems to be model-specific.
In some models it corresponds to errors in means, in other models
errors in probabilities. A good illustration of this point can be found
in Shleifer (2000) where the formal treatment of sentiment varies from
section to section. The absence of a clear formal definition of sentiment
makes it difficult to ask generally how sentiment is manifest in asset
prices, and specifically how it impacts the cross-section.
2.5.4.2 Representative Investor
Like their neoclassical counterparts, behavioral asset pricing theorists
are prone to build representative investor models. Whereas neoclassical
asset pricing theorists build models in which a rational representative
investor sets prices, behavioral asset pricing theorists build models in
which the representative investor exhibits the behavior traits of indi-
46 Survey of Surveys
viduals, as documented in the psychology literature. However, as I dis-
cuss in Section 5, aggregation theory strongly indicates that in the
presence of heterogeneous investors, the representative investor will
reflect but not share the traits exhibited by individual investors. As
a result, economists who use standard representative investor models
are prone to “representativeness bias,” a common pattern documented
in the behavioral decision literature. This statement applies to neo-
classical economists working in the tradition of Mehra and Prescott
(1985), whose asset pricing model features a representative investor
with constant relative risk aversion. Moreover, the statement applies
equally to behavioral economists whose models feature representative
investors with behavioral beliefs and preferences identified with indi-
viduals. Examples include Barberis et al. (1998) and Barberis et al.
(2001a).
3
Behavioralizing Beliefs and Preferences
Generally speaking, sentiment reflects the influence of psychology on
beliefs and preferences. The focus of the present section is the behav-
ioralization of neoclassical beliefs and preferences.
An excellent discussion of behavioral beliefs and preferences can be
found in Fox and See (2003). I draw from some of their discussion as
well as Shefrin (2007) to describe how behaviorists have introduced
psychological elements into the neoclassical framework.
In neoclassical finance, investor preferences are assumed to obey
the axioms of expected utility theory, where the expectation is taken
with respect to probability beliefs. In this respect, the expected utility
function combines both beliefs (probability distributions) and prefer-
ences (utility function). Survey evidence indicates that the proportion
of the population whose behavior conforms to expected utility the-
ory is small. See Bruhin et al. (2009a) who suggest a figure of about
20 percent, which provides an indication of the relative importance of
developing a coherent behavioral approach.
Expected utility theory models risky prospects as random variables
or more generally as stochastic processes. Here each realization ˜ x of a
random process x generates a utility u(˜ x). Expected utility E[u] is the
47
48 Behavioralizing Beliefs and Preferences
expectation of utility u with respect to the probabilities governing x.
A natural question to ask is whether there are behavioral counterparts
to the neoclassical notions of utility function, probability distribution,
and process of maximization? The purpose of this section is to address
this question, with a view toward identifying structure.
3.1 Behavioral Beliefs: Heuristics and Biases
In neoclassical finance, investors’ beliefs are typically assumed to be
free from bias, meaning that investors’ beliefs correspond to the actual
probabilities governing x. However, the heuristics and biases literature
suggest the presence of systematic biases, thereby leading to distorted
beliefs. Some types of phenomena, such as optimism, overconfidence,
extrapolation bias, and gambler’s fallacy, can all be modeled as biases
to the probabilities representing investors’ beliefs. However, other phe-
nomena such as ambiguity aversion are inconsistent with beliefs repre-
sented as probability densities. I discuss both types of situations below.
The concept “change of measure” is well established in asset pricing
theory, where it is commonly used to convert physical (or true) proba-
bilities into risk neutral probabilities. In this section, I discuss how the
change of measure approach can also capture the impact of particular
psychological biases, relative to neoclassical beliefs.
To illustrate the concept of change of measure, consider a random
variable x which is normally distributed N(0, 1). The probability den-
sity function (pdf) of x is
1


e

1
2
x
2
. Consider multiplying the pdf of x
by a change of measure function ξ(x) = e
xµ−
1
2
µ
2
. Doing so, and group-
ing terms in the exponent, leads to the function
1


e

1
2
[x−µ]
2
. This, of
course, is the pdf for a random variable which is normally distributed
N(µ, 1). Hence, the operation of multiplying the pdf of x by the change
of measure ξ has led to a shift in mean.
Some biases can be modeled as simple shifts in mean. For example,
let x be the abnormal return to a security which is normally distributed
N(0, 1). Interpret the zero mean as the security being fairly priced in
the market. An investor who is unrealistically optimistic might believe
that the mean of x is µ > 0. In this case, optimism bias can be modeled
as a right shift in the mean of x, which is accomplished through a change
3.1 Behavioral Beliefs: Heuristics and Biases 49
of measure. On the other hand pessimism bias corresponds to a shift
in mean where µ < 0.
By definition, the change of measure function ξ is the ratio of the
probability density of the mean-shifted random variable to the proba-
bility density of the original random variable. Consider the log-change
of measure ln(ξ). The log-change of measure, evaluated at x, effectively
measures the percentage difference between the mean-shifted probabil-
ity and original probability associated with x. In effect, the log-change
of measure is a metric for the probability bias associated with each
potential realization of x.
Notice that in the mean-shift example above, ln(ξ) is just a linear
function of x, µx −
1
2
µ
2
. Consider the case of optimism bias, so that
µ > 0. For sufficiently low values of x, the log-change of measure will be
negative, meaning that probabilities assigned to very negative returns
will be too low. For sufficiently high values of x, the log-change of
measure will be positive, meaning the probabilities assigned to very
positive returns will be too high (Figure 3.1).
-6
-4
-2
0
2
4
6
-5 -4 -3 -2 -1 0 1 2 3 4 5
x
Fig. 3.1 Illustration of a log-change of measure corresponding to optimism bias.
50 Behavioralizing Beliefs and Preferences
If x is N(µ, σ) then its pdf has the form
1

2πσ
2
exp(−
(x−µ)
2

2
). Con-
sider two normally distributed random variables x
1
and x
2
. The ratio
of the pdf of x
2
to the pdf of x
1
provides the change of measure ξ
1,2
which transforms x
1
into x
2
. If x
1
is the true random variable, and x
2
represents an investor’s beliefs about this variable, then the log-change
of measure ln(ξ
1,2
) is a metric for the investor’s probability bias func-
tion. If σ
1
= σ
2
, then the log-change of measure will not be linear in
the values of x. For example, suppose that µ
1
= µ
2
and σ
1
> σ
2
. When
it comes to tail events, the investor will associate probabilities to these
events which are too low. Hence, the log-change of measure will be
negative in both the upper tail and the lower tail, but positive in the
mid-range. In particular the log-change of measure will have the shape
of an inverted-U.
Overconfidence is typically understood to imply that investors
underestimate risk. For example, if the true return process is given by
x
1
in the preceding paragraph, and an investor believes that the return
process is given by x
2
, then overconfidence is represented by the condi-
tion σ
2
< σ
1
. Hence, overconfidence is associated with a log-change of
measure whose shape is an inverted-U (Figure 3.2).
-50
-40
-30
-20
-10
0
10
-5 -4 -3 -2 -1 0 1 2 3 4 5
Fig. 3.2 Illustration of a log-change of measure corresponding to overconfidence bias.
3.2 Behavioral Preferences 51
Extrapolation bias (also known as “hot hand fallacy”) and gambler’s
fallacy both stem from judgmental heuristics. Notably, both can give
rise to time varying optimism and pessimism. These biases are dynamic,
referring as they do to biases in conditional probabilities related to
a stochastic process. If the recent portion of the realized trajectory
features a run of predominantly favorable outcomes, then hot hand
fallacy gives rise to excessive optimism, while gambler’s fallacy gives
rise to excessive pessimism.
Typically, optimism is associated with a shift in mean, whereas over-
confidence is associated with a shift in variance. However, hot hand
fallacy and gambler’s fallacy can also apply to beliefs about variance.
Hot hand fallacy predicts that after a recent period of high volatility,
subsequent volatility will continue to be high, whereas gambler’s fallacy
predicts a decline in subsequent volatility.
As was mentioned above, some beliefs are not representable as prob-
abilities. Aversion to ambiguity can be represented in several ways. One
way is through the use of Choquet integrals, where decision weights
take the place of probabilities and do not integrate to unity over the
entire population of possible outcomes (Fox and See, 2003). Notably,
the ratio of a Choquet density to a probability density is also a change
of measure, transforming a probability density to a Choquet density.
Therefore, the change of measure approach applies here as well.
Over time, economists are making increasing use of insights from
the psychology literature. An example is support theory, in which prob-
abilities are attached not to events per se, but to descriptions of events
(Tversky and Koehler, 1994; Diecidue and La-Ornual, 2009).
3.2 Behavioral Preferences
In expected utility theory, preferences are completely captured by
the utility function. In contrast, behavioral preferences are embodied
within both the counterpart to a utility function and in decision weights
associated with probabilities. In this section I discuss two psychologi-
cally based approaches to risk preferences. The first is the cumulative
version of prospect theory (CPT), and the second is SP/A theory.
Both theories focus on the cumulative (or decumulative)
distribution function, and both involve weighting functions whose shape
52 Behavioralizing Beliefs and Preferences
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5 0.55 0.6 0.65 0.7 0.75 0.8 0.85 0.9 0.95 1
Probability
Fig. 3.3 Illustration of a weighting function for decumulative probabilities with an inverse-S
shape.
resembles an inverse-S (Figure 3.3). In CPT, the inverse-S shape reflects
the principle of diminishing sensitivity, known as “psychophysics.” In
SP/A theory, the inverse-S shape reflects the impact of emotions, most
notably fear and hope. Notably, the inverse-S shaped weighting function
gives rise to a U-shaped log-change of measure in which probabilities
of extreme events are overweighted and probabilities of intermediate
events are underweighted.
To introduce some notation, denote the set of possible outcomes by
a finite set of real numbers X = {x
1
, . . . x
n
}, ordered from the lowest
outcome x
1
to the highest x
n
. Formally, a risky alternative or prospect
is a random variable that takes on values in X. One way to describe
the probabilities attached to a random variable is to use a decumulative
distribution function D, where D(x) is the probability that the outcome
payoff is at least x. The same probabilistic information is also conveyed
by the cumulative distribution function, which measures the probability
that the outcome payoff is no greater than x, or the probability density
3.2 Behavioral Preferences 53
function which measures the probability that the outcome payoff is
exactly x.
3.2.1 Prospect Theory
Prospect theory is a theory of choice developed by psychologists Daniel
Kahneman and Amos Tversky (Kahneman and Tversky, 1979; Tversky
and Kahneman, 1992).
1
Prospect theory has four distinctive features.
First, the carriers of utility are changes, meaning gains and losses rel-
ative to a reference point, not final position.
Second, the utility function (known as a value function in prospect
theory) is concave in gains and concave in losses, typically with a point
of nondifferentiability at the origin so that the function is more steeply
sloped for losses (to the left of the origin) than for gains (to the right
of the origin) (Figure 3.4). Hence, the utility function is S-shaped with
a kink at the origin. Tversky and Kahneman (1992) suggest using
-20
-15
-10
-5
0
5
10
-
1
0
-
8
.
7
5
-
7
.
5
-
6
.
2
5
-
5
-
3
.
7
5
-
2
.
5
-
1
.
2
5 0
1
.
2
5
2
.
5
3
.
7
5 5
6
.
2
5
7
.
5
8
.
7
5
1
0
Gain/loss
Fig. 3.4 Illustration of an S-shaped value function.
1
See Wang et al. (2009) for a survey-based analysis of cross-country differences in responses
to prospect theoretic questions.
54 Behavioralizing Beliefs and Preferences
a utility function u(x) with the form x
α
in the domain of gains (x ≥ 0)
and −λ(−x)
β
in the domain of losses (x ≤ 0).
2
Third, probabilities are weighted (or distorted) when prospects are
evaluated. In the original 1979 version of prospect theory (OPT), the
weighting function π has probability density p as its argument. The
π-function in OPT is convex, with π(p) > p for small positive values of
p and π(p) < p for high values of p less than 1. Tversky and Kahneman
(1992) proposed a cumulative version of prospect theory (CPT) that
uses rank-dependent utility. Unlike OPT, where probability weights
depend only on probability density, the weights in CPT depend on
outcome rank.
Tversky and Kahneman (1992) use two weighting functions, one
function for gains and one function for losses. Both functions take
decumulative probabilities as their arguments, where the decumula-
tive distribution pertains to the absolute value of the gain or loss,
respectively. The weighting function is similar to the h-transform used
by Lopes. It features an inverse S-shape, which Tversky–Kahneman
generate using the ratio of a power function to a H¨ older average; that
is, p
γ
/(p
γ
+ (1 − p)
γ
)
1/γ
. As a result, in CPT it is the probabilities of
extreme outcomes that are overweighted (very large losses and very
large gains). As I mentioned in the previous section, Ingersoll (2008)
points out that the shape of the weighting function only resembles an
inverse-S for part of its range.
Both the S-shape of the utility function and the inverse S-shape of
the weighting functions reflect psychophysics, meaning the diminished
sensitivity to successive changes. For the utility function, the changes
pertain to differences relative to the reference point. For the weighting
function, the changes pertain to differences relative to the endpoints 0
and 1.
Fourth, decision makers engage in editing or framing before formally
evaluating risky prospects. There are several types of editing issues.
Perhaps, the simplest editing issue is the choice of reference point.
2
Strictly speaking, the first derivatives of these functions at zero approach infinity, and
therefore do not display a kink. However, a suitably small translation of both functions
leads the first derivatives to be finite at zero, with the kink in question.
3.2 Behavioral Preferences 55
Kahneman and Tversky illustrate this issue by describing a medical
task in which the data can be presented, or framed, in one of two ways.
The first way is in terms of lives saved, while the second way is in
terms of lives lost. The “lives saved” frame implicitly sets the reference
point at 100 percent fatalities. The “lives lost” frame implicitly sets the
reference point at 0 percent fatalities. Although the data underlying the
two frames are identical, physicians tend to act as if they are more risk
averse when presented with the data framed in terms of “lives saved”
than when the data are framed in terms of “lives lost.” This choice
pattern is consistent with the S-shaped utility function.
A more complex framing issues is the segmentation of a compli-
cated decision task into a series of subtasks. The structure of each
subtask is called a “mental account” and the segmentation process is
known as “narrow framing.” Because narrow framing tends to overlook
interdependencies between mental accounting structures, the segmen-
tation process is often suboptimal. Tversky and Kahneman present
examples in which narrow framing leads to the selection of stochasti-
cally dominated choices. Prospect theory is a descriptive framework,
not a normative framework. People who choose stochastically domi-
nated alternatives do so because they do not always grasp the complete
structure of the decision tasks they confront. The complete structure is
typically opaque, not transparent, and people lack the ability to frame
complex decision tasks transparently.
3.2.2 SP/A Theory
Lopes (1987) and Lopes and Oden (1999) developed a psychologically
based approach known as SP/A theory to explain choice among risky
alternatives. Lopes titled her 1987 article “The Psychology of Risk:
Between Hope and Fear” to capture the idea that the emotions of
hope and fear play key roles in choice among risky alternatives. In this
respect, the letters that make up SP/A refer to specific concepts which
measure or impact the degree of fear or hope experienced by a decision
maker. Notably, S stands for “security,” P stands for “potential,” and
A stands for “aspiration.”
56 Behavioralizing Beliefs and Preferences
In a typical decision task, a decision maker chooses a best alter-
native from a menu {D} of risky alternatives. The role of the theory
is to describe the criteria leading to the choice. In SP/A theory, risky
alternatives are evaluated using an objective function whose arguments
reflect security S, potential P, and aspiration A. See Diecidue and van
de Ven (2008) and Bruhin et al. (2009b) for some recent treatments
of the role of aspiration in economic choice. Payne (2005) describes
an experiment whose results suggest that because of aspiration, SP/A
theory better predicts choice than prospect theory.
Roughly speaking, increased fear stems from reduced security, and
reduced security corresponds to an increase in the probability attached
to the occurrence of some unfavorable event. Suppose we consider
two decision makers who face an identical risk, or prospect D, but
experience different degrees of fear. Intuitively, the decision maker who
experiences more fear attaches greater importance to the probability
associated with unfavorable events than the decision maker who expe-
riences less fear.
Formally, Lopes uses rank-dependent utility to capture the effects
of fear. In rank-dependent utility, a decision maker who faced a risky
prospect D acts as if the decumulative density function is the trans-
form h(D) rather than D itself. In SP/A theory, fear operates on the
probabilities associated with unfavorable events. Because D is a decu-
mulative distribution function, the probability attached to the least
favorable event x
1
is given by Prob{x
1
} = D(x
1
) − D(x
2
).
Under the transform h(D), the probability attached to the least
favorable event x
1
is given by Prob{x
1
} = h(D(x
1
)) − h(D(x
2
)). In
Lopes’ framework, a decision maker who is fearful about exposure to
event x
1
overweights the probability attached to x
1
. That is, the fearful
decision maker employs a transform h which satisfies:
h(D(x
1
)) − h(D(x
2
)) > D(x
1
) − D(x
2
). (3.1)
Because x
1
is the least favorable outcome, the probability that the
actual outcome turns out to be x
1
or higher is 1. In other words,
D(x
1
) = 1. What Equation (3.1) effectively states is that increased fear
leads to an increase in the slope of the h-function in the neighborhood
of 1.
3.2 Behavioral Preferences 57
Rank-dependent utility also captures the manner in which the deci-
sion maker experiences hope. Hope is associated with upside poten-
tial, or potential for short. Hope induces a decision maker to attach
greater weight to the most favorable events. Because D is a decumula-
tive distribution function, the probability attached to the most favor-
able event x
n
is given by Prob{x
n
} = D(x
n
). Therefore, the emotion
of hope leads a decision maker to employ a transform h which sat-
isfies h(D(x
n
)) > D(x
n
). For the second most favorable outcome, the
corresponding inequality would read:
h(D(x
n
)) − h(D(x
n−1
)) > D(x
n
) − D(x
n−1
). (3.2)
This inequality indicates that increased hope leads to an increase in
the slope of the h-function in the neighborhood of 0.
In Lopes’ framework, a person who neither experiences fear nor
hope is associated with an h-function that is the identify function:
h(D) = D. A decision maker who experiences only fear, but not hope,
is associated with an h-function that is strictly convex in D: it is steep
in the neighborhood of 1 and flat in the neighborhood of 0. Formally,
Lopes uses a power function h
S
(D) = D
q
, q > 1 for this case. A deci-
sion maker who experiences only hope is associated with an h-function
that is strictly concave in D. Formally, Lopes uses a power function
h
P
(D) = 1 − (1 − D)
p
, p > 1 for this case. A person who experiences
both fear and hope is associated with an h-function that has an inverse-
S shape. It is concave in the neighborhood of the origin and convex in
the neighborhood of 1. Formally, Lopes uses a convex combination of
the power functions h
S
and h
P
to capture this case (Figure 3.5).
In SP/A theory, the degree to which fear and hope are experienced
depends on the degree to which risky prospects offer security S and
potential P. To capture the impact of both security and potential,
Lopes uses an expected utility function with probabilities derived from
the h-transform. She calls the function SP for security-potential, and
it has the form:
SP =
n
¸
i=1
(h(D
i
) − h(D
i+1
))u(x
i
). (3.3)
58 Behavioralizing Beliefs and Preferences
0
0.2
0.4
0.6
0.8
1
1.2
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
D
h1(D)
h2(D)
h(D)
Fig. 3.5 Illustration of three decumulative transformation functions in Lopes’ SP/A frame-
work.
In Equation (4.1), u is a utility function whose argument is outcome x.
Although Lopes uses the assumption u(x) = x in most of her analysis,
Lopes and Oden (1999) comment that in practice, u might display a
bit of concavity.
The “A” in SP/A denotes aspiration. Aspiration pertains to a target
value α (or range) to which the decision maker aspires. Aspiration
points reflect different types of goals. For example, a decision maker
might wish to generate an outcome that would allow the purchase of
a particular good or service. Alternatively, the aspiration point might
reflect a status quo position that corresponds to the notion of no gain or
loss. In Lopes’ framework, aspiration-risk is measured as the probability
A = Prob{x ≥ α} that the random outcome x meets or exceeds the
aspiration level α.
In SP/A theory, the decision maker maximizes an objective function
V (SP, A) in deciding which alternative D to choose from the menu of
available prospects. V is strictly monotone increasing in both of its
arguments. Therefore, there are situations in which a decision maker is
willing to trade off some SP in exchange for a higher value of A.
3.3 Regret 59
3.3 Regret
In the early development of prospect theory, Kahneman and Tversky
focused on the role of regret. Regret is the psychological pain associated
with recognizing after the fact that taking a different decision would
have produced a better outcome. Kahneman and Tversky eventually
built prospect theory using the S-shaped value function, but in Kah-
neman and Tversky (1982) continued to emphasize the importance of
regret. They pointed out that regret will be magnified by the ease with
which a person can easily imagine taking a different decision. In conse-
quence, regret avoidance discourages deviations from behavior viewed
as conventional, a version of status quo bias.
The volume edited by Loomes and Sugden (1982) contains a for-
mal treatment of decision making when preferences are determined
by regret. Shefrin and Statman (1984, 1985) modify the concept of
regret to reflect pride in the domain of gains. Gollier and Salani´e (2006)
develop a regret-based model to analyze the character of the risk pre-
mium in an asset pricing model.
3.4 Self-Control
Self-control refers to situations when a person is conflicted, and thinks
he or she should take one decision, but emotionally feels like taking a
different decision. Studies of self-control in financial economics tend to
emphasize the difficulty in delaying gratification. However, self-control
applies more broadly, and in particular applies when the emotion of
regret prevents a person from taking a decision that they “think” is
appropriate.
Shefrin and Thaler (1988) extend the neoclassical life cycle frame-
work to incorporate psychological elements associated with self-control
and editing. The behavioral life cycle framework features a two-self
framework focused on the interaction between a person’s rational self
(a “planner” corresponding to activity in the prefrontal cortex of the
brain) and an impulsive self (a “doer” corresponding to activity in
the limbic system of the brain). In addition to traditional neoclassical
variables such as consumption, the behavioral life cycle framework adds
60 Behavioralizing Beliefs and Preferences
psychological variables to represent “temptation,” “willpower,” “rules,”
and associated psychological costs for exhibiting self-control. In partic-
ular, rules are represented by constraints which resemble corner solu-
tions. People may choose to violate a rule they themselves establish,
but doing so entails a psychological cost.
When the work of Shefrin and Thaler (1988) was published, the
field of neuroeconomics was nonexistent. However, neuroeconomics is
now flourishing, and providing neurological support for psychologically
based models. For example, Hare et al. (2009) identify a specific region
in the prefrontal cortex associated with self-control, the DLPFC. Knut-
son et al. (2008) identify hope with a region of the brain called the
nucleus accumbens. Knoch et al. (2009) study the impact of breakdowns
in prefrontal cortex activity on reputation. With advances in neuroe-
conomics, economists are beginning to develop sophisticated models of
brain activity (Brocas and Carrillo, 2008).
4
Behavioralizing Portfolio Selection Theory
In their survey article, Barberis and Thaler (2003) identify empirical
features characterizing portfolio selection by individual investors. These
include the following:
— lack of diversification associated with the 1/n-heuristic,
— home bias associated with familiarity and ambiguity aversion,
— excessive trading associated with overconfidence,
— the disposition effect associated with prospect theory, regret,
and self-control
— the attention hypothesis associated with availability bias, and
— dividends associated with self-control.
In this section, I focus attention on the following series of additional
issues:
— desire for positively skewed returns,
— role of dividends as a self-control device and managing risk, and
— status quo bias.
The sheer number of effects, most at odds with the neoclassi-
cal mean-variance approach, indicates that the behavioralization of
61
62 Behavioralizing Portfolio Selection Theory
portfolio theory will be very rich. At present, it is also a work in
progress.
1
The purpose of this section is to describe the implications
of some of the behavioral preference frameworks for portfolio selection,
and relate these to empirical evidence.
A broad issue that arises in this section concerns the role of opti-
mization. A first step in the behavioralization of portfolio theory is to
replace neoclassical preferences with behavioral preferences and under-
take a full maximization. Doing so yields some interesting insights.
However, there is ample evidence that individual investors are imper-
fect maximizers. Therefore, the issues described in this section begin
with full maximization and then move on to quasi-maximization.
4.1 Preference for Positively Skewed Returns:
Full Maximization
4.1.1 Empirical Issues
In contrast to neoclassical portfolio theory, which emphasizes mean-
variance efficiency, behavioral portfolio theory emphasizes the taste for
positive skewness. There is strong evidence that people have a taste
for positively skewed returns. In 2009, LaFleur’s magazine reported
that the average household in Massachusetts spends $1,760 annually on
lotteries, roughly one-third more than the average annual contribution
Americans make to their 401(k) plans. The desire for positively skewed
returns is related to thrill-seeking. Grinblatt and Keloharju (2009) use
a clever database to study thrill-seeking behavior by investors.
A good example of a security with a lottery feature is the British
Premium Bond. The Premium Bond was launched by the U.K. gov-
ernment in 1956, and is issued by National Savings and Investments,
a government department. Any British citizen can purchase Premium
Bonds, with the minimum investment being £100, or with a monthly
standing order, £50. The maximum amount a person can invest is
£30,000. As with a regular bond, the Premium Bond returns princi-
pal on a nominal basis. However, in lieu of monthly interest, investors’
1
Investor preferences also reflect concerns other than returns, such as social responsibility
(Statman, 2008).
4.1 Preference for Positively Skewed Returns: Full Maximization 63
returns are determined by a lottery, and moreover are tax-free. For
each single pound invested, there is one chance to win. Lobe and H¨ olzl
(2008) document the degree of popularity of this security. The fraction
of the U.K. population which invests in Premium Bonds grew from
approximately 12 percent in 1957 to 40 percent in 2005. During the
1980s, approximately 45 percent of the population invested in Pre-
mium Bonds. Lobe and H¨ olzl (2008) points out that the tax advantage
of Premium Bonds suggests that there are neoclassical reasons to jus-
tify why higher income investors would choose to hold Premium Bonds.
However, for lower income investors, the attractiveness of these securi-
ties is more likely to lie with the manner in which the bonds package
safety of principal with the lottery feature of coupon payments.
Statman (2002) and Kumar (2009) investigate the extent to which
individual investors hold stocks with lottery-like characteristics. Kumar
notes that lotteries offer a low probability of a high payoff (positive
skewness), feature low prices per ticket, and have negative expected
returns. Similarly, lottery-like stocks feature positively skewed return
distributions, low prices per share, and possibly negative expected
returns.
After classifying stocks into lottery stocks and non-lottery stocks,
Kumar compares the portfolio weights individual investors assign to lot-
tery stocks to both a random assignment as well as the portfolio weights
institutional investors assign to lottery stocks. A random assignment
would allocate 0.74 percent of the portfolio to lottery stocks. On aver-
age, individual investors allocate 8.3 percent of their portfolios to lot-
tery stocks. Kumar also reports that even the most wealthy individual
investors allocate 7.7 percent of their portfolios to lottery stocks. In
contrast, institutional investors allocate 0.28 percent of their portfolios
to lottery stocks.
At the opposite end of the spectrum lie stocks that are the mirror
images of lottery stocks. These stocks feature low volatility, are less
positively skewed, have higher prices per share, and higher expected
returns. The random assignment would allocate 54 percent of a port-
folio to stocks that are diametrically opposite to lottery stocks. The
weight assigned by individual investors is 33 percent, and the weight
assigned by institutional investors is 59 percent.
64 Behavioralizing Portfolio Selection Theory
Polkovnichenko (2002) reports that the wealthiest group of investors
hold almost the same amount in direct equity as indirect equity. Simi-
larly, Kumar reports that investors who hold larger mutual fund port-
folios invest more in lottery stocks. Kumar also finds that the highest
turnover is among buyers of lottery stocks.
In line with risk-seeking behavior in the domain of losses, Kumar
finds that when economic conditions deteriorate, people increase their
allocations to lottery stocks. In particular, he finds that on a risk-
adjusted basis, those who invest in lottery stocks earn 5.9 percent a
year less than those who do not. Those who invest heavily in lottery
stocks earn 8.9 percent less than those who invest moderately in lottery
stocks. The 8.9 percent actually corresponds to 13.1 percent on a risk-
adjusted basis.
4.1.2 Theoretical Issues
Consider the traditional Markowitz portfolio selection problem when
security returns are normally distributed. Formally, let w is a vector
of portfolio weights, Σ be a return covariance matrix, R be a vector of
expected returns, and τ ∈ [0, ∞) measure the degree of risk tolerance.
Then the mean-variance efficient frontier is the family of solutions to the
w-minimization of w
T
Σw − τR
T
w, for varying τ. Associated with each
mean-variance portfolio is a normally distributed return distribution.
A key feature of the mean-variance efficient portfolios is the two-
fund separation property. In this case, any mean-variance efficient port-
folio is a convex combination of a unique risky portfolio and the risk-free
portfolio.
4.1.2.1 Full Maximization with Behavioral Preferences
The starting point for behavioralizing portfolio theory involves the
replacement of a mean-variance objective function with a behavioral
objective function in the optimization procedure. The two main objec-
tive functions used in the literature are the prospect theory weighting
function
¸
W
i
v(x
i
) and the SP/A function V (SP, A).
Consider how different an optimized behavioral portfolio return dis-
tribution is from a mean-variance return distribution. To describe the
4.1 Preference for Positively Skewed Returns: Full Maximization 65
Fig. 4.1 Illustration of the payoff pattern to the risk-free security and a combination of the
risk-free security and unique mean-variance risky security. The x-axis is the gross return to
the unique risky mean-variance security.
contrast, consider a series of graphs with the return to the unique risky
mean-variance portfolio on the x-axis. Figure 4.1 illustrates the returns
to two portfolios relative to the unique risky portfolio. One of the two
portfolios is risk-free, returning the risk-free rate, and the other com-
bines the risk-free security and the unique risky security.
Figure 4.2 illustrates the return to a behavioral portfolio associ-
ated with prospect theory preferences when the weighting function is
the identity function, meaning the weights are themselves probabili-
ties. Notice that the returns to loss states are very low, and in fact
are associated with a lower bound on achievable return. This bound-
ary property stems from the fact that the value function is convex in
the domain of losses. The kink at the origin of the value function, in
combination with subcertainty, can produce a flat region with neither
gain nor loss. In the domain of gains, the behavioral return is similar
to a neoclassical return, as the value function is concave in gains. See
also Barberis and Huang (2008).
Figure 4.3 illustrates the possible impact of the weighting func-
tion in the domain of gains. With overweighting of probabilities at the
extremes, the behavioral return distribution corresponds to an inverse-S
66 Behavioralizing Portfolio Selection Theory
Fig. 4.2 Illustration of the payoff pattern to a portfolio associated with optimization of
a prospect theory criterion function with probabilities as weights. The x-axis is the gross
return to the unique risky mean-variance security.
Fig. 4.3 Illustration of the payoff pattern to a portfolio associated with optimization of a
prospect theory criterion function with an inverse-S-shaped weighting function. The x-axis
is the gross return to the unique risky mean-variance security.
in the domain of gains. This occurs because the probabilities associated
with intermediate gains tend to be underweighted, whereas the proba-
bilities associated with extreme gains tend to be overweighted, thereby
giving rise to positive skewness, in that there is a small probability of
an especially large gain.
4.1 Preference for Positively Skewed Returns: Full Maximization 67
Notably, the weighting function is rank dependent, thereby intro-
ducing kinks into the indifference map. For losses, overweighting of
extreme loss probabilities can introduce a convex feature that operates
in the opposite direction from the shape of the value function in the
domain of losses. In the domain of gains, the overweighting of extreme
gain probabilities introduces a non-convex feature that operates in the
opposite direction from the shape of the value function in the domain of
gains. The effect of probability weighting can produce a non-boundary
flat return in the domain of losses, and accentuated risk seeking in the
domain of gains.
Prospect theory portfolios tend to feature either boundary solutions
or unbounded solutions, a feature noted in the early drafts of Shefrin
and Statman (2000), and explored more fully in Shefrin (2008a). See
also De Giorgi et al. (2009) and He and Zhou (2009). He and Zhou call
this an “ill posed problem.”
The early versions of Shefrin and Statman (2000) used prospect
theory as the basis for a behavioral portfolio theory. In commenting
on these drafts, Daniel Kahneman suggested to Statman and me that
Lopes’ SP/A theory might provide a superior approach to CPT. The
published version of Shefrin and Statman (2000) develops an SP/A-
based behavioral portfolio selection theory, emphasizing the roles of
fear, hope, and aspiration. See Diecidue and van de Ven (2008) and
De Giorgi (2009) who focus on the impact of aspiration on portfolio
returns.
Figure 4.4 illustrates the contrast between a typical SP/A-based
portfolio return and the risky mean-variance return. Notice that the
SP/A return function has three segments. Begin with the middle seg-
ment. It is flat and the return corresponds to the achievement of the
aspiration level. To the left is a region where the investor fails to
achieve the aspiration level. With concave utility, and in the absence of
probability weighting, this region will typically be strictly concave and
increasing. Probability weighting tends to flatten this region, perhaps
to the point of rendering it completely flat as depicted in the figure.
With linear utility, this region will lie on the boundary. At the right is a
region where the investor achieves a return above aspiration. Depend-
ing on the utility function and specific parameters, one or more of these
68 Behavioralizing Portfolio Selection Theory
Fig. 4.4 Illustration of the payoff pattern to a portfolio associated with optimization of
an SP/A criterion function with an inverse-S-shaped weighting function. The x-axis is the
gross return to the unique risky mean-variance security.
regions might disappear. Because of probability weighting, the return
pattern typically has the shape of an inverse-S. With concave utility,
and absent probability weighting, the shape is typically concave and
increasing. With linear utility, the region is very small, but features a
spike at the highest return.
Recall the discussion in Section 3 concerning the use of change
of measure to model behavioral beliefs. In this regard, the various
behavioral return patterns below can be expressed as expected util-
ity maximizing solutions, relative to a change of measure. To see why,
let equilibrium state prices be given by ν. Begin with the equilibrium
consumption profile c
S
chosen by an investor with SP/A preferences
based on utility function u. Now ask whether we can identify a prob-
ability density function P

such that an expected utility maximizing
investor with utility function u and beliefs P

would have selected c
S
?
Let i and k be two states. For equilibrium prices ν, choose P

i
and
P

k
such that:
P

i
u
i
/P

k
u
k
= ν
i

k
, (4.1)
where u
i
and u
k
are marginal utilities of consumption evaluated at c
S
.
Then, normalize so that the components P

sum to unity.
4.1 Preference for Positively Skewed Returns: Full Maximization 69
Fig. 4.5 Illustration of the change of measure associated with the payoff pattern to a port-
folio associated with optimization of an SP/A criterion function with an inverse-S-shaped
weighting function. The x-axis is the gross return to the unique risky mean-variance security.
If Π denotes the objective probability density function, then the
function L = ln(P

/Π) is a log-change of measure that captures the
impact of the SP/A investor’s behavioral preferences and behavioral
beliefs. For example, Figure 4.5 adds a stylized change of measure to
Figure 4.4. Notice that the pattern can be constructed by beginning
with a U-like shape, corresponding to the change of measure associ-
ated with the inverse-S shaped weighting function in SP/A theory.
The U-shape reflects the fact that the probabilities of extreme events
are overweighted and intermediate events are underweighted. In Fig-
ure 4.5, the right side of the U is weaker than the left side. Notice
that the middle section features a jump at the discontinuity marking
the boundary of the middle region, typically followed by a decline to
reflect the fact that the behavioral portfolio return is flat at the aspi-
ration point, whereas expected utility maximization typically calls for
an increasing return profile.
Shefrin and Statman (2000) point out that SP/A-based portfolio
theory provides a rationale for the attractiveness of securities like the
Premium Bond. An SP/A portfolio combines safety in the form of a
70 Behavioralizing Portfolio Selection Theory
risk-free bond, a risky bond, and a risky security with the features of an
out-of-the-money call option. They also point out that the behavioral
portfolio approach explains the attractiveness of many structured prod-
ucts. Recent behavioral approaches to structured products includes
Rieger (2008) and Bernard and Boyle (2008).
4.1.2.2 Bounded Rationality
Shefrin and Statman (2000) also suggest that bounded rationality typi-
cally prevents investors from choosing portfolios using full optimization.
Instead, investors decompose decision tasks into simpler sub-task and
use quasi-optimization. For example, investors might frame the port-
folio selection task by identifying two aspirations, one for downside
protection and the other for upside potential. Each task is associated
with its own mental account and aspiration point. The downside pro-
tection account features a low aspiration level (avoiding poverty) and
a high probability of achieving this level. The upside potential account
features a high aspiration (riches), and a low or modest probability of
achieving this aspiration level. In their model, the investor establishes a
nonnegative allocation w = (w
D
, w
U
), w
D
+ w
U
= 1, which gives rise to
sub-utilities (V
D
, V
U
), respectively. An allocation w is evaluated using a
criterion function F(V
D
, V
U
), with the portfolio decision being to solve
max
w
F. Shefrin and Statman call such a portfolio a “layered pyramid.”
By its nature, mental accounting gives rise to incomplete diversifica-
tion, as investors ignore covariance relationships.
Although this section has emphasized preferences without regard
to biased beliefs, be aware that the impact of probability weighting is
similar to the impact of biased beliefs.
4.2 Quasi-Maximization
The disposition effect, attention hypothesis, favoring of dividend-
paying stocks, status quo bias, and the propensity to trade are all symp-
toms of bounded rationality. It seems to me that modeling bounded
rationality can be more difficult than modeling full rationality. In this
respect, behavioral portfolio models can be delicate. For example, Bar-
beris and Huang (2008) develop a dynamic prospect theory model to
4.2 Quasi-Maximization 71
study the disposition effect. In this regard, they use static theory to
identify the solution, and then recast the model in a dynamic context.
What they find is that prospect theory can lead investors to sell their
losers and purchase more of their winners, in apparent contradiction
of the disposition effect. This is an interesting result, which supports
the original contention in Shefrin and Statman (1985) that the dispo-
sition effect is likely driven by several psychological features, such as
mental accounting, regret, and self-control. As I mentioned in Section 2,
prospect theory value function, by itself, cannot explain the disposition
effect.
Results like those in Barberis and Xiong (2009), together with the
ill posed property of prospect theory-based portfolios, suggest that
extreme caution is necessary in developing models to capture behav-
ioral portfolio selection.
4.2.1 Empirical Issues
Section 2 covers many of the key empirical findings about individual
investors’ portfolios. In this section, I focus on issues which emphasize
mental accounting issues in a dynamic environment, such as diversifi-
cation, the disposition effect, cash dividends, and options. Below I offer
a few brief comments about diversification and the disposition effect,
and then concentrate on cash dividends and options.
Goetzmann and Kumar (2008) document the existence of skilled
investors who hold concentrated portfolios. They point out that as
investors increase the number of stocks in their portfolios, they tend to
choose stocks which co-move, thereby depriving themselves of the ben-
efits of diversification. They also report that older investors have bet-
ter diversified portfolios and trade less aggressively than their younger
counterparts. See also Dorn and Huberman (2005).
Feng and Seasholes (2005) find that susceptibility to the disposi-
tion effect declines by 72 percent with sophistication and experience,
especially in connection with riding losers. Their finding provides an
indication that some investors learn to achieve increased self-control.
Seru et al. (2009) report a similar pattern, with the disposition of
the median investor improving by 4 percent after a year of trading
72 Behavioralizing Portfolio Selection Theory
experience. In respect to regret, Lehenkari (2008) uses Finnish data to
provide intriguing evidence that investors are more prone to the dis-
position effect for stocks they purchased than for stocks they received
as gifts or as part of an inheritance. Recent experimental evidence by
Fogel and Berry (2006) strongly supports the importance of regret in
explaining the disposition effect.
4.2.1.1 Cash Dividends
The first paper in behavioral finance written by economists is She-
frin and Statman (1984). They develop a behavioral explanation for
why some investors framework find cash dividends attractive. Their
approach emphasizes the roles of self-control, mental accounting,
hedonic editing, and regret. Self-control and mental accounting sug-
gest several hypotheses linking consumption and dividends, along with
a clientele effect. The theory suggests that older, retired investors rely
on a heuristic in which they finance consumption out of an income
mental account, but are reluctant to “dip into capital” which resides
in an asset mental account. Hedonic editing and regret suggest a con-
nection between stock market movements and dividends. In particular,
investors might prefer to finance consumption out of dividends than
out of capital in order to avoid the regret of selling off stock which sub-
sequently appreciates by a large amount. About 20 years after being
proposed, most of these hypotheses have come to be tested, with the
overall results being supportive of the behavioral approach.
2
Graham and Kumar (2006) measure the preference for dividends by
portfolio dividend yield. They report that irrespective of income, older
households prefer dividend-paying stocks over non-dividend-paying
stocks. For younger households, those with low income exhibit a some-
what stronger preference for dividend-paying stocks than households
with higher income. Moreover, the preference for dividends is relatively
stable, with dividend yields of individual investors’ portfolios changing
slowly over time. These findings support the predictions of behavioral
dividend theory about the impact of age and income.
2
The material in this section is taken from Shefrin (2008b).
4.2 Quasi-Maximization 73
In terms of stock selection, Graham and Kumar report that older,
retired investors hold about 80 percent of their portfolios in dividend-
paying stocks, while younger investors hold about 65 percent of their
portfolios in dividend-paying stocks. In particular, older investors with
low incomes attach more weight to utility stocks than do younger
investors with high incomes. The reverse is true for the stocks of firms
in the “computers” and “business services” industries. This pattern is
consistent with the idea that older, low-income investors focus more on
industries with a reputation for paying higher dividends.
Further support for dividend clienteles stems from Graham and
Kumar’s analysis of household trading behavior. They report that
households trade around dividend events such as ex-dividend days,
dividend announcement dates, and dividend initiations. Before the ex-
dividend date, households that are older or have less income are net
buyers of stock. These households also increase their holdings after div-
idend initiations and exhibit abnormal buying behavior after dividend
announcements. These behavior patterns appear to affect the prices
for the stocks of small firms, where individual investors’ trades play an
especially important role.
Consistent with the predictions of behavioral dividend theory,
retirement status increases the demand for dividends. Graham and
Kumar regress portfolio yield on a series of variables including age
and a retirement status dummy. They report that the age variable has
a positive and significant coefficient estimate, as does a dummy for
retirement status.
Dong et al. (2005) develop panel data from a survey of individ-
ual investors in the Netherlands. On a weekly basis, their survey
asked questions about personal finance and consumption matters. Of
the 2,035 survey respondents, 555 own or previously owned shares
in exchange-listed companies or investment funds. Of these investors,
less than 20 percent own individual stocks. Also, 40 percent of the
investors are at least 55 years of age and 20 percent are at least 65 years
of age.
Consistent with the findings of Graham and Kumar (2006) and
Baker et al. (2007a), Dong et al. (2005) find that consumption from div-
idends is confined to investors who are either older or have low incomes.
74 Behavioralizing Portfolio Selection Theory
Outside this group, individual investors do not consume a large portion
of their dividends but instead re-invest their dividend income. In con-
trast to the findings reported by Baker et al., they find that the propen-
sity to consume from dividend income is less than from regular income.
Dong et al. (2005) assert that the findings described previously run
counter to the behavioral predictions from Shefrin and Statman (1984).
However, as pointed out earlier, behavioral theory does not predict that
all households consume dividends. Instead, behavioral theory only pre-
dicts that older, retired households and low-income households favor
dividend-paying stocks to finance consumption. Younger investors with
moderate to high incomes have little need to finance consumption with
dividends.
According to Dong et al. (2005), their findings support the behav-
ioral prediction for stock dividends. They find that in cases where firms
cannot pay a cash dividend, investors indicate that they would prefer
companies to pay a stock dividend, as opposed to no dividends. This
is a framing effect, because stock dividends involve costs but do not
increase the fundamental value of investors’ positions. As Dong et al.
(p. 124) indicate, “in principle stock dividends are no more than stock
splits.”
Baker et al. (2007a) provide the strongest evidence for a connection
between consumption and dividends. Their analysis makes use of two
types of data sets: the Consumer Expenditure Survey (CEX) from the
Bureau of Labor Statistics and brokerage account data. The CEX pro-
vides panel data on household consumption, income, wealth, dividends,
and demographic information. The brokerage data provide information
about how individual investors behave when they receive dividends.
The CEX data used by Baker et al. (2007a) cover the period 1988–
2001 and involve 3,272 household-year observations. Total expenditure
including durables is about $50,000 (measured in December 2001 in
dollars). Total income, which includes dividends but not capital gains,
has a mean of $56,789. Financial wealth is typically around a third
of total wealth. For the mean household, interest income is $1,207 and
dividend income is $891. On average, interest and dividends account for
4 percent and 2 percent of total income, respectively. The distribution
is skewed, with the median household reporting zero dividend income.
4.2 Quasi-Maximization 75
The BLC predicts that the marginal propensity to consume from
the income mental account exceeds the marginal propensity to consume
from the asset mental account. Shefrin and Statman (1984) suggest that
dividends flow through the income mental account and that capital
gains flow through the asset account. This implies that the marginal
propensity to consume from dividend income is about the same as
the marginal propensity to consume from total income, which in turn
exceeds the marginal propensity to consume from capital gains.
Baker et al. (2007a) estimate that the marginal propensity to con-
sume from dividend income is 0.49, the same as the marginal propen-
sity to consume out of total income, whereas the marginal propensity
to consume from total current-year returns is close to zero.
4.2.1.2 Options and Structured Products
Shefrin and Statman (1993a) apply the framework developed in Shefrin
and Statman (1984) to other aspects of portfolio design. The general
issue Shefrin and Statman (1993a) address how psychological issues
impact the design and marketing of financial products, especially struc-
tured products.
Shefrin and Statman (1984) focus on the use of option strategies,
most notably covered call writing, which has been the most popular
option strategy for individual investors. Although considerations from
neoclassical finance there indicate little merit to covered call writing,
the same cannot be said for behavioral finance. Shefrin and Statman
(1993a) use prospect theory to emphasize the framing elements inherent
in a covered call strategy, noting how writing a call option which is fully
covered by a holding the underlying stock masks a loss on the option
when the stock price increases. Moreover, by segregating the option
premium received from other cash flows such as dividends received,
hedonic editing allows the premium to be emphasized psychologically,
thereby generating an additional psychological benefit.
The analysis in Shefrin and Statman (1993a) identifies conditions
under which investors will find it attractive to use options in particular
ways such as writing call options on stocks in their possession, using put
options for protection on stocks they hold, and purchasing call options
76 Behavioralizing Portfolio Selection Theory
for speculative purposes. Because of hedonic editing, the premium to
be paid when purchasing put options for protection is detracting, just
as the premium received when writing call options is enhancing.
Evidence discussed in Lakonishok et al. (2007) updates some of the
discussion in Shefrin and Statman (1993a). Lakonishok et al. (2007)
document the relative frequency of different option trading strategies
by different classes of investors. They find that a large percentage of call
writing stems from covered call positions. Surprisingly, they also find
that volatility trading, using techniques emphasized in finance text-
books such as straddle positions, only accounts for a small fraction of
option trading volume. Interestingly, most individual investors do not
use protective puts. However, some do tend to use call options for spec-
ulative purposes. During the dot-com bubble, many did so for growth
stocks, but not value stocks. Moreover, empirical evidence indicates
that individual investors do not follow rational exercise strategies, as
many exercise options prematurely (Poteshman and Serbin, 2003).
The mental accounting treatment of covered call writing involves the
construction of mental accounts using both a stock and a call option.
This feature indicates that mental accounts can be more complex than
simple position holders for single assets. In this regard, McConnell and
Schwartz (1992) describe how individual investors use the interest, but
not principal, from their money market accounts to fund the purchase
of out-of-the-money call options. This strategy provides both downside
protection and upside potential, in accordance with the features empha-
sized in behavioral portfolio theory. McConnell and Schwartz (1992)
describe this investing pattern as the motivation for the development
of the liquid yield option note (LYON), a particular financial product
with complex option properties. A key takeaway from the discussion
is that when cognitive limits can prevent investors from making use
of multiple securities to form complex mental accounts, financial engi-
neers might be able to create value by doing so for them, by creating
financial products such as the LYON.
4.2.2 Theoretical Issues
A traditional model of dynamic portfolio choice involves an expected
utility maximizing investor choosing, at each time t, consumption (c
t
)
4.2 Quasi-Maximization 77
and securities (x
t
= x
t,1
· · · , x
t,J
) given initial wealth (W), a stochastic
stream of labor income (L
t
), and stochastic prices (q
t
) (Merton, 1971;
Foldes, 1978). The standard Euler condition for this problem involves
the purchase of a marginal unit of security j at time t, and requires
that the marginal benefit of this purchase be equal to the marginal cost.
The marginal benefit is the expected marginal utility of consumption
at time t + 1 generated by the marginal investment in security j. The
marginal cost is the foregone marginal utility of consumption at time t,
as the increased expenditure on security j comes at the expense of less
consumption at time t.
In the traditional model, the expected utility function has the form:
E
¸
¸
t
δ
t
u(c
t
)
¸
, (4.2)
where δ is a subjective time preference discount factor, and the expec-
tation is taken over a subjective probability measure (P) which an
investor associates with the underlying stochastic process. The Euler
condition has the following form:
q
t,j
∂u
∂c
t
= E
¸
q
t+1,j
δ
∂u
∂c
t+1

. (4.3)
In this setting, the purpose of the portfolio is to manage the risk
profile of the investor’s consumption stream, based on initial wealth
(W) and stochastic labor income (L). This means that the portfolio
serves to hedge uncertain labor income so as to smooth consumption
over time. Unless labor income is highly volatile, most trading activ-
ity would involve marginal adjustments to a diversified portfolio with
the purpose of rebalancing or dealing with liquidity needs to finance
consumption.
Consider a behavioral analog to the traditional framework, with a
view to modeling issues such as the disposition effect, attention hypoth-
esis, use of dividends to finance consumption, status quo bias, and pref-
erence for positive skewness. See Hoffmann et al. (2010).
Begin with a full maximization extension to Equation (4.3), which
we subsequently develop into a version featuring quasi-maximization.
Write the analog of expected utility as an objective function U. Let
U have as its arguments consumption stream c = [c
t
], portfolio choices
78 Behavioralizing Portfolio Selection Theory
x = [x
t
], changes in portfolio positions y = [x
t
− x
t−1
], prices q = [q
t
],
and probability beliefs (P). The arguments c, x, and y are random vari-
ables. The inclusion of x, y, and q as arguments allows for an investor’s
preferences to reflect not only consumption, but also the performance
of his or her portfolio and the impact of gains and losses from trading.
In the neoclassical framework, investors with predictable streams of
labor income make small but frequent adjustments to their portfolios,
by weighting the costs of foregone marginal current consumption asso-
ciated with a marginal security purchase against the expected marginal
future consumption so generated. Notice that the criterion driving port-
folio choice is consumption and savings.
In the corresponding behavioral framework, the consumption–
savings feature is augmented by additional considerations. When a
behavioral investor contemplates a marginal increase in his or her hold-
ings x
t,j
of security j at time t, (s)he adds three additional compo-
nents to the neoclassical calculus. Those components take the form of
∂U/∂x
t,j
, ∂U/∂y
t,j
, and ∂U/∂y
t+1,j
. The behavioral Euler condition is:
q
t,j
∂U
∂c
t
=
¸
t+1

q
t+1,j
∂U
∂c
t+1

+
∂U
∂x
t
+
∂U
∂y
t

¸
t+1
∂U
∂y
t+1
. (4.4)
The term
¸
t+1

q
t+1,j
∂U/∂c
t+1

is the analog to E

q
t+1,j
δ∂u/
∂c
t+1

, with the summation
¸
t+1
over the support of outcomes
at t + 1. Note that because of probability weighting, the behavioral
expression might not correspond to an expectation with respect to the
investor’s beliefs. The term ∂U/∂x
t
captures the effects of marginal
evaluation utility, meaning the psychological effect the investor expe-
riences from the value of his portfolio at different points in time. Here
an investor’s sense of well-being at a given moment, apart from his
consumption, is enhanced when his or her portfolio grows, and is dimin-
ished when it falls. The terms ∂U/∂y
t
and ∂U/∂y
t+1
capture realiza-
tion utility (Barberis and Huang, 2008), meaning the impact of trading
a position. The minus sign associated with the term
¸
t+1
∂U/∂y
t+1
in Equation (4.4) reflects the fact that increasing x
t
reduces y
t+1
=
x
t+1
− x
t
. In this respect, an investor who sells at a gain might expe-
rience positive realization utility, whereas an investor who sells at a
4.2 Quasi-Maximization 79
loss might experience negative realization utility (Thaler and Johnson,
1990).
A behavioral investor who contemplates a marginal increase in his or
her holdings of security j at time t views the benefits in terms of utility
from additional consumption at t + 1, increased evaluation utility at
t + 1, and possibly increased realization utility at t. The corresponding
cost is measured as foregone utility of consumption at t. Notably, the
behavioral terms are unsigned. Depending on circumstances and the
investor’s characteristics, the additional behavioral terms can either
serve to augment or reduce the benefits of a higher x
t,j
.
The behavioral extension to the neoclassical portfolio choice model
can be structured to incorporate the behavioral phenomena mentioned
above. The extension involves issues related to preferences as well as
issues related to biased beliefs. As to preferences, mental accounting,
evaluation utility, and realization utility stem from the dependence of U
upon x and y. Reference points and aspiration levels for both consump-
tion and portfolio values can be embedded in the specific form for U.
Gains and losses enter through the dependence of U on y. A simi-
lar statement applies to the disposition effect, with realization utility
associated with gains or losses on a security, relative to a pre-specified
reference point being a prime driver of whether or not to sell a security
in the portfolio (Shefrin and Statman, 1985).
Status quo bias leads investors’ reluctance to trade resulting, for
example, from regret avoidance Samuelson and Zeckhauser (1988).
3
Benartzi and Thaler (2007) and Benartzi et al. (2009) document the
strength of status quo bias in saving behavior. Mitchell et al. (2006)
report that 80 percent of participants in 401(k) accounts initiate no
trades in a two-year period, and an additional 11 percent make only
one trade. Status quo bias significantly contributes to investors’ lack
of diversification. In our behavioral Euler condition (Equation (4.4))
3
Regret is one component of realization utility, and relates to the consequence stemming
from a reference action. The behavioral approach stresses the importance of reference
points and aspiration levels. Notably, these might be based on perceived performance by
others in an investor’s peer group. A new line of inquiry investigates the role of social
networks on investors’ decisions (Kaustia and Kn¨ upfer, 2009).
80 Behavioralizing Portfolio Selection Theory
status quo bias is captured by ∂U/∂y
t
< 0 for y
t
> 0 and ∂U/∂y
t
> 0
for y
t
< 0 with a point of non-differentiability (kink) at y
t
= 0.
Status quo bias does not mean that investors refrain from trading
altogether, only that other forces must be strong enough to overcome
their reluctance. Certainly, if the needs for hedging, rebalancing, and
liquidity are sufficiently strong, investors will overcome status quo bias
and trade. Likewise, investors can overcome status quo bias if they have
enough conviction in their stock picking skills to feel little potential for
regret (Kahneman et al., 1991) or derive sufficiently high evaluation
utility from their portfolios.
In this last regard, Huddart et al. (2009) report that volume tends
to spike when the price of a stock crosses a previous low or high. More-
over, the longer the time since the previous low or high, the larger the
increase in volume. The effect is accentuated during periods when sen-
timent is high, as measured by Baker and Wurgler (2007), and is more
concentrated in smaller stocks that are difficult to value, thereby sug-
gesting that beliefs lie at the root of the issue. See also Statman et al.
(2006) who document that trading volume is higher in rising markets
than declining markets.
In the framework being sketched here, belief effects such as biases
and ambiguity are encapsulated within the criterion function U. An
example of a bias is overconfidence, which can lead investors to under-
estimate risk. Barber and Odean (2000) emphasize that overconfidence
leads investors to trade too much, just the opposite of status quo bias.
In one sense, the juxtaposition between the two effects highlights the
nature of heterogeneity within the population. In another sense, the
two coexist, in that active traders tend to concentrate their trades in
just a few stocks.
Most individual investors have only the vaguest notion of how secu-
rity returns are jointly distributed (Benartzi, 2001). For this reason,
investors are unlikely to maximize fully. Instead, they are prone to
engage in mental accounting, both to simplify the decision task and
to exert self-control. Mental accounting is a key reason underlying
the violation of stochastic dominance in the prospect theoretic choice
experiments reported in Kahneman and Tversky (1979). This violation
underscores the fact that prospect theory is not a framework for full
4.2 Quasi-Maximization 81
maximization, but a heuristic-based framework to explain how peo-
ple structure decision tasks and engage in quasi-optimization. In this
regard, choice in prospect theory features an evaluation function which
mimics the structure of an expected utility function, albeit imperfectly.
With this perspective in mind, I note that the beliefs used in con-
nection with Equation (4.4) across securities might not be compatible
with a single set of beliefs. Instead investors might apply Equation (4.4)
on a security by security basis, or perhaps a mental account by mental
account basis. As a result, the process for portfolio selection will resem-
ble the process for consumer choice, with diversification corresponding
to a “taste for variety.” Moreover, quasi-maximization is likely to be
accomplished in the context of a set of specific heuristics, as in Shefrin
and Thaler (1988), which provides the formal BLC framework under-
lying the explanation for dividends advanced in Shefrin and Statman
(1984). Behavioral portfolio models are also likely to feature a mixture
of diffusion processes and jump processes, with portfolio composition
changes reflecting jumps, as status quo bias is overcome.
5
Behavioralizing Asset Pricing Theory
Modern neoclassical asset pricing theory is built around the concept of
a stochastic discount factor (SDF) (Cochrane, 2005; Foldes, 2000). The
concept is very powerful, and allows most asset pricing models to be
expressed as special cases of a general framework. The neoclassical SDF
is structured as a monotone decreasing function, often interpreted as
the marginal rate of substitution for a representative investor. In log–
log space, the negative of the slope of the SDF is often interpreted as
the representative investor’s coefficient of relative risk aversion.
In many neoclassical models, the SDF is treated as time invariant.
However, some authors use models displaying time variation in the
SDF. Examples are Constantinides (1990) and Campbell and Cochrane
(1999) which feature stochastic risk aversion, the result of habit for-
mation. Habit formation involves the notion of reference consumption
levels, and is therefore related to reference point-based behavioral pref-
erence models. As was mentioned in Sections 2 and 3, Kamstra et al.
(2009) develops a model in which time varying, but deterministic, risk
aversion reflects the medical condition “seasonal affective disorder”
(SAD). SAD involves the impact of number of hours of sunlight on
rates of depression in the population. In their previous published work,
82
5.1 Stochastic Discount Factor 83
the authors find that in both the Northern and Southern hemispheres,
stock returns tend to rise in the autumn as the number of hours of
daylight falls, and rise in the spring as the number of hours of daylight
increases.
As it happens, empirical estimates of the SDF show it to be oscillat-
ing rather than being monotone decreasing (A¨ıt-Sahalia and Lo, 2000;
Rosenberg and Engle, 2004). In this section I discuss how behavioral
elements explain the oscillating shape. This section describes how the
neoclassical SDF framework can be extended to capture behavioral
beliefs and preferences, thereby behavioralizing asset pricing theory.
The extension serves to identify both strengths and weaknesses in the
behavioral approach. Much of the discussion below involves strengths.
An example of a weakness involves market models in which investor
preferences conform to prospect theory. De Giorgi et al. (2009) demon-
strate that under plausible conditions, equilibrium will feature investors
either holding (i) a risk-free portfolio or (ii) an unbounded portfolio.
As noted in Section 2, the behavioral asset pricing literature has
mostly developed around two-asset models where the two assets are
the risk-free security and a risky security. This statement applies to
models designed to analyze the cross-section, which is clearly problem-
atic, given that the cross-section features a multitude of securities.
The discussion below provides a formal definition for the SDF,
briefly outlines the nature of a neoclassical SDF, describes the behav-
ioral extension, and analyzes the implications of a behavioral SDF for
particular asset classes.
1
5.1 Stochastic Discount Factor
Consider the fundamental SDF-based asset pricing equation that the
price q of an asset with random payoff x is the expected value of its
discounted payoff, where m is the discount factor used to capture the
effects of both time value of money and risk. Formally,
q = E[mx]. (5.1)
1
The first part of this section is drawn from Shefrin (2008c).
84 Behavioralizing Asset Pricing Theory
In Equation (5.1), both m and x are random variables. That is, the
discount factor m typically varies across payoff levels in order to reflect
that risk is priced differently across payoff levels.
2
The SDF is a powerful concept, in that absent pure arbitrage oppor-
tunities, it underlies prices, returns, and risk premia associated with
all assets, be they fixed income securities, equities, options, and other
derivatives.
5.1.1 Neoclassical SDF
Think of the SDF m as state price per unit probability. In the neoclas-
sical framework, m takes on higher values in unfavorable states than
favorable states, giving rise to monotone declining shape across states.
The capital asset pricing model (CAPM) is an important special
case of the SDF framework, and corresponds to the case when m is a
linear function of the market return. Another special neoclassical case
occurs when equilibrium prices are set as if there is a representative
investor with correct beliefs and preferences that conform to constant
relative risk aversion (CRRA). In this case, log(m) is linear in the log
of the gross market return.
As I mentioned above, in the neoclassical representative investor
framework, m can be interpreted as a conditional marginal rate of
substitution, the amount of current consumption the representative
investor is willing to sacrifice in exchange for a marginal unit of future
consumption in a particular state. In this regard, the negative slope of
the neoclassical SDF reflects aversion to risk. The greater the degree of
risk aversion, the steeper the slope. Risk neutrality corresponds to the
case of a zero slope.
The SDF m also serves as a change of measure, transforming the
true (also known as objective or physical) probability density function
into the risk-neutral measure (Cochrane, 2005).
5.1.2 Behavioral SDF
Shefrin (2008a) establishes that when all investors have power utility,
the log-SDF can be interpreted as the sum of a component relating
2
Dividing by q leads to a similar expression on a return or per dollar basis, namely where
R denotes the security return.
5.1 Stochastic Discount Factor 85
to fundamentals and a component relating to market sentiment. The
market sentiment term Λ is the sum of a log-change of measure ln(ξ)
and a time preference shift term parameter δ
S
.
Λ = ln(ξ) + ln(δ
S
). (5.2)
The log-change of measure has the same interpretation as in Section 3,
but for the probability density function associated with equilibrium
prices rather than an individual investor. Just as in Section 3, bias is rel-
ative to probabilities associated with market efficiency. In this respect,
prices of all securities are efficient when the corresponding log-change
of measure is zero. Note that sentiment is a stochastic process whose
underlying probabilities derive from the objective process for funda-
mentals. Sentiment reflects investors’ misjudgments and misreactions
to changing fundamentals, a result of suboptimal learning processes.
The fundamental variables entering the SDF are aggregate gross
consumption growth g, a term γ measuring the market coefficient of
relative risk aversion, and a term δ measuring the market time discount
factor. The source of γ and δ is obtained from an aggregation procedure
which is discussed below. Formally, the equation relating the log-SDF
and sentiment has the form:
ln(m) = ln(δ) − γln(g) + Λ. (5.3)
Equation (5.3) indicates the manner in which sentiment impacts
market prices through the SDF, especially its shape. In this regard,
Equation (5.3) implies that when sentiment is zero, the log-SDF cor-
responds to the neoclassical case. However, if the magnitude of sen-
timent is sufficiently large, then Λ will drive the shape of the SDF.
Figure 5.1 contrasts a neoclassical log-SDF and a behavioral log-SDF,
both graphed against ln(g). The neoclassical SDF is linear with a nega-
tive slope. The behavioral SDF depicted in the figure oscillates, with the
probabilities of extreme negative events overweighted and extreme pos-
itive events underweighted. This type of SDF occurs when the market
is dominated by a combination of overconfident optimists and under-
confident pessimists, a case discussed in further detail below.
Using options data, A¨ıt-Sahalia and Lo (2000) and Rosenberg
and Engle (2004) estimate the empirical SDF, or more precisely its
86 Behavioralizing Asset Pricing Theory
-30%
-20%
-10%
0%
10%
20%
30%
40%
50%
60%
-
4
%
-
3
%
-
1
%
1
%
3
%
4
%
6
%
ln(g)
-ln(g) = neoclassical log-SDF
Sentiment Function
behavioral log-(SDF)
Fig. 5.1 Illustration of the decomposition in Equation (5.3), contrasting a behavioral SDF
and a neoclassical SDF.
projection in respect to returns on the S&P 500. Both papers find that
the empirical SDF has the behavioral shape portrayed in Figure 5.1.
The Rosenberg–Engle paper is especially interesting, in that the
authors estimate the SDF in two ways. First, they restrict the SDF to
have the traditional neoclassical shape displayed in Figure 5.1. Second,
they use a free form Chebyshev polynomial procedure that involves no
such restriction. This approach can be viewed as an inadvertent test
of whether the empirical SDF is behavioral or not. Their empirical
findings provide strong support that the SDF is behavioral.
Barone-Adesi et al. (2008) estimate the empirical SDF using a
GARCH-based filtered historical simulation technique. Notably, their
approach allows the objective pdf and risk-neutral pdf associated
with the representative investor to feature different volatilities. In this
regard, behavioral equilibrium naturally features differential volatil-
ities for the objective pdf and risk-neutral pdf, largely because of
the role played by sentiment Λ in the SDF m. The new GARCH-
based filtered simulation approach produces a shape for the SDF which
5.1 Stochastic Discount Factor 87
features dampened oscillations relative to the Gaussian-based approach
of Rosenberg–Engle.
The behavioral SDF is more volatile than its neoclassical counter-
part. This difference in volatility has profound implications for the mag-
nitudes of risk premiums and Sharpe ratios. Sharpe ratios are bounded
from above by the coefficient of variation of the SDF. As a result, the
more volatile behavioral SDF admits higher risk premiums and Sharpe
ratios than does the less volatile neoclassical SDF.
In Shefrin (2008a), I suggest that the shape of the SDF reflects
the joint distribution of investors’ biases. In this respect, I suggest
that the shape of the behavioral SDF depicted in Figure 5.1 reflects a
negative correlation between investors’ errors about the first and second
moments of returns. Using survey data, the 2005 edition of Shefrin
(2008a) predicted that after 2001, pessimistic investors would become
less underconfident, thereby resulting in a flatter left tail for the log-
SDF depicted in Figure 5.1, see Figure 5.2. The findings reported in
Barone-Adesi et al. (2008) indicate that the tail did indeed flatten in
this period.
-60%
-50%
-40%
-30%
-20%
-10%
0%
10%
-
4
%
-
3
%
-
1
%
1
%
3
%
4
%
6
%
ln(g)
-ln(g) = neoclassical log-SDF
Sentiment
Function
behavioral log-(SDF)
Fig. 5.2 Illustration of the impact of less underconfidence by pessimists in Figure 5.1.
88 Behavioralizing Asset Pricing Theory
5.2 How Markets Aggregate Investor Attributes
The empirical evidence strongly indicates that there is considerable
heterogeneity in respect to investors’ beliefs, degree of risk aversion,
and rate of impatience. Some of this evidence is direct, meaning that
it stems from surveys concerning these variables. Other evidence is
indirect, being based on investors’ decisions, such as the composition
of their portfolios.
3
Therefore, a core issue in asset pricing theory is to
understand how markets aggregate investors’ beliefs and preferences.
In regard to preferences, recall from the discussion in Section 3 that the
change of measure technique applies both to preferences and to beliefs.
5.2.1 Aggregating Beliefs, Risk Tolerance, and
Rates of Impatience
The study of the aggregation question dates back at least as far as Lint-
ner (1969). Shefrin (2008a) presents an equilibrium aggregation result
for a complete markets model featuring power utility. He demonstrates
that the equilibrium market pdf is a generalized weighted H¨ older aver-
age of the individual investors’ pdfs, with the weights reflecting rel-
ative wealth or consumption, and the exponents being coefficients of
relative risk aversion. Shefrin (2008a) also analyzes aggregation under
the assumption of constant absolute risk aversion (CRRA), although
he suggests that CARA is unrealistic from a behavioral perspective.
Jouini and Napp (2006, 2007) establish similar results. They focus on
the special case when investors share the same coefficient of relative risk
aversion. Jouini and Napp’s analysis of aggregation under the assump-
tion of constant absolute risk aversion is more general.
For a detailed discussion of the aggregation theorems, readers are
referred to Shefrin (2008a) and the works of Jouini and Napp (2007).
For the purpose of the current discussion, it is sufficient to describe
the general features of the representative investor for the case of
power utility (CRRA). The log-SDF for the case when investors have
3
In a series of papers, Barber et al. (2005), Barber et al. (2009a), and Barber et al. (2009b)
provide evidence that individual investors as a group influence prices, but lose money to
professional investors. However, they also document considerable heterogeneity, demon-
strating that some individual investors possess superior skills.
5.2 How Markets Aggregate Investor Attributes 89
CARA utility functions is similar in form, except that all averages are
unweighted, rather than being consumption-weighted as is the case with
CRRA utility (Shefrin, 2008a).
Heterogeneity leads the representative investor’s beliefs to be differ-
ent from those of the individual investors. For example, suppose that
all investors have log-utility, share the same intertemporal discount
factor δ, and have normally distributed beliefs with different means. In
this case, the generalized weighted H¨older average is a simple wealth-
weighted average of probability density functions. If the means are suf-
ficiently different across investors, such a pdf can be multi-modal, and
therefore non-normal. If investors differ in their intertemporal discount
factors, the representative investor’s stochastic process will typically
violate Bayes’ rule.
At any moment in time, the representative investor’s degree of
risk tolerance will be a consumption weighted convex combination of
the individual investors’ degree of risk tolerance. Because consump-
tion shares are time varying, heterogeneity leads the representative
investor’s utility exponent γ to be state dependent. This means that
γ will not measure the representative investor’s degree of relative risk
aversion.
The representative investor’s intertemporal discount rate δ is a non-
linear function of the individual investors’ discount rates. However,
because the aggregation process relates to discount factors, not dis-
count rates, the representative investor’s discount factor will typically
be non-exponential, even though every investor has an exponential dis-
count factor. Moreover, because H¨ older averages do not always sum
to unity, the representative investor’s discount rate will also reflect
probability aggregation. Stated somewhat differently, the representa-
tive investor might exhibit aversion to ambiguity, even if none of the
individual investors do.
The characteristics of the aggregation process hold clear lessons
for asset pricing theorists of all stripes, neoclassical and behavioral.
Typically, the market does not aggregate heterogeneous beliefs and
preferences in a way that leads to the existence of a typical neoclassical
representative investor, or a representative investor whose preferences
and beliefs feature typical behavioral characteristics.
90 Behavioralizing Asset Pricing Theory
5.2.2 Liquidity Premium and Asymmetric Volatility
The preceding discussion describes the manner in which sentiment
reflects the aggregation of investors’ beliefs and preferences, while
Equation (5.3) reflects the manner in which the SDF incorporates
sentiment. In this respect, consider how market prices reflect a liquidity
premium.
In Section 4, I made the point that liquidity issues can arise because
at a given time t only a subset of investors actively trade. The others
exhibit status quo bias which leads them to refrain from trading as a
result of being situated at kinks in their preference maps. As a result,
particular investors wishing to engage in trade at t might find that
inducing sufficiently large positions on the part of their active trading
partners requires market prices to move by a “nonnegligible” amount.
In this respect, market prices at t, and hence sentiment, would reflect
the beliefs and preferences of investors who engage in trade, but not
investors who refrain from trade. It is in this sense that in the short-run
sentiment reflects a liquidity premium.
As a general matter, the degree of status quo bias might be condi-
tional on circumstances. For example, if regret potential increases after
declines in aggregate consumption growth, then status quo bias might
increase following declines in the return to the market portfolio. To
the extent that the liquidity premium increases after such a decline,
the result would be asymmetric volatility in the returns to particular
securities. One indication of the extent to which status quo bias plays
a role is the change in trading volume. Typically, an increase in status
quo bias would lead to a decline in trading volume. In this regard, see
Statman et al. (2006).
Asymmetric volatility might reflect asymmetric hot hand fallacy on
the part of some investors, in the sense that investors predict stronger
continuation after downturns than after upturns. At the same time,
asymmetric hot hand fallacy would tend not to predict sharp upturns
after steep declines, whereas status quo bias would.
5.2.3 Risk Aversion Puzzles
Assuming the existence of a typical neoclassical representative investor
can lead to puzzles. For example, Jackwerth (2000) estimated the
5.3 Aggregation and the Shape of Sentiment 91
coefficient of absolute risk aversion for the market in the case when
sentiment is (implicitly) zero. The papers by A¨ıt-Sahalia and Lo (2000)
and Rosenberg and Engle (2004) also estimate market risk aversion
under the assumption that sentiment is zero. All of these studies find
coefficients of risk aversion for the market that are extreme. Reasonable
benchmark values for the coefficient of relative risk aversion (CRRA)
involve the range 1.5–3.5, with a few outliers between 0 and 1 and
between 5 and 6.5.
The A¨ıt-Sahalia-Lo range for CRRA is 1–60. The Rosenberg–Engle
estimated range for the CRRA is 2–12, with a mean of 7.6. Moreover,
they find that market risk aversion is highly variable over time. Jack-
werth (2000) estimates imply that the market’s coefficient of CRRA
ranges between −14 and 27. Negative values, which indicate risk seek-
ing, are especially surprising. Blackburn and Ukhov (2006), which
extends Jackwerth’s analysis from the market to individual stocks, also
features highly variable rates of risk aversion that take on exotic values.
However, negative values for CRRA and values over 6 are at odds
with survey data. Jackwerth points out that the shape of the absolute
risk aversion function, which he estimates for the market, is not mono-
tone declining and time invariant, as neoclassical theory would suggest.
Instead, it mostly tends to be U-shaped and time varying.
The culprit in these risk aversion estimation exercises is the assump-
tion that sentiment is zero, with the SDF conforming to a monotone
decreasing function. If sentiment is nonzero in practice, and oscillates,
then the upward sloping segments will appear to imply negative risk
aversion. Incorrectly assuming it to be zero in theory forces the mod-
els’ risk aversion parameters to pick up the effects of market sentiment,
such as optimism, pessimism, and overconfidence.
5.3 Aggregation and the Shape of Sentiment
The easiest way to understand the shape of sentiment is by working
with examples featuring log-utility. Consider a complete market model
featuring two investors, whose initial wealth levels are equal. Suppose
that both investors have log-utility functions. Assume that the log of
consumption growth is normally distributed, and that both investors
believe it to be normally distributed. However, let the two investors
92 Behavioralizing Asset Pricing Theory
disagree about the first moment. Assume that one investor is excessively
bullish about mean consumption growth, while the second investor is
excessively bearish about mean consumption growth. For now, let both
investors hold correct beliefs about the second moment. Notably, in the
special case of log-utility, the H¨ older average becomes a simple convex
combination.
As pointed out in Section 3, associated with the bullish investor
is a log-change of measure which is linear with positive slope. Associ-
ated with the bearish investor is a log-change of measure which is linear
with negative slope. Recall that the market equilibrium aggregates pdfs,
not log-change of measure functions. The resulting log-change of mea-
sure for the representative investor will be U-shaped. Therefore, the
bearish investor drives the shape of market sentiment for unfavorable
states, whereas the bullish investor drives the shape of market senti-
ment for favorable states. Notably, a U-shape implies that the market
overweights the probabilities of extreme events, be they favorable or
unfavorable.
If investors are overconfident, then they will underestimate the prob-
abilities of extreme events. As a result, the sentiment function can
have an inverse U-shape. The inverse-U can be interpreted as reflecting
the “black swan effect:” (Taleb, 2007). See Odean (1998b) for the first
theoretical treatment dedicated to the question of how overconfidence
impacts asset prices.
Finally, if bearish investors tend to be underconfident, in that they
overpredict the probabilities of crashes, while bullish investors tend to
be overconfident, then the sentiment function will tend to be oscillating,
as depicted in Figure 5.1.
5.4 Risk and Return
5.4.1 Behavioral Mean-Variance Returns
A mean-variance (MV) portfolio is a portfolio of assets that maximizes
expected return for a specified return variance. In neoclassical asset
pricing, where sentiment is zero, the risk premium for a security is based
on its return covariance with any risky mean-variance portfolio. For
example, in the CAPM the market portfolio is mean-variance efficient,
5.4 Risk and Return 93
and this is why risk premiums are based on the covariance between the
security’s returns and the returns to the market portfolio. In the case
of CAPM, the SDF is linear and negatively sloped.
How does sentiment impact the relationship between risk and
return? The short answer is in the same way as in neoclassical asset
pricing. Because risk premiums for all securities are based on return
covariance with MV portfolios, even when sentiment is nonzero, the
key lies in understanding how sentiment impacts the nature of the
return distributions for both MV portfolios and individual assets. This
means appreciating that systematic risk, meaning risk which is priced,
has both a fundamental component and a sentiment component.
The SDF can be used to price all assets, including portfolios. Hence,
the SDF also prices all mean-variance efficient portfolios, and therefore
the SDF can be used to generate the return distribution for an MV
portfolio. Just as in neoclassical asset pricing theory, the SDF generates
the mean-variance frontier. The relationship is the same. Effectively,
the return to a risky mean-variance porfolio is a linear function of the
SDF, with a negative slope coefficient. Therefore, the shape of a mean-
variance portfolio return function is obtained by inverting and tilting
the SDF (Cochrane, 2005; Shefrin, 2005).
Figure 5.3 contrasts the return pattern for a behavioral mean-
variance portfolio to the return pattern for a neoclassical mean-variance
portfolio. Each curve in Figure 5.3 is a plot of the linear function
described in Shefrin (2008a) that links the return of an MV portfolio
to the consumption growth rate through the SDF. The behavioral MV
curve corresponds to the case in which sentiment is given by Figure 5.1,
whereas the neoclassical MV curve corresponds to the case when sen-
timent is zero. Notice that the shapes in Figure 5.3 correspond to the
inverted, tilted shapes in Figure 5.1.
Figure 5.3 has three points worthy of note. First, the return to
a traditional MV portfolio is approximately the return to a portfolio
consisting of the market portfolio and the risk-free security. For the
purpose of illustration, the weight attached to the risk-free asset in
both portfolios is high.
Second, the return to a behavioral MV portfolio is more volatile than
the return to a traditional MV portfolio. This is because the peaks and
94 Behavioralizing Asset Pricing Theory
0.75
0.8
0.85
0.9
0.95
1
1.05
1.1
-
4
%
-
3
%
-
3
%
-
2
%
-
1
%
0
%
1
%
2
%
3
%
3
%
4
%
5
%
6
%
ln(g)
M
e
a
n
-
v
a
r
i
a
n
c
e

R
e
t
u
r
n

(
G
r
o
s
s
)
Behavioral MV Portfolio Return
Efficient MV Portfolio Return
Fig. 5.3 This figure contrasts the return pattern for a neoclassical MV portfolio and a
behavioral MV portfolio associated with Figure 5.1.
valleys in the behavioral MV portfolio correspond to exposure from
risky arbitrage. This is a true MV portfolio, based on the objective
pdf. Therefore, maximizing expected return involves the exploitation
of nonzero sentiment.
Notice that in Figure 5.1, sentiment is positive at the extreme left
and negative at the extreme right. This means that extreme out-of-
the-money put options on the market portfolio are overpriced, while
extreme out-of-the-money call options are underpriced. As a result, a
true mean-variance portfolio would feature a short position in extreme
out-of-the-money put options on the market portfolio and a long posi-
tion in extreme out-of-the-money call options on the market portfolio.
Third, expected returns and risk premiums for securities are deter-
mined by security betas, taken with respect to a risky mean-variance
portfolio benchmark. In this case the risk premium to a security is
the product of the security’s beta and the risk premium of the mean-
variance benchmark.
5.4 Risk and Return 95
5.4.2 Coskewness
MV-returns are very negatively skewed relative to the market portfolio
because returns to a behavioral MV-portfolio are not only extremely
low in low-growth states, but fall off quickly as the rate of consumption
growth declines. Some readers might be surprised that skewness is an
issue at all in a discussion about mean-variance portfolios. After all,
skewness pertains to the third moment, and mean-variance preferences
are neutral in respect to all moments higher than the second. The rea-
son why skewness can be an issue for a behavioral MV portfolio involves
the risky arbitrage feature of true MV portfolios. An investor who takes
a large short position in extreme out-of-the-money put options on the
market portfolio will earn a very low return when the return on the
market portfolio is very low. An investor who takes a long position
in extreme out-of-the-money call options on the market portfolio will
earn a high return when the return on the market portfolio is very
high.
To understand the implications that the behavioral MV shape in
Figure 5.3 has for the nature of risk premiums for individual securities,
consider what systematic risk entails. Systematic risk is risk associ-
ated with the returns to an MV portfolio. Securities whose returns are
highly correlated with the returns to an MV portfolio will be associ-
ated with high degrees of systematic risk. In this regard, remember that
the returns to the behavioral MV portfolio depicted in Figure 5.3 are
negatively skewed relative to the market portfolio. Therefore, securities
whose returns are high in systematic risk will feature negative skewness
relative to the market portfolio.
Following Barone-Adesi and Talwar (1983), Harvey and Siddique
(2000) study the cross-section of stock returns using coskewness relative
to the market portfolio. They propose several definitions of coskewness.
One definition of coskewness is the beta of the security’s return rela-
tive to the squared market return, controlling for covariation with the
market return. With this definition, coskewness measures the extent to
which a security’s return covaries with the squared market return.
The definition of coskewness is particularly appropriate when the
SDF has a quadratic-like U-shape. Recall that the SDF will be
96 Behavioralizing Asset Pricing Theory
U-shaped when the sentiment component has a U-shape and is large
relative to the fundamental component of the SDF. In this case, the
shape of the MV function is an inverted-U. Observe that this type of
MV return function will feature negative coskewness with respect to the
market portfolio. In this situation, a security that is high in system-
atic risk, meaning risk that is priced, will mimic the MV-return pattern,
and therefore also feature negative coskewness. This is because coskew-
ness measures the amount of mean-variance skewness that a security’s
return adds to an investor’s portfolio. Coskewness plays a role analo-
gous to a factor loading in a multifactor asset pricing model. Notably,
there is a negative premium to holding stocks whose returns exhibit
positive coskewness relative to the market portfolio.
Of course, the shape of the behavioral MV return pattern depicted
in Figure 5.3 is not exactly an inverted-U. In the right portion of Fig-
ure 5.3, the behavioral MV-return pattern does not have the same
feature as an inverted-U. However, in the left portion, the patterns are
similar. This suggests that when sentiment has the shape depicted in
Figure 5.1, negative coskewness with respect to the market portfolio
will capture some, but not all, aspects of systematic risk. See Barone-
Adesi et al. (2004) for evidence that this is the case.
Harvey and Siddique (2000) study coskewness using a four-factor
model, where the factors are, respectively, the market return, size,
book-to-market equity, and momentum. Harvey and Siddique find that
the correlation between coskewness and mean returns of portfolios
sorted by size, book-to-market equity, and momentum is −0.71. This
means that much of the explanatory power of size, book-to-market
equity (B/M), and momentum in the returns of individual stocks plau-
sibly derives from coskewness.
Harvey and Siddique point out that the market factor by itself
explains only 3.5 percent of cross-sectional returns. However, the com-
bination of the market factor and coskewness explains 68.1 percent of
cross-sectional returns. This value rivals the 71.8 percent associated
with the use of the market factor, size, and B/M. For momentum,
recent winners feature lower coskewness than that of recent losers. All
of these findings are consistent with MV portfolio returns having the
shape depicted in Figure 5.3.
5.4 Risk and Return 97
Dittmar (2002) discusses the superiority of nonlinear models over
linear factor models such as the Fama–French three-factor model. He
develops a flexible nonlinear pricing kernel approach which is in the
tradition of a zero sentiment representative investor model. Notably, he
finds that the empirical SDF in his analysis is not monotone decreasing
over its range, but instead has a U-shaped pattern. In this regard, Poti
(2006) uses a quadratic U-shaped SDF to extend the Harvey–Siddique
analysis.
5.4.3 Decomposition of Risk Premiums
Let Z be an arbitrary security with return r
Z
. Letting Φ = exp(Λ)
define
h
Z
=
E[δΦg
−γ
r
Z
]
E[δg
−γ
r
Z
]
, (5.4)
where the expectation is taken with respect to the true underlying
stochastic process Π. Keeping in mind that Φ is proportional to a
change of measure, the impact of the numerator of Equation (5.4) is
to shift probability weight across different realizations of the product
g
−γ
r
Z
. Suppose that we shift probability to higher valued realizations,
and take the expectation with respect to the true underlying process.
In this case h
Z
> 1, and so the third term on the right-hand side of
Equation (5.5) is negative. Relative to fundamental value E[δg
−γ
r
Z
],
h
Z
represents the degree to which Z is overpriced (when h
Z
> 1) and
underpriced (when h
Z
< 1).
Shefrin (2008a) establishes that the risk premium associated with
Z is given by:
E[r
Z
] − i
1
= (i
1,e
− i
1
) −
Cov[g
−γ
r
Z
]
E[g
−γ
]
+ i
1,e
1 − h
Z
h
Z
, (5.5)
where i
1
is the market risk-free rate of interest, and i
1,e
is the value of
the risk-free rate when sentiment is zero and prices are efficient. When
prices are efficient, i
1
= i
1,e
. Notice that when investor errors lead to
an increase in the risk-free rate, the risk premium on Z declines.
Equation (5.5) has a straightforward interpretation. When senti-
ment Λ is zero, then Equation (5.4) implies that h
Z
= 1. In this case,
98 Behavioralizing Asset Pricing Theory
the third term on the right-hand side of Equation (5.5) is zero. In addi-
tion, i
1,e
− i
1
= 0 so that when prices are efficient, Equation (5.5) is just
equal to the middle term on the right-hand side. When sentiment Λ is
nonzero, Equation (5.4) implies that h
Z
may not be equal to 1. Notice
that Z being overpriced causes its risk premium to decline relative to
the case of market efficiency.
4
Observe that the expression (5.5) for the risk premium decomposes
into a fundamental component (the middle term on the right-hand
side) and a sentiment component (the first and third terms). In this
respect, when prices are inefficient, the first and third terms are typi-
cally nonzero. The variable h
Z
measures the degree to which security
Z is overpriced. For example, if h
Z
= 2, then Z is overpriced by 100
percent. Equation (5.5) indicates that the overvaluation of Z causes the
expected return to Z to change by
i
1,e
(1−h
Z
)
h
Z
which in this case equals
50 percent(=
1−2
2
) of what the gross risk free rate would be were prices
efficient.
5.4.3.1 Cross-Section
Equation (5.5) specifies how the cross-section is driven by a combina-
tion of fundamental risk and sentiment-based risk.
5
Fundamental risk
is determined by the return covariance with the fundamental compo-
nent of the SDF. Conditional on the interest rate, sentiment risk is
determined by the term
1−h
Z
h
Z
.
6
In addition to the discussion about coskewness, and its connection
to the cross-section, I also want to mention some experimental evi-
dence. For the last decade, I have surveyed money managers, analysts,
4
This statement is conditional on the equilibrium return distribution r
Z
. Notably, the equi-
librium return distribution to Z might be different in the case when prices are inefficient
relative to the case when they are efficient. Think about an out-of-the-money call option in
an example when the representative investor is excessively optimistic. The excessive opti-
mism will increase the price of the option, thereby left shifting and distorting its objective
return distribution at a given node in the uncertainty tree.
5
Equation (5.5) should be understood with reference to the projection of sentiment onto
the returns for individual securities.
6
In Section 4 I mentioned that the framework for behavioral portfolio selection resembles
the framework for consumer choice, as some investors build portfolios with only a vague
understanding of covariance. This feature does not negate Equation (5.5), as long as some
investors are able to exploit, and eliminate, pure arbitrage opportunities.
5.4 Risk and Return 99
and students to elicit their judgments about risk and return in con-
nection with the cross-section. The general finding from this survey is
that although those surveyed believe that in theory risk and return are
positively related, in practice they form judgments as if the two are
negatively related.
Judgments about risk are consistent with the neoclassical frame-
work. Most of those surveyed judge that risk is positively related to
market beta, book-to-market equity, and recent momentum, while neg-
atively related to market capitalization. In contrast to the neoclassical
framework, most also believe that the relationship between expected
returns and these same variables has the opposite sign as with risk.
Not surprisingly, judgments about risk and return turn out to be neg-
atively related.
Two psychological principles might be at work in explaining why
people form judgments as if they believe that risk and return are neg-
atively related. The first is representativeness. People might rely on a
heuristic whereby they believe that the stocks of good companies are
representative of good stocks. The second is that both low risk and
high returns generate positive affect, and so the affect heuristic can
also explain the negative relationship.
Baker and Wurgler (2006, 2007) discuss various measures of senti-
ment that have been used in behavioral studies. They then develop a
new index of sentiment based on the following six specific measures:
— closed-end fund discount,
— detrended log-turnover,
— number of IPO,
— first-day return on IPOs,
— dividend premium, and
— equity share in new issues.
Baker and Wurgler suggest that sentiment will impact prices of
some stocks more than others. They suggest that the returns to stocks
which are difficult both to value and to arbitrage will be more sensitive
to sentiment than stocks which are both easier to value and easier to
arbitrage. The heart of their argument involves the concept of a senti-
ment seesaw. When sentiment is high, speculative difficult-to-arbitrage
100 Behavioralizing Asset Pricing Theory
stocks become overvalued, and safe easy-to-arbitrage stocks become
undervalued. When sentiment is low, the reverse occurs. A notable fea-
ture of the analysis is that because of mispricing, stocks that appear to
be riskier in terms of fundamentals, can feature lower expected returns
than safer stocks.
In Baker–Wurgler, a stock’s sensitivity to sentiment is measured in
terms of a sentiment beta. A sentiment beta explains the sentiment
component of the risk premium. Conditional on the risk-free rate, this
component effectively corresponds to the third term in the right-hand
side of Equation (5.5); unconditionally, it corresponds to the sum of
the first and third terms.
Empirically, Baker–Wurgler report that returns are predictable,
conditional on the value of sentiment in the prior month. When past
sentiment has been high, subsequent returns to speculative stocks,
more difficult-to-arbitrage stocks are indeed lower than returns to safer
stocks which are easier to arbitrage. Conversely, when past sentiment
has been low, subsequent returns to speculative stocks, more difficult-
to-arbitrage stocks are higher than to safer stocks which are easier to
arbitrage.
5.5 Long-run Fitness
The neoclassical perspective has tended to be that information
investors will exploit any inefficiencies immediately, so that prices are
efficient at all times. In contrast, the behavioral position is different,
namely that in the short-run, limits to arbitrage can prevent mar-
ket prices from being efficient. The conditions for efficiency essen-
tially require that the average investor error be zero, so that investor
bias is unsystematic, and that the covariance between investor error
and investor wealth be zero, so that any errors are not concentrated
within the investor population (Shefrin and Statman, 1994). In contrast
to the neoclassical perspective, the behavioral perspective holds that
information traders, meaning investors who are free from error, will
not guarantee market efficiency, because information investors are typ-
ically reluctant to take on the risk associated with the large positions
required to eliminate inefficiencies completely.
5.5 Long-run Fitness 101
What about the long-run? Will the existence of information traders
eliminate any price inefficiencies? The neoclassical answer to this ques-
tion is yes, with the argument being that investors who are subject to
error will lose their wealth to information investors. In contrast, the
behavioral answer has been not necessarily, that investors who are sub-
ject to error can take on more risk than information investors, thereby
earning higher returns in the long-run, and surviving.
Lawrence Summers first proposed the intuition underlying the
behavioral perspective, and the idea was formally discussed by DeLong
et al. (1991). Their framework features CRRA utility functions, and
describes conditions under which investors who are prone to error sur-
vive in the long-run and impact prices.
The analysis in DeLong et al. (1991) is controversial. In particular,
the main result is at odds with the general literature on long-term
survivability or fitness. This literature has recently been surveyed by
Blume and Easley (2008). At the center of the fitness lies the concept
of entropy. Given two discrete probability density functions Π and P
over a series of K states, define
I
Π
(P) =
K
¸
k=1
Π(k)ln(Π(k)/P(k)), (5.6)
and call this variable the relative entropy of P with respect to Π.
Consider a set of investors indexed by j. Suppose that objective
probabilities correspond to Π. Blume and Easley establish that the
time average of investor j’s marginal utility to investor k’s marginal
utility is given by:
[ln(δ
k
) − I
Π
(P
k
)] − [(ln(δ
j
) − I
Π
(P
j
)]. (5.7)
They call the difference [ln(δ
j
) − I
Π
(P
j
)] investor j’s survival index.
Suppose that investor j has a lower survival index than investor k.
Then with probability one, the ratio of investor j’s marginal utility
approaches +∞. However, if physical resources are bounded from above
and investor utility is CRRA with risk aversion parameter γ
j
≥ 1, then
over time investor j’s consumption must go to zero.
Notice that there are two components to the survival index. The first
component relates to patience, and the second to errors in beliefs. The
102 Behavioralizing Asset Pricing Theory
survival condition suggests that long-run fitness is determined by being
sufficiently patient relative to other investors and/or being sufficiently
better informed about objective probabilities.
Setting aside impatience, the results in Blume and Easley (2008)
indicate that informed investors will dominate pricing in the long-run.
This means that there is good reason to believe that the fundamental
assertion put forth in DeLong et al. (1991) is incorrect. Indeed, Blume
and Easley (2008) state that they have difficulty following the logic
in the latter paper. In this respect, the model appears to be vaguely
specified, with the formal argument being loose.
Of course, it is entirely possible that investors with erroneous beliefs
persist in the long run, but for reasons different from those suggested
in the analysis of DeLong et al. (1991). For example, there may be no
informed investors at all, in that all investors are subject to some form
of error or bias. Alternatively, the market can feature incompleteness in
combination with overlapping generations with finite lifetimes. There-
fore, better informed investors might not live long enough to extract
most of the wealth from their less informed counterparts. Moreover,
even if they did so, their heirs might not possess the same degree of
superiority.
5.6 Pricing Dynamics
A stream of literature is developing that focuses on asset pricing dynam-
ics. The volume edited by Hens and Schenk-Hopp´e (2008) contains
many interesting contributions, not the least being the survey by Blume
and Easley mentioned above. Dumas et al. (2009) build on the work
of Scheinkman and Xiong (2003), who assume that the aggregate con-
sumption growth rate follows a mean-reverting diffusion process with a
time varying stochastic drift. Investors do not observe the drift directly.
Instead they receive a noisy signal whose change is equal to the sum
of drift plus a noise term that is uncorrelated with the change in drift.
All investors make forecasts of the change in drift. However, some
investors overreact to their forecast errors for the rate of change in
the signal. In this respect, they read more into their signal forecast
errors than is appropriate, because they believe that the noisy signal
5.6 Pricing Dynamics 103
features a positive correlation with the change in drift. As a result, they
overweight the contribution of their signal forecast errors when predict-
ing how the drift term will subsequently change. This overweighting
leads them to underestimate drift volatility, thereby leading to over-
confidence on their part.
The term structure of interest rates provides fertile ground for
applying behavioral asset pricing models. From a neoclassical perspec-
tive, interest rate spreads and movements pose major puzzles. Two
prominent examples are the expectations hypothesis and the credit
spread puzzle for corporate debt. The expectations hypothesis requires
that the expected risk adjusted return to a long default-free bonds
should be equal to the expected return from rolling over short-term risk-
free bonds over the life of the long bond. Realized bond yield volatility
exceeds the upper bound limit implied by the expectations hypothesis
(Shiller, 1978). The credit spread puzzle relates to the fact that the com-
mon factors explaining default probabilities explain at most 25 percent
of the variation in spreads. Moreover, there appears to be a common
latent factor which explains 75 percent of the residual. Bond yields are
more sensitive to macroeconomic announcements (Gurkaynak et al.,
2005; Duffie et al., 2007). In this regard, Equation (5.5) comes to mind,
both as a factor in explaining the failure of the expectations hypothesis
and the existence of the credit spread puzzle.
5.6.1 Behavioral Interest Rate Models
In a general SDF-based framework, Equation (5.1) stipulates that the
price of any security is given by q = E[mx]. For zero coupon risk-free
bonds, x can be taken to be 1, and so bond prices are simply given by
E[m], where the expectation is taken with respect to objective proba-
bilities. Of course, for risky bonds, x = 1.
Equivalently, in a CRRA-framework the price of zero coupon risk-
free bonds can be expressed as q = E[δ(t)g
−γ
], where the expectation is
taken with respect to the representative investor’s pdf (Shefrin, 2008a).
With g
−γ
being a convex function, E[g
−γ
] ≥ E[g]
−γ
. Beginning with
the case of certainty, in which case E[g
−γ
] = E[g]
−γ
. Notice that apply-
ing a mean preserving spread will increase the bond price, thereby low-
ering the associated interest rate.
104 Behavioralizing Asset Pricing Theory
A clue to why the expectations hypothesis fails comes from
understanding that bond prices are determined by the representative
investor’s pdf in q = E[δ(t)g
−γ
], rather than the objective pdf. At the
margin, the representative investor is quite willing to substitute a roll-
over strategy involving short-bonds for a long-bond. However, if sen-
timent is nonzero, then this substitution would be ill advised for a
representative investor with correct beliefs. In other words, the expec-
tations hypothesis might hold subjectively, relative to the beliefs that
underlie equilibrium prices, but not objectively, where it is the objective
probabilities that underlie the dynamics of bond returns and yields.
One thing that distinguishes a behavioral theory of bond pricing
from a neoclassical theory is the nature of the SDF. A neoclassical
SDF is monotone decreasing and stable over time, whereas a behavioral
SDF oscillates and is prone to being volatile over time. In a behavioral
approach, Equation (5.5) captures how nonzero sentiment leads bond
returns to feature both fundamental and sentiment-based components.
Consider an example of a behavioral term structure theory as devel-
oped in Shefrin (2008a). The example features log-utility investors with
heterogeneous beliefs. As was mentioned above, the equilibrium market
pdf is a generalized weighted H¨ older average of the individual investors’
pdfs, with the weights reflecting relative wealth or consumption, and
the exponents being coefficients of relative risk aversion. In the special
case of log-utility, the H¨ older average is a simple convex combination.
As a result, the SDF will also have the form of a convex combination.
This is very convenient, as it leads market prices to be expressed as
convex combinations of prices that would occur in a series of hypo-
thetical homogeneous belief models. In particular, bond prices will be
convex combinations of bond prices that would occur in a series of
homogeneous belief examples.
The preceding section about aggregation and the shape of sentiment
discusses an example with just two investors, one bullish and the other
bearish. In this two-investor example, homogeneous beliefs which are
independent and identically distributed (i.i.d.) lead the yield curve to
be flat. However, heterogeneous beliefs lead the yield curve to be down-
ward sloping, a consequence of yield being a convex function of bond
price. In the homogeneous case, bond prices decline with maturity to
5.6 Pricing Dynamics 105
preserve constant yield. Notably, the decline is larger for the higher
yield. Therefore, as the maturity increases, consider what happens to
the ratio of the bond price for the investor associated with lower yield
to the bond price for the investor associated with the higher yield. This
ratio grows with maturity. Hence, the wealth weighted average convex
combination of bond prices at different maturities will lead to the lower
yield being emphasized. The result will be a declining yield curve.
5.6.1.1 Modeling Interest Rate Dynamics
Consider how the style of models developed in Scheinkman and Xiong
(2003) and Dumas et al. (2009) provide insight into the dynamics of
interest rates. In a promising approach, Xiong and Yan (2009) extend
the log-utility example from Shefrin (2008a) from a discrete exchange
economy to a continuous production economy. Just as Shefrin (2008a)
suggests that the two investor log-utility heterogeneity model can cap-
ture the oscillating shape displayed by the empirical SDF, Xiong and
Yan (2009) suggest that a two-investor log-utility model can capture
important features of interest rate dynamics. Notably, they confirm the
failure of the expectations hypothesis. They also confirm the tendency
for wealth fluctuations to amplify volatility and time variation in bond
prices and premia. Most importantly, they offer an explanation for a
tent-shaped linear combination of forward rates to explain bond return
dynamics.
At the heart of the dynamics in Xiong and Yan (2009) are a set of
assumptions about consumption growth and investment. The returns
to investment dI
t
/I
t
follow a diffusion process with instantaneous drift
f
t
, Brownian motion Z
I
(t), and standard deviation σ
I
. The equation
governing dI
t
/I
t
is:
dI
t
I
t
= f
t
dt + σ
I
dZ
I
(t). (5.8)
Because Equation (5.8) is a linear production process relating to the
instantaneous marginal rate of transformation between current and
future consumption, Xiong and Yan (2009) show that it effectively
determines the instantaneous risk-free rate of interest r
t
= f
t
− σ
2
I
.
106 Behavioralizing Asset Pricing Theory
However, f
t
itself is determined by a mean reverting diffusion pro-
cess with a tendency to revert to an unobservable long-run mean l
t
according to:
df
t
= −λf(f
t
− l
t
)dt + σ
f
dZ
f
(t), (5.9)
where λ
f
governs the mean reverting speed of f
t
. Finally, l
t
follows an
Ornstein–Uhlenbeck process:
dl
t
= −λ
l
(l
t

¯
l)dt + σ
l
dZ
l
(t), (5.10)
where λ
l
governs the mean reverting speed of l
t
. The variables in these
equations that are not identified relate to the associated Brownian
motions and their coefficients.
Together, Equations (5.8)–(5.10) imply:
dr
t
= −λ
f
(t)[r
t
− (l
t
− σ
2
I
)]dt + σ
f
dZ
f
. (5.11)
Equation (5.14) is common in the bond pricing literature (Balduzzi
et al., 1998).
Equations (5.8)–(5.10) provide the basis for the objective pro-
cesses governing the evolution of aggregate consumption growth. These
are conditional on publicly available information at t represented by
{f
τ
}
t
τ=0
, and determine a process for a conditional objective density for
{l
t
|{f
τ
}
t
τ=0
}. The two investors in Xiong and Yan (2009) hold biased
beliefs. They observe a white noise signal dS
t
= dZ
S
(t) which they mis-
takenly believe to be informative about the fundamental shock l
t
. More-
over, the two investors’ reactions to dS
t
are equal and opposite, so that
one investor acts bullishly and the other bearishly. Each investor’s sub-
jective estimate of l
t
at t is conditioned on {f
τ
, S(τ)}
t
τ=0
. Denote an
estimate of l
t
by
ˆ
l
t
.
Xiong and Yan (2009) establish that in equilibrium featuring homo-
geneity, the bond price at time t of a zero coupon default-free bond with
maturity τ has the form:
B
H

τ, f
t
,
ˆ
l
t

= e
−a
f
(τ)ft−a
l
(τ)
ˆ
lt−b(τ)
. (5.12)
The corresponding yield is given by:
Y
H

τ, f
t
,
ˆ
l
t

= −
1
τ
log(B
H
). (5.13)
5.6 Pricing Dynamics 107
In terms of yield, the loading factor on f
t
has a value of 1 for maturity
τ = 0, which monotonically decreases to zero with maturity length.
The loading factor on the estimate
ˆ
l
t
turns out to be hump shaped,
achieving a maximum impact at intermediate maturities, and plays a
central role in the results of Xiong and Yan (2009). The yields of bonds
with very long maturities have low exposure to
ˆ
l
t
, as l
t
reverts to
¯
l in
the long-run.
As in Shefrin (2008a), heterogeneous beliefs lead equilibrium bond
prices in Xiong and Yan (2009) to be wealth-weighted convex combi-
nations of bond prices in homogeneous belief economies. Wealth shifts
from trading based on differential beliefs lead to excess volatility in
bond prices, a feature emphasized in Shefrin and Statman (1994). The
SDF in Xiong and Yan (2009) follows a process dm
t
/m
t
which decom-
poses into a fundamental component and a sentiment component. The
fundamental component is −r
t
dt − σ
I
dZ
I
. The sentiment component is
proportional to the difference between the objective estimate
ˆ
l
t
and the
representative investor’s estimate formed as a wealth-weighted convex
combination of the individual investors’ estimates. The stochastic term
in the sentiment component is the Brownian motion d
ˆ
Z
R
f
underlying
the objective estimate
ˆ
l
t
.
The equation for dm
t
/m
t
indicates the percentage change in the
SDF over the next interval of length dt. Because the instantaneous
interest rate is given by the expected value of m, consumption growth
fundamentals will lead the log-SDF to decline at rate r plus a random
shock to aggregate consumption growth. Sentiment introduces a sec-
ond shock, based on the magnitude of the error in the representative
investor’s estimate of l
t
relative to the statistically optimal estimate. If
the representative investor’s estimate is too low and d
ˆ
Z
R
f
> 0, then the
sentiment component will exert upward pressure on dm
t
/m
t
, thereby
counteracting the tendency of state prices to fall with time.
An important contribution of Xiong and Yan (2009) is to ana-
lyze the relationship between bond maturity and return volatility.
They show that in a homogeneous belief economy, the relationship
is monotone decreasing. However, in a heterogeneous belief econ-
omy, the relationship is hump shaped, similar to the empirical results
in Dai and Singleton (2003). Xiong and Yan (2009) explain that the
108 Behavioralizing Asset Pricing Theory
reason for the different shape in the heterogeneous beliefs case, rela-
tive to the homogeneous beliefs case, lies with the factor weightings in
Equations (5.12) and (5.13). Essentially, the weighting on the f
t
fac-
tor dominates the weighting on the
ˆ
l
t
in the homogeneous beliefs case,
whereas wealth shift dynamics can induce sufficient volatility in
ˆ
l
t
to
inject the hump.
One way to think about the expectations hypothesis is through the
following equation developed in Campbell and Shiller (1991).
Y
t+1
(n − 1) − Y
t
(n) = α
n
+ β
n
Y
t
(n) − Y
t
(1)
n − 1
. (5.14)
When the expectations hypothesis holds β
n
= 1. This is because when
Y
t
(n) − Y
t
(1) > 0, there will be an advantage to investing in the long-
bond, unless its price falls sufficiently, thereby driving up its yield.
However, the empirical evidence indicates that β
n
is close to zero and
positive for 2-month maturities and then declines to about −4 for 10-
year maturities. Therefore, in practice, when the yield spread is posi-
tive, long-term yields tend to rise, not fall.
A behavioral explanation for this finding is as follows. Suppose that
the representative investor’s forecast of future short rates is biased
upward. In this case, the market will excessively discount long bonds,
thereby unduly driving down their prices. Therefore, yields on current
long-term bonds rise. Because the instantaneous short-term interest
rate is determined by the production sector in Xiong and Yan (2009),
the spread will also widen. However, the long-term bond is under-
priced relative to fundamentals. Therefore over time, the price can be
expected to rise, thereby driving down future yields, in contradiction of
the expectations hypothesis. The asset pricing literature also attributes
the failure of the expectations hypothesis to time varying risk aversion.
As discussed above, market aggregation implies that wealth shifts can
lead to time varying risk aversion. It may well be that the failure of the
expectations hypothesis is driven by both features.
Cochrane and Piazzesi (2005) indicate that a tent-shaped function
of three forward rates corresponding to maturities of one, three and five
years appears to forecast holding period bond returns of all maturities
better than the maturity specific forward spreads. Using simulation
5.6 Pricing Dynamics 109
analysis, Xiong and Yan (2009) use their model to advance an inter-
esting explanation for the tent-shaped factor. They note that yields to
intermediate maturities display the greatest sensitivity to beliefs about
l
t
. In their parametric specification, the greatest sensitivity occurs at
a maturity length of three years, making the three-year forward rate
more sensitive than the one year or five year. When wealth shifts to
the bullish investor, the representative investor’s estimate
ˆ
l
t
increases,
thereby increasing the impact of the three-year forward rate relative
to the other two. As a result, the value of the tent-factor increases,
because of the disproportionate weight on the three-year forward rate.
A higher value for the tent-shaped factor predicts a higher bond return.
The higher bond return occurs because, as described above, the bullish
investor’s beliefs unduly drives down the current price, which mean
reverts in the future.
5.6.2 Behavioral Option Pricing
Shefrin (2008a) shows that the linear properties discussed in connec-
tion with bond prices also apply to option prices. In particular, Shefrin
(2008a) presents a specific example involving heterogenous beliefs in
which equilibrium option prices can be represented as wealth-weighted
convex combinations of Black–Scholes prices that would apply in homo-
geneous belief economies.
One of the major empirical features associated with option pricing is
the option “smirk” or “smile,” which refers to the shape of the implied
volatility function (IVF). Given a set of options which vary only by
exercise price, the IVF maps the exercise price to the implied volatility
derived from applying the Black–Scholes option formula to the market
price. If the market treated the return of the underlying security as
following a Brownian motion, then the IVF would be flat instead of
decreasing (the smirk) or U-shaped (the smile). Notably, the option
smirk in the IVF occurs in conjunction with an oscillating SDF, where
there is disagreement about investors about the volatility of the asset
underlying the options.
Heterogeneity typically prevents equilibrium option prices from
satisfying Black–Scholes because heterogeneity leads the risk neutral
110 Behavioralizing Asset Pricing Theory
density to be non-normal. In the continuous time example presented
in Shefrin (2008a), the interest rate and volatility that serve as inputs
into the Black–Scholes formula are correct. When equilibrium option
prices do not satisfy Black–Scholes, it is not because the input values
are necessarily incorrect. Black–Scholes fails because the risk-neutral
density is non-normal, even though the objective return distribution
for the asset underlying the option is normal.
Black–Scholes intuition has developed around partial equilibrium
models. In these models, the risk-neutral pdf is derived from an objec-
tive normal pdf as a change of measure. In this case, the risk-neutral
pdf is itself normal with the same volatility as the objective pdf. In
the behavioral framework, the risk-neutral pdf is obtained, not from
the objective pdf, but from the representative investor’s pdf. In this
regard, heterogeneity typically leads the representative investor’s pdf
to be non-normal, even when the objective density is normal.
Smile effects occur in the implied volatility function (IVF) because
equilibrium option prices do not obey Black–Scholes. In the continuous
time example described in Shefrin (2008a), all investors have common
knowledge of the objective volatility, which makes it possible to use
that volatility to compute the Black–Scholes value associated with any
option. By way of contrast, the IVF backs out volatilities implied by
the risk-neutral pdf based on the representative investor’s pdf. In the
example portrayed in Shefrin (2008a), the representative investor’s pdf
features greater volatility than the common knowledge volatility, lead-
ing the IVF function to lie above its common knowledge counterpart.
The analysis easily extends to the case when investors differ in their
views about volatility, although in this case there is a question as to
which volatility to use in the Black–Scholes formula. One possibility is
to use the volatility associated with the representative investor’s pdf.
However, the analysis is little changed if the objective volatility were
used, or a convex combination of the investors’ subjective volatilities.
As I mentioned above, the change of measure used to derive risk-
neutral densities from objective densities is based on a behavioral SDF
rather than a neoclassical SDF. Therefore, the behavioral risk-neutral
density need not feature the same volatility as the objective density, as
is the case in neoclassical models.
5.6 Pricing Dynamics 111
Much of the empirical work estimating the SDF and market risk
aversion is based on index option prices. In this respect, option smiles
and smirks and the oscillating shape of the empirical SDF represent dif-
ferent sides of the same coin. To the extent that the sentiment compo-
nent of the SDF is germane, the same will be true of index option prices.
In this regard, Han (2008) finds that changes in sentiment explain time
variation in the slope of index option smile which traditional factors
cannot. He reports that when market sentiment becomes more bearish,
the index IVF smile is steeper and the risk-neutral skewness of monthly
index returns is more negative .
Constantinides et al. (2009) examine possible mispricing of SPX
options for the periods 1988–1995 and 1997–2002. Their analysis con-
tains some novel elements, specifically the inclusion of transaction costs.
Notably, they ask whether market prices were such that in any month
of their study, an investor with objectively correct beliefs could have
increased his expected utility by engaging in a zero-net-cost trade. They
find that there were many months when this would have been possi-
ble, thereby leading them to conclude that SPX options were mispriced
over time. In particular, they note that mispricing was pronounced for
out-of-the-money call options. Interestingly, they suggest that before
the crash of 1987, option traders might have navely used the Black–
Scholes formula when it did not apply. Doing so corresponds to the use
of the formula as a heuristic.
The analysis in Constantinides et al. (2009) implies that the pricing
of SPX options cannot be explained in terms of a neoclassical repre-
sentative investor holding correct beliefs. This analysis neither makes
explicit use of sentiment, nor takes into account how sentiment injects
oscillation into the shape of the SDF. Nevertheless, the authors implic-
itly acknowledge sentiment-based explanations for their results.
6
Behavioralizing Corporate Finance
Behavioral corporate finance can be thought of as the intersection of
finance and management Gaston-Breton and Lejarraga (2009). The
groundwork for behavioral corporate finance was laid out in a series of
papers Meir Statman and a series of coauthors wrote about project deci-
sions. See Statman and Tyebjee (1985), Statman and Caldwell (1987),
and Statman and Sepe (1989). In Shefrin (2001), I suggested that
enough was known about the behavioral aspects of corporate finance
to identify a subfield one could label “behavioral corporate finance.”
Other early contributions included Heaton (2002) and Gervais et al.
(2003). Heaton (2002) explored the impact of managerial overconfi-
dence on spending and finance decisions. Gervais et al. (2003) pointed
out that overconfidence can exert a positive impact on the value of the
firm by mitigating agency conflicts, a point also developed in Goel and
Thakor (2008) and Hackbarth (2009).
1
1
The framework developed in Hackbarth (2009) is contingent claims based, and should
prove especially useful for integrating behavioral corporate issues into a general asset
pricing model.
112
6.1 Open Questions Raised by Baker et al. 113
6.1 Open Questions Raised by Baker et al.
In this section, I take as my starting point the behavioral corporate
framework developed in Baker et al. (2007b) discussed in Section 2
above. That framework goes a long way toward behavioralizing cor-
porate finance. Baker et al. (2007b) conclude with a series of open
questions about the future direction of behavioral corporate finance.
I wrote my behavioral corporate textbook (Shefrin, 2005) at the
same time that Baker et al. were writing their survey. I intended my text
for classroom instruction, in order to provide a behavioral supplement
to the traditional material taught in corporate finance courses. In many
ways, the textbook approach parallels the approach in Baker et al.
(2007b). In this regard, I call criterion functions such as Equations (2.1)
and (2.4) “behavioral adjusted present value.” The text offers a series
of cases to make the material concrete for students. I suggest that some
of the case discussions provide insight into a list of open questions that
conclude Baker et al. (2007b).
Debt Puzzle
The first open question in Baker et al. (2007b) asks why managers do
not pursue the tax benefits of debt more aggressively, an issue raised
by Graham (2000). In Shefrin (2005) I present a case about the phar-
maceutical firm Merck, suggesting that the loss aversion property in
prospect theory offers a potential explanation, as bankruptcy losses
loom much larger than the gains associated with tax shield benefits.
In private correspondence, Graham has indicated that he concurs, sug-
gesting that managers do not want to get too close to the edge of the
cliff.
Irrational Managers and Inefficient Prices
The second open question asks whether a model that incorporates both
irrational managers and inefficient prices would lead to new insights.
There is no reason why irrational managers should not balance per-
ceived short-term gains against perceived long-term value. Theoret-
ically, it is straightforward to combine Equations (2.1) and (2.4) in
114 Behavioralizing Corporate Finance
order to capture both phenomena. One way to do this is to modify
Equation (2.1), replacing output f(K; ) with (1 + γ)f(K; ) from Equa-
tion (2.4), and changing the base from which the mispricing term δ(·)
is measured from 0 to −γf(K; ). Notably, mispricing is now understood
in the sense of “perceived” mispricing. These substitutions lead to the
criterion function:
λ[(1 + γ)f(K; ) − K + (e(δ(·) − γf(K; )))] + (1 − λ)[δ(·) − γf(K; )].
(6.1)
As to how the combination of irrational managers and inefficient
prices might be germane, in Shefrin (2005) I present a case about the
acquisition of Time-Warner by AOL, one of the largest in history. I sug-
gest that the financial features of this case reflect both irrational man-
agers and inefficient prices. In line with the Shleifer–Vishny framework,
the executives at AOL did use overvalued equity to purchase Time-
Warner. However, Time-Warner’s CEO Gerald Levin was not skeptical
about mispricing, in contrast to the general evidence presented by Mal-
mendier and Tate (2008).
Debiasing
One of the most important open questions involves debiasing, how cor-
porations can learn to reduce susceptibility to managerial pitfalls? In
Shefrin (2005), I devote considerable space to this issue, including a
section focusing on the role psychological phenomena play in business
process. To be sure, psychological pitfalls might well lead managers
to make decisions that are not in the best interests of shareholders.
From a neoclassical perspective, this represents an agency conflict, a
concept that is already central to corporate finance. However, because
the conflict is not between rational principals and rational agents, the
usual incentive remedies for traditional agency conflicts may not apply.
Indeed there is already a literature in addressing behavioral bias within
organizations (Heath et al., 1998).
Shefrin (2005) emphasizes that behavioral phenomena and agency
phenomena are both central to corporate finance. There are two main
issues involving the relationship between agency phenomena and behav-
ioral phenomena. First, a key challenge is to differentiate them clearly,
6.1 Open Questions Raised by Baker et al. 115
so behavioral issues do not come to be regarded as rational agency
issues. Second, there is a whole behavioral dimension to the treatment
of agency phenomena. In Shefrin (2005) I devote a section to discussing
interactions between nonrational principals and nonrational agents.
Shefrin (2005) is organized to identify how psychological pitfalls cre-
ate gaps between recommended courses of action in corporate finance
textbooks and what corporate managers actually do. Topics include
the spectrum from valuation analysis to real options. The literature
on these gaps continues to grow. For example, Tiwana et al. (2007)
document a series of cases in which managers make imperfect use of
real option analysis. In particular, managers are apt to downplay real
option analysis for high NPV projects relative to low NPV projects,
thereby underestimating the value of the latter.
Themes in the Remainder of the Section
In the body of this section, I focus on three main issues pertaining to
the behavioralization of corporate finance. All relate to delving more
deeply into understanding the impact of managers’ personality charac-
teristics on their decisions. The first is based on the analysis of surveys
to identify the impact of managerial optimism decisions about debt
ratios, debt maturity, dividend policy, and acquisition frequency. The
second is a study about the impact of social networks to which CEOs
belong, especially when it comes to vulnerability to groupthink. The
third is evidence about the psychological traits of entrepreneurs, which
builds on the discussion in Baker et al. (2007b).
The three themes upon which I focus are not exhaustive. There
have been numerous new papers in behavioral corporate finance that
build on earlier ideas. Some examples are as follows: Degeorge et al.
(2007) document that managers tend to rely on bookbuilding to go
public, which is less cost effective than auction techniques. Huang and
Ritter (2009) analyze the empirical issues associated with achieving tar-
get capital structures. Spalt (2008) studies the degree to which firms
use option-based compensation to exploit managerial optimism. Peyer
and Vermaelen (2009) establish that the buyback puzzle has not disap-
peared. Campbell et al. (2009) use turnover rates to test and confirm
116 Behavioralizing Corporate Finance
the predictions in Goel and Thakor (2008) that there is an optimal
degree of overconfidence to balance agency conflicts.
6.2 Managerial Psychology and Decisions
6.2.1 Results Based on FEI Survey
Ben-David et al. (2007) provide interesting insights into the connection
between overconfidence and managerial decisions. Their study makes
use of a survey Duke University andCFO magazine jointly run of the
members of Financial Executives International (FEI), an association of
top financial executives. The survey features a unique panel of nearly
7,000 observations of managers’ probability distributions regarding the
stock market. Survey results show the financial executives to be mis-
calibrated: Over a six-year period, realized market returns are within
the executives’ 80 percent confidence intervals only 38 percent of the
time.
Given that raising external capital is costly, firms whose projects
have high NPV can do their shareholders a favor by keeping their divi-
dend payouts low, and using the cash to fund projects instead. After all,
high NPV means that the company is able to generate higher expected
returns through its projects than investors can earn if instead paid
the cash. Overconfidence leads executives to underestimate the risk
attached to project cash flows, resulting in overestimates of projects’
NPV. As a result, the greater the degree of overconfidence, the more
inclined are executives to adopt projects, and undertake acquisitions.
Overconfidence can also impact capital structure. Think about unre-
alistically optimistic, overconfident managers seeking to balance the tax
shield benefits of debt against the expected costs of financial distress,
especially bankruptcy. They will be inclined to underestimate the like-
lihood of going into bankruptcy, and therefore to underestimate the
expected cost bankruptcy. Therefore, overconfident managers will be
prone to take on more debt than is appropriate.
The general evidence from the Ben-David et al. (2007) is that the
greater the degree of unrealistic optimism and overconfidence on the
part of executives, the more prone they are to take on excessive debt. In
the Duke/FEI survey, the average debt ratio was 23 percent. Imagine
6.2 Managerial Psychology and Decisions 117
dividing executives into ten groups according to their overconfidence.
Suppose that for each group, we compute the average debt ratio of the
company that the executive works for. Ben-David et al. (2007) report
that as we migrate from the group having the lowest overconfidence to
the group having the highest overconfidence, the debt ratio increases
by 0.5 percent every time we move up to a new group.
In addition, overconfident executives tend to take out debt with
longer maturities. The average maturity for the survey firms was 3.7
years. However, the group with the highest overconfidence rating took
out longer loans than the group with the lowest overconfidence rating,
by about one year. The longer maturity leaves a company with less
financial flexibility, and more interest rate risk.
6.2.2 Results Based on Personality Tests
Graham et al. (2009) may well be the first study which measures atti-
tudes of senior management directly through personality tests and then
relates the results to firm level policies such as the usage of debt and
mergers and acquisitions. In 2006, the authors surveyed both CEOs
and CFOs. Most of the executives surveyed were identified as readers
of Chief Executive, CFO magazine, and for CFOs the publications CFO
Europe and CFO Asia magazines. The responses generated a database
with 1,180 CEOs with 1,017 CEOs who work for firms headquartered in
the United States. For CFOs, there were correspondingly 549 responses.
In their survey, Graham et al. (2009) elicit responses to indicate
degree of risk aversion, dispositional optimism, time preference, and
aversion to a sure loss. To assess risk aversion, the authors use a pro-
cedure developed by Barsky et al. (1997). To assess optimism, they
administer a standard psychometric test known as Scheier and Carvers
Life Orientation Test Revised or LOT-R test. To measure time pref-
erences, they assess time predilection for gains and losses. To gauge
aversion to a sure loss, they present managers with a gamble that, if
taken, indicates that the survey respondent is averse to a sure loss. In
addition, Graham et al. (2009) elicit other measures that be related,
such as height as a measure of confidence, age, tenure, career track,
education, and prestige of the executives undergraduate college.
118 Behavioralizing Corporate Finance
In respect to debt, Graham et al. (2009) report that CEOs with a
background in finance and accounting tend to take on more debt than
CEO with other backgrounds. This might be a reflection of the avail-
ability heuristic, as managers with formal backgrounds in finance and
accounting will be more familiar with the advantages of debt. Interest-
ingly, males, optimists, and executives from private companies are more
likely to use a higher proportion of short term debt than others. See
Landier and Thesmar (2009) whose theory suggests that optimists are
more likely to take on short-term debt. If male gender is positively cor-
related with overconfidence, as suggested by Barber and Odean (2000),
then overconfidence appears to lead to more usage of short-term debt.
Graham et al. (2009) find that CEOs who are more risk tolerant are
also more likely to make acquisitions, as are CEOs who are averse to a
sure loss. There may be several explanations for the second finding. In
Shefrin (2005) I suggest that the managers of poorly performing firms,
who register subpar performance as a psychological loss, are prone to
engage in risky acquisition activity. Graham et al. (2009) point out
that managers who are averse to a sure loss are more prone to engage
in multiple acquisitions, perhaps because they keep bidding until
they win.
Interpreting height as a measure of confidence, the results in Gra-
ham et al. (2009) suggest a matching story whereby young, confident,
risk-tolerant CEOs are more likely to work for high growth compa-
nies and companies with high anticipated growth rates. In terms of
relying on incentive-based compensation, three personal characteristics
appear to be significant: risk-taking, gender, and rate of time prefer-
ence. Graham et al. (2009) find that risk-tolerant CEOs are more likely
to obtain proportionately larger remuneration through incentives such
as stock, bonus payments, and option grants relative to a large salary.
In this regard, Spalt (2008) reports a complementary finding, that some
firms use stock option pay for lower-level employees to take advantage
of biased probability judgments or weightings. Not surprisingly, males
are more likely to be compensated through incentives, as are executives
who are more patient. Moreover, Graham et al. (2009) hypothesize that
firms set compensation policies to hire managers who, over time, will
perpetuate corporate policies, such as capital structure.
6.3 Social Networks 119
An intriguing finding of Graham et al. (2009) involves the bench-
marking of the dispositional optimism of U.S. CEOs against others. On
a 0–4 range, the average response for the general population is about
2.5. However, more than 80 percent of CEOs’ responses to the opti-
mism questions average about 3, so that CEOs are significantly more
optimistic than the lay population.
CEOs are also more optimistic than CFOs, with only 66 percent of
CFOs being classified as very optimistic. Interestingly, when asked to
rate CEO optimism relative to their own, CFOs confirm this finding.
CFO suggests that CEOs are more optimistic about all aspects of life,
not just business prospects. There are other ways CEOs differ from
CFOs. CEOs are less likely to have MBA degrees than CFOs. CEOs
are likely to be older, have longer tenure, and to have attended more
prestigious universities. Strikingly, CFOs are more averse to sure losses
and more patient than CEOs.
Finally, Graham et al. (2009) report that both CEOs and CFOs
from the United States are more risk tolerant than the CEOs and CFOs
of 25 companies which are not located in the United States. In addition,
non-U.S. CEOs and CFOs are less patient than their U.S. counterparts.
Foreign CEOs also have a higher aversion to sure losses than do U.S.
CEOs.
6.3 Social Networks
Fracassi and Tate (2009) test whether connections, in the sense of social
networks, between management and potential directors influence direc-
tor selection and subsequent firm performance.
2
Their study uses panel
data for firms in the S&P 1500, thereby focusing on social ties in large
U.S. corporations. In this respect, Fracassi and Tate (2009) identify
directors who belong (or belonged) to the same golf clubs, charities,
and other non-professional organizations as the CEO. They consider
employees’ histories, identifying directors who, with the exception of
the current firm, were employed by the same firm as the CEO in the
2
Studies focusing on social networking are gaining interest. See Cai and Sevilir (2009) and
Ishii and Xuan (2009) who study the role of social networks involving directors and/or
managers in M&A.
120 Behavioralizing Corporate Finance
past. Finally, they identify directors who attended the same educational
institutions as the CEO.
In Shefrin (2005), I discuss a series of psychological impediments
to effective governance processes. A key impediment is “groupthink”
(Janis, 1972; B´enabou, 2008). Groupthink is a form of collective confir-
mation bias for groups, whereby there is insufficient dissent within the
group, resulting in the amplification of individual biases. Groupthink
can result when group members are too similar to one another, all shar-
ing similar perspectives. Therefore, close ties between the CEO and the
board, based on similarities in backgrounds and experiences, increase
the potential for groupthink. In this regard, Fracassi and Tate (2009)
point out that directors might be more willing to give the benefit of
the doubt to management when they have a closer relationship with,
and trust in, the CEO.
In Shefrin (2005), I point out that when it comes to heterogeneous
board composition, some boards might look good on paper, yet be very
vulnerable to groupthink. Indeed U.S. exchanges require the major-
ity of directors in listed firms to be independent, and in recent years
have strengthened independence requirements for key board commit-
tees. Alas, this is no guarantee that directors with network connections
to the CEO qualify as independent directors, but are not unbiased
monitors.
Social Connectivity and Firm Value
Fracassi and Tate (2009) find that the relationship between social con-
nectivity and firm value, measured using Tobin’s Q, is negative. More-
over, the effect is strongest when shareholder rights are weak. In the
remainder of this section, I discuss some of the specific issues associated
with the negative relationship.
Consider alternative measures of CEO power such as being both
board chair and president, long tenure, a high entrenchment index,
and being much more highly paid than the next best paid executive.
Fracassi and Tate (2009) report that CEOs who are powerful are prone
to add new directors having existing network ties to the CEO. Such
directors are more likely to buy company stock at the same time as the
6.3 Social Networks 121
CEO. Specifically, relative to other outside directors, directors closely
connected to the CEO through social networks are more likely to pur-
chase stock within five days of a CEO stock purchase. In addition, if
the CEO is a net buyer, then over the fiscal year these directors are
more likely to be net buyers.
More importantly, there is some evidence that the presence of direc-
tors closely connected to the CEO is correlated with weaker monitor-
ing. For example, firms with closely connected directors are less likely
to initiate earnings restatements which are internally prompted. Espe-
cially, interesting is the fact that firms with boards closely connected
to the CEO make more frequent acquisitions. Fracassi and Tate (2009)
state that the merger bids of these firms destroy $407 million of share-
holder value on average, $293 million more than the bids of other firms.
This feature complements the discussion in the prior section about the
characteristics of CEOs of firms who are frequent acquirers.
Not surprisingly, such poor decisions lead to lower overall market
valuations, especially when shareholder rights are weak when it comes
to substituting for board monitoring. Specifically, Fracassi and Tate
(2009) find that the average cumulative abnormal return for the three-
day window surrounding merger announcements is lower for firms with
a higher percentage of connected directors. They also find that the
average value created by the deals (for acquiring shareholders) is nega-
tive, and that the value destruction is concentrated in firms with weak
shareholder rights as measured by Gompers et al. (2003).
Plausibly, acquisition-based value destruction might well reflect a
monitoring failure or an advisory failure on the part of the board. Major
acquisitions might be initiated by the CEO; however, they require
board approval. Directors connected to the CEO might be unwilling
to oppose value-destroying policies that provide private benefits to the
CEO. Alternatively, such directors might withhold information in their
possession that does not support the CEO’s proposal.
Fracassi and Tate (2009) indicate that their analysis does not
provide much insight into the motivation of connected independent
directors. They identify several possibilities. For example, connected
directors might be seeking to expedite the CEO’s agenda, in which case
they are complicit in the associated value destruction. Alternatively,
122 Behavioralizing Corporate Finance
connected directors might be motivated by an interest in preserving
social ties with the CEO. Finally, connected directors might simply
share the CEO’s perspective, or cognitive biases, by dint of having the
same background and moving in the same social circles.
6.4 Entrepreneurs
This section has two goals.
3
The first goal is to broaden the psycho-
logical framework used to characterize managers. The second goal is
to identify the psychological profile of entrepreneurs with the goal of
understanding why they tolerate inferior returns. In this respect, the
discussion about the psychological profile of entrepreneurs will serve as
a vehicle for both goals.
Shefrin (2005) and Baker et al. (2007b) both indicate that among
the many psychological phenomena, behavioral corporate finance
mainly focuses on excessive optimism and overconfidence. In this
respect, I note that the behavioral decision literature does not treat
these concepts as primitives, but identifies factors that explain what
makes people excessively optimistic and overconfident. A key factor is
control. Weinstein (1980) and Flynn et al. (1994) indicate that control
is a determinant of both excessive optimism and overconfidence. Below
I discuss how psychologists measure control, and relate the discussion
to entrepreneurs.
Baker et al. (2007b) devote several paragraphs to entrepreneurs.
This is because entrepreneurs appear to accept inferior returns. Hamil-
ton (2000) establishes that entrepreneurs accept lower median life-
time earnings than similarly skilled wage-earners. Moskowitz and
Vissing-Jorgensen (2002) establish that entrepreneurs earn low risk-
adjusted returns.
4
Moreover, entrepreneurs appear to hold poorly
diversified portfolios. Instead they concentrate their wealth in their
own private business (Gentry and Hubbard, 2001; Moskowitz and
Vissing-Jorgensen, 2002; Heaton and Lucas, 2000). In light of the
3
I draw the material for this section from Shefrin (2010).
4
The average return to all private equity is actually similar to that of the public market
equity index. However, the risk is higher. Survival rates of private firms are only around
34 percent over the first 10 years of the firms life.
6.4 Entrepreneurs 123
evidence, Bitler et al. (2004) investigate the structuring of incen-
tives for entrepreneurs and ask why people would choose to become
entrepreneurs.
Puri and Robinson (2008) provide some insight into why people
would choose to become entrepreneurs. They ask whether entrepreneurs
are risk-takers. They suggest that entrepreneurs might derive substan-
tial non-pecuniary benefits from self-employment. Based on their prior
work (Puri and Robinson, 2007), they hypothesize that entrepreneurs
might be optimistic about their entrepreneurial prospects. Puri and
Robinson’s analysis of the data in the Survey of Consumer Finances
(SCF) leads them to a series of conclusions. First, they find that
entrepreneurs are more risk-loving and optimistic than the rest of the
population. Second, these traits are separable in that the correlation
between risk tolerance and optimism is low. Third, entrepreneurs tend
to have long planning horizons, good health practices, and strong fam-
ily ties. In respect to planning, entrepreneurs are almost three times
more likely to indicate that they never intend to retire. Moreover, peo-
ple who do not plan to retire work about 3 percent longer per week. In
the main, these findings are supported by Wadhwa et al. (2009).
Consider how the configuration of entrepreneurs’ characteristics
relates to the suboptimality of the returns which they earn. Unless
entrepreneurs are actually risk seeking, being more risk-loving than
the general population does not explain why entrepreneurs earn sub-
optimal risk-adjusted returns. Notably, optimism might provide part of
the explanation. Puri and Robinson focus on dispositional optimism as
opposed to unrealistic optimism. They point out that the psychology
literature distinguishes between dispositional optimism and optimistic
bias. The former views optimism as a positive personality trait associ-
ated with positive generalized expectations of the future (Scheier and
Carver, 1985). The latter views optimism as a negative, domain-specific
bias in expectations (Weinstein, 1980).
It is straightforward to see how optimistic bias can lead to inferior
returns. Of course, dispositional optimism can also involve optimistic
bias, and therefore produce inferior returns as well. This is not to say
that the net impact of dispositional optimism is negative. In this regard,
Bitler et al. (2004) find that effort increases firm performance. In their
124 Behavioralizing Corporate Finance
analysis, Puri and Robinson focus on dispositional optimism. They do
not measure optimism bias directly, but infer it from self-reported esti-
mates of own life expectancy. Shefrin (2010) extends their analysis by
surveying entrepreneurs for dispositional optimism and comparing their
responses with the responses of non-entrepreneurs.
While the combination of risk-seeking choices and optimism possibly
explains why entrepreneurs earn low risk-adjusted returns, it seems
far more plausible that non-pecuniary benefits lie at the heart of the
issue. After all, entrepreneurs work longer hours and plan to retire
later than their non-entrepreneur counterparts. What might those non-
pecuniary benefits be? I suggest that a major non-pecuniary benefit for
entrepreneurs is the exercise of control over their working environment,
the control that derives from starting a new business and managing
that business. In this regard, I hypothesize that the desire for control
is a psychological need which might be higher for entrepreneurs than
non-entrepreneurs.
Puri and Robinson document that relative to non-entrepreneurs,
entrepreneurs are more likely to be married and have larger families
(see also Wadhwa et al., 2009). This suggests that entrepreneurs might
be more socially focused than others. I hypothesize that this is the
case. In particular, I test whether entrepreneurs are less anxious in
social settings than non-entrepreneurs, and more aware of social cues
and their own behavior.
If entrepreneurs derive significant non-pecuniary benefits from
entrepreneurial activity, then it seems plausible that entrepreneurs
would report being happier than their non-entrepreneurial counter-
parts. I test whether this is the case using two instruments, one pertain-
ing to well-being (life satisfaction) and the other pertaining to degree
of affect (positive, neutral, or negative). The major findings are as fol-
lows: entrepreneurs exhibit more dispositional optimism than their non-
entrepreneurial counterparts. Notably, the most significant difference
between entrepreneurs and non-entrepreneurs is the desire for control:
entrepreneurs place much more emphasis on control. Entrepreneurs
appear to be more comfortable in social situations, and feel that they
are more sensitive to social cues. They also report being more posi-
tive and more satisfied with their lives. Taken together, these findings
6.4 Entrepreneurs 125
suggest that the non-pecuniary benefits that entrepreneurs experience
are substantial.
6.4.1 Data
The data for this study involve responses to a series of psychological
surveys. The surveys relate to instruments for measuring dispositional
optimism (Scheier and Carver, 1985), desirability of control (Burger
and Cooper, 1979), social anxiousness (Leary, 1983), self-monitoring
behavior (Snyder, 1974; Lennox and Wolfe, 1984), life satisfaction
(Diener et al., 1985), and affect (Watson and Clark, 1988). The surveys
were administered to two different groups: (1) a group of entrepreneurs
and (2) a group of MBA and undergraduate business students (non-
entrepreneurs).
6.4.2 Desire for Control
Puri and Robinson classify respondents to the SCF as entrepreneurs if
they own some or all of at least one privately owned business, and are
full-time self-employed. It seems plausible to suggest that people who fit
this description exert more control over their working environment than
others, and that part of the appeal of being an entrepreneur is achieving
this degree of control. Put somewhat differently, entrepreneurs possess
a strong need for control which leads them to choose a career in which
they seek to meet that need. This statement reflects a view in the
literature on motivation about the importance of nonmonetary rewards
relative to monetary rewards.
Burger and Cooper (1979) introduce a survey instrument (DC)
which is designed to measure the desirability of control. The survey
instrument consists of twenty questions. Two representative questions
are: “I prefer a job where I have a lot of control over what I do and
when I do it,” and “I would prefer to be a leader than a follower.”
The range of possible responses varies from 1 to 7 where 1 means “The
statement does not apply to me at all,” and 7 means “The statement
always applies to me.” For 15 of the questions, 7 is associated with the
strongest desire for control, while for 5 of the questions, 1 is associated
126 Behavioralizing Corporate Finance
with the strongest desire for control. For sake of consistency, the score
for each of the 5 questions is replaced by 7-score.
The results suggest that desire for control is a major non-pecuniary
benefit associated with being an entrepreneur. The average DC score
per question is 5.4 for entrepreneurs and 4.6 for non-entrepreneurs. To
interpret these results, note that 4 means “I am unsure about whether
or not the statement applies to me or it applies to me about half the
time,” 5 means “The statement applies more often than not,” and 6
means “The statement usually applies to me.”
6.4.3 Optimism
Using their life expectancy proxy for optimism, Puri and Robinson
find that entrepreneurs exhibit more dispositional optimism than non-
entrepreneurs. I augment their analysis by using an instrument to mea-
sure dispositional optimism (Scheier and Carver, 1985).
The optimism instrument (OP) consists of eight questions whose
responses fall on a five-point scale where 1 means “strongly disagree”
and 5 means “strongly agree.” Two of the questions are: “I’m always
optimistic about my future,” and “I hardly ever expect things to go
my way.”
Adjusting for signs, the mean OP-score for entrepreneurs was 3.5
and for non-entrepreneurs it was 3.2. Notably, both groups feature
OP-scores between Neutral and Agree, and are therefore optimistic.
Although entrepreneurs appear to be more optimistic than non-
entrepreneurs, the difference in scores is not statistically significant
for the particular sample in my study. However, the results are quite
similar to the larger sample described in Puri and Robinson (2008),
where the differences are statistically significant.
6.4.4 Social Anxiousness and Self-monitoring
Consider psychological attributes which relate to the finding that
entrepreneurs marry at a higher rate than non-entrepreneurs and
have more children. This finding suggests that entrepreneurs like
people. This subsection considers two psychological instruments that
relate to interpersonal relationships. The first stems from the work
6.4 Entrepreneurs 127
of Leary (1983), who studied social anxiousness, the degree to which
people are uncomfortable in social situations. The second relates to
self-monitoring, a concept studied by Snyder (1974) and Lennox and
Wolfe (1984).
The survey instrument used to study social anxiousness (SA) fea-
tures fifteen questions. Two of the questions are: “I often feel nervous
even in casual get-togethers,” and “I get nervous when I speak to some-
one in a position of authority.” Responses are on a 5-point scale where
1 means “Not at all characteristic,” and 5 means “Extremely charac-
teristic.”
A lower SA score signals less social anxiety. Entrepreneurs have a
mean SA score of 2.0, whereas non-entrepreneurs’ mean SA score is
2.9. A score of 2 means “Slightly characteristic” and a score of 3 means
“Moderately characteristic.”
Self-monitoring involves being sensitive to social cues, and being
able to adapt to those cues. The self-monitoring survey instrument is
SM. It features questions such as the following: “I am often able to
read people’s true emotions correctly through their eyes.” “In social
situations, I have the ability to alter my behavior if I feel something
else is called for.” The range of possible responses comprises a 6-point
scale from “Always false” to “Always true.”
The mean score for entrepreneurs is 4.4 and for non-entrepreneurs
is 3.9. Here 3 means “Somewhat false, but with exception,” 4 means
“Somewhat true, but with exception,” and 5 means “Generally true.”
6.4.5 Affect and Well-Being
Are entrepreneurs happier than non-entrepreneurs? What makes this
question especially interesting is that entrepreneurs work more than
non-entrepreneurs and earn lower risk-adjusted returns on their invest-
ments. To investigate the question, I use survey instruments based on
the work of Watson and Clark (1988), who study affect, and Diener
et al. (1985), who study life satisfaction or well-being. I refer to the
associated instruments as AF and WB, respectively.
The affect survey asks respondents to indicate how frequently
they experience a variety of emotional states such as “interested,”
128 Behavioralizing Corporate Finance
“distressed,” and “excited.” Responses are on a 5-point scale, in which
1 means “very slightly or not at all,” and 5 means “extremely.” Some
states connote negative affect and some connote positive affect. An AF
score of zero connotes neutral overall affect. The mean AF-score for
entrepreneurs was 1.2 and the mean AF-score for non-entrepreneurs
was 0.8.
The life satisfaction survey asks respondents twenty-five true/false
questions such as “I always seem to have something pleasant to look
forward to,” and “Often I get irritated at little annoyances.” Of the
25 questions, 11 relate to feelings of positive life satisfaction and 14 to
negative feelings. Relative to an adjusted neutral score of 0, the mean
WB-score for entrepreneurs was −1.0 and for non-entrepreneurs was
−1.2. Entrepreneurs score higher on the WB-survey. However, both
groups’ mean scores are negative responses. Overall, the evidence indi-
cates that entrepreneurs are happier than non-entrepreneurs.
6.4.6 Risk Preferences
Puri and Robinson (2008) find that entrepreneurs are more risk-loving
and optimistic than the rest of the population, although the correla-
tion between risk tolerance and optimism is low. The analysis suggests
that desire for control is a major motivator for why people choose
to be entrepreneurs, stronger than optimism, and perhaps risk atti-
tude as well. In terms of risk, it is possible that people choose to be
entrepreneurs because they have a preference for lottery-type returns.
To test for this possibility, I ran a supplementary survey, based on the
concepts in SP/A theory (see Section 3) which posed the following two
questions to entrepreneurs:
1. On a scale of 1 to 7 where 1 is unimportant and 7 is extremely
important, how important to you is upside potential as a
reason being an entrepreneur?
2. On a scale of 1 to 7 where 1 is unimportant and 7 is extremely
important, how important to you is being in control of your
work world as a reason being an entrepreneur?
6.4 Entrepreneurs 129
The mean response to the first question was 5.8 and to the second
was 6.3. Clearly, both factors are important. However, between the
two, control is more important.
The supplementary survey also posed qualitative questions designed
to elicit features emphasized by SP/A theory, such as the strength of
emotions such as fear and hope, and whether entrepreneurs establish
specific aspirational goals. These respectively correspond to the three
main variables in the theory, S, P, and A. The responses indicate that
entrepreneurs feel that fear is a weak emotion for them, hope is a mod-
erately strong emotion, and aspirational goals are strong. Many indicate
that they would accept lottery-type returns if the downside is small, but
are more reluctant if the downside is not small, if doing so significantly
diminishes the probability of achieving their specific aspirations.
7
Behavioralizing the Approach to Financial
Market Regulation
The behavioral aspect of financial market regulation is a particularly
rich area of study. A major reason why is its complexity, in that it
combines behavioral aspects of portfolio selection, derivatives, asset
pricing, corporate finance, and political economy. Yet, until the global
financial crisis erupted in 2008, financial economists paid little attention
to financial market regulation in general and its behavioral aspects in
particular.
To the best of my knowledge, the first behavioral treatment of finan-
cial market regulation is found in Shefrin and Statman (1992, 1993b).
The literature then lay dormant until Hirshleifer (2008b), who was
unaware of our earlier work, and addressed some of the same issues.
Statman (2009) and Shefrin and Statman (2009) describe how their
framework applies to a series of issues preceding and including the
financial crisis. The series includes credit card and bank regulations,
the evolution of insider trading regulations, Regulation FD, trading
halts, the Global Settlement, and the Sarbanes–Oxley Act.
In this section, I review the basic behavioral approach to financial
market regulation, with reference to both historical and recent financial
crises.
130
7.1 Dynamic Tug-of-War 131
7.1 Dynamic Tug-of-War
7.1.1 Fairness and Efficiency
The regulation of financial markets in the United States can only be
understood in reference to the continuous debate that has shaped it.
In this respect, financial market regulation reflects concerns for both
efficiency and fairness, a point emphasized by Shefrin and Statman
(1992).
1
Regulations are the outcome of a dynamic tug-of-war between
parties with differing views about efficiency and fairness, in which rel-
ative strength continually shifts from side to side.
There are two notions of efficiency in financial economics. The first
is Pareto efficiency and the second is informational efficiency. An allo-
cation is said to be Pareto efficient when no other feasible allocation of
resources and technology can improve one person’s situation without
harming another’s. By its nature, Pareto efficiency precludes the wast-
ing of resources. It also ensures that the riskiness of investment projects
are in line with the attitudes of investors toward risk. Notably, Pareto
efficiency is consistent with errors and biases. In a Pareto-efficient
allocation, investors might well take on more risk than is objectively
appropriate or overreact to information and cause security prices to be
excessively volatile.
Informational efficiency is achieved when all investors share objec-
tive beliefs and information so that competitive prices accurately
reflect that information. Efficient prices provide proper guidance to
entrepreneurs, managers, and investors. By observing prices, managers
can avoid projects with negative net present value, and investors can
select efficient portfolios. However, some managers and investors might
misinterpret the information conveyed within efficient prices and make
supoptimal choices. Asymmetric information destroys informational
efficiency and moral hazard interferes with it.
Shefrin and Statman (1992) build on Kahneman et al. (1991), who
provided the first behavioral characterization of fairness in terms of
“reference transactions.” Shefrin and Statman (1992) propose that in
financial markets fairness is understood as entitlements to particular
1
This section is drawn from some of the main ideas in Shefrin and Statman (1992, 1993b).
132 Behavioralizing the Approach to Financial Market Regulation
classes of transactions. They identify seven classes of fairness by the
following entitlements:
— A transaction is fair when the parties enter into it voluntarily.
Freedom from coercion entitles people to avoid being coerced
into a transaction against their will, and being permitted to
engage in a transaction of their choosing.
— Freedom from misrepresentation entitles people to rely on vol-
untarily disclosed information being accurate.
— Fairness in equal information entitles people to equal access to
a particular set of information.
— Fairness in equal processing power entitles people not only to
equal access to information but also to a “competency floor”
of information processing skills. This class of fairness recognizes
that some people commit cognitive errors. Protecting them may
take the form of compulsory disclosure or the outright prohibi-
tion of certain types of transactions.
— Fairness as freedom from impulse entitles people to protection
from possible imperfect self-control.
— Fairness in efficient prices entitles people to prices they perceive
to be efficient. This is the notion of fairness inherent in the term
“fair and orderly market.”
— Fairness in equal bargaining power entitles people to equal
power in negotiations leading to a transaction. Bargaining
power inequality can occur, for example, when one party to
a transaction is wealthier than the other.
Policymakers operate as if they have preferences which depend on
both efficiency and fairness. Some combinations of efficiency and fair-
ness dominate other combinations that are higher in one or the other
attribute. A conceptual efficiency/fairness frontier is composed of com-
binations that are not dominated. Choices along the frontier involve
reductions in fairness to increase fairness or reductions in fairness to
increase efficiency.
7.1 Dynamic Tug-of-War 133
7.1.2 Capture Theory
Shefrin and Statman (2009) augment the framework described above
to reflect capture theory. This theory postulates that interest groups
regularly enlist politicians and regulators in their battles with one
another over regulatory proposals. See Stigler (1971) who noted that
each interest group — bankers, lawyers, union members, and employ-
ers — wants regulations that maximize its wealth. Politicians have the
power to direct regulators to benefit one interest group or another.
At the same time, politicians need resources such as campaign contri-
butions to maximize their chances of re-election. Similarly, regulators
want to steer the regulatory process in directions that benefit them, in
prestige or industry jobs once they leave public service. The political
process involves competition among interest groups each attempting to
capture politicians and regulators by some combination of votes, contri-
butions, and favors in exchange for enacting and executing regulations
which transfer wealth to them.
Stigler emphasized that an interest group is likely to capture its
regulators when the per-capita benefits to the members of the inter-
est group are large relative to per-capita benefits to the general public.
Peltzman (1976) augmented capture theory, noting that interest groups
would not capture their regulators when the total benefits to the general
public are sufficiently large, even if the per-capita benefits are relatively
small. Politicians and regulators who allow interest groups to capture
them under such circumstances might lose more political support than
they gain. Politicians and regulators have limited power to tilt regula-
tion toward interest groups and their power varies by the environment
in which they operate. Economic booms and rising financial markets
placate the general public, reducing its vigilance and making it eas-
ier for politicians to tilt regulations toward interest groups. However,
recessions and declining financial markets enrage the general public,
increasing its vigilance and its clamor for regulatory protection from
interest groups. Public outrage is a fire which must be hot enough for
a period long enough to shape regulations into a form that benefits
the general public. Stoking the fire of public outrage is often difficult
because the cost of public mobilization is relatively high. In contrast,
134 Behavioralizing the Approach to Financial Market Regulation
interest groups have ready mechanisms for lobbying which they can
mobilize quickly to take advantage of even small changes in the eco-
nomic, financial, political, legal, and technological environments.
7.1.3 Specific Regulations
Conceptually, each configuration of regulations gives rise to a config-
uration in efficiency/fairness space. Each regulation is on the frontier
unless another regulation improves both fairness and efficiency. Shefrin
and Statman (1992) discuss six specific regulations, placing each into
the historical context which led the regulations being established. The
discussion below emphasizes how the essential issues in the debates can
be viewed through the lens of tradeoffs involving differing notions of
efficiency and fairness. Shefrin and Statman (1992) emphasize the issues
of fairness loom large at the time the regulations were put in place.
1. Merit (“blue-sky”) regulations: These diminish Pareto effi-
ciency by restricting choices of those offering and those buy-
ing securities. Merit regulations may cause deviations from
informationally efficient prices, thus adversely affecting the
right to correct (efficient) prices. As to fairness, merit regu-
lation interferes with the right to freedom from coercion, but
its paternalistic characteristics enhance rights to equal pro-
cessing power freedom from impulse. Merit regulations take
power from entrepreneurs and give it to investors, a move
toward equal power if entrepreneurs have more power than
do investors in a mandatory disclosure environment.
2. Disclosure regulations: These improve informational effi-
ciency relative to a buyer beware world. Less clear is whether
mandatory disclosure improves Pareto efficiency, on account
of its costs, and the potential elimination of some risky ven-
tures. The implications of disclosure regulations for fairness
are similar to those for merit regulation.
3. Margin regulations: In the deliberations leading up to the
passage of the 1934 Securities Exchange Act, three dis-
tinct motivations can be seen underlying the regulation
of margin: the need to protect investors from their own
7.1 Dynamic Tug-of-War 135
poor judgment, the limitation of price volatility induced by
low-margin requirements, and the desire to allocate credit
to productive investment instead of financial speculation.
The associated issues involve freedom from coercion, efficient
prices (fair and orderly markets), equal processing power, and
freedom from impulse.
4. Insider trading regulations: The efficiency aspects associated
with these regulations pertain to adverse selection. Some
argue that the regulations improve informational efficiency,
while others argue the opposite, that insiders create new
information when they trade. This issue is largely one of tim-
ing, however, because most material information is released
to the public eventually. The fairness question has played a
more prominent part in the debate than the efficiency ques-
tion, particularly as to a disclosure-or-abstain rule. Two fair-
ness threads can be seen one involving the right to equal
information, and the other involving the right to be protected
from one’s own mistakes. The former issue, which centers on
the notion of the “level playing field,” has been the more
prominent of the two. A fair investment game is perceived
to be a similar game of skill, with equal access to public
information from which the investor can infer which secu-
rities are undervalued and which are overvalued. In a fair
game of skill, winners and losers are expected, but trading
on inside information is unfair because it conveys a non-skill-
based advantage to one party in the trade.
5. Suitability regulations: Suitability rules present a clear case
where some fairness rights win over other rights and over
efficiency. A move from a no-paternalism to a full-paternalism
world enhances the right to freedom from cognitive errors and
the right to freedom from impulse. The paternalistic nature
of suitability rules is evident in the incentives they provide
to brokers to err on the side of low risk rather than on the
side of high expected returns.
6. Trading interruption regulations which followed the stock
market crash of 1987: trading halts are detrimental to Pareto
136 Behavioralizing the Approach to Financial Market Regulation
efficiency because they prevent parties from entering into
agreements viewed as beneficial to themselves. A counter-
argument is that trading halts enhance efficiency during a
sharp decline by permitting settling up to ensure that all
parties are solvent. Any impact on informational efficiency
depends on whether such efficiency is achieved in the absence
of interruptions. One critic notes that natural trading inter-
ruptions, such as the closing bell on the exchange, apparently
do not dampen volatility during turbulent periods. As to fair-
ness, the main issues involve coercion, efficient prices (fair
and orderly markets), equal information processing, freedom
from impulse, and equal bargaining power.
7.1.4 Theoretical Example
7.1.4.1 Absence of Constraints
To fix ideas, consider a numerical binomial example, a special case
of the framework described in Section 5, where T = 2. Let aggregate
consumption be equal to 100 units at t = 0. Thereafter the growth rate
of aggregate consumption per period is stochastic, and is equal either
to u > 1 or d = 1/u < 1. I take u = 1.05, so that the economy grows
either at 5 percent per period or shrinks at 4.76 percent. Let there be
two types of investors in the model, called type 1 and type 2, with type
1 holding 65 percent of the initial wealth of the economy.
Both types of investors share a common rate of time preference
(δ = 0.99). However, the two types of investors differ in their beliefs.
Type 1 investors believe that at t = 1, the probability that u occurs is
0.8, whereas type 2 investors believe that the probability is 0.67. For
t > 2, type 1 investors believe that the probability that a u follows a u is
0.95 and that a d follows a d is 0.8. In contrast, type 2 investors believe
that aggregate consumption growth evolves according to a process that
is independent and identically distributed (i.i.d.), with the probability
of u being 0.76 at all dates. For the sake of argument, assume that
type 2 investors are information traders, meaning they possess correct
beliefs. In this case, type 1 investors will be excessively optimistic after
7.1 Dynamic Tug-of-War 137
economic growth has been positive and excessively pessimistic after
economic growth has been negative, reflecting “hot hand fallacy.”
At the beginning of each date, investors decide how to divide their
portfolio wealth into current consumption and an end of period portfo-
lio such that the value of consumption together with the end-of-period
portfolio is equal to the value of the beginning of period portfolio.
Because the example is binomial, only two securities are needed
to complete the market, which I take to be the market portfolio and
the one-period risk-free security. In equilibrium, the market portfolio
returns 6.1 percent at t = 1 if the economy has grown and −3.8 percent
if it contracts. At t = 1, the interest rate is 3.5 percent. (Returns are
calculated using Equation (5.1), as the SDF embodies all price informa-
tion. Keep in mind that in this model, the market portfolio corresponds
to unlevered equity, unlike, say, the S&P 500.)
At t = 0, type 1 investors have total wealth equal to $193, and type 2
investors have total wealth equal to $104. Both types save 66.3 percent
of their wealth and consume the remainder. At the end of t = 0, type 1
investors borrow $314.6 and hold $442.7 of the market portfolio. Their
equity holdings amount to 346 percent of the value of their portfolios.
Put another way, type 1 investors purchase 71 percent of their equity
holdings on margin. When debt is exclusively used to purchase equity,
the degree of margin corresponds to the debt-to-equity ratio. Type 2
investors hold $314.6 of the risk-free security, and take a short position
of −$245.7 in the market portfolio.
At t = 0, type 1 investors are more optimistic than type 2 investors,
which is why they use leverage to increase their holdings of the market
portfolio. Type 1 investors believe that the expected return on the
market is 4.1 percent, and therefore the equity premium is 0.6 percent.
In contrast, type 2 investors short the market. They believe that the
equity premium is −0.7 percent.
The differences in beliefs at t = 0 lead to an equilibrium with
nonzero sentiment Λ. Notably, equilibrium returns are established as if
the market overestimates the probability of an up-move by 12 percent,
and underestimates the probability of a down-move by 31 percent. As
Equation (5.1) indicates, nonzero sentiment impacts the SDF underly-
ing equilibrium returns.
138 Behavioralizing the Approach to Financial Market Regulation
If growth is positive at t = 1, type 1 investors will earn high returns
from their levered portfolios. They will increase their consumption by
11.3 percent. In contrast, type 2 investors will decrease their consump-
tion by 6.8 percent.
If growth is negative at t = 1, type 1 investors will earn low returns
from their levered portfolios. They will decrease their consumption by
22 percent. In contrast, type 2 investors will increase their consumption
by 28 percent.
In an example featuring more periods, type 2 investors alternate
between acting as if they are excessively optimistic and excessively
pessimistic. Volatile beliefs translate into excessive trading. For this
reason, some investors might jump from holding portfolios that fea-
ture too much leverage to shorting the market portfolio, and back
again. For example, if aggregate consumption declines at t = 1, type 1
investors will become pessimistic and short $882 of the market portfo-
lio, while holding $987 of the risk-free security. In this situation, type 2
investors will end up holding levered positions. These contrasting port-
folios reflect contrasting beliefs about the equity premium. After a down
move at t = 1, type 1 investors believe that the equity premium is −1.9
percent, whereas type 2 investors believe it to be 2.7 percent.
This example illustrates that investors who hold incorrect beliefs
choose non-optimal portfolios that feature inappropriate risk exposure
and excessive trading. This state of affairs may well occur even when
sentiment is zero and prices are efficient. If prices are inefficient, then
sentiment typically induces excessive volatility in the market risk pre-
mium, thereby exacerbating the degree of non-optimality.
7.1.4.2 Regulatory Constraints, Fairness, and Efficiency
In this framework, financial market regulation operates by impos-
ing constraints on investor’s portfolios. For example, regulation can
limit leverage and short sales by imposing upper and lower bounds
on the portfolio proportion allocated to the market portfolio. The
intent of such limits is to force investors to act as if their beliefs were
less extreme. From a modeling perspective, the impact of regulations
induces investors to act as they hold different beliefs than they actually
7.1 Dynamic Tug-of-War 139
do, say {Q
i
} instead of {P
i
}. As in Section 5, a change of measure
captures the essential issue. Notably, such constraints typically impact
market prices, for example by reducing the degree of volatility in the
equity premium.
To see the effect of the change of measure, I modify the example
as follows. Consider what happens if economic growth is positive at
t = 1. In this case, type 1 investors increase their estimate of the equity
premium to 1 percent (from 0.6 percent at t = 0), and increase the
amount of equity purchased on margin to 88 percent (from 71 percent
at the end of t = 0).
Suppose a regulator, concerned that the market was irrationally
exuberant at t = 0, were to step in and limit the degree to which
investors can purchase equity on margin to 50 percent in all subse-
quent periods. What difference would this make to investors’ positions,
imputed beliefs, and market prices? At both t = 0 and t = 1 after an up-
move, type 1 investors would reduce the amount of equity they purchase
on margin to 50 percent. At t = 1 after a down move, type 2 investors
would do likewise. The higher margin requirements would force type 1
investors to trade at t = 0 as if they assigned a probability of 72 per-
cent to an up-move at t = 1, rather than their true belief of 80 percent.
In turn, this would decrease the volatility of the equity premium. At
t = 1, the ex post premium would be either 1.9 percent or −4.1 percent
instead of 2.6 percent or 7.3 percent. The objective expected premium
would increase from −0.7 percent to −0.1 percent, thereby reducing a
substantial portion of the excessive optimism prevailing in the market.
The imposition of regulatory constraints has implications for fair-
ness and efficiency. Suppose that every investor’s allocation in the con-
strained equilibrium features higher expected utility than in the uncon-
strained equilibrium (when both are measured with respect to the cor-
rect probability vector Π). Does this mean that the imposition of regu-
latory constraints has resulted in an increase in fairness coupled with a
loss in efficiency? That depends on both the definitions of fairness and
efficiency.
Constraining investors to act as if they are smarter, or less vulnera-
ble to emotion, might improve fairness defined in terms of freedom from
misrepresentation, equal information, equal processing ability, freedom
140 Behavioralizing the Approach to Financial Market Regulation
from impulse, and equal bargaining power. However, by their nature
constraints involve coercion, thereby leading to reduced fairness defined
as freedom from coercion.
As to efficiency, forcing investors to trade as if they hold beliefs
{Q
i
} which are different from {P
i
} moves the equilibrium allocation
off the Pareto-efficiency frontier defined by {P
i
}, but onto the Pareto-
efficiency frontier defined by {Q
i
}. Hence, when investors view markets
through the lens of {P
i
}, they perceive that the constraints result in
a move off the Pareto-efficient frontier. However, efficiency can also be
defined as informational efficiency. If regulatory constraints lead the
representative investor to hold beliefs in the Q-equilibrium which are
closer to Π than in the P-equilibrium, regulations might lead to an
improvement in efficiency defined relative to correct prices. By defini-
tion, such a situation would also be viewed as an increase in fairness,
when fairness is defined as the right to trade at correct prices.
7.2 Biased Cost–Benefit Assessment
Hirshleifer (2008b) focuses attention on the manner in which systematic
biases on the part of participants in the political process impact reg-
ulatory outcomes. He calls his framework the psychological attraction
approach to regulation, because particular parties advocating increased
regulation exploit psychological biases to attract attention and support.
The basis for Hirshleifer’s framework is the following series of seven
phenomena.
1. Salience and vividness effects: Regulatory debates are influ-
enced heavily by extreme events, and by heart-rending per-
sonal stories. Hirshleifer points out that accounting scan-
dals such as occurred at firms such as Enron and WorldCom
precipitated passage of the Sarbanes–Oxley Act of 2002 (or
SOX). He points out that the costs of disclosure imposed by
SOX are less vivid than poignant stories about families who
suffered large losses.
2
He suggests that the focus on individual
2
The constitutionality of SOX is currently before the U.S. Supreme Court, which will rule
on the issue in 2010.
7.2 Biased Cost–Benefit Assessment 141
losses is amplified through conversation and media reporting,
which are biased toward the communication of adverse and
emotionally charged news. This tendency applies to losses
suffered from derivatives trading, thereby leading the pub-
lic to attach negative perceptions of derivatives. As a result,
the public tends to overlook benefits associated with deriva-
tives, such as hedging, leading to a disposition to regulate
derivatives.
2. Omission bias: This bias involves the favoring of omissions
over commissions. As an example, regulation intended to pro-
tect unsophisticated investors from securities lacking merit
might also prevent them from risk-reducing diversification.
3. Scapegoating and xenophobia: When things go wrong, peo-
ple instinctively seek someone to blame, such as members of
a visible, disliked, and relatively weak out-group. Economic
and stock market downturns increase pressure for regulation
targeting specific culprits. Hirshleifer cites the Securities Acts
of 1933 and 1934, and the Sarbanes–Oxley Act as examples.
4. Fairness and reciprocity norms: Hirshleifer’s view of fairness
partly overlaps with the discussion above. He focuses on three
norms, namely reciprocity, equality, and charity. Reciprocity
involves quid pro quo. Equality involves equal division of
resources. Charity involves acts intended to relieve distress.
According to Hirshleifer, reciprocity motivates scapegoat-
ing, particularly as intermediaries are seen by the public as
adding little value. Equality underlies hostility to lenders,
a view underlying regulations to prevent usury. Charity is
directed at addressing recent losses. Considerations of fair-
ness can explain regulations intended to limit speculation,
restrict short-selling, and penalize short-term capital gains.
5. Overconfidence: According to Hirshleifer, overconfidence
helps to explain the existence of excessive regulation. He cites
the work of Adam Smith and Friedrich Hayek in respect
to a spontaneous order, understood as a web of coadapta-
tions resulting from individual decisions, that is too complex
for a central planner to comprehend. Hirshleifer asserts that
142 Behavioralizing the Approach to Financial Market Regulation
overconfidence leads public officials to believe that they
can institute regulatory mechanisms which can improve on
market allocation. As examples, he mentions transactions
taxes to limit speculation and monetary policy to address
irrationally exuberant equity markets or to provide stimulus
during an economic downturn. He suggests that the better-
than-average effect induces encourages regulators to believe
they are more skilled than investors at identifying the value
of the market. He suggests that the illusion of control leads
regulators to conclude that they are able to avert bubbles
and crashes. In this last regard, he indicates that following
adverse outcomes, regulators come under criticism for failing
to prevent unfavorable outcome, which leads to calls for more
strict regulation.
6. Mood effects and availability cascades: Hirshleifer makes the
point that short-term moods impact judgments and decisions
relating to the long-term. Moreover, mood can be contagious,
thereby aggregating at the social level. In this respect, the
call for greater regulation by a few can lead others to infer
that there is good reason to do so. Information cascades
can develop, whereby the increased social discourse about an
issue leads to the perception that the issue is truly important.
7. Ideological replicators: Ideologies, be they religious, politi-
cal, or economic, shape financial regulation through word
of mouth. Hirshleifer suggests that conspiracy theories are
prone to develop during market crashes. He makes the point
that the general public have a poor understanding of how
major players in financial markets affect risks. In this regard,
he singles out hedge fund managers as being especially sus-
pect, as they operate in secrecy, and are often foreigners.
7.3 Regulation and Financial Crises
Elizabeth Warren is a member of Harvard University Law faculty
who specializes in bankruptcy. Currently, she also heads the commit-
tee which oversees the dispensing of funds from the Troubled Assets
7.3 Regulation and Financial Crises 143
Relief Plan (TARP) passed by Congress in 2008 to rescue key finan-
cial institutions in distress. In Rather (2009), Warren reminds us of
some history, that a financial crisis including a credit squeeze accom-
panied the 1792 crisis, the first economic crisis in U.S. history. Major
financial crises occurred every 15–20 years between 1792 and 1929. As
Shefrin and Statman (2009) point out, the stock market crash of 1929
and Great Depression following heated regulatory irons to extremely
high temperatures, facilitating strikes that changed their shapes radi-
cally, including the Securities Act of 1933, the Securities Exchange Act
of 1934, and the Glass-Steagall Act which separated commercial and
investment banking.
Warren pointed out that no major financial crises occurred between
1934 and the 1980s that were comparable in scale to those occurring
before the Great Depression. However, since 1980, a process of dereg-
ulation has been underway, which she notes has been accompanied by
two major financial crises in the United States, the savings and loan
(S&L) crisis of the 1980s, and the financial crisis in 2008–2009.
The parallels between the S&L crisis of the 1980s and the financial
crisis of 2008–2009 are striking, and strongly suggests the collective
failure to learn. For both crises, the key features involved the
following:
— the market for real estate;
— a political and regulatory environment encouraging home own-
ership;
— fixed income securities at the core of the crisis;
— a period of interest rate volatility featuring low real interest
rates followed by high real interest rates;
— S&L institutions as key players;
— a period of active deregulation featuring low capital ratios;
— risk-seeking behavior by financial institutions;
— subprime loans, featuring acquisition, development, and con-
struction loans in the S&L crisis, and subprime mortgages in
the 2008–2009 crisis;
— the funding of projects with a high probability of generating
poor cash flows;
144 Behavioralizing the Approach to Financial Market Regulation
— agency issues on the part of real estate appraisers in the S&L
crisis and rating firms in the financial crisis of 2008–2009;
— high accounting profits reported by financial institutions;
— large bonuses to executives;
— pyramid (Ponzi) schemes;
— government agencies becoming insolvent: the FSLIC insuring
S&Ls becoming insolvent in 1987, and in 2009 the FDIC was
nearing insolvency; and
— subsequent turbulence in equity markets.
In the discussion below I highlight some of the parallels in the above
list, first by describing the S&L crisis, and then the financial crisis of
2008–2009.
7.3.1 S&L Crisis in the 1980s
Until the 1980s, regulations restricted S&Ls to deposits and home mort-
gage loans. Notably, interest rate ceilings prevented S&Ls from paying
competitive interest rates on deposits. During the late 1970s, when the
critical seeds of the S&L crisis were sown, nominal capital requirements
were low (6 percent of liabilities). Because of inflation, nominal interest
rates rose dramatically at this time, leading to an outflow of deposits
from S&Ls. For the industry as a whole, these increases caused the mar-
ket value of their assets to fall below the market value of their liabilities.
However, there was no fair value accounting in place at the time, and
so this state of affairs was not recognized within the industry’s financial
statements.
The industry response to the changing economic environment fea-
tured status quo bias. Essentially the industry continued to hold mort-
gages it originated, but did not
— lengthen term of its deposits,
— hedge its risk exposures,
— shorten the term of its mortgages, and
— sell mortgages and buy shorter term assets.
Regulatory reform began to take shape in the period 1980–1982.
Legislators granted the S&L industry the power to enter new areas of
7.3 Regulation and Financial Crises 145
business, intended to increase S&L profits as opposed to promoting
housing and homeownership. For example, the Depository Institutions
Deregulation and Monetary Control Act (DIDMCA) made the follow-
ing changes:
— removed interest rate ceiling on deposit accounts,
— expanded authority for federal S&Ls to make ADC loans, and
— raised the deposit insurance limit raised to $100,000 from
$40,000.
The Garn-St. Germain Act of 1982 was a Reagan administration ini-
tiative for federally chartered S&Ls. It eliminated previous statutory
limit on the loan to value ratio, and permitted the portfolios of these
S&Ls to comprise up to
— 40 percent of assets in commercial mortgages,
— 30 percent of assets in consumer loans,
— 10 percent of assets in commercial loans, and
— 10 percent of assets in commercial leases.
At the same time that regulatory restrictions on the S&L industry
were being lifted, the ability of regulatory authorities to monitor the
industry was reduced. The Federal Home Loan Bank Board was the
agency responsible for regulating S&Ls. Between 1980 and 1985 Bank
Board
— reduced net worth requirements for insured S&L institutions,
— removed limits on the amounts of brokered deposits an S&L
could hold,
— permitted lax accounting standards, and
— reduced the size of its regulatory and supervisory staff.
During the early 1980s, most S&Ls were reporting losses. Garn-
St. Germain and DIDMCA increased S&L’s risk-seeking opportuni-
ties for funding high risk development projects. Many institutions took
advantage of these opportunities, in what might have featured aver-
sion to a sure loss, but definitely featured fraud. For those readers
who instinctively hold a laissez fa
ˆ
ire philosophy and believe in market
146 Behavioralizing the Approach to Financial Market Regulation
efficiency, should find the following value destructive sequence of events
instructive.
Value destroying S&Ls seized newfound opportunities in the market
for commercial property loans, by granting loans to developers of low-
quality projects for amounts that overstated the value of these projects.
In doing so, the S&Ls communicated the size of the proposed loan
to potential appraisers, asking for preliminary appraisals. S&Ls then
selected high appraisers and rewarded them with a continuing stream
of business. The same phenomenon manifested itself 20 years later with
ratings firms. Notably, these large loans featured an “anticipated profit
component,” reflecting the inflated values of their projects.
The loans just described were for interest only, with principal due
in two-to-three years. Notably, the terms of these loans held back the
interest payments due, and these payments were subsequently booked
as income on S&Ls’ financial statements, along with points and fees.
S&Ls preferred weak developers, as they would agree to high interest
rates.
For those readers who believe the discipline of the market will surely
curtail such behavior, read on. To be sure, at some point it would
become apparent and did, especially to the S&Ls, that the cash flows
from most of these low quality, high risk projects, were low. Could the
S&Ls avoid recognizing the losses in question? Yes, if they could find
buyers for these projects, which they did, in a scheme called “cash for
trash.” The scheme involved locating other weak developers search-
ing for loans, who agreed to purchase the failing projects of others at
inflated prices, together with a profit sharing arrangement. The inflated
prices in “cash for trash” resulted in book profits for these projects, and
the S&Ls extracted a 50–50 profit split. Being market transactions, the
associated gains were registered as income on these S&Ls’ financial
statements. This made the firms appear to be profitable, induced clean
opinions by auditors, and attracted more depositors, who would be
protected by higher deposit insurance limits. See Black (2005).
Eventually, this pyramid scheme would have to end, and the S&L
would fail. However, that does not mean that the S&L executives would
fail. Indeed, executives would have paid themselves substantial bonuses
for having turned around a nearly failed bank at a time when the whole
7.3 Regulation and Financial Crises 147
industry was in decline. Perhaps, they were sufficiently overconfident
to believe that their fraudulent behavior would go undetected.
In 1983, the Bank Board came under new leadership, with the
appointment of Ed Gray. Gray made a strong effort to reverse the
direction of deregulation, and to recapitalize the FSLIC, which was in
the process of going insolvent. His attempts to investigate one partic-
ular S&L, Lincoln Savings and Loan, led him to be summoned by five
Senators who questioned the appropriateness of his actions. In the end,
Lincoln’s failure cost U.S. taxpayers over $2 billion. The total cost of
addressing the failures in the S&L industry was approximately $150
Billion.
7.3.2 Financial Crisis of 2008–2009
Opinions about the root cause of the financial crisis of 2008–2009 dif-
fer. Posner (2009) argues that the root lies with a weak regulatory
structure, within which private-sector decisions were largely rational.
Akerlof and Shiller (2009) argue that the root lies with irrational deci-
sions associated with the occurrence of a housing market bubble, a
surge in subprime mortgage lending, and the breakdown of the ratio-
nal adverse selection (“lemons”) paradigm. Brunnermeier et al. (2008)
and Taylor (2009) also argue that an asset pricing bubble was a major
issue, which they trace to low U.S. interest rates during the period
2003–2006. Sullivan (2009) blames poor corporate governance, explicit
corruption, and unwise governmental mandates and guarantees. Differ-
entiating among these various views requires a search for the devil in
the details.
In Shefrin (2009), I suggest that each of the above explanations con-
tains an element of truth. The regulatory structure was weak, interest
rates were low, there was a housing market bubble, the surge in sub-
prime mortgage lending was highly problematic, and there were failures
in governance at financial firms.
One lesson not learned from the S&L crisis involves failing to recog-
nize the dangers from simultaneously expanding risky alternatives to
financial institutions and reducing regulatory oversight, if not ignoring
the warnings from regulators. In this regard, the Gramm–Leach–Bliley
148 Behavioralizing the Approach to Financial Market Regulation
Act of 1999 fragmented regulatory oversight so that no regulator had a
comprehensive picture of each financial firm. The Commodity Futures
Modernization Act left the derivatives market without regulatory over-
sight. This occurred despite the warnings issued by the head of the
Commodity Futures Trading Commission, Brooksley Born, who was
subsequently relieved of her position for issuing those warnings. See
PBS (2009) which documents Born’s concerns, and the opposition to
it from Federal Reserve Chair Alan Greenspan, and Clinton adminis-
tration officials Robert Rubin and Larry Summers.
Another regulator whose warnings went in vain was Comptroller of
the Currency John Dugan (CNN, 2008). In mid-2005, he was among
the first to sound the alarm about risky lending in the housing market,
saying that many buyers, particularly those with bad credit, would soon
be unable to afford their mortgage payments. He added that if housing
prices declined, homeowners would be unable to sell their homes, and
foreclosure rates would rise dramatically.
During 2005, bank regulators observed deteriorating conditions in
the mortgage market, and proposed new guidelines for banks writing
risky loans. In line with capture theory, the banking industry lobbied
against the proposal. The governments banking agencies spent nearly
a year debating the rules, which required unanimous agreement among
different regulatory bodies, including the Federal Reserve. The industry
succeeded in achieving a one-year delay, with new rules released in late
2006, and the strictest of the proposed provisions gone. The provisions
stripped from the final rules included the following:
— Banks that bundle and sell mortgages must disclose to investors
exactly what they are buying;
— Banks must advise home buyers that interest rates might
increase dramatically, resulting in payments becoming due
sooner than expected;
— Banks must verify that buyers actually have jobs and can afford
houses; and
— Banks must cap their risky mortgages so that a string of
defaults would not be crippling.
7.3 Regulation and Financial Crises 149
Consider the comments of Alan Greenspan, who chaired the Federal
Reserve during the key years in which the seeds of the crisis were sown.
In June 2005, Greenspan testified before Congress that some local hous-
ing markets exhibited “froth.” He pointed to the use of risky financing
by some homeowners and suggested that the price increases in those
local markets were unsustainable. However, he concluded that there
was no national housing bubble and that the economy was not at risk.
In the same vein, Greenspan’s successor at the Fed, Ben Bernanke, gave
a speech on 17 May 2007 in which he stated, “[W]e do not expect sig-
nificant spillovers from the subprime market to the rest of the economy
or to the financial system.”
In 2008, Greenspan testified before the House of Representatives
Committee on Oversight and Government Reform. Below is an excerpt
from his appearance, beginning with a question from committee chair
Henry Waxman.
Henry Waxman: You were the longest serving chairman
of the Federal Reserve in history and during this period
of time were perhaps the leading proponent of dereg-
ulation of our financial markets. Certainly you were
the most influential voice for deregulation. You have
been a staunch advocate for letting markets regulate
themselves. . . . My question for you is simple, were you
wrong?
Alan Greenspan: I made a mistake in presuming that
the self-interest of organizations, specifically banks and
others, was capable of protecting their own shareholders
and the equity in the firms, and it has been my expe-
rience, having worked both as a regulator for 18 years
and similar quantities in the private sector, especially
10 years at a major international bank, that the loan
officers of those institutions knew far more about the
risks involved and the people to whom they lent money
than I saw even our best regulators at the fed capable of
doing. So the problem here is something which looked
150 Behavioralizing the Approach to Financial Market Regulation
to be a very solid edifice and indeed a critical pillar to
market competition and free markets did break down
and I think that shocked me. I still do not fully under-
stand why it happened and obviously to the extent that
I figure out where it happened and why, I will change
my views. . . .
Henry Waxman: This is your statement. “I do have an
ideology. My judgment is that free competitive mar-
kets are by far the unrivaled way to organize economies.
We have tried regulation, none meaningfully worked.”
That was your quote. You had the authority to prevent
irresponsible lending practices that led to the subprime
mortgage crisis. You were advised to do so by many oth-
ers. And now our whole economy is paying its price. Do
you feel that your ideology pushed you to make deci-
sions that you wish you had not made?
Alan Greenspan: I found a flaw in the model that I per-
ceived as the critical functioning structure that defines
how the world works. . . . [T]his crisis, however, has
turned out to be much broader than anything I could
have imagined · · · Those of us who have looked to the
self-interest of lending institutions to protect sharehold-
ers’ equity-myself especially-are in a state of shocked
disbelief.
I suggest that the flaw in Greenspan’s model, and the missing pieces
in his understanding of what happened, lies in behavioral finance. In
this regard, psychological issues such as overconfidence, excessive opti-
mism, confirmation bias, and aversion to a sure loss played major roles
in the decision processes at investment banks, rating agencies, insur-
ance firms, institutional investors, and regulatory agencies.
In Shefrin (2009), I present a series of case studies demonstrat-
ing that the financial crisis of 2008–2009 was driven by a combination
of sentiment in asset prices and organizational issues that are at the
7.3 Regulation and Financial Crises 151
heart of behavioral corporate finance. The case studies highlight (to a
greater or lesser degree) irrational decision making at key points in the
financial product supply chain. I emphasize that psychological pitfalls
dominated the key processes at financial firms. As mentioned in Shefrin
(2009), these processes pertained to goal setting, planning, incentives,
and information sharing. Below are highlights from the cases, all of
which highlight the importance of debiasing.
AIG
American International Group (AIG) is an insurance firm which facil-
itated the explosive growth in subprime mortgage lending in 2004
and 2005 by selling credit default spreads (CDS) that insured against
default. AIG’s financial product division irrationally failed to track the
proportion of subprime mortgages in the pools being insured, thereby
misgauging the risk of those assets and causing the associated CDS
contracts to be mispriced. Interestingly, this failure occurred despite
employee incentives that balanced long-term performance against
short-term performance. Moreover, conversations that AIG had with its
investment bank trading partners indicate the presence of a widespread
conservatism bias regarding the assumption that historical mortgage
default rates would continue to apply.
In 2004 and 2005, the activities of financial institutions might have
been rational because they were able to shift default risk to AIG by
purchasing credit default swap protection. However, after AIG stopped
selling CDS contracts, many financial firms took on the risk themselves,
apparently under the illusion that housing prices would continue to rise
and that default rates would not be affected by the increased ratio of
subprime to prime mortgages. At one point, Merrill Lynch used CDS
contracts to create synthetic collateralized debt obligations (CDOs)
because the number of subprime mortgages available to create tradi-
tional CDOs was insufficient relative to the firm’s aspirations.
UBS
UBS’s investment banking division admitted to misgauging subprime
mortgage risk by not “looking through” the CDO structures and by
152 Behavioralizing the Approach to Financial Market Regulation
assuming that historical default rates would continue to apply. UBS’s
underperformance relative to competitors led them to exhibit reference
point-induced risk seeking. This behavior was compounded by poor
incentive structures at UBS that emphasized short-term performance
over long-term performance.
UBS placed no operational limits on the size of its CDO warehouse.
It was not alone. Investment banks are typically intermediaries, not
end investors planning to hold large positions in subprime mortgages.
Some hold the equity tranches of CDOs as a signal to the buyers of
the less risky tranches. What created many of the losses for investment
banks, however, was inventory risk–risk stemming from warehousing
the subprime positions that underlay CDOs. As the housing market
fell into decline, many investment banks found that they could not
locate buyers for the CDOs and, as a result, inadvertently became end
investors.
Rating Agencies: S&P and Moody’s
The rating agencies and investors’ reliance on them played a huge role
in the financial crisis. Both (supposedly) sophisticated investors, such
as the investment bankers at UBS, and naive end investors, relied on
the risk assessments of rating agencies. However, the rating agencies
explicitly indicated that their ratings were premised on accepting the
information they received as accurate, even if the mortgages featured
limited documentation. For this reason, these agencies suggested that
their ratings be treated as only one piece of information when assessing
risk. By this argument, users who accepted their ratings at face value
behaved irrationally.
Did the rating agencies exhibit irrational behavior by weakening
their risk assessment criteria to cultivate more business? This question
is different from asking whether the ratings agencies behaved ethically.
The major problem with the behavior of rating agencies might arise
more from a conflict of interests (the principal–agent conflict) than
irrationality, in the sense that the issuers of securities, not the end
investors in the securities, pay for ratings. At the same time, I suggest
that reference point-induced risk seeking and groupthink played major
roles at rating agency S&P.
7.3 Regulation and Financial Crises 153
In this regard, Griffin and Tang (2009) find that ratings by Moody’s
and S&P were higher than what their models implied. They conclude
that tranches that were rated AAA actually corresponded to BBB when
rated according to the firms’ models and default standards.
SEC
The SEC came under intense criticism for its lax oversight of investment
banking practices and for failing to detect a large hedge fund Ponzi
scheme run by Bernard Madoff. A focal point of the criticism of invest-
ment bank oversight involved a meeting that took place at the SEC
on April 28, 2004, when the commission was chaired by William Don-
aldson. That meeting established the Consolidated Supervised Entities
(CSE) program, a voluntary regulatory program that allowed the SEC
to review the capital structure and risk management procedures of par-
ticipating financial institutions. Five investment banks joined the CSE
program, as did two bank holding companies. The investment banks
were Bear Stearns, Goldman Sachs, Lehman Brothers, Merrill Lynch,
and Morgan Stanley, and the bank holding companies were Citigroup
and JPMorgan Chase. Subsequent to the April 2004 meeting, the SEC
failed to provide qualified staff in sufficient numbers to effectively mon-
itor these financial institutions.
As to the Bernard Madoff Ponzi scheme, the largest in history, the
SEC issued an internal report in the summer of 2009. The report indi-
cates that between 1992 and 2008, the agency received six distinct com-
plaints about Madoff’s operations, one of which involved three separate
versions. Several of the complaints suggested that Madoff was running
a Ponzi scheme. The report reveals that the SEC conducted investiga-
tions and examinations related to Madoff’s investment advisory busi-
ness but failed to uncover the fraud, largely because it ignored pertinent
evidence that it uncovered. I suggest that confirmation bias lay at the
heart of the SEC’s failure to detect Madoff’s Ponzi scheme.
Regulatory Response
In June 2009, President Obama proposed a broad new regulatory
framework for financial markets, emphasizing a centralized structure
154 Behavioralizing the Approach to Financial Market Regulation
for oversight, with the Federal Reserve Bank playing a central role.
Among the other proposals is the creation of a consumer protection
agency, suggested by Elizabeth Warren. Notably, there is opposition to
this last proposal from the financial services industry. The SEC’s Divi-
sion of Enforcement Management is in the process of developing a plan
to restructure its approach through five new specialized units. These
units would focus on investment companies (including hedge funds and
private equity firms), market abuses, complex derivatives, the munici-
pal securities market, and bribery issues in foreign countries involving
U.S. companies.
In October 2009, the House Financial Services Committee voted to
pass the “Accountability and Transparency in Rating Agencies Act” in
order to address some of the issues mentioned above. This Act allows
investors to sue rating agencies for knowingly or recklessly failing to
review key information in their ratings, provides the SEC with author-
ity to sanction supervisors at credit rating agencies for failing to super-
vise employees, and contains requirements to mitigate the conflicts of
interest that result because it is the issuers who actually pay rating
agencies, not the investors.
7.3.3 Market Prices
Figure 7.1 displays the monthly time series of stock prices between
1984 and 2009. As can be seen, both the period associated with the
S&L crisis and the financial crisis of 2008–2009 featured considerable
turbulence. The stock market crash of October 1987 occurred in the
midst of the S&L crisis.
Figure 7.2 displays an update of the analysis in Shiller (1981, 2005).
The point Shiller makes in this figure is that market prices are exces-
sively volatile relative to dividends. In making this point, he uses three
different discount rates. Two discount rates are time varying and are
derived from aggregate consumption and market interest rates, respec-
tively. The third discount rate is time invariant.
Notice from Figure 7.2 that after 1991, market prices exceeded
intrinsic value no matter which of the three proxies for intrinsic value is
used. According to this figure, the financial crisis of 2008–2009 merely
7.3 Regulation and Financial Crises 155
0
200
400
600
800
1000
1200
1400
1600
1800
2000
1
9
8
4
1
9
8
5
1
9
8
6
1
9
8
7
1
9
8
8
1
9
8
9
1
9
9
0
1
9
9
1
1
9
9
2
1
9
9
3
1
9
9
4
1
9
9
5
1
9
9
6
1
9
9
7
1
9
9
8
1
9
9
9
2
0
0
0
2
0
0
1
2
0
0
2
2
0
0
3
2
0
0
4
2
0
0
5
2
0
0
6
2
0
0
7
2
0
0
8
2
0
0
9
Date
R
e
a
l

S
t
o
c
k

P
r
i
c
e

(
S
h
i
l
l
e
r

I
n
d
e
x
)
Fig. 7.1 Display of the monthly time series of stock prices between 1984 and 2009. Source:
Shiller (2005) and http://www.irrationalexuberance.com/index.htm.
served to bring stock prices into alignment with fundamental value.
The data displayed in Figure 7.2 are annual, with the 2009 value cor-
responding to June. Stock prices were approximately 18 percent lower
at the trough in March, and ended 2009 20 percent higher than in
June. Given that Shiller’s measures for intrinsic value move much more
slowly than market prices, one might infer that the market was 20 per-
cent undervalued at the trough in March, but moved to become 20
percent overvalued by year end.
Shiller bases his measures for intrinsic value on aggregate dividend
flows. However, Boudoukh et al. (2007) analyze how intrinsic values for
stocks should reflect the present value of future payouts, not just divi-
dends. Payouts include share repurchases. Their analysis suggests that
as the ratio of dividends to payouts declined after the 1970s, Shiller’s
measures might have become biased downward. This might explain why
stock prices continued well above intrinsic value, even after the collapse
of the dot-com bubble.
A final point to note is that along the decomposition of total returns
along the intrinsic value paths features a much larger contribution from
156 Behavioralizing the Approach to Financial Market Regulation
10
100
1000
10000
1860 1880 1900 1920 1940 1960 1980 2000 2020
Year
P
r
i
c
e
Actual Real Stock Price
Present Value of
Dividends Using Constant
Discount Rate
Present Value
Dividends Using
Actual Future
Interest Rates
Consumption-Discounted
Dividends
Fig. 7.2 Comparing actual real stock price with three alternative PDVs of future real divi-
dends. Source: Shiller (2005) and http://www.irrationalexuberance.com/index.htm.
dividend yield than is the case with market value. With market value,
the lion’s share of total return stems from stock price appreciation, not
dividend yield. However, for the measure of intrinsic value associated
with the time invariant discount rate, price changes are very gradual,
thereby forcing most of the return to be generated by the dividend
yield.
7.4 Differences in Perspective
Hirshleifer’s perspective emphasizes overregulation, stemming from the
psychological underweighting of regulatory costs relative to benefits. In
many ways, his position is similar to the general position that guided
Alan Greenspan’s approach to regulation when he chaired the Federal
Reserve.
On the other hand, capture theory suggests that private interests
will seek to exploit the regulatory framework to their advantage, rela-
tive to the general public. Much of the time, their efforts will succeed,
and they will achieve greater power than those seeking stronger reg-
ulation that favors the general public. However, the efforts of special
7.4 Differences in Perspective 157
interests will also create conditions in which the financial system is
prone to a crisis. In a crisis, the balance of power shifts to those seek-
ing regulations that favor the general public.
A crisis creates a window of opportunity for those seeking stronger
regulation. People are prone to recency bias. They have short memories.
For this reason, over time regulations will wax and wane in strength,
reflecting external shocks, the elements of capture theory, and the tug-
of-war between varying notions of fairness. No group is immune to
psychological pitfalls, be they investors, corporate managers, traders,
bankers, regulators, or politicians. Free markets which are perfectly self-
regulating constitute an ideal, a point which Alan Greenspan learned
the hard way. At the same time, regulations established in response
to specific crises might be highly imperfect, representing either overre-
action or underreaction to the crises in question. There is a literature
suggesting that some of the provisions in the Sarbanes–Oxley Act are
in this vein. Regulatory agencies are imperfect, as the discussion about
the SEC demonstrates. It is easy to have debates about regulation on
broad principles. However, any serious discussion of financial market
regulation needs to highlight behavioral biases and be based on facts
rather than ideology.
8
Concluding Remarks
Behavioral finance is in the process of transforming the financial
paradigm from a neoclassical framework to a framework that explicitly
incorporates psychological phenomena. In this monograph I describe
highlights of the behavioral finance literature, focusing attention on
both its strengths and its weaknesses.
Behavioral finance has brought a realistic dimension to finance,
replacing unrealistic assumptions neoclassical finance makes about
investors and managers being fully rational and market prices being effi-
cient. At the same time, a significant portion of the behavioral finance
literature lacks rigor, and indeed some of its claims appear to be in
error. In addition to surveying the foundations underlying the behav-
ioral literature, I use this monograph to identify trends — possible
future directions for behavioral finance. I suggest that the future of
academic finance will combine the realism offered by behavioral finance
with the rigor offered by neoclassical finance. Sections 4 through 7 of
this monograph provide some specific ideas for what shape the behav-
ioralization of finance might take.
The approach to behavioral preferences begins with the behavior
that motivated prospect theory: gains and losses relative to a reference
point as carriers of value, the four-fold pattern for risk attitudes,
158
159
psychophysics, and editing. Notably, prospect theory ignores the role
of emotion in decision making. In addition, its formalization possesses
some structural weaknesses. For these reasons, behavioral preferences
will add features in other psychological-based theories of risk such as
regret, self-control, and the emotions emphasized in SP/A theory. My
view of behavioral beliefs is that they are best modeled as imperfect
learning systems that are heuristic-driven and violate Bayes’ rule. For-
mally, it will be convenient to represent behavioral beliefs using change
of measure techniques, especially for studying the impact of beliefs on
asset pricing.
Behavioral beliefs and preferences serve as the basis for behavior-
alizing portfolio theory. Behavioral portfolios are structured to satisfy
a variety of financial and psychological needs. Behavioral portfolios
reflect a series of psychological phenomena such as status quo bias,
the attention hypothesis, the disposition effect, overconfidence, and the
predilection for lottery stocks, cash dividends, structured products, and
covered call option positions. Behavioral portfolios tend to be undiver-
sified. In the future, behavioral portfolio models will feature combined
diffusion/jump processes to capture the tendency of investors to make
adjustments to their portfolios intermittently rather than continuously
in time. These models will feature the combination of status quo bias
and other psychological traits.
The behavioralization of asset pricing theory will replace unrealis-
tic neoclassical assumptions with more realistic psychologically based
assumptions in models built around the SDF. In this regard, system-
atic risk will feature both fundamentals and sentiment. Sentiment will
be expressed as a log-change of measure. Risk premiums will be deter-
mined through the SDF and mean-variance (MV) frontier in analogous
ways to neoclassical theory, except that the SDF and MV-frontier will
reflect sentiment. Behavioral asset pricing models have already begun
to generalize neoclassical asset pricing models, especially for the term
structure of interest rates and option prices. In addition, work on iden-
tifying return anomalies will continue. A recent example pertains to
the U.S. Presidential election cycle. See Kr¨ aussl et al. (2009).
Some of the most recent behavioral advances have occurred in
corporate finance. I focus attention on three areas that strike me as
160 Concluding Remarks
especially germane to future contributions. They are: (1) the influence
of corporate managers’ personal characteristics on their corporate deci-
sions; (2) the impact of social networks on corporate decisions and
corporate value; and (3) the psychological profiles of entrepreneurs.
At present there is but a small literature in financial market regula-
tion, which might explain why financial economists were ill equipped to
comment on the aftermath of the financial crisis of 2008–2009. It seems
to me that psychological pitfalls lie at the heart of the financial crisis
of 2008–2009, and for this reason it will be important going forward for
financial economists to study the role of psychology in financial market
regulation. I suggest that by and large, the lessons of the S&L crisis
of the 1980s largely went unheeded, and the seeds of that crisis were
allowed to germinate and wreck havoc two decades later.
If there is a lesson in this monograph for economists, it is this. Going
forward, neoclassical economists need to incorporate more psychology
into their models, and behavioral economists need to incorporate more
rigor into theirs. The future lies with both groups meeting in the middle.
References
A¨ıt-Sahalia, Y. and A. Lo (2000), ‘Nonparametric risk management
and implied risk aversion’. Journal of Econometrics 94, 9–51.
Akerlof, G. and R. Shiller (2009), Animal Spirits: How Human Psychol-
ogy Drives the Economy, and Why it Matters for Global Capitalism.
Princeton, NJ: Princeton University Press.
Allais, M. (1953), ‘La psychologie de l’hˆ omme rationnel devant le risque:
critique des postulats et axiomes de l’ ´ecole am´ericaine’. Economet-
rica 21, 503–546.
Allport, G. W. and L. J. Postman (1947), The Psychology of Rumor.
Holt, New York.
Anderson, T. W. (1932), Remembering: A Study in Experimental and
Social Psychology. London York: Cambridge University Press.
Andreassen, P. B. and S. J. Kraus (1990), ‘Judgemental extrapolation
and the salience of change’. Journal of Forecasting 9, 347–372.
Ang, J. and Y. Cheng (2009), ‘Direct evidence on the market-driven
acquisitions theory’. Journal of Financial Research, forthcoming.
Arkes, H., L. Herren, and A. Isen (1988), ‘The role of potential loss in
the influence of affect on risk-taking behavior’. Organizational Behav-
ior and Human Decision Processes 66, 228–236.
161
162 References
Asch, S. (1956), ‘Studies of independence and conformity: A minority
of one against a unanimous majority’. Psychological Monographs 70,
Whole No. 416.
Baker, M., J. Coval, and J. Stein (2004), ‘Corporate financing decisions
when investors take the path of least resistance’. Journal of Financial
Economics 84, 266–298.
Baker, M., S. Nagel, and J. Wurgler (2007a), ‘The Effect of Dividends
on Consumption’. Brookings Papers on Economic Activity pp. 277–
291.
Baker, M., R. Ruback, and J. Wurgler (2007b), ‘Behavioral corporate
finance: A survey’. In: E. Eckbo (ed.): The Handbook of Corporate
Finance: Empirical Corporate Finance. New York: Elsevier/North
Holland.
Baker, M., J. Stein, and J. Wurgler (2003), ‘When does the market
matter? Stock prices and the investment of equity-dependent firms’.
Quarterly Journal of Economics 118, 969–1006.
Baker, M. and J. C. Stein (2004), ‘Market liquidity as a sentiment
indicator’. Journal of Financial Markets 7, 271–99.
Baker, M. and J. Wurgler (2002), ‘Market timing and capital structure’.
Journal of Finance 57, 1–32.
Baker, M. and J. Wurgler (2004a), ‘Appearing and disappearing div-
idends: The link to catering incentives’. Journal of Financial Eco-
nomics 73, 271–288.
Baker, M. and J. Wurgler (2004b), ‘A catering theory of dividends’.
Journal of Finance 59, 271–288.
Baker, M. and J. Wurgler (2006), ‘Investor sentiment and the cross-
section of stock returns’. Journal of Finance 61, 1645–1680.
Baker, M. and J. Wurgler (2007), ‘Investor sentiment in the stock mar-
ket’. Journal of Economics Perspectives 21, 129–151.
Bakshi, G. S., C. Cao, and Z. Chen (2000), ‘Do call prices and the
underlying stock always move in the same direction’. Review of
Financial Studies 13, 549–584.
Balduzzi, P., D. Das, and S. Foresi (1998), ‘The central tendency: a
second factor in bond yields’. Review of Economics and Statistics
80, 62–72.
References 163
Barber, B., Y. Lee, Y. Liu, and T. Odean (2005), ‘Do individual day
traders make money? Evidence from Taiwan’. Working Paper, Uni-
versity of California, Davis.
Barber, B., Y. Lee, Y. Liu, and T. Odean (2009a), ‘Just how much do
individual investors lose by trading?’. Review of Financial Studies
22, 609–632.
Barber, B. and T. Odean (2000), ‘Trading is hazardous to your wealth:
The common stock performance of individual investors’. Journal of
Finance 55, 773–806.
Barber, B. and T. Odean (2008), ‘All that glitters: The effect of atten-
tion on the buying behavior of individual and institutional investors’.
Review of Financial Studies 21, 785–818.
Barber, B., T. Odean, and N. Zhu (2009b), ‘Do retail trades move
markets’. Review of Financial Studies 22, 151–186.
Barberis, N. and M. Huang (2001), ‘Mental accounting, loss aversion
and individual stock returns’. Journal of Finance 56, 1247–1292.
Barberis, N. and M. Huang (2008), ‘Stocks as lotteries: The impli-
cations of probability weighting for security prices’. American Eco-
nomic Review 98, 2066–2100.
Barberis, N., M. Huang, and T. Santos (2001a), ‘Prospect theory and
asset prices’. Quarterly Journal of Economics 116, 1–53.
Barberis, N. and A. Shleifer (2003), ‘Style investing’. Journal of Finan-
cial Economics 68, 161–199.
Barberis, N., A. Shleifer, and R. Vishny (1998), ‘A model of investor
sentiment’. Journal of Financial Economics 49, 307–345.
Barberis, N., A. Shleifer, and J. Wurgler (2001b), ‘Comovement’. Work-
ing Paper, University of Chicago.
Barberis, N. and R. Thaler (2003), ‘A Survey of behavioral finance’.
In: G. Constantinides, R. Stulz, and M. Harris (eds.): Handbook of
the Economics of Finance. North Holland.
Barberis, N. and W. Xiong (2009), ‘What drives the disposition effect?
An analysis of a long-standing preference-based explanation’. Journal
of Finance 64, 751–784.
Barone-Adesi, G., R. Engle, and L. Mancini (2008), ‘A GARCH option
pricing model with filtered historical simulation’. Review of Financial
Studies 21, 1223–1258.
164 References
Barone-Adesi, G., P. Gagliardini, and G. Urga (2004), ‘Testing asset
pricing models with coskewness’. Journal of Business & Economic
Statistics 22, 474–485.
Barone-Adesi, G. and P. Talwar (1983), ‘Market models and het-
eroscedasticity of residual security returns’. Journal of Business &
Economic Statistics 1, 163–168.
Barsky, R. B., M. S. Kimball, F. T. Juster, and M. D. Shapiro (1997),
‘Preference parameters and behavioral heterogeneity: An experimen-
tal approach in the health and retirement survey’. Quarterly Journal
of Economics 112, 537–579.
Ben-David, I., J. Graham, and C. Harvey (2007), ‘Managerial overcon-
fidence and corporate policies’. Working paper, Duke University.
B´enabou, R. (2008), ‘Groupthink: Collective delusions in organizations
and markets’. Working Paper, Princeton University.
Benartzi, S. (2001), ‘Excessive extrapolation and the allocation of
401(k) accounts to company stock’. Journal of Finance 56, 1747–
1764.
Benartzi, S., E. Peleg, and R. Thaler (2009), ‘Choice architecture and
retirement saving plans’. In: E. Shafir (ed.): The Behavioral Foun-
dations of Policy. Russell Sage Foundation and Princeton University
Press, forthcoming.
Benartzi, S. and R. Thaler (1995), ‘Myopic loss aversion and the equity
premium puzzle’. Quarterly Journal of Economics pp. 73–92.
Benartzi, S. and R. Thaler (2001), ‘Nave diversification strategies in
defined contribution savings plans’. American Economic Review 91,
79–98.
Benartzi, S. and R. Thaler (2007), ‘Heuristics and biases in retirement
savings behavior’. Journal of Economic Perspectives 21, 81–104.
Bergstresser, D., M. Desai, and J. Rauh (2004), ‘Earnings manipula-
tion and managerial investment decisions: Evidence from sponsored
pension plans’. Quarterly Journal of Economics 121, 157–195.
Bernard, C. and P. Boyle (2008), ‘Structured investment products with
caps and floors’. Working paper, University of Waterloo.
Bitler, M., T. Moskowitz, and A. Vissing-Jorgensen (2004), ‘Test-
ing agency theory with entrepreneur effort and wealth’. Journal of
Finance 60, 539–576.
References 165
Black, W. (2005), ‘Control fraud as an explanation for white-collar
crime waves: The case of the savings & loan debacle’. Crime, Law
and Social Change 43, 1–29.
Blackburn, D. and A. Ukhov (2006), ‘Estimating preferences toward
risk: Evidence from Dow Jones’. Working Paper, University of
Indiana.
Blume, L. and D. Easley (2008), ‘Market selection and asset pricing’.
In: T. Hens and K. Schenk-Hopp´e (eds.): Handbook of Financial Mar-
kets: Dynamics and Evolution. Amsterdam: Elsevier.
Boudoukh, J., R. Michaely, M. Richardson, and M. Roberts (2007), ‘On
the importance of measuring payout yield: Implications for empirical
asset pricing’. Journal of Finance 62, 877–915.
Bouwman, C., K. Fuller, and A. Nain (2009), ‘The performance of
stock-price driven acquisitions’. Review of Financial Studies 22, 633–
679.
Brav, A., J. Graham, C. Harvey, and R. Michaely (2005), ‘Payout
policy in the 21st century’. Journal of Financial Economics 77,
483–527.
Brocas, I. and J. Carrillo (2008), ‘The brain as a hierarchical organiza-
tion’. American Economic Review 98, 1312–1346.
Bruhin, A., H. Fehr-Duda, and T. Epper (2009a), ‘Risk and rationality:
Uncovering heterogeneity in probability distortion’. Working paper,
ET: Zurich.
Bruhin, A., H. Fehr-Duda, and T. Epper (2009b), ‘Uncertainty breeds
decreasing impatience: The role of risk preferences in time discount-
ing’. Working paper (ET: Zurich).
Brunnermeier, M., D. Diamond, and R. Rajam (2008), ‘Deciphering
the liquidity and credit crunch 2007–08’. NBER Working Paper No.
14612.
Burger, J. and H. Cooper (1979), ‘The desirability of control’. Motiva-
tion and Emotion 3, 381–393.
Cai, Y. and M. Sevilir (2009), ‘Board connections and M&A transac-
tions’. Working paper, University of North Carolina.
Camerer, C. (1995), ‘Individual decision making’. In: J. Kagel and A.
Roth (eds.): The Handbook of Experimental Economics. Princeton
University Press, pp. 587–703.
166 References
Campbell, C., S. Johnson, J. Rutherford, and B. Stanley (2009),
‘CEO confidence and forced turnover’. Working paper, Texas A&M
University.
Campbell, J. (2000), ‘Asset pricing at the millenium’. Journal of
Finance.
Campbell, J. and R. Shiller (1991), ‘Yield spreads and interest rate
movements: A bird’s eye view’. Review of Economic Studies 58, 495–
514.
Campbell, J. Y. and J. Cochrane (1999), ‘By Force of Habit: A
consumption-based explanation of aggregate stock market behavior’.
Journal of Political Economy 107, 205–251.
Chan, W. S., R. Frankel, and S. P. Kothari (2003), ‘Testing behavioural
finance theories using trends and sequences in financial performance’.
Working paper (MIT).
Chen, J., H. Hong, and J. Stein (2002), ‘Breadth of ownership and
stock returns’. Journal of Financial Economics 66, 171–205.
Chirinko, R. and H. Schaller (2001), ‘Business fixed investment and
’bubbles’: The Japanese case’. American Economic Review 91, 663–
680.
Chirinko, R. and H. Schaller (2004), ‘Glamour vs. value: The real story’.
Working paper, Emory University.
Chopra, N., J. Lakonishok, and J. Ritter (1992), ‘Measuring abnormal
performance: do stocks overreact?’. Journal of Financial Economics
31, 235–268.
CNN, December 1, (2008), ‘Government warned of mortgage meltdown:
Regulators ignored warnings about risky mortgages, delayed regula-
tions on the industry’.
Cochrane, J. (2000), Asset Pricing. Princeton: Princeton University
Press, 1st edition.
Cochrane, J. (2005), Asset Pricing. Princeton: Princeton University
Press, 2nd edition.
Cochrane, J. and M. Piazzesi (2005), ‘Bond Risk Premia’. American
Economic Review 95, 138–160.
Constantinides, G. (1990), ‘Habit formation: A resolution of the equity
premium puzzle’. Journal of Political Economy 98, 519–543.
References 167
Constantinides, G., J. Jackwerth, and S. Perrakis (2009), ‘Mispricing of
S&P 500 index options’. Review of Financial Studies 22, 1247–1277.
Cooper, A., C. Woo, and W. Dunkelberg (1988), ‘Entrepreneurs’ per-
ceived chances for success’. Journal of Business Venturing 3, 97–108.
Cooper, M., O. Dimitrov, and P. R. Rau (2001), ‘A rose.com by any
other name’. Journal of Finance 56, 2371–2388.
Coval, J. and T. Shumway (2005), ‘Do behavioral biases affect prices?’.
Journal of Finance 60, 1–34.
Dai, Q. and K. Singleton (2003), ‘Term structure dynamics in theory
and reality’. Review of Financial Studies 16, 631–678.
Daniel, K., D. Hirshleifer, and A. Subrahmanyam (1998), ‘Investor psy-
chology and security market under- and overreactions’. Journal of
Finance 53, 1839–1885.
Daniel, K., D. Hirshleifer, and A. Subrahmanyam (2001), ‘Overconfi-
dence, arbitrage and equilibrium asset pricing’. Journal of Finance
56, 921–965.
Daniel, K. D. and S. Titman (2006), ‘Market reactions to tangible and
intangible information’. Journal of Finance 61, 1605–1643.
De Bondt, W., H. Shefrin, G. Muradoglu, and S. Staikouras (2008),
‘Behavioral finance: Quo vadis?’. Journal of Applied Finance 18(2),
7–21.
De Giorgi, E. (2009), ‘Goal-based investing with cumulative prospect
theory and satisficing behavior’. Working paper, University of
Lugano.
De Giorgi, E., T. Hens, and M. Rieger (2009), ‘Financial market equi-
libria with cumulative prospect theory’. Working paper, University
of Zurich.
DeBondt, W. and R. Thaler (1985), ‘Does the stock market overreact?’.
Journal of Finance 40, 793–808.
Degeorge, F., F. Derrien, and K. Womack (2007), ‘Analyst Hype in
IPOs: Explaining the popularity of bookbuilding’. Review of Finan-
cial Studies 20, 1021–1058.
DeLong, J. B., A. Shleifer, L. Summers, and R. Waldmann (1990a),
‘Noise trader risk in financial markets’. Journal of Political Economy
98, 703–738.
168 References
DeLong, J. B., A. Shleifer, L. Summers, and R. Waldmann (1990b),
‘Positive feedback investment strategies and destabilizing rational
speculation’. Journal of Finance 45, 375–395.
DeLong, J. B., A. Shleifer, L. Summers, and R. Waldmann (1991), ‘The
survival of noise traders in financial markets’. Journal of Business 64,
1–20.
Detemple, J. and S. Murthy (1994), ‘Intertemporal asset pricing with
heterogeneous beliefs’. Journal of Economic Theory 62, 294–320.
Dichev, I. (2004), ‘What are stock investors’ actual historical returns?’.
University of Michigan working paper.
Diecidue, E. and D. La-Ornual (2009), ‘Reconciling support theory and
the book-making principle’. Journal of Risk and Uncertainty, forth-
coming.
Diecidue, E. and J. van de Ven (2008), ‘Aspiration level, probability
of success and failure, and expected utility’. International Economic
Review 49, 683–700.
Diener, E., R. Emmons, R. Larsen, and S. Griffin (1985), ‘The sat-
isfaction with life scale’. Journal of Personality Assessment 49,
71–75.
Diether, K., C. Malloy, and A. Scherbina (2004), ‘Differences of opinion
and the cross-section of stock returns’. Journal of Finance 57, 2113–
2141.
Dittmar, R. (2002), ‘Nonlinear pricing kernels, kurtosis preference, and
evidence from the cross section of equity returns’. Journal of Finance
57, 369–403.
Dong, M., D. Hirshleifer, S. Richardson, and S. Teoh (2003), ‘Does
investor misvaluation drive the takeover market?’. Journal of
Finance 61, 725–762.
Dong, M., C. Robinson, and C. Veld (2005), ‘Why individual investors
want dividends’. Journal of Corporate Finance 12, 121–158.
Dorn, D. and G. Huberman (2005), ‘Talk and action: What individual
investors say and what they do’. Review of Finance 9, 437–481.
Doukas, J. A. and P. J. McKnight (2005), ‘European momentum strate-
gies, information diffusion, and investor conservatism’. European
Financial Management 11, 313–338.
References 169
Doukas, J. A. and D. Petmezas (2005), ‘Acquisitions, overconfident
managers and self-attribution bias’. Working Paper, Old Dominion
University.
Duffie, D., L. Saita, and K. Wang (2007), ‘Multi-period corporate
default prediction with stochastic variates’. Journal of Financial Eco-
nomics 83, 635–665.
Dumas, B., A. Kurshev, and R. Uppal (2009), ‘What can rational
investors do about excessive volatility and sentiment fluctuations?’.
Journal of Finance, forthcoming.
Edmans, A., D. Garcia, and O. Norli (2007), ‘Sports sentiment and
stock returns’. Journal of Finance 62, 1967–1998.
Edwards, W. (1968), ‘Conservatism in human information processing’.
In: B. Kleinmutz (ed.): Formal Representation of Human Judgement.
New York: Wiley.
Fama, E. and K. French (1993), ‘Common risk factors in the returns
of bonds and stocks’. Journal of Financial Economics 33, 3–56.
Fama, E. and K. French (1995), ‘Size and book-to-market factors in
earnings and returns’. Journal of Finance 50, 131–155.
Fama, E. and K. French (1996), ‘Multifactor explanations of asset pric-
ing anomalies’. Journal of Finance 51, 55–84.
Fama, E. and K. French (2001), ‘Disappearing dividends: Changing
firm characteristics or lower propensity to pay?’. Journal of Financial
Economics 60, 3–43.
Feng, L. and M. Seasholes (2005), ‘Do investor sophistication and
trading experience eliminate behavioral biases in finance markets?’.
Review of Finance 9, 305–351.
Flynn, J., P. Slovic, and C. K. Mertz (1994), ‘Gender, race, and the
perception of environmental health risks’. Risk Analysis 14, 1101–
1198.
Fogel, S. and T. Berry (2006), ‘The disposition effect and individual
investor decisions: The roles of regret and counterfactual alterna-
tives’. Journal of Behavioral Finance 7, 107–116.
Foldes, L. (1978), ‘Martingale conditions for optimal saving: Discrete
time’. Journal of Mathematical Economics 5, 83–96.
Foldes, L. (2000), ‘Valuation and martingale properties of shadow
prices: An exposition’. Journal of Economic Dynamics and Control
24, 1641–1701.
170 References
Fox, C. and K. See (2003), ‘Belief and preference in decision under
uncertainty’. In: D. Hardman and L. Macchi (eds.): Thinking: Psy-
chological Perspectives on Reasoning, Judgment and Decision Mak-
ing. John Wiley and Sons.
Fracassi, C. and G. Tate (2009), ‘External networking and internal firm
governance’. Working paper (UCLA).
French, K. and J. Poterba (1991), ‘Investor diversification and interna-
tional equity markets’. American Economic Review 81, 222–226.
Frieder, L. (2004), ‘Earnings announcements, order flow, and returns’.
Working Paper, Purdue University.
Frieder, L. and A. Subrahmanyam (2005), ‘Brand perceptions and the
market for common stock’. Journal of Financial and Quantitative
Analysis 40, 57–85.
Froot, K. and E. Dabora (1999), ‘How are stock prices affected by the
location of trade?’. Journal of Financial Economics 53, 189–216.
Gaston-Breton, C. and J. Lejarraga (2009), ‘Biases in learning from
past strategic decisions’. Working paper, Universidad Carlos III de
Madrid.
Genesove, D. and C. Mayer (2001), ‘Loss aversion and seller behavior:
Evidence from the housing market’. Quarterly Journal of Economics
116, 1233–1260.
Gentry, W. and G. Hubbard (2001), ‘Entrepreneurship and household
saving’. NBER Working Paper No. 7894.
Gervais, S., J. B. Heaton, and T. Odean (2003), ‘Overconfidence, invest-
ment policy, and executive stock options’. Working paper, Duke Uni-
versity.
Gilchrist, S., C. Himmelberg, and G. Huberman (2004), ‘Do stock price
bubbles influence corporate investment?’. Journal of Monetary Eco-
nomics 52, 805–827.
Goel, A. and A. Thakor (2008), ‘Overconfidence, CEO selection, and
corporate governance’. Journal of Finance 63, 2737–2784.
Goetzmann, W. and N. Zhu (2005), ‘Rain or shine: Where is the
weather effect?’. European Financial Management 11, 559–578.
Goetzmann, W. N. and A. Kumar (2008), ‘Equity portfolio diversifica-
tion’. Review of Finance 12, 433–463.
References 171
Gollier, C. and B. Salani´e (2006), ‘Individual decisions under risk, risk
sharing and asset prices with regret’. Working paper (University of
Toulouse).
Gompers, P., J. Ishii, and A. Metrick (2003), ‘Corporate governance
and equity prices’. Quarterly Journal of Economics 118, 107–155.
Graham, J. and C. Harvey (2001), ‘The theory and practice of corporate
finance: Evidence from the field’. Journal of Financial Economics 60,
187–243.
Graham, J., C. Harvey, and M. Puri (2009), ‘Managerial attitudes and
corporate actions’. Working Paper, Duke University.
Graham, J., C. Harvey, and S. Rajgopal (2004), ‘The economic impli-
cations of corporate financial reporting’. Journal of Accounting and
Economics 40, 3–73.
Graham, J. and A. Kumar (2006), ‘Dividend preference of retail
investors: Do dividend clienteles exist?’. Journal of Finance 6, 1305–
1336.
Graham, J. R. (2000), ‘How big are the tax benefits of debt?’. Journal
of Finance 55, 1901–1941.
Griffin, D. and A. Tversky (1992), ‘The weighing of evidence and the
determinants of overconfidence’. Cognitive Psychology 24, 411–435.
Griffin, J. and D. Tang (2009), ‘Did subjectivity play a role in CDO
credit ratings?’. Working paper, University of Texas.
Grinblatt, M. and B. Han (2004), ‘Prospect theory, mental accounting
and momentum’. Journal of Financial Economics 78, 311–333.
Grinblatt, M. and M. Keloharju (2001a), ‘How distance, language, and
culture influence stockholdings and trades’. Journal of Finance 56,
1053–1073.
Grinblatt, M. and M. Keloharju (2001b), ‘What makes investors
trade?’. Journal of Finance 56, 589–616.
Grinblatt, M. and M. Keloharju (2009), ‘Sensation seeking, overconfi-
dence, and trading activity’. Journal of Finance 64, 549–578.
Guedes, J. and T. Opler (1996), ‘The determinants of the maturity of
corporate debt issues’. Journal of Finance 51, 1809–1833.
Guedj, I. and D. Scharfstein (2004), ‘Organizational scope and invest-
ment: Evidence from the drug development strategies of biopharma-
ceutical firms’. NBER Working Paper No. w10933.
172 References
Gurkaynak, R., B. Sack, and E. Swanson (2005), ‘The excess sensitivity
of long-term interest rates: Evidence and implications for macroeco-
nomic models’. American Economic Review 95, 425–436.
Hackbarth, D. (2009), ‘Determinants of corporate borrowing: A behav-
ioral perspective’. Journal of Corporate Finance 15, 389–411.
Hamilton, B. (2000), ‘Does entrepreneurship pay? An empirical analysis
of the returns to self-employment’. Journal of Political Economy 108,
604–631.
Han, B. (2008), ‘Investor sentiment and option prices’. Review of Finan-
cial Studies 21, 387–414.
Hansen, L. P. and R. Jagannathan (1991), ‘Implications of security
market data for models of dynamic economies’. Journal of Political
Economy 99, 225–262.
Hare, T., C. Camerer, and A. Rangel (2009), ‘Self-control in decision-
making involves modulation of the vmPFC valuation system’. Sci-
ence 324, 646–648.
Harrison, J. M. and D. Kreps (1978), ‘Speculative investor behavior in
a stock market with heterogeneous expectations’. Quarterly Journal
of Economics 92, 323–336.
Harvey, C. and A. Siddique (2000), ‘Conditional skewness in asset pric-
ing tests’. Journal of Finance 55, 1263–1295.
Haugen, R. A. and N. L. Baker (1996), ‘Commonality in the determi-
nants of expected stock returns’. Journal of Financial Economics 41,
401–439.
He, X. and X. Zhou (2009), ‘Behavioral portfolio choice: an analytical
treatment’. Working paper, Oxford University.
Heath, C., R. Larrick, and J. Klayman (1998), ‘Cognitive repairs: How
organizational practices can compensate for individual shortcom-
ings’. Research in Organizational Behavior 20, 1–37.
Heaton, J. and D. Lucas (2000), ‘Asset pricing and portfolio choice: The
importance of entrepreneurial risk’. Journal of Finance 55, 1163–
1198.
Heaton, J. B. (2002), ‘Managerial optimism and corporate finance’.
Financial Management 31, 33–45.
Henderson, B., N. Jegadeesh, and M. Weisbach (2009), ‘World mar-
kets for raising new capital’. Journal of Financial Economics 82,
63–101.
References 173
Hens, T. and K. Schenk-Hopp´e (2008), Handbook of Financial Markets:
Dynamics and Evolution. Amsterdam: Elsevier.
Hens, T. and M. Vlcek (2005), ‘Does prospect theory explain the dis-
position effect?’. Working paper, University of Zurich.
Hirshleifer, D. (2008a), ‘Investor psychology and asset pricing’. Journal
of Finance 56, 1533–1598.
Hirshleifer, D. (2008b), ‘Psychological bias as a driver of financial reg-
ulation’. European Financial Management 14, 856–874.
Hirshleifer, D. and T. Shumway (2003), ‘Good day sunshine: Stock
returns and the weather’. Journal of Finance 58, 1009–1032.
Hirshleifer, D., A. Subrahmanyam, and S. Titman (2006), ‘Feedback
and the success of irrational traders’. Journal of Financial Economics
81, 311–388.
Hoffmann, A., H. Shefrin, and J. Pennings (2010), ‘Behavioral portfolio
analysis of individual investors’. Working Paper, Maastricht Univer-
sity and Netspar.
Hong, H., T. Lim, and J. Stein (2000), ‘Bad news travels slowly:
Size, analyst coverage, and the profitability of momentum strategies’.
Journal of Finance 55, 265–295.
Hong, H. and J. Stein (1999), ‘A unified theory of underreaction,
momentum trading, and overreaction in asset markets’. Journal of
Finance 54, 2143–2184.
Hong, H. and J. Stein (2003), ‘Differences of opinion, short-sale con-
straints and market crashes’. Review of Financial Studies 16, 487–
525.
Hong, H. G., J. C. Stein, and J. Yu (2007), ‘Simple forecasts and
paradigm shifts’. Journal of Finance 62, 1207–1242.
Hovakimian, A. (2006), ‘Are observed capital structures determined
by equity market timing?’. Journal of Financial and Quantitative
Analysis 41, 221–243.
Huang, R. and J. Ritter (2009), ‘Testing theories of capital structure
and estimating the speed of adjustment’. Journal of Financial and
Quantitative Analysis 44, 237–271.
Huberman, G. (2001), ‘Familiarity breeds investment’. Review of
Financial Studies 14, 659–680.
174 References
Huddart, S., M. Lang, and M. Yetman (2009), ‘Volume and price pat-
terns around a stock’s 52-week highs and lows: Theory and evidence’.
Management Science 55, 16–31.
Hvidkjaer, S. (2005), ‘Small trades and the cross-section of stock
returns’. Working paper, University of Maryland.
Hvidkjaer, S. (2006), ‘A trade-based analysis of momentum’. Review of
Financial Studies 19, 457–491.
Ikenberry, D., J. Lakonishok, and T. Vermaelen (1995), ‘Market under-
reaction to open market share repurchases’. Journal of Financial
Economics 39, 181–208.
Ikenberry, D., J. Lakonishok, and T. Vermaelen (2000), ‘Stock repur-
chases in Canada: Performance and strategic trading’. Journal of
Finance 55, 2373–2398.
Ingersoll, J. (2008), ‘Non-Monotonicity of the Tversky-Kahneman
probability weighting function: A cautionary note’. European Finan-
cial Management 14, 385–390.
Ishii, J. and Y. Xuan (2009), ‘Acquirer-target social ties and merger
outcomes’. Working Paper, Stanford University.
Jackwerth, J. C. (2000), ‘Recovering risk aversion from options prices
and realized returns’. Review of Financial Studies 13, 433–451.
Janis, I. (1972), Victims of Groupthink: A Psychological Study of For-
eign Policy Decisions and Fiascoes. Houghton Mifflin.
Jenter, D. (2005), ‘Market timing and managerial portfolio decisions’.
Journal of Finance 60, 1903–1949.
Jones, C. and O. Lamont (2002), ‘Short-sale constraints and stock
returns’. Journal of Financial Economics 66, 207–239.
Jouini, E. and C. Napp (2006), ‘Heterogeneous beliefs and asset pricing
in discrete time’. Journal of Economic Dynamics and Control 30,
1233–1260.
Jouini, E. and C. Napp (2007), ‘Consensus consumer and intertemporal
asset pricing with heterogeneous beliefs’. Review of Economic Studies
74, 1149–1174.
Kahneman, D., J. Knetsch, and R. Thaler (1991), ‘Fairness as a
constraint on profit seeking entitlements in the market’. In: R.
Thaler (ed.): Quasi-Rational Economics. New York: Russell Sage
Foundation.
References 175
Kahneman, D., P. Slovic, and A. Tversky (1982), Judgment Under
Uncertainty: Heuristics and Biases. Cambridge: Cambridge Univer-
sity Press.
Kahneman, D. and A. Tversky (1973), ‘On the psychology of predic-
tion’. Psychological Review 80, 237–251.
Kahneman, D. and A. Tversky (1979), ‘Prospect theory: An analysis
of decision making under risk’. Econometrica pp. 263–291.
Kahneman, D. and A. Tversky (1982), ‘The psychology of preferences’.
Scientific American 246, 160–173.
Kamstra, M., L. Kramer, M. Levi, and T. Wang (2009), ‘Time-varying
preferences and SAD: Evidence from an asset pricing model’. Work-
ing paper, University of Toronto.
Kamstra, M. J., L. A. Kramer, and M. D. Levi (2000), ‘Losing sleep
at the market: The daylight-savings anomaly’. American Economic
Review 90, 1005–1011.
Kausar, A. and R. Taffler (2006), ‘Testing behavioural finance models
of market under and overreaction: Do they really work?’. Working
Paper (University of Edinburgh.
Kaustia, M. (2004), ‘Market-wide impact of the disposition effect: Evi-
dence from IPO trading volume’. Journal of Financial Markets 7,
207–235.
Kaustia, M. (2009), ‘Prospect theory and the disposition effect’. Jour-
nal of Financial and Quantitative Analysis, forthcoming.
Kaustia, M. and S. Kn¨ upfer (2009), ‘Learning from the outcomes
of others: Evidence from the stock market’. Working paper, Aalto
University.
Klayman, J. and Y. Ha (1987), ‘Confirmation, disconfirmation and
information in hypothesis testing’. Psychological Review 94, 211–228.
Klein, P. (2001), ‘Were the acquisitive conglomerates inefficient?’.
RAND Journal of Economics 32, 745–761.
Knoch, D., F. Schneiderb, D. Schunk, M. Hohmann, and E. Fehr (2009),
‘Disrupting the prefrontal cortex diminishes the human ability to
build a good reputation’. PNAS 106, 20895–20899.
Knutson, B. and C. Kuhnen (2009), ‘The impact of affect on beliefs,
preferences and financial decisions’. Journal of Financial and Quan-
titative Analysis, forthcoming.
176 References
Knutson, B., C. Kuhnen, G. E. Wimmer, and P. Winkielman (2008),
‘Nucleus accumbens activation mediates the influence of reward cues
on financial risk-taking’. NeuroReport 19, 509–513.
Kr¨ aussl, R., A. Lucas, D. Rijsbergen, P. van der Sluis, and E. Vrugt
(2009), ‘Washington meets wall street: A closer examination of the
presidential cycle puzzle’. Working paper, University of Amsterdam.
Kumar, A. (2009), ‘Who gambles in the stock market?’. Journal of
Finance 64, 1889–1933.
Kuran, T. and C. Sunstein (1999), ‘Availability cascades and risk reg-
ulation’. Stanford Law Review 51, 683–768.
Kurz, M. (1997), Endogenous Economic Fluctuations: Studies in the
Theory of Rational Beliefs. Studies in Economic Theory No. 6,
Springer-Verlag: Berlin and New York.
Kyle, A. and F. A. Wang (1997), ‘Speculation duopoly with agreement
to disagree: Can overconfidence survive the market test?’. Journal of
Finance 52, 2073–2090.
La Porta, R., J. Lakonishok, A. Shleifer, and R. Vishny (1997), ‘Good
news for value stocks: further evidence on market efficiency’. Journal
of Finance 49, 1541–1578.
Lakonishok, J., I. Lee, N. Pearson, and A. Poteshman (2007), ‘Option
market activity’. The Review of Financial Studies 20, 813–857.
Lakonishok, J., A. Shleifer, and R. W. Vishny (1994), ‘Contrarian
investment, extrapolation and risk’. Journal of Finance 49, 1541–
1578.
Lamont, O. (2000), ‘Investment plans and stock returns’. Journal of
Finance 55, 2719–2745.
Landier, A. and D. Thesmar (2009), ‘Financial contracting with opti-
mistic entrepreneurs: Theory and evidence’. Review of Financial
Studies 22, 117–150.
Leary, M. (1983), ‘Social anxiousness: The construct and its measure-
ment’. Journal of Personality Assessment 47(1), 66–75.
Lee, C., A. Shleifer, and R. Thaler (1991), ‘Investor sentiment and the
closed-end fund puzzle’. Journal of Finance 46, 75–110.
Lehenkari, M. (2008), ‘The disposition effect: Underlying mechanisms
and implications for individual investors’. Working paper, University
of Oulu, Finland.
References 177
Lennox, R. and R. Wolfe (1984), ‘Revision of the self-monitoring scale’.
Journal of Personality and Social Psychology 46, 1349–1364.
LeRoy, S. and R. Porter (1981), ‘The present-value relation: Tests based
on implied variance bounds’. Econometrica 49, 97–113.
Lintner, J. (1969), ‘The aggregation of investors’ diverse judgments and
preferences in pure competitive markets’. Journal of Financial and
Quantitative Analysis 4, 347–400.
Lobe, S. and A. H¨ olzl (2008), ‘Happy savers, happy issuer: The UK
lottery bond’. Center of Finance, University of Regensburg, Forth-
coming Revue Bancaire et Financi`ere - Bank- en Financiewezen J.E.
Loomes, G. and R. Sugden (1982), ‘Regret theory: An alternative
theory of rational choice under uncertainty’. Economic Journal 92,
805–824.
Lopes, L. (1987), ‘Between hope and fear: The psychology of risk’.
Advances in Experimental Social Psychology 20, 255–295.
Lopes, L. L. and G. C. Oden (1999), ‘The role of aspiration level in
risk choice: A comparison of cumulative prospect theory and SP/A
theory’. Journal of Mathematical Psychology 43, 286–313.
Loughran, T. and J. Ritter (1995), ‘The new issues puzzle’. Journal of
Finance 50, 23–51.
Loughran, T. and J. Ritter (2002), ‘Why don’t issuers get upset about
leaving money on the table in IPOs?’. Review of Financial Studies
15, 413–443.
Loughran, T., J. Ritter, and K. Rydqvist (1994), ‘Initial public offer-
ings: International insights’. Pacific-Basin Finance Journal 2, 165–
199.
Loughran, T. and A. Vijh (1997), ‘Do long-term shareholders benefit
from corporate acquisitions?’. Journal of Finance 52, 1765–1790.
Malmendier, U. and G. Tate (2005), ‘CEO overconfidence and corpo-
rate investment’. Journal of Finance 60, 2661–2700.
Malmendier, U. and G. Tate (2008), ‘Who makes acquisitions? CEO
overconfidence and the market’s reaction’. Journal of Financial Eco-
nomics 89, 20–43.
Marsh, P. (1982), ‘The choice between equity and debt: An empirical
study’. Journal of Finance 37, 121–144.
178 References
McConnell, J. and E. Schwartz (1992), ‘The origin of LYONs: A case
study in financial innovation’. Journal of Applied Corporate Finance
pp. 40–47.
Mehra, R. and E. Prescott (1985), ‘The equity premium: A puzzle’.
Journal of Monetary Economics 15, 145–161.
Merrow, E., K. Phillips, and C. Myers (1981), Understanding Cost
Growth and Performance Shortfalls in Pioneer Process Plants. Santa
Monica, CA: Rand.
Merton, R. C. (1971), ‘Optimum consumption and portfolio rules
in a continuous-time model’. Journal of Economic Theory 3, 373–
413.
Miller, E. (1977), ‘Risk, uncertainty, and divergence of opinion’. Journal
of Finance 32, 1151–1168.
Mitchell, O., G. Mottola, S. Utkus, and T. Yamaguchi (2006), ‘The
inattentive participant: Portfolio trading behavior in 401(k) plans’.
Working Paper, Wharton.
Moskowitz, T. and A. Vissing-Jorgensen (2002), ‘The returns to
entrepreneurial investment: A private equity premium puzzle?’.
American Economic Review 92, 745–778.
Muelbroek, L. (1992), ‘An empirical analysis of illegal insider trading’.
Journal of Finance 47, 1661–1699.
Nagel, S. (2005), ‘Short sales institutional investors and the cross-
section of stock returns’. Journal of Financial Economics 78,
277–309.
Odean, T. (1998a), ‘Are investors reluctant to realize their losses?’.
Journal of Finance 53, 1775–1798.
Odean, T. (1998b), ‘Volume, volatility, price, and profit when all traders
are above average’. Journal of Finance 53, 1887–1934.
Odean, T. (1999), ‘Do investors trade too much?’. American Economic
Review 89, 1279–1298.
Panageas, S. (2003), ‘Speculation, overpricing, and investment: Theory
and empirical evidence’. Working Paper, MIT.
Payne, J. (2005), ‘It is whether you win or lose: The importance of the
overall probabilities of winning or losing in risky choice’. The Journal
of Risk and Uncertainty 30, 5–19.
PBS (2009), ‘The warning’. Frontline television documentary.
References 179
Peltzman, S. (1976), ‘Toward a more general theory of regulation’.
Journal of Law and Economics 19, 211–240.
Peyer, U. and T. Vermaelen (2009), ‘The nature and persistence of
buyback anomalies’. Review of Financial Studies 22, 1693–1745.
Polk, C. and P. Sapienza (2004), ‘The real effects of investor sentiment’.
NBER working paper no. 10563.
Polkovnichenko, V. (2002), ‘Household portfolio diversification: A case
for rank-dependent preferences’. Review of Financial Studies 18,
1467–1501.
Posner, R. A. (2009), A Failure of Capitalism: The Crisis of ’08 and
the Descent into Depression. Cambridge, MA: Harvard University
Press.
Poteshman, A. (2001), ‘Underreaction, overreaction and increasing mis-
reaction to information in the options market’. Journal of Finance
56, 851–876.
Poteshman, A. and V. Serbin (2003), ‘Clearly irrational financial mar-
ket behavior: Evidence from the early exercise of exchange traded
stock options’. Journal of Finance 58, 37–70.
Poti, V. (2006), ‘The coskewness puzzle in the cross-section of industry
portfolio returns’. Working paper, Dublin City University Business
School.
Puri, M. and D. Robinson (2007), ‘Optimism and economic choice’.
Journal of Financial Economics 86, 71–99.
Puri, M. and D. Robinson (2008), ‘Who are entrepreneurs and why do
they behave that way?’. Working Paper, Duke University.
Rather, D. (2009), ‘Sold!’. Dan Rather Reports, Episode Number: 407,
February 17.
Rau, P. and T. Vermaelen (1998), ‘Glamour, value and the post-
acquisition performance of acquiring firms’. Journal of Financial
Economics 49, 223–253.
Ravenscraft, D. and F. Scherer (1987), Mergers, Sell-Offs, and Eco-
nomic Efficiency. Washington, DC: Brookings Institution.
Richardson, S. and R. Sloan (2003), ‘External financing and future
stock returns’. Working paper, University of Pennsylvania.
Rieger, M. (2008), ‘Probability misestimation and preferences in finan-
cial investment decision’. Working Paper, University of Zurich.
180 References
Roll, R. (1986), ‘The hubris hypothesis of corporate takeovers’. Journal
of Business 59, 197–216.
Rosenberg, J. and R. Engle (2004), ‘Empirical pricing kernels’. Journal
of Financial Economics 64, 341–372.
Ross, L., D. Green, and P. House (1977), ‘The ‘false consensus effect’:
An egocentric bias in social perception and attribution processes’.
Journal of Experimental Social Psychology 13, 279–301.
Ross, S. (2005), Neoclassical Finance. Princeton, NJ: Princeton Uni-
versity Press.
Samuelson, W. and R. Zeckhauser (1988), ‘Status quo bias in decision
making’. Journal of Risk and Uncertainty 1, 7–59.
Saunders, E. M. J. (1993), ‘Stock prices and Wall Street weather’.
American Economic Review 83, 1337–1345.
Scarpetta, S., P. Hemmings, T. Tressel, and J. Woo (2002), ‘The role of
policy and institutions for productivity and firm dynamics: Evidence
from micro and industry data’. OECD working paper.
Scheier, M. F. and C. S. Carver (1985), ‘Optimism, coping and health:
Assessment and implications of generalized outcome expectancies’.
Health Psychology 4, 219–247.
Scheinkman, J. and W. Xiong (2003), ‘Overconfidence and speculative
bubbles’. Journal of Political Economy 111, 1183–1219.
Seru, A., T. Shumway, and N. Stoffman (2009), ‘Learning by trading’.
Review of Financial Studies, forthcoming.
Seyhun, H. N. (1992), ‘Why does aggregate insider trading predict
future stock returns?’. The Quarterly Journal of Economics 107,
1303–1331.
Shefrin, H. (2001), ‘Behavioral corporate finance’. Journal of Applied
Corporate Finance 14, 113–124.
Shefrin, H. (2005), Behavioral Corporate Finance. Burr Ridge:
McGraw-Hill.
Shefrin, H. (2007), ‘Behavioral portfolio selection’. In: Encyclopedia of
Quantitative Finance. Wiley, forthcoming.
Shefrin, H. (2008a), A Behavioral Approach to Asset Pricing. Boston:
Elsevier Academic Press, 2nd edition.
Shefrin, H. (2008b), ‘Behavioral explanations for why individ-
ual investors find dividends attractive’. In: H. K. Baker (ed.):
References 181
Chapter 13, the Blackwell Companion to Dividends and Dividend
Policy. Blackwell.
Shefrin, H. (2008c), ‘Risk and return in behavioral SDF-based asset
pricing models’. Journal of Investment Management 6, 4–22.
Shefrin, H. (2009), ‘How psychological pitfalls generated the global
financial crisis’. In: L. Siegel (ed.): Understanding the Global Finan-
cial Crisis.
Shefrin, H. (2010), ‘Insights into the psychological profiles of
entrepreneurs’. In: R. Yazdipour (ed.): Advances in Entrepreneurial
Finance. New York, NY: Springer, forthcoming.
Shefrin, H. and M. Statman (1984), ‘Explaining investor preference for
cash dividends’. Journal of Financial Economics 13, 253–282.
Shefrin, H. and M. Statman (1985), ‘The disposition to sell winners
too early and ride losers too long: theory and evidence’. Journal of
Finance 40, 777–790.
Shefrin, H. and M. Statman (1992), Ethics, Fairness, Efficiency, and
Financial Markets. Charlottesville, VA: CFA Institute Publications.
Shefrin, H. and M. Statman (1993a), ‘Behavioral aspects of the design
and marketing of financial products’. Financial Management 22,
123–134.
Shefrin, H. and M. Statman (1993b), ‘Ethics, fairness and efficiency in
financial markets’. Financial Analysts Journal pp. 21–29.
Shefrin, H. and M. Statman (1994), ‘Behavioral capital asset pric-
ing theory’. Journal of Financial and Quantitative Analysis 29,
323–349.
Shefrin, H. and M. Statman (2000), ‘Behavioral portfolio theory’. Jour-
nal of Financial and Quantitative Analysis 35, 127–151.
Shefrin, H. and M. Statman (2009), ‘Striking regulatory irons while
hot’. Journal of Investment Management 7, 29–42.
Shefrin, H. and R. Thaler (1988), ‘The behavioral life cycle hypothesis’.
Economic Inquiry 24, 609–643.
Shiller, R. (1978), ‘The volatility of long-term interest rates and the
expectations models of the term structure’. Journal of Political Econ-
omy 87, 1190–1219.
Shiller, R. (1981), ‘Do stock prices move too much to be justified by
subsequent changes in dividends?’. American Economic Review 71,
421–436.
182 References
Shiller, R. (1999), ‘Human behavior and the efficiency of the financial
system’. In: J. Taylor and M. Woodford (eds.): Handbook of Macroe-
conomics. Amsterdam: Elsevier.
Shiller, R. (2005), Irrational Exuberance. Princeton: Princeton Univer-
sity Press.
Shiller, R. and J. Pound (1989), ‘Survey evidence on the diffusion
of interest and information among investors’. Journal of Economic
Behavior and Organization 12, 46–66.
Shleifer, A. (2000), Inefficient Markets. Oxford: Oxford University
Press.
Shleifer, A. and R. Vishny (2003), ‘Stock market driven acquisitions’.
Journal of Financial Economics 70, 295–312.
Sloan, R. (1996), ‘Do stock prices fully reflect information in accruals
and cash flows about future earnings?’. The Accounting Review 71,
289–316.
Slovic, P. (1972), ‘Psychological study of human judgment: Implications
for investment decision making’. Journal of Finance 27, 779–799.
Smith, V. (2003), ‘Constructivist and ecological rationality in eco-
nomics’. American Economic Review 93, 465–507.
Snyder, M. (1974), ‘Self-monitoring of expressive behavior’. Journal of
Personality and Social Psychology 30, 526–537.
Sorescu, S. and A. Subrahmanyam (2006), ‘The cross section of analyst
recommendations’. Journal of Financial and Quantitative Analysis
41, 139–168.
Spalt, O. (2008), ‘Probability weighting and employee stock options’.
Working paper, University of Mannheim.
Stasser, G., L. Taylor, and C. Hanna (1989), ‘Information sampling
in structured and unstructured discussions of three- and six-person
groups’. Journal of Personality and Social Psychology 57, 67–78.
Stasser, G. and W. Titus (1985), ‘Pooling of unshared information in
group decision making: Biased information sampling during discus-
sion’. Journal of Personality and Social Psychology 48, 1467–1478.
Statman, M. (2002), ‘Lottery players: Stock traders’. Financial Ana-
lysts Journal 58, 14–21.
Statman, M. (2008), ‘The expressive nature of socially responsible
investors’. Journal of Financial Planning pp. 40–46.
References 183
Statman, M. (2009), ‘Regulating financial markets: Protecting us from
ourselves and others’. Financial Analysts Journal 65, 22–31.
Statman, M. and D. Caldwell (1987), ‘Applying behavioral finance to
capital budgeting: Project termination’. Financial Management pp.
7–15.
Statman, M. and J. Sepe (1989), ‘Project termination announcements
and the market value of the firm’. Financial Management pp. 1–8.
Statman, M., S. Thorley, and K. Vorkink (2006), ‘Investor overconfi-
dence and trading volume’. Review of Financial Studies 19, 1531–
1565.
Statman, M. and T. Tyebjee (1985), ‘Optimistic capital budgeting
forecasts: An experiment’. Financial Management pp. 27–33.
Stein, J. (1989), ‘Overreactions in the options market’. Journal of
Finance 44, 1011–1022.
Stein, J. (1996), ‘Rational capital budgeting in an irrational world’.
Journal of Business 69, 429–455.
Stigler, G. (1971), ‘The theory of economic regulation’. The Bell Jour-
nal of Economics and Management Science 2, 3–21.
Subrahmanyam, A. (2007), ‘Behavioural finance: A review and synthe-
sis’. European Financial Management 14(1), 12–29.
Sullivan, R. (2009), ‘Governance: Travel and destinations’. Financial
Analysts Journal 65, 6–10.
Taleb, N. (2007), The Black Swan. New York: Random House.
Taylor, J. (2009), ‘The financial crisis and the policy responses: An
empirical analysis of what went wrong’. NBER Working Paper No.
14631.
Teoh, S. H., I. Welch, and T. J. Wong (1998a), ‘Earnings manage-
ment and the long-run market performance of initial public offerings’.
Journal of Finance 53, 1935–1974.
Teoh, S. H., I. Welch, and T. J. Wong (1998b), ‘Earnings management
and the underperformance of seasoned equity offerings’. Journal of
Financial Economics 50, 63–99.
Thaler, R. (1993), Advances in Behavioral Finance: Volume I. New
York: Russell Sage Foundation.
Thaler, R. and E. Johnson (1990), ‘Gambling with the house money
and trying to break even: The effects of prior outcomes on risky
choice’. Management Science 36, 643–660.
184 References
Titman, S., K. C. J. Wei, and F. Xie (2004), ‘Capital investments and
stock returns’. Journal of Financial and Quantitative Analysis, Inter-
national Review of Finance 9, 111–131.
Tiwana, A., J. Wang, M. Keil, and P. Ahluwalia (2007), ‘The bounded
rationality bias in managerial valuation of real options: Theory and
evidence from IT projects’. Decision Sciences 38, 157–181.
Todd, P. and G. Gigerenzer (2000), ‘Simple heuristics that make us
smart’. Behavioral and Brain Sciences 23, 727–780.
Trivers, R. (1985), Social Evolution. Benjamin/Cummings, Menlo
Park.
Trivers, R. (1991), ‘Deceit and Self-deception’. In: R. Robinson and L.
Tiger (eds.): Man and Beast Revisited. Washington D.C.: Smithso-
nian Press.
Tversky, A. and D. Kahneman (1992), ‘Advances in prospect the-
ory: Cumulative representation of uncertainty’. Journal of Risk and
Uncertainty 5, 297–323.
Tversky, A. and D. J. Koehler (1994), ‘Support theory: A nonexten-
sional representation of subjective probability’. Psychological Review
101, 547–567.
Wadhwa, V., R. Aggarwal, K. Holly, and A. Salkever (2009), The
Anatomy of an Entrepreneur Family Background and Motivation.
Ewing Marion Kauffman Foundation of Entrepreneurship.
Wang, M., M. Rieger, and T. Hens (2009), ‘International test on risk
attitudes’. Working paper, University of Zurich.
Watson, D. and L. Clark (1988), ‘Development and validation of brief
measures of positive and negative affect: The PANAS scales’. Journal
of Personality and Social Psychology 54, 1063–1070.
Weinstein, N. (1980), ‘Unrealistic optimism about future life events’.
Journal of Personality and Social Psychology 39, 806–820.
Welch, I. (2000), ‘Views of financial economists on the equity premium
and on professional controversies’. Journal of Business 73, 501–537.
Xiong, W. and H. Yan (2009), ‘Heterogeneous expectations and bond
markets’. Review of Financial Studies, forthcoming.
Zhang, X. F. (2006), ‘Information uncertainty and analyst forecast
behavior’. Journal of Finance 61, 105–136.

4 Behavioralizing Portfolio Selection Theory 4.1 4.2 Preference for Positively Skewed Returns: Full Maximization Quasi-Maximization

61 62 70 82 83 88 91 92 100 102 112 113 116 119 122

5 Behavioralizing Asset Pricing Theory 5.1 5.2 5.3 5.4 5.5 5.6 Stochastic Discount Factor How Markets Aggregate Investor Attributes Aggregation and the Shape of Sentiment Risk and Return Long-run Fitness Pricing Dynamics

6 Behavioralizing Corporate Finance 6.1 6.2 6.3 6.4 Open Questions Raised by Baker et al. Managerial Psychology and Decisions Social Networks Entrepreneurs

7 Behavioralizing the Approach to Financial Market Regulation 7.1 7.2 7.3 7.4 Dynamic Tug-of-War Biased Cost–Benefit Assessment Regulation and Financial Crises Differences in Perspective

130 131 140 142 156 158 161

8 Concluding Remarks References

Electronic copy available at: http://ssrn.com/abstract=1597934

Foundations and Trends R in Finance Vol. 4, Nos. 1–2 (2009) 1–184 c 2010 H. Shefrin DOI: 10.1561/0500000030

Behavioralizing Finance
Hersh Shefrin
Mario L. Belotti Professor of Finance, Leavey School of Business, Santa Clara University, Santa Clara, CA 95053, USA hshefrin@scu.edu

Abstract
Finance is in the midst of a paradigm shift, from a neoclassical based framework to a psychologically based framework. Behavioral finance is the application of psychology to financial decision making and financial markets. Behavioralizing finance is the process of replacing neoclassical assumptions with behavioral counterparts. This monograph surveys the literature in behavioral finance, and identifies both its strengths and weaknesses. In doing so, it identifies possible directions for behavioralizing the frameworks used to study beliefs, preferences, portfolio selection, asset pricing, corporate finance, and financial market regulation. The intent is to provide a structured approach to behavioral finance in respect to underlying psychological concepts, formal framework, testable hypotheses, and empirical findings. A key theme of this monograph is that the future of finance will combine realistic assumptions from behavioral finance and rigorous analysis from neoclassical finance.

coherence. Among its main strengths is its use of assumptions based on findings from the psychology literature about how people deviate from fully rational behavior. In this monograph I suggest that finance is moving to a new paradigm that will combine structural features from neoclassical finance and realistic assumptions from behavioral finance. Viewed as a process. Among its main weaknesses is its reliance on the unrealistic assumption that all decision makers are fully rational.1 Introduction Viewed as a field. I refer to this process as “behavioralizing finance. and consistency. behavioral finance is about the transformation of the financial paradigm from a neoclassical based framework to a psychologically based framework. 2 . To develop this position. and relate it to the issues associated with structure. rigorous framework. I describe the outlines of what such a synthesis entails. behavioral finance is the application of psychology to financial decision making and financial markets. Among its main strengths is its systematic. Among its main weaknesses is the reliance by its proponents on an ad hoc collection of models that lack mutual consistency and a unifying structure. Behavioral finance also has both strengths and weaknesses.” Neoclassical finance has both strengths and weaknesses.

(2009). For years. some behavioral economists are beginning to develop models that are as rigorous as their neoclassical counterparts. rather than the cross-section itself. However. The behavioral result contends that irrational investors can survive in the long run because they take on more risk than rational investors. The second shift features behavioral economists developing a systematic. see Ross (2005). this situation is changing slowly as some of these theorems have come to be replicated in the work of more traditional economists. At the same time.3 The behavioralization of finance involves intellectual shifts by two groups. For example. However. such as Jouini and Napp (2006). (2009). Some neoclassical economists have begun to develop behavioral models. the behavioral pricing kernel framework. this contention is at odds with theorems in the general literature about long-run survivability. systematic approach. My own experience has been that both neoclassical economists and behavioral economists have been resistant to the proposal that each needs to move to a common middle ground where psychological assumptions and rigor come together. The first shift features neoclassical economists explicitly incorporating psychological elements into their models. the tendency to overweight evidence that confirms one’s views relative to evidence that disconfirms those views. rigorous framework. These shifts are taking place against a historical backdrop in which financial economists from both camps have displayed some of the same psychological traits that are the focus of behavioral finance. one well-known behavioral result might well be incorrect. whose formal framework shares much in common with Dumas et al. . Two prominent examples are Jouini and Napp (2006) and Dumas et al. For a sense of neoclassical resistance. Shifts necessary for the behavioralization of finance are underway. Take for example. In addition. arguing that a sentiment-based pricing kernel framework is insufficiently behavioral. Likewise. most of the current behavioral explanations for the crosssection of returns rely on models that feature a single risky security. neoclassical economists have tended to claim that the behavioral pricing kernel theorems discussed in Section 5 of this monograph are wrong. Confirmation bias may well be at the top of the list. which is an approach I favor. behavioral economists have resisted the unified. A good example is Xiong and Yan (2009).

In the remainder of the monograph. Although the Journal of Finance published the first formal paper in behavioral finance in 1972 (see Slovic (1972).1 Brief History Behavioral finance is relatively new as a subject of study. These foundations are the subject of Section 2. A behavioral example is Shefrin and Statman (1994). the common middle ground involves models that identify how markets aggregate heterogeneous investors’ beliefs. My approach to this monograph is to describe the highlights of the behavioral finance literature. Two neoclassical examples are Detemple and Murthy (1994) and Kurz (1997). When it comes to asset pricing. the behavioral assumptions specify that the source of the heterogeneity is a series of heuristics that give rise to biased beliefs. financial economists did not begin to . For example. or which those surveys overlooked for one reason or another. My first objective is to discuss works which have emerged since the past surveys appeared. as expressed through prior surveys.4 Introduction In the past. Kurz describes the agents in his framework as holding “rational beliefs. moving from the neoclassical position to the behavioral position requires a philosophical leap. 1. I write this survey as some neoclassical economists and behavioral economists have begun the process of moving toward a common middle ground. both neoclassical and behavioral economists have chosen to exclude papers relating to the behavioral pricing kernel at their respective NBER meetings. the neoclassical assumptions involve rational behavior by all agents. some of the models employed by neoclassical economists are structurally similar to those employed by behavioral economists. The modelig distance between the two approaches is short. Notably.” In contrast. The contents of these sections represent my thinking about the nature of the common middle ground which brings together the best of neoclassical finance and behavioral finance. In some respects. However. My second objective is to present some ideas about trends toward a unifying framework for behavioral finance which captures some of the rigor in neoclassical finance. and to identify some of the weaknesses of this literature. I have two main objectives.

then President-elect of the American Finance Association. and myself. a formal framework. testable hypotheses. Subrahmanyam (2007). Amos Tversky. and Andrei Shleifer. invited Meir Statman and me to organize the first behavioral session at the annual meetings. Hirshleifer (2008a). Barberis and Thaler (2003). At the session Statman presented our paper on the disposition effect. The first published work in behavioral finance by economists was Shefrin and Statman (1984). Larry Summers. In December 1984. At one extreme you will still find the view expressed by some that behavioral finance is nothing more than a collection of findings about anomalies relative to the efficient market hypothesis. Fischer Black. Despite the beginnings of a convergence that I mentioned above. In addition to De Bondt. but overall still lacks coherence and consistency. and De Bondt et al. (2008) survey the general area. which ultimately led to the NBER Behavioral Finance program. Paul Andreassen. Between the extremes are views that acknowledge that behavioral finance possesses some structure and can be credited with some successes. Robert Shiller. Shefrin and Statman (1985). Subsequently. . Even then. currently there is a wide spectrum of views about behavioral finance. At the other extreme you will find the view that the behavioral approach already constitutes a coherent paradigm. first at the Sloan Foundation. In the last decade several surveys of behavioral finance have been written. the main participants at the Sloan and Sage Foundation included Daniel Kahneman. Thaler collected many of the papers written by members of this group into an edited collection: see Thaler (1993). one pertaining to irrationality in investor behavior and the other to inefficiency in asset pricing. and De Bondt presented his work with Richard Thaler on overreaction.1. it took more than a decade for the behavioral approach to gain traction. Richard Roll. with a systematic approach involving psychological concepts. and then later at the Sage Foundation. Thaler. Statman. David Dreman. and empirical findings. the late Fischer Black. These two papers set the stage for two main streams in the behavioral finance literature. Thaler organized a behavioral research group.1 Brief History 5 apply the concepts pioneered by Slovic and other psychologists working in behavioral decision making until the early 1980s. DeBondt and Thaler (1985).

6 Introduction while Baker et al. Its strengths emanate from basing its foundations on a rich literature in psychology. . and the value maximizing agency theory approach to corporate finance. and the time has come to move in the direction of increased rigor and discipline. as behavioral finance is new. (2007b) survey behavioral corporate finance. mean-variance portfolios. and lacking in discipline. thereby stifling major new insights. These are: mean-variance portfolios. it seems to me that many proponents of behavioral finance have been excessively resistant to rigorous analysis. asset pricing theory. Section 2 highlights the foundations of behavioral finance by surveying a sequence of surveys. In contrast. or more broadly. ad hoc. Instead. Any discussion of paradigm transformation would be incomplete without addressing how behavioral assumptions will impact the major building blocks lying at the heart of the neoclassical framework. and it might be a mistake to narrow its range of inquiry too early. 1. the efficient market hypothesis. There is little point in fully replicating what can already be found in those surveys. There is no generally accepted coherent behavioral counterparts to the workhorses of neoclassical finance. my intent in this monograph is to summarize the key features of those surveys. and relate these features to the process of behavioralizing finance. In one sense the lack of discipline is a positive. unsystematic. the Black–Scholes option pricing formula. However. much of the literature in behavioral finance is scattered. and the Modigliani–Miller principle. This monograph will describe how the substitution of neoclassical assumptions with behavioral assumptions will impact these building blocks. Its weaknesses emanate from an overall approach that has been piecemeal. the behavioralization of finance. The remaining five sections focus on trends relating to a more rigorous approach to behavioral finance.2 Organization of Monograph: Foundations and Trends I have organized this monograph to highlight both the strengths and weaknesses of the behavioral approach. the capital asset pricing model.

SP/A theory. and sets the stage for identifying both the strengths and weaknesses of the behavioral approach. regret. overconfidence. Among the preference-related concepts are prospect theory. These concepts divide naturally into beliefs and preferences. Section 4 deals with behavioral portfolio selection. and representativeness. and realization decisions are influenced by a phenomenon known as “the disposition effect. Section 3 introduces the key psychological concepts used in behavioral finance.1. The literature on behavioral asset pricing theory is eclectic. as seen through the eyes of some of its key contributors. and features a wide variety of models. centered on the concept of sentiment. The section summarizes the major contributions to behavioral finance. Beliefs pertain to the manner in which investors’ judgments about risk and return reflect bias. gambler’s fallacy. Among the beliefrelated psychological concepts discussed in the section are unrealistic optimism. in which purchasing decisions are influenced by salience or attention. and exhibit a taste for positive skewness. Section 5 describes how behavioral portfolio choices impact asset pricing. Preferences pertain to the manner in which investors frame and evaluate cash flow representations. Behavioral beliefs involve a series of biases such as hot hand fallacy. and describes findings from tests of these hypotheses. Many of the models in the behavioral asset pricing literature were developed to analyze overreaction and underreaction in financial . The remaining sections provide an opportunity to present some thoughts on what a systematic behavioral approach might entail.” The disposition effect is the tendency to sell winners too early and ride losers too long in all months of the year except December. This discussion provides an overview of the field. and self-control. Behavioral preferences also explain how investors segment their portfolios into risk layers. Framing effects lead many investors to adopt a piecemeal approach to portfolio selection. The section describes a series of hypotheses about the preference for dividend-paying stocks that stem from the behavioral finance literature. both of which reflect human psychology.2 Organization of Monograph: Foundations and Trends 7 Chapters 2–8 Section 2 is a survey of recent surveys of behavioral finance. and home bias.

showing how these features combine to impact the mean-variance frontier. the impact of group composition on collective corporate financial decisions and corporate value. The section describes how capture theory. especially issues pertaining to the global financial crisis that erupted in 2008. investment banks. In the section I focus on three themes that strike me as areas providing fertile ground for future work. perceptions of fairness. and the manner in which psychological characteristics of entrepreneurs impact the choices they make about risk and return. behavioral preferences. equity prices. The three areas are the study of how specific individual traits impact corporate financial decisions. Section 6 discusses behavioral corporate finance. In this respect. more systematic approach to asset pricing. With the intent of describing a less eclectic. the discussion in the section emphasizes the latter two applications. the section focuses on a behavioral SDF-based approach. corporate governance. This approach brings together the powerful asset pricing tools favored by neoclassical asset pricing theorists and the realistic assumptions favored by behavioral asset pricing theorists. Issues which are more contemporary involve the regulation of financial derivatives. and the credit card industry. the Securities Exchange Act of 1934. Sarbanes– Oxley. . ratings agencies. The literature on behavioral corporate finance involves the study of how psychology impacts capital budgeting. and regulations covering the Savings & Loan industry. the Securities Act of 1933. the term structure of interest rates. capital structure. Behavioral corporate finance is concerned with the manner in which behavioral beliefs. Examples of past regulation are merit regulation (“blue sky laws”). and inefficient prices impact the corporate financial decisions made by managers. hedge funds. The section points out the importance of psychological forces in the shaping of the major legislation underlying financial market regulation in the United States. and mergers and acquisitions. Section 7 discusses the role which psychological factors play in the structure of financial market regulation. The behavioral SDF-based approach offers a unified treatment of behavioral beliefs and behavioral preferences. and option prices. and shocks to the economic and financial system impact the regulatory system.8 Introduction markets.

Surveys are part review.3 Home Bias and Omissions As I mentioned above. and finance generally. particularly the work of Smith: See his Nobel address. No single survey is comprehensive. its strengths. its weaknesses. All of these surveys reflect the biases of their authors in terms of perspective and emphasis. go. and quo vadis–where it is going? I plan to use this survey as a vehicle to communicate my perspective about the direction in which I would like to see behavioral finance specifically.1. part synthesis. In the interest of disclosure about home bias and omissions. (2007b). Home bias is particularly evident. none of these surveys discusses the experimental work on asset pricing bubbles. I hold particular views about the state of behavioral finance. and Subrahmanyam (2007). beginning with Hirshleifer (2008a) and moving on to Barberis and Thaler (2003). meaning authors are prone to emphasize their own work. meaning it will also feature home bias and omissions. Smith (2003). . For example. I should say at the outset that this survey will follow suit. Section 2 provides a survey of recent surveys. Baker et al.3 Home Bias and Omissions 9 Section 8 contains some concluding remarks which highlight the main points in this monograph. 1. and part editorial.

I select portions of each to identify key concepts and trace the different perspectives. unrealistic 10 . I discuss the perspective of a more recent general survey by Subrahmanyam (2007).2 Survey of Surveys This section describes the highlights of four surveys of behavioral finance that have appeared in the past decade. The list includes overconfidence. For the discussion of corporate finance. I describe the highlights of the survey written by Baker et al. I then present the highlights of a survey by Barberis and Thaler (2003). (2007b). I begin with the survey by Hirshleifer (2008a). Proponents of behavioral finance draw on a rich literature in psychology which documents phenomena that interfere with rational beliefs and decisions. In discussing Barberis and Thaler’s survey. and to a lesser extent corporate finance. I focus on Hirshleifer’s review of the psychology and some of the broad issues in asset pricing. Because there is a fair amount of overlap in these surveys. Finally. He also discusses how many of these psychological concepts have been applied to asset pricing. I focus on their discussion of specific issues in asset pricing. Hirshleifer presents a thorough treatment of the underlying psychology upon which proponents of behavioral finance draw.

Examples of these biases relate to such heuristic principles as availability. future outcomes. For example. Anchoring is the tendency to formulate an estimate by using a process that begins with an initial number (the anchor) and then making adjustments relative to the anchor. but to underweight information that is not readily available. confirmation bias. The volume edited by Kahneman et al. statistics. anchoring. a choice framework developed by Kahneman and Tversky (1979). Availability is the tendency to form judgments based on information which is readily available. availability bias. a person might underestimate the danger from riptides and overestimate the danger from shark attacks because the media tends to report most shark attacks but rarely reports less dramatic incidents involving riptides. Anchoring bias is the tendency for the associated adjustments to be too small. and probability weighting. For example.11 optimism. and the associated biases in those judgments. A heuristic is a crude rule of thumb for making judgments about probabilities. However. The heuristics and biases approach studies the heuristics upon which people rely to form judgments. A bias is a predisposition toward making a particular judgmental error. Representativeness is the tendency to rely on stereotypes to make a judgment. Glossary of Biases Some biases are associated with specific heuristics. aversion to a sure loss. This is because they form the judgment that favorable past performance is representative of good stocks. self-attribution bias. anchoring. loss aversion. The last three phenomena are major features of prospect theory. representativeness. Heuristics and biases associated with availability reflect the importance of salience and attention. (1982) contains most of the foundation contributions to the heuristics and biases literature. etc. and representativeness. . All other phenomena are part of an approach known as heuristics and biases. conservatism. a person who relies on representativeness might be especially bold in predicting that the future return of a particular stock will be very favorable because its past long-term performance has been very favorable.

Unrealistic optimism involves overestimating the probabilities of favorable events and underestimating the probabilities of unfavorable events. The illusion of control is overestimating the role of skill relative to luck in the determination of outcomes. Extrapolation bias leads people to overreact. People who are overconfident about their knowledge tend to think they know more than they actually do. Status quo bias predisposes people to favor inaction over action. Conservatism is the tendency to underreact by overweighting base rate information relative to new (or singular) information. people tend to be overconfident about both their abilities and their knowledge. because of insufficient attention to factors that induce regression to the mean. Categorization bias can be severe if the members of the same category are different from each other in meaningful ways. Hindsight bias is the tendency to view outcomes in hindsight as being more likely than warranted. Reliance on the affect heuristic is often described as “gut feeling” or intuition. Although some biases relate directly to specific heuristics. Categorization is the act of partitioning objects into general categories and ignoring the differences among members of the same category. For example. and to underweight information which disconfirms those views. developing unwarranted forecasts that recent changes will continue into the future. In particular. perhaps because of regret aversion. Confirmation bias leads people to overweight information which confirms their prior views. people who are overconfident about their knowledge tend to establish excessively narrow confidence intervals. other biases stem from a variety of factors.12 Survey of Surveys representativeness leads such predictions to be overly bold. . People who are overconfident about their abilities overestimate those abilities. Other examples of biases that do not stem directly from specific heuristics are unrealistic optimism and the illusion of control. That is. The affect heuristic refers to basing judgments on feelings rather than on underlying fundamentals. See Samuelson and Zeckhauser (1988). ex post the ex ante probability of the event that actually occurred is judged to be higher than the ex ante judgment of that ex ante probability. Such people end up being surprised at their mistakes more often than they had anticipated.

which lies at the center of the psychological-based approach to choice.1 2. misattribution stemming from halo effects. In this last respect. the work of Trivers (1985.2. people prefer familiar and similar individuals for reasons having to do with genetic relatedness. Hirshleifer provides financial examples for some of these items. This occurs because people who are well calibrated find it impossible to signal strong confidence credibly. Trivers (1985) suggests that for evolutionary reasons.1 Review of Psychology Literature and Link to Asset Pricing 13 The discussion of prospect theory. In this regard. Because of a shared evolutionary history. he mentions work on fast and frugal heuristics. The first category relates to attention. misjudgments about truth stemming from ease of processing. in order that people signal strong confidence to others. and therefore be subject to the same biases. memory. such as the halo effect explaining why growth . people rely on heuristics because people face cognitive limitations. cue competition. Heuristic Simplification Hirshleifer divides the items associated with heuristic simplification into several subcategories. First.1. 1991) suggests that people inadvertently signal their inner states to others. is deferred until later in this section and the section following. as described in Todd and Gigerenzer (2000). and processing. Items include salience and availability effects such as underweighting the probability of events not explicitly available for consideration.1 Hirshleifer (2001): Review of Psychology Literature and Link to Asset Pricing Broad Issues Concerning Judgment and Decision Bias Hirshleifer provides a set of organizing principles to place the above phenomena into a systematic structure. He argues that three principles provide a unified explanation for most biases. nature might have selected for traits such as overconfidence. people might be predisposed to rely on the same heuristics. people’s judgments and decisions are subject to their own emotions as well as to their reason. Second. Third. 2. and familiarity bias. the use of habits as a way of economizing on thinking. For this reason.

perhaps because time and cognitive resources are limited. but in the presence of a strong trend exhibit hot hand fallacy. Similarly. who found that in the presence of modest stock price fluctuations. The second subcategory for heuristic simplification pertains to narrow framing. Biases associated with this subcategory largely relate to oversimplification of a decision task.” Representativeness can lead people to overweight recent evidence (recency bias). Prospect theory. draw inferences which are too strong in the face of small samples (“law of small numbers”). loss aversion. The challenge is to identify the relevant situations which give rise to each. and reference points. regret. and reference effects. The subcategory also includes a variety of effects motivated by or associated with prospect theory. and familiarity explaining why investors appear to exhibit a preference for local stocks. Part of the explanation might involve cognitive . Hirshleifer cites a few financial applications such as the disposition effect in connection with mental accounting and money illusion in connection with framing. whereas Edwards (1968) finds evidence that in some types of situations. subjects exhibit gambler’s fallacy. with its focus on framing effects especially the isolation effect. representativeness suggests that people underweight base rate information. representativeness can lead people to underestimate unconditional population frequencies. Hirshleifer cites the experimental work of Andreassen and Kraus (1990). the tendency to overrely on how representative an object is of the population from which it is drawn. In some situations. which is discussed in detail below. The third category relates to representativeness. status quo bias. mental accounting. a phenomenon known as “base rate underweighting. and overpredict continuation when they infer the process from recent observations (“hot hand fallacy”). people underweight singular information relative to base rate information. such as the house money effect. preference reversal.14 Survey of Surveys stocks tend to be overpriced. and anchoring. lies at the center of this subcategory. whereas in others they predict continuation. the endowment effect. The fourth and final subcategory relates to subtleties in the updating of beliefs. overpredict reversals when they know the process governing transitions (“gambler’s fallacy”). In some types of situations people tend to predict reversals. As an application.

possibly because they worry about hostile manipulation of the odds they face. Klayman and Ha (1987) suggest that confirmatory bias might be an efficient mental shortcut rather than the result of self-deception. even if the choices were nonoptimal. Griffin and Tversky (1992) present a framework to address this issue. hindsight bias. Mood also appears to be an important determinant of how people feel about taking risk. At the same time. Arkes et al. and that external factors were responsible for their failures. Confirmatory bias leads people to feel better about their views because they underweight evidence that disconfirms their views. but overweight evidence that supports their views.2. along with cognitive dissonance. and confirmatory bias. Self-deception Overconfidence is a form of self-deception. Overconfident people take pride in feeling that they have more ability than is actually the case and that they know more than is actually the case. (1988) found that lottery ticket sales in Ohio increased after an Ohio State University football victory. based on the distinction between the “strength” of evidence (such as how extreme or salient is the information) and the “weight” of evidence (such as how statistically relevant is the information to prediction). Hindsight bias leads people to feel that their forecasting skills are better than is actually the case. Many fear choices that are ambiguous about the risk involved. but overreact to scenarios and concrete examples which are more easily processed. people feel better about themselves than they might were they bias-free.1 Review of Psychology Literature and Link to Asset Pricing 15 processing costs and salience. Cognitive dissonance leads people to feel better about choices they made. This is why people might be predisposed to underweight information that is abstract and statistical such as probabilistic information and sample size. Biased self-attribution leads people to feel that their skills were responsible for their successes. In all of these cases. biased self-attribution. Emotions and Self-control The emotion of fear exerts an important influence on the way people make choices about risky alternatives. .

where discount factors decrease more rapidly than is the case with exponential discounting. Stasser et al. liquidity premia.2 Social Interactions By and large the literature in behavioral finance draws from the psychology of individual behavior rather than social psychology. Hirshleifer points out that mood exerts more of an impact on judgments about abstract issues than concrete issues for which people possess information. A key bias studied in social psychology is the “conformity effect. (1989) report that people tend to discuss issues . However. Kuran and Sunstein (1999) suggest that such interactions can produce “availability cascades” whereby topics under discussion publicly generate momentum.1.” the tendency of people’s judgments and decisions to be skewed toward supporting other members of one’s group. Shiller and Pound (1989) report that interpersonal communication drew investors’ attention to the stocks they most recently had purchased. Shiller (1999) emphasizes the role of culture in communication. thereby reinforcing prior discussion. and the equity premium puzzle. even when they believe others to be in error. there have been some applications of social psychology to issues involving both investments and corporate finance. A key issue for investments concerns the process by which investors acquire information. stressing that because of limited attention. investors attach special importance to ideas or facts which are reinforced by conversation or ritual. An example is Asch (1956) who found that people will report judgments that conform to the judgments of others in their group. Hyperbolic discounting also gives rise to potential self-control difficulties when it comes to immediate gratification and the delay of unpleasantness. Time Preference and Self-control Neoclassical models tend to use exponential discounting. There is a literature in psychology about hyperbolic discounting.16 Survey of Surveys presumably because the win elevated the moods of Ohio residents. 2. meaning a constant time-invariant discount factor to reflect impatience. In this respect. Applications of self-control to financial issues include savings.

(Kahneman and Tversky. Probabilities are distorted by a probability weighting function in which extremely low probabilities are treated as zero. 2. The value function is S-shaped so that it is concave in the domain of gains and convex in the domain of losses.2. in the “false consensus effect. where it is steeper for losses than gains.1 Review of Psychology Literature and Link to Asset Pricing 17 about which they have shared information rather than sharing which they hold privately.3 Modeling Alternatives to Expected Utility Allais (1953) presented a set of paradoxes demonstrating that people are unlikely to behave in accordance with expected utility theory. relating to what he calls “the heart of the grand debate . overestimate the importance of disposition and underestimate the role of external circumstances. Hirshleifer describes several applications of these biases. 1979). (1977). For example. Welch (2000) describes the results of a survey of financial economists in connection with the equity premium. Camerer (1995) describes a series of alternative frameworks. and moderately high probabilities are underweighted.1. 2. and reports a false consensus effect in this regard.” people come to believe that others share their beliefs to a greater degree than is actually the case: see Ross et al. In prospect theory.4 Broad Asset Pricing Issues Hirshleifer’s survey focuses on the impact of investor psychology on asset pricing. Finally. Another example is “fundamental attribution error” where people. For example. Anderson (1932). moderately low probabilities are overweighted. the carriers of value (meaning arguments of the utility function) are gains and losses relative to a pre-specified reference point. Stasser and Titus (1985) present evidence that people will withhold private information to avoid contradicting the views of others in their group. because people tend to emphasize main points in conversation and deemphasize qualifying details. The value function also features a kink at the origin. in the course of determining the behavior of others.1. The literature in social psychology has identified other biases. listeners tend to form beliefs that are too extreme Allport and Postman (1947). of which the most prominent is prospect theory.

Campbell (2000) and Cochrane (2000) who emphasize how expected returns are related to “objective external sources of risk. systematic factors as well as idiosyncratic factors might be persistently mispriced. Moreover. idiosyncratic corners of the market. One possibility is where information is sparse and arrives slowly. In this regard. there might not be any perfectly rational investors in the market. Whereas rational investors move prices closer to their efficient levels. because of cognitive limitations.b. equilibrium prices reflect a weighted average of the different investors’ beliefs. The same argument might serve to explain how irrational investors induce persistent crosssection return predictability over the long-term. Consider where mispricing is most likely to be severe. 1991) is that if foolish investors underestimate risk and take on more than would be rational. As a result. As a result. the additional premium earned might counter the flow of wealth to rational investors. the magnitude of mispricing might depend on the degree to which investors not subject to bias might have detected a profitable opportunity and are seeking to exploit it. rational investors need not eliminate mispricing through their trades. namely that in equilibrium arbitrage is a double-edged sword. what Hirshleifer describes as “the dusty. The behavioral view. the wealth flow to rational investors which stem from the mistakes of irrational . The neoclassical view is that wealth flows from foolish to smart investors.” However.18 Survey of Surveys in finance spanning the last two decades.” He begins his discussion with two presentations of asset pricing. He asserts that in consequence. he describes a conceptual starting point. Hirshleifer makes an interesting point about the impact of irrational investors on asset prices. one associated with objective risk and the other with a proxy for mispricing. developed in a series of papers by DeLong et al. irrational investors move prices away from efficient levels.” Hirshleifer suggests that what differentiates the behavioral approach from the neoclassical approach is its focus on how expected returns are related to both risk and to investor misvaluation. Again. (1990a. a linear framework like the CAPM in which expected return is a function of two variables. so that in the long-run prices are set by rational investors. Rational investors might have difficulty in knowing the degree to which other like minded investors are attempting to exploit the same profit opportunity.

market. The behavioral challenge is identify how irrational investors can persistently affect prices when they are losing wealth to rational investors.2. return volatility. risk-free rate.2. namely psychology and the limits to arbitrage. Hirshleifer describes challenges respectively faced by those in the neoclassical school and those in the behavioral school. and predictability of returns. 2.2 Barberis and Thaler (2003): Review of Behavioral Asset Pricing Literature Barberis and Thaler (2003) suggest that behavioral finance has two main building blocks.9 percent. And it implies an absence of predictability. in contrast to evidence of predictability. to explain the magnitude of the historical equity premium. The neoclassical challenge is to identify objective risks to explain the various puzzles and anomalies. not the historical 3. . market volatility. The source of these puzzles is the inability of a standard neoclassical representative investor model. The neoclassical model implies an equity premium of 0. not the historical 18 percent. Limits to arbitrage refer to obstacles preventing arbitrage by fully rational agents from eliminating price inefficiencies. with plausible parameters.S. It implies a return standard deviation of 12 percent. Departures from Bayes’ rule occur because of heuristics and biases described in Hirsheleifer’s survey. and return predictability in the U. Barberis and Thaler also discuss a phenomenon known as ambiguity aversion. By psychology they mean systematic departures from rationality of beliefs (Bayes’ rule) and rationality of preferences (expected utility maximization). expected utility maximization is replaced by prospect theory. In the behavioral models discussed by Barberis and Thaler.1 percent. 2. which applies to the case of uncertainty.1 Aggregate Stock Market One of the most important applications of behavioral ideas to explaining aggregate stock market movements involves a triplet of puzzles involving the equity premium.2 Barberis and Thaler (2003): Review of Behavioral Asset Pricing Literature 19 investors is countered by the irrational investors earning a higher risk premium from their exposure to aggregate market risk.

20 Survey of Surveys The main behavioral approach to explaining the equity premium involves the work of Benartzi and Thaler (1995) and Barberis et al. thereby prompting the question of whether volatility in the dividend growth rate or required return is sufficient to explain the puzzle. degree of market risk. r denotes required return. In addition. the price–dividend ratio is quite volatile. based on the work of LeRoy and Porter (1981) and Shiller (1981). then the neoclassical theory predicts that the price–dividend ratio should also be stable. In practice. and g denotes the expected dividend growth rate. thereby putting upward pressure on the equity premium. A quick way of introducing the main issue is through the well-known Gordon constant expected growth formula which can 1 be stated as DPt = r−g . Both approaches use a representative investor whose preferences are based on prospect theory. if investors are concerned they have ambiguous beliefs about stock market returns. (2001a). prospect theory emphasizes loss aversion in respect to changes in consumption or wealth. thereby inducing a high equity premium in the market. Behavioral theories of the equity premium puzzle focus on how short investment time horizons might induce loss averse investors to behave as if they were extremely averse to risk. Although there is little evidence to suggest that the long-run dividend growth rate is volatile. As to volatile risk aversion. Behavioral explanations for the volatility puzzle are based on both beliefs and preferences. The starting point for the puzzle is the price– dividend ratio. This formula indicates the price–dividend ratio is driven by required return and expected dividend growth. While neoclassical expected utility theory emphasizes risk aversion in respect to total consumption or wealth on the part of investors. where P denotes price. they mention the work of Campbell and Cochrane (1999) who develop a habit formation model in which investors become . then they might demand an ambiguity aversion premium. If these are stable over time. and overall risk aversion. the required return on stocks is a function of the risk-free rate. Barberis and Thaler point out that it is difficult to reconcile the volatility of the price– dividend ratio in terms of volatile interest rates or volatile risk. D denotes divit+1 dend.

if the drop in the market leads investors to experience losses relative to the amount they originally invested. For example.2. investors might view a recent spurt in dividend growth as “representative” of the long-run sustainable future growth rate.2 Cross-section of Average Returns Behavioral finance is most associated with the anomalies literature. in line with hedonic editing. Among the most notable anomalies that cannot be explained on a CAPM risk-adjusted basis are the following: 1. investors become more tolerant of a loss after a rise in the market has generated gains for investors.2 Barberis and Thaler (2003): Review of Behavioral Asset Pricing Literature 21 more risk averse as their consumption falls toward the level to which they have become habituated. thereby leading subjective g to exceed objective g.2. whereby patterns associated with the cross-section of average stock returns cannot be explained in terms of the capital asset pricing model. and more averse to a loss after a drop in the market reduces that gain. A similar statement might apply to perceived risk. in which investors are assumed to be risk averse and have stable attitudes toward risk. A behavioral explanation of the volatility puzzle based on preferences is similar in spirit to the behavioral explanation for the equity premium puzzle. In contrast to neoclassical expected utility theory. Size premium: stocks with low market capitalization have subsequently earned higher returns than stocks with high market capitalization. Moreover. (2001a) develop a model with a representative investor whose preferences reflect elements of prospect theory and hedonic editing (Thaler and Johnson. prospect theory emphasizes that attitude toward risk is nonconstant and varies according to whether outcomes are registered as gains or as losses. Barberis et al. In their model. Barberis and Thaler argue that behavioral explanations based on representativeness suggest that investors’ expectations about future dividend growth and risk display excessive volatility relative to recent events. . 1990). 2. then they will become much more averse to further loss.

In this regard. 4. 1996) develop a three-factor model which empirically captures many of the anomalies. the risk to a security is . 7. and the difference in returns between high and low stocks ranked by BM. Momentum: winners based on prior short-term month performance (less than one year) have subsequently outperformed losers based on the same criterion during the next year. whereas stocks associated with dividend omissions have subsequently underperformed the market. the difference in returns between small firms and large firms. Barberis and Thaler raise two concerns in connection with the Fama–French risk-based explanation. the return to low BM stocks tends to be below the risk-free rate. 6. Scaled price-ratios: stocks associated with high ratios for book-to-market (BM) and earnings-to-price (EP) have subsequently outperformed stocks associated with lower ratios during the next year. and for which they provide a neoclassical risk-based interpretation. Post-earnings-announcement drift: stocks associated with positive earnings surprises have subsequently outperformed stocks associated with negative earnings surprises during the next year. Long-term reversals: losers based on prior 36 month performance have subsequently outperformed winners based on the same criterion. 3. Second. neoclassical theory measures risk by the covariance of returns with the marginal utility of aggregate consumption growth. over 60 months. stocks associated with dividend initiations have subsequently outperformed the market. Fama and French (1993. The factors relate to the market return. First. 5. New Issues and Repurchases: Stocks of firms that have issued new (repurchased) equity have subsequently underperformed (outperformed) a control group of firms matched on size and book-to-market equity over a five (four) year period. Dividend initiations and omissions: For a year after the announcement.22 Survey of Surveys 2.

in which case representativeness leads the representative investor to overreact to the run. when the excessive weight applied to the private signal leads the risky security to be mispriced. 2001) develop a model in which prices reflect the beliefs of a risk-neutral investor whose biases stem from overconfidence and self-attribution bias. Therefore.2 Barberis and Thaler (2003): Review of Behavioral Asset Pricing Literature 23 that it will pay a low return in bad times when the marginal utility of consumption growth is high. thereby increasing the investor’s overconfidence. However. overreaction and a subsequent correction eventually result. Daniel et al. a security earns a risk premium for generating a low return in bad times when aggregate consumption growth is low. The resulting momentum and subsequent positive feedback effect operate like a mechanical thermostat which does not click off when the correct . selfattribution bias leads the price-setting investor to be overconfident about information which he has generated on his own (his “private signal”) relative to information which is publicly available.2. In this behavioral combination. at which point momentum traders amplify the effect. The model produces momentum when external events confirm the price-setting investor’s private signal. (1998) develop a model involving a risk neutral representative investor whose beliefs are biased by conservatism and representativeness. Privately observed “news” diffuses slowly into prices. Barberis et al. Their model features a risk-free security and a risky security. there is little if any empirical support that stocks earning above average returns perform especially poorly during periods of low aggregate consumption growth. Hong and Stein (1999) present a model featuring the interaction between “newswatchers” who base their trades on privately observed fundamentals and “momentum traders” who base their trades on the most recent price change. (1998. a subsequent correction eventually results. When overreaction leads the risky security to become mispriced. These models emphasize the combination of momentum in the short-term and reversals stemming from overreaction in the long-term. Conservatism leads the representative investor to underreact to new information unless the new information is part of a run. However. There are a variety of behavioral belief-based models that have been developed to explain cross-sectional patterns.

were extreme. and Jones and Lamont (2002). Chen et al. (1997) find that longterm reversals associated with past winners and losers as well (low BM stocks) appear to occur at the time of earnings are announced. They suggest that this finding supports a hypothesis stemming from the Hong–Stein model. the evidence indicates that increased breadth of ownership or higher costs of shorting will lead to upward bias in price and subsequent lower returns. (1992) and La Porta et al. Short-sales restrictions create an asymmetry in which the beliefs of optimistic investors carry more weight in price formation than the beliefs of pessimistic investors. See Miller (1977). The model developed by Barberis and Huang (2001) explains the value premium associated with BM by suggesting that loss aversion and narrow framing lead investors to act as if they are more averse to . The resulting overreaction eventually leads to a correction. Perhaps. investors who believe a security is overvalued might choose to buy the security in anticipation of selling it later to investors who are especially optimistic. A series of papers discuss the implications associated with heterogeneous beliefs in combination with restrictions on short sales. momentum traders suffer from confirmation bias and the opposite of aversion to ambiguity. driven by representativeness. Barberis and Thaler suggest that these findings reflect investors receiving news that their past forecasts. Diether et al. Hong et al. Hong and Stein (2003) suggest that pessimistic investors who cannot short an overvalued stock will buy the stock after its price declines to fundamental value. although the traits of “newswatchers” are similar to those of the price-setting agents in the model developed by Daniel et al. (2000) find that momentum is more pronounced in the stocks of small cap firms and firms with low analyst coverage. Harrison and Kreps (1978) and Scheinkman and Xiong (2003) point out that in the presence of short-sale constraints. In this regard. (2002). Hong and Stein do not emphasize particular psychological biases in their approach.24 Survey of Surveys temperature is reached. (2004). but not if its price increases. Chopra et al. This leads to an asymmetric volatility response leading to skewness in the return distribution. to the extent that they rely solely on technical analysis and fail to generate information about fundamentals.

The puzzle concerns the existence of a closed-end fund discount (or premium. Although the behavioral models described above were developed to explain the cross-section of returns. The sentiment-based explanation suggests that closed-end fund prices will covary in response to changing sentiment.. it is worth noting that they all feature a single risky asset rather than multiple assets whose returns are jointly determined. One of the most important aspects of the behavioral approach to risk premiums is the idea that the premium can be decomposed into a fundamental component and a systematic sentiment component. The behavioral approach to the puzzle is based on the notion that investor sentiment varies over time. high BM stocks. 1995) showing the existence of common factors in earnings and returns associated with small cap stocks.2. they emphasize that while such common factors exist. In this respect. 1991). they are weaker for earnings than for returns. They also review work by Fama and French (1993. are at best suggestive of the role played by underlying psychological phenomena. A similar statement holds in respect to explanations of the closedend puzzle (Lee et al. Barberis and Shleifer (2003) suggest that investors are prone to “categorization” and treat certain members of certain groups of stocks (such as small cap stocks) . being “systematic” means that sentiment should also explain the premiums associated with small cap stocks. and so on. (2001a). Pessimistic sentiment produces closed-end fund discounts whereas optimistic sentiment produces closed-end fund premiums.2 Barberis and Thaler (2003): Review of Behavioral Asset Pricing Literature 25 risk after a stock has declined in price. fluctuating between degrees of optimism and pessimism. This means that the explanations associated with these models. The dynamics are similar to the model developed by Barberis et al. thereby suggesting that other issues are at work. Barberis and Thaler point out that the discount attached to closed-end funds as a class and premium attached to small cap stocks covary. where sentiment is understood to mean excessive optimism or pessimism. In this regard. when positive) relative to net asset value. rational investors will demand a positive risk premium as compensation for bearing the component of risk associated with changes in investor sentiment as well as the component associated with changing fundamentals. In equilibrium.

risk-adjusted portfolio returns are negatively correlated with the frequency of trading. Huberman (2001) documents a regional bias in the holding of stocks of regional telephone companies. Benartzi (2001) reports a strong bias toward own company stock in 401(k) plans. They introduce the discussion by mentioning the trend toward increased participation in equities and the increased emphasis on defined contribution plans in retirement planning. Barber and Odean (2000) find that for individual investors. categorization produces common factors in returns to stocks in the same group. They attribute this result to a combination of overconfidence and transaction . Japan. As a result. A pertinent example of a group is the S&P 500. French and Poterba (1991) document the high concentrations of domestic equity in the equity portion of investors’ portfolios in the USA. Many investors trade excessively and unprofitably.3 Investor Behavior Barberis and Thaler discuss how psychology influences the behavior of individual investors. Investors might also choose poorly diversified portfolios because they use crude heuristics. Individual investors do not choose mean-variance portfolios. and the United Kingdom Grinblatt and Keloharju (2001a) find that Finnish investors concentrate on the stocks of firms that are located close to them geographically.2. See Benartzi and Thaler (2001) who report evidence of instances in which the relative number of equity funds appears to drive the portion of participants’ allocations toward equity. Instead.” the tendency to overweight securities with which investors are familiar.26 Survey of Surveys as being more similar than fundamentals warrant. (2001b) who document the impact of a stock’s beta when the stock is first included in the index. an issue they describe as being less controversial than issues involving the determinants of market prices. See Barberis et al. Benartzi and Thaler (2001) suggest that ambiguity aversion and the desire for familiarity lead investors to exhibit “home bias. 2. such as the “1/n” heuristic which involves dividing assets equally across choices. their portfolios tend to be insufficiently diversified.

Using discount brokerage data. In terms of the selling decision. mental accounting. investors are prone to pay more tax than necessary because they sell winners too quickly and hold losers too long. Genesove and Mayer (2001) present evidence from the housing market that homeowners are reluctant to sell their homes at prices below original purchase price. Momentum results after losses when investors who are eager to hold their stocks after bad news retard the adjustment of price to fundamental value. The first prong pertains to the impact of irrational investor behavior on rational corporate . A key feature of the disposition effect for stocks is that the original purchase price serves as the reference point. Operationally. They describe work by Grinblatt and Han (2004) which indicates that the disposition effect might underlie momentum. overconfident investors mistakenly believe that their skills are sufficiently high to compensate for the costs of trading.3 Baker et al. Shefrin and Statman (1985) call this behavior “the disposition effect” and attribute it to the following series of psychological phenomena: the reference point property. Professional traders in the Treasury Bond futures pit at the Chicago Board of Trade take greater intraday risks in the afternoon if they have incurred losses in the morning.2. Momentum results after gains when investors who are eager to take profits after good news retard the adjustment of price to fundamental value.3 Baker et al. (2007): Survey of Corporate Finance Baker et al. and stocks with news announcements. (2007): Survey of Corporate Finance 27 costs. attention for stocks is based on one or more of the following traits: abnormally high trading volume. As study by Coval and Shumway (2005) demonstrates that the disposition effect is not confined to individual investors. and self-control. Barberis and Thaler do touch on issues of pricing in connection with the disposition effect. regret aversion. Odean (1998a) provides strong evidence of the disposition effect. extreme returns. Odean (1999) and Barber and Odean (2008) find that investors are prone to buy stocks to which their attention has recently been drawn. That is. In terms of the buying decision. (2007b) provide a two-pronged approach in surveying the literature on behavioral corporate finance. 2.

and Jenter (2005). f (K. Second. Seyhun (1992). In this respect there is evidence that corporate managers earn abnormal returns on their own trades. they are less constrained by the limits of arbitrage than equally smart money managers. develop a model to capture the criterion function a rational manager uses to balance these objectives. suggest that these managers balance three objectives: (1) maximizing fundamental value. think about a neoclassical firm whose investment and financing strategies are determined independently. ) − K + eδ(·)] + (1 − λ)δ(·). 2. The product eδ represents the value transfer to original shareholders from exploiting the mispricing. e) be a function specifying the present value of future cash flows inclusive of financing effects. The criterion function has the form: λ[f (K.1) .3.1 The Irrational Investors Approach There are several reasons to believe that some corporate managers might have the ability to identify mispricing. and λ be a number between 0 and 1 measuring the degree to which the manager’s horizon is long-term. See Muelbroek (1992). Third. Define e as the deviation in the firm’s book value of equity from its optimal neoclassical level. Baker et al. Theory Baker et al. (2) undertaking actions to boost the market value of their firms above fundamental value (catering). Let K be the amount of capital invested. they have superior information about their own firms. Consider how rational managers might function in a world in which irrational investors cause inefficiencies in market prices.28 Survey of Surveys managers. (2. while the second prong pertains to the manner in which irrational decisions by corporate managers impact corporate value. and (3) undertaking financing decisions to exploit temporary mispricings of the securities issued by their firms (market timing). managers might have access to efficient heuristics: see Baker and Stein (2004). For this purpose. First.

the firm takes on too much equity relative to debt.2) would imply low investment as fK > 1. Suppose that δK > 0. In this analogy. in which case fe < 0. A firm that is financially constrained would find itself unable or unwilling to raise external financing. λ (2.2) implies that fK < 1. In this respect.2) states that fK = 1 (meaning the net present value of investment is zero at the margin) and e = fe (0) = 0. In this case. Consider the case in which δ > 0 (overpricing) combined with δe < 0.3) is analogous to the microeconomic problem faced by a monopolist who chooses quantity to equate marginal revenue and marginal cost. The inequality means that when the firm issues more equity.3 Baker et al. so that managers cater to investor sentiment by increasing K relative to the neoclassical case.3) indicates that the firm chooses e > 0. In this case. In this case. up to the point in the monopoly analogy whereby marginal revenue equals marginal cost. The right-hand side of Equation (2. meaning higher K exacerbates overpricing. (2007): Survey of Corporate Finance 29 The first-order conditions for this problem are as follows: fK = 1 − e+ (1 − λ) δK . Begin with Equation (2.2) −fe = δ(·) + e+ (2. Such firms are reluctant to issue underpriced equity in order to finance investment. Consider how choices of K and e are impacted by λ < 1 and δ = 0. There is also the case δ < 0 (underpricing). Equation (2. Equation (2. when e > 0.1) managers have a long-term horizon (λ = 1) and there is no mispricing (δ = 0). . the resulting selling pressure leads the degree of overpricing to decline. λ (1 − λ) δe .3). while the left-hand side corresponds to marginal cost. so that e might be negative with fe > 0.3) In the neoclassical benchmark for Equation (2.2).2. Equation (2. e corresponds to quantity and δ corresponds to price. Equation (2. Next consider Equation (2. thereby losing tax shield and related benefits. Equation (2.3) corresponds to the two terms making up marginal revenue. For managers with long enough horizons (λ sufficiently close to 1).

as happens when share turnover is higher. (2003) study whether investment will be most sensitive to mispricing for equitydependent firms. point out that open questions remain about the magnitude of the mispricing-investment effect and the associated efficiency implications. In this last respect. . Nevertheless. Baker et al. Titman et al. Polk and Sapienza also show that this impact is greater for shorter managerial horizons. see Loughran and Vijh (1997) and Rau and Vermaelen (1998). Managers might use mergers and acquisitions to cater to investor sentiment. and claim not to be convinced that the pattern stems from catering. 2004). Polk and Sapienza (2004) find that in the cross-section. beginning with Keynes’ discussion about “animal spirits” during the 1920s. and moving on to other bubble episodes involving electronics (1959–1962). growth stocks (1967–1968). It also explains why cash acquirers subsequently outperform stock acquirers (who earn negative long-run returns). These studies sought to control for fundamental value and searched for a residual effect on investment from stock prices. The authors of later studies searched for evidence that proxies for mispricing impact investment. do not view these results as strong evidence of inefficiency in an ex ante sense. and Lamont (2000) reports something similar in a time series analysis. Baker et al. Baker et al. or by engaging in a combination for which sentiment is positive. Baker et al. (2004) find such evidence. perhaps by using the overvalued stock of their own firms to acquire less overvalued targets. Panageas (2003). firms that invest more earn lower subsequent returns. see Shleifer and Vishny (2003). the case discussed above where e < 0. This theory offers some insight into the time-series relationship between merger volume and stock prices. review some of the early literature on this topic. Chirinko and Schaller (2001. and dot-com stocks (late 1990s). and Gilchrist et al. (2004) find the same pattern. Following Stein (1996). They point out that the first modern empirical studies were inconclusive about the impact of mispricing on the magnitude of investment. Polk and Sapienza (2004).30 Survey of Surveys Empirical Studies Involving Real Investment Most research about real investment and mispricing focuses on the case of overpricing.

Moreover. They discuss the work of Klein (2001) who finds a diversification premium of 36 percent in the period 1966–1968. Jenter (2005) presents related evidence that managers conduct secondary equity offerings (SEOs) while simultaneously engaging in insider selling. (2004) also propose that investors might respond more passively to accepting the acquirer’s shares as part of the merger (a form of status quo bias). (1994) report a high positive . for example by issuing new shares when equity is overpriced and by repurchasing shares when equity is underpriced. mispricing can lead a firm’s managers to engage in market timing. Baker et al. suggest that the 1960’s conglomerate wave reflected a desire by managers to cater to investors’ preference for conglomerates. but would not actively purchase those shares when issued directly. as reported in Ravenscraft and Scherer (1987). Loughran et al. Baker et al. Baker et al.3 Baker et al.2. There is considerable empirical evidence suggesting market timing in respect to equity. (2007): Survey of Corporate Finance 31 Evidence of a positive correlation between market-level mispricing and merger volume can be found in Dong et al. subsequent returns to mergers conducted during high-valuation periods tend to be poor (Bouwman et al. 2009). followed by a decline between July 1968 and June 1970. An issue related to mergers and acquisition is the diversification discount. pose an interesting question about mispricing-driven mergers: why would an overvalued firm use a merger instead of issuing more equity? They suggest a framing-based explanation that a merger might be more effective at hiding the underlying misvaluation. Although there is little in the behavioral finance literature dealing with this issue. These studies report that undervalued targets are more likely to be hostile to acquisition. Survey evidence presented by Graham and Harvey (2001) indicates that financial managers view such market timing activities as very important. Empirical Studies Involving Financial Policy As was mentioned in the theoretical section above. and Ang and Cheng (2009).. (2003). and that overpriced acquirers tend to pay higher takeover premiums.

Using a sample of U. Indeed. firms tend to issue short-term debt when the yield curve is steep.S. Brav et al. and when waiting for long-term rates to fall. returns tend to be low. firms. empirically subsequent bond returns are positively correlated with the slope of the yield curve. As for seasoned equity offerings (SEOs). Marsh (1982) reports that in his sample of U. firms issuing equity subsequently underperform the market by between 20 and 40 percent over five years. Moreover. and longer maturity debt when the yield curve is flatter or negatively sloped.32 Survey of Surveys correlation between volume of initial public offerings (IPOs) and international stock market valuations. they also point out that the equity market timing issue is controversial. After reviewing the relevant literature.K. and is still a subject of considerable debate. In this respect. after five years the average IPO underperforms its nonIPO peer by 30 percent. the survey evidence presented by Graham and Harvey (2001) does indicate that firms issue debt when they feel that interest rates are low. (2009) report that subsequent to periods of high foreign issues. . Marsh (1982) reports that equity issuance increases after recent stock market appreciation. thereby supporting the market timing decisions managers make about debt issuance. In this respect. Dichev (2004) reports that firms who engaged in market timing reduced their cost of capital by 1. (2000). firms. Guedes and Opler (1996) report that firms do indeed issue short maturity debt when the yield curve is steeply sloped. (2005) report survey evidence indicating that the great majority of firms agree that they repurchase shares when they view their shares as underpriced in the market. In terms of share repurchases.3 percent a year. Relative to a sizematched firm. At the same time.S. Loughran and Ritter (1995) provide evidence suggesting the presence of mispricing. the Graham– Harvey survey evidence indicates that 44 percent of firms engage in cross-border market timing to exploit differential interest rates. while Henderson et al. interest rates impacted the debt–equity choice. Froot and Dabora (1999) present evidence of cross-border mispricing. Moreover. conclude that on average U. Baker et al. Ikenberry et al. See also Ikenberry et al. (1995) find that the average repurchaser outperformed firms matched by size and book-to-market equity by 12 percent over the subsequent four years. Moving from equity to debt.

3 Baker et al. A high value of this ratio suggests that the firm raised the bulk of its external equity or debt when its market-to-book ratio was high. then firms which issue equity when they believe their stocks are overpriced will tend to feature low leverage. equity overvaluation might make debt cheap. point out that the issue is still under debate. where the weights are based on the degree of external financing. and are also subject to status quo bias. and omissions increase when the dividend premium is negative. (2007): Survey of Corporate Finance 33 Interestingly. Other Corporate Decisions Besides investment and financial policy. For example. Richardson and Sloan (2003) find that equity returns tend to be low after debt issuance. See Hovakimian (2006). earnings management. Baker and Wurgler (2004a.2. Baker and Wurgler (2002) provide such a market timing-based explanation of capital structure. They claim to find evidence of catering by noting that initiations increase when the dividend premium is high. behavioral phenomena play a role in other corporate decisions such as dividends. suggesting that firms might be timing their idiosyncratic credit quality. Baker et al. The major issue in respect to dividends is whether firms cater to investor sentiment by initiating dividends when sentiment is positive and omitting dividends when sentiment is negative. there are other explanations for the pattern. According to the theory. as equity valuation is a component of the cost of debt. and the overvalued equity subsequently corrects itself. (2003) measure a “dividend premium” as the difference between average market-to-book equity between payers and nonpayers. Their analysis employs a weighted measure of market-tobook equity.b) suggest that catering might explain the result described in Fama and French (2001) . Baker et al. Therefore. If managers engage in market timing. One issue is that the relationship Baker and Wurgler identify might stem from growth opportunities rather than market timing. and executive compensation. firms issue debt because it is cheap. choice of firm name. Despite the evidence supporting the prediction. in the cross-section the weighted average will be negatively related to debt-to-assets. That said.

2. 2. catering as defined above. 2. . regret aversion.2 The Irrational Managers Approach Consider the opposite side of the coin. (2004) found that financial managers believe that investors care more about earnings per share (EPS) than cash flows.3. Bergstresser et al. earnings manipulation is a form of catering. does not address other psychological issues associated with dividends such as self-control. Here unrealistic optimism is typically interpreted as overestimating the first moment of cash flows or returns. They noted that consistent with a catering motive. Cooper et al. not the magnitude of dividends. many small firms changed their name to include a “dotcom” reference. such name changes resulted in an average announcement effect of 74 percent. 1984). (2004) found that as CEO compensation becomes more sensitive to share price. earnings management increases. but corporate managers are subject to psychological pitfalls. Teoh et al. A survey conducted by Graham et al. The most common pitfalls in the behavioral corporate finance literature are unrealistic optimism and overconfidence. and hedonic editing (Shefrin and Statman. For this reason. where market prices are efficient. and for this reason are willing to sacrifice net present value for higher EPS. In addition. However.1 Theory The theoretical framework for the case of irrational managers operating within efficient markets is similar in structure to the model used to study the case of rational managers operating within irrational markets. they also note that the catering theory of dividends is limited in that it only explains initiations and omissions.34 Survey of Surveys about the sharp decline between 1978 and 1999 of the number of firms that pay dividends. Sloan (1996) finds that firms associated with high accruals tend to feature low subsequent returns. (2001) presented evidence that during the Internet bubble. and overconfidence is interpreted as underestimating the second moment (risk as measured by the standard deviation).b) report that firms who most aggressively manage earnings are associated with the highest underperformance in connection with equity issuance.3. (1998a.

The first-order conditions for this problem are as follows: fK = 1 . ) + efe (K. This happens when the firm is cash poor and chooses to issue more equity than is associated with the neoclassical case..6) (2. Suppose that e > 0.4) (1 + γ)fe = γ(f (K. When γ > 0 managers overinvest.4) γ = 0. Equation (6. 1 + (1 − e)γ (2. and balance a shift away from the neoclassical capital structure against perceived losses from market timing. The criterion function has the form: (1 + γ)f (K. thereby leading to cash flow sensitivity.5) (2. then a financial pecking order results.2. 2009). 2003. 2. ) − K − eγf (K. A collection of papers in the behavioral literature points out that unrealistic optimism and overconfidence can produce benefits in the presence of agency-related managerial traits that lead to underinvestment. such as market-to-book as an ex ante measure and abnormal subsequent realized returns as an ex post measure.1) implies that the degree of overinvestment falls. ).2. An unbiased manager is associated with γ = 0.2 Empirical Issues A key empirical task associated with the rational managers/inefficient market approach involves identifying proxies for mispricing. )). a key empirical task associated with the irrational managers/efficient market approach involves identifying proxies . If there is an upper bound on leverage (fe > 0). (Gervais et al. discuss a variety of possible measures. This results in the usual first-order conditions fK = 1 and fe = 0. In the neoclassical benchmark for Equation (2. Goel and Thakor. 2008. By the same token.3 Baker et al. Baker et al. Hackbarth.3. such as risk aversion or debt overhang. as managers never sell equity they perceive to be underpriced unless it is absolutely necessary in order to fund investment. (2007): Survey of Corporate Finance 35 A new variable γ is added to the model to represent the combination of managerial excessive optimism and overconfidence.

with γ representing private managerial benefits and eγf (K. half of startups do not survive more than three years (Scarpetta et al. (1981) and Statman and Tyebjee (1985) present evidence of excessive optimism in project cost forecasts. only 38 percent expect development and 17 percent expect difficulties. They find that 56 percent initially estimate that there will be “development” in the near future and only 6 percent expect difficulties.. To counter excessive optimism . However. identifying proxies which distinguish between the two possibilities is a critical task. overconfident CEOs would wait the expiration year to exercise. whereas excessively optimistic. For this reason. The point of the proxy is that because CEOs portfolios are overly concentrated in the value of their own firms.36 Survey of Surveys for excessive optimism and overconfidence. Cooper et al. Landier and Thesmar (2009) survey French entrepreneurs. and especially high when the firm is equity dependent. There is ample evidence suggesting that excessive optimism and overconfidence lead to overinvestment. Malmendier and Tate find that cash flow sensitivity is pronounced among firms whose CEOs are longholders. (1988) report that 68 percent of entrepreneurs believe that their own startups are more likely to succeed than comparable firms. and pecking order financing have both agency cost explanations and behavioral explanations. Given the undiversified nature of entrepreneurial investment. This is especially important in that the behavioral structure represented by Equation (2. Merrow et al. and one-third perceive the risk of failure to be close to zero. rational CEOs would exercise in-themoney options long before expiration. Malmendier and Tate (2005) suggest a longholder proxy for excessive optimism and overconfidence.4) is formally similar to an agency structure. cash flow sensitivity. a point to which I return in Section 6. ) being replaced by a function c(e) corresponding to the agency cost of raising external equity. Moskowitz and Vissing-Jorgensen (2002) suggest that returns to private equity seem too low. 2002). The longholder proxy identifies chief executive officers (CEOs) by their propensity to hold in-the-money executive options on their own firm’s stock until expiration. only 5 percent believe the odds to be worse. Indeed. underinvestment. This means that overinvestment. three years into their ventures. Moreover.

“all derived from investor rationality. In addition. 1989. Landier and Thesmar discuss contracts which transfer control from entrepreneurs to investors when the former persist in following their initial plans. Roll (1986) suggested that hubris leads managers of acquiring firms to overpay for targets. For example. perhaps because they are least concerned about financing the acquisition with equity they believe to be underpriced. despite signals suggesting changes. More than half of firms use target debt–equity ratios. and target debt–equity ratios tend to be based on book values instead of market values as called for in the neoclassical framework. The argument is that this gain makes owner/managers more willing to accept underpricing. less than 10 percent set strict targets.4 Subrahmanyam (2007): Additional Perspectives Subrahmanyam (2007) begins by identifying what he calls the central paradigms of modern finance. Malmendier and Tate (2008) apply their longholder analysis to mergers and acquisitions. especially diversifying mergers. longholder CEOs of the least equity dependent firms are more apt to engage in acquisitions. thereby losing control of their enterprises to investors more frequently than they planned. the Graham–Harvey survey evidence indicates that many financial managers rely on heuristics such as the payback rule and use firm-wide discount rates to compute hurdle rates instead of risk-specific hurdle rates.4 Subrahmanyam (2007): Additional Perspectives 37 and overconfidence. thereby suffering from “the winner’s curse. thereby generating a perceived gain relative to the file range reference point. Managers also exhibit reference point effects. they choose short-term financing over long-term financing. They also suggest that because excessively optimistic. overconfident entrepreneurs underestimate short-term risks.” . They are prone to “throw good money after bad:” (Statman and Sepe. also discuss behavioral managerial patterns distinct from excessive optimism and overconfidence.” As it happens. Baker et al. 2. Guedj and Scharfstein. Loughran and Ritter (2002) apply a portion of the framework presented by Shefrin and Statman (1984) to suggest that initial underpricing for IPOs is pronounced when the offer price lies above the upper end of the file range.2. 2004). longholder CEOs complete more mergers than longholders.

2. focusing on whether factors proxy for both fundamentally based risk and sentiment. However. In this respect. (2007) explain momentum. This type of model provides insight into the implications of investors relying on heuristics rather than full optimizing models.’s approach. They report that stocks initially exhibit momentum in response to an announcement that the firm is in distress (a going-concern audit report). where managers . Behavioral analyses of the cross-section focus on momentum in the short-run followed by reversals in the long-run.38 Survey of Surveys The paradigms are: (i) portfolio allocation based on expected return and risk. To minimize discussion which overlaps the preceding three surveys. attention will focus on works that appeared subsequent to the earlier surveys. (1998). Doukas and Petmezas (2005) find support for the self-attribution hypothesis. (ii) risk-based asset pricing models such as the CAPM and other similar frameworks. asset bubbles. Part of this debate focuses on the Fama and French (1993) three-factor model and a critique by Daniel and Titman (1997) centered on whether it is firm characteristics or factor loadings that explain returns. bringing into question the “distressrisk” explanation. (iii) the Miller–Modigliani theorem and its augmentation by the theory of agency.1 Cross-Section of Stock Returns The debate about whether risk or psychology explains the cross-section continues. More recently. they argue that the component of this ratio which relates to fundamentals does not actually forecast returns. one of the elements of the Daniel et al. Moreover. Hong et al. Kausar and Taffler (2006) provide evidence which supports the approach advanced by Daniel et al. and (iv) the pricing of contingent claims. the debate has become more nuanced. or ideas not covered. they find the phenomenon in the market for corporate control.4. and other phenomena using a model in which investors use overly simplified models to evaluate stocks. Daniel and Titman (2006) suggest that the book-to-market effect is driven by overreaction to that part of the bookto-market ratio not related to accounting fundamentals. but later exhibit reversals. His survey then focuses on how behavioral ideas have impacted these paradigms.

pessimism is not adequately reflected in stock prices. and vice versa. (1994) find no evidence that this is the case. However. Given that marginal utility is inversely related to consumption growth. For theory to support a positive premium in respect to size and book-to-market equity. trading volume and accounting ratios such as return on equity and price/earnings. (2000) hold in Europe. thereby providing out-of-sample confirmation for the Hong and Stein (1999) hypotheses. this implies that the returns associated with small cap firms and stocks with high book-to-market equity would need to be particularly high in good times. Notably. a high Sharpe ratio implies highly variable marginal utility across states. Haugen and Baker (1996) find that the strongest determinants of expected returns are past returns. because of short-selling constraints. there is mixed evidence about the regime switching hypotheses advanced by Barberis et al. Baker and Stein argue that optimistic investors generate volume. At the same time. the returns of small and high book-to-market stocks would need to covary negatively with marginal utility. Subrahmanyam’s treatment of this issue mentions the work of Baker and Stein (2004). (2003) find no support suggesting extrapolation after a sequence of news events within returns data. Subrahmanyam points out that it is difficult for the neoclassical approach to justify the magnitude of Sharpe ratios observed in practice. 1991). Chan et al. Frieder (2004) does find supporting evidence in order flow data around earnings announcements. The Barberis–Thaler survey discusses belief-based models with institutional friction. the optimism is reversed in subsequent periods. Notably. for which there is little if any empirical support. . In a related vein. Doukas and McKnight (2005) find that the results reported by Hong et al.2. Lakonishok et al. However. (Hansen and Jagannathan. they find no evidence that risk measures such as systematic or total volatility are material for the cross-section of equity returns. producing a negative relation between returns and past volume. However. (1998).4 Subrahmanyam (2007): Additional Perspectives 39 appear to earn successfully smaller returns in each successive acquisition because they become increasingly overconfident with each successful acquisition. Based on the Euler equation from a neoclassical representative investor model.

40 Survey of Surveys In cross-sectional regressions.2 Other Psychological Influences on Returns As was mentioned in Hirshleifer’s survey. and Hirshleifer et al. 2006). followed by the reversal of their sentiment. Subrahmanyam discusses several papers concerning one of the central claims made in the behavioral finance literature. Griffin and Tversky (1992) provide a psychological theory based on weight and strength explaining why some situations feature underweighting of base rates. Some findings appear to lack mutual consistency. Hvidkjaer (2005) reports a negative relationship between trade imbalances of small investors and future stock returns in the cross-section. they find some support for the Griffin and Tversky hypothesis. in that shortselling is typically cheaper for institutions. (2006).4. especially mood. namely that investors who are imperfectly irrational are able to persist in the long run. The specific papers he mentions in this regard are DeLong et al. Doing so. 2. Kyle and Wang (1997). on returns. which is the seminal work mentioned earlier on this point. thereby generating stock return predictability. This result is consistent with an overreaction effect by small investors. They proxy weight by analyst experience and strength by the number of categories spanned by analyst revisions. consider the possibility that mood is positively . (1991). Hirshleifer’s survey mentions the role of emotions. Sorescu and Subrahmanyam (2006) test the argument that investors overreact to the strength of a signal and underreact to its weight. For example. stocks of small firms which have low analyst following appear to feature greater mispricing in terms of return predictability from book-to-market and momentum. Yet. There is evidence that mispricing is most severe for stocks where institutional ownership is lowest (Nagel. small investors buy loser momentum stocks and subsequently become net sellers in these stocks. whereas other situations feature overweighting of base rates. 2005) who suggests that institutional ownership is a proxy for short-selling constraints. This suggests that they may create momentum by underreacting to negative information. Subrahmanyam mentions a finding by Hvidkjaer (2006) whereby on net. (Zhang.

Subrahmanyam describes a comprehensive study of trading activity by Grinblatt and Keloharju (2001b). A related approach focuses on seasonal affective disorder (SAD). and suggest that the cause is induced depression of investors subject to SAD. Finnish data is exceptionally rich. In a different type of mood study. mood has also become a specific object of study in the neurofinance literature: see Knutson and Kuhnen (2009). the disposition effect will be clearly identifiable. They confirm the disposition effect in Finnish data. They find that returns are positive when the outcome is positive and vice versa. Although Subrahmanyam does not mention it explicitly. in that it is comprehensive on an account and trade-by-trade basis. (2000) focus on returns for weekends involving the switch from daylight savings time to standard time. He also finds that there is a sharp upsurge in volume after the price rises above the offer price for the first time.2. Frieder and Subrahmanyam (2005) present evidence that individual investors prefer stocks with high brand recognition. For example. Hirshleifer and Shumway (2003) confirm this finding. Poteshman and Serbin (2003) present evidence that option investors do not optimally exercise options. Extending this study to a number of international markets. more wealthy investors. Kaustia (2004) also provides insight into the disposition effect by focusing on IPO markets. They find these returns to be very negative. In discussing the disposition effect. Kaustia finds that volume is lower when the stock price is below the offer price. As to diversification.4 Subrahmanyam (2007): Additional Perspectives 41 correlated with exposure to sunlight. At the same time. Edmans et al. they exercise options . Kamstra et al. In support of the familiarity hypothesis. Goetzmann and Kumar (2008) find that the portfolios held by individual investors who are young and less wealthy are less diversified than older. (2007) study how stock market returns in a country are impacted by the outcomes of sporting events involving that country. A study by Saunders (1993) documents that stock returns on the New York Stock Exchange tend to be positive on sunny days but mediocre on cloudy days. The idea is that because the offer price is a common purchase price. Goetzmann and Zhu (2005) find little evidence that the return pattern stems from individual investors’ trading patterns.

5 Critique: Weaknesses in the Behavioral Approach Behavioral finance has both strengths and weaknesses.5. and need to be addressed for the behavioralizing of finance to be successful. 2. To be sure.42 Survey of Surveys when they should instead sell them. Evidence presented by Bakshi et al. (2000) suggests that traders often cause movements in option prices to be inconsistent with comparative statics based on traditional rationality assumptions. in that some of the weaknesses in the behavioral approach are foundational in nature. Unfortunately. prospect theory offers important insights. In particular the shape is at odds with the principles underlying the theory (Ingersoll. Discussions about preferences in the behavioral finance literature are overwhelmingly dominated by prospect theory. It is incomplete in respect to framing. but not as a good job in articulating its weaknesses. The shape of the functional form used to capture probability weighting is sensitive to the choice of parameter value. when in fact they are not. The convex portion of its value function gives rise to . systematic behavioral approach to preferences. However. This issue is important. 2008). Addressing these issues underlies much of the remainder of this monograph. I provide a brief description of what I see as key weaknesses in the behavioral approach. only Baker et al. The prior surveys do a good job of articulating its strengths. and captures a wide variety of behavioral patterns observed in experimental settings. there is a difference between describing the empirical patterns and modeling those patterns.1 Preferences We do not yet have a coherent. strive to present an approach which is balanced by focusing on weaknesses as well as strengths. Indeed. The literature in behavioral finance has a tendency to treat the patterns and the theory as one and the same. In this section. prospect theory is incomplete and its formal structure possesses some serious flaws. 2. Both Stein (1989) and Poteshman (2001) report that options implied volatility appears to reflect short-term overreaction. for all its richness.

regret. Instead they rely on . Statman and I suggest that original purchase price serves as a likely reference point. However. In contrast. which is consistent with the self-control discussion in Shefrin and Statman (1985). Feng and Seasholes (2005) present evidence that people can learn to reduce their vulnerability to the disposition effect. They also use probabilities rather than weights.5 Critique: Weaknesses in the Behavioral Approach 43 boundary solutions that are highly unrealistic (Shefrin.. the tendency to exhibit the disposition effect. 2009). Barberis and Huang (2008) uses nonlinear probability weighting. we were quick to indicate that prospect theory cannot provide a complete explanation for the disposition effect. We proposed that a series of additional psychological phenomena underlies the behavior. (2001a) emphasize hedonic editing. 2008a). concavity of the value function in the domain of gains suggests a tendency to realize winners. Our discussion of self-control indicated that some people could learn to reduce. if not eliminate. and self-control. Applications of prospect theory in the behavioral finance literature have also been piecemeal. This feature presents difficulties in respect to the character of equilibrium (De Giorgi et al. Barberis and Xiong. as called for by Tversky and Kahneman (1992).2. Shefrin and Statman (1985) begin with elements of prospect theory.5. Kaustia. In this regard. In introducing the disposition effect. 2009). 2009. 2005.2 Cross-Section Most of the behavioral asset pricing models to explain cross-sectional patterns do not actually model the cross-section. Barberis et al. These phenomena included a realization utility mental accounting component. My sense is that finance academics have been very slow to pick up on the nuances in Shefrin and Statman (1985). Only recently have there been a wave of papers addressing the inability of prospect theory to explain the disposition effect (Hens and Vlcek. and convexity of the value function in the domain of losses suggests a tendency to hold losers. In addition. 2. For example. but use a piecewise linear value function that does not conform to the specification in Tversky and Kahneman (1992).

44 Survey of Surveys models that feature. the findings about underreaction took behaviorists by surprise. the contributions to this literature are something of a behavioral hodge podge. who develops asset pricing theory around the concept of a stochastic discount factor (SDF). Rigor matters. Hong and Stein (1999) develop a model based on differential heuristics. the results in the latter paper are inconsistent with the general literature on long-run survivability (Blume and Easley. coherent. 2.3 Long-Run Dynamics There is reason to believe that a key behavioral contention is in error. taken together they do not provide for a unified.5. The most influential paper in this regard is DeLong et al. Taken together.5. reverse engineering has characterized the behavioral approach to underreaction. As it happens. Barberis et al. 2. This is problematic. (1998) develop a model in which a representative investor makes rational use of Bayes’ rule. systematic behavioral approach to asset pricing. However.4 Contingent Claims The contingent claims approach has provided finance with a framework to unify a series of different models of asset pricing and portfolio choice. However. An excellent treatment of this approach is Cochrane (2005). (1998) develop a model based on time varying overconfidence and biased self-attribution. In addition. see Subrahmanyam (2007). but a single risky security. in that behavioral models have been constructed to explain the observed patterns rather than predict them. This mention . This is the claim that in the presence of rational investors. (1991). Underlying all these models are a variety of interesting stories. Subrahmanyam (2007) mentions contingent claims as one of the four central “paradigms” of finance. irrational investors will survive in the long-term because they take on excessive risk. An exception is DeBondt and Thaler (1985) which formulated its prediction of overreaction based on Kahneman and Tversky (1973). but relative to a flawed regime switching model. Daniel et al. and focus on overreaction and underreaction. 2008).

It seems to me that behavioralizing the contingent claims approach lies at the heart of behavioralizing finance. 2. however. This discussion will address two particular topics. A good illustration of this point can be found in Shleifer (2000) where the formal treatment of sentiment varies from section to section. 2. behavioral asset pricing theorists are prone to build representative investor models. Instead. Whereas neoclassical asset pricing theorists build models in which a rational representative investor sets prices.2 Representative Investor Like their neoclassical counterparts. either to explain it or to provide some form of justification. In some models it corresponds to errors in means. Yet. in other models errors in probabilities.1 Sentiment Sentiment is one of the central concepts of behavioral asset pricing. The absence of a clear formal definition of sentiment makes it difficult to ask generally how sentiment is manifest in asset prices. I discuss the nature of a behavioral contingent claims approach. a risky security is priced relative to one SDF.4. In Section 5. all securities can be priced in terms of a single stochastic discount factor (SDF). (2001a) do not discuss the issue. and the default-free security is priced relative to another SDF.2. This dual structure implies the existence of arbitrage profits. behavioral. However. absent from much of the behavioral asset pricing literature is a coherent definition of sentiment.4. A major result in neoclassical asset pricing is that in the absence of full arbitrage. and the character of a behavioral representative investor. in Barberis et al.5.5 Critique: Weaknesses in the Behavioral Approach 45 occurs in his introduction. sentiment.5. None of the other surveys accord the contingent claims approach much emphasis. (2001a). and specifically how it impacts the cross-section. Yet. he does not return to the topic in the body of his survey. Barberis et al. I regard this deficiency as significant. or otherwise. behavioral asset pricing theorists build models in which the representative investor exhibits the behavior traits of indi- . sentiment seems to be model-specific.

(2001a). Moreover. . as documented in the psychology literature.46 Survey of Surveys viduals. as I discuss in Section 5. aggregation theory strongly indicates that in the presence of heterogeneous investors. (1998) and Barberis et al. the statement applies equally to behavioral economists whose models feature representative investors with behavioral beliefs and preferences identified with individuals. This statement applies to neoclassical economists working in the tradition of Mehra and Prescott (1985). As a result.” a common pattern documented in the behavioral decision literature. the representative investor will reflect but not share the traits exhibited by individual investors. whose asset pricing model features a representative investor with constant relative risk aversion. Examples include Barberis et al. However. economists who use standard representative investor models are prone to “representativeness bias.

sentiment reflects the influence of psychology on beliefs and preferences. See Bruhin et al. (2009a) who suggest a figure of about 20 percent. In this respect. where the expectation is taken with respect to probability beliefs. Expected utility E[u] is the x 47 . Here each realization x of a ˜ random process x generates a utility u(˜). In neoclassical finance. Expected utility theory models risky prospects as random variables or more generally as stochastic processes. which provides an indication of the relative importance of developing a coherent behavioral approach. Survey evidence indicates that the proportion of the population whose behavior conforms to expected utility theory is small.3 Behavioralizing Beliefs and Preferences Generally speaking. I draw from some of their discussion as well as Shefrin (2007) to describe how behaviorists have introduced psychological elements into the neoclassical framework. the expected utility function combines both beliefs (probability distributions) and preferences (utility function). investor preferences are assumed to obey the axioms of expected utility theory. The focus of the present section is the behavioralization of neoclassical beliefs and preferences. An excellent discussion of behavioral beliefs and preferences can be found in Fox and See (2003).

thereby leading to distorted beliefs. overconfidence. I discuss how the change of measure approach can also capture the impact of particular psychological biases. The concept “change of measure” is well established in asset pricing theory. Some biases can be modeled as simple shifts in mean. meaning that investors’ beliefs correspond to the actual probabilities governing x.48 Behavioralizing Beliefs and Preferences expectation of utility u with respect to the probabilities governing x. 3. of 2π course. the operation of multiplying the pdf of x by the change of measure ξ has led to a shift in mean. can all be modeled as biases to the probabilities representing investors’ beliefs. such as optimism. extrapolation bias. leads to the function √1 e− 2 [x−µ] . where it is commonly used to convert physical (or true) probabilities into risk neutral probabilities. I discuss both types of situations below. with a view toward identifying structure. 1). probability distribution. This. For example. and process of maximization? The purpose of this section is to address this question. The probability den1 2 sity function (pdf) of x is √1 e− 2 x . Some types of phenomena. 1).1 Behavioral Beliefs: Heuristics and Biases by a change of measure function ξ(x) = exµ− 2 µ . is the pdf for a random variable which is normally distributed N (µ. relative to neoclassical beliefs. which is accomplished through a change In neoclassical finance. Consider multiplying the pdf of x 2π 1 2 . investors’ beliefs are typically assumed to be free from bias. other phenomena such as ambiguity aversion are inconsistent with beliefs represented as probability densities. and group1 2 ing terms in the exponent. A natural question to ask is whether there are behavioral counterparts to the neoclassical notions of utility function. In this case. let x be the abnormal return to a security which is normally distributed N (0. consider a random variable x which is normally distributed N (0. Doing so. However. the heuristics and biases literature suggest the presence of systematic biases. However. An investor who is unrealistically optimistic might believe that the mean of x is µ > 0. 1). Interpret the zero mean as the security being fairly priced in the market. Hence. In this section. optimism bias can be modeled as a right shift in the mean of x. To illustrate the concept of change of measure. and gambler’s fallacy.

By definition. .1 Behavioral Beliefs: Heuristics and Biases 49 of measure.1 Illustration of a log-change of measure corresponding to optimism bias. meaning the probabilities assigned to very positive returns will be too high (Figure 3. the log-change of measure will be positive. For sufficiently low values of x. For sufficiently high values of x. Consider the log-change of measure ln(ξ). Notice that in the mean-shift example above. effectively measures the percentage difference between the mean-shifted probability and original probability associated with x. Consider the case of optimism bias. The log-change of measure. evaluated at x. ln(ξ) is just a linear function of x. the log-change of measure will be negative. On the other hand pessimism bias corresponds to a shift in mean where µ < 0. 6 4 2 0 -5 -4 -3 -2 -1 0 1 2 3 4 5 -2 -4 -6 x Fig.3. µx − 1 µ2 .1). meaning that probabilities assigned to very negative returns will be too low. In effect. so that 2 µ > 0. the log-change of measure is a metric for the probability bias associated with each potential realization of x. the change of measure function ξ is the ratio of the probability density of the mean-shifted random variable to the probability density of the original random variable. 3.

and x2 represents an investor’s beliefs about this variable. Con2σ 2 sider two normally distributed random variables x1 and x2 . σ) then its pdf has the form √2πσ2 exp(− (x−µ) ). overconfidence is associated with a log-change of measure whose shape is an inverted-U (Figure 3. The ratio of the pdf of x2 to the pdf of x1 provides the change of measure ξ1.2 which transforms x1 into x2 . but positive in the mid-range. suppose that µ1 = µ2 and σ1 > σ2 .2 ) is a metric for the investor’s probability bias function. When it comes to tail events. If σ1 = σ2 . the investor will associate probabilities to these events which are too low. For example. . 10 0 -5 -4 -3 -2 -1 0 1 2 3 4 5 -10 -20 -30 -40 -50 Fig. 3. the log-change of measure will be negative in both the upper tail and the lower tail. Overconfidence is typically understood to imply that investors underestimate risk. then the log-change of measure ln(ξ1. If x1 is the true random variable. then overconfidence is represented by the condition σ2 < σ1 . then the log-change of measure will not be linear in the values of x.50 Behavioralizing Beliefs and Preferences 2 1 If x is N (µ. and an investor believes that the return process is given by x2 . Hence. Hence. For example. if the true return process is given by x1 in the preceding paragraph.2 Illustration of a log-change of measure corresponding to overconfidence bias.2). In particular the log-change of measure will have the shape of an inverted-U.

Both theories focus on the cumulative (or decumulative) distribution function. These biases are dynamic. The first is the cumulative version of prospect theory (CPT). economists are making increasing use of insights from the psychology literature. In this section I discuss two psychologically based approaches to risk preferences. An example is support theory. Notably.2 Behavioral Preferences In expected utility theory. One way is through the use of Choquet integrals. and the second is SP/A theory. 3. preferences are completely captured by the utility function. Aversion to ambiguity can be represented in several ways. Notably. some beliefs are not representable as probabilities. and both involve weighting functions whose shape . but to descriptions of events (Tversky and Koehler. 2009). Therefore. transforming a probability density to a Choquet density. optimism is associated with a shift in mean. However. Over time. Typically. hot hand fallacy and gambler’s fallacy can also apply to beliefs about variance.3. If the recent portion of the realized trajectory features a run of predominantly favorable outcomes. both can give rise to time varying optimism and pessimism. whereas overconfidence is associated with a shift in variance. where decision weights take the place of probabilities and do not integrate to unity over the entire population of possible outcomes (Fox and See. referring as they do to biases in conditional probabilities related to a stochastic process. Hot hand fallacy predicts that after a recent period of high volatility. whereas gambler’s fallacy predicts a decline in subsequent volatility. 2003). the ratio of a Choquet density to a probability density is also a change of measure. in which probabilities are attached not to events per se. behavioral preferences are embodied within both the counterpart to a utility function and in decision weights associated with probabilities. subsequent volatility will continue to be high. while gambler’s fallacy gives rise to excessive pessimism. the change of measure approach applies here as well. 1994. In contrast. Diecidue and La-Ornual.2 Behavioral Preferences 51 Extrapolation bias (also known as “hot hand fallacy”) and gambler’s fallacy both stem from judgmental heuristics. then hot hand fallacy gives rise to excessive optimism. As was mentioned above.

which measures the probability that the outcome payoff is no greater than x. denote the set of possible outcomes by a finite set of real numbers X = {x1 .” In SP/A theory. a risky alternative or prospect is a random variable that takes on values in X. most notably fear and hope.2 0.1 0 0 Behavioralizing Beliefs and Preferences 0. xn }.8 0.05 0.1 0.5 0. Formally. Notably.5 0. the inverse-S shape reflects the principle of diminishing sensitivity.9 0. To introduce some notation. ordered from the lowest outcome x1 to the highest xn .95 1 Probability Fig.7 0.3 0. the inverse-S shaped weighting function gives rise to a U-shaped log-change of measure in which probabilities of extreme events are overweighted and probabilities of intermediate events are underweighted.45 0.4 0.4 0.6 0.85 0. .8 0.65 0.52 1 0. known as “psychophysics. or the probability density . resembles an inverse-S (Figure 3. .2 0.3 Illustration of a weighting function for decumulative probabilities with an inverse-S shape.35 0. One way to describe the probabilities attached to a random variable is to use a decumulative distribution function D. The same probabilistic information is also conveyed by the cumulative distribution function.9 0.3).55 0. where D(x) is the probability that the outcome payoff is at least x.3 0.7 0.75 0.25 0.6 0. .15 0. In CPT. 3. the inverse-S shape reflects the impact of emotions.

the carriers of utility are changes.25 -3. Tversky and Kahneman (1992) suggest using 10 5 0 -7. typically with a point of nondifferentiability at the origin so that the function is more steeply sloped for losses (to the left of the origin) than for gains (to the right of the origin) (Figure 3. meaning gains and losses relative to a reference point.75 -6. 3.3.4).25 3. 10 . 1 See Wang et al. 3.25 1. the utility function is S-shaped with a kink at the origin. the utility function (known as a value function in prospect theory) is concave in gains and concave in losses. 1992). First. (2009) for a survey-based analysis of cross-country differences in responses to prospect theoretic questions.5 5 -10 -8.25 7.1 Prospect theory has four distinctive features.4 Illustration of an S-shaped value function.2 Behavioral Preferences 53 function which measures the probability that the outcome payoff is exactly x.5 -2.75 -5 -10 -15 -20 Gain/loss Fig. 1979. Hence.1 Prospect Theory Prospect theory is a theory of choice developed by psychologists Daniel Kahneman and Amos Tversky (Kahneman and Tversky.2.75 6. not final position. Tversky and Kahneman.75 -1.5 8. Second.5 -5 0 2.

. the changes pertain to differences relative to the endpoints 0 and 1. decision makers engage in editing or framing before formally evaluating risky prospects. However. Fourth. The π-function in OPT is convex. the weights in CPT depend on outcome rank. The weighting function is similar to the h-transform used by Lopes. For the utility function. probabilities are weighted (or distorted) when prospects are evaluated. where probability weights depend only on probability density. As a result. in CPT it is the probabilities of extreme outcomes that are overweighted (very large losses and very large gains). with π(p) > p for small positive values of p and π(p) < p for high values of p less than 1. pγ /(pγ + (1 − p)γ )1/γ . Tversky and Kahneman (1992) proposed a cumulative version of prospect theory (CPT) that uses rank-dependent utility.54 Behavioralizing Beliefs and Preferences a utility function u(x) with the form xα in the domain of gains (x ≥ 0) and −λ(−x)β in the domain of losses (x ≤ 0). 2 Strictly speaking. the changes pertain to differences relative to the reference point. There are several types of editing issues. In the original 1979 version of prospect theory (OPT). meaning the diminished sensitivity to successive changes. Perhaps. the simplest editing issue is the choice of reference point.2 Third. which Tversky–Kahneman generate using the ratio of a power function to a H¨lder average. and therefore do not display a kink. For the weighting function. Unlike OPT. that o is. As I mentioned in the previous section. Ingersoll (2008) points out that the shape of the weighting function only resembles an inverse-S for part of its range. the weighting function π has probability density p as its argument. Tversky and Kahneman (1992) use two weighting functions. a suitably small translation of both functions leads the first derivatives to be finite at zero. Both functions take decumulative probabilities as their arguments. one function for gains and one function for losses. where the decumulative distribution pertains to the absolute value of the gain or loss. Both the S-shape of the utility function and the inverse S-shape of the weighting functions reflect psychophysics. with the kink in question. It features an inverse S-shape. the first derivatives of these functions at zero approach infinity. respectively.

3. In this respect. while the second way is in terms of lives lost. Prospect theory is a descriptive framework.” and A stands for “aspiration.” P stands for “potential. physicians tend to act as if they are more risk averse when presented with the data framed in terms of “lives saved” than when the data are framed in terms of “lives lost. the letters that make up SP/A refer to specific concepts which measure or impact the degree of fear or hope experienced by a decision maker. The complete structure is typically opaque. in one of two ways. not transparent. Lopes titled her 1987 article “The Psychology of Risk: Between Hope and Fear” to capture the idea that the emotions of hope and fear play key roles in choice among risky alternatives. Although the data underlying the two frames are identical.2 SP/A Theory Lopes (1987) and Lopes and Oden (1999) developed a psychologically based approach known as SP/A theory to explain choice among risky alternatives. People who choose stochastically dominated alternatives do so because they do not always grasp the complete structure of the decision tasks they confront.2 Behavioral Preferences 55 Kahneman and Tversky illustrate this issue by describing a medical task in which the data can be presented.3. The “lives saved” frame implicitly sets the reference point at 100 percent fatalities. the segmentation process is often suboptimal.” This choice pattern is consistent with the S-shaped utility function.” Because narrow framing tends to overlook interdependencies between mental accounting structures.2. The “lives lost” frame implicitly sets the reference point at 0 percent fatalities. not a normative framework. A more complex framing issues is the segmentation of a complicated decision task into a series of subtasks. Notably. Tversky and Kahneman present examples in which narrow framing leads to the selection of stochastically dominated choices. The first way is in terms of lives saved.” . S stands for “security. or framed. The structure of each subtask is called a “mental account” and the segmentation process is known as “narrow framing. and people lack the ability to frame complex decision tasks transparently.

1) Because x1 is the least favorable outcome. Because D is a decumulative distribution function. What Equation (3. D(x1 ) = 1. a decision maker who is fearful about exposure to event x1 overweights the probability attached to x1 . the probability attached to the least favorable event x1 is given by Prob{x1 } = h(D(x1 )) − h(D(x2 )). The role of the theory is to describe the criteria leading to the choice. a decision maker chooses a best alternative from a menu {D} of risky alternatives.56 Behavioralizing Beliefs and Preferences In a typical decision task. That is. the probability attached to the least favorable event x1 is given by Prob{x1 } = D(x1 ) − D(x2 ). (2009b) for some recent treatments of the role of aspiration in economic choice. Under the transform h(D). In other words. but experience different degrees of fear. and aspiration A. In SP/A theory. See Diecidue and van de Ven (2008) and Bruhin et al. risky alternatives are evaluated using an objective function whose arguments reflect security S. SP/A theory better predicts choice than prospect theory. potential P . a decision maker who faced a risky prospect D acts as if the decumulative density function is the transform h(D) rather than D itself. Intuitively. Payne (2005) describes an experiment whose results suggest that because of aspiration. (3. In rank-dependent utility. Lopes uses rank-dependent utility to capture the effects of fear. Formally. fear operates on the probabilities associated with unfavorable events. the decision maker who experiences more fear attaches greater importance to the probability associated with unfavorable events than the decision maker who experiences less fear. Roughly speaking. In Lopes’ framework. Suppose we consider two decision makers who face an identical risk. In SP/A theory. .1) effectively states is that increased fear leads to an increase in the slope of the h-function in the neighborhood of 1. the probability that the actual outcome turns out to be x1 or higher is 1. and reduced security corresponds to an increase in the probability attached to the occurrence of some unfavorable event. increased fear stems from reduced security. or prospect D. the fearful decision maker employs a transform h which satisfies: h(D(x1 )) − h(D(x2 )) > D(x1 ) − D(x2 ).

Formally. A decision maker who experiences only hope is associated with an h-function that is strictly concave in D. Lopes uses a convex combination of the power functions hS and hP to capture this case (Figure 3. p > 1 for this case. It is concave in the neighborhood of the origin and convex in the neighborhood of 1. Therefore. the degree to which fear and hope are experienced depends on the degree to which risky prospects offer security S and potential P . In SP/A theory. q > 1 for this case.3) . She calls the function SP for security-potential. Lopes uses an expected utility function with probabilities derived from the h-transform.3. For the second most favorable outcome. Hope induces a decision maker to attach greater weight to the most favorable events. the probability attached to the most favorable event xn is given by Prob{xn } = D(xn ). Formally. the corresponding inequality would read: h(D(xn )) − h(D(xn−1 )) > D(xn ) − D(xn−1 ). Formally. In Lopes’ framework. is associated with an h-function that is strictly convex in D: it is steep in the neighborhood of 1 and flat in the neighborhood of 0.5). the emotion of hope leads a decision maker to employ a transform h which satisfies h(D(xn )) > D(xn ). but not hope. Lopes uses a power function hS (D) = Dq .2) This inequality indicates that increased hope leads to an increase in the slope of the h-function in the neighborhood of 0. Because D is a decumulative distribution function. To capture the impact of both security and potential.2 Behavioral Preferences 57 Rank-dependent utility also captures the manner in which the decision maker experiences hope. a person who neither experiences fear nor hope is associated with an h-function that is the identify function: h(D) = D. and it has the form: n SP = i=1 (h(Di ) − h(Di+1 ))u(xi ). A decision maker who experiences only fear. Lopes uses a power function hP (D) = 1 − (1 − D)p . (3. or potential for short. A person who experiences both fear and hope is associated with an h-function that has an inverseS shape. (3. Hope is associated with upside potential.

.2 0. V is strictly monotone increasing in both of its arguments. 3. Therefore.58 1.5 Illustration of three decumulative transformation functions in Lopes’ SP/A framework.3 0. u is a utility function whose argument is outcome x. u might display a bit of concavity.2 Behavioralizing Beliefs and Preferences 1 h2(D) 0.1 0. the decision maker maximizes an objective function V (SP. Alternatively. Aspiration pertains to a target value α (or range) to which the decision maker aspires. A) in deciding which alternative D to choose from the menu of available prospects. Aspiration points reflect different types of goals.2 0 0 0. aspiration-risk is measured as the probability A = P rob{x ≥ α} that the random outcome x meets or exceeds the aspiration level α.4 h1(D) 0.5 D 0.1). there are situations in which a decision maker is willing to trade off some SP in exchange for a higher value of A. In Equation (4. Lopes and Oden (1999) comment that in practice. Although Lopes uses the assumption u(x) = x in most of her analysis. a decision maker might wish to generate an outcome that would allow the purchase of a particular good or service. For example.9 1 Fig. In SP/A theory.6 0. In Lopes’ framework.8 0.6 0. The “A” in SP/A denotes aspiration.8 h(D) 0.4 0. the aspiration point might reflect a status quo position that corresponds to the notion of no gain or loss.7 0.

but in Kahneman and Tversky (1982) continued to emphasize the importance of regret. but emotionally feels like taking a different decision. the behavioral life cycle framework adds . regret avoidance discourages deviations from behavior viewed as conventional. Gollier and Salani´ (2006) e develop a regret-based model to analyze the character of the risk premium in an asset pricing model. In consequence. Kahneman and Tversky eventually built prospect theory using the S-shaped value function. They pointed out that regret will be magnified by the ease with which a person can easily imagine taking a different decision. Studies of self-control in financial economics tend to emphasize the difficulty in delaying gratification.3. However.3 Regret In the early development of prospect theory. and thinks he or she should take one decision. and in particular applies when the emotion of regret prevents a person from taking a decision that they “think” is appropriate.3 Regret 59 3. Shefrin and Statman (1984. The volume edited by Loomes and Sugden (1982) contains a formal treatment of decision making when preferences are determined by regret. Regret is the psychological pain associated with recognizing after the fact that taking a different decision would have produced a better outcome.4 Self-Control Self-control refers to situations when a person is conflicted. Kahneman and Tversky focused on the role of regret. 1985) modify the concept of regret to reflect pride in the domain of gains. Shefrin and Thaler (1988) extend the neoclassical life cycle framework to incorporate psychological elements associated with self-control and editing. self-control applies more broadly. In addition to traditional neoclassical variables such as consumption. The behavioral life cycle framework features a two-self framework focused on the interaction between a person’s rational self (a “planner” corresponding to activity in the prefrontal cortex of the brain) and an impulsive self (a “doer” corresponding to activity in the limbic system of the brain). a version of status quo bias. 3.

With advances in neuroeconomics. (2008) identify hope with a region of the brain called the nucleus accumbens. (2009) study the impact of breakdowns in prefrontal cortex activity on reputation. 2008). Hare et al.” and associated psychological costs for exhibiting self-control. rules are represented by constraints which resemble corner solutions. For example.60 Behavioralizing Beliefs and Preferences psychological variables to represent “temptation. . the DLPFC. People may choose to violate a rule they themselves establish. In particular. (2009) identify a specific region in the prefrontal cortex associated with self-control.” “willpower. When the work of Shefrin and Thaler (1988) was published. Knoch et al. neuroeconomics is now flourishing.” “rules. the field of neuroeconomics was nonexistent. but doing so entails a psychological cost. economists are beginning to develop sophisticated models of brain activity (Brocas and Carrillo. and providing neurological support for psychologically based models. Knutson et al. However.

and — dividends associated with self-control. The sheer number of effects. regret. indicates that the behavioralization of 61 . and — status quo bias. the disposition effect associated with prospect theory. most at odds with the neoclassical mean-variance approach. — role of dividends as a self-control device and managing risk. These include the following: — — — — lack of diversification associated with the 1/n-heuristic. excessive trading associated with overconfidence. I focus attention on the following series of additional issues: — desire for positively skewed returns. and self-control — the attention hypothesis associated with availability bias. home bias associated with familiarity and ambiguity aversion.4 Behavioralizing Portfolio Selection Theory In their survey article. In this section. Barberis and Thaler (2003) identify empirical features characterizing portfolio selection by individual investors.

However. 2008). . and relate these to empirical evidence. government in 1956. such as social responsibility (Statman. Doing so yields some interesting insights. The maximum amount a person can invest is £30. 4. At present. behavioral portfolio theory emphasizes the taste for positive skewness. which emphasizes meanvariance efficiency.K.62 Behavioralizing Portfolio Selection Theory portfolio theory will be very rich. However. with the minimum investment being £100. There is strong evidence that people have a taste for positively skewed returns.1 4. investors’ 1 Investor preferences also reflect concerns other than returns. or with a monthly standing order. A broad issue that arises in this section concerns the role of optimization.000. LaFleur’s magazine reported that the average household in Massachusetts spends $1. the Premium Bond returns principal on a nominal basis. The desire for positively skewed returns is related to thrill-seeking. Grinblatt and Keloharju (2009) use a clever database to study thrill-seeking behavior by investors. in lieu of monthly interest.1 The purpose of this section is to describe the implications of some of the behavioral preference frameworks for portfolio selection. £50. roughly one-third more than the average annual contribution Americans make to their 401(k) plans. A first step in the behavioralization of portfolio theory is to replace neoclassical preferences with behavioral preferences and undertake a full maximization. it is also a work in progress.760 annually on lotteries. The Premium Bond was launched by the U. Therefore. In 2009. a government department.1 Preference for Positively Skewed Returns: Full Maximization Empirical Issues In contrast to neoclassical portfolio theory. and is issued by National Savings and Investments. A good example of a security with a lottery feature is the British Premium Bond. As with a regular bond. there is ample evidence that individual investors are imperfect maximizers.1. Any British citizen can purchase Premium Bonds. the issues described in this section begin with full maximization and then move on to quasi-maximization.

3 percent of their portfolios to lottery stocks. and possibly negative expected returns.28 percent of their portfolios to lottery stocks. After classifying stocks into lottery stocks and non-lottery stocks.74 percent of the portfolio to lottery stocks. The weight assigned by individual investors is 33 percent. are less positively skewed. During the 1980s. The random assignment would allocate 54 percent of a portfolio to stocks that are diametrically opposite to lottery stocks. A random assignment would allocate 0.4. there is one chance to win. Kumar also reports that even the most wealthy individual investors allocate 7.K. For each single pound invested. At the opposite end of the spectrum lie stocks that are the mirror images of lottery stocks. Lobe and H¨lzl o (2008) document the degree of popularity of this security. Kumar notes that lotteries offer a low probability of a high payoff (positive skewness). and have negative expected returns.7 percent of their portfolios to lottery stocks. have higher prices per share. In contrast. approximately 45 percent of the population invested in Preo mium Bonds. On average. These stocks feature low volatility. and moreover are tax-free. Statman (2002) and Kumar (2009) investigate the extent to which individual investors hold stocks with lottery-like characteristics.1 Preference for Positively Skewed Returns: Full Maximization 63 returns are determined by a lottery. feature low prices per ticket. However. The fraction of the U. Similarly. . population which invests in Premium Bonds grew from approximately 12 percent in 1957 to 40 percent in 2005. low prices per share. institutional investors allocate 0. Lobe and H¨lzl (2008) points out that the tax advantage of Premium Bonds suggests that there are neoclassical reasons to justify why higher income investors would choose to hold Premium Bonds. and higher expected returns. lottery-like stocks feature positively skewed return distributions. individual investors allocate 8. and the weight assigned by institutional investors is 59 percent. the attractiveness of these securities is more likely to lie with the manner in which the bonds package safety of principal with the lottery feature of coupon payments. Kumar compares the portfolio weights individual investors assign to lottery stocks to both a random assignment as well as the portfolio weights institutional investors assign to lottery stocks. for lower income investors.

1. Kumar finds that when economic conditions deteriorate.9 percent less than those who invest moderately in lottery stocks. In particular. A key feature of the mean-variance efficient portfolios is the twofund separation property.1. those who invest in lottery stocks earn 5. any mean-variance efficient portfolio is a convex combination of a unique risky portfolio and the risk-free portfolio.9 percent actually corresponds to 13. The 8. Associated with each mean-variance portfolio is a normally distributed return distribution. Σ be a return covariance matrix. 4. R be a vector of expected returns.1 percent on a riskadjusted basis.9 percent a year less than those who do not.1 Full Maximization with Behavioral Preferences The starting point for behavioralizing portfolio theory involves the replacement of a mean-variance objective function with a behavioral objective function in the optimization procedure. In this case.2 Theoretical Issues Consider the traditional Markowitz portfolio selection problem when security returns are normally distributed. To describe the . ∞) measure the degree of risk tolerance. and τ ∈ [0. In line with risk-seeking behavior in the domain of losses. for varying τ . The two main objective functions used in the literature are the prospect theory weighting function Wi v(xi ) and the SP/A function V (SP. 4. he finds that on a riskadjusted basis. people increase their allocations to lottery stocks. Consider how different an optimized behavioral portfolio return distribution is from a mean-variance return distribution. Similarly. A). Kumar reports that investors who hold larger mutual fund portfolios invest more in lottery stocks. Those who invest heavily in lottery stocks earn 8. Then the mean-variance efficient frontier is the family of solutions to the w-minimization of wT Σw − τ RT w. let w is a vector of portfolio weights.64 Behavioralizing Portfolio Selection Theory Polkovnichenko (2002) reports that the wealthiest group of investors hold almost the same amount in direct equity as indirect equity. Kumar also finds that the highest turnover is among buyers of lottery stocks. Formally.2.

3 illustrates the possible impact of the weighting function in the domain of gains.2 illustrates the return to a behavioral portfolio associated with prospect theory preferences when the weighting function is the identity function. Figure 4. in combination with subcertainty. meaning the weights are themselves probabilities.1 Preference for Positively Skewed Returns: Full Maximization 65 Fig. In the domain of gains.4. With overweighting of probabilities at the extremes. The x-axis is the gross return to the unique risky mean-variance security. The kink at the origin of the value function. Figure 4. contrast. See also Barberis and Huang (2008).1 illustrates the returns to two portfolios relative to the unique risky portfolio. This boundary property stems from the fact that the value function is convex in the domain of losses. the behavioral return is similar to a neoclassical return. as the value function is concave in gains. and in fact are associated with a lower bound on achievable return.1 Illustration of the payoff pattern to the risk-free security and a combination of the risk-free security and unique mean-variance risky security. 4. and the other combines the risk-free security and the unique risky security. One of the two portfolios is risk-free. can produce a flat region with neither gain nor loss. Figure 4. returning the risk-free rate. the behavioral return distribution corresponds to an inverse-S . Notice that the returns to loss states are very low. consider a series of graphs with the return to the unique risky mean-variance portfolio on the x-axis.

4. in that there is a small probability of an especially large gain. 4. thereby giving rise to positive skewness.2 Illustration of the payoff pattern to a portfolio associated with optimization of a prospect theory criterion function with probabilities as weights. The x-axis is the gross return to the unique risky mean-variance security. whereas the probabilities associated with extreme gains tend to be overweighted. Fig.3 Illustration of the payoff pattern to a portfolio associated with optimization of a prospect theory criterion function with an inverse-S-shaped weighting function. in the domain of gains. This occurs because the probabilities associated with intermediate gains tend to be underweighted. .66 Behavioralizing Portfolio Selection Theory Fig. The x-axis is the gross return to the unique risky mean-variance security.

Begin with the middle segment. thereby introducing kinks into the indifference map. For losses. It is flat and the return corresponds to the achievement of the aspiration level. this region will typically be strictly concave and increasing. Prospect theory portfolios tend to feature either boundary solutions or unbounded solutions. this region will lie on the boundary. With concave utility. He and Zhou call this an “ill posed problem.4. In the domain of gains. (2009) and He and Zhou (2009). perhaps to the point of rendering it completely flat as depicted in the figure. and accentuated risk seeking in the domain of gains. overweighting of extreme loss probabilities can introduce a convex feature that operates in the opposite direction from the shape of the value function in the domain of losses. Probability weighting tends to flatten this region. the overweighting of extreme gain probabilities introduces a non-convex feature that operates in the opposite direction from the shape of the value function in the domain of gains. emphasizing the roles of fear. Notice that the SP/A return function has three segments. Figure 4. To the left is a region where the investor fails to achieve the aspiration level.1 Preference for Positively Skewed Returns: Full Maximization 67 Notably. The effect of probability weighting can produce a non-boundary flat return in the domain of losses. With linear utility. See also De Giorgi et al. one or more of these . hope. Daniel Kahneman suggested to Statman and me that Lopes’ SP/A theory might provide a superior approach to CPT. and explored more fully in Shefrin (2008a).” The early versions of Shefrin and Statman (2000) used prospect theory as the basis for a behavioral portfolio theory. and aspiration.4 illustrates the contrast between a typical SP/A-based portfolio return and the risky mean-variance return. The published version of Shefrin and Statman (2000) develops an SP/Abased behavioral portfolio selection theory. the weighting function is rank dependent. a feature noted in the early drafts of Shefrin and Statman (2000). See Diecidue and van de Ven (2008) and De Giorgi (2009) who focus on the impact of aspiration on portfolio returns. Depending on the utility function and specific parameters. and in the absence of probability weighting. At the right is a region where the investor achieves a return above aspiration. In commenting on these drafts.

Because of probability weighting. In this regard. let equilibrium state prices be given by ν. 4. the return pattern typically has the shape of an inverse-S. With concave utility. The x-axis is the gross return to the unique risky mean-variance security. To see why. regions might disappear. With linear utility. Begin with the equilibrium consumption profile cS chosen by an investor with SP/A preferences based on utility function u.68 Behavioralizing Portfolio Selection Theory Fig. For equilibrium prices ν. (4.4 Illustration of the payoff pattern to a portfolio associated with optimization of an SP/A criterion function with an inverse-S-shaped weighting function. the region is very small. Then. Recall the discussion in Section 3 concerning the use of change of measure to model behavioral beliefs. normalize so that the components P ∗ sum to unity. relative to a change of measure. the shape is typically concave and increasing. the various behavioral return patterns below can be expressed as expected utility maximizing solutions. . but features a spike at the highest return. choose Pi∗ and ∗ such that: Pk ∗ Pi∗ ui /Pk uk = νi /νk .1) where ui and uk are marginal utilities of consumption evaluated at cS . and absent probability weighting. Now ask whether we can identify a probability density function P ∗ such that an expected utility maximizing investor with utility function u and beliefs P ∗ would have selected cS ? Let i and k be two states.

5. For example. An SP/A portfolio combines safety in the form of a . 4.5 adds a stylized change of measure to Figure 4. whereas expected utility maximization typically calls for an increasing return profile.4. corresponding to the change of measure associated with the inverse-S shaped weighting function in SP/A theory. Shefrin and Statman (2000) point out that SP/A-based portfolio theory provides a rationale for the attractiveness of securities like the Premium Bond. In Figure 4.5 Illustration of the change of measure associated with the payoff pattern to a portfolio associated with optimization of an SP/A criterion function with an inverse-S-shaped weighting function. The U-shape reflects the fact that the probabilities of extreme events are overweighted and intermediate events are underweighted. Figure 4. then the function L = ln(P ∗ /Π) is a log-change of measure that captures the impact of the SP/A investor’s behavioral preferences and behavioral beliefs. the right side of the U is weaker than the left side. Notice that the pattern can be constructed by beginning with a U-like shape.4. Notice that the middle section features a jump at the discontinuity marking the boundary of the middle region. If Π denotes the objective probability density function.1 Preference for Positively Skewed Returns: Full Maximization 69 Fig. The x-axis is the gross return to the unique risky mean-variance security. typically followed by a decline to reflect the fact that the behavioral portfolio return is flat at the aspiration point.

wD + wU = 1. It seems to me that modeling bounded rationality can be more difficult than modeling full rationality. which gives rise to sub-utilities (VD . VU ). with the portfolio decision being to solve maxw F . 4. be aware that the impact of probability weighting is similar to the impact of biased beliefs. and the propensity to trade are all symptoms of bounded rationality. mental accounting gives rise to incomplete diversification. Instead. wU ).2 Quasi-Maximization The disposition effect. The downside protection account features a low aspiration level (avoiding poverty) and a high probability of achieving this level. 4. The upside potential account features a high aspiration (riches). behavioral portfolio models can be delicate. investors might frame the portfolio selection task by identifying two aspirations.2 Bounded Rationality Shefrin and Statman (2000) also suggest that bounded rationality typically prevents investors from choosing portfolios using full optimization. the investor establishes a nonnegative allocation w = (wD . They also point out that the behavioral portfolio approach explains the attractiveness of many structured products.” By its nature. investors decompose decision tasks into simpler sub-task and use quasi-optimization. favoring of dividendpaying stocks. Although this section has emphasized preferences without regard to biased beliefs. as investors ignore covariance relationships. and a low or modest probability of achieving this aspiration level. attention hypothesis. VU ). Barberis and Huang (2008) develop a dynamic prospect theory model to .70 Behavioralizing Portfolio Selection Theory risk-free bond. a risky bond. Each task is associated with its own mental account and aspiration point. Shefrin and Statman call such a portfolio a “layered pyramid. one for downside protection and the other for upside potential. For example. respectively.1. For example. status quo bias. An allocation w is evaluated using a criterion function F (VD . Recent behavioral approaches to structured products includes Rieger (2008) and Bernard and Boyle (2008). In their model. In this respect.2. and a risky security with the features of an out-of-the-money call option.

Goetzmann and Kumar (2008) document the existence of skilled investors who hold concentrated portfolios. Seru et al. suggest that extreme caution is necessary in developing models to capture behavioral portfolio selection. I focus on issues which emphasize mental accounting issues in a dynamic environment. prospect theory value function. and then recast the model in a dynamic context.2. they tend to choose stocks which co-move. such as diversification.2 Quasi-Maximization 71 study the disposition effect. the disposition effect. cash dividends. with the disposition of the median investor improving by 4 percent after a year of trading . In this regard. Below I offer a few brief comments about diversification and the disposition effect. together with the ill posed property of prospect theory-based portfolios. and then concentrate on cash dividends and options. What they find is that prospect theory can lead investors to sell their losers and purchase more of their winners. and self-control.1 Empirical Issues Section 2 covers many of the key empirical findings about individual investors’ portfolios. As I mentioned in Section 2. regret. 4. thereby depriving themselves of the benefits of diversification. which supports the original contention in Shefrin and Statman (1985) that the disposition effect is likely driven by several psychological features. (2009) report a similar pattern. such as mental accounting. See also Dorn and Huberman (2005). They also report that older investors have better diversified portfolios and trade less aggressively than their younger counterparts. They point out that as investors increase the number of stocks in their portfolios. Their finding provides an indication that some investors learn to achieve increased self-control. and options. cannot explain the disposition effect. they use static theory to identify the solution. in apparent contradiction of the disposition effect. Feng and Seasholes (2005) find that susceptibility to the disposition effect declines by 72 percent with sophistication and experience.4. Results like those in Barberis and Xiong (2009). In this section. by itself. especially in connection with riding losers. This is an interesting result.

2 The material in this section is taken from Shefrin (2008b). In particular. Self-control and mental accounting suggest several hypotheses linking consumption and dividends. Moreover.2 Graham and Kumar (2006) measure the preference for dividends by portfolio dividend yield. Their approach emphasizes the roles of self-control. Lehenkari (2008) uses Finnish data to provide intriguing evidence that investors are more prone to the disposition effect for stocks they purchased than for stocks they received as gifts or as part of an inheritance. In respect to regret. along with a clientele effect. They report that irrespective of income. retired investors rely on a heuristic in which they finance consumption out of an income mental account. with dividend yields of individual investors’ portfolios changing slowly over time. About 20 years after being proposed. older households prefer dividend-paying stocks over non-dividend-paying stocks. These findings support the predictions of behavioral dividend theory about the impact of age and income. those with low income exhibit a somewhat stronger preference for dividend-paying stocks than households with higher income. 4. They develop a behavioral explanation for why some investors framework find cash dividends attractive. Hedonic editing and regret suggest a connection between stock market movements and dividends. Recent experimental evidence by Fogel and Berry (2006) strongly supports the importance of regret in explaining the disposition effect. the preference for dividends is relatively stable. but are reluctant to “dip into capital” which resides in an asset mental account.1. with the overall results being supportive of the behavioral approach.2. hedonic editing. .72 Behavioralizing Portfolio Selection Theory experience. For younger households. investors might prefer to finance consumption out of dividends than out of capital in order to avoid the regret of selling off stock which subsequently appreciates by a large amount.1 Cash Dividends The first paper in behavioral finance written by economists is Shefrin and Statman (1984). most of these hypotheses have come to be tested. The theory suggests that older. and regret. mental accounting.

Dong et al. Graham and Kumar regress portfolio yield on a series of variables including age and a retirement status dummy. Graham and Kumar report that older. They report that households trade around dividend events such as ex-dividend days. low-income investors focus more on industries with a reputation for paying higher dividends. their survey asked questions about personal finance and consumption matters. Of these investors. . Also. Consistent with the predictions of behavioral dividend theory. On a weekly basis. 555 own or previously owned shares in exchange-listed companies or investment funds. less than 20 percent own individual stocks. households that are older or have less income are net buyers of stock. (2007a). (2005) develop panel data from a survey of individual investors in the Netherlands. retired investors hold about 80 percent of their portfolios in dividendpaying stocks. as does a dummy for retirement status. retirement status increases the demand for dividends. Before the exdividend date. The reverse is true for the stocks of firms in the “computers” and “business services” industries. while younger investors hold about 65 percent of their portfolios in dividend-paying stocks. Dong et al. dividend announcement dates. 40 percent of the investors are at least 55 years of age and 20 percent are at least 65 years of age. (2005) find that consumption from dividends is confined to investors who are either older or have low incomes. They report that the age variable has a positive and significant coefficient estimate.035 survey respondents. and dividend initiations. These households also increase their holdings after dividend initiations and exhibit abnormal buying behavior after dividend announcements. Further support for dividend clienteles stems from Graham and Kumar’s analysis of household trading behavior. Of the 2.2 Quasi-Maximization 73 In terms of stock selection.4. In particular. where individual investors’ trades play an especially important role. This pattern is consistent with the idea that older. These behavior patterns appear to affect the prices for the stocks of small firms. older investors with low incomes attach more weight to utility stocks than do younger investors with high incomes. Consistent with the findings of Graham and Kumar (2006) and Baker et al.

because stock dividends involve costs but do not increase the fundamental value of investors’ positions. has a mean of $56. They find that in cases where firms cannot pay a cash dividend. income. with the median household reporting zero dividend income. dividends.. as opposed to no dividends. The CEX provides panel data on household consumption.74 Behavioralizing Portfolio Selection Theory Outside this group. interest income is $1. The brokerage data provide information about how individual investors behave when they receive dividends.” Baker et al. individual investors do not consume a large portion of their dividends but instead re-invest their dividend income. respectively. This is a framing effect. On average. Financial wealth is typically around a third of total wealth. Total expenditure including durables is about $50. investors indicate that they would prefer companies to pay a stock dividend. wealth. which includes dividends but not capital gains. Dong et al. However. behavioral theory only predicts that older.272 household-year observations. their findings support the behavioral prediction for stock dividends. and demographic information. Instead. Total income. (2007a) cover the period 1988– 2001 and involve 3. they find that the propensity to consume from dividend income is less than from regular income. Younger investors with moderate to high incomes have little need to finance consumption with dividends. The distribution is skewed. (2007a) provide the strongest evidence for a connection between consumption and dividends. Their analysis makes use of two types of data sets: the Consumer Expenditure Survey (CEX) from the Bureau of Labor Statistics and brokerage account data. (2005) assert that the findings described previously run counter to the behavioral predictions from Shefrin and Statman (1984). According to Dong et al. For the mean household. The CEX data used by Baker et al. .207 and dividend income is $891. behavioral theory does not predict that all households consume dividends. retired households and low-income households favor dividend-paying stocks to finance consumption. (p. interest and dividends account for 4 percent and 2 percent of total income.000 (measured in December 2001 in dollars). As Dong et al.789. (2005). 124) indicate. as pointed out earlier. In contrast to the findings reported by Baker et al. “in principle stock dividends are no more than stock splits.

Moreover.1.49. especially structured products.2 Options and Structured Products Shefrin and Statman (1993a) apply the framework developed in Shefrin and Statman (1984) to other aspects of portfolio design. Although considerations from neoclassical finance there indicate little merit to covered call writing. Baker et al. 4. and purchasing call options . hedonic editing allows the premium to be emphasized psychologically. which has been the most popular option strategy for individual investors. using put options for protection on stocks they hold.2 Quasi-Maximization 75 The BLC predicts that the marginal propensity to consume from the income mental account exceeds the marginal propensity to consume from the asset mental account. the same as the marginal propensity to consume out of total income. noting how writing a call option which is fully covered by a holding the underlying stock masks a loss on the option when the stock price increases. thereby generating an additional psychological benefit.4. the same cannot be said for behavioral finance. by segregating the option premium received from other cash flows such as dividends received. The general issue Shefrin and Statman (1993a) address how psychological issues impact the design and marketing of financial products. Shefrin and Statman (1984) focus on the use of option strategies. The analysis in Shefrin and Statman (1993a) identifies conditions under which investors will find it attractive to use options in particular ways such as writing call options on stocks in their possession. most notably covered call writing. Shefrin and Statman (1984) suggest that dividends flow through the income mental account and that capital gains flow through the asset account. Shefrin and Statman (1993a) use prospect theory to emphasize the framing elements inherent in a covered call strategy.2. (2007a) estimate that the marginal propensity to consume from dividend income is 0. which in turn exceeds the marginal propensity to consume from capital gains. whereas the marginal propensity to consume from total current-year returns is close to zero. This implies that the marginal propensity to consume from dividend income is about the same as the marginal propensity to consume from total income.

Lakonishok et al. using techniques emphasized in finance textbooks such as straddle positions. consumption (ct ) . just as the premium received when writing call options is enhancing. a particular financial product with complex option properties. However. empirical evidence indicates that individual investors do not follow rational exercise strategies. This feature indicates that mental accounts can be more complex than simple position holders for single assets. Interestingly. (2007) updates some of the discussion in Shefrin and Statman (1993a). from their money market accounts to fund the purchase of out-of-the-money call options. During the dot-com bubble. many did so for growth stocks. 2003). some do tend to use call options for speculative purposes. they also find that volatility trading. Surprisingly. 4. Moreover. The mental accounting treatment of covered call writing involves the construction of mental accounts using both a stock and a call option. only accounts for a small fraction of option trading volume. but not value stocks.2. most individual investors do not use protective puts. in accordance with the features emphasized in behavioral portfolio theory. McConnell and Schwartz (1992) describe how individual investors use the interest. at each time t.2 Theoretical Issues A traditional model of dynamic portfolio choice involves an expected utility maximizing investor choosing. They find that a large percentage of call writing stems from covered call positions. Evidence discussed in Lakonishok et al. In this regard. but not principal. (2007) document the relative frequency of different option trading strategies by different classes of investors. This strategy provides both downside protection and upside potential.76 Behavioralizing Portfolio Selection Theory for speculative purposes. Because of hedonic editing. A key takeaway from the discussion is that when cognitive limits can prevent investors from making use of multiple securities to form complex mental accounts. financial engineers might be able to create value by doing so for them. by creating financial products such as the LYON. the premium to be paid when purchasing put options for protection is detracting. as many exercise options prematurely (Poteshman and Serbin. McConnell and Schwartz (1992) describe this investing pattern as the motivation for the development of the liquid yield option note (LYON).

1971. In the traditional model. most trading activity would involve marginal adjustments to a diversified portfolio with the purpose of rebalancing or dealing with liquidity needs to finance consumption. Consider a behavioral analog to the traditional framework.2 Quasi-Maximization 77 and securities (xt = xt.2) where δ is a subjective time preference discount factor. This means that the portfolio serves to hedge uncertain labor income so as to smooth consumption over time. attention hypothesis. Write the analog of expected utility as an objective function U . and requires that the marginal benefit of this purchase be equal to the marginal cost. and preference for positive skewness. use of dividends to finance consumption. xt. a stochastic stream of labor income (Lt ). (2010).j ∂u ∂u .3). and the expectation is taken over a subjective probability measure (P ) which an investor associates with the underlying stochastic process. The marginal benefit is the expected marginal utility of consumption at time t + 1 generated by the marginal investment in security j. (4. Foldes. The standard Euler condition for this problem involves the purchase of a marginal unit of security j at time t. with a view to modeling issues such as the disposition effect. Unless labor income is highly volatile. Let U have as its arguments consumption stream c = [ct ]. 1978). based on initial wealth (W ) and stochastic labor income (L). = E qt+1.4. status quo bias. The Euler condition has the following form: qt. The marginal cost is the foregone marginal utility of consumption at time t. which we subsequently develop into a version featuring quasi-maximization. and stochastic prices (qt ) (Merton.3) In this setting.1 · · · . the expected utility function has the form: E t δ t u(ct ) . as the increased expenditure on security j comes at the expense of less consumption at time t.J ) given initial wealth (W ). the purpose of the portfolio is to manage the risk profile of the investor’s consumption stream. See Hoffmann et al. Begin with a full maximization extension to Equation (4. portfolio choices .j δ ∂ct ∂ct+1 (4.

by weighting the costs of foregone marginal current consumption associated with a marginal security purchase against the expected marginal future consumption so generated.78 Behavioralizing Portfolio Selection Theory x = [xt ].4) reflects the fact that increasing xt reduces yt+1 = xt+1 − xt . 2008).j δ∂u/ ∂ct+1 .j ∂U = ∂ct qt+1. x. Those components take the form of ∂U/∂xt. The behavioral Euler condition is: qt. The terms ∂U/∂yt and ∂U/∂yt+1 capture realization utility (Barberis and Huang. In the corresponding behavioral framework. investors with predictable streams of labor income make small but frequent adjustments to their portfolios.j t+1 ∂U ∂ct+1 + ∂U ∂U + − ∂xt ∂yt t+1 ∂U . prices q = [qt ]. apart from his consumption. and q as arguments allows for an investor’s preferences to reflect not only consumption. whereas an investor who sells at a .j . The minus sign associated with the term t+1 ∂U/∂yt+1 in Equation (4. In the neoclassical framework. and y are random variables. is enhanced when his or her portfolio grows.j . with the summation t+1 over the support of outcomes at t + 1. In this respect. ∂U/∂yt.j . and ∂U/∂yt+1.j ∂U/∂ct+1 is the analog to E qt+1. meaning the impact of trading a position. the consumption– savings feature is augmented by additional considerations. an investor who sells at a gain might experience positive realization utility. (s)he adds three additional components to the neoclassical calculus. meaning the psychological effect the investor experiences from the value of his portfolio at different points in time. The term ∂U/∂xt captures the effects of marginal evaluation utility. the behavioral expression might not correspond to an expectation with respect to the investor’s beliefs. but also the performance of his or her portfolio and the impact of gains and losses from trading. The inclusion of x. Notice that the criterion driving portfolio choice is consumption and savings. Note that because of probability weighting. and is diminished when it falls. y. The arguments c. Here an investor’s sense of well-being at a given moment.j of security j at time t. When a behavioral investor contemplates a marginal increase in his or her holdings xt. ∂yt+1 (4.4) The term t+1 qt+1. and probability beliefs (P ). changes in portfolio positions y = [xt − xt−1 ].

from regret avoidance Samuelson and Zeckhauser (1988). A similar statement applies to the disposition effect. with realization utility associated with gains or losses on a security. mental accounting. and relates to the consequence stemming from a reference action. the behavioral terms are unsigned. 1985).j . Notably. Gains and losses enter through the dependence of U on y. (2009) document the strength of status quo bias in saving behavior.3 Benartzi and Thaler (2007) and Benartzi et al. these might be based on perceived performance by others in an investor’s peer group. Mitchell et al. evaluation utility. The extension involves issues related to preferences as well as issues related to biased beliefs. As to preferences. and realization utility stem from the dependence of U upon x and y. The behavioral approach stresses the importance of reference points and aspiration levels. and an additional 11 percent make only one trade.4)) 3 Regret is one component of realization utility. and possibly increased realization utility at t. Reference points and aspiration levels for both consumption and portfolio values can be embedded in the specific form for U . The corresponding cost is measured as foregone utility of consumption at t. increased evaluation utility at t + 1. 1990). Depending on circumstances and the investor’s characteristics. 2009). u . relative to a pre-specified reference point being a prime driver of whether or not to sell a security in the portfolio (Shefrin and Statman. In our behavioral Euler condition (Equation (4. A new line of inquiry investigates the role of social networks on investors’ decisions (Kaustia and Kn¨ pfer. Status quo bias significantly contributes to investors’ lack of diversification. the additional behavioral terms can either serve to augment or reduce the benefits of a higher xt. Notably. for example.4. The behavioral extension to the neoclassical portfolio choice model can be structured to incorporate the behavioral phenomena mentioned above. A behavioral investor who contemplates a marginal increase in his or her holdings of security j at time t views the benefits in terms of utility from additional consumption at t + 1. (2006) report that 80 percent of participants in 401(k) accounts initiate no trades in a two-year period.2 Quasi-Maximization 79 loss might experience negative realization utility (Thaler and Johnson. Status quo bias leads investors’ reluctance to trade resulting.

the larger the increase in volume. and liquidity are sufficiently strong. Huddart et al. Certainly. just the opposite of status quo bias. which can lead investors to underestimate risk. and is more concentrated in smaller stocks that are difficult to value. An example of a bias is overconfidence. rebalancing. (2009) report that volume tends to spike when the price of a stock crosses a previous low or high. In the framework being sketched here. the longer the time since the previous low or high. the two coexist. belief effects such as biases and ambiguity are encapsulated within the criterion function U . as measured by Baker and Wurgler (2007). Status quo bias does not mean that investors refrain from trading altogether. The effect is accentuated during periods when sentiment is high. investors can overcome status quo bias if they have enough conviction in their stock picking skills to feel little potential for regret (Kahneman et al. if the needs for hedging. See also Statman et al.. In another sense. Most individual investors have only the vaguest notion of how security returns are jointly distributed (Benartzi. In this last regard. Moreover. the juxtaposition between the two effects highlights the nature of heterogeneity within the population. (2006) who document that trading volume is higher in rising markets than declining markets. Barber and Odean (2000) emphasize that overconfidence leads investors to trade too much. For this reason.80 Behavioralizing Portfolio Selection Theory status quo bias is captured by ∂U/∂yt < 0 for yt > 0 and ∂U/∂yt > 0 for yt < 0 with a point of non-differentiability (kink) at yt = 0. in that active traders tend to concentrate their trades in just a few stocks. investors will overcome status quo bias and trade. they are prone to engage in mental accounting. This violation underscores the fact that prospect theory is not a framework for full . In one sense. Instead. Mental accounting is a key reason underlying the violation of stochastic dominance in the prospect theoretic choice experiments reported in Kahneman and Tversky (1979). only that other forces must be strong enough to overcome their reluctance. Likewise. 2001). investors are unlikely to maximize fully. 1991) or derive sufficiently high evaluation utility from their portfolios. both to simplify the decision task and to exert self-control. thereby suggesting that beliefs lie at the root of the issue.

With this perspective in mind. albeit imperfectly. . as in Shefrin and Thaler (1988).4) across securities might not be compatible with a single set of beliefs. with portfolio composition changes reflecting jumps. As a result. with diversification corresponding to a “taste for variety.2 Quasi-Maximization 81 maximization. I note that the beliefs used in connection with Equation (4.” Moreover. Instead investors might apply Equation (4. or perhaps a mental account by mental account basis. the process for portfolio selection will resemble the process for consumer choice. In this regard.4) on a security by security basis. but a heuristic-based framework to explain how people structure decision tasks and engage in quasi-optimization. as status quo bias is overcome. choice in prospect theory features an evaluation function which mimics the structure of an expected utility function. quasi-maximization is likely to be accomplished in the context of a set of specific heuristics.4. Behavioral portfolio models are also likely to feature a mixture of diffusion processes and jump processes. which provides the formal BLC framework underlying the explanation for dividends advanced in Shefrin and Statman (1984).

82 . The concept is very powerful. the SDF is treated as time invariant. risk aversion reflects the medical condition “seasonal affective disorder” (SAD). SAD involves the impact of number of hours of sunlight on rates of depression in the population. 2005. Foldes. the result of habit formation. Kamstra et al.5 Behavioralizing Asset Pricing Theory Modern neoclassical asset pricing theory is built around the concept of a stochastic discount factor (SDF) (Cochrane. In log– log space. the negative of the slope of the SDF is often interpreted as the representative investor’s coefficient of relative risk aversion. but deterministic. some authors use models displaying time variation in the SDF. and is therefore related to reference point-based behavioral preference models. Habit formation involves the notion of reference consumption levels. (2009) develops a model in which time varying. As was mentioned in Sections 2 and 3. 2000). and allows most asset pricing models to be expressed as special cases of a general framework. often interpreted as the marginal rate of substitution for a representative investor. In their previous published work. Examples are Constantinides (1990) and Campbell and Cochrane (1999) which feature stochastic risk aversion. However. In many neoclassical models. The neoclassical SDF is structured as a monotone decreasing function.

1 Stochastic Discount Factor 83 the authors find that in both the Northern and Southern hemispheres.1 Stochastic Discount Factor Consider the fundamental SDF-based asset pricing equation that the price q of an asset with random payoff x is the expected value of its discounted payoff. thereby behavioralizing asset pricing theory. .5. This statement applies to models designed to analyze the cross-section. (2009) demonstrate that under plausible conditions. given that the cross-section features a multitude of securities. where m is the discount factor used to capture the effects of both time value of money and risk. Formally. and analyzes the implications of a behavioral SDF for particular asset classes. 1 The (5. the behavioral asset pricing literature has mostly developed around two-asset models where the two assets are the risk-free security and a risky security. equilibrium will feature investors either holding (i) a risk-free portfolio or (ii) an unbounded portfolio.1 5. describes the behavioral extension. De Giorgi et al. An example of a weakness involves market models in which investor preferences conform to prospect theory. empirical estimates of the SDF show it to be oscillating rather than being monotone decreasing (A¨ ıt-Sahalia and Lo. which is clearly problematic. Much of the discussion below involves strengths. As noted in Section 2. 2000. 2004). q = E[mx]. briefly outlines the nature of a neoclassical SDF. and rise in the spring as the number of hours of daylight increases. In this section I discuss how behavioral elements explain the oscillating shape. As it happens. The extension serves to identify both strengths and weaknesses in the behavioral approach. stock returns tend to rise in the autumn as the number of hours of daylight falls. Rosenberg and Engle. The discussion below provides a formal definition for the SDF. This section describes how the neoclassical SDF framework can be extended to capture behavioral beliefs and preferences.1) first part of this section is drawn from Shefrin (2008c).

m takes on higher values in unfavorable states than favorable states. In this case. That is. in that absent pure arbitrage opportunities. giving rise to monotone declining shape across states. The greater the degree of risk aversion. In this regard. and risk premia associated with all assets. Risk neutrality corresponds to the case of a zero slope. The SDF m also serves as a change of measure. 2005). 5. options.1. transforming the true (also known as objective or physical) probability density function into the risk-neutral measure (Cochrane.1. in the neoclassical representative investor framework. the log-SDF can be interpreted as the sum of a component relating 2 Dividing by q leads to a similar expression on a return or per dollar basis. the steeper the slope. namely where R denotes the security return. it underlies prices. In the neoclassical framework. both m and x are random variables. be they fixed income securities. returns. 5. the discount factor m typically varies across payoff levels in order to reflect that risk is priced differently across payoff levels. equities.84 Behavioralizing Asset Pricing Theory In Equation (5. and other derivatives. the negative slope of the neoclassical SDF reflects aversion to risk. and corresponds to the case when m is a linear function of the market return.1). log(m) is linear in the log of the gross market return. the amount of current consumption the representative investor is willing to sacrifice in exchange for a marginal unit of future consumption in a particular state. . m can be interpreted as a conditional marginal rate of substitution. Another special neoclassical case occurs when equilibrium prices are set as if there is a representative investor with correct beliefs and preferences that conform to constant relative risk aversion (CRRA).2 The SDF is a powerful concept. As I mentioned above.1 Neoclassical SDF Think of the SDF m as state price per unit probability.2 Behavioral SDF Shefrin (2008a) establishes that when all investors have power utility. The capital asset pricing model (CAPM) is an important special case of the SDF framework.

In this regard. if the magnitude of sentiment is sufficiently large. Figure 5. The market sentiment term Λ is the sum of a log-change of measure ln(ξ) and a time preference shift term parameter δS . prices of all securities are efficient when the corresponding log-change of measure is zero.2) The log-change of measure has the same interpretation as in Section 3. A¨ ıt-Sahalia and Lo (2000) and Rosenberg and Engle (2004) estimate the empirical SDF. both graphed against ln(g). a result of suboptimal learning processes. but for the probability density function associated with equilibrium prices rather than an individual investor. Sentiment reflects investors’ misjudgments and misreactions to changing fundamentals. The behavioral SDF depicted in the figure oscillates. In this respect.5. However. The neoclassical SDF is linear with a negative slope. with the probabilities of extreme negative events overweighted and extreme positive events underweighted. Formally. a case discussed in further detail below. This type of SDF occurs when the market is dominated by a combination of overconfident optimists and underconfident pessimists. The source of γ and δ is obtained from an aggregation procedure which is discussed below. bias is relative to probabilities associated with market efficiency.3) indicates the manner in which sentiment impacts market prices through the SDF. then Λ will drive the shape of the SDF. or more precisely its . (5. Using options data. especially its shape. (5.1 Stochastic Discount Factor 85 to fundamentals and a component relating to market sentiment. Equation (5. Λ = ln(ξ) + ln(δS ). Just as in Section 3. a term γ measuring the market coefficient of relative risk aversion.3) Equation (5. Note that sentiment is a stochastic process whose underlying probabilities derive from the objective process for fundamentals. the equation relating the log-SDF and sentiment has the form: ln(m) = ln(δ) − γln(g) + Λ. The fundamental variables entering the SDF are aggregate gross consumption growth g. the log-SDF corresponds to the neoclassical case.3) implies that when sentiment is zero.1 contrasts a neoclassical log-SDF and a behavioral log-SDF. and a term δ measuring the market time discount factor.

First.1 Illustration of the decomposition in Equation (5. Both papers find that the empirical SDF has the behavioral shape portrayed in Figure 5. This approach can be viewed as an inadvertent test of whether the empirical SDF is behavioral or not.1. largely because of the role played by sentiment Λ in the SDF m. The Rosenberg–Engle paper is especially interesting. they restrict the SDF to have the traditional neoclassical shape displayed in Figure 5. projection in respect to returns on the S&P 500. 5. Notably. in that the authors estimate the SDF in two ways. behavioral equilibrium naturally features differential volatilities for the objective pdf and risk-neutral pdf.86 60% Behavioralizing Asset Pricing Theory 50% 40% 30% behavioral log-(SDF) 20% Sentiment Function 10% 0% -4% -3% -1% 1% 3% 4% 6% -10% -ln(g) = neoclassical log-SDF -20% -30% ln(g) Fig. Barone-Adesi et al. Their empirical findings provide strong support that the SDF is behavioral. their approach allows the objective pdf and risk-neutral pdf associated with the representative investor to feature different volatilities. (2008) estimate the empirical SDF using a GARCH-based filtered historical simulation technique.3). Second. In this regard.1. contrasting a behavioral SDF and a neoclassical SDF. they use a free form Chebyshev polynomial procedure that involves no such restriction. The new GARCHbased filtered simulation approach produces a shape for the SDF which .

(2008) indicate that the tail did indeed flatten in this period. . I suggest that the shape of the behavioral SDF depicted in Figure 5.1 Stochastic Discount Factor 87 features dampened oscillations relative to the Gaussian-based approach of Rosenberg–Engle.1.2. As a result. pessimistic investors would become less underconfident. Sharpe ratios are bounded from above by the coefficient of variation of the SDF. the 2005 edition of Shefrin (2008a) predicted that after 2001.1 reflects a negative correlation between investors’ errors about the first and second moments of returns. This difference in volatility has profound implications for the magnitudes of risk premiums and Sharpe ratios. Using survey data. I suggest that the shape of the SDF reflects the joint distribution of investors’ biases.5. The findings reported in Barone-Adesi et al. The behavioral SDF is more volatile than its neoclassical counterpart.1. 10% behavioral log-(SDF) 0% -4% -3% -1% 1% 3% 4% 6% Sentiment Function -10% -ln(g) = neoclassical log-SDF -20% -30% -40% -50% -60% ln(g) Fig. 5. thereby resulting in a flatter left tail for the logSDF depicted in Figure 5. see Figure 5. In this respect. In Shefrin (2008a).2 Illustration of the impact of less underconfidence by pessimists in Figure 5. the more volatile behavioral SDF admits higher risk premiums and Sharpe ratios than does the less volatile neoclassical SDF.

Barber et al. In regard to preferences. although he suggests that CARA is unrealistic from a behavioral perspective. They focus on the special case when investors share the same coefficient of relative risk aversion. Risk Tolerance. . but lose money to professional investors. being based on investors’ decisions. However. degree of risk aversion.3 Therefore. and rate of impatience.1 Aggregating Beliefs. they also document considerable heterogeneity.2 How Markets Aggregate Investor Attributes The empirical evidence strongly indicates that there is considerable heterogeneity in respect to investors’ beliefs. recall from the discussion in Section 3 that the change of measure technique applies both to preferences and to beliefs. with the weights reflecting relative wealth or consumption. Jouini and Napp (2006. Jouini and Napp’s analysis of aggregation under the assumption of constant absolute risk aversion is more general. Barber et al. (2009b) provide evidence that individual investors as a group influence prices.88 Behavioralizing Asset Pricing Theory 5. and Barber et al. Shefrin (2008a) presents an equilibrium aggregation result for a complete markets model featuring power utility. The log-SDF for the case when investors have 3 In a series of papers. meaning that it stems from surveys concerning these variables. Other evidence is indirect. Shefrin (2008a) also analyzes aggregation under the assumption of constant absolute risk aversion (CRRA). Some of this evidence is direct. readers are referred to Shefrin (2008a) and the works of Jouini and Napp (2007). He demonstrates that the equilibrium market pdf is a generalized weighted H¨lder avero age of the individual investors’ pdfs. such as the composition of their portfolios. For a detailed discussion of the aggregation theorems. a core issue in asset pricing theory is to understand how markets aggregate investors’ beliefs and preferences. 5.2. and the exponents being coefficients of relative risk aversion. (2009a). it is sufficient to describe the general features of the representative investor for the case of power utility (CRRA). 2007) establish similar results. For the purpose of the current discussion. demonstrating that some individual investors possess superior skills. (2005). and Rates of Impatience The study of the aggregation question dates back at least as far as Lintner (1969).

This means that γ will not measure the representative investor’s degree of relative risk aversion. For example. heterogeneity leads the representative investor’s utility exponent γ to be state dependent. and have normally distributed beliefs with different means. even though every investor has an exponential discount factor. the market does not aggregate heterogeneous beliefs and preferences in a way that leads to the existence of a typical neoclassical representative investor. the representative investor might exhibit aversion to ambiguity. the representative investor’s stochastic process will typically violate Bayes’ rule. If investors differ in their intertemporal discount factors. Because consumption shares are time varying.2 How Markets Aggregate Investor Attributes 89 CARA utility functions is similar in form. Moreover. the representative investor’s discount factor will typically be non-exponential. neoclassical and behavioral. rather than being consumption-weighted as is the case with CRRA utility (Shefrin. . and therefore non-normal. The characteristics of the aggregation process hold clear lessons for asset pricing theorists of all stripes. even if none of the individual investors do. Stated somewhat differently. because H¨lder averages do not always sum o to unity.5. share the same intertemporal discount factor δ. 2008a). the representative investor’s degree of risk tolerance will be a consumption weighted convex combination of the individual investors’ degree of risk tolerance. Typically. If the means are sufficiently different across investors. However. Heterogeneity leads the representative investor’s beliefs to be different from those of the individual investors. or a representative investor whose preferences and beliefs feature typical behavioral characteristics. because the aggregation process relates to discount factors. except that all averages are unweighted. not discount rates. In this case. the generalized weighted H¨lder average is a simple wealtho weighted average of probability density functions. suppose that all investors have log-utility. The representative investor’s intertemporal discount rate δ is a nonlinear function of the individual investors’ discount rates. the representative investor’s discount rate will also reflect probability aggregation. At any moment in time. such a pdf can be multi-modal.

For example.2. in the sense that investors predict stronger continuation after downturns than after upturns. if regret potential increases after declines in aggregate consumption growth. whereas status quo bias would. In this respect. the result would be asymmetric volatility in the returns to particular securities. The others exhibit status quo bias which leads them to refrain from trading as a result of being situated at kinks in their preference maps. would reflect the beliefs and preferences of investors who engage in trade. I made the point that liquidity issues can arise because at a given time t only a subset of investors actively trade. One indication of the extent to which status quo bias plays a role is the change in trading volume. the degree of status quo bias might be conditional on circumstances. an increase in status quo bias would lead to a decline in trading volume. To the extent that the liquidity premium increases after such a decline. consider how market prices reflect a liquidity premium. 5. For example.3) reflects the manner in which the SDF incorporates sentiment. see Statman et al. while Equation (5. In this regard.2 Liquidity Premium and Asymmetric Volatility The preceding discussion describes the manner in which sentiment reflects the aggregation of investors’ beliefs and preferences.3 Risk Aversion Puzzles Assuming the existence of a typical neoclassical representative investor can lead to puzzles. As a result. market prices at t. but not investors who refrain from trade. In Section 4.2. then status quo bias might increase following declines in the return to the market portfolio. particular investors wishing to engage in trade at t might find that inducing sufficiently large positions on the part of their active trading partners requires market prices to move by a “nonnegligible” amount.90 Behavioralizing Asset Pricing Theory 5. As a general matter. and hence sentiment. Typically. (2006). At the same time. Jackwerth (2000) estimated the . Asymmetric volatility might reflect asymmetric hot hand fallacy on the part of some investors. It is in this sense that in the short-run sentiment reflects a liquidity premium. asymmetric hot hand fallacy would tend not to predict sharp upturns after steep declines. In this respect.

However. they find that market risk aversion is highly variable over time. The culprit in these risk aversion estimation exercises is the assumption that sentiment is zero. then the upward sloping segments will appear to imply negative risk aversion. Consider a complete market model featuring two investors. If sentiment is nonzero in practice. All of these studies find coefficients of risk aversion for the market that are extreme.3 Aggregation and the Shape of Sentiment The easiest way to understand the shape of sentiment is by working with examples featuring log-utility. it mostly tends to be U-shaped and time varying. Incorrectly assuming it to be zero in theory forces the models’ risk aversion parameters to pick up the effects of market sentiment. The A¨ ıt-Sahalia-Lo range for CRRA is 1–60. Jackwerth (2000) estimates imply that the market’s coefficient of CRRA ranges between −14 and 27. Negative values. with a mean of 7. However. also features highly variable rates of risk aversion that take on exotic values. and that both investors believe it to be normally distributed.6. which indicate risk seeking. Instead.5. Reasonable benchmark values for the coefficient of relative risk aversion (CRRA) involve the range 1. 5. Assume that the log of consumption growth is normally distributed. pessimism. are especially surprising. and overconfidence. negative values for CRRA and values over 6 are at odds with survey data. Jackwerth points out that the shape of the absolute risk aversion function.5–3. such as optimism. which extends Jackwerth’s analysis from the market to individual stocks. The Rosenberg–Engle estimated range for the CRRA is 2–12. Moreover. The papers by A¨ ıt-Sahalia and Lo (2000) and Rosenberg and Engle (2004) also estimate market risk aversion under the assumption that sentiment is zero.3 Aggregation and the Shape of Sentiment 91 coefficient of absolute risk aversion for the market in the case when sentiment is (implicitly) zero. Blackburn and Ukhov (2006). let the two investors .5. is not monotone declining and time invariant. and oscillates. which he estimates for the market. with a few outliers between 0 and 1 and between 5 and 6. with the SDF conforming to a monotone decreasing function.5. as neoclassical theory would suggest. Suppose that both investors have log-utility functions. whose initial wealth levels are equal.

1 Risk and Return Behavioral Mean-Variance Returns A mean-variance (MV) portfolio is a portfolio of assets that maximizes expected return for a specified return variance. in that they overpredict the probabilities of crashes. Finally.4 5. as depicted in Figure 5. the H¨lder average becomes a simple convex o combination. whereas the bullish investor drives the shape of market sentiment for favorable states. Notably. As a result. then the sentiment function will tend to be oscillating. Assume that one investor is excessively bullish about mean consumption growth. For example. a U-shape implies that the market overweights the probabilities of extreme events. in the special case of log-utility. The resulting log-change of measure for the representative investor will be U-shaped. the risk premium for a security is based on its return covariance with any risky mean-variance portfolio. be they favorable or unfavorable. For now. Notably. Therefore. . while the second investor is excessively bearish about mean consumption growth. 2007). the sentiment function can have an inverse U-shape. in the CAPM the market portfolio is mean-variance efficient.1. if bearish investors tend to be underconfident. Associated with the bearish investor is a log-change of measure which is linear with negative slope.4. Recall that the market equilibrium aggregates pdfs. In neoclassical asset pricing. the bearish investor drives the shape of market sentiment for unfavorable states. 5. See Odean (1998b) for the first theoretical treatment dedicated to the question of how overconfidence impacts asset prices. As pointed out in Section 3. let both investors hold correct beliefs about the second moment. while bullish investors tend to be overconfident. then they will underestimate the probabilities of extreme events.92 Behavioralizing Asset Pricing Theory disagree about the first moment. The inverse-U can be interpreted as reflecting the “black swan effect:” (Taleb. where sentiment is zero. associated with the bullish investor is a log-change of measure which is linear with positive slope. If investors are overconfident. not log-change of measure functions.

For the purpose of illustration. meaning risk which is priced. the shape of a meanvariance portfolio return function is obtained by inverting and tilting the SDF (Cochrane.1.4 Risk and Return 93 and this is why risk premiums are based on the covariance between the security’s returns and the returns to the market portfolio. Shefrin. the SDF also prices all mean-variance efficient portfolios. Notice that the shapes in Figure 5. 2005. tilted shapes in Figure 5. the weight attached to the risk-free asset in both portfolios is high. In the case of CAPM. The SDF can be used to price all assets. Therefore. This is because the peaks and . Effectively. with a negative slope coefficient. the key lies in understanding how sentiment impacts the nature of the return distributions for both MV portfolios and individual assets. even when sentiment is nonzero.3 contrasts the return pattern for a behavioral meanvariance portfolio to the return pattern for a neoclassical mean-variance portfolio. and therefore the SDF can be used to generate the return distribution for an MV portfolio. How does sentiment impact the relationship between risk and return? The short answer is in the same way as in neoclassical asset pricing. Figure 5. Because risk premiums for all securities are based on return covariance with MV portfolios. First. Each curve in Figure 5. whereas the neoclassical MV curve corresponds to the case when sentiment is zero.3 correspond to the inverted.3 is a plot of the linear function described in Shefrin (2008a) that links the return of an MV portfolio to the consumption growth rate through the SDF.1.3 has three points worthy of note. Just as in neoclassical asset pricing theory. the return to a traditional MV portfolio is approximately the return to a portfolio consisting of the market portfolio and the risk-free security. the SDF generates the mean-variance frontier. the SDF is linear and negatively sloped. The behavioral MV curve corresponds to the case in which sentiment is given by Figure 5. including portfolios. This means appreciating that systematic risk. the return to a risky mean-variance porfolio is a linear function of the SDF. The relationship is the same. Hence.5. 2005). has both a fundamental component and a sentiment component. the return to a behavioral MV portfolio is more volatile than the return to a traditional MV portfolio. Second. Figure 5.

while extreme out-of-the-money call options are underpriced.9 0. This means that extreme out-ofthe-money put options on the market portfolio are overpriced.05 Mean-variance Return (Gross) 1 Efficient MV Portfolio Return 0. Notice that in Figure 5.1.1.1 Behavioralizing Asset Pricing Theory Behavioral MV Portfolio Return 1. sentiment is positive at the extreme left and negative at the extreme right.75 -4% -3% -3% -2% -1% 0% 1% 2% 3% 3% 4% 5% ln(g) 6% Fig.95 0. In this case the risk premium to a security is the product of the security’s beta and the risk premium of the meanvariance benchmark. maximizing expected return involves the exploitation of nonzero sentiment. valleys in the behavioral MV portfolio correspond to exposure from risky arbitrage.85 0. As a result.3 This figure contrasts the return pattern for a neoclassical MV portfolio and a behavioral MV portfolio associated with Figure 5. a true mean-variance portfolio would feature a short position in extreme out-of-the-money put options on the market portfolio and a long position in extreme out-of-the-money call options on the market portfolio. Third.8 0. Therefore. expected returns and risk premiums for securities are determined by security betas. . based on the objective pdf.94 1. 5. This is a true MV portfolio. taken with respect to a risky mean-variance portfolio benchmark.

but fall off quickly as the rate of consumption growth declines. The definition of coskewness is particularly appropriate when the SDF has a quadratic-like U-shape. Harvey and Siddique (2000) study the cross-section of stock returns using coskewness relative to the market portfolio. Securities whose returns are highly correlated with the returns to an MV portfolio will be associated with high degrees of systematic risk. To understand the implications that the behavioral MV shape in Figure 5. Recall that the SDF will be . skewness pertains to the third moment.5. Following Barone-Adesi and Talwar (1983).2 Coskewness MV-returns are very negatively skewed relative to the market portfolio because returns to a behavioral MV-portfolio are not only extremely low in low-growth states. controlling for covariation with the market return.4.4 Risk and Return 95 5. and mean-variance preferences are neutral in respect to all moments higher than the second.3 has for the nature of risk premiums for individual securities. remember that the returns to the behavioral MV portfolio depicted in Figure 5. consider what systematic risk entails. coskewness measures the extent to which a security’s return covaries with the squared market return. One definition of coskewness is the beta of the security’s return relative to the squared market return. In this regard. Systematic risk is risk associated with the returns to an MV portfolio. The reason why skewness can be an issue for a behavioral MV portfolio involves the risky arbitrage feature of true MV portfolios.3 are negatively skewed relative to the market portfolio. They propose several definitions of coskewness. After all. With this definition. Some readers might be surprised that skewness is an issue at all in a discussion about mean-variance portfolios. securities whose returns are high in systematic risk will feature negative skewness relative to the market portfolio. An investor who takes a large short position in extreme out-of-the-money put options on the market portfolio will earn a very low return when the return on the market portfolio is very low. An investor who takes a long position in extreme out-of-the-money call options on the market portfolio will earn a high return when the return on the market portfolio is very high. Therefore.

8 percent associated with the use of the market factor. the shape of the MV function is an inverted-U. the patterns are similar. the market return. and B/M. size. a security that is high in systematic risk. the combination of the market factor and coskewness explains 68. See BaroneAdesi et al. there is a negative premium to holding stocks whose returns exhibit positive coskewness relative to the market portfolio. but not all. meaning risk that is priced. the shape of the behavioral MV return pattern depicted in Figure 5.3. Of course. This suggests that when sentiment has the shape depicted in Figure 5. and momentum is −0. . However. Harvey and Siddique (2000) study coskewness using a four-factor model. in the left portion. aspects of systematic risk.71. where the factors are. In this case. Coskewness plays a role analogous to a factor loading in a multifactor asset pricing model. For momentum. size. Harvey and Siddique find that the correlation between coskewness and mean returns of portfolios sorted by size. and momentum. However. will mimic the MV-return pattern.1 percent of cross-sectional returns. (2004) for evidence that this is the case. This value rivals the 71. the behavioral MV-return pattern does not have the same feature as an inverted-U. Notably.1. Observe that this type of MV return function will feature negative coskewness with respect to the market portfolio. book-to-market equity (B/M). This is because coskewness measures the amount of mean-variance skewness that a security’s return adds to an investor’s portfolio. In this situation. and therefore also feature negative coskewness.3 is not exactly an inverted-U.96 Behavioralizing Asset Pricing Theory U-shaped when the sentiment component has a U-shape and is large relative to the fundamental component of the SDF. respectively. In the right portion of Figure 5.5 percent of cross-sectional returns. All of these findings are consistent with MV portfolio returns having the shape depicted in Figure 5. book-to-market equity. recent winners feature lower coskewness than that of recent losers.3. This means that much of the explanatory power of size. Harvey and Siddique point out that the market factor by itself explains only 3. negative coskewness with respect to the market portfolio will capture some. book-to-market equity. and momentum in the returns of individual stocks plausibly derives from coskewness.

. In this case hZ > 1. E[δg −γ rZ ] (5. hZ represents the degree to which Z is overpriced (when hZ > 1) and underpriced (when hZ < 1). the impact of the numerator of Equation (5. he finds that the empirical SDF in his analysis is not monotone decreasing over its range.4) where the expectation is taken with respect to the true underlying stochastic process Π.3 Decomposition of Risk Premiums Let Z be an arbitrary security with return rZ . i1 = i1. Notice that when investor errors lead to an increase in the risk-free rate.5) is negative. and take the expectation with respect to the true underlying process. Keeping in mind that Φ is proportional to a change of measure. Suppose that we shift probability to higher valued realizations.4 Risk and Return 97 Dittmar (2002) discusses the superiority of nonlinear models over linear factor models such as the Fama–French three-factor model.e is the value of the risk-free rate when sentiment is zero and prices are efficient. the risk premium on Z declines.5) has a straightforward interpretation.4) implies that hZ = 1.e − i1 ) − 1 − hZ Cov[g −γ rZ ] .4. Notably. + i1. 5.5) where i1 is the market risk-free rate of interest. Shefrin (2008a) establishes that the risk premium associated with Z is given by: E[rZ ] − i1 = (i1. Relative to fundamental value E[δg −γ rZ ]. Letting Φ = exp(Λ) define hZ = E[δΦg −γ rZ ] . and i1. He develops a flexible nonlinear pricing kernel approach which is in the tradition of a zero sentiment representative investor model.e E[g −γ ] hZ (5. When sentiment Λ is zero. When prices are efficient. In this regard. but instead has a U-shaped pattern. Poti (2006) uses a quadratic U-shaped SDF to extend the Harvey–Siddique analysis. then Equation (5. Equation (5.4) is to shift probability weight across different realizations of the product g −γ rZ . In this case. and so the third term on the right-hand side of Equation (5.5.e .

i1.5) is zero. In this respect.4) implies that hZ may not be equal to 1.5) for the risk premium decomposes into a fundamental component (the middle term on the right-hand side) and a sentiment component (the first and third terms). Equation (5. Notably.5) specifies how the cross-section is driven by a combination of fundamental risk and sentiment-based risk.4 Observe that the expression (5. if hZ = 2. Think about an out-of-the-money call option in an example when the representative investor is excessively optimistic. then Z is overpriced by 100 percent. 5 Equation (5. pure arbitrage opportunities.e hZ Z which in this case equals 50 percent(= 1−2 ) of what the gross risk free rate would be were prices 2 efficient.5).6 hZ In addition to the discussion about coskewness. the first and third terms are typically nonzero.5 Fundamental risk is determined by the return covariance with the fundamental component of the SDF.e − i1 = 0 so that when prices are efficient. Notice that Z being overpriced causes its risk premium to decline relative to the case of market efficiency. thereby left shifting and distorting its objective return distribution at a given node in the uncertainty tree.1 Cross-Section Equation (5. In addition.5) should be understood with reference to the projection of sentiment onto the returns for individual securities. Equation (5. The variable hZ measures the degree to which security Z is overpriced. 5. When sentiment Λ is nonzero. For example. . the equilibrium return distribution to Z might be different in the case when prices are inefficient relative to the case when they are efficient. sentiment risk is determined by the term 1−hZ . as some investors build portfolios with only a vague understanding of covariance. I also want to mention some experimental evidence.3. when prices are inefficient. For the last decade. as long as some investors are able to exploit. 6 In Section 4 I mentioned that the framework for behavioral portfolio selection resembles the framework for consumer choice. Equation (5. I have surveyed money managers. 4 This statement is conditional on the equilibrium return distribution rZ . and eliminate. analysts.4.98 Behavioralizing Asset Pricing Theory the third term on the right-hand side of Equation (5.5) is just equal to the middle term on the right-hand side. and its connection to the cross-section.5) indicates that the overvaluation of Z causes the i (1−h ) expected return to Z to change by 1. This feature does not negate Equation (5. Conditional on the interest rate. The excessive optimism will increase the price of the option.

speculative difficult-to-arbitrage . Most of those surveyed judge that risk is positively related to market beta. dividend premium. number of IPO. book-to-market equity. When sentiment is high. 2007) discuss various measures of sentiment that have been used in behavioral studies. while negatively related to market capitalization. in practice they form judgments as if the two are negatively related. The second is that both low risk and high returns generate positive affect. and equity share in new issues. In contrast to the neoclassical framework. judgments about risk and return turn out to be negatively related. The general finding from this survey is that although those surveyed believe that in theory risk and return are positively related. People might rely on a heuristic whereby they believe that the stocks of good companies are representative of good stocks.5. Baker and Wurgler suggest that sentiment will impact prices of some stocks more than others. and recent momentum.4 Risk and Return 99 and students to elicit their judgments about risk and return in connection with the cross-section. Not surprisingly. They suggest that the returns to stocks which are difficult both to value and to arbitrage will be more sensitive to sentiment than stocks which are both easier to value and easier to arbitrage. Judgments about risk are consistent with the neoclassical framework. detrended log-turnover. They then develop a new index of sentiment based on the following six specific measures: — — — — — — closed-end fund discount. The first is representativeness. first-day return on IPOs. Baker and Wurgler (2006. and so the affect heuristic can also explain the negative relationship. Two psychological principles might be at work in explaining why people form judgments as if they believe that risk and return are negatively related. The heart of their argument involves the concept of a sentiment seesaw. most also believe that the relationship between expected returns and these same variables has the opposite sign as with risk.

limits to arbitrage can prevent market prices from being efficient. it corresponds to the sum of the first and third terms. can feature lower expected returns than safer stocks. 5. Empirically. and that the covariance between investor error and investor wealth be zero. subsequent returns to speculative stocks. when past sentiment has been low. namely that in the short-run. more difficult-to-arbitrage stocks are indeed lower than returns to safer stocks which are easier to arbitrage. this component effectively corresponds to the third term in the right-hand side of Equation (5. meaning investors who are free from error. the behavioral position is different. In contrast. A sentiment beta explains the sentiment component of the risk premium. a stock’s sensitivity to sentiment is measured in terms of a sentiment beta.100 Behavioralizing Asset Pricing Theory stocks become overvalued. subsequent returns to speculative stocks. so that investor bias is unsystematic. so that any errors are not concentrated within the investor population (Shefrin and Statman. stocks that appear to be riskier in terms of fundamentals. Baker–Wurgler report that returns are predictable. unconditionally. because information investors are typically reluctant to take on the risk associated with the large positions required to eliminate inefficiencies completely. the behavioral perspective holds that information traders. In Baker–Wurgler. Conditional on the risk-free rate. conditional on the value of sentiment in the prior month. . Conversely. and safe easy-to-arbitrage stocks become undervalued. The conditions for efficiency essentially require that the average investor error be zero. will not guarantee market efficiency. so that prices are efficient at all times. the reverse occurs.5). A notable feature of the analysis is that because of mispricing. more difficultto-arbitrage stocks are higher than to safer stocks which are easier to arbitrage.5 Long-run Fitness The neoclassical perspective has tended to be that information investors will exploit any inefficiencies immediately. When sentiment is low. 1994). When past sentiment has been high. In contrast to the neoclassical perspective.

Blume and Easley establish that the time average of investor j’s marginal utility to investor k’s marginal utility is given by: [ln(δk ) − IΠ (Pk )] − [(ln(δj ) − IΠ (Pj )].5 Long-run Fitness 101 What about the long-run? Will the existence of information traders eliminate any price inefficiencies? The neoclassical answer to this question is yes. and the idea was formally discussed by DeLong et al.7) They call the difference [ln(δj ) − IΠ (Pj )] investor j’s survival index. that investors who are subject to error can take on more risk than information investors. (5. the ratio of investor j’s marginal utility approaches +∞. In contrast. Suppose that investor j has a lower survival index than investor k. (5. with the argument being that investors who are subject to error will lose their wealth to information investors. The analysis in DeLong et al. However. In particular. This literature has recently been surveyed by Blume and Easley (2008). and surviving. Their framework features CRRA utility functions. Notice that there are two components to the survival index. then over time investor j’s consumption must go to zero. Given two discrete probability density functions Π and P over a series of K states. thereby earning higher returns in the long-run. and the second to errors in beliefs. Lawrence Summers first proposed the intuition underlying the behavioral perspective. The first component relates to patience. (1991). The . define K IΠ (P ) = k=1 Π(k)ln(Π(k)/P (k)).5. (1991) is controversial. if physical resources are bounded from above and investor utility is CRRA with risk aversion parameter γj ≥ 1. and describes conditions under which investors who are prone to error survive in the long-run and impact prices.6) and call this variable the relative entropy of P with respect to Π. Suppose that objective probabilities correspond to Π. the behavioral answer has been not necessarily. At the center of the fitness lies the concept of entropy. the main result is at odds with the general literature on long-term survivability or fitness. Consider a set of investors indexed by j. Then with probability one.

it is entirely possible that investors with erroneous beliefs persist in the long run. Setting aside impatience. In this respect. However. even if they did so. there may be no informed investors at all. because they believe that the noisy signal . with the formal argument being loose. All investors make forecasts of the change in drift. Dumas et al. in that all investors are subject to some form of error or bias.6 Pricing Dynamics A stream of literature is developing that focuses on asset pricing dynamics. not the least being the survey by Blume and Easley mentioned above. Therefore. Blume and Easley (2008) state that they have difficulty following the logic in the latter paper. who assume that the aggregate consumption growth rate follows a mean-reverting diffusion process with a time varying stochastic drift. Moreover. (2009) build on the work of Scheinkman and Xiong (2003). Alternatively. their heirs might not possess the same degree of superiority. Investors do not observe the drift directly. (1991). This means that there is good reason to believe that the fundamental assertion put forth in DeLong et al. the market can feature incompleteness in combination with overlapping generations with finite lifetimes. they read more into their signal forecast errors than is appropriate. Indeed. (1991) is incorrect.102 Behavioralizing Asset Pricing Theory survival condition suggests that long-run fitness is determined by being sufficiently patient relative to other investors and/or being sufficiently better informed about objective probabilities. Instead they receive a noisy signal whose change is equal to the sum of drift plus a noise term that is uncorrelated with the change in drift. but for reasons different from those suggested in the analysis of DeLong et al. the results in Blume and Easley (2008) indicate that informed investors will dominate pricing in the long-run. 5. For example. In this respect. the model appears to be vaguely specified. better informed investors might not live long enough to extract most of the wealth from their less informed counterparts. some investors overreact to their forecast errors for the rate of change in the signal. Of course. The volume edited by Hens and Schenk-Hopp´ (2008) contains e many interesting contributions.

in a CRRA-framework the price of zero coupon riskfree bonds can be expressed as q = E[δ(t)g −γ ]... where the expectation is taken with respect to objective probabilities. they overweight the contribution of their signal forecast errors when predicting how the drift term will subsequently change. Duffie et al. x = 1. E[g −γ ] ≥ E[g]−γ . 2007). 5. Moreover. .1) stipulates that the price of any security is given by q = E[mx]. For zero coupon risk-free bonds. In this regard. thereby leading to overconfidence on their part. With g −γ being a convex function. for risky bonds.5) comes to mind. interest rate spreads and movements pose major puzzles. x can be taken to be 1. Realized bond yield volatility exceeds the upper bound limit implied by the expectations hypothesis (Shiller.6 Pricing Dynamics 103 features a positive correlation with the change in drift. Bond yields are more sensitive to macroeconomic announcements (Gurkaynak et al. Equation (5. From a neoclassical perspective.6. Two prominent examples are the expectations hypothesis and the credit spread puzzle for corporate debt. 2008a). The credit spread puzzle relates to the fact that the common factors explaining default probabilities explain at most 25 percent of the variation in spreads. thereby lowering the associated interest rate. where the expectation is taken with respect to the representative investor’s pdf (Shefrin. in which case E[g −γ ] = E[g]−γ . Equation (5. 1978). As a result. The expectations hypothesis requires that the expected risk adjusted return to a long default-free bonds should be equal to the expected return from rolling over short-term riskfree bonds over the life of the long bond. Of course. Notice that applying a mean preserving spread will increase the bond price. 2005. This overweighting leads them to underestimate drift volatility. Beginning with the case of certainty.1 Behavioral Interest Rate Models In a general SDF-based framework. there appears to be a common latent factor which explains 75 percent of the residual. The term structure of interest rates provides fertile ground for applying behavioral asset pricing models. both as a factor in explaining the failure of the expectations hypothesis and the existence of the credit spread puzzle. and so bond prices are simply given by E[m]. Equivalently.5.

However. rather than the objective pdf. as it leads market prices to be expressed as convex combinations of prices that would occur in a series of hypothetical homogeneous belief models. bond prices decline with maturity to . o As a result. whereas a behavioral SDF oscillates and is prone to being volatile over time. The example features log-utility investors with heterogeneous beliefs. The preceding section about aggregation and the shape of sentiment discusses an example with just two investors. the expectations hypothesis might hold subjectively. then this substitution would be ill advised for a representative investor with correct beliefs. Consider an example of a behavioral term structure theory as developed in Shefrin (2008a). In the homogeneous case. relative to the beliefs that underlie equilibrium prices. if sentiment is nonzero. and the exponents being coefficients of relative risk aversion. However. the representative investor is quite willing to substitute a rollover strategy involving short-bonds for a long-bond. the SDF will also have the form of a convex combination. In this two-investor example. homogeneous beliefs which are independent and identically distributed (i.) lead the yield curve to be flat. As was mentioned above. bond prices will be convex combinations of bond prices that would occur in a series of homogeneous belief examples. One thing that distinguishes a behavioral theory of bond pricing from a neoclassical theory is the nature of the SDF. In other words. where it is the objective probabilities that underlie the dynamics of bond returns and yields. but not objectively.i. a consequence of yield being a convex function of bond price. This is very convenient.5) captures how nonzero sentiment leads bond returns to feature both fundamental and sentiment-based components. In the special case of log-utility. the H¨lder average is a simple convex combination. In particular. one bullish and the other bearish. with the weights reflecting relative wealth or consumption.d. Equation (5. the equilibrium market pdf is a generalized weighted H¨lder average of the individual investors’ o pdfs. heterogeneous beliefs lead the yield curve to be downward sloping. In a behavioral approach. A neoclassical SDF is monotone decreasing and stable over time. At the margin.104 Behavioralizing Asset Pricing Theory A clue to why the expectations hypothesis fails comes from understanding that bond prices are determined by the representative investor’s pdf in q = E[δ(t)g −γ ].

In a promising approach. Notably. Xiong and Yan (2009) extend the log-utility example from Shefrin (2008a) from a discrete exchange economy to a continuous production economy. Most importantly. The returns to investment dIt /It follow a diffusion process with instantaneous drift ft . they offer an explanation for a tent-shaped linear combination of forward rates to explain bond return dynamics. Hence.8) Because Equation (5.1. It (5. consider what happens to the ratio of the bond price for the investor associated with lower yield to the bond price for the investor associated with the higher yield. (2009) provide insight into the dynamics of interest rates.1 Modeling Interest Rate Dynamics Consider how the style of models developed in Scheinkman and Xiong (2003) and Dumas et al. Therefore. Xiong and Yan (2009) suggest that a two-investor log-utility model can capture important features of interest rate dynamics. This ratio grows with maturity. Xiong and Yan (2009) show that it effectively 2 determines the instantaneous risk-free rate of interest rt = ft − σI .8) is a linear production process relating to the instantaneous marginal rate of transformation between current and future consumption. Just as Shefrin (2008a) suggests that the two investor log-utility heterogeneity model can capture the oscillating shape displayed by the empirical SDF.5. 5.6. they confirm the failure of the expectations hypothesis. Brownian motion ZI (t). At the heart of the dynamics in Xiong and Yan (2009) are a set of assumptions about consumption growth and investment. as the maturity increases. They also confirm the tendency for wealth fluctuations to amplify volatility and time variation in bond prices and premia. and standard deviation σI . . The result will be a declining yield curve. the decline is larger for the higher yield. Notably.6 Pricing Dynamics 105 preserve constant yield. the wealth weighted average convex combination of bond prices at different maturities will lead to the lower yield being emphasized. The equation governing dIt /It is: dIt = ft dt + σI dZI (t).

They observe a white noise signal dSt = dZS (t) which they mistakenly believe to be informative about the fundamental shock lt . ft ..14) is common in the bond pricing literature (Balduzzi et al. and determine a process for a conditional objective density for τ {lt |{fτ }t =0 }. the two investors’ reactions to dSt are equal and opposite. (5.10) imply: 2 drt = −λf (t)[rt − (lt − σI )]dt + σf dZf .11) Equation (5.10) provide the basis for the objective processes governing the evolution of aggregate consumption growth. the bond price at time t of a zero coupon default-free bond with maturity τ has the form: ˆ ˆ BH τ. (5. lt = − log(B H ). τ (5.13) . lt = e−af (τ )ft −al (τ )lt −b(τ ) .8)–(5. Finally. Equations (5. Denote an τ ˆ estimate of lt by lt .9) where λf governs the mean reverting speed of ft . so that one investor acts bullishly and the other bearishly.106 Behavioralizing Asset Pricing Theory However. These are conditional on publicly available information at t represented by {fτ }t =0 . Each investor’s subjective estimate of lt at t is conditioned on {fτ . ft . The variables in these equations that are not identified relate to the associated Brownian motions and their coefficients. lt follows an Ornstein–Uhlenbeck process: l)dt dlt = −λl (lt − ¯ + σl dZl (t).8)–(5. Equations (5.12) The corresponding yield is given by: 1 ˆ YH τ. S(τ )}t =0 . 1998). Moreover.10) where λl governs the mean reverting speed of lt . (5. The two investors in Xiong and Yan (2009) hold biased τ beliefs. Xiong and Yan (2009) establish that in equilibrium featuring homogeneity. (5. ft itself is determined by a mean reverting diffusion process with a tendency to revert to an unobservable long-run mean lt according to: dft = −λf (ft − lt )dt + σf dZf (t). Together.

similar to the empirical results in Dai and Singleton (2003). The sentiment component is ˆ proportional to the difference between the objective estimate lt and the representative investor’s estimate formed as a wealth-weighted convex combination of the individual investors’ estimates.6 Pricing Dynamics 107 In terms of yield. in a heterogeneous belief economy. the relationship is hump shaped. the relationship is monotone decreasing. heterogeneous beliefs lead equilibrium bond prices in Xiong and Yan (2009) to be wealth-weighted convex combinations of bond prices in homogeneous belief economies. However. Sentiment introduces a second shock.5. The equation for dmt /mt indicates the percentage change in the SDF over the next interval of length dt. consumption growth fundamentals will lead the log-SDF to decline at rate r plus a random shock to aggregate consumption growth. The SDF in Xiong and Yan (2009) follows a process dmt /mt which decomposes into a fundamental component and a sentiment component. the loading factor on ft has a value of 1 for maturity τ = 0. based on the magnitude of the error in the representative investor’s estimate of lt relative to the statistically optimal estimate. ˆ The loading factor on the estimate lt turns out to be hump shaped. achieving a maximum impact at intermediate maturities. and plays a central role in the results of Xiong and Yan (2009). The stochastic term ˆR in the sentiment component is the Brownian motion dZf underlying ˆ the objective estimate lt . The yields of bonds ˆ l with very long maturities have low exposure to lt . a feature emphasized in Shefrin and Statman (1994). which monotonically decreases to zero with maturity length. As in Shefrin (2008a). as lt reverts to ¯ in the long-run. The fundamental component is −rt dt − σI dZI . then the sentiment component will exert upward pressure on dmt /mt . Because the instantaneous interest rate is given by the expected value of m. thereby counteracting the tendency of state prices to fall with time. An important contribution of Xiong and Yan (2009) is to analyze the relationship between bond maturity and return volatility. If ˆR the representative investor’s estimate is too low and dZf > 0. Wealth shifts from trading based on differential beliefs lead to excess volatility in bond prices. They show that in a homogeneous belief economy. Xiong and Yan (2009) explain that the .

One way to think about the expectations hypothesis is through the following equation developed in Campbell and Shiller (1991). the weighting on the ft facˆ tor dominates the weighting on the lt in the homogeneous beliefs case. thereby driving up its yield. the price can be expected to rise. Essentially. Therefore. However. relative to the homogeneous beliefs case. Therefore. As discussed above. Yt+1 (n − 1) − Yt (n) = αn + βn Yt (n) − Yt (1) . Suppose that the representative investor’s forecast of future short rates is biased upward. Therefore over time. there will be an advantage to investing in the longbond. market aggregation implies that wealth shifts can lead to time varying risk aversion.13). In this case. Because the instantaneous short-term interest rate is determined by the production sector in Xiong and Yan (2009). in practice. ˆ whereas wealth shift dynamics can induce sufficient volatility in lt to inject the hump. when the yield spread is positive. yields on current long-term bonds rise. It may well be that the failure of the expectations hypothesis is driven by both features. Using simulation .14) When the expectations hypothesis holds βn = 1. the market will excessively discount long bonds. However.12) and (5. thereby driving down future yields. unless its price falls sufficiently. Cochrane and Piazzesi (2005) indicate that a tent-shaped function of three forward rates corresponding to maturities of one. long-term yields tend to rise. in contradiction of the expectations hypothesis. not fall. lies with the factor weightings in Equations (5. the spread will also widen. This is because when Yt (n) − Yt (1) > 0. three and five years appears to forecast holding period bond returns of all maturities better than the maturity specific forward spreads. n−1 (5. thereby unduly driving down their prices. the long-term bond is underpriced relative to fundamentals. The asset pricing literature also attributes the failure of the expectations hypothesis to time varying risk aversion. A behavioral explanation for this finding is as follows. the empirical evidence indicates that βn is close to zero and positive for 2-month maturities and then declines to about −4 for 10year maturities.108 Behavioralizing Asset Pricing Theory reason for the different shape in the heterogeneous beliefs case.

5.6 Pricing Dynamics

109

analysis, Xiong and Yan (2009) use their model to advance an interesting explanation for the tent-shaped factor. They note that yields to intermediate maturities display the greatest sensitivity to beliefs about lt . In their parametric specification, the greatest sensitivity occurs at a maturity length of three years, making the three-year forward rate more sensitive than the one year or five year. When wealth shifts to ˆ the bullish investor, the representative investor’s estimate lt increases, thereby increasing the impact of the three-year forward rate relative to the other two. As a result, the value of the tent-factor increases, because of the disproportionate weight on the three-year forward rate. A higher value for the tent-shaped factor predicts a higher bond return. The higher bond return occurs because, as described above, the bullish investor’s beliefs unduly drives down the current price, which mean reverts in the future. 5.6.2 Behavioral Option Pricing

Shefrin (2008a) shows that the linear properties discussed in connection with bond prices also apply to option prices. In particular, Shefrin (2008a) presents a specific example involving heterogenous beliefs in which equilibrium option prices can be represented as wealth-weighted convex combinations of Black–Scholes prices that would apply in homogeneous belief economies. One of the major empirical features associated with option pricing is the option “smirk” or “smile,” which refers to the shape of the implied volatility function (IVF). Given a set of options which vary only by exercise price, the IVF maps the exercise price to the implied volatility derived from applying the Black–Scholes option formula to the market price. If the market treated the return of the underlying security as following a Brownian motion, then the IVF would be flat instead of decreasing (the smirk) or U-shaped (the smile). Notably, the option smirk in the IVF occurs in conjunction with an oscillating SDF, where there is disagreement about investors about the volatility of the asset underlying the options. Heterogeneity typically prevents equilibrium option prices from satisfying Black–Scholes because heterogeneity leads the risk neutral

110

Behavioralizing Asset Pricing Theory

density to be non-normal. In the continuous time example presented in Shefrin (2008a), the interest rate and volatility that serve as inputs into the Black–Scholes formula are correct. When equilibrium option prices do not satisfy Black–Scholes, it is not because the input values are necessarily incorrect. Black–Scholes fails because the risk-neutral density is non-normal, even though the objective return distribution for the asset underlying the option is normal. Black–Scholes intuition has developed around partial equilibrium models. In these models, the risk-neutral pdf is derived from an objective normal pdf as a change of measure. In this case, the risk-neutral pdf is itself normal with the same volatility as the objective pdf. In the behavioral framework, the risk-neutral pdf is obtained, not from the objective pdf, but from the representative investor’s pdf. In this regard, heterogeneity typically leads the representative investor’s pdf to be non-normal, even when the objective density is normal. Smile effects occur in the implied volatility function (IVF) because equilibrium option prices do not obey Black–Scholes. In the continuous time example described in Shefrin (2008a), all investors have common knowledge of the objective volatility, which makes it possible to use that volatility to compute the Black–Scholes value associated with any option. By way of contrast, the IVF backs out volatilities implied by the risk-neutral pdf based on the representative investor’s pdf. In the example portrayed in Shefrin (2008a), the representative investor’s pdf features greater volatility than the common knowledge volatility, leading the IVF function to lie above its common knowledge counterpart. The analysis easily extends to the case when investors differ in their views about volatility, although in this case there is a question as to which volatility to use in the Black–Scholes formula. One possibility is to use the volatility associated with the representative investor’s pdf. However, the analysis is little changed if the objective volatility were used, or a convex combination of the investors’ subjective volatilities. As I mentioned above, the change of measure used to derive riskneutral densities from objective densities is based on a behavioral SDF rather than a neoclassical SDF. Therefore, the behavioral risk-neutral density need not feature the same volatility as the objective density, as is the case in neoclassical models.

5.6 Pricing Dynamics

111

Much of the empirical work estimating the SDF and market risk aversion is based on index option prices. In this respect, option smiles and smirks and the oscillating shape of the empirical SDF represent different sides of the same coin. To the extent that the sentiment component of the SDF is germane, the same will be true of index option prices. In this regard, Han (2008) finds that changes in sentiment explain time variation in the slope of index option smile which traditional factors cannot. He reports that when market sentiment becomes more bearish, the index IVF smile is steeper and the risk-neutral skewness of monthly index returns is more negative . Constantinides et al. (2009) examine possible mispricing of SPX options for the periods 1988–1995 and 1997–2002. Their analysis contains some novel elements, specifically the inclusion of transaction costs. Notably, they ask whether market prices were such that in any month of their study, an investor with objectively correct beliefs could have increased his expected utility by engaging in a zero-net-cost trade. They find that there were many months when this would have been possible, thereby leading them to conclude that SPX options were mispriced over time. In particular, they note that mispricing was pronounced for out-of-the-money call options. Interestingly, they suggest that before the crash of 1987, option traders might have navely used the Black– Scholes formula when it did not apply. Doing so corresponds to the use of the formula as a heuristic. The analysis in Constantinides et al. (2009) implies that the pricing of SPX options cannot be explained in terms of a neoclassical representative investor holding correct beliefs. This analysis neither makes explicit use of sentiment, nor takes into account how sentiment injects oscillation into the shape of the SDF. Nevertheless, the authors implicitly acknowledge sentiment-based explanations for their results.

and should prove especially useful for integrating behavioral corporate issues into a general asset pricing model. (2003). In Shefrin (2001). Heaton (2002) explored the impact of managerial overconfidence on spending and finance decisions.6 Behavioralizing Corporate Finance Behavioral corporate finance can be thought of as the intersection of finance and management Gaston-Breton and Lejarraga (2009). Gervais et al. See Statman and Tyebjee (1985). The groundwork for behavioral corporate finance was laid out in a series of papers Meir Statman and a series of coauthors wrote about project decisions. (2003) pointed out that overconfidence can exert a positive impact on the value of the firm by mitigating agency conflicts.” Other early contributions included Heaton (2002) and Gervais et al. a point also developed in Goel and Thakor (2008) and Hackbarth (2009). 112 .1 1 The framework developed in Hackbarth (2009) is contingent claims based. I suggested that enough was known about the behavioral aspects of corporate finance to identify a subfield one could label “behavioral corporate finance. Statman and Caldwell (1987). and Statman and Sepe (1989).

” The text offers a series of cases to make the material concrete for students. I suggest that some of the case discussions provide insight into a list of open questions that conclude Baker et al.6. suggesting that the loss aversion property in prospect theory offers a potential explanation. (2007b) asks why managers do not pursue the tax benefits of debt more aggressively. I call criterion functions such as Equations (2. (2007b) discussed in Section 2 above.1 Open Questions Raised by Baker et al. Theoretically. Baker et al. in order to provide a behavioral supplement to the traditional material taught in corporate finance courses. 113 6. I intended my text for classroom instruction. In many ways.1) and (2. were writing their survey. In Shefrin (2005) I present a case about the pharmaceutical firm Merck. There is no reason why irrational managers should not balance perceived short-term gains against perceived long-term value.4) “behavioral adjusted present value. I take as my starting point the behavioral corporate framework developed in Baker et al.1 Open Questions Raised by Baker et al. the textbook approach parallels the approach in Baker et al. Debt Puzzle The first open question in Baker et al. Graham has indicated that he concurs. suggesting that managers do not want to get too close to the edge of the cliff. (2007b) conclude with a series of open questions about the future direction of behavioral corporate finance. Irrational Managers and Inefficient Prices The second open question asks whether a model that incorporates both irrational managers and inefficient prices would lead to new insights. an issue raised by Graham (2000). In this section. That framework goes a long way toward behavioralizing corporate finance.4) in . (2007b).1) and (2. as bankruptcy losses loom much larger than the gains associated with tax shield benefits. it is straightforward to combine Equations (2. (2007b). I wrote my behavioral corporate textbook (Shefrin. In this regard. 2005) at the same time that Baker et al. In private correspondence.

the executives at AOL did use overvalued equity to purchase TimeWarner. In line with the Shleifer–Vishny framework. Notably. 1998). I devote considerable space to this issue. ) with (1 + γ)f (K. ) − K + (e(δ(·) − γf (K. One way to do this is to modify Equation (2. psychological pitfalls might well lead managers to make decisions that are not in the best interests of shareholders.4).1). (6. one of the largest in history.114 Behavioralizing Corporate Finance order to capture both phenomena. Shefrin (2005) emphasizes that behavioral phenomena and agency phenomena are both central to corporate finance. These substitutions lead to the criterion function: λ[(1 + γ)f (K. I suggest that the financial features of this case reflect both irrational managers and inefficient prices. a key challenge is to differentiate them clearly. the usual incentive remedies for traditional agency conflicts may not apply. in contrast to the general evidence presented by Malmendier and Tate (2008).. this represents an agency conflict. )]. in Shefrin (2005) I present a case about the acquisition of Time-Warner by AOL. ). a concept that is already central to corporate finance. replacing output f (K. First. Time-Warner’s CEO Gerald Levin was not skeptical about mispricing. There are two main issues involving the relationship between agency phenomena and behavioral phenomena. However. mispricing is now understood in the sense of “perceived” mispricing. From a neoclassical perspective. Debiasing One of the most important open questions involves debiasing. because the conflict is not between rational principals and rational agents. and changing the base from which the mispricing term δ(·) is measured from 0 to −γf (K. . )))] + (1 − λ)[δ(·) − γf (K. To be sure. including a section focusing on the role psychological phenomena play in business process.1) As to how the combination of irrational managers and inefficient prices might be germane. However. how corporations can learn to reduce susceptibility to managerial pitfalls? In Shefrin (2005). Indeed there is already a literature in addressing behavioral bias within organizations (Heath et al. ) from Equation (2.

6.1 Open Questions Raised by Baker et al.

115

so behavioral issues do not come to be regarded as rational agency issues. Second, there is a whole behavioral dimension to the treatment of agency phenomena. In Shefrin (2005) I devote a section to discussing interactions between nonrational principals and nonrational agents. Shefrin (2005) is organized to identify how psychological pitfalls create gaps between recommended courses of action in corporate finance textbooks and what corporate managers actually do. Topics include the spectrum from valuation analysis to real options. The literature on these gaps continues to grow. For example, Tiwana et al. (2007) document a series of cases in which managers make imperfect use of real option analysis. In particular, managers are apt to downplay real option analysis for high NPV projects relative to low NPV projects, thereby underestimating the value of the latter. Themes in the Remainder of the Section In the body of this section, I focus on three main issues pertaining to the behavioralization of corporate finance. All relate to delving more deeply into understanding the impact of managers’ personality characteristics on their decisions. The first is based on the analysis of surveys to identify the impact of managerial optimism decisions about debt ratios, debt maturity, dividend policy, and acquisition frequency. The second is a study about the impact of social networks to which CEOs belong, especially when it comes to vulnerability to groupthink. The third is evidence about the psychological traits of entrepreneurs, which builds on the discussion in Baker et al. (2007b). The three themes upon which I focus are not exhaustive. There have been numerous new papers in behavioral corporate finance that build on earlier ideas. Some examples are as follows: Degeorge et al. (2007) document that managers tend to rely on bookbuilding to go public, which is less cost effective than auction techniques. Huang and Ritter (2009) analyze the empirical issues associated with achieving target capital structures. Spalt (2008) studies the degree to which firms use option-based compensation to exploit managerial optimism. Peyer and Vermaelen (2009) establish that the buyback puzzle has not disappeared. Campbell et al. (2009) use turnover rates to test and confirm

116

Behavioralizing Corporate Finance

the predictions in Goel and Thakor (2008) that there is an optimal degree of overconfidence to balance agency conflicts.

6.2
6.2.1

Managerial Psychology and Decisions
Results Based on FEI Survey

Ben-David et al. (2007) provide interesting insights into the connection between overconfidence and managerial decisions. Their study makes use of a survey Duke University andCFO magazine jointly run of the members of Financial Executives International (FEI), an association of top financial executives. The survey features a unique panel of nearly 7,000 observations of managers’ probability distributions regarding the stock market. Survey results show the financial executives to be miscalibrated: Over a six-year period, realized market returns are within the executives’ 80 percent confidence intervals only 38 percent of the time. Given that raising external capital is costly, firms whose projects have high NPV can do their shareholders a favor by keeping their dividend payouts low, and using the cash to fund projects instead. After all, high NPV means that the company is able to generate higher expected returns through its projects than investors can earn if instead paid the cash. Overconfidence leads executives to underestimate the risk attached to project cash flows, resulting in overestimates of projects’ NPV. As a result, the greater the degree of overconfidence, the more inclined are executives to adopt projects, and undertake acquisitions. Overconfidence can also impact capital structure. Think about unrealistically optimistic, overconfident managers seeking to balance the tax shield benefits of debt against the expected costs of financial distress, especially bankruptcy. They will be inclined to underestimate the likelihood of going into bankruptcy, and therefore to underestimate the expected cost bankruptcy. Therefore, overconfident managers will be prone to take on more debt than is appropriate. The general evidence from the Ben-David et al. (2007) is that the greater the degree of unrealistic optimism and overconfidence on the part of executives, the more prone they are to take on excessive debt. In the Duke/FEI survey, the average debt ratio was 23 percent. Imagine

6.2 Managerial Psychology and Decisions

117

dividing executives into ten groups according to their overconfidence. Suppose that for each group, we compute the average debt ratio of the company that the executive works for. Ben-David et al. (2007) report that as we migrate from the group having the lowest overconfidence to the group having the highest overconfidence, the debt ratio increases by 0.5 percent every time we move up to a new group. In addition, overconfident executives tend to take out debt with longer maturities. The average maturity for the survey firms was 3.7 years. However, the group with the highest overconfidence rating took out longer loans than the group with the lowest overconfidence rating, by about one year. The longer maturity leaves a company with less financial flexibility, and more interest rate risk. 6.2.2 Results Based on Personality Tests

Graham et al. (2009) may well be the first study which measures attitudes of senior management directly through personality tests and then relates the results to firm level policies such as the usage of debt and mergers and acquisitions. In 2006, the authors surveyed both CEOs and CFOs. Most of the executives surveyed were identified as readers of Chief Executive, CFO magazine, and for CFOs the publications CFO Europe and CFO Asia magazines. The responses generated a database with 1,180 CEOs with 1,017 CEOs who work for firms headquartered in the United States. For CFOs, there were correspondingly 549 responses. In their survey, Graham et al. (2009) elicit responses to indicate degree of risk aversion, dispositional optimism, time preference, and aversion to a sure loss. To assess risk aversion, the authors use a procedure developed by Barsky et al. (1997). To assess optimism, they administer a standard psychometric test known as Scheier and Carvers Life Orientation Test Revised or LOT-R test. To measure time preferences, they assess time predilection for gains and losses. To gauge aversion to a sure loss, they present managers with a gamble that, if taken, indicates that the survey respondent is averse to a sure loss. In addition, Graham et al. (2009) elicit other measures that be related, such as height as a measure of confidence, age, tenure, career track, education, and prestige of the executives undergraduate college.

In terms of relying on incentive-based compensation. as managers with formal backgrounds in finance and accounting will be more familiar with the advantages of debt. Interpreting height as a measure of confidence. (2009) hypothesize that firms set compensation policies to hire managers who. who register subpar performance as a psychological loss. confident. will perpetuate corporate policies. and rate of time preference. perhaps because they keep bidding until they win. as are CEOs who are averse to a sure loss. and option grants relative to a large salary. are prone to engage in risky acquisition activity. (2009) suggest a matching story whereby young. optimists. Graham et al. (2009) find that risk-tolerant CEOs are more likely to obtain proportionately larger remuneration through incentives such as stock. three personal characteristics appear to be significant: risk-taking. Moreover. If male gender is positively correlated with overconfidence. males are more likely to be compensated through incentives. . that some firms use stock option pay for lower-level employees to take advantage of biased probability judgments or weightings. This might be a reflection of the availability heuristic.118 Behavioralizing Corporate Finance In respect to debt. then overconfidence appears to lead to more usage of short-term debt. and executives from private companies are more likely to use a higher proportion of short term debt than others. See Landier and Thesmar (2009) whose theory suggests that optimists are more likely to take on short-term debt. Spalt (2008) reports a complementary finding. In this regard. Interestingly. In Shefrin (2005) I suggest that the managers of poorly performing firms. such as capital structure. the results in Graham et al. as suggested by Barber and Odean (2000). risk-tolerant CEOs are more likely to work for high growth companies and companies with high anticipated growth rates. over time. as are executives who are more patient. Graham et al. males. Not surprisingly. (2009) find that CEOs who are more risk tolerant are also more likely to make acquisitions. bonus payments. (2009) report that CEOs with a background in finance and accounting tend to take on more debt than CEO with other backgrounds. Graham et al. gender. Graham et al. (2009) point out that managers who are averse to a sure loss are more prone to engage in multiple acquisitions. There may be several explanations for the second finding. Graham et al.

corporations.S. In this respect. non-U. and to have attended more prestigious universities. CEOs. See Cai and Sevilir (2009) and Ishii and Xuan (2009) who study the role of social networks involving directors and/or managers in M&A. between management and potential directors influence director selection and subsequent firm performance. more than 80 percent of CEOs’ responses to the optimism questions average about 3. In addition. (2009) involves the benchmarking of the dispositional optimism of U. identifying directors who. not just business prospects. However. CFOs confirm this finding. with only 66 percent of CFOs being classified as very optimistic.S. Strikingly.2 Their study uses panel data for firms in the S&P 1500. CFO suggests that CEOs are more optimistic about all aspects of life. CFOs are more averse to sure losses and more patient than CEOs. CEOs and CFOs are less patient than their U. so that CEOs are significantly more optimistic than the lay population. were employed by the same firm as the CEO in the 2 Studies focusing on social networking are gaining interest. Graham et al. CEOs are also more optimistic than CFOs.3 Social Networks 119 An intriguing finding of Graham et al.S. Interestingly.S. have longer tenure. CEOs against others. 6. CEOs are likely to be older. thereby focusing on social ties in large U. charities.S. On a 0–4 range. . with the exception of the current firm. Finally. and other non-professional organizations as the CEO. Fracassi and Tate (2009) identify directors who belong (or belonged) to the same golf clubs. Foreign CEOs also have a higher aversion to sure losses than do U. There are other ways CEOs differ from CFOs. (2009) report that both CEOs and CFOs from the United States are more risk tolerant than the CEOs and CFOs of 25 companies which are not located in the United States. when asked to rate CEO optimism relative to their own. counterparts. CEOs are less likely to have MBA degrees than CFOs.3 Social Networks Fracassi and Tate (2009) test whether connections.5. in the sense of social networks. They consider employees’ histories.6. the average response for the general population is about 2.

but are not unbiased monitors.S. I discuss a series of psychological impediments to effective governance processes. Such directors are more likely to buy company stock at the same time as the . B´nabou. I point out that when it comes to heterogeneous board composition. some boards might look good on paper. this is no guarantee that directors with network connections to the CEO qualify as independent directors. In Shefrin (2005). and trust in. a high entrenchment index. based on similarities in backgrounds and experiences. I discuss some of the specific issues associated with the negative relationship. A key impediment is “groupthink” (Janis. Social Connectivity and Firm Value Fracassi and Tate (2009) find that the relationship between social connectivity and firm value. the effect is strongest when shareholder rights are weak. increase the potential for groupthink. the CEO. close ties between the CEO and the board. Moreover. resulting in the amplification of individual biases. In the remainder of this section. Groupthink is a form of collective confire mation bias for groups.120 Behavioralizing Corporate Finance past. whereby there is insufficient dissent within the group. and in recent years have strengthened independence requirements for key board committees. exchanges require the majority of directors in listed firms to be independent. Consider alternative measures of CEO power such as being both board chair and president. Therefore. measured using Tobin’s Q. Fracassi and Tate (2009) point out that directors might be more willing to give the benefit of the doubt to management when they have a closer relationship with. long tenure. is negative. yet be very vulnerable to groupthink. In Shefrin (2005). and being much more highly paid than the next best paid executive. Finally. Fracassi and Tate (2009) report that CEOs who are powerful are prone to add new directors having existing network ties to the CEO. Alas. 1972. 2008). they identify directors who attended the same educational institutions as the CEO. In this regard. Groupthink can result when group members are too similar to one another. Indeed U. all sharing similar perspectives.

Especially. there is some evidence that the presence of directors closely connected to the CEO is correlated with weaker monitoring. (2003). acquisition-based value destruction might well reflect a monitoring failure or an advisory failure on the part of the board. in which case they are complicit in the associated value destruction. For example. In addition. such poor decisions lead to lower overall market valuations. if the CEO is a net buyer. Fracassi and Tate (2009) state that the merger bids of these firms destroy $407 million of shareholder value on average. especially when shareholder rights are weak when it comes to substituting for board monitoring. relative to other outside directors. connected directors might be seeking to expedite the CEO’s agenda. Fracassi and Tate (2009) find that the average cumulative abnormal return for the threeday window surrounding merger announcements is lower for firms with a higher percentage of connected directors. . then over the fiscal year these directors are more likely to be net buyers. More importantly. Alternatively. Plausibly. For example.6. Fracassi and Tate (2009) indicate that their analysis does not provide much insight into the motivation of connected independent directors. Alternatively. interesting is the fact that firms with boards closely connected to the CEO make more frequent acquisitions. Not surprisingly. firms with closely connected directors are less likely to initiate earnings restatements which are internally prompted. Specifically. Directors connected to the CEO might be unwilling to oppose value-destroying policies that provide private benefits to the CEO. $293 million more than the bids of other firms. however. directors closely connected to the CEO through social networks are more likely to purchase stock within five days of a CEO stock purchase. Major acquisitions might be initiated by the CEO. they require board approval. This feature complements the discussion in the prior section about the characteristics of CEOs of firms who are frequent acquirers. They identify several possibilities.3 Social Networks 121 CEO. They also find that the average value created by the deals (for acquiring shareholders) is negative. Specifically. such directors might withhold information in their possession that does not support the CEO’s proposal. and that the value destruction is concentrated in firms with weak shareholder rights as measured by Gompers et al.

In this respect. or cognitive biases. 6. the discussion about the psychological profile of entrepreneurs will serve as a vehicle for both goals. entrepreneurs appear to hold poorly diversified portfolios. Heaton and Lucas. Below I discuss how psychologists measure control.3 The first goal is to broaden the psychological framework used to characterize managers. In this respect. (2007b) both indicate that among the many psychological phenomena. Baker et al.122 Behavioralizing Corporate Finance connected directors might be motivated by an interest in preserving social ties with the CEO. connected directors might simply share the CEO’s perspective. 2000). average return to all private equity is actually similar to that of the public market equity index. Weinstein (1980) and Flynn et al. behavioral corporate finance mainly focuses on excessive optimism and overconfidence. (1994) indicate that control is a determinant of both excessive optimism and overconfidence. and relate the discussion to entrepreneurs. Moskowitz and Vissing-Jorgensen (2002) establish that entrepreneurs earn low riskadjusted returns. Hamilton (2000) establishes that entrepreneurs accept lower median lifetime earnings than similarly skilled wage-earners. In light of the 3I 4 The draw the material for this section from Shefrin (2010).4 Entrepreneurs This section has two goals. Survival rates of private firms are only around 34 percent over the first 10 years of the firms life. 2001. but identifies factors that explain what makes people excessively optimistic and overconfident. Moskowitz and Vissing-Jorgensen. Instead they concentrate their wealth in their own private business (Gentry and Hubbard. This is because entrepreneurs appear to accept inferior returns. Shefrin (2005) and Baker et al. I note that the behavioral decision literature does not treat these concepts as primitives. 2002. (2007b) devote several paragraphs to entrepreneurs. .4 Moreover. The second goal is to identify the psychological profile of entrepreneurs with the goal of understanding why they tolerate inferior returns. by dint of having the same background and moving in the same social circles. Finally. A key factor is control. However. the risk is higher.

1985). The former views optimism as a positive personality trait associated with positive generalized expectations of the future (Scheier and Carver. Puri and Robinson’s analysis of the data in the Survey of Consumer Finances (SCF) leads them to a series of conclusions. dispositional optimism can also involve optimistic bias. entrepreneurs are almost three times more likely to indicate that they never intend to retire. First. The latter views optimism as a negative. being more risk-loving than the general population does not explain why entrepreneurs earn suboptimal risk-adjusted returns.4 Entrepreneurs 123 evidence. In the main. They suggest that entrepreneurs might derive substantial non-pecuniary benefits from self-employment. Notably. In this regard. Second. Puri and Robinson (2008) provide some insight into why people would choose to become entrepreneurs. (2009). good health practices. and strong family ties. Bitler et al. Of course. Unless entrepreneurs are actually risk seeking. Based on their prior work (Puri and Robinson. This is not to say that the net impact of dispositional optimism is negative. It is straightforward to see how optimistic bias can lead to inferior returns. They ask whether entrepreneurs are risk-takers. and therefore produce inferior returns as well. people who do not plan to retire work about 3 percent longer per week. 1980). Third. these findings are supported by Wadhwa et al. In their . (2004) investigate the structuring of incentives for entrepreneurs and ask why people would choose to become entrepreneurs. these traits are separable in that the correlation between risk tolerance and optimism is low. they hypothesize that entrepreneurs might be optimistic about their entrepreneurial prospects. domain-specific bias in expectations (Weinstein. they find that entrepreneurs are more risk-loving and optimistic than the rest of the population.6. Consider how the configuration of entrepreneurs’ characteristics relates to the suboptimality of the returns which they earn. Moreover. 2007). Bitler et al. (2004) find that effort increases firm performance. entrepreneurs tend to have long planning horizons. optimism might provide part of the explanation. Puri and Robinson focus on dispositional optimism as opposed to unrealistic optimism. In respect to planning. They point out that the psychology literature distinguishes between dispositional optimism and optimistic bias.

one pertaining to well-being (life satisfaction) and the other pertaining to degree of affect (positive. I hypothesize that the desire for control is a psychological need which might be higher for entrepreneurs than non-entrepreneurs. Notably. Shefrin (2010) extends their analysis by surveying entrepreneurs for dispositional optimism and comparing their responses with the responses of non-entrepreneurs. This suggests that entrepreneurs might be more socially focused than others. They also report being more positive and more satisfied with their lives. Puri and Robinson document that relative to non-entrepreneurs. Taken together. then it seems plausible that entrepreneurs would report being happier than their non-entrepreneurial counterparts. the control that derives from starting a new business and managing that business. After all. and feel that they are more sensitive to social cues. but infer it from self-reported estimates of own life expectancy. the most significant difference between entrepreneurs and non-entrepreneurs is the desire for control: entrepreneurs place much more emphasis on control.. The major findings are as follows: entrepreneurs exhibit more dispositional optimism than their nonentrepreneurial counterparts. 2009). Puri and Robinson focus on dispositional optimism. In particular. I hypothesize that this is the case. If entrepreneurs derive significant non-pecuniary benefits from entrepreneurial activity.124 Behavioralizing Corporate Finance analysis. entrepreneurs are more likely to be married and have larger families (see also Wadhwa et al. They do not measure optimism bias directly. I test whether entrepreneurs are less anxious in social settings than non-entrepreneurs. these findings . it seems far more plausible that non-pecuniary benefits lie at the heart of the issue. What might those nonpecuniary benefits be? I suggest that a major non-pecuniary benefit for entrepreneurs is the exercise of control over their working environment. neutral. I test whether this is the case using two instruments. or negative). In this regard. Entrepreneurs appear to be more comfortable in social situations. While the combination of risk-seeking choices and optimism possibly explains why entrepreneurs earn low risk-adjusted returns. entrepreneurs work longer hours and plan to retire later than their non-entrepreneur counterparts. and more aware of social cues and their own behavior.

The surveys relate to instruments for measuring dispositional optimism (Scheier and Carver. The survey instrument consists of twenty questions. life satisfaction (Diener et al.4 Entrepreneurs 125 suggest that the non-pecuniary benefits that entrepreneurs experience are substantial. 1985). 6. This statement reflects a view in the literature on motivation about the importance of nonmonetary rewards relative to monetary rewards. 1984). 1983).” For 15 of the questions. It seems plausible to suggest that people who fit this description exert more control over their working environment than others.4.” The range of possible responses varies from 1 to 7 where 1 means “The statement does not apply to me at all. 1988). Two representative questions are: “I prefer a job where I have a lot of control over what I do and when I do it. while for 5 of the questions. Lennox and Wolfe. and that part of the appeal of being an entrepreneur is achieving this degree of control. The surveys were administered to two different groups: (1) a group of entrepreneurs and (2) a group of MBA and undergraduate business students (nonentrepreneurs).4. 1979).” and “I would prefer to be a leader than a follower. desirability of control (Burger and Cooper. 6. Put somewhat differently.6. and affect (Watson and Clark.. Burger and Cooper (1979) introduce a survey instrument (DC) which is designed to measure the desirability of control. and are full-time self-employed. 1974. 1985).” and 7 means “The statement always applies to me.1 Data The data for this study involve responses to a series of psychological surveys. 7 is associated with the strongest desire for control. self-monitoring behavior (Snyder.2 Desire for Control Puri and Robinson classify respondents to the SCF as entrepreneurs if they own some or all of at least one privately owned business. social anxiousness (Leary. 1 is associated . entrepreneurs possess a strong need for control which leads them to choose a career in which they seek to meet that need.

126

Behavioralizing Corporate Finance

with the strongest desire for control. For sake of consistency, the score for each of the 5 questions is replaced by 7-score. The results suggest that desire for control is a major non-pecuniary benefit associated with being an entrepreneur. The average DC score per question is 5.4 for entrepreneurs and 4.6 for non-entrepreneurs. To interpret these results, note that 4 means “I am unsure about whether or not the statement applies to me or it applies to me about half the time,” 5 means “The statement applies more often than not,” and 6 means “The statement usually applies to me.” 6.4.3 Optimism

Using their life expectancy proxy for optimism, Puri and Robinson find that entrepreneurs exhibit more dispositional optimism than nonentrepreneurs. I augment their analysis by using an instrument to measure dispositional optimism (Scheier and Carver, 1985). The optimism instrument (OP) consists of eight questions whose responses fall on a five-point scale where 1 means “strongly disagree” and 5 means “strongly agree.” Two of the questions are: “I’m always optimistic about my future,” and “I hardly ever expect things to go my way.” Adjusting for signs, the mean OP-score for entrepreneurs was 3.5 and for non-entrepreneurs it was 3.2. Notably, both groups feature OP-scores between Neutral and Agree, and are therefore optimistic. Although entrepreneurs appear to be more optimistic than nonentrepreneurs, the difference in scores is not statistically significant for the particular sample in my study. However, the results are quite similar to the larger sample described in Puri and Robinson (2008), where the differences are statistically significant. 6.4.4 Social Anxiousness and Self-monitoring

Consider psychological attributes which relate to the finding that entrepreneurs marry at a higher rate than non-entrepreneurs and have more children. This finding suggests that entrepreneurs like people. This subsection considers two psychological instruments that relate to interpersonal relationships. The first stems from the work

6.4 Entrepreneurs

127

of Leary (1983), who studied social anxiousness, the degree to which people are uncomfortable in social situations. The second relates to self-monitoring, a concept studied by Snyder (1974) and Lennox and Wolfe (1984). The survey instrument used to study social anxiousness (SA) features fifteen questions. Two of the questions are: “I often feel nervous even in casual get-togethers,” and “I get nervous when I speak to someone in a position of authority.” Responses are on a 5-point scale where 1 means “Not at all characteristic,” and 5 means “Extremely characteristic.” A lower SA score signals less social anxiety. Entrepreneurs have a mean SA score of 2.0, whereas non-entrepreneurs’ mean SA score is 2.9. A score of 2 means “Slightly characteristic” and a score of 3 means “Moderately characteristic.” Self-monitoring involves being sensitive to social cues, and being able to adapt to those cues. The self-monitoring survey instrument is SM. It features questions such as the following: “I am often able to read people’s true emotions correctly through their eyes.” “In social situations, I have the ability to alter my behavior if I feel something else is called for.” The range of possible responses comprises a 6-point scale from “Always false” to “Always true.” The mean score for entrepreneurs is 4.4 and for non-entrepreneurs is 3.9. Here 3 means “Somewhat false, but with exception,” 4 means “Somewhat true, but with exception,” and 5 means “Generally true.” 6.4.5 Affect and Well-Being

Are entrepreneurs happier than non-entrepreneurs? What makes this question especially interesting is that entrepreneurs work more than non-entrepreneurs and earn lower risk-adjusted returns on their investments. To investigate the question, I use survey instruments based on the work of Watson and Clark (1988), who study affect, and Diener et al. (1985), who study life satisfaction or well-being. I refer to the associated instruments as AF and WB, respectively. The affect survey asks respondents to indicate how frequently they experience a variety of emotional states such as “interested,”

128

Behavioralizing Corporate Finance

“distressed,” and “excited.” Responses are on a 5-point scale, in which 1 means “very slightly or not at all,” and 5 means “extremely.” Some states connote negative affect and some connote positive affect. An AF score of zero connotes neutral overall affect. The mean AF-score for entrepreneurs was 1.2 and the mean AF-score for non-entrepreneurs was 0.8. The life satisfaction survey asks respondents twenty-five true/false questions such as “I always seem to have something pleasant to look forward to,” and “Often I get irritated at little annoyances.” Of the 25 questions, 11 relate to feelings of positive life satisfaction and 14 to negative feelings. Relative to an adjusted neutral score of 0, the mean WB-score for entrepreneurs was −1.0 and for non-entrepreneurs was −1.2. Entrepreneurs score higher on the WB-survey. However, both groups’ mean scores are negative responses. Overall, the evidence indicates that entrepreneurs are happier than non-entrepreneurs.

6.4.6

Risk Preferences

Puri and Robinson (2008) find that entrepreneurs are more risk-loving and optimistic than the rest of the population, although the correlation between risk tolerance and optimism is low. The analysis suggests that desire for control is a major motivator for why people choose to be entrepreneurs, stronger than optimism, and perhaps risk attitude as well. In terms of risk, it is possible that people choose to be entrepreneurs because they have a preference for lottery-type returns. To test for this possibility, I ran a supplementary survey, based on the concepts in SP/A theory (see Section 3) which posed the following two questions to entrepreneurs: 1. On a scale of 1 to 7 where 1 is unimportant and 7 is extremely important, how important to you is upside potential as a reason being an entrepreneur? 2. On a scale of 1 to 7 where 1 is unimportant and 7 is extremely important, how important to you is being in control of your work world as a reason being an entrepreneur?

P. . and A. Many indicate that they would accept lottery-type returns if the downside is small.8 and to the second was 6.6. The supplementary survey also posed qualitative questions designed to elicit features emphasized by SP/A theory. between the two. The responses indicate that entrepreneurs feel that fear is a weak emotion for them. However. S. both factors are important. hope is a moderately strong emotion. These respectively correspond to the three main variables in the theory. and whether entrepreneurs establish specific aspirational goals. and aspirational goals are strong. Clearly.4 Entrepreneurs 129 The mean response to the first question was 5. such as the strength of emotions such as fear and hope. but are more reluctant if the downside is not small. if doing so significantly diminishes the probability of achieving their specific aspirations.3. control is more important.

Statman (2009) and Shefrin and Statman (2009) describe how their framework applies to a series of issues preceding and including the financial crisis. financial economists paid little attention to financial market regulation in general and its behavioral aspects in particular. To the best of my knowledge. asset pricing. 1993b). and the Sarbanes–Oxley Act.7 Behavioralizing the Approach to Financial Market Regulation The behavioral aspect of financial market regulation is a particularly rich area of study. corporate finance. 130 . in that it combines behavioral aspects of portfolio selection. The literature then lay dormant until Hirshleifer (2008b). and addressed some of the same issues. the first behavioral treatment of financial market regulation is found in Shefrin and Statman (1992. In this section. Regulation FD. I review the basic behavioral approach to financial market regulation. the Global Settlement. with reference to both historical and recent financial crises. The series includes credit card and bank regulations. until the global financial crisis erupted in 2008. Yet. derivatives. A major reason why is its complexity. the evolution of insider trading regulations. trading halts. and political economy. who was unaware of our earlier work.

1. In this respect. and investors. managers can avoid projects with negative net present value.1 Regulations are the outcome of a dynamic tug-of-war between parties with differing views about efficiency and fairness. However.1 Dynamic Tug-of-War Fairness and Efficiency The regulation of financial markets in the United States can only be understood in reference to the continuous debate that has shaped it. By observing prices. (1991). Informational efficiency is achieved when all investors share objective beliefs and information so that competitive prices accurately reflect that information. 1993b).1 7. There are two notions of efficiency in financial economics. It also ensures that the riskiness of investment projects are in line with the attitudes of investors toward risk. financial market regulation reflects concerns for both efficiency and fairness. in which relative strength continually shifts from side to side. investors might well take on more risk than is objectively appropriate or overreact to information and cause security prices to be excessively volatile. By its nature. and investors can select efficient portfolios. who provided the first behavioral characterization of fairness in terms of “reference transactions. Pareto efficiency precludes the wasting of resources. some managers and investors might misinterpret the information conveyed within efficient prices and make supoptimal choices.7.” Shefrin and Statman (1992) propose that in financial markets fairness is understood as entitlements to particular 1 This section is drawn from some of the main ideas in Shefrin and Statman (1992.1 Dynamic Tug-of-War 131 7. managers. a point emphasized by Shefrin and Statman (1992). An allocation is said to be Pareto efficient when no other feasible allocation of resources and technology can improve one person’s situation without harming another’s. In a Pareto-efficient allocation. Asymmetric information destroys informational efficiency and moral hazard interferes with it. Pareto efficiency is consistent with errors and biases. . The first is Pareto efficiency and the second is informational efficiency. Notably. Shefrin and Statman (1992) build on Kahneman et al. Efficient prices provide proper guidance to entrepreneurs.

132 Behavioralizing the Approach to Financial Market Regulation classes of transactions. Some combinations of efficiency and fairness dominate other combinations that are higher in one or the other attribute. — Freedom from misrepresentation entitles people to rely on voluntarily disclosed information being accurate. Choices along the frontier involve reductions in fairness to increase fairness or reductions in fairness to increase efficiency. This class of fairness recognizes that some people commit cognitive errors. Protecting them may take the form of compulsory disclosure or the outright prohibition of certain types of transactions. — Fairness in equal information entitles people to equal access to a particular set of information. Policymakers operate as if they have preferences which depend on both efficiency and fairness. — Fairness in equal processing power entitles people not only to equal access to information but also to a “competency floor” of information processing skills. A conceptual efficiency/fairness frontier is composed of combinations that are not dominated. . Bargaining power inequality can occur. They identify seven classes of fairness by the following entitlements: — A transaction is fair when the parties enter into it voluntarily. — Fairness as freedom from impulse entitles people to protection from possible imperfect self-control. This is the notion of fairness inherent in the term “fair and orderly market. Freedom from coercion entitles people to avoid being coerced into a transaction against their will. when one party to a transaction is wealthier than the other.” — Fairness in equal bargaining power entitles people to equal power in negotiations leading to a transaction. — Fairness in efficient prices entitles people to prices they perceive to be efficient. for example. and being permitted to engage in a transaction of their choosing.

However.1. increasing its vigilance and its clamor for regulatory protection from interest groups. This theory postulates that interest groups regularly enlist politicians and regulators in their battles with one another over regulatory proposals. Stoking the fire of public outrage is often difficult because the cost of public mobilization is relatively high. Politicians and regulators who allow interest groups to capture them under such circumstances might lose more political support than they gain. Similarly. Economic booms and rising financial markets placate the general public. regulators want to steer the regulatory process in directions that benefit them. Stigler emphasized that an interest group is likely to capture its regulators when the per-capita benefits to the members of the interest group are large relative to per-capita benefits to the general public. noting that interest groups would not capture their regulators when the total benefits to the general public are sufficiently large. Public outrage is a fire which must be hot enough for a period long enough to shape regulations into a form that benefits the general public.7. politicians need resources such as campaign contributions to maximize their chances of re-election. and favors in exchange for enacting and executing regulations which transfer wealth to them. even if the per-capita benefits are relatively small.2 Capture Theory Shefrin and Statman (2009) augment the framework described above to reflect capture theory. reducing its vigilance and making it easier for politicians to tilt regulations toward interest groups. contributions.1 Dynamic Tug-of-War 133 7. At the same time. recessions and declining financial markets enrage the general public. lawyers. Politicians have the power to direct regulators to benefit one interest group or another. Peltzman (1976) augmented capture theory. in prestige or industry jobs once they leave public service. See Stigler (1971) who noted that each interest group — bankers. union members. Politicians and regulators have limited power to tilt regulation toward interest groups and their power varies by the environment in which they operate. . The political process involves competition among interest groups each attempting to capture politicians and regulators by some combination of votes. In contrast. and employers — wants regulations that maximize its wealth.

3. Merit regulations take power from entrepreneurs and give it to investors. Merit regulations may cause deviations from informationally efficient prices. As to fairness. 2. 7. three distinct motivations can be seen underlying the regulation of margin: the need to protect investors from their own . merit regulation interferes with the right to freedom from coercion. each configuration of regulations gives rise to a configuration in efficiency/fairness space. financial. Shefrin and Statman (1992) discuss six specific regulations. a move toward equal power if entrepreneurs have more power than do investors in a mandatory disclosure environment. Shefrin and Statman (1992) emphasize the issues of fairness loom large at the time the regulations were put in place. on account of its costs.134 Behavioralizing the Approach to Financial Market Regulation interest groups have ready mechanisms for lobbying which they can mobilize quickly to take advantage of even small changes in the economic. thus adversely affecting the right to correct (efficient) prices. Disclosure regulations: These improve informational efficiency relative to a buyer beware world. and technological environments.1. Less clear is whether mandatory disclosure improves Pareto efficiency. but its paternalistic characteristics enhance rights to equal processing power freedom from impulse. placing each into the historical context which led the regulations being established. legal. Margin regulations: In the deliberations leading up to the passage of the 1934 Securities Exchange Act. political. The implications of disclosure regulations for fairness are similar to those for merit regulation. and the potential elimination of some risky ventures. Merit (“blue-sky”) regulations: These diminish Pareto efficiency by restricting choices of those offering and those buying securities. 1. The discussion below emphasizes how the essential issues in the debates can be viewed through the lens of tradeoffs involving differing notions of efficiency and fairness.3 Specific Regulations Conceptually. Each regulation is on the frontier unless another regulation improves both fairness and efficiency.

Suitability regulations: Suitability rules present a clear case where some fairness rights win over other rights and over efficiency. 5. and the desire to allocate credit to productive investment instead of financial speculation. and the other involving the right to be protected from one’s own mistakes. Insider trading regulations: The efficiency aspects associated with these regulations pertain to adverse selection. The associated issues involve freedom from coercion.” has been the more prominent of the two. particularly as to a disclosure-or-abstain rule. that insiders create new information when they trade. The fairness question has played a more prominent part in the debate than the efficiency question. Some argue that the regulations improve informational efficiency.7. however. The paternalistic nature of suitability rules is evident in the incentives they provide to brokers to err on the side of low risk rather than on the side of high expected returns. winners and losers are expected. A fair investment game is perceived to be a similar game of skill.1 Dynamic Tug-of-War 135 poor judgment. 6. with equal access to public information from which the investor can infer which securities are undervalued and which are overvalued. A move from a no-paternalism to a full-paternalism world enhances the right to freedom from cognitive errors and the right to freedom from impulse. and freedom from impulse. which centers on the notion of the “level playing field. because most material information is released to the public eventually. the limitation of price volatility induced by low-margin requirements. Trading interruption regulations which followed the stock market crash of 1987: trading halts are detrimental to Pareto . efficient prices (fair and orderly markets). but trading on inside information is unfair because it conveys a non-skillbased advantage to one party in the trade. equal processing power. 4. This issue is largely one of timing. while others argue the opposite. Two fairness threads can be seen one involving the right to equal information. In a fair game of skill. The former issue.

In this case. type 1 investors believe that the probability that a u follows a u is 0.8. whereas type 2 investors believe that the probability is 0. type 2 investors believe that aggregate consumption growth evolves according to a process that is independent and identically distributed (i. Let aggregate consumption be equal to 100 units at t = 0. and is equal either to u > 1 or d = 1/u < 1. consider a numerical binomial example. Thereafter the growth rate of aggregate consumption per period is stochastic. a special case of the framework described in Section 5. Both types of investors share a common rate of time preference (δ = 0.i.76 percent.67.1 Theoretical Example Absence of Constraints To fix ideas.4. efficient prices (fair and orderly markets). assume that type 2 investors are information traders. For the sake of argument. freedom from impulse. the probability that u occurs is 0. such as the closing bell on the exchange. and equal bargaining power. Type 1 investors believe that at t = 1.99). For t > 2. so that the economy grows either at 5 percent per period or shrinks at 4. called type 1 and type 2. the two types of investors differ in their beliefs. As to fairness.136 Behavioralizing the Approach to Financial Market Regulation efficiency because they prevent parties from entering into agreements viewed as beneficial to themselves. with the probability of u being 0.95 and that a d follows a d is 0. One critic notes that natural trading interruptions. A counterargument is that trading halts enhance efficiency during a sharp decline by permitting settling up to ensure that all parties are solvent. However. type 1 investors will be excessively optimistic after . Any impact on informational efficiency depends on whether such efficiency is achieved in the absence of interruptions. I take u = 1. meaning they possess correct beliefs. where T = 2.8.4 7.76 at all dates. In contrast. apparently do not dampen volatility during turbulent periods.05.d.).1. equal information processing. 7. with type 1 holding 65 percent of the initial wealth of the economy. Let there be two types of investors in the model.1. the main issues involve coercion.

1 Dynamic Tug-of-War 137 economic growth has been positive and excessively pessimistic after economic growth has been negative. Notably. When debt is exclusively used to purchase equity. which I take to be the market portfolio and the one-period risk-free security.7. Their equity holdings amount to 346 percent of the value of their portfolios. and underestimates the probability of a down-move by 31 percent.3 percent of their wealth and consume the remainder. the market portfolio corresponds to unlevered equity. type 2 investors short the market.6 percent.1 percent. and take a short position of −$245.8 percent if it contracts. only two securities are needed to complete the market.7 of the market portfolio.” At the beginning of each date.) At t = 0.6 and hold $442. The differences in beliefs at t = 0 lead to an equilibrium with nonzero sentiment Λ.7 in the market portfolio. as the SDF embodies all price information. type 1 investors are more optimistic than type 2 investors. At t = 0. investors decide how to divide their portfolio wealth into current consumption and an end of period portfolio such that the value of consumption together with the end-of-period portfolio is equal to the value of the beginning of period portfolio. They believe that the equity premium is −0. the degree of margin corresponds to the debt-to-equity ratio. Keep in mind that in this model. equilibrium returns are established as if the market overestimates the probability of an up-move by 12 percent. Because the example is binomial. the S&P 500. and type 2 investors have total wealth equal to $104. Type 2 investors hold $314. At the end of t = 0. the interest rate is 3.7 percent. type 1 investors have total wealth equal to $193. .6 of the risk-free security. Type 1 investors believe that the expected return on the market is 4. type 1 investors borrow $314.1 percent at t = 1 if the economy has grown and −3.1). the market portfolio returns 6. As Equation (5. In contrast. Both types save 66. Put another way. (Returns are calculated using Equation (5. say. reflecting “hot hand fallacy. and therefore the equity premium is 0.1) indicates. unlike. nonzero sentiment impacts the SDF underlying equilibrium returns. which is why they use leverage to increase their holdings of the market portfolio. In equilibrium. At t = 1.5 percent. type 1 investors purchase 71 percent of their equity holdings on margin.

They will decrease their consumption by 22 percent.7 percent. type 2 investors will end up holding levered positions.8 percent. They will increase their consumption by 11. and Efficiency In this framework. whereas type 2 investors believe it to be 2. type 2 investors will increase their consumption by 28 percent. regulation can limit leverage and short sales by imposing upper and lower bounds on the portfolio proportion allocated to the market portfolio. if aggregate consumption declines at t = 1. thereby exacerbating the degree of non-optimality. After a down move at t = 1. In this situation.9 percent. If growth is negative at t = 1.138 Behavioralizing the Approach to Financial Market Regulation If growth is positive at t = 1. 7. type 2 investors alternate between acting as if they are excessively optimistic and excessively pessimistic. In contrast. For example. This example illustrates that investors who hold incorrect beliefs choose non-optimal portfolios that feature inappropriate risk exposure and excessive trading. type 2 investors will decrease their consumption by 6. This state of affairs may well occur even when sentiment is zero and prices are efficient. If prices are inefficient. In an example featuring more periods. type 1 investors will earn high returns from their levered portfolios.4. type 1 investors believe that the equity premium is −1. For this reason. From a modeling perspective.2 Regulatory Constraints.1. type 1 investors will earn low returns from their levered portfolios. For example. while holding $987 of the risk-free security. Fairness.3 percent. then sentiment typically induces excessive volatility in the market risk premium. type 1 investors will become pessimistic and short $882 of the market portfolio. These contrasting portfolios reflect contrasting beliefs about the equity premium. In contrast. financial market regulation operates by imposing constraints on investor’s portfolios. and back again. Volatile beliefs translate into excessive trading. the impact of regulations induces investors to act as they hold different beliefs than they actually . some investors might jump from holding portfolios that feature too much leverage to shorting the market portfolio. The intent of such limits is to force investors to act as if their beliefs were less extreme.

or less vulnerable to emotion. The imposition of regulatory constraints has implications for fairness and efficiency. As in Section 5.1 percent. rather than their true belief of 80 percent. In this case.1 Dynamic Tug-of-War 139 do. What difference would this make to investors’ positions. type 1 investors would reduce the amount of equity they purchase on margin to 50 percent. such constraints typically impact market prices. At t = 1 after a down move.6 percent at t = 0). and increase the amount of equity purchased on margin to 88 percent (from 71 percent at the end of t = 0). imputed beliefs.3 percent. a change of measure captures the essential issue. At t = 1. Suppose a regulator. might improve fairness defined in terms of freedom from misrepresentation.1 percent instead of 2. equal information. for example by reducing the degree of volatility in the equity premium. were to step in and limit the degree to which investors can purchase equity on margin to 50 percent in all subsequent periods. The higher margin requirements would force type 1 investors to trade at t = 0 as if they assigned a probability of 72 percent to an up-move at t = 1.7.6 percent or 7. type 1 investors increase their estimate of the equity premium to 1 percent (from 0. type 2 investors would do likewise. and market prices? At both t = 0 and t = 1 after an upmove. freedom . say {Qi } instead of {Pi }.9 percent or −4. equal processing ability. Consider what happens if economic growth is positive at t = 1. Does this mean that the imposition of regulatory constraints has resulted in an increase in fairness coupled with a loss in efficiency? That depends on both the definitions of fairness and efficiency. this would decrease the volatility of the equity premium. concerned that the market was irrationally exuberant at t = 0. the ex post premium would be either 1. In turn. To see the effect of the change of measure. I modify the example as follows. Constraining investors to act as if they are smarter. Notably. The objective expected premium would increase from −0. thereby reducing a substantial portion of the excessive optimism prevailing in the market. Suppose that every investor’s allocation in the constrained equilibrium features higher expected utility than in the unconstrained equilibrium (when both are measured with respect to the correct probability vector Π).7 percent to −0.

By definition. and equal bargaining power. If regulatory constraints lead the representative investor to hold beliefs in the Q-equilibrium which are closer to Π than in the P-equilibrium. However.2 He suggests that the focus on individual 2 The constitutionality of SOX is currently before the U. Hence. forcing investors to trade as if they hold beliefs {Qi } which are different from {Pi } moves the equilibrium allocation off the Pareto-efficiency frontier defined by {Pi }. He points out that the costs of disclosure imposed by SOX are less vivid than poignant stories about families who suffered large losses. . As to efficiency. He calls his framework the psychological attraction approach to regulation. efficiency can also be defined as informational efficiency. which will rule on the issue in 2010. by their nature constraints involve coercion. The basis for Hirshleifer’s framework is the following series of seven phenomena. such a situation would also be viewed as an increase in fairness. they perceive that the constraints result in a move off the Pareto-efficient frontier. thereby leading to reduced fairness defined as freedom from coercion. 1.2 Biased Cost–Benefit Assessment Hirshleifer (2008b) focuses attention on the manner in which systematic biases on the part of participants in the political process impact regulatory outcomes. However. regulations might lead to an improvement in efficiency defined relative to correct prices. when investors view markets through the lens of {Pi }. Hirshleifer points out that accounting scandals such as occurred at firms such as Enron and WorldCom precipitated passage of the Sarbanes–Oxley Act of 2002 (or SOX).S. 7. Salience and vividness effects: Regulatory debates are influenced heavily by extreme events.140 Behavioralizing the Approach to Financial Market Regulation from impulse. when fairness is defined as the right to trade at correct prices. but onto the Paretoefficiency frontier defined by {Qi }. because particular parties advocating increased regulation exploit psychological biases to attract attention and support. and by heart-rending personal stories. Supreme Court.

This tendency applies to losses suffered from derivatives trading. As a result. and relatively weak out-group. Charity is directed at addressing recent losses. overconfidence helps to explain the existence of excessive regulation. understood as a web of coadaptations resulting from individual decisions. Hirshleifer asserts that . Overconfidence: According to Hirshleifer. leading to a disposition to regulate derivatives. restrict short-selling. Equality underlies hostility to lenders. Considerations of fairness can explain regulations intended to limit speculation. disliked. Scapegoating and xenophobia: When things go wrong. Omission bias: This bias involves the favoring of omissions over commissions. regulation intended to protect unsophisticated investors from securities lacking merit might also prevent them from risk-reducing diversification. Economic and stock market downturns increase pressure for regulation targeting specific culprits. Equality involves equal division of resources. people instinctively seek someone to blame. and penalize short-term capital gains. 3. thereby leading the public to attach negative perceptions of derivatives. Reciprocity involves quid pro quo. a view underlying regulations to prevent usury. Fairness and reciprocity norms: Hirshleifer’s view of fairness partly overlaps with the discussion above. 5. losses is amplified through conversation and media reporting. He cites the work of Adam Smith and Friedrich Hayek in respect to a spontaneous order. 4. namely reciprocity.7. the public tends to overlook benefits associated with derivatives. According to Hirshleifer. such as members of a visible. that is too complex for a central planner to comprehend.2 Biased Cost–Benefit Assessment 141 2. which are biased toward the communication of adverse and emotionally charged news. equality. He focuses on three norms. and the Sarbanes–Oxley Act as examples. Charity involves acts intended to relieve distress. particularly as intermediaries are seen by the public as adding little value. and charity. As an example. reciprocity motivates scapegoating. such as hedging. Hirshleifer cites the Securities Acts of 1933 and 1934.

the call for greater regulation by a few can lead others to infer that there is good reason to do so. she also heads the committee which oversees the dispensing of funds from the Troubled Assets . political. He suggests that the illusion of control leads regulators to conclude that they are able to avert bubbles and crashes. Currently. In this regard. Hirshleifer suggests that conspiracy theories are prone to develop during market crashes. He suggests that the betterthan-average effect induces encourages regulators to believe they are more skilled than investors at identifying the value of the market. whereby the increased social discourse about an issue leads to the perception that the issue is truly important. he singles out hedge fund managers as being especially suspect. or economic. mood can be contagious. Information cascades can develop. shape financial regulation through word of mouth.3 Regulation and Financial Crises Elizabeth Warren is a member of Harvard University Law faculty who specializes in bankruptcy. Moreover. he mentions transactions taxes to limit speculation and monetary policy to address irrationally exuberant equity markets or to provide stimulus during an economic downturn. He makes the point that the general public have a poor understanding of how major players in financial markets affect risks. In this last regard. he indicates that following adverse outcomes. As examples. and are often foreigners. 6. which leads to calls for more strict regulation. 7. Ideological replicators: Ideologies.142 Behavioralizing the Approach to Financial Market Regulation overconfidence leads public officials to believe that they can institute regulatory mechanisms which can improve on market allocation. 7. be they religious. as they operate in secrecy. In this respect. thereby aggregating at the social level. Mood effects and availability cascades: Hirshleifer makes the point that short-term moods impact judgments and decisions relating to the long-term. regulators come under criticism for failing to prevent unfavorable outcome.

The parallels between the S&L crisis of the 1980s and the financial crisis of 2008–2009 are striking. and subprime mortgages in the 2008–2009 crisis. — subprime loans. the key features involved the following: — the market for real estate. the Securities Exchange Act of 1934. history. featuring acquisition. and strongly suggests the collective failure to learn. As Shefrin and Statman (2009) point out. the stock market crash of 1929 and Great Depression following heated regulatory irons to extremely high temperatures. and the financial crisis in 2008–2009. — a political and regulatory environment encouraging home ownership. that a financial crisis including a credit squeeze accompanied the 1792 crisis. In Rather (2009). — S&L institutions as key players. and construction loans in the S&L crisis. Major financial crises occurred every 15–20 years between 1792 and 1929. which she notes has been accompanied by two major financial crises in the United States. the savings and loan (S&L) crisis of the 1980s. facilitating strikes that changed their shapes radically. — a period of interest rate volatility featuring low real interest rates followed by high real interest rates. However. development. — the funding of projects with a high probability of generating poor cash flows. Warren pointed out that no major financial crises occurred between 1934 and the 1980s that were comparable in scale to those occurring before the Great Depression. — a period of active deregulation featuring low capital ratios. including the Securities Act of 1933.S. the first economic crisis in U. — fixed income securities at the core of the crisis. Warren reminds us of some history.7. . — risk-seeking behavior by financial institutions. since 1980. For both crises. and the Glass-Steagall Act which separated commercial and investment banking. a process of deregulation has been underway.3 Regulation and Financial Crises 143 Relief Plan (TARP) passed by Congress in 2008 to rescue key financial institutions in distress.

Essentially the industry continued to hold mortgages it originated. but did not — — — — lengthen term of its deposits. — pyramid (Ponzi) schemes. and — subsequent turbulence in equity markets. 7. shorten the term of its mortgages. Because of inflation.144 Behavioralizing the Approach to Financial Market Regulation — agency issues on the part of real estate appraisers in the S&L crisis and rating firms in the financial crisis of 2008–2009. nominal capital requirements were low (6 percent of liabilities). hedge its risk exposures. and then the financial crisis of 2008–2009. — government agencies becoming insolvent: the FSLIC insuring S&Ls becoming insolvent in 1987. regulations restricted S&Ls to deposits and home mortgage loans. However. interest rate ceilings prevented S&Ls from paying competitive interest rates on deposits. there was no fair value accounting in place at the time. first by describing the S&L crisis. and in 2009 the FDIC was nearing insolvency. Regulatory reform began to take shape in the period 1980–1982. — high accounting profits reported by financial institutions. In the discussion below I highlight some of the parallels in the above list. and sell mortgages and buy shorter term assets. Legislators granted the S&L industry the power to enter new areas of .1 S&L Crisis in the 1980s Until the 1980s. Notably. leading to an outflow of deposits from S&Ls. — large bonuses to executives. During the late 1970s. when the critical seeds of the S&L crisis were sown. For the industry as a whole. these increases caused the market value of their assets to fall below the market value of their liabilities. nominal interest rates rose dramatically at this time. The industry response to the changing economic environment featured status quo bias.3. and so this state of affairs was not recognized within the industry’s financial statements.

most S&Ls were reporting losses. For those readers who instinctively hold a laissez faˆ philosophy and believe in market ire . For example. The Federal Home Loan Bank Board was the agency responsible for regulating S&Ls. but definitely featured fraud.3 Regulation and Financial Crises 145 business. At the same time that regulatory restrictions on the S&L industry were being lifted. Germain and DIDMCA increased S&L’s risk-seeking opportunities for funding high risk development projects.7. — permitted lax accounting standards. — removed limits on the amounts of brokered deposits an S&L could hold. and permitted the portfolios of these S&Ls to comprise up to — — — — 40 30 10 10 percent percent percent percent of of of of assets assets assets assets in in in in commercial mortgages. the Depository Institutions Deregulation and Monetary Control Act (DIDMCA) made the following changes: — removed interest rate ceiling on deposit accounts. It eliminated previous statutory limit on the loan to value ratio. and — raised the deposit insurance limit raised to $100. Many institutions took advantage of these opportunities.000. — expanded authority for federal S&Ls to make ADC loans. in what might have featured aversion to a sure loss. and — reduced the size of its regulatory and supervisory staff. consumer loans. Between 1980 and 1985 Bank Board — reduced net worth requirements for insured S&L institutions. and commercial leases. During the early 1980s. GarnSt. intended to increase S&L profits as opposed to promoting housing and homeownership. commercial loans. The Garn-St.000 from $40. the ability of regulatory authorities to monitor the industry was reduced. Germain Act of 1982 was a Reagan administration initiative for federally chartered S&Ls.

read on.” reflecting the inflated values of their projects. who would be protected by higher deposit insurance limits. by granting loans to developers of lowquality projects for amounts that overstated the value of these projects. executives would have paid themselves substantial bonuses for having turned around a nearly failed bank at a time when the whole . The loans just described were for interest only. Notably. along with points and fees.146 Behavioralizing the Approach to Financial Market Regulation efficiency. and the S&Ls extracted a 50–50 profit split. asking for preliminary appraisals. especially to the S&Ls. To be sure. if they could find buyers for these projects. should find the following value destructive sequence of events instructive. For those readers who believe the discipline of the market will surely curtail such behavior. Indeed. See Black (2005). and these payments were subsequently booked as income on S&Ls’ financial statements.” The scheme involved locating other weak developers searching for loans. this pyramid scheme would have to end. Being market transactions. with principal due in two-to-three years. the terms of these loans held back the interest payments due. in a scheme called “cash for trash. that the cash flows from most of these low quality. the associated gains were registered as income on these S&Ls’ financial statements. and the S&L would fail. the S&Ls communicated the size of the proposed loan to potential appraisers. S&Ls then selected high appraisers and rewarded them with a continuing stream of business. In doing so. induced clean opinions by auditors. high risk projects. which they did. were low. Value destroying S&Ls seized newfound opportunities in the market for commercial property loans. Could the S&Ls avoid recognizing the losses in question? Yes. Eventually. S&Ls preferred weak developers. However. these large loans featured an “anticipated profit component. at some point it would become apparent and did. that does not mean that the S&L executives would fail. This made the firms appear to be profitable. The same phenomenon manifested itself 20 years later with ratings firms. Notably. together with a profit sharing arrangement. The inflated prices in “cash for trash” resulted in book profits for these projects. as they would agree to high interest rates. who agreed to purchase the failing projects of others at inflated prices. and attracted more depositors.

taxpayers over $2 billion. a surge in subprime mortgage lending.S.3. Akerlof and Shiller (2009) argue that the root lies with irrational decisions associated with the occurrence of a housing market bubble. explicit corruption. interest rates during the period 2003–2006. I suggest that each of the above explanations contains an element of truth. Posner (2009) argues that the root lies with a weak regulatory structure. which they trace to low U. there was a housing market bubble.3 Regulation and Financial Crises 147 industry was in decline. In this regard. Brunnermeier et al. Sullivan (2009) blames poor corporate governance. Lincoln’s failure cost U. Gray made a strong effort to reverse the direction of deregulation.7. In 1983. Perhaps. His attempts to investigate one particular S&L. Differentiating among these various views requires a search for the devil in the details. within which private-sector decisions were largely rational. and to recapitalize the FSLIC. Lincoln Savings and Loan. In Shefrin (2009).S. they were sufficiently overconfident to believe that their fraudulent behavior would go undetected. interest rates were low. 7. the surge in subprime mortgage lending was highly problematic. and there were failures in governance at financial firms. The total cost of addressing the failures in the S&L industry was approximately $150 Billion. and unwise governmental mandates and guarantees. In the end. the Bank Board came under new leadership. the Gramm–Leach–Bliley . led him to be summoned by five Senators who questioned the appropriateness of his actions. if not ignoring the warnings from regulators. and the breakdown of the rational adverse selection (“lemons”) paradigm. which was in the process of going insolvent.2 Financial Crisis of 2008–2009 Opinions about the root cause of the financial crisis of 2008–2009 differ. (2008) and Taylor (2009) also argue that an asset pricing bubble was a major issue. One lesson not learned from the S&L crisis involves failing to recognize the dangers from simultaneously expanding risky alternatives to financial institutions and reducing regulatory oversight. The regulatory structure was weak. with the appointment of Ed Gray.

including the Federal Reserve. and the opposition to it from Federal Reserve Chair Alan Greenspan. and foreclosure rates would rise dramatically. who was subsequently relieved of her position for issuing those warnings. and the strictest of the proposed provisions gone. homeowners would be unable to sell their homes. with new rules released in late 2006. the banking industry lobbied against the proposal. saying that many buyers. He added that if housing prices declined. During 2005. he was among the first to sound the alarm about risky lending in the housing market. — Banks must advise home buyers that interest rates might increase dramatically. bank regulators observed deteriorating conditions in the mortgage market. would soon be unable to afford their mortgage payments. and — Banks must cap their risky mortgages so that a string of defaults would not be crippling. 2008). The provisions stripped from the final rules included the following: — Banks that bundle and sell mortgages must disclose to investors exactly what they are buying. In mid-2005. which required unanimous agreement among different regulatory bodies. and Clinton administration officials Robert Rubin and Larry Summers. particularly those with bad credit. and proposed new guidelines for banks writing risky loans. See PBS (2009) which documents Born’s concerns. resulting in payments becoming due sooner than expected. The governments banking agencies spent nearly a year debating the rules. . The industry succeeded in achieving a one-year delay. Another regulator whose warnings went in vain was Comptroller of the Currency John Dugan (CNN. Brooksley Born. This occurred despite the warnings issued by the head of the Commodity Futures Trading Commission. The Commodity Futures Modernization Act left the derivatives market without regulatory oversight.148 Behavioralizing the Approach to Financial Market Regulation Act of 1999 fragmented regulatory oversight so that no regulator had a comprehensive picture of each financial firm. In line with capture theory. — Banks must verify that buyers actually have jobs and can afford houses.

Henry Waxman: You were the longest serving chairman of the Federal Reserve in history and during this period of time were perhaps the leading proponent of deregulation of our financial markets. . having worked both as a regulator for 18 years and similar quantities in the private sector. that the loan officers of those institutions knew far more about the risks involved and the people to whom they lent money than I saw even our best regulators at the fed capable of doing.7. who chaired the Federal Reserve during the key years in which the seeds of the crisis were sown. However. . Greenspan’s successor at the Fed. In the same vein. .” He pointed to the use of risky financing by some homeowners and suggested that the price increases in those local markets were unsustainable. beginning with a question from committee chair Henry Waxman. Certainly you were the most influential voice for deregulation. Greenspan testified before Congress that some local housing markets exhibited “froth.” In 2008. My question for you is simple. Greenspan testified before the House of Representatives Committee on Oversight and Government Reform. he concluded that there was no national housing bubble and that the economy was not at risk. were you wrong? Alan Greenspan: I made a mistake in presuming that the self-interest of organizations. specifically banks and others. “[W]e do not expect significant spillovers from the subprime market to the rest of the economy or to the financial system. In June 2005. Ben Bernanke.3 Regulation and Financial Crises 149 Consider the comments of Alan Greenspan. and it has been my experience. You have been a staunch advocate for letting markets regulate themselves. Below is an excerpt from his appearance. was capable of protecting their own shareholders and the equity in the firms. So the problem here is something which looked . gave a speech on 17 May 2007 in which he stated. especially 10 years at a major international bank.

lies in behavioral finance. . confirmation bias. however. I present a series of case studies demonstrating that the financial crisis of 2008–2009 was driven by a combination of sentiment in asset prices and organizational issues that are at the . Henry Waxman: This is your statement. and aversion to a sure loss played major roles in the decision processes at investment banks. . none meaningfully worked. and regulatory agencies. insurance firms. rating agencies. You had the authority to prevent irresponsible lending practices that led to the subprime mortgage crisis. . and the missing pieces in his understanding of what happened. You were advised to do so by many others. My judgment is that free competitive markets are by far the unrivaled way to organize economies. In this regard. And now our whole economy is paying its price. . psychological issues such as overconfidence. .150 Behavioralizing the Approach to Financial Market Regulation to be a very solid edifice and indeed a critical pillar to market competition and free markets did break down and I think that shocked me. We have tried regulation. Do you feel that your ideology pushed you to make decisions that you wish you had not made? Alan Greenspan: I found a flaw in the model that I perceived as the critical functioning structure that defines how the world works. “I do have an ideology. excessive optimism. I suggest that the flaw in Greenspan’s model. [T]his crisis.” That was your quote. . institutional investors. has turned out to be much broader than anything I could have imagined · · · Those of us who have looked to the self-interest of lending institutions to protect shareholders’ equity-myself especially-are in a state of shocked disbelief. I will change my views. I still do not fully understand why it happened and obviously to the extent that I figure out where it happened and why. In Shefrin (2009).

However. AIG American International Group (AIG) is an insurance firm which facilitated the explosive growth in subprime mortgage lending in 2004 and 2005 by selling credit default spreads (CDS) that insured against default. Moreover. AIG’s financial product division irrationally failed to track the proportion of subprime mortgages in the pools being insured. this failure occurred despite employee incentives that balanced long-term performance against short-term performance. incentives. apparently under the illusion that housing prices would continue to rise and that default rates would not be affected by the increased ratio of subprime to prime mortgages. Merrill Lynch used CDS contracts to create synthetic collateralized debt obligations (CDOs) because the number of subprime mortgages available to create traditional CDOs was insufficient relative to the firm’s aspirations. all of which highlight the importance of debiasing. many financial firms took on the risk themselves. UBS UBS’s investment banking division admitted to misgauging subprime mortgage risk by not “looking through” the CDO structures and by . planning.3 Regulation and Financial Crises 151 heart of behavioral corporate finance. The case studies highlight (to a greater or lesser degree) irrational decision making at key points in the financial product supply chain. and information sharing. In 2004 and 2005. thereby misgauging the risk of those assets and causing the associated CDS contracts to be mispriced. the activities of financial institutions might have been rational because they were able to shift default risk to AIG by purchasing credit default swap protection. I emphasize that psychological pitfalls dominated the key processes at financial firms. Interestingly. conversations that AIG had with its investment bank trading partners indicate the presence of a widespread conservatism bias regarding the assumption that historical mortgage default rates would continue to apply. these processes pertained to goal setting.7. after AIG stopped selling CDS contracts. Below are highlights from the cases. As mentioned in Shefrin (2009). At one point.

Did the rating agencies exhibit irrational behavior by weakening their risk assessment criteria to cultivate more business? This question is different from asking whether the ratings agencies behaved ethically. I suggest that reference point-induced risk seeking and groupthink played major roles at rating agency S&P. such as the investment bankers at UBS. . UBS placed no operational limits on the size of its CDO warehouse. however. many investment banks found that they could not locate buyers for the CDOs and. these agencies suggested that their ratings be treated as only one piece of information when assessing risk. not the end investors in the securities. For this reason. as a result.152 Behavioralizing the Approach to Financial Market Regulation assuming that historical default rates would continue to apply. Some hold the equity tranches of CDOs as a signal to the buyers of the less risky tranches. pay for ratings. users who accepted their ratings at face value behaved irrationally. It was not alone. As the housing market fell into decline. the rating agencies explicitly indicated that their ratings were premised on accepting the information they received as accurate. in the sense that the issuers of securities. and naive end investors. was inventory risk–risk stemming from warehousing the subprime positions that underlay CDOs. Investment banks are typically intermediaries. not end investors planning to hold large positions in subprime mortgages. At the same time. What created many of the losses for investment banks. Rating Agencies: S&P and Moody’s The rating agencies and investors’ reliance on them played a huge role in the financial crisis. even if the mortgages featured limited documentation. relied on the risk assessments of rating agencies. Both (supposedly) sophisticated investors. UBS’s underperformance relative to competitors led them to exhibit reference point-induced risk seeking. inadvertently became end investors. By this argument. The major problem with the behavior of rating agencies might arise more from a conflict of interests (the principal–agent conflict) than irrationality. However. This behavior was compounded by poor incentive structures at UBS that emphasized short-term performance over long-term performance.

The report indicates that between 1992 and 2008. and the bank holding companies were Citigroup and JPMorgan Chase. 2004. They conclude that tranches that were rated AAA actually corresponded to BBB when rated according to the firms’ models and default standards. The report reveals that the SEC conducted investigations and examinations related to Madoff’s investment advisory business but failed to uncover the fraud. and Morgan Stanley. A focal point of the criticism of investment bank oversight involved a meeting that took place at the SEC on April 28. the agency received six distinct complaints about Madoff’s operations. largely because it ignored pertinent evidence that it uncovered. one of which involved three separate versions. Several of the complaints suggested that Madoff was running a Ponzi scheme. As to the Bernard Madoff Ponzi scheme. as did two bank holding companies. The investment banks were Bear Stearns. I suggest that confirmation bias lay at the heart of the SEC’s failure to detect Madoff’s Ponzi scheme. That meeting established the Consolidated Supervised Entities (CSE) program. when the commission was chaired by William Donaldson. Griffin and Tang (2009) find that ratings by Moody’s and S&P were higher than what their models implied. Merrill Lynch. Five investment banks joined the CSE program.7.3 Regulation and Financial Crises 153 In this regard. President Obama proposed a broad new regulatory framework for financial markets. Regulatory Response In June 2009. a voluntary regulatory program that allowed the SEC to review the capital structure and risk management procedures of participating financial institutions. the SEC issued an internal report in the summer of 2009. Subsequent to the April 2004 meeting. SEC The SEC came under intense criticism for its lax oversight of investment banking practices and for failing to detect a large hedge fund Ponzi scheme run by Bernard Madoff. Lehman Brothers. the largest in history. Goldman Sachs. the SEC failed to provide qualified staff in sufficient numbers to effectively monitor these financial institutions. emphasizing a centralized structure .

3.2 that after 1991. provides the SEC with authority to sanction supervisors at credit rating agencies for failing to supervise employees. market abuses. These units would focus on investment companies (including hedge funds and private equity firms). the financial crisis of 2008–2009 merely . The third discount rate is time invariant.154 Behavioralizing the Approach to Financial Market Regulation for oversight.S. he uses three different discount rates. The SEC’s Division of Enforcement Management is in the process of developing a plan to restructure its approach through five new specialized units. suggested by Elizabeth Warren. In October 2009. This Act allows investors to sue rating agencies for knowingly or recklessly failing to review key information in their ratings. companies. 2005). Notice from Figure 7. the municipal securities market.2 displays an update of the analysis in Shiller (1981. The stock market crash of October 1987 occurred in the midst of the S&L crisis. market prices exceeded intrinsic value no matter which of the three proxies for intrinsic value is used. The point Shiller makes in this figure is that market prices are excessively volatile relative to dividends. Notably. the House Financial Services Committee voted to pass the “Accountability and Transparency in Rating Agencies Act” in order to address some of the issues mentioned above. In making this point. and contains requirements to mitigate the conflicts of interest that result because it is the issuers who actually pay rating agencies. with the Federal Reserve Bank playing a central role. According to this figure. Figure 7. Among the other proposals is the creation of a consumer protection agency. respectively. 7. As can be seen. and bribery issues in foreign countries involving U.1 displays the monthly time series of stock prices between 1984 and 2009. both the period associated with the S&L crisis and the financial crisis of 2008–2009 featured considerable turbulence. there is opposition to this last proposal from the financial services industry. not the investors. complex derivatives.3 Market Prices Figure 7. Two discount rates are time varying and are derived from aggregate consumption and market interest rates.

Shiller’s measures might have become biased downward. even after the collapse of the dot-com bubble. but moved to become 20 percent overvalued by year end. 7. Shiller bases his measures for intrinsic value on aggregate dividend flows. and ended 2009 20 percent higher than in June.htm. Payouts include share repurchases. with the 2009 value corresponding to June. A final point to note is that along the decomposition of total returns along the intrinsic value paths features a much larger contribution from . Given that Shiller’s measures for intrinsic value move much more slowly than market prices.1 Display of the monthly time series of stock prices between 1984 and 2009.com/index. not just dividends. served to bring stock prices into alignment with fundamental value.7. The data displayed in Figure 7. Boudoukh et al. one might infer that the market was 20 percent undervalued at the trough in March.3 Regulation and Financial Crises 2000 1800 1600 Real Stock Price (Shiller Index) 1400 1200 1000 800 600 400 200 0 19 84 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999 000 001 002 003 004 005 006 007 008 009 2 2 2 2 2 2 1 2 2 1 2 1 1 1 2 1 1 1 1 1 1 1 1 1 1 155 Date Fig. However. Their analysis suggests that as the ratio of dividends to payouts declined after the 1970s.2 are annual. Stock prices were approximately 18 percent lower at the trough in March.irrationalexuberance. (2007) analyze how intrinsic values for stocks should reflect the present value of future payouts. This might explain why stock prices continued well above intrinsic value. Source: Shiller (2005) and http://www.

156 10000 Behavioralizing the Approach to Financial Market Regulation Present Value Dividends Using Actual Future Interest Rates 1000 Present Value of Dividends Using Constant Discount Rate Actual Real Stock Price Price 100 Consumption-Discounted Dividends 10 1860 1880 1900 1920 1940 Year 1960 1980 2000 2020 Fig. the efforts of special . thereby forcing most of the return to be generated by the dividend yield. Source: Shiller (2005) and http://www.irrationalexuberance. Much of the time.htm. relative to the general public.com/index.2 Comparing actual real stock price with three alternative PDVs of future real dividends. 7. not dividend yield. price changes are very gradual. their efforts will succeed. 7.4 Differences in Perspective Hirshleifer’s perspective emphasizes overregulation. capture theory suggests that private interests will seek to exploit the regulatory framework to their advantage. On the other hand. With market value. and they will achieve greater power than those seeking stronger regulation that favors the general public. his position is similar to the general position that guided Alan Greenspan’s approach to regulation when he chaired the Federal Reserve. However. the lion’s share of total return stems from stock price appreciation. for the measure of intrinsic value associated with the time invariant discount rate. dividend yield than is the case with market value. stemming from the psychological underweighting of regulatory costs relative to benefits. However. In many ways.

or politicians. representing either overreaction or underreaction to the crises in question. They have short memories. . and the tugof-war between varying notions of fairness. a point which Alan Greenspan learned the hard way. No group is immune to psychological pitfalls. be they investors. regulators. Regulatory agencies are imperfect.4 Differences in Perspective 157 interests will also create conditions in which the financial system is prone to a crisis. any serious discussion of financial market regulation needs to highlight behavioral biases and be based on facts rather than ideology. the elements of capture theory. corporate managers. At the same time. It is easy to have debates about regulation on broad principles. regulations established in response to specific crises might be highly imperfect. reflecting external shocks. as the discussion about the SEC demonstrates. Free markets which are perfectly selfregulating constitute an ideal. over time regulations will wax and wane in strength. the balance of power shifts to those seeking regulations that favor the general public. A crisis creates a window of opportunity for those seeking stronger regulation. People are prone to recency bias. However. For this reason. In a crisis.7. traders. bankers. There is a literature suggesting that some of the provisions in the Sarbanes–Oxley Act are in this vein.

I suggest that the future of academic finance will combine the realism offered by behavioral finance with the rigor offered by neoclassical finance. 158 . Behavioral finance has brought a realistic dimension to finance. Sections 4 through 7 of this monograph provide some specific ideas for what shape the behavioralization of finance might take. a significant portion of the behavioral finance literature lacks rigor. In addition to surveying the foundations underlying the behavioral literature. In this monograph I describe highlights of the behavioral finance literature. focusing attention on both its strengths and its weaknesses. At the same time.8 Concluding Remarks Behavioral finance is in the process of transforming the financial paradigm from a neoclassical framework to a framework that explicitly incorporates psychological phenomena. the four-fold pattern for risk attitudes. The approach to behavioral preferences begins with the behavior that motivated prospect theory: gains and losses relative to a reference point as carriers of value. and indeed some of its claims appear to be in error. replacing unrealistic assumptions neoclassical finance makes about investors and managers being fully rational and market prices being efficient. I use this monograph to identify trends — possible future directions for behavioral finance.

self-control. especially for the term structure of interest rates and option prices. it will be convenient to represent behavioral beliefs using change of measure techniques.159 psychophysics. In this regard. My view of behavioral beliefs is that they are best modeled as imperfect learning systems that are heuristic-driven and violate Bayes’ rule. and the emotions emphasized in SP/A theory. Behavioral portfolios reflect a series of psychological phenomena such as status quo bias. See Kr¨ussl et al. the attention hypothesis. a Some of the most recent behavioral advances have occurred in corporate finance. its formalization possesses some structural weaknesses. behavioral preferences will add features in other psychological-based theories of risk such as regret. Risk premiums will be determined through the SDF and mean-variance (MV) frontier in analogous ways to neoclassical theory. the disposition effect. and the predilection for lottery stocks. especially for studying the impact of beliefs on asset pricing. Presidential election cycle. cash dividends. Behavioral beliefs and preferences serve as the basis for behavioralizing portfolio theory. and editing. structured products. In the future. (2009). systematic risk will feature both fundamentals and sentiment. overconfidence. Behavioral asset pricing models have already begun to generalize neoclassical asset pricing models. Behavioral portfolios tend to be undiversified. except that the SDF and MV-frontier will reflect sentiment. Behavioral portfolios are structured to satisfy a variety of financial and psychological needs. For these reasons. Formally. In addition. Notably. Sentiment will be expressed as a log-change of measure.S. A recent example pertains to the U. prospect theory ignores the role of emotion in decision making. I focus attention on three areas that strike me as . behavioral portfolio models will feature combined diffusion/jump processes to capture the tendency of investors to make adjustments to their portfolios intermittently rather than continuously in time. These models will feature the combination of status quo bias and other psychological traits. In addition. The behavioralization of asset pricing theory will replace unrealistic neoclassical assumptions with more realistic psychologically based assumptions in models built around the SDF. and covered call option positions. work on identifying return anomalies will continue.

the lessons of the S&L crisis of the 1980s largely went unheeded. and behavioral economists need to incorporate more rigor into theirs. .160 Concluding Remarks especially germane to future contributions. and for this reason it will be important going forward for financial economists to study the role of psychology in financial market regulation. it is this. and (3) the psychological profiles of entrepreneurs. which might explain why financial economists were ill equipped to comment on the aftermath of the financial crisis of 2008–2009. At present there is but a small literature in financial market regulation. (2) the impact of social networks on corporate decisions and corporate value. If there is a lesson in this monograph for economists. and the seeds of that crisis were allowed to germinate and wreck havoc two decades later. neoclassical economists need to incorporate more psychology into their models. It seems to me that psychological pitfalls lie at the heart of the financial crisis of 2008–2009. The future lies with both groups meeting in the middle. I suggest that by and large. They are: (1) the influence of corporate managers’ personal characteristics on their corporate decisions. Going forward.

Kraus (1990). forthcoming. Journal of Forecasting 9. and Why it Matters for Global Capitalism. ‘Direct evidence on the market-driven acquisitions theory’. Anderson. ‘Nonparametric risk management and implied risk aversion’. Cheng (2009). Organizational Behavior and Human Decision Processes 66. Allport. London York: Cambridge University Press. (1932). W. New York. M. and R. Economete e rica 21. Journal of Econometrics 94. and A. L. B. Herren. J. Holt. J. Isen (1988). 503–546. ‘La psychologie de l’hˆmme rationnel devant le risque: o critique des postulats et axiomes de l’ ´cole am´ricaine’. Shiller (2009). T. Remembering: A Study in Experimental and Social Psychology. and L. Arkes. G. Princeton. Y. 9–51. W. and S. Akerlof. and A. 228–236. (1953). H. P. NJ: Princeton University Press. ‘Judgemental extrapolation and the salience of change’.References A¨ ıt-Sahalia. The Psychology of Rumor. Journal of Financial Research. J. 347–372. Lo (2000). 161 . Animal Spirits: How Human Psychology Drives the Economy. Postman (1947). ‘The role of potential loss in the influence of affect on risk-taking behavior’. Allais. G. Ang.. and Y. Andreassen.

‘Market timing and capital structure’. Journal of Finance 57. ‘Do call prices and the underlying stock always move in the same direction’. 271–99. and J. Review of Economics and Statistics 80. Wurgler (2007b). 969–1006. M. Baker. and J. Coval. (1956). Whole No. M. 1–32. Brookings Papers on Economic Activity pp. R. Quarterly Journal of Economics 118. 271–288. Das. 416. ‘When does the market matter? Stock prices and the investment of equity-dependent firms’. Stein (2004). Wurgler (2007).. Baker. and J. M. In: E. Baker. M. ‘Studies of independence and conformity: A minority of one against a unanimous majority’. Psychological Monographs 70. S. Wurgler (2003). M. and S. and J. Baker. J. ‘Corporate financing decisions when investors take the path of least resistance’. and J.. Journal of Finance 59. Review of Financial Studies 13. ‘Behavioral corporate finance: A survey’. 549–584.. Journal of Financial Markets 7. ‘Investor sentiment in the stock market’. S. ‘Investor sentiment and the crosssection of stock returns’. Journal of Financial Economics 84. M.. G. Cao. Wurgler (2002). and J. Wurgler (2006). M. Stein (2004). ‘Market liquidity as a sentiment indicator’. M. ‘The Effect of Dividends on Consumption’. and J. M. Baker. M. 129–151. and Z. Ruback. J.): The Handbook of Corporate Finance: Empirical Corporate Finance. 62–72. and J. and J. Wurgler (2004a). 271–288. ‘Appearing and disappearing dividends: The link to catering incentives’. Baker. Baker. New York: Elsevier/North Holland. Baker. P. . Nagel. Baker. ‘The central tendency: a second factor in bond yields’.162 References Asch. 277– 291. Bakshi. Chen (2000). C.. ‘A catering theory of dividends’. Eckbo (ed. Stein. Wurgler (2007a). and J. 266–298. S. Journal of Finance 61. Balduzzi. 1645–1680. C. Journal of Economics Perspectives 21. Baker. Journal of Financial Economics 73. D.. Wurgler (2004b). Foresi (1998).

Journal of Finance 56. Barber. Barber.. and T. Stulz. Lee. Barberis. Barone-Adesi.. Shleifer (2003). loss aversion and individual stock returns’. Working Paper. Barber. Davis. Vishny (1998). 161–199. 1–53. ‘Just how much do individual investors lose by trading?’. and L. Wurgler (2001b). ‘Stocks as lotteries: The implications of probability weighting for security prices’. Barberis. N. Barberis. A. Review of Financial Studies 21. Huang. Harris (eds. ‘A GARCH option pricing model with filtered historical simulation’. Huang (2008). Review of Financial Studies 22. ‘A model of investor sentiment’. North Holland. Shleifer. Constantinides. Y. ‘Comovement’. Y. Review of Financial Studies 22. B. Working Paper.References 163 Barber. and A. ‘Trading is hazardous to your wealth: The common stock performance of individual investors’. Journal of Financial Economics 68. Journal of Finance 55. 1223–1258. . Shleifer. N. and T. Odean (2005). ‘Prospect theory and asset prices’. Journal of Finance 64. American Economic Review 98. Engle. Barberis. Odean. 307–345... University of Chicago. Quarterly Journal of Economics 116. and N. ‘Mental accounting. Barberis.): Handbook of the Economics of Finance. Liu. Liu.. ‘Do individual day traders make money? Evidence from Taiwan’. N. N. 1247–1292. B. T. 751–784.. Odean (2009a). and T. ‘A Survey of behavioral finance’. ‘Do retail trades move markets’. Zhu (2009b).. Barber. ‘What drives the disposition effect? An analysis of a long-standing preference-based explanation’. G. and M. M. Y. Thaler (2003). University of California. N. ‘All that glitters: The effect of attention on the buying behavior of individual and institutional investors’. Barberis. Mancini (2008). Huang (2001). ‘Style investing’. and T. Barberis. Review of Financial Studies 21. Odean (2000). In: G. N. 773–806. 2066–2100. R. Barberis. and R. 609–632. N. and M. A. Odean (2008). Y. B. and J. N. and R. Lee. Xiong (2009). and W. Santos (2001a). and M. 151–186. B. B. R. and T. Journal of Financial Economics 49. 785–818.

I. 539–576. Quarterly Journal of Economics 121. Kimball. and R. Russell Sage Foundation and Princeton University Press. ‘Market models and heteroscedasticity of residual security returns’. ‘Testing agency theory with entrepreneur effort and wealth’. Moskowitz. Quarterly Journal of Economics pp. 537–579. (2001). E. Benartzi. Harvey (2007). F. Journal of Business & Economic Statistics 1. 157–195. American Economic Review 91. Duke University. S. Benartzi. Working Paper. B. R. Journal of Finance 56. Juster. Talwar (1983). Thaler (2007). 73–92. S.. and A. Desai.. ‘Structured investment products with caps and floors’. and R. S.164 References Barone-Adesi. (2008). M. and M. Bernard. S. Boyle (2008). Working paper. Working paper. Peleg. ‘Excessive extrapolation and the allocation of 401(k) accounts to company stock’. Gagliardini. G. ‘Preference parameters and behavioral heterogeneity: An experimental approach in the health and retirement survey’. and P. Urga (2004). 474–485. and P. B´nabou. Vissing-Jorgensen (2004). S. ‘Heuristics and biases in retirement savings behavior’. and G. Quarterly Journal of Economics 112. ‘Myopic loss aversion and the equity premium puzzle’. and R. ‘Managerial overconfidence and corporate policies’. J. G. Graham. 81–104. ‘Groupthink: Collective delusions in organizations e and markets’. 1747– 1764. T. ‘Nave diversification strategies in defined contribution savings plans’. Princeton University. Benartzi.): The Behavioral Foundations of Policy. ‘Earnings manipulation and managerial investment decisions: Evidence from sponsored pension plans’. In: E. Barsky. ‘Choice architecture and retirement saving plans’. and C. D. Journal of Finance 60. T. Benartzi. ‘Testing asset pricing models with coskewness’. M. Journal of Business & Economic Statistics 22.. R.. Journal of Economic Perspectives 21. Thaler (2009). P. Rauh (2004). Thaler (1995). and R.. S. Bitler. Bergstresser. 163–168. Shapiro (1997). Barone-Adesi. University of Waterloo. M. Benartzi. 79–98. and J. Shafir (ed. D.. C. forthcoming. . Thaler (2001). Ben-David.

Working Paper. Crime. Nain (2009). pp. J.. D. Roberts (2007). Kagel and A. I. H. Michaely.. M. and R.. ‘Control fraud as an explanation for white-collar crime waves: The case of the savings & loan debacle’. Law and Social Change 43. Brunnermeier. ‘The brain as a hierarchical organization’. Motivation and Emotion 3. Review of Financial Studies 22. and H. Diamond. C. ‘Uncertainty breeds decreasing impatience: The role of risk preferences in time discounting’.. Ukhov (2006). and M. ET: Zurich. Y. Epper (2009a). American Economic Review 98. Brocas. Cooper (1979). K. Harvey. ‘The desirability of control’. Amsterdam: Elsevier. Schenk-Hopp´ (eds. Blume. In: T. and R.): The Handbook of Experimental Economics. Bruhin. C. 483–527. Richardson. 587–703. Blackburn. ‘On the importance of measuring payout yield: Implications for empirical asset pricing’. C. W. ‘Individual decision making’. Working paper. ‘Estimating preferences toward risk: Evidence from Dow Jones’. ‘The performance of stock-price driven acquisitions’. 14612. (2005). and A. J. ‘Board connections and M&A transactions’. R.. A. Bouwman. Easley (2008). Working paper. Michaely (2005). Journal of Finance 62. 1312–1346. In: J. Boudoukh.References 165 Black. 633– 679. and M. Cai.. Camerer. L. 1–29. D. A. and A. Epper (2009b). Roth (eds. Hens and K. ‘Risk and rationality: Uncovering heterogeneity in probability distortion’. Bruhin. and T. and T. Burger. 877–915. NBER Working Paper No. . M. ‘Deciphering the liquidity and credit crunch 2007–08’. Fehr-Duda.): Handbook of Financial Mare kets: Dynamics and Evolution. and D. Carrillo (2008). Working paper (ET: Zurich). Fehr-Duda. Rajam (2008). Sevilir (2009). Princeton University Press. H. Fuller. A. ‘Market selection and asset pricing’. ‘Payout policy in the 21st century’. Graham. (1995). Journal of Financial Economics 77. J. University of Indiana. University of North Carolina. 381–393. and J. Brav.

Stanley (2009). Chan. S. Schaller (2001). Johnson. and H. 1st edition. C. 235–268. Campbell. Chen. Stein (2002). Journal of Political Economy 98. Hong. and S. ‘Glamour vs. ‘By Force of Habit: A consumption-based explanation of aggregate stock market behavior’.. J. . Review of Economic Studies 58. 138–160. 171–205. ‘CEO confidence and forced turnover’. 663– 680. Journal of Finance.. and B. R. Cochrane. J. G. Y. Schaller (2004). Shiller (1991). American Economic Review 91. Chopra. Asset Pricing. CNN. Working paper. Working paper (MIT). Journal of Financial Economics 31. value: The real story’. J. Asset Pricing.166 References Campbell. and H. N. Frankel. P. J. J. Emory University. J. Cochrane. Princeton: Princeton University Press. Campbell. Working paper. American Economic Review 95. and J. Rutherford. and R. Cochrane. Ritter (1992). J. H. ‘Bond Risk Premia’. Constantinides. R. ‘Government warned of mortgage meltdown: Regulators ignored warnings about risky mortgages. (2008). (2000). (1990). ‘Habit formation: A resolution of the equity premium puzzle’. 519–543. ‘Testing behavioural finance theories using trends and sequences in financial performance’. December 1. Journal of Political Economy 107. J.. and J. Princeton: Princeton University Press. Cochrane (1999). Chirinko. (2000). Kothari (2003). ‘Measuring abnormal performance: do stocks overreact?’. Journal of Financial Economics 66. 495– 514. W. 205–251. Texas A&M University. ‘Asset pricing at the millenium’. Campbell. and J. delayed regulations on the industry’. 2nd edition.. Lakonishok. R. Piazzesi (2005). ‘Breadth of ownership and stock returns’. S. J. ‘Business fixed investment and ’bubbles’: The Japanese case’. ‘Yield spreads and interest rate movements: A bird’s eye view’. Chirinko. and M. (2005).

M. Jackwerth.. 703–738. J.. Dunkelberg (1988). and W. W. and S. W. D. C. Dai. 1839–1885. Singleton (2003). Woo. ‘Market reactions to tangible and intangible information’.. 921–965. and T. ‘Entrepreneurs’ perceived chances for success’. and R. Daniel. Journal of Finance 56. and A. 631–678. K. Review of Financial Studies 22. Shumway (2005). DeLong. Journal of Finance 60. 793–808. A. Review of Financial Studies 20. Journal of Political Economy 98. and S. Journal of Applied Finance 18(2). University of Lugano. L. Womack (2007). and S. Subrahmanyam (1998).. ‘Does the stock market overreact?’.References 167 Constantinides. ‘A rose. Waldmann (1990a). De Giorgi. and P. Perrakis (2009). 2371–2388. Cooper. B. ‘Financial market equilibria with cumulative prospect theory’. and R.. D. Thaler (1985). ‘Term structure dynamics in theory and reality’. and K. 1247–1277. 1–34. ‘Investor psychology and security market under. Review of Financial Studies 16. ‘Do behavioral biases affect prices?’. and M. 97–108. and K. G. ‘Overconfidence. A. Journal of Business Venturing 3. Dimitrov. Shefrin. H.. Subrahmanyam (2001). Journal of Finance 53. J. E. 7–21. J. Working paper. ‘Noise trader risk in financial markets’. ‘Analyst Hype in IPOs: Explaining the popularity of bookbuilding’. Summers.. Rau (2001).com by any other name’. K. Working paper. arbitrage and equilibrium asset pricing’. F. De Bondt. Hens. 1605–1643. DeBondt. Hirshleifer. K. Q. Rieger (2009). and A. Staikouras (2008). . ‘Behavioral finance: Quo vadis?’. F. Cooper. Daniel. Degeorge. University of Zurich. Journal of Finance 61. T. Coval. Shleifer. ‘Mispricing of S&P 500 index options’. Titman (2006). G. (2009). Journal of Finance 56. R.and overreactions’.. Derrien. Muradoglu. Daniel. Journal of Finance 40. De Giorgi.. ‘Goal-based investing with cumulative prospect theory and satisficing behavior’. 1021–1058. E. D. O. Hirshleifer.

probability of success and failure. and J. B. R. S. and S. Dong. Griffin (1985). E. 313–338. Dong. Doukas. C. R. B. K. Robinson. Shleifer. and investor conservatism’. A. European Financial Management 11. E.. 683–700. Teoh (2003). Journal of Finance 45.. and S. 369–403.. Journal of Risk and Uncertainty. and S. Journal of Economic Theory 62. van de Ven (2008). Diether. La-Ornual (2009). Summers. and P.. I. ‘The satisfaction with life scale’. C. Emmons. Diener. 437–481. 725–762. Journal of Finance 57. Murthy (1994). 1–20. Hirshleifer. Dittmar. 375–395. and A. forthcoming. ‘Positive feedback investment strategies and destabilizing rational speculation’. L. ‘The survival of noise traders in financial markets’. M. ‘Aspiration level. ‘European momentum strategies. 2113– 2141. McKnight (2005). Journal of Business 64. ‘Nonlinear pricing kernels. Malloy. Waldmann (1991). A. Waldmann (1990b). Journal of Personality Assessment 49. Shleifer. and G. (2002). 121–158. DeLong. Summers. kurtosis preference. and R. ‘Differences of opinion and the cross-section of stock returns’. ‘Intertemporal asset pricing with heterogeneous beliefs’. A. Diecidue.. Review of Finance 9. 294–320. and expected utility’.. and evidence from the cross section of equity returns’. Larsen. ‘Reconciling support theory and the book-making principle’. J. Journal of Finance 57. and R. J. information diffusion. L. Detemple. International Economic Review 49. 71–75. Richardson. Journal of Finance 61. Scherbina (2004).168 References DeLong. Journal of Corporate Finance 12. University of Michigan working paper. M. J. Huberman (2005). Veld (2005). ‘Talk and action: What individual investors say and what they do’. Diecidue. E. Dichev. Dorn. ‘What are stock investors’ actual historical returns?’. J. D. R. . J. and D. (2004). ‘Does investor misvaluation drive the takeover market?’. and C. ‘Why individual investors want dividends’. D.

French (1993). B. Petmezas (2005). ‘Size and book-to-market factors in earnings and returns’. Fama. Duffie. Review of Finance 9. Dumas. Journal of Finance 51. In: B. Uppal (2009). ‘What can rational investors do about excessive volatility and sentiment fluctuations?’. E. Working Paper. Foldes. ‘Acquisitions. Journal of Finance 62. Edwards. ‘Martingale conditions for optimal saving: Discrete time’. 131–155. French (2001). Foldes. ‘Sports sentiment and stock returns’. and the perception of environmental health risks’. Risk Analysis 14. Flynn. W. P. . Kurshev.. Saita. Fama. L. and K. Seasholes (2005). L. New York: Wiley. Journal of Financial Economics 83. Journal of Financial Economics 60. Slovic. 3–43.. Berry (2006). 1641–1701. and K. 1101– 1198. and R. Journal of Finance 50. E. L. 1967–1998.References 169 Doukas. ‘Conservatism in human information processing’. Mertz (1994). A. forthcoming. 55–84. Edmans. S. and K. race. 3–56. J. and T. Fama. D. Feng. (2000). E. and K. Old Dominion University. D. ‘Do investor sophistication and trading experience eliminate behavioral biases in finance markets?’. Fogel. Wang (2007). (1968). E. ‘Disappearing dividends: Changing firm characteristics or lower propensity to pay?’.): Formal Representation of Human Judgement. 305–351. ‘Valuation and martingale properties of shadow prices: An exposition’. Journal of Economic Dynamics and Control 24. 83–96. ‘Gender. K. French (1995). A. 635–665. and M. Journal of Finance. and O. overconfident managers and self-attribution bias’. L. French (1996). ‘Multi-period corporate default prediction with stochastic variates’. 107–116. ‘Common risk factors in the returns of bonds and stocks’. ‘Multifactor explanations of asset pricing anomalies’. and K. Fama. Kleinmutz (ed. J. and D. Norli (2007). and C. Journal of Financial Economics 33. A... ‘The disposition effect and individual investor decisions: The roles of regret and counterfactual alternatives’. Garcia. (1978). Journal of Mathematical Economics 5. Journal of Behavioral Finance 7.

and G. European Financial Management 11. ‘Entrepreneurship and household saving’. C.. See (2003).. Fracassi. J. ‘How are stock prices affected by the location of trade?’. and executive stock options’. Odean (2003). ‘Overconfidence. Huberman (2004). C. Poterba (1991). ‘Rain or shine: Where is the weather effect?’. 805–827.): Thinking: Psychological Perspectives on Reasoning. and A. W. Frieder. Working Paper. Froot. and C. Journal of Financial and Quantitative Analysis 40. Judgment and Decision Making. S. and T. Gilchrist. French. order flow. Goetzmann. investment policy. Gervais. and returns’. Journal of Financial Economics 53. 1233–1260. ‘Overconfidence. 2737–2784. 189–216. Heaton. Subrahmanyam (2005). 57–85. Review of Finance 12. John Wiley and Sons. Goel. Kumar (2008). and E.170 References Fox. American Economic Review 81. ‘Equity portfolio diversification’. Duke University. 559–578. ‘Earnings announcements. Journal of Finance 63. and corporate governance’. and A. ‘Investor diversification and international equity markets’. Gaston-Breton. K. Hardman and L. Zhu (2005). Lejarraga (2009). (2004). Dabora (1999). Gentry. W. Hubbard (2001). and G. . Goetzmann. A. Mayer (2001). and K. and J. Journal of Monetary Economics 52. Thakor (2008). In: D. Working paper. 222–226. ‘External networking and internal firm governance’. Macchi (eds. K. C. Quarterly Journal of Economics 116. C. NBER Working Paper No. Universidad Carlos III de Madrid. L. 7894. Working paper (UCLA). ‘Biases in learning from past strategic decisions’. N. ‘Do stock price bubbles influence corporate investment?’. S. CEO selection. ‘Brand perceptions and the market for common stock’. and A. and N. Working paper. ‘Loss aversion and seller behavior: Evidence from the housing market’. and J. Himmelberg. L. W. B. Purdue University. D. ‘Belief and preference in decision under uncertainty’. 433–463. Tate (2009). and G. Frieder. Genesove.

107–155. and D. Ishii. ‘Managerial attitudes and corporate actions’. and trading activity’. mental accounting and momentum’. Guedes. Grinblatt. D. ‘How distance. ‘How big are the tax benefits of debt?’. Graham. Journal of Financial Economics 78. Graham. Rajgopal (2004). Kumar (2006). 1901–1941. J. 1809–1833. and culture influence stockholdings and trades’. M. Tversky (1992). Tang (2009). ‘Dividend preference of retail investors: Do dividend clienteles exist?’. Grinblatt. Harvey. ‘Sensation seeking. ‘Prospect theory. ‘The weighing of evidence and the determinants of overconfidence’. Journal of Finance 64. risk e sharing and asset prices with regret’. and A. and M. Puri (2009). Journal of Finance 6. 187–243. M. Journal of Accounting and Economics 40. 311–333. J. Graham. Guedj. Working paper. and B. Keloharju (2009). J. language. and T. Journal of Finance 51.. ‘The theory and practice of corporate finance: Evidence from the field’. University of Texas. Graham. Journal of Finance 56. . Scharfstein (2004). I. Han (2004). Opler (1996). 549–578. Salani´ (2006). ‘Individual decisions under risk. and D. M. Graham. and S. overconfidence. Harvey. 1305– 1336. and A. and M. Keloharju (2001a). 411–435. C. and B. Gompers. Journal of Finance 55. and M. J. C. J. J. Cognitive Psychology 24. M. P. ‘Did subjectivity play a role in CDO credit ratings?’. and C. Quarterly Journal of Economics 118. ‘What makes investors trade?’. 589–616. ‘The determinants of the maturity of corporate debt issues’. NBER Working Paper No.. 1053–1073.. ‘Corporate governance and equity prices’. Harvey (2001). Grinblatt. Metrick (2003). C. ‘Organizational scope and investment: Evidence from the drug development strategies of biopharmaceutical firms’. ‘The economic implications of corporate financial reporting’. (2000). R. J.References 171 Gollier. Griffin. Working paper (University of Toulouse). Journal of Finance 56. and M. Griffin. Grinblatt. and A. Keloharju (2001b). Working Paper. J. Duke University. Journal of Financial Economics 60. 3–73. w10933.

Jagannathan (1991). Jegadeesh. Kreps (1978). Oxford University. 401–439. Baker (1996). B. ‘Investor sentiment and option prices’. ‘Behavioral portfolio choice: an analytical treatment’. Journal of Financial Economics 41. 225–262. and X. Hansen. (2002). Heaton. and A. 33–45. B. Rangel (2009). ‘Conditional skewness in asset pricing tests’. ‘Self-control in decisionmaking involves modulation of the vmPFC valuation system’. (2009). Journal of Financial Economics 82. L. Hackbarth. Financial Management 31. and E. R. Heaton. Quarterly Journal of Economics 92. Weisbach (2009). 646–648. N. and R.. and D. ‘World markets for raising new capital’. P. B. Zhou (2009). C. and M. Larrick. X. R. and J. Sack. Siddique (2000). ‘Implications of security market data for models of dynamic economies’. and N. C. J. 387–414. (2000). Swanson (2005).. 323–336. and A.. Hamilton. ‘Speculative investor behavior in a stock market with heterogeneous expectations’. Science 324. 1–37. (2008). C. B. M. Camerer. 1163– 1198. 1263–1295. ‘Managerial optimism and corporate finance’. ‘Does entrepreneurship pay? An empirical analysis of the returns to self-employment’. R. Lucas (2000). ‘Commonality in the determinants of expected stock returns’.. Journal of Political Economy 108. Heath. American Economic Review 95. Harvey. Henderson. J. ‘Determinants of corporate borrowing: A behavioral perspective’. Working paper.172 References Gurkaynak. Han. Hare. Journal of Corporate Finance 15. ‘Cognitive repairs: How organizational practices can compensate for individual shortcomings’. J. Klayman (1998). He. Journal of Political Economy 99. 63–101. Journal of Finance 55. 604–631. 389–411. L. Haugen. . D. A. Harrison. 425–436. B. ‘The excess sensitivity of long-term interest rates: Evidence and implications for macroeconomic models’. T. Research in Organizational Behavior 20. Review of Financial Studies 21. ‘Asset pricing and portfolio choice: The importance of entrepreneurial risk’. and D. Journal of Finance 55.

H. ‘Differences of opinion. and T. Huberman. ‘Psychological bias as a driver of financial regulation’. Working Paper. Subrahmanyam. 265–295. ‘Investor psychology and asset pricing’. D. analyst coverage. (2006). ‘Are observed capital structures determined by equity market timing?’. A. Working paper. Hirshleifer. H. Pennings (2010). Titman (2006). and J. Ritter (2009). ‘Feedback and the success of irrational traders’. ‘A unified theory of underreaction. R. Journal of Finance 58. 237–271. Hong. Vlcek (2005). H. Schenk-Hopp´ (2008). and the profitability of momentum strategies’. 487– 525. ‘Bad news travels slowly: Size. University of Zurich. D. ‘Does prospect theory explain the disposition effect?’. and M. and overreaction in asset markets’. A. 856–874. Journal of Finance 55. ‘Good day sunshine: Stock returns and the weather’. T.. Journal of Finance 54. J. .. Hirshleifer. Shefrin. H. and S. Hoffmann. Yu (2007). Lim. T. 1009–1032. and J. 311–388. Journal of Financial Economics 81. 1207–1242. 659–680. (2008a). Journal of Finance 62. 221–243. Review of Financial Studies 16. Shumway (2003). 1533–1598. Hovakimian. Stein (2003). Hens. Journal of Financial and Quantitative Analysis 44. (2001). Handbook of Financial Markets: e Dynamics and Evolution.. Amsterdam: Elsevier. 2143–2184. and J. Hong. G.. Stein. D. and J. Review of Financial Studies 14. Hong.References 173 Hens. ‘Simple forecasts and paradigm shifts’. ‘Familiarity breeds investment’. Huang. Maastricht University and Netspar. H. Hirshleifer. Stein (2000). and J. D. Journal of Finance 56. G. T. Stein (1999). Hirshleifer. European Financial Management 14. Hong. A. (2008b). ‘Testing theories of capital structure and estimating the speed of adjustment’. short-sale constraints and market crashes’. Journal of Financial and Quantitative Analysis 41. and J. and K. ‘Behavioral portfolio analysis of individual investors’. C. momentum trading.

. E. Management Science 55. Jones. Thaler (1991). 457–491. (2006). ‘Small trades and the cross-section of stock returns’. 385–390. Ishii. and O. Journal of Financial Economics 66. Review of Financial Studies 13. C. ‘Stock repurchases in Canada: Performance and strategic trading’. I. 181–208. ‘Market timing and managerial portfolio decisions’. Napp (2007). ‘Non-Monotonicity of the Tversky-Kahneman probability weighting function: A cautionary note’. and R. ‘Market underreaction to open market share repurchases’. Ingersoll. Review of Economic Studies 74. ‘Volume and price patterns around a stock’s 52-week highs and lows: Theory and evidence’. European Financial Management 14. Jouini. Kahneman. ‘A trade-based analysis of momentum’. (2008). Lang. D. Knetsch. J. Stanford University. Journal of Economic Dynamics and Control 30. (2005). M. Vermaelen (1995). 1903–1949.174 References Huddart. Working paper. In: R. Hvidkjaer. Ikenberry. 2373–2398. ‘Short-sale constraints and stock returns’. and Y. ‘Heterogeneous beliefs and asset pricing in discrete time’. Thaler (ed. Review of Financial Studies 19. D. (1972). and C. S. 433–451. Yetman (2009). Vermaelen (2000). 207–239. Jouini. New York: Russell Sage Foundation. University of Maryland. 16–31. 1149–1174. (2000). J. S. J. D.. and T. J. C. J. Lakonishok. J. and C. Working Paper. and M. Jenter. Lakonishok. (2005). Ikenberry. Hvidkjaer. Houghton Mifflin.. Journal of Finance 55. Xuan (2009). ‘Fairness as a constraint on profit seeking entitlements in the market’.. ‘Recovering risk aversion from options prices and realized returns’. 1233–1260. Journal of Financial Economics 39. Victims of Groupthink: A Psychological Study of Foreign Policy Decisions and Fiascoes. ‘Acquirer-target social ties and merger outcomes’. ‘Consensus consumer and intertemporal asset pricing with heterogeneous beliefs’. Jackwerth. D. Journal of Finance 60. Janis. Napp (2006). and T. . S. Lamont (2002).): Quasi-Rational Economics. E.

References

175

Kahneman, D., P. Slovic, and A. Tversky (1982), Judgment Under Uncertainty: Heuristics and Biases. Cambridge: Cambridge University Press. Kahneman, D. and A. Tversky (1973), ‘On the psychology of prediction’. Psychological Review 80, 237–251. Kahneman, D. and A. Tversky (1979), ‘Prospect theory: An analysis of decision making under risk’. Econometrica pp. 263–291. Kahneman, D. and A. Tversky (1982), ‘The psychology of preferences’. Scientific American 246, 160–173. Kamstra, M., L. Kramer, M. Levi, and T. Wang (2009), ‘Time-varying preferences and SAD: Evidence from an asset pricing model’. Working paper, University of Toronto. Kamstra, M. J., L. A. Kramer, and M. D. Levi (2000), ‘Losing sleep at the market: The daylight-savings anomaly’. American Economic Review 90, 1005–1011. Kausar, A. and R. Taffler (2006), ‘Testing behavioural finance models of market under and overreaction: Do they really work?’. Working Paper (University of Edinburgh. Kaustia, M. (2004), ‘Market-wide impact of the disposition effect: Evidence from IPO trading volume’. Journal of Financial Markets 7, 207–235. Kaustia, M. (2009), ‘Prospect theory and the disposition effect’. Journal of Financial and Quantitative Analysis, forthcoming. Kaustia, M. and S. Kn¨pfer (2009), ‘Learning from the outcomes u of others: Evidence from the stock market’. Working paper, Aalto University. Klayman, J. and Y. Ha (1987), ‘Confirmation, disconfirmation and information in hypothesis testing’. Psychological Review 94, 211–228. Klein, P. (2001), ‘Were the acquisitive conglomerates inefficient?’. RAND Journal of Economics 32, 745–761. Knoch, D., F. Schneiderb, D. Schunk, M. Hohmann, and E. Fehr (2009), ‘Disrupting the prefrontal cortex diminishes the human ability to build a good reputation’. PNAS 106, 20895–20899. Knutson, B. and C. Kuhnen (2009), ‘The impact of affect on beliefs, preferences and financial decisions’. Journal of Financial and Quantitative Analysis, forthcoming.

176

References

Knutson, B., C. Kuhnen, G. E. Wimmer, and P. Winkielman (2008), ‘Nucleus accumbens activation mediates the influence of reward cues on financial risk-taking’. NeuroReport 19, 509–513. Kr¨ussl, R., A. Lucas, D. Rijsbergen, P. van der Sluis, and E. Vrugt a (2009), ‘Washington meets wall street: A closer examination of the presidential cycle puzzle’. Working paper, University of Amsterdam. Kumar, A. (2009), ‘Who gambles in the stock market?’. Journal of Finance 64, 1889–1933. Kuran, T. and C. Sunstein (1999), ‘Availability cascades and risk regulation’. Stanford Law Review 51, 683–768. Kurz, M. (1997), Endogenous Economic Fluctuations: Studies in the Theory of Rational Beliefs. Studies in Economic Theory No. 6, Springer-Verlag: Berlin and New York. Kyle, A. and F. A. Wang (1997), ‘Speculation duopoly with agreement to disagree: Can overconfidence survive the market test?’. Journal of Finance 52, 2073–2090. La Porta, R., J. Lakonishok, A. Shleifer, and R. Vishny (1997), ‘Good news for value stocks: further evidence on market efficiency’. Journal of Finance 49, 1541–1578. Lakonishok, J., I. Lee, N. Pearson, and A. Poteshman (2007), ‘Option market activity’. The Review of Financial Studies 20, 813–857. Lakonishok, J., A. Shleifer, and R. W. Vishny (1994), ‘Contrarian investment, extrapolation and risk’. Journal of Finance 49, 1541– 1578. Lamont, O. (2000), ‘Investment plans and stock returns’. Journal of Finance 55, 2719–2745. Landier, A. and D. Thesmar (2009), ‘Financial contracting with optimistic entrepreneurs: Theory and evidence’. Review of Financial Studies 22, 117–150. Leary, M. (1983), ‘Social anxiousness: The construct and its measurement’. Journal of Personality Assessment 47(1), 66–75. Lee, C., A. Shleifer, and R. Thaler (1991), ‘Investor sentiment and the closed-end fund puzzle’. Journal of Finance 46, 75–110. Lehenkari, M. (2008), ‘The disposition effect: Underlying mechanisms and implications for individual investors’. Working paper, University of Oulu, Finland.

References

177

Lennox, R. and R. Wolfe (1984), ‘Revision of the self-monitoring scale’. Journal of Personality and Social Psychology 46, 1349–1364. LeRoy, S. and R. Porter (1981), ‘The present-value relation: Tests based on implied variance bounds’. Econometrica 49, 97–113. Lintner, J. (1969), ‘The aggregation of investors’ diverse judgments and preferences in pure competitive markets’. Journal of Financial and Quantitative Analysis 4, 347–400. Lobe, S. and A. H¨lzl (2008), ‘Happy savers, happy issuer: The UK o lottery bond’. Center of Finance, University of Regensburg, Forthcoming Revue Bancaire et Financi`re - Bank- en Financiewezen J.E. e Loomes, G. and R. Sugden (1982), ‘Regret theory: An alternative theory of rational choice under uncertainty’. Economic Journal 92, 805–824. Lopes, L. (1987), ‘Between hope and fear: The psychology of risk’. Advances in Experimental Social Psychology 20, 255–295. Lopes, L. L. and G. C. Oden (1999), ‘The role of aspiration level in risk choice: A comparison of cumulative prospect theory and SP/A theory’. Journal of Mathematical Psychology 43, 286–313. Loughran, T. and J. Ritter (1995), ‘The new issues puzzle’. Journal of Finance 50, 23–51. Loughran, T. and J. Ritter (2002), ‘Why don’t issuers get upset about leaving money on the table in IPOs?’. Review of Financial Studies 15, 413–443. Loughran, T., J. Ritter, and K. Rydqvist (1994), ‘Initial public offerings: International insights’. Pacific-Basin Finance Journal 2, 165– 199. Loughran, T. and A. Vijh (1997), ‘Do long-term shareholders benefit from corporate acquisitions?’. Journal of Finance 52, 1765–1790. Malmendier, U. and G. Tate (2005), ‘CEO overconfidence and corporate investment’. Journal of Finance 60, 2661–2700. Malmendier, U. and G. Tate (2008), ‘Who makes acquisitions? CEO overconfidence and the market’s reaction’. Journal of Financial Economics 89, 20–43. Marsh, P. (1982), ‘The choice between equity and debt: An empirical study’. Journal of Finance 37, 121–144.

‘It is whether you win or lose: The importance of the overall probabilities of winning or losing in risky choice’. Mottola. C. ‘Speculation. 5–19. (2005). S. and E. 1279–1298. Nagel.178 References McConnell. Schwartz (1992). T. ‘The inattentive participant: Portfolio trading behavior in 401(k) plans’. Journal of Finance 53. J. American Economic Review 92. 1775–1798. and E. (2003). (1992). Moskowitz. T. and divergence of opinion’. L. Miller. Yamaguchi (2006). and C. Working Paper. T. Myers (1981). and profit when all traders are above average’. Frontline television documentary. ‘Do investors trade too much?’. (2005). Merton. . Journal of Monetary Economics 15. T. ‘Are investors reluctant to realize their losses?’. K. and T. O. Mitchell. S. Working Paper. 373– 413. 1151–1168. R. J. (1998a). 40–47. ‘Short sales institutional investors and the crosssection of stock returns’. PBS (2009). and investment: Theory and empirical evidence’. E. Payne. Journal of Financial Economics 78. Santa Monica. (1999). Journal of Finance 47. G. S. E. uncertainty. Journal of Economic Theory 3. 1887–1934. Odean. Odean.. Merrow. (1971). Journal of Finance 53. ‘Volume. Vissing-Jorgensen (2002). MIT. and A. Journal of Applied Corporate Finance pp. Understanding Cost Growth and Performance Shortfalls in Pioneer Process Plants. 1661–1699. (1977). ‘The returns to entrepreneurial investment: A private equity premium puzzle?’. Muelbroek. 277–309.. Journal of Finance 32. ‘The equity premium: A puzzle’. ‘The origin of LYONs: A case study in financial innovation’. Odean. American Economic Review 89. volatility. ‘An empirical analysis of illegal insider trading’. 745–778. 145–161. Mehra. (1998b). Prescott (1985). Utkus. R. Wharton. overpricing. ‘The warning’. ‘Risk. CA: Rand. The Journal of Risk and Uncertainty 30. Phillips. Panageas. price. ‘Optimum consumption and portfolio rules in a continuous-time model’.

Polkovnichenko. Washington. ‘Household portfolio diversification: A case for rank-dependent preferences’. and D. Puri. (2008). Sapienza (2004). Richardson. M. Rieger. U. ‘Clearly irrational financial market behavior: Evidence from the early exercise of exchange traded stock options’. and F. Rather. Review of Financial Studies 18. Working paper. 223–253. Journal of Finance 58. University of Pennsylvania. D. ‘Glamour. (2009). Polk. 10563. Puri. 851–876. Posner. A Failure of Capitalism: The Crisis of ’08 and the Descent into Depression. and R. A. ‘Who are entrepreneurs and why do they behave that way?’. Cambridge. value and the postacquisition performance of acquiring firms’. 37–70. Dan Rather Reports. Journal of Law and Economics 19.References 179 Peltzman. Ravenscraft. Poteshman. ‘External financing and future stock returns’. and T. Dublin City University Business School. and Economic Efficiency. Duke University. . 1693–1745. V. 71–99. Sloan (2003). overreaction and increasing misreaction to information in the options market’. Working Paper. (2006). Poteshman. Journal of Financial Economics 49. NBER working paper no. ‘Probability misestimation and preferences in financial investment decision’. (2002). and T. and P. P. (2001). A. Serbin (2003). 1467–1501. ‘Toward a more general theory of regulation’. MA: Harvard University Press. DC: Brookings Institution. R. Rau. ‘Underreaction. A. S. ‘The nature and persistence of buyback anomalies’. University of Zurich. Review of Financial Studies 22. C. February 17. V. and V. and D. Vermaelen (2009). Vermaelen (1998). Peyer. S. Episode Number: 407. Journal of Finance 56. 211–240. M. D. (1976). Robinson (2007). (2009). ‘The real effects of investor sentiment’. Working paper. Mergers. M. Journal of Financial Economics 86. Scherer (1987). Poti. ‘Optimism and economic choice’. ‘Sold!’. Robinson (2008). ‘The coskewness puzzle in the cross-section of industry portfolio returns’. Sell-Offs. Working Paper.

Scheinkman. N. Neoclassical Finance. Seru. J. Stoffman (2009). R. Xiong (2003). J. In: Encyclopedia of Quantitative Finance. H. W. H. and P..180 References Roll. Wiley. 7–59. Scarpetta. Burr Ridge: McGraw-Hill. L.): . and R. (2008a). coping and health: Assessment and implications of generalized outcome expectancies’. ‘Empirical pricing kernels’. Rosenberg. K. J. 197–216. (2007). ‘Stock prices and Wall Street weather’. 1183–1219. Journal of Political Economy 111. House (1977). F. M. Samuelson. Shumway. T. M. 341–372. S. and R. and N. T. Boston: Elsevier Academic Press. 1337–1345. (1992). ‘The hubris hypothesis of corporate takeovers’. H. A Behavioral Approach to Asset Pricing. ‘Overconfidence and speculative bubbles’. The Quarterly Journal of Economics 107. ‘Learning by trading’. Carver (1985). Health Psychology 4. Behavioral Corporate Finance. Zeckhauser (1988). H. Shefrin. and C. ‘Behavioral corporate finance’. 2nd edition. (1986). Ross. 219–247. ‘Behavioral explanations for why individual investors find dividends attractive’. Journal of Financial Economics 64. Shefrin. In: H. and J.. Tressel. Engle (2004). ‘Why does aggregate insider trading predict future stock returns?’. Hemmings. Woo (2002). (2005). Journal of Experimental Social Psychology 13. ‘Behavioral portfolio selection’. (2008b). 113–124. Green. (2001). Ross.. Shefrin. P. Shefrin. (1993). Review of Financial Studies. forthcoming. ‘Status quo bias in decision making’. Scheier. Seyhun. S. A. Journal of Applied Corporate Finance 14. Shefrin. American Economic Review 83. S. ‘Optimism. ‘The role of policy and institutions for productivity and firm dynamics: Evidence from micro and industry data’. 1303–1331. Baker (ed. D. forthcoming. and W. H. NJ: Princeton University Press. E. 279–301. (2005). Princeton. Saunders. OECD working paper. H. Journal of Risk and Uncertainty 1. ‘The ‘false consensus effect’: An egocentric bias in social perception and attribution processes’. Journal of Business 59.

): Understanding the Global Financial Crisis. 123–134. Shiller. Statman (1993a). Thaler (1988). Fairness. Statman (1985). Shefrin. Journal of Financial and Quantitative Analysis 35. 777–790. New York. (1981). Financial Management 22. In: L. Statman (1993b). H. ‘The volatility of long-term interest rates and the expectations models of the term structure’. ‘The disposition to sell winners too early and ride losers too long: theory and evidence’. Efficiency. Journal of Financial and Quantitative Analysis 29. Shefrin. and M. 323–349. fairness and efficiency in financial markets’. ‘Behavioral capital asset pricing theory’. 21–29. 29–42. 127–151. (2008c). Statman (2000). Statman (1992). Ethics. . Blackwell. Shefrin. H. Charlottesville. and M. ‘Explaining investor preference for cash dividends’. Economic Inquiry 24. Shefrin. and M. Journal of Finance 40. (2009). and M. 1190–1219. 421–436. Financial Analysts Journal pp. Shefrin. ‘Insights into the psychological profiles of entrepreneurs’. Shefrin. H. H. Shiller. and M. ‘Behavioral portfolio theory’. and M. and R. Shefrin. ‘Behavioral aspects of the design and marketing of financial products’. Shefrin. the Blackwell Companion to Dividends and Dividend Policy. Journal of Investment Management 7. Yazdipour (ed. forthcoming. (2010). (1978). and M.): Advances in Entrepreneurial Finance. NY: Springer.References 181 Chapter 13. Shefrin. 4–22. H. Journal of Investment Management 6. 609–643. VA: CFA Institute Publications. H. Shefrin. ‘Striking regulatory irons while hot’. In: R. R. ‘How psychological pitfalls generated the global financial crisis’. ‘The behavioral life cycle hypothesis’. Statman (1984). H. ‘Do stock prices move too much to be justified by subsequent changes in dividends?’. Siegel (ed. H. and M. 253–282. Shefrin. and Financial Markets. Journal of Political Economy 87. Statman (2009). R. H. Statman (1994). H. H. American Economic Review 71. ‘Ethics. ‘Risk and return in behavioral SDF-based asset pricing models’. H. Journal of Financial Economics 13. Shefrin.

G. The Accounting Review 71. 40–46. R. ‘Lottery players: Stock traders’. Amsterdam: Elsevier. 779–799. Journal of Financial and Quantitative Analysis 41. Journal of Financial Planning pp. Shiller. Smith. (2002). Taylor. Sorescu. ‘Constructivist and ecological rationality in economics’. Hanna (1989). R. Journal of Economic Behavior and Organization 12. A. P. Taylor and M. ‘Psychological study of human judgment: Implications for investment decision making’. 139–168. (2008). R. In: J. Subrahmanyam (2006). ‘Pooling of unshared information in group decision making: Biased information sampling during discussion’. (2008). 526–537. Pound (1989). M. Shleifer. 14–21. American Economic Review 93. A.. M. 46–66. M. . Titus (1985). Slovic. Shleifer. Statman. Sloan. (1996). and W. Financial Analysts Journal 58. (2003). Stasser.and six-person groups’. 289–316. (1999). Journal of Finance 27. ‘Probability weighting and employee stock options’. Inefficient Markets. (2000). Irrational Exuberance. 465–507. and R. and C. Journal of Personality and Social Psychology 57. Statman. Working paper. Journal of Financial Economics 70. University of Mannheim. ‘Human behavior and the efficiency of the financial system’. Shiller. Princeton: Princeton University Press. (2005). ‘The cross section of analyst recommendations’. Vishny (2003). 67–78. (1974). L. ‘The expressive nature of socially responsible investors’. G. ‘Do stock prices fully reflect information in accruals and cash flows about future earnings?’. V. 295–312. ‘Self-monitoring of expressive behavior’. ‘Stock market driven acquisitions’. and A. S. (1972). Snyder.182 References Shiller. ‘Survey evidence on the diffusion of interest and information among investors’. ‘Information sampling in structured and unstructured discussions of three. O. Journal of Personality and Social Psychology 48. 1467–1478. Oxford: Oxford University Press. and J. R.): Handbook of Macroeconomics. Woodford (eds. Spalt. Journal of Personality and Social Psychology 30. Stasser.

Thaler.. M.References 183 Statman. Advances in Behavioral Finance: Volume I. Welch. R. R. Review of Financial Studies 19. 1011–1022. ‘Earnings management and the long-run market performance of initial public offerings’. Tyebjee (1985). 429–455. Stein. 27–33. M. 12–29. The Bell Journal of Economics and Management Science 2. Teoh. (2007). (1993). 7–15. Journal of Business 69. (2009). Johnson (1990). Financial Management pp. Teoh. (1989). ‘Optimistic capital budgeting forecasts: An experiment’. I. 14631. Wong (1998b). H. M. Financial Analysts Journal 65.. R. Journal of Finance 53. (2007). Welch. (1996). ‘The theory of economic regulation’. A. ‘Investor overconfidence and trading volume’. The Black Swan. M. J. and J. Journal of Finance 44. ‘Project termination announcements and the market value of the firm’. Thaler. ‘The financial crisis and the policy responses: An empirical analysis of what went wrong’. 1935–1974. (2009). Sullivan. Taylor. Statman. N. ‘Governance: Travel and destinations’. ‘Applying behavioral finance to capital budgeting: Project termination’. Stein. ‘Earnings management and the underperformance of seasoned equity offerings’. Wong (1998a). 1–8. 22–31. I. ‘Regulating financial markets: Protecting us from ourselves and others’. J. ‘Overreactions in the options market’. Journal of Financial Economics 50. Financial Analysts Journal 65. G. 643–660. Financial Management pp. and T. Caldwell (1987). . Taleb. J. and K. Subrahmanyam. European Financial Management 14(1). Statman. ‘Behavioural finance: A review and synthesis’. S. 3–21. New York: Russell Sage Foundation. and D. Management Science 36. NBER Working Paper No. 63–99. and T. (1971). H. M. J. J. and E. and T. ‘Rational capital budgeting in an irrational world’. Financial Management pp. (2009). Statman. Stigler. Vorkink (2006). Sepe (1989). S. S. 1531– 1565.. Thorley. ‘Gambling with the house money and trying to break even: The effects of prior outcomes on risky choice’. 6–10. New York: Random House. Statman.

184 References Titman. 297–323. ‘Information uncertainty and analyst forecast behavior’. Journal of Personality and Social Psychology 39. Menlo Park. A. N. ‘Deceit and Self-deception’. and D. (1980). (1991). F. Wadhwa. Wang. Keil. X. Trivers. A. Wang. Weinstein. ‘The bounded rationality bias in managerial valuation of real options: Theory and evidence from IT projects’. The Anatomy of an Entrepreneur Family Background and Motivation. Ahluwalia (2007). Journal of Financial and Quantitative Analysis. M. Hens (2009). J. ‘International test on risk attitudes’. Decision Sciences 38. K. Washington D. (2006). 1063–1070. Todd. and A. Ewing Marion Kauffman Foundation of Entrepreneurship. A. Journal of Business 73. Tversky. and P. Salkever (2009). 111–131. J. Tiger (eds. Koehler (1994). and D.. Social Evolution. R. ‘Capital investments and stock returns’. Benjamin/Cummings. Clark (1988).. Review of Financial Studies. C. Psychological Review 101.): Man and Beast Revisited. P. Wei. Robinson and L. M. Kahneman (1992). D.. Xie (2004). M. Yan (2009). Welch. (1985). Working paper. V. 105–136. ‘Simple heuristics that make us smart’. and F. 806–820. Zhang. Journal of Personality and Social Psychology 54. ‘Views of financial economists on the equity premium and on professional controversies’. Rieger.C. Journal of Finance 61. K. ‘Development and validation of brief measures of positive and negative affect: The PANAS scales’. University of Zurich. Behavioral and Brain Sciences 23. S. and H.: Smithsonian Press. 157–181. J. Xiong. 727–780. Tversky. ‘Support theory: A nonextensional representation of subjective probability’. Holly. forthcoming. and L. 501–537. . R. Journal of Risk and Uncertainty 5. ‘Advances in prospect theory: Cumulative representation of uncertainty’. Aggarwal. Watson. (2000). 547–567. ‘Unrealistic optimism about future life events’. I. and T. W. Gigerenzer (2000). Tiwana.. ‘Heterogeneous expectations and bond markets’. and G. International Review of Finance 9. In: R. R. Trivers.

Sign up to vote on this title
UsefulNot useful