Munich Personal RePEc Archive

Investor Psychology and Asset Pricing

Hirshleifer, David

Fisher College of Business, The Ohio State University

10 February 2001

Online at https://mpra.ub.uni-muenchen.de/5300/

MPRA Paper No. 5300, posted 13 Oct 2007 UTC

February 9, 2001

Investor Psy hology and Asset Pri ing

The paper begins with an unnumbered, unheadered introdu tory se tion.Then:

Contents

I Judgment and De ision Biases 6I.1 Heuristi Simpli� ation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9

I.1.1 Attention/Memory/Ease-of-Pro essing E�e ts . . . . . . . . . . . 9I.1.2 Narrow Framing/Mental A ounting/Referen e E�e ts . . . . . . 10I.1.3 The Representativeness Heuristi . . . . . . . . . . . . . . . . . . 12I.1.4 Belief Updating: Combining E�e ts . . . . . . . . . . . . . . . . . 13

I.2 Self-De eption . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16I.3 Emotions and Self-Control . . . . . . . . . . . . . . . . . . . . . . . . . . 18

I.3.1 Distaste for Ambiguity . . . . . . . . . . . . . . . . . . . . . . . . 18I.3.2 Mood and De isions . . . . . . . . . . . . . . . . . . . . . . . . . 19I.3.3 Time Preferen e and Self-Control . . . . . . . . . . . . . . . . . . 19

I.4 So ial Intera tions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20I.5 Modelling Alternatives to Expe ted Utility and to Bayesian Updating . . 22

II Eviden e of Risk and Mispri ing E�e ts 23II.1 Predi tability of Se urity Returns . . . . . . . . . . . . . . . . . . . . . . 23

II.1.1 Predi tability Based upon Fa tor Risk Measures . . . . . . . . . . 24II.1.2 Predi tability Based upon Pri e and Ben hmark Value Measures . 24II.1.3 Predi tability Based upon Past Returns: Momentum and Reversal 26II.1.4 Predi tability Based upon Publi Versus Private News Events . . 27II.1.5 Predi tability Based upon Mood Proxies . . . . . . . . . . . . . . 29

II.2 Equity Premium and Riskfree Rate Puzzles . . . . . . . . . . . . . . . . . 30II.3 A tions Possibly Taken in Response to Mispri ing . . . . . . . . . . . . . 30II.4 A tions Possibly Taken to Create Misvaluation . . . . . . . . . . . . . . . 30II.5 Quality of Information Aggregation . . . . . . . . . . . . . . . . . . . . . 31II.6 Investor Behavior . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31

IIIAsset Pri ing Theories Based on Investor Psy hology 32III.1 Stati Asset Pri ing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33III.2 Dynami Asset Pri ing . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35

III.2.1 Pure Noise Trading . . . . . . . . . . . . . . . . . . . . . . . . . . 35III.2.2 Positive Feedba k Trading . . . . . . . . . . . . . . . . . . . . . . 36III.2.3 Mistaken Beliefs . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37III.2.4 Alternative Preferen es . . . . . . . . . . . . . . . . . . . . . . . . 40III.2.5 Evolving Populations . . . . . . . . . . . . . . . . . . . . . . . . . 42

III.3 Empiri ally Distinguishing Pri ing Theories . . . . . . . . . . . . . . . . 43III.3.1 Distinguishing Explanations for Size and Value E�e ts . . . . . . 43III.3.2 Distinguishing Explanations for Post-Event Continuation . . . . . 44III.3.3 Distinguishing Explanations for Momentum and Reversal . . . . . 45

IVCon lusion 46

2

February 9, 2001

Investor Psy hology and Asset Pri ing

David Hirshleifer�

This survey was written for presentation at the Ameri an Finan e Asso iation AnnualMeetings in New Orleans, LA, January, 2001. I espe ially thank the editor, GeorgeConstantinides, for valuable omments and suggestions. I also thank the dis ussant,Ni holas Barberis, Franklin Allen, Robert Bloom�eld, Mi hael Brennan, Markus Brun-nermeier, Joshua Coval, Ming Dong, Kent Daniel, Ja k Hirshleifer, Harrison Hong, RaviJagannathan, Narasimhan Jegadeesh, Andrew Karolyi, Charles Lee, Seongyeon Lim,Deborah Lu as, Rajnish Mehra, Jayanta Sen, Tyler Shumway, Ren�e Stulz, AvanidharSubrahmanyam, Siew Hong Teoh, Sheridan Titman, Yue Wang, Ivo Wel h, and parti i-pants of the Di e Finan e Seminar at Ohio State University for very helpful dis ussionsand omments.

�Fisher College of Business, The Ohio State University, 740A Fisher Hall, 2100 NeilAvenue, Columbus, OH 43210-1144; (614) 292-5174; hirshleifer.2�osu.edu

Investor Psy hology and Asset Pri ing

Abstra t

The basi paradigm of asset pri ing is in vibrant ux. The purely rational approa h

is being subsumed by a broader approa h based upon the psy hology of investors. In

this approa h, se urity expe ted returns are determined by both risk and misvaluation.

This survey sket hes a framework for understanding de ision biases, evaluates the a

priori arguments and the apital market eviden e bearing on the importan e of investor

psy hology for se urity pri es, and reviews re ent models.

The best plan is : : : to pro�t by the folly of others.

| Pliny the Elder, from John Bartlett, omp. Familiar Quotations, 9th ed. 1901.

In the muddled days before the rise of modern �nan e, some otherwise-reputable

e onomists, su h as Adam Smith, Irving Fisher, John Maynard Keynes, and Harry

Markowitz, thought that individual psy hology a�e ts pri es.1 What if the reators of

asset pri ing theory had followed this thread? Pi ture a s hool of so iologists at the Uni-

versity of Chi ago proposing the De� ient Markets Hypothesis: that pri es ina urately

re e t all available information. A brilliant Stanford psy hologist, all him Bill Blunte,

invents the Deranged Anti ipation and Per eption Model (or DAPM), in whi h prox-

ies for market misvaluation are used to predi t se urity returns. Imagine the euphoria

when resear hers dis overed that these mispri ing proxies (su h as book/market, earn-

ings/pri e, and past returns), and mood indi ators su h as amount of sunlight, turned

out to be strong predi tors of future returns. At this point, it would seem that the

de� ient markets hypothesis was the best- on�rmed theory in the so ial s ien es.

To be sure, dissatis�ed pra titioners would have omplained that it is harder to a tu-

ally make money than ivory tower theorists laim. One an even imagine some a ademi

hereti s do umenting rapid short-term sto k market responses to news arrival in event

studies, and arguing that se urity return predi tability results from rational premia for

bearing risk. Would the old guard surrender easily? Not when they ould appeal to in-

tertemporal versions of the DAPM, in whi h mispri ing is only orre ted slowly. In su h

a setting, short-window event studies annot un over the market's ineÆ ient response to

new information. More generally, given the strong theoreti al underpinnings of market

ineÆ ien y, the rebels would probably have an uphill �ght.

This alternative history suggests that the traditional view that �nan ial e onomists

have had about the rationality of asset pri es was not as inevitable as it may seem.

Despite many empiri al studies, s holarly viewpoints on the rationality of asset pri ing

have not onverged. This is probably a result of strong prior beliefs on both sides. On

one side, strong priors are re e ted in the methodologi al laim that we should adhere to

1Smith analyzed how the `overweening on eit' of mankind aused labor to be underpri ed in moreenterprising pursuits. Young workers do not arbitrage away pay di�erentials be ause they are proneto overestimate their ability to su eed. Fisher wrote a book on money illusion; in The Theory of

Interest ((1930), h. 21, pp. 493-94) he argued that nominal interest rates systemati ally fail to adjustsuÆ iently for in ation, and explained savings behavior in relation to self- ontrol, foresight, and habits.Keynes (1936) famously ommented on animal spirits in sto k markets. Markowitz (1952) proposedthat people fo us on gains and losses relative to referen e points, and that this helps explain the pri ingof insuran e and lotteries.

1

rational explanations unless the eviden e ompels reje tion; and in the use of the term

`risk premium' inter hangeably with `mean return in ex ess of the riskfree rate'. For

those on the opposite side, risk often omes quite late in the list of possible explanations

for return predi tability.

Often advo ates of one approa h or the other have ast the �rst stone out the door of

their own glass house. There is in fa t a notable parallelism among obje tions to the two

approa hes, illustrated in orresponding fashion in Table 1. (Lining up ea h obje tion

with its ounterpart does not imply parity in the validity of the arguments.)

This survey assesses the theory and eviden e regarding investor psy hology as a

determinant of asset pri es. This issue is at the heart of a grand debate in �nan e

spanning the last two de ades. In the last few years, �nan ial e onomists have grown

more re eptive to imperfe t rational explanations. Over time I believe that the purely

rational paradigm will be subsumed by a broader psy hologi al paradigm that in ludes

full rationality as a signi� ant spe ial ase.

Two superb re ent presentations of the asset pri ing �eld (Campbell (2000), Co hrane

(2000)) emphasize obje tive external sour es of risk. As Campbell puts it, \... asset

pri ing is on erned with the sour es of risk and the e onomi for es that determine the

rewards for bearing risk." For Co hrane, \The entral task of �nan ial e onomi s is to

�gure out what are the real risks that drive asset pri es and expe ted returns."

In ontrast, I argue here that the entral task of asset pri ing is to examine how

expe ted returns are related to risk and to investor misvaluation. Campbell's survey

emphasizes the stability of the �nan e paradigm over the last two de ades. I will argue

that the basi paradigm of asset pri ing is in vigorous and produ tive ux.

Figure 1 illustrates stati asset pri ing (analogous to the CAPM) when investors

misvalue assets and se urities. Returns are in reasing with risk (measured here by

CAPM beta) and with urrent market undervaluation of the asset. There are several

potential noisy proxies for the degree of underpri ing, su h as pri e- ontaining variables

(e.g., book/market, market value, earnings/pri e), measures of publi mood (e.g., the

weather), or a tions possibly taken to exploit mispri ing (e.g., re ent o uren e of a

sto k repur hase, insider pur hases). Risk and mispri ing e�e ts do not ne essarily take

su h a simple linearly separable form (see the models des ribed in Se tion IV), but it is

still useful to keep the two notions on eptually distin t.

This pi ture is only a starting point. Just as the stati risk e�e ts of the CAPM have

been generalized to intertemporal asset pri ing, so the dynami behavior of mispri ing

must be a ounted for as well. After de ades of study, the sour es of risk premia in purely

2

rational dynami models are well understood. In ontrast, dynami psy hology-based

asset pri ing theory is in its infan y.

In the remainder of the introdu tion, I dis uss market for es that an maintain or

to eliminate mispri ing, and why we annot dismiss mispri ing on on eptual grounds.

Se tion I of the survey presents relevant psy hologi al biases, and argue that many of

the important biases grow naturally from just a few deep roots. Se tion II summa-

rizes eviden e on apital market and investor behavior regarding the importan e of risk

and misvaluation e�e ts. Se tion III presents asset pri ing theories based on imperfe t

rationality. Se tion IV on ludes with further dire tions for resear h.

To think about whether mispri ing is viable, onsider the traditional argument for

rational pri e-setting. In this a ount, smart traders spot dollar bills lying on the ground

and grab them, whi h does away with mispri ing. Setting aside the dynami s of wealth

momentarily, the arbitrage story is in omplete in two ways. First, equilibrium pri es

re e t a weighted average of the beliefs of the rational and irrational traders.2 So long

as ea h group has signi� ant risk-bearing apa ity, both in uen e pri es signi� antly.

Arbitrage is a double-edged blade: just as rational investors arbitrage away ineÆ ient

pri ing, foolish traders arbitrage away eÆ ient pri ing. Se ond, in some respe ts all

investors may be imperfe tly rational. Even in the Olympi s, no one runs at the speed

of light; some ognitive tasks are too hard for any of us.

The traditional argument further asserts that wealth ows from foolish to wise in-

vestors. This point arries onsiderable weight. Suppose that some rational individuals

are immune from bias, and that all markets are liquid. Suppose that terminal dividends

obey a linear fa tor model with K systemati and N idiosyn rati payo� omponents (I

will all these systemati and idioysyn rati `fa tors'). An irrational investor on average

trades and loses on every fa tor that he misvalues. If the number of fa tors N + K is

large, and if a nontrivial fra tion of them are substantially mispri ed, then on average

irrational investors lose a very large amount of money almost surely. Soon superior

rationality will prevail.

Thus, as long as some investors are rational and markets are perfe t, there an be

substantial mispri ing in only a small fra tion of the N + K fa tors. If N >> K,

then some or all of the systemati fa tors an be substantially mispri ed, but only a

small fra tion of the N idiosyn rati omponents an be (see Daniel, Hirshleifer, and

Subrahmanyam (2001a)).

2See, e.g., Campbell and Kyle (1993), DeLong, Shleifer, Summers, and Waldmann (1990a), Figlewski(1978), Shefrin and Statman (1994), and Shiller (1984).

3

On the other hand, people are likely to be more prone to bias in valuing se urities

for whi h information is sparse. This suggests that misper eptions are strongest in the

dusty, idiosyn rati orners of the market pla e. One way to re on ile both intuitions

is to re ognize that there are biases that almost noone is immune to. In this ase there

an be widespread idiosyn rati mispri ing whi h only be omes apparent ex post.

Although misper eptions are probably most severe when information is sparse and ar-

rives slowly, there is no reason to think that onfusion is on�ned purely to idiosyn rati

fa tors. Market timers trade based on what they per eive to be superior information

about the market or about industry plays su h as high-te h. Investors (whether wisely

or not) pur hase ma roe onomi fore asts. So if investors sometimes misinterpret infor-

mation, they will make systemati as well as idiosyn rati errors. Indeed, to the extent

that misper eptions are onveyed through so ial pro esses, mistakes may be greatest for

systemati fa tors along with a few well-known se urities.

The fa t that several empiri al patterns of predi tability are strongest in small (pre-

sumably less liquid) �rms suggests that illiquid markets may be less eÆ ient. This is

less obvious than it sounds|the �ndings may result from the sparser information avail-

able about small, illiquid �rms. Sin e arbitrage is double-edged, holding wealth onstant

there is no presumption that liquidity immediately redu es mispri ing. It does, however,

speed the ow of wealth between between smart and foolish traders, whi h may in the

long run do so.3

It is often suggested that the expertise of hedge funds or investment banks will

improve arbitrage enough to eliminate any signi� ant mispri ing. This works if foolish

investors are wise enough to delegate to sound managers. However, intermediaries have

in entives to serve or exploit the irrationalities of potential lients. It is not obvious

that layering agen y over folly improves de isions.4 So misvaluation does not require

that there be fri tions or spe ial impediments to fund-raising by smart players. Su h

fri tions, however, an slow the ow of wealth between smart and foolish smart traders,

perhaps allowing mispri ing to persist longer.

When substantial mispri ing is limited to a few fa tors and residuals, less rational

3Liquidity makes it easier for smart traders to arbitrage away mispri ing, but also easier for foolishtraders to arbitrage away eÆ ient pri ing. Barber and Odean (1999) �nd that traders who swit hto online brokerages trade more aggressively yet subsequently perform poorly|their greater liquidityen ouraged bad trades.

4Furthermore, regardless of whether there are intermediaries, it is exa tly when a se urity or se torbe omes more mispri ed that smarter investors be ome poorer. This weakens rational arbitrage (andstrengthens irrational anti-arbitrage) in an untimely way (Shleifer and Vishny (1997), Kyle and Xiong(2000)).

4

investors do not ne essarily lose on average to wiser ones. Investors who underestimate

risk take larger long positions in risky assets, and thereby a hieve higher expe ted re-

turns (DeLong et al (1990a, 1991)). It ould further be argued that trading pressure

by irrational investors indu es ross-se tional return predi tability; that these investors

thereby lose money; but that on average they make up their losses by bearing more

aggregate market risk. However, if over on�dent investors irrationally overbuy the mar-

ket, this should result in a low expe ted return. This does not jibe well with the equity

premium puzzle of Mehra and Pres ott (1985).

There are other means by whi h imperfe tly rational individuals an earn high ex-

pe ted returns. Over on�dent investors who buy and sell aggressively in response to

valid private information signals may exploit liquidity traders more pro�tably than ra-

tional investors (Hirshleifer and Luo (2001)). In an imperfe tly ompetitive se urities

market, over on�dent traders an bene�t by intimidating ompeting informed traders

(Kyle and Wang (1997)).5 Over on�dent individuals are also likely to overinvest in

a quiring private information, at the expense of leisure.6

However, what eviden e we have suggests that aggressively trading individual in-

vestors do badly.7 Despite the ingenious explanations for pro�table foolishness, it is

quite plausible that in fa t fools and their money are soon parted. Even if so, a misper-

eption that derives from a fundamental human psy hologi al trait an remain important

for asset pri es in the long term. There are two related reasons.

First, wealth is reshu�ed in the pro ess of generational su ession. Se ond, in the

pro ess of getting ri h, individuals an learn to be less rational. For example, biased self-

attribution (Se tion I.2) auses individuals to attribute su esses to their own qualities

and failures to han e. As a result, losses by over on�dent individuals an be o�set by

the rising on�den e of the nouveau ri he (see Daniel, Hirshleifer, and Subrahmanyam

(1998), Gervais and Odean (2001)).

It is hallenging to �nd a sour e of risk to explain rationally the magnitude of ross-

se tional predi tability (see Se tion II). The hallenge for the mispri ing theory is to

explain how irrational investors an remain important while hemorrhaging a great deal

of ash. The disappearan e of the size e�e t in the mid-1980s and the in onsisten y of

5See also Benos (1998), Fis her and Verre hia (1999), and Wang (1998)).6Other means by whi h the imperfe tly rational an do well or poorly have been des ribed as well;

see Blume and Easley (1982, 1990, 2000), Palomino (1996), Luo (1998), and Sandroni (2000).7See Barber and Odean (1999, 2000a, 2000b) and Odean (1999). However, most of the theoreti al

models imply that only investors with moderate over on�den e will do well. The data may be pi kingup the poor performan e of the extremely over on�dent.

5

the value e�e t in the last few years is suggestive.8

There is a further problem. Having dete ted a return pattern statisti ally, it is hard

for an investor to know whether other investors have yet dete ted and a ted upon it. In

1984, how ould an investor be sure whether other investors were overexploiting the size

e�e t (Daniel and Titman (1999))? This un ertainty suggests that sometimes patterns of

mispri ing will be arbitraged away too slowly, and other times will overshoot. Con eiv-

ably the long life of the momentum e�e t has resulted from arbitrageurs ea h mistakenly

fearing that others have started to trade aggressively. As Yogi Berra ommented about

a popular restaurant: \No one goes there any more be ause it's too rowded."

The other possible reason for persistent mispri ing is that some relevant pie es of

publi information are ignored or misused by everyone. This an o ur either be ause

the signals are obs urely lo ated or be ause our shared model of the world is just not

sophisti ated enough to make their relevan e lear. A pri ing error of this sort may

disappear on e a smart e onometri ian identi�es it.

It is impossible to be omprehensive on a topi of this s ope. Several important

topi s have been dis ussed in greater depth elsewhere.9 My fo us is on the psy hology

of imperfe t rationality, not psy hologi al determinants of rational risk aversion or time

preferen e. My ben hmark for omparison throughout is the traditional asset pri ing

paradigm; I do not over market imperfe tions, nor models of rational bubbles.

I Judgment and De ision Biases

This se tion des ribes some psy hologi al e�e ts that are potentially relevant for se uri-

ties markets, with hints at possible explanations based upon adaptiveness.10 E onomists

have traditionally been skepti al of the varied array of seemingly arbitrary biases o�ered

8The U.S. small �rm e�e t was strongly positive every year during 1974-83, and then was negative forsix out of the next seven years; The two losing years of the millenium, whi h followed the publi ationof an important paper on \...Good News for Value Sto ks" (LaPorta et al (1997)) were the worst yearsfor value sto ks sin e 1928, though 2000 was better.

9I generally fo us on asset market regularities involving a time horizon of at least a month, and donot onsider seasonalities (for re ent eviden e see Hawawini, Keim, and Ziemba (2000)). Several surveysexamine the equity premium puzzle in greater depth (see, e.g., Campbell (1999, 2000), Ko herlakota(1996), and Mehra and Pres ott (2001)). Experiments in psy hology and e onomi s are surveyed inBossaerts (2000), Camerer (1995, 1998), and Hertwig and Ortmann (2001).

10See also the surveys of Camerer (1995), DeBondt and Thaler (1995), Rabin (1998), and Shiller(1999). There are also important literatures that build `fast and frugal' heuristi s based upon ex ante onsiderations [Gigerenzer, Todd, and ABC-Group (1999)℄, and that model the de ision onsequen esof bounds on rationality [Conlisk (1996)℄.

6

by experimental psy hology. The empiri al �ndings gain reden e if we an understand

what auses them. I argue here that these patterns generally derive from ommon roots.

Sin e time and ognitive resour es are limited, we annot analyze the data the envi-

ronment provides us with optimally. Instead, natural sele tion has designed minds that

implement rules of thumb (`algorithms', `heuristi s', or `mental modules') sele tively to

a subset of ues (see Simon (1956)). Su h heuristi s are e�e tive when applied to ap-

propriate problems. But their inevitable biases an be ome agrant when used outside

their ideal domain of appli ability.

E onomists often argue that errors are independent a ross individuals, and therefore

an el out in equilibrium. However, people share similar heuristi s, those that worked

well in our evolutionary past. So on the whole we should be subje t to similar biases.

Systemati biases ( ommon to most people, and predi table based upon the nature of the

de ision problem) have been on�rmed in a vast literature in experimental psy hology.

There is mu h debate about exa tly how good a job heuristi s do. While psy hologists

su h as Kahneman and Tversky have made lear that heuristi s an play a positive role,

in the last de ade, evolutionary (Darwinian) psy hologists have strongly emphasized the

adaptiveness of ognitive pro esses. In many ases biases diminish but do not vanish

when probabilities are reexpressed as numeri al frequen ies,11 and when problems are

posed in visual formats. However, there is no guarantee that �nan ial de ision problems

will be presented to individuals in a manner that favors the most a urate de isions.

The modern environment di�ers greatly from the prehistori environment of evo-

lutionary adaption for whi h human ognitive me hanisms were designed by natural

sele tion. Modern humans deal with new abstra tions su h as se urities, money, imper-

sonal markets, probabilities, and government; and with temptations su h as easy a ess

to fats and sugars, gambling asinos, and real-time internet trading.

The general fa t that ognitive resour e onstraints for e the use of heuristi s to

make de isions I will all heuristi simpli� ation. (For ognitive resour e onstraints,

read limited attention, pro essing power and memory.) A se ond sour e of bias arises

indire tly from ognitive onstraints. This is that natural sele tion probably did not

design human minds solely to make good de isions. Trivers (1985, 1991) dis usses evi-

den e that people annot perfe tly ontrol indi ators of their true internal states. This

reates sele tion for the ability to read subtle ues su h as fa ial expression, eye onta t,

posture, tone of voi e, and spee h tempo to infer the mental states of other individu-

11See Cosmides and Tooby (1996), Gigerenzer (1991, 1996), Gigerenzer and Ho�rage (1995), Kahne-man and Tversky (1996) and Tversky and Kahneman (1983)).

7

als. In Trivers' self-de eption theory, individuals are designed to think they are better

(smarter, stronger, better friends) than they really are. Truly believing this helps the

individual fool others about these qualities.

I argue that heuristi simpli� ation and self-de eption together provide a uni�ed

explanation for most of the judgment and de ision biases identi�ed in experimental

psy hology. This framework an provide guidan e as to whi h biases identi�ed in exper-

iments represent general me hanisms, and whi h are onditional side-e�e ts.12

Why don't people simply learn their way out of biased judgments? To some extent

they do. One barrier is that learning is just too hard. The other barrier arises from

self-de eption. Individuals who think they are already ompetent may be slow to adjust

their de ision pro edures (e.g., Einhorn and Hogarth (1978)).

Mu h of the eviden e des ribed here derives from experiments by e onomists and

psy hologists; their methods are somewhat di�erent. Finan ial e onomists are familiar

with riti isms of psy hologi al experiments: that the stakes are low, that subje ts

have little experien e with the experimental setting, that there is weak in entive to pay

attention or tell the truth, and that publi ation depends on �nding an e�e t. What may

not be as familiar is that there is data addressing these issues. On the whole training

and in reasing rewards and number of repetitions often redu es, but does not eliminate

biases. Lessons learned through repetition often do not arry over well a ross seemingly

similar tasks. The well-known biases have been subje ted to repli ation.13 Many (though

not all) of the ognitive biases are stronger for individuals with low ognitive ability or

skills than for those with high ability or skills, onsistent with biases being genuine errors

(see Stanovi h and West (2000)).

Subse tions I.1 and I.2 onsider individual biases organized by proposed auses

(heuristi simpli� ation and self-de eption). Subse tion I.3 onsiders emotion and self-

ontrol, Subse tion I.4 dis usses so ial intera tions, and Subse tion I.5 dis usses model-

ing alternatives to expe ted utility theory and to Bayesian updating.

12Explanations based upon ognitive adaptiveness are subje t to the obje tion that it is too easy to ome up with `just-so' stories that �t the data ex post. However, my goal here is not to make the ase that the eviden e supports the adaptiveness approa h (see Barkow, Cosmides, and Tooby (1992)).Rather, my point is that it is hard to make sense out of biases without a on eptual framework.Adaptiveness is about the only one we have.

13See, e.g., Camerer (1995), Rabin (1998) and Hertwig and Ortmann (2001).

8

I.1 Heuristi Simpli� ation

I.1.1 Attention/Memory/Ease-of-Pro essing E�e ts

Limited attention, memory, and pro essing apa ities for e a fo us on subsets of available

information. Un ons ious asso iations also reate sele tive fo us. In many studies,

priming subje ts with (possibly irrelevant) verbal information triggers asso iations that

in uen e judgments (see, e.g., Gilovi h (1981), Higgins (1996)).

Sele tive triggering of asso iations auses salien e and availability e�e ts (e.g., Kah-

neman and Tversky (1973)). An information signal is salient if it has hara teristi s

(e.g., di�ering from the ba kground or from a past state) that are good at hooking our

attention or at reating asso iations that fa ilitate re all. In the availability heuristi

(Tversky and Kahneman (1973)), items that are easier to re all are judged to be more

ommon. This generally makes sense, sin e things that are more ommon are noti ed or

reported more often, making them easier to remember. Shiller (2000b) suggested that

the ease with whi h regular Web users an think of examples relating to the internet

revolution en ouraged the market boom of the late 1990s.

One reason people are in uen ed by the the format of de ision problems is that

they annot perfe tly retrieve relevant information from memory (Tversky and Kahne-

man (1973), Pennington and Hastie (1988)). People underweight the probabilities of

ontingen ies that are not expli itly available for onsideration (Fis hho�, Slovi , and

Li htenstein (1978)). This suggests a kind of over on�den e (see Subse tion I.2), and

apparent market overrea tion when unforeseen ontingen ies do o ur.

A ording to self-per eption theory (Bem (1972)), \Individuals ome to know their

own attitudes, emotions and internal states by inferring them from observations of their

own behavior and ir umstan es in whi h they o ur." The need to infer an result from

memory loss, or from simple la k of a ess to un ons ious internal states. A tenden y

to form habits an be an optimal me hanism to address memory loss, re e ting an

impli it self-per eption that a tions taken before probably had a good reason (Hirshleifer

and Wel h (2000)). Habits also e onomize on thinking. Habits, in luding the habitual

adheren e to self-imposed rules an also play a role in self-regulation strategies (e.g,

onsume only out of dividends, not prin ipal; see Shefrin and Statman (1984), Thaler

and Shefrin (1981)).

The halo e�e t auses someone who likes one outstanding hara teristi of an in-

dividual to extend this favorable evaluation to the individual's other hara teristi s

(Nisbett and Wilson (1977a)). An analogous misattribution bias ould potentially ause

9

sto k market mispri ing. In an eÆ ient market, a sto k being good in terms of growth

prospe ts says nothing about its prospe ts for future risk-adjusted returns (whi h are on

average zero). If people mistakenly extend their favorable evaluation of a sto k's earn-

ings prospe ts to its return prospe ts, growth sto ks will be overpri ed (see Lakonishok,

Shleifer, and Vishny (1994), Shefrin and Statman (1995)).

Familiar signal ombinations (e.g., yellow with banana) are easier to per eive than

unfamiliar ones (Bruner, Postman, and Rodrigues (1951)). There is a strong and robust

mere exposure e�e t in whi h exposure to an unreinfor ed stimulus tends to make people

like it more (see, e.g., Bornstein and D'Agostino (1992), Moreland and Bea h (1992)).

The basis for this heuristi may be that what is familiar, being understood better,

is often less risky. However, this an be taken too far, as when people prefer to bet

on a matter about whi h they know more than another equivalent gamble (Heath and

Tversky (1991)). People also like similarity in hoi e of friends and mates (Bers heid and

Reis (1998)). A ording to evolutionary psy hology, people prefer familiar and similar

individuals be ause these were indi ators of geneti relatedness (e.g., Trivers (1985)).

These biases suggest a tenden y to prefer lo al investments (see also Huberman (1999)).

A literature in psy hology has examined how subje ts learn by observation over time

to predi t a variable that is sto hasti ally related to multiple ues (see, e.g., Krus hke and

Johansen (1999)). A pervasive �nding is that animals and people do not a hieve orre t

understanding of the orrelation stru ture. Instead, ue ompetition o urs: salient ues

weaken the e�e ts of less salient ones, and the presen e of irrelevant ues auses subje ts

to use relevant ues and base rates (un onditional frequen ies) less. There is also learned

utilization of irrelevant ues. Cue ompetition raises interesting questions about how

information ooding through the internet will a�e t misvaluation.

The learned usage of irrelevant ues omes lose to magi al thinking, the belief in

relations between ausally unrelated a tions or events (as with astrology and other su-

perstitions). A type of magi al thinking alled the illusion of ontrol onsists of the

belief that a person an favorably in uen e unrelated han e events. A possible exam-

ple example is that people value lottery ti ket numbers they sele t more than randomly

assigned ones (Langer (1975)).

I.1.2 Narrow Framing/Mental A ounting/Referen e E�e ts

Narrow framing (see Kahneman and Lovallo (1993), Read, Loewenstein, and Rabin

(1999)) involves analyzing problems in too isolated a fashion. This makes ex ellent

sense when time and ognitive resour es are limited. Many problems an be ompart-

10

mentalized safely. An impli ation is that the form of presentation of logi ally identi al

de ision problems, su h as the highlighting of a di�erent referen e for omparison of out-

omes an have large framing e�e ts on hoi es (Tversky and Kahneman (1981, 1986)).

Optimizing with respe t to a problem-spe i� referen e point, and having a dire t prefer-

en e over deviations (instead of over total onsumption) e onomizes on thinking. Money

illusion is another do umented example of sensitivity to irrelevant des ription features

(Sha�r, Diamond, and Tversky (1997)).

By using di�erent presentation or pro edures, experimenters an eli it preferen e

reversals. Fa ed with a hoi e between a binary lottery with a high probability but

relatively low maximum payo�, versus another with lower probability and higher maxi-

mum payo�, subje ts often tend to prefer the high probability lottery, yet pla e a higher

valuation on the high-maximum-payo� lottery!14 There are also ontext e�e ts, in whi h

the presen e of a non-sele ted hoi e alternative a�e ts whi h alternative is sele ted.

Mental a ounting (Thaler (1985)) is a kind of narrow framing that involves keeping

tra k of gains and losses related to de isions in separate mental a ounts, and to reex-

amine ea h a ount only intermittently when a tion-relevant. Mental a ounting may

explain the disposition e�e t (Shefrin and Statman (1985)), an ex essive propensity to

hold on to se urities that have de lined in value and to sell winners. Having observa-

tion of gains and losses trigger pleasant or unpleasant feelings seems a sensible mental

design to motivate pro�table a tions. Su h a me hanism may, however, be sidetra ked

when the individual avoids re ognizing losses. Self-de eption theory reinfor es this argu-

ment, be ause a loss is an indi ator of low de ision ability, and a self-de eiver maintains

self-esteem by avoiding re ognition of su h indi ators.

Related arguments an explain the house money e�e t (Thaler and Johnson (1990))|

a greater willingness to gamble with money that was re ently won. The unpleasantness

of a loss of re ently-won money may be diluted by aggregating it with the earlier gain.

An horing (Tversky and Kahneman (1974)) is the phenomenon that people tend

to be unduly in uen ed in their assessment of some quantity by arbitrary quantities

mentioned in the statement of the problem, even when the quantities are learly unin-

formative. Some re ent authors o�er and test possible explanations in whi h the pro ess

of evaluating the an hor makes an hor- onsistent arguments more a essible.15

A ording to expe ted utility theory, utility derives solely from the probability dis-

tribution of payo�s resulting from a hoi e. However, people seem to be regret averse in

14Li htenstein and Slovi (1971), Grether and Plott (1979), Tversky, Slovi , and Kahneman (1990).15Chapman and Johnson (1999), Mussweiler and Stra k (1999).

11

their hoi es (e.g., Josephs et al (1996), Ritov (1996)). They seem to be on erned not

just that a hoi e may lead to low onsumption, but that onsumption may be lower

than the out ome provided by an alternative hoi e.

An eÆ ient heuristi method of omparing de ision alternatives may be to line up and

ompare possible out omes by state of the world (rather than evaluating the expe ted

utility of ea h alternative separately and then omparing). Thus, having feelings be

triggered by omparison of out omes may be an e�e tive me hanism for motivating

good hoi es. Regret avoidan e may also re e t a self-de eption me hanism designed to

prote t self-esteem about de isionmaking ability (Josephs et al (1996)).

Regret is stronger for de isions that involve a tion rather than passivity (Kahneman

and Tversky (1982)), an e�e t sometimes alled the omission bias (Ritov and Baron

(1990)). Regret aversion an explain the endowment e�e t, a preferen e for people to

hold on to what they have rather than ex hange for a better alternative, as with the

refusal of individuals to swap a lottery ti ket for an equivalent one plus ash.16 The status

quo bias (Samuelson and Ze khauser (1988)) involves preferring the hoi e designated

as the default or status quo among a list of alternatives.

Loss aversion is the phenomenon that people tend to be averse even to very small risks

relative to a referen e point, suggesting a kink in the utility fun tion. This may result

from the ognitive eÆ ien y of mentally dis retizing ontinuous variables, as re e ted in

the use of terms like `gain,' `break even,' and `loss', whi h make the distin tion between

a gain and loss more salient.

I.1.3 The Representativeness Heuristi

The representativeness heuristi (Grether (1980), Kahneman and Tversky (1973), Tver-

sky and Kahneman (1974)) involves assessing the probability of a state of the world

based on the degree to whi h the eviden e is per eived as similar to or typi al of the

state of the world. Similarity an be viewed as an indi ator of the onditional probability

of the eviden e given the state of the world versus other states. However, a Bayesian also

takes into a ount heavily the prior probability of the out omes, whereas people tend to

underweight statements about un onditional population frequen ies in performing on-

ditional updating|base-rate underweighting. Furthermore, people's per eptions of how

`representative' a pie e of eviden e is of a state of the world may mat h its onditional

probability poorly. For example, people tend to rely too heavily on small samples (the

16See Bar-Hillel and Neter (1996), Kahneman, Knets h, and Thaler (1991), Knez, Smith, andWilliams (1985).

12

`law of small numbers') and rely too little on large samples, inadequately dis ount for the

regression phenomenon, and dis ount inadequately for sele tion bias in the generation or

reporting of eviden e.17 Representativeness e�e ts have been dete ted in experimental

markets (see Camerer (1995), Se tion II.C.4).

The idea that a sample should resemble the population is often orre t, espe ially in

a large unbiased independent sample. The pre eding errors amount to applying an infer-

en e too weakly within its realm of validity (large sample size) and too strongly beyond

its realm of validity (small sample size). This is a natural onsequen e of the tradeo�s

involved with the design of an eÆ ent heuristi . The resulting errors are not random:

here, the error is predi table based on the sample size. The law of small numbers suggests

that newly-popular theories about the market drawn from re ent investment experien e

may ause overrea tion.

Misunderstanding of how randomness works an also ause a phenomenon of gam-

bler's falla y. This is the belief that in an independent sample the re ent o uren e of

one out ome in reases the odds that the next out ome will di�er. In fa t people avoid

betting on a lottery number that was a winner sometime over the pre eding few days

(Clotfelter and Cook (1993)).

On the other hand, use of the representativeness heuristi an ause trend- hasing,

be ause people are to ready to believe that trends have systemati auses. Statisti ians

refer to the lustering illusion, wherein people per eive random lusters as re e ting a

ausal pattern. People mistakenly believe in `hot hands' among sports players even when

a tual performan e is very lose to serially independent (Gilovi h, Vallone, and Tversky

(1985)). In an experimental market, onsistent with gambler's falla y, Andreassen and

Kraus (1990) found that when exogenous pri es u tuate modestly, subje ts buy on dips

and sell on rises. However, when a trend appears subje ts do less of this tra king, and

possibly swit h to hasing trends. There is further eviden e from experiments and from

surveys that real estate and sto k market investors extrapolate trends in fore asting

pri e movements.18

I.1.4 Belief Updating: Combining E�e ts

Edwards (1968) identi�ed the phenomenon of onservatism, that under appropriate ir-

umstan es individuals do not hange their beliefs as mu h as would a rational Bayesian

17See, e.g., Brenner, Koehler, and Tversky (1996), GriÆn and Tversky (1992), Kahneman and Tver-sky (1973), Nisbett and Ross (1980) h.4 and referen es therein, and Tversky and Kahneman (1971).

18See DeBondt (1993), Case and Shiller (1990), and Shiller (1988).

13

in the fa e of new eviden e. The more useful the eviden e, the greater the shortfall

between a tual updating and rational updating.

Having a framework for assessing biases an help when they seem to on i t. For

example, onservatism implies underweighting of new eviden e. Yet if we view prior

beliefs as a base-rate, onservatism would seem to ontradi ts base-rate underweighting.

Perhaps onservatism is a onsequen e of an horing upon an initial probability estimate.

Yet the representativeness heuristi predi ts that people will extrapolate too strongly

from patterns in small samples, and salien e bias also auses people to overrea t to

ertain kinds of information. Whi h bias do we believe?

To resolve on i ts like this requires a fo us on underlying auses, and how they will

operate in a parti ular setting. For example, self-de eption an ause onservatism in

a stable environment, be ause an individual who has expli itly adopted a belief may

be relu tant to admit to himself that he made a mistake. On the other hand, if the

environment is volatile, there may be no dishonor in re ognizing that di�erent beliefs

are alled for.

One explanation for onservatism is that pro essing new information and updating

beliefs is ostly. There is eviden e that information that is presented in a ognitively

ostly form is weighed less: information that is abstra t and statisti al, su h as sam-

ple size and probabilisti base-rate information. Furthermore, people may overrea t to

information that is easily pro essed, i.e., s enarios and on rete examples.

The ostly-pro essing argument an be extended to explain base rate underweighting.

If an individual underweights new information re eived about population frequen ies

(base rates), then base rate underweighting is really a form of onservatism. Indeed,

base rates are underweighted less when they are presented in more salient form or in

a fashion whi h emphasizes their ausal relation to the de ision problem (see Koehler

(1996)). This ostly-pro essing-of-new-information argument does not suggest that an

individual will underweight his pre-existing internalized prior belief. On the other hand,

if base rate underweighting is a onsequen e of the use of the representativeness heuristi ,

there should be underweighting of priors.

GriÆn and Tversky (1992) suggest that base-rate underweighting and onservatism,

interpreted as under- versus over-rea tion to signals, an be understood as results of

ex essive relian e on the strength of information signals and underrelian e on the weight

of information signals. The strength of an information realization is how `extreme' the

eviden e is (in some sense), and the weight of eviden e is its reliability or pre ision.

For example, a large sample of onditionally i.i.d. signals has high weight. But if

14

the preponderan e of favorable over unfavorable signals is modest, it has low strength.

Conservatism arises when people rely too little on high weight eviden e su h as a long

sample, and base rate negle t when people rely too heavily on high-strength eviden e

su h as a few signals all in one dire tion.

In summary, di�erent experimental settings an lead to under- or over-relian e on new

signals; people seem to make judgments di�erently in di�erent situations (see Grether

(1992), Payne, Bettman, and Johnson (1992)). Given the di�erent possible e�e ts,

invoking the name of a bias does not provide ompelling support for assuming under-

or over-rea tion in a �nan ial model. Further support an be provided by omparing

the e onomi de ision environment of the model with the spe i� experimental de ision

setting in whi h the bias was do umented, and espe ially by running new experiments

that mat h losely the de ision environment in the �nan ial model.

Most studies of pri e fore asts �nd biased that is predi table using urrent observ-

ables. For example, fore asts are often found to be adaptive, i.e., they respond partially

to past fore ast errors.19 Su h biases are potentially onsistent either with Bayesian

learning with an unknown distribution, or with over on�den e. Experimental studies

involving a �xed distribution generally also yield biases, and fore asts are adaptive in

most fore ast experiments involving endogenously determined pri es as well (see Camerer

(1995), Se tion II.E). Consistent with over on�den e, fore asters seem to put too little

weight on the known fore asts of other fore asters (Bat helor and Dua (1992)).

Analyst fore asts of earnings are over-optimisti at long time horizons and pessimisti

at short horizons (e.g., Ri hardson, Teoh, and Wyso ki (1999)). Su h biases may ome

from misper eptions or from agen y in entives. However, we would normally expe t

rational agents to provide at least a positive in remental value in their fore asting a -

tivities. There is on i ting eviden e as to whether sto k market analysts' fore asts

of earnings do better or worse than a time-series fore ast (see the review of Kothari

(2000)). A large literature shows that real-world de isionmakers su h as PhD admission

ommittees or do tors do not predi t out omes as well as me hani al de ision rules based

on simple linear ombinations of obje tive input measures (see Camerer (1991)). This

suggests that the rise of arbitrage based upon modern statisti al analysis in se urities

markets will indeed redu e mispri ing.

19See the dis ussions in Lovell (1986) and Williams (1987), but see also Keane and Runkle (1990).There is a similar �nding for survey fore asts of ma roe onomi variables (e.g., Aggarwal, Mohanty,and Song (1995)).

15

I.2 Self-De eption

The self-de eption theory implies over on�den e, a very well-do umented bias (as re-

viewed, e.g., in Odean (1998b)).20 An extensive literature on alibration shows that

people believe their knowledge is more a urate than it really is.21 For example, their

predi tions of probabilities of events are too extreme (too high relative to the true fre-

quen y when they think the event probably will o ur, too low when they think it will

not). The on�den e intervals they provide for quantities are too narrow, e.g., 98% on�-

den e intervals ontain the true quantity only 60% of the time (Alpert and Rai�a (1982)).

Experts are well- alibrated in some ontexts but not others (see Camerer (1995) p. 592-

3). Experts an be more prone to over on�den e than non-experts when predi tability is

low and eviden e is ambiguous (GriÆn and Tversky (1992)). Over on�den e is greater

for hallenging judgment tasks, and individuals tend to be more over on�dent when

feedba k on their information or de isions is deferred or in on lusive.22

Over on�den e is sometimes reversed for very easy items (Li htenstein and Fis ho�

(1977)).23 Over on�den e implies overoptimism about the individual's ability to su eed

in his endeavors. Su h optimism has been found in a number of di�erent settings (Miller

and Ross (1975)). Men tend to be more over on�dent than women, though the size of

the di�eren e depends on whether the task is per eived to be mas uline or feminine.24

Sin e people fail more often than they expe t to, rational learning over time would

tend to eliminate over on�den e. So for self-de eption to su eed, nature must provide

me hanisms that bias the learning pro ess. This is onsistent with self-enhan ing biased

self-attribution. People tend to attribute good out omes to their own abilities, and

bad out omes to external ir umstan es.25 Over on�den e and biased self-attribution

are stati and dynami ounterparts; self-attribution auses individuals to learn to be

20Bernardo and Wel h (2000) provide an alternative theory of over on�den e based on group infor-mational bene�ts.

21See, e.g., Keren (1991), Li htenstein, Fis ho�, and Phillips (1982), M Clelland and Bolger (1994),and Yates (1990).

22Einhorn (1980), Fis hho�, Slovi , and Li htenstein (1977), GriÆn and Tversky (1992), Li htenstein,Fis ho�, and Phillips (1982), Yates (1990).

23This is not surprising on me hani al grounds; in the extreme ase of perfe t de ision a ura y, itis possible to be under- but not over- on�dent about a ura y. It has been suggested that apparentover on�den e ould be an artifa t of the hoi e of questions that are not a \representative sample of theknowledge domain" (e.g., Gigerenzer, Ho�rage, and Kleinbolting (1991)), but over on�den e remainswhen questions are randomly sele ted from the knowledge domain, and has been do umented in manypra ti al hoi e settings (GriÆn and Tversky (1992), Brenner et al (1996), Soll (1996)).

24Deaux and Emswiller (1974), Lenney (1977), Lundeberg, Fox, and Pun o har (1994).25Fis ho� (1982), Langer and Roth (1975), Miller and Ross (1975), and Taylor and Brown (1988).

16

over on�dent rather than onverging to an a urate self-assessment.

Self-de eption also explains why there are a tion-indu ed attitude hanges of the sort

that motivate the theory of ognitive dissonan e.26 In one experiment people who hose

between two produ ts downgraded their assessments of the one they did not pi k. In an-

other, women who had to exert greater e�ort to gain entry to a group subsequently liked

the group more. In other experiments, people who were indu ed with mild in entives or

by request to express opinions be ame more sympatheti to those opinions. A tenden y

to be ex essively atta hed to a tivities for whi h one has expended resour es, the sunk

ost e�e t, has been on�rmed in several ontexts (Arkes and Blumer (1985)). The self-

de eption theory suggests that a tenden y to adjust attitudes to mat h past a tions is

a me hanism designed to persuade the individual that he is a skillful de isionmaker (see

also Nel, Helmrei h, and Aronson (1969) and Steele and Liu (1983).

Similar reasoning an explain hindsight bias (e.g., Hawkins and Hastie (1990))| it

helps our self-esteem to think we `knew it all along'; and the phenomenon of rationalization|

onstru ting a plausible ex post rationale for past hoi es helps an individual feel better

about his de ision ompeten e. People are very ready to devise and apparently believe

their explanations for alleged fa ts about the world as well as themselves.27

People tend to interpret ambiguous eviden e in a fashion onsistent with their own

prior beliefs. They give areful s rutiny to in onsistent fa ts and explain them as due

to lu k or faulty data-gathering (see Gilovi h (1991) h.4). This on�rmatory bias an

help maintain self-esteem, onsistent with self-de eption. Exposure to eviden e should

tend to ause rational Bayesians with di�ering beliefs to onverge, whereas the attitudes

of experimental subje ts exposed to mixed eviden e tend to be ome more polarized

(e.g., Isenberg (1986), Lord, Ross, and Lepper (1979)). Forsythe et al (1992) �nd that

individuals more subje t to this on�rmation bias lose money in an experimental market

to those who are less subje t to it. Con�rmatory bias may ause some investors to sti k

to unsu essful trading strategies, ausing mispri ing to persist.

Some general biases toward on�rmation of hypotheses do not rely on self-de eption.

In evaluating hypotheses assigned by the experimenter about the relation of two kinds

of variables (e.g., studying the night before an exam, and getting a good grade), a large

literature �nds that people put too mu h weight on on�rming eviden e. This involves

fo using on ases in whi h both study and a good grade o urred, and negle ting other

26Festinger and Carlsmith (1959), Cooper and Fazio (1984) and Harmon-Jones and Mills (1999) h.1).27See, e.g., Gazzaniga (1988) pp.12-14, Nisbett and Wilson (1977b), Ross et al (1977).

17

information ( ases in whi h one but not the other o urred).28 It has been argued that

su h a bias is an eÆ ient short ut in many ontexts (Klayman and Ha (1987)).

People are also biased toward seeking on�rmatory information. In the famous Wa-

son task experiments (Wason (1966)), subje ts were asked to turn over ards to evaluate

a hypothesis. They often turned over ards whi h potentially ould provide instan es

onsistent with the hypothesis, and often left unturned ards that ould on lusively

reje t the hypothesis. A possible explanation is that positive ases are easier to pro-

ess ognitively. There is eviden e that people are more in uen ed by the information

re e ted in the o urren e of an event than the non-o urren e.29

I.3 Emotions and Self-Control

Emotions probably play a role in su h traditionally rational onsiderations as time and

risk preferen e, and in most or all of the e�e ts des ribed earlier. I dis uss some further

aspe ts of emotion here.

I.3.1 Distaste for Ambiguity

Choi es are in uen ed by the stru ture of gambles above and beyond the overall prob-

ability distribution of onsumption out omes that the gambles provide. The Ellsberg

paradoxes (Ellsberg (1961)) suggested that people are averse to ambiguity, ausing irra-

tional hoi es. Ambiguity aversion has been on�rmed in market experimental settings.

It seems to re e t a more general tenden y for emotions su h as fear to a�e t risky

hoi es (see Peters and Slovi (1996)). As suggested by Camerer (1995), ambiguity

aversion may in rease risk premia unduly when new �nan ial markets are introdu ed,

be ause of the layering of un ertainty about both the stru ture of the e onomi envi-

ronment and about resulting out omes. A possible explanation for ambiguity aversion

is that the obvious absen e of an identi�able parameter of the de ision problem may

often be asso iated with higher risk and the possibility of hostile manipulation. This

justi�es a fo us on missing information, but su h an heuristi an go astray when there

is no hostile manipulation. In a related vein, the eviden e of Heath and Tversky (1991)

indi ates that, holding probabilities onstant, people prefer gambles that give them a

sense of understanding or ompeten e.

28See e.g., Cro ker (1982), Fis hho� and Beyth-Marom (1983), and Jenkins and Ward (1965).29E.g., Newman, Wol�, and Hearst (1980), Nisbett and Ross (1980).

18

I.3.2 Mood and De isions

Risk aversion, regret aversion, and loss aversion may re e t a al ulated avoidan e of

unpleasant future feelings. However, mood and emotions felt today also a�e t risk tak-

ing. For example, sales of State of Ohio lottery ti kets were found to in rease in the days

following a football vi tory by Ohio State University (Arkes, Herren, and Isen (1988)).

More generally, people who are in good moods are more optimisti in their hoi es and

judgments than those in bad moods (see, e.g., Wright and Bower (1992)). Feelings a�e t

people's per eptions of and hoi es with respe t to risk (see, e.g., Mann (1992)). Bad

moods are asso iated with more detailed and riti al strategies of evaluating information

(Petty, Glei her, and Baker (1991)). The in uen e of mood and emotion on pur hase

plans and the e�e ts of advertising have been studied by marketing resear hers as well.30

A�e tive states (feelings or moods) ontain information that individuals an use

to draw inferen es about the environment.31 However, people often attribute arousal or

feelings to the wrong sour e, leading to in orre t judgments or misattribution biases (see,

e.g., Ross (1977)). For example, people feel happier on sunny days than on rainy days,

but priming them by asking them about the weather a�e ts their judgment of how happy

they are (S hwarz and Clore (1983)). Moods states tend to a�e t relatively abstra t

judgments more than spe i� ones about whi h people have on rete information.32

This suggests, for example, that if the weather in New York puts sto k market traders

in a bad mood, their pessimism may on ern long-term market growth prospe ts rather

than whether the Fed is going to lower interest rates next week.

I.3.3 Time Preferen e and Self-Control

The onventional representation of de isions over time has an additively separable utility

fun tion with exogenous, de lining exponential weights. However, eviden e from psy-

hology suggests that dis ount rates hange with ir umstan es. Deferring onsumption

involves self- ontrol, and is therefore related to mood and feelings. There is eviden e

that dis ount rates are sometimes remarkably high, that gains are dis ounted more heav-

ily than losses, that small magnitudes are dis ounted more heavily than large, that the

framing of a hoi e as a delay versus an advan e has a large e�e t on de isions, that

time preferen e di�ers greatly in di�erent de ision domains (e.g., money versus health),

30Barone, Miniard, and Romeo (2000), Cohen and Areni (1991), Erevelles (1998), Mano (1999).31See e.g., Clore, S hwarz, and Conway (1994), Wilson and S hooler (1991).32Clore, S hwarz, and Conway (1994), Forgas (1995).

19

and that vis eral in uen es su h as pain or hunger a�e t intertemporal hoi es.33

The exponential spe i� ation is time onsistent. However, experimental studies sug-

gest that people and non-human animals are time-in onsistent. Spe i� ally, they tend

to dis ount a deferral of onsumption from date t to t + 1 more heavily as date t ap-

proa hes, onsistent for example with a hyperboli form for dis ount rates.34 This auses

hoi e reversals even when no new information arrives. Hyperboli dis ounting has been

disputed.35 Nevertheless, re ent e onomi studies have applied time-in onsistent dis-

ounting to a wide range of issues in luding savings, liquidity premia and the equity

premium puzzle.36

I.4 So ial Intera tions

Finan ial e onomists have borrowed more from the psy hology of the individual than

from so ial psy hology. Finan ial theorists have examined how information is trans-

mitted by pri es, volume or orporate a tions. However, person-to-person and media

ontagion of ideas and behavior also seems important. People tend to onform with

the judgments and behaviors of others, as do umented in the famous length estima-

tion experiments of As h (1956). A meta-analysis of 133 related studies (Bond and

Smith (1996)) on�rmed the onformity e�e t, whi h is, however, history- and ulture-

dependent. There are rational informational reasons to learn by observing the a tions

of others.37 However, a fully des riptive analysis will have to en ompass imperfe t ra-

tionality (see e.g., Ellison and Fudenberg (1995)).

Conversation is riti al in the ontagion of popular ideas about �nan ial markets, as

emphasized by Shiller (2000a).38 In a survey of individual investors, Shiller and Pound

(1989) found that almost all of the investors who re ently pur hased a sto k had their

attention drawn to it through dire t interpersonal ommuni ation. The in uen e of

onversation on trading may arise from individuals' over on�den e about their ability

to distinguish pertinent information from noise or propaganda; examples of large pri e

33See e.g., the dis ussions of Chapman (1998), Loewenstein and Prele (1992), and Loewenstein(1996, 2000).

34See Ainslie (1975), Kirby and Herrnstein (1995), Loewenstein and Prele (1992), Thaler (1981).35See Fernandez-Villaverde and Mukherji (2000), Mulligan (1996) Rubinstein (2000).36See Harris and Laibson (2001), Laibson (1997), Luttmer and Mariotti (2000), O'Donoghue and

Rabin (1999).37See, e.g., Banerjee (1992), Bikh handani, Hirshleifer, and Wel h (1992); on the possibility of infor-

mational as ades in se urities markets, see Avery and Zemsky (1998), Lee (1998).38See the onversational learning models of Banerjee and Fudenberg (1999) and Cao and Hirshleifer

(2000).

20

movements triggered by internet hat omes to mind.

As dis ussed by Shiller (1999), be ause of limited attention people tend to pay mu h

more attention to ideas or fa ts that are reinfor ed by onversation, ritual and symbols.

In onsequen e ulture be omes an important determinant of behavior, and expression

of ideas an be self-reinfor ing. Kuran and Sunstein (1999) des ribe the pro ess of

belief formation as leading to `availability as ades', wherein an expressed per eption is

per eived to be more plausible as a onsequen e of its in reased availability in publi

dis ourse.

Conversation pools information surprisingly poorly. Groups of people tend to talk

mu h more about information signals that they already share than individuals-spe i�

signals (Stasser, Taylor, and Hanna (1989)). As a result groups sometimes fail to de-

te t patterns that are dis ernable by ombining individual-spe i� signals (Stasser and

Titus (1985)). Environmental pressures su h as rowding and unusual ir umstan es

ause group members to experien e ` ognitive overload,' and rigid thinking (adheren e

to habitual behaviors; see Argote, Turner, and Fi hman (1988)).

When ommuni ating information, people tend to sharpen and level, i.e., emphasize

what they onstrue to be the main point, and deemphasize qualifying details that might

onfuse this point. This is ne essary for larity given ognitive onstraints (Allport

and Postman (1947), Anderson (1932)), but tends to ause listener beliefs to move to

extremes. A losely related point is that auses tend to be oversimpli�ed, distorting

listener beliefs. There are also systemati message distortions related to a desire to be

entertaining or to manipulate the listener (see Gilovi h (1991), h. 6). These fa ts point

to the need for analysis of onversation and rumors in se urities markets.

The fundamental attribution error (Ross (1977)) is the tenden y for individuals to

underestimate the importan e of external ir umstan es and overestimate the impor-

tan e of disposition in determining the behavior of others. In a �nan ial ontext, su h

a bias might ause observers of a repur hase to on lude that the CEO dislikes holding

ex ess ash rather than that the CEO is responding to market undervaluation of the

sto k. This would suggest market underrea tion to orporate events.

People mistakenly believe that others share their beliefs more than they really do, the

false onsensus e�e t (e.g., Ross, Green, and House (1977)). Self-de eption may en our-

age this by making the individual relu tant to onsider the possibility that he is making

a deviant error. False onsensus may also result from availability (sin e like-minded

people tend to asso iate together). The urse of knowledge (Camerer, Loewenstein, and

Weber (1989)) is a tenden y to think that others who are less informed are more similar

21

in their beliefs to the observer than they really are.

I.5 Modelling Alternatives to Expe ted Utility and to Bayesian

Updating

Expe ted utility theory has dominated �nan ial modelling be ause it aptures rational

de isionmaking elegantly. However, the paradoxes of Allais (1953) and subsequent on-

�rmations showed systemati violations of expe ted utility; people seem to be in uen ed

by `irrelevant alternatives'. Further violations have multiplied. Eviden e of systemati

preferen e reversals suggests that hoi e may not be well des ribed by maximization

of a utility fun tion at all. A less radi al departure from the traditional approa hes

is to onsider alternative obje tives (Camerer (1995, 1998) provides an in-depth treat-

ment). Camerer dis usses generalizations that involve fun tional forms on probability

weightings and utility fun tions, in some ases expli itly derived from modi�ed axioms

of hoi e.

In prospe t theory (Kahneman and Tversky (1979), Tversky and Kahneman (1992)),

individuals maximize a weighted sum of `values' (analogous to utilities), where the

weights are fun tions of probabilities (instead of true probabilities). Extremely low

probabilities are treated as impossibilities, and extremely high probabilities as ertain-

ties. In ontrast, very (but not extremely) low probabilities are overestimated, and very

(but not extremely) high probabilities are overestimated. For intermediate probabilities,

the weighting fun tion in reases with a slope less than one. The value fun tion is kinked

at the `referen e point' (loss aversion).39 The value fun tion is on ave to the right of

the referen e point and onvex to the left, re e ting risk aversion among gambles that

involve only gains and risk seeking among gambles involving only losses.

The advantage of this approa h is that it an apture many of the known patterns of

individual hoi e under risk, as well as �nan ial regularities (see, e.g., Camerer (1998),

Shiller (1999)). Indeed, Camerer (1998) argues that a form of prospe t theory �ts the

data better than either expe ted utility theory or the other generalizations that have

been proposed.

Several other generalizations of time-additively separable expe ted utility have been

applied to asset pri ing issues, espe ially the equity premium puzzle. Epstein and Zin

39First-order risk averse preferen es (Epstein and Zin (1990)), like loss aversion, involve a utilityfun tion that depends on a referen e point, and in whi h there is nontrivial aversion even to small risks.In the ase of disappointment aversion (Gul (1991)), investors weigh out omes that are worse than the ertainty-equivalent out ome more heavily than favorable out omes.

22

(1989) developed a lass of intertemporal utility fun tions that allow for non-additivity

and non-expe ted utility behavior. Priming is a phenomenon in whi h exposure to a

stimulus a�e ts a subje t's later response to further presentation of the same or a re-

lated stimulus. Eviden e of priming e�e ts does not tell us how people rea t to repeated

onsumption hoi es (self-administered stimuli of a sort), but is broadly suggestive that

past onsumption levels may in uen e how people respond to future onsumption lev-

els. Su h dependen e is re e ted in habit formation preferen es (Constantinides (1990),

Sundaresan (1989)), in whi h the utility derived from urrent onsumption also depends

on a habitual level of onsumption.

Gilboa and S hmeidler (1995) o�er a ase based de ision theory whi h, unlike ex-

pe ted utility theory, is not based on evaluating out omes and their probabilities. A

ase is a menu of de ision options. Choi es are evaluated based on out omes of past

hoi es and how similar those hoi es are to those in the urrent menu.

The eviden e on heuristi s and biases also suggests that Bayesian updating is not

fully des riptive of human behavior. However, Bayes theorem is non-arbitrary, whi h

is a useful dis ipline for modelling. Some re ent models des ribe updating based on

self-attribution bias and on�rmatory bias.40

II Eviden e of Risk and Mispri ing E�e ts

I lassify the eviden e bearing on asset mispri ing into �ve ategories: (1) return pre-

di tability; (2) the equity premium puzzle; (3) eviden e as to whether �rms take a tions

in response to mispri ing; (4) whether �rms take a tions in order to reate mispri ing;

and (5) eviden e of investment errors.41 My emphasis here is on �ndings that have

re eived on�rmation over time and lo ation. However, su h onsisten y is not a pre-

requisite for a pattern to be interesting. If widespread and fairly stable patterns of

mispri ing exist, then almost surely transient and situation-spe i� ones do too.

II.1 Predi tability of Se urity Returns

Return predi tability resear h is haunted by the spe ter of datamining. Some of the

patterns des ribed here are probably just vagaries of han e. However, predi tability

40See Daniel, Hirshleifer, and Subrahmanyam (1998), Gervais and Odean (2001), Rabin (2000), Rabinand S hrag (1999), and Yariv (2001).

41This topi is vast; for re ent reviews of di�erent aspe ts of the eviden e pertaining to mispri ing,see Fama (1991, 1998), Hirshleifer and Teoh (2001), Kothari (2000), and Lee (2001). I do not dis ussa tions by outsiders su h as mutual funds to exploit predi tability.

23

is a generi predi tion of modern asset pri ing theories. So autious skepti ism rather

than profound suspi ion is alled for.

Most of the patterns of return predi tability summarized here have dual (and du-

elling) explanations based on either risk premia or mispri ing. Empiri al papers on

predi tability often interpret psy hologi al explanations naively. Several authors in-

terpret eviden e that fa tor loadings or aggregate onditioning variables an apture

predi tability as ounter to the psy hologi al approa h. But the psy hologi al approa h

re ognizes that investors should are about fa tor risk. To attribute a return pattern to

rational fa tor pri ing requires not just a �nding that fa tors matter, but measurement

of whether expe ted returns are ommensurate with the relevant risks. Furthermore,

the psy hologi al approa h predi ts that fa tors, not just residuals, will be mispri ed.

The onditioning variables and the variables used to identify fa tors, su h as aggregate

dividend yield, the term premium, the default premium, book/market, and size, are very

natural proxies for fa tor misvaluation, as will be dis ussed.

II.1.1 Predi tability Based upon Fa tor Risk Measures

I fo us here on CAPM beta and the fa tor loadings of Fama and Fren h (1993). A

positive univariate relation of beta with expe ted returns is found in most studies, but

depends on the ountry, time period, empiri al implementation, and form of the CAPM

being tested.42 Beta has in remental power to predi t future returns after ontrolling

for market value and/or fundamental/pri e ratios in some studies but not others.43

II.1.2 Predi tability Based upon Pri e and Ben hmark Value Measures

A natural way to identify mispri ing is to ompare an asset's pri e to a related value

measure. A remarkably onsistent empiri al pattern is that almost any su h pairing

that resear hers try predi ts future returns in the right dire tion{ the ` heap' se urity

on average appre iates relative to a risk-adjusted ben hmark, or relative to an `expensive'

se urity. EÆ ient markets fans will on lude, however, that the se urity is heap be ause

it is riskier, and that the risk adjustment is misspe i�ed.44

42See, e.g., Bla k, Jensen, and S holes (1972), Bossaerts (2000), Fama and Fren h (1992), Fama andMa Beth (1973), Handa, Kothari, andWasley (1993), Harvey (1989), Heston, Rouwenhorst, andWessels(1999), Kim (1997), Kothari, Shanken, and Sloan (1995), Kothari and Shanken (2000), Rouwenhorst(1999), Solnik (1974).

43See, e.g., Fama and Fren h (1992, 1996a), Handa, Kothari, and Wasley (1993), Heston, Rouwen-horst, and Wessels (1999), Jagannathan and Wang (1996), Kim (1997), Knez and Ready (1997), Kothariand Shanken (2000), and Kothari, Shanken, and Sloan (1995).

44This insight has been applied to sto ks by Ball (1978), Berk (1995), and Keim (1988).

24

In several ases the market value of a parent �rm has been substantially less than one

of its parts, and managers undertook transa tions apparently suitable for exploiting the

overpri ing of a division.45 Closed end funds often trade at dis ounts and premia relative

to net asset value; these dis ounts predi t future small sto k returns.46 Se urities that

are virtually perfe t substitutes are sometimes traded at di�erent pri es by di�erent

lienteles (Froot and Dabora (1999), Rosenthal and Young (1990)).

A short-term yield provides a value ben hmark for a long term bond. Dis repan-

ies between long- and short-term yields positively predi t the holding period returns

on long-term bonds.47 Bonds denominated in di�erent urren ies provide mutal ben h-

marks. Investing in a ountry's bonds that have re ently be ome heaper (higher nominal

yield) relative to another ountry's bonds on average earns higher returns|the forward

premium puzzle (see, e.g., Engel (1996)).

Sto k ben hmarks in lude fundamental measures su h as book value, earnings, or

even a onstant (for the size e�e t). Cross-se tionally, equity-pri e-related variables

(e.g., 1/pri e, book/market, earnings/pri e, debt/equity) predi t high sto k returns in

U.S. and many other ountries, even after ontrolling for beta.48 For the sto k market

as a whole, high fundamental/pri e ratios (dividend yield or book/market) predi t future

index returns in the U.S. and internationally in several, though not all studies.49 A better

predi tor of ross-se tional and aggregate returns an be formed by normalizing pri e

with earnings-based indi es of fundamental value.50 Market returns are also predi table

based on term and default spreads.51

45Cornell and Liu (2000), Lamont and Thaler (2000), S hill and Zhou (1999).46See Swaminathan (1996), Neal andWheatley (1998). Bodurtha, Kim, and Lee (1995) �nd that U.S.-

traded losed-end ountry fund premia and dis ounts are often large, and omove primarily be ause oftheir ommon sensitivity to the U.S. market. Country fund premia predi t returns on U.S. size-rankedportfolios and fund sto k returns.

47Bekaert, Hodri k, and Marshall (1997b), Campbell and Shiller (1991), Co hrane (2000) Se . 20.1,Fama and Bliss (1987), Mankiw and Summers (1984), Mankiw (1986), and Shiller, Campbell, andS hoenholtz (1983).

48See e.g., Banz (1981), Basu (1983), Bhandari (1988), Chan, Hamao, and Lakonishok (1991), Daniel,Titman, and Wei (2001), Davis (1994), Davis, Fama, and Fren h (2000), DeBondt and Thaler (1987),Fama and Fren h (1992, 1998), Haugen and Baker (1996), Hawawini and Keim (2000), Heston, Rouwen-horst, and Wessels (1995), Jagannathan, Kubota, and Takehara (1998), Rosenberg, Reid, and Lanstein(1985), Rouwenhorst (1999), and Stattman (1980).

49See Bossaerts and Hillion (1999), Campbell and Shiller (1988a), Fama and Fren h (1988a), Goet-zmann, Ibbotson, and Peng (2001), Goyal and Wel h (1999), Hodri k (1992), Kothari and Shanken(1997), Lewellen and Shanken (2000), Pesaran and Timmermann (1995), and Ponti� and S hall (1998).

50See Abarbanell and Bushee (1998), Chang, Chen, and Dong (1999), Frankel and Lee (1998, 1999),Lee, Myers, and Swaminathan (1999).

51See Campbell (1987), Fama and Fren h (1989), Keim and Stambaugh (1986).

25

Size and value portfolios are asso iated with a fa tor or fa tors distin t from the

sto k market portfolio.52 The loadings on three fa tors based on size, value and the

market predi t the returns on portfolios sorted on various hara teristi s, but do not

explain short-term momentum; a global two-fa tor model predi ts international returns

(Fama and Fren h (1996b, 1998)).

Several studies report very high Sharpe ratios a hievable based on ross-se tional

value e�e ts,53 a point reinfor ed by low international orrelations of some size and value

strategies (Hawawini and Keim (1995)). This raises the question of whether the implied

variability of marginal utility a ross states under rational asset pri ing is implausibly

high (see Hansen and Jagannathan (1991)). Chen (2000) �nds that book/market and

momentum-based portfolios do not ontain enough information about future returns on

aggregate wealth to be strongly pri ed as state variables in a Merton ICAPM.

Fama and Fren h suggest that size and book/market fa tors may be orrelated with

harms su�ered by individuals when �rms are distressed. Di�ering on lusions have been

drawn about the asso iation of size and book/market with distress.54 The book/market

e�e t remains strong after ontrolling for distress (GriÆn and Lemon (2001)). The

voluntary allo ation by employees of personal retirement funds into shares of their own

�rms (Benartzi (1997)) opposes the distress-risk hypothesis.

Con lusions di�er as to whether ` hara teristi s' (size, book/market) or fa tor load-

ings do a better job predi ting returns.55 Perhaps the most ompelling eviden e for

expe tational errors is that, after portfolios are formed, growth sto ks on average re-

spond very negatively to subsequent earnings announ ements for several years, and

value sto ks do not (La Porta et al (1997), Skinner and Sloan (2000)).

II.1.3 Predi tability Based upon Past Returns: Momentum and Reversal

In many asset and se urity lasses internationally there is positive short-lag auto orre-

lation and negative long-lag auto orrelation.56 Cross-se tionally, U.S., European, and

52See Fama and Fren h (1993, 1995), Liew and Vassalou (2000). This of ourse does not guaranteethat the loadings on these fa tors are pri ed separately from market beta; for example, under the CAPMthey would not be.

53Brennan, Chordia, and Subrahmanyam (1998), Lakonishok, Shleifer, and Vishny (1994), Ma Kinlay(1995).

54Chan and Chen (1991), Chen and Zhang (1998), Di hev (1998), Shumway (1996).55See Daniel and Titman (1997), Daniel, Titman, and Wei (2001), Davis, Fama, and Fren h (2000),

Jagannathan, Kubota, and Takehara (1998), and Lewellen (1999).56See Balvers, Wu, and Gilliland (2000), Barkham and Geltner (1995), Case and Shiller (1990),

Chan, Hameed, and Tong (2000), Cutler, Poterba, and Summers (1991), Fama and Fren h (1988b),Gyourko and Keim (1992), Ng and Fu (2000), Poterba and Summers (1988) and Ri hards (1997). On

26

emerging market sto ks that have done very well in the re ent past (about 3-12 months)

tend to do well over the next month.57 Long term reversals in the ross-se tion were

do umented by DeBondt and Thaler (1985).58 Momentum is stronger in small �rms,

growth �rms, �rms with low analyst following, and in the se urity-spe i� (non-market)

omponent of sto k returns.59 Volume intera ts with momentum in predi ting future re-

turns, suggesting a possible y le of overrea tion and orre tion (Lee and Swaminathan

(2000b)). Chordia and Shivakumar (2000) report that momentum pro�ts an be ap-

tured based on se urity sensitivities to a few aggregate variables (see also Ahn, Conrad,

and Dittmar (2000)). Lewellen (2000) provides eviden e of negative auto orrelation and

ross-serial orrelation in industry and size portfolios, onsistent with negative market

auto orrelation during the study's time period.60

Past winners earn substantially higher returns than do past losers at the dates of

quarterly earnings announ ements o uring in the 7 months following portfolio forma-

tion.61 This is suprising from a rational risk perspe tive be ause high momentum �rms

should be ome less leveraged and less risky. Also, �rms with extremely low returns over

the several months are having trouble, so the distress fa tor view of value e�e ts suggests

that negative momentum �rms should earn high future returns.

II.1.4 Predi tability Based upon Publi Versus Private News Events

Several event studies have do umented abnormal returns subsequent to the event date.

One explanation, event sele tion, is that a �rm's de ision whether and when to engage

in the event depends on whether there is market misvaluation. A se ond possibility, ma-

nipulation, is that around the time of the a tion the �rm re on�gures other information

methodologi al and robustness issues for sto ks, see Carmel and Young (1997), Jegadeesh (1991), Kim,Nelson, and Startz (1988), Ri hardson and Sto k (1989), and Ri hardson and Smith (1994). Thereis also a literature on whether sto k returns are ex essively volatile relative to dividend variability(Campbell and Shiller (1988a, 1988b), Kleidon (1986), LeRoy and Porter (1981), Marsh and Merton(1986), Shiller (1979, 1981), and West (1988). This is equivalent to the issue of whether there is ex essivelong-run reversal in sto k pri es (see Co hrane (1991)), sin e any overrea tion must in rease volatility.

57Jegadeesh and Titman (1993), Rouwenhorst (1998), Rouwenhorst (1999).58On methodologi al issues and the robustness of this �nding, see Ball and Kothari (1989), Ball,

Kothari, and Shanken (1995), Chan (1988), and Chopra, Lakonishok, and Ritter (1992).59See Daniel and Titman (1999), Grinblatt and Moskowitz (1999), Grundy and Martin (2001), Hong,

Lim, and Stein (2000), and Jegadeesh and Titman (1993). Both industry and non-industry omponentsof momentum help predi t future returns (Grundy and Martin (2001), Moskowitz and Grinblatt (1999)).

60A given serial ovarian e stru ture is potentially subje t to very di�erent ausal interpretations.Jegadeesh and Titman (1995) provide a de omposition that distinguishes fa tors from residuals, andtherefore lends itself to a distin tion between fa tor versus residual auto orrelation.

61Jegadeesh and Titman (1993); see also Chan, Jegadeesh, and Lakonishok (1996).

27

reported to investors in order to indu e misvaluation.

There is eviden e suggesting that both sele tion and manipulation o ur. Regarding

sele tion, a remarkable pattern emerges from studies of dis retionary orporate events

(a tions hosen by management or other potentially informed parties). The average

abnormal sto k return in the 3-5 years subsequent to the event has the same sign as the

event-date sto k pri e rea tion. I all this regularity post-event return ontinuation.62

The eviden e that has appeared sin e this post-event return ontinuation hypothe-

sis was proposed by Daniel, Hirshleifer, and Subrahmanyam (1998) has generally been

supportive over new time periods and events. There has been little study of post-event

performan e for events that are not taken at the dis retion of management or analysts

with in entives to rea t to mispri ing. However, Cornett, Mehran, and Tehranian (1998)

�nd that there is post-event ontinuation when bank sto ks issue equity, ex ept when

equity issuan e is for ed by reserve requirements.

Fama (1998) argues that these return patterns are sensitive to empiri al methodol-

ogy. Several re ent studies have on luded that there is limited or no underperforman e

of new issue �rms.63 However, some re ent methods minimize the power to dete t mis-

valuation e�e ts (Loughran and Ritter (2000)). Jegadeesh (1999) reports large post-SEO

underperforman e even relative to several (ex essively) stringent return ben hmarks.

The argument that post-IPO underperforman e is eliminated by an appropriate

ben hmark is ounterintuitive, be ause it amounts to saying that IPO �rms have un-

62Events for whi h this has been found in lude sto k splits (Grinblatt, Masulis, and Titman (1984),Desai and Jain (1997b), Ikenberry, Rankine, and Sti e (1996), Ikenberry and Ramnath (2000)); tendero�er and open market repur hases (Lakonishok and Vermaelen (1990), Ikenberry, Lakonishok and Ver-maelen (1995, 2000)); equity arveouts (Vijh (1999)); spino�s (Cusatis, Miles, and Woolridge (1993),Desai and Jain (1997a)); a ounting writeo�s (Bartov, Lindahl, and Ri ks (1998)); analyst earningsfore ast revisions (Chan, Jegadeesh, and Lakonishok (1996), Lin (2000a, 2000b)); analyst sto k re -ommendations (Barber et al (2001), Bjerring, Lakonishok, and Vermaelen (1983), Elton, Gruber, andGultekin (1984), Groth et al (1979), Kris he and Lee (2000), Mi haely and Woma k (1999), Wom-a k (1996)); dividend initiations (Mi haely, Thaler, and Woma k (1995), Boehme and Sores u (2000));dividend omissions (Mi haely, Thaler, and Woma k (1995)); seasoned issues of debt (Spiess and A�e k-Graves (1999)); seasoned issues of ommon sto k (Cornett, Mehran, and Tehranian (1998), Foerster andKarolyi (2000), Jegadeesh (1999), Loughran and Ritter (1995), Spiess and A�e k-Graves (1995), Teoh,Wel h, and Wong (1998b), but see Kang, Kim, and Stulz (1999)); publi announ ement of previousinsider trades (Seyhun (1988) and Roze� and Zaman (1988)); and venture apital share distributions(Gompers and Lerner (1998)). The hypothesis has not been tested for IPOs sin e we do not observethe pri e rea tion to the announ ement that an IPO will o ur. The pattern does not hold for ex hangelisting (Dharan and Ikenberry (1995), Foerster and Karolyi (1999), and M Connell and Sanger (1987));and private pla ements (Hertzel et al (1999)), whi h may involve informed dis retion on the part of thebuying as well as the selling party.

63See Brav, Ge zy, and Gompers (2000), E kbo and Norli (2000), E kbo, Masulis, and Norli (2000),Gompers and Lerner (2000), and Mit hell and Sta�ord (2000).

28

usually low risk. Risk redu tion may justify a low return ben hmark for SEO �rms,

but risk in reases would seem to imply a higher ben hmark after debt issues or bond

rating downgrades, making the underperforman e after these events64 even stronger.

Poor post-downgrade performan e also opposes the distress-risk-fa tor theory of return

predi tability. New issue �rms perform espe ially badly at subsequent earnings an-

noun ement dates, whi h is hard to interpret as a negative risk premium.65

Irrelevant, redundant or old news a�e ts se urity pri es when presented saliently.66

These demonstrable examples of mispri ing suggest that less blatant mispri ing may

o ur routinely. Little of sto k pri e or orange jui e futures pri e variability has been ex-

plained empiri ally by relevant publi news.67 Histori al rashes and spe ulative episodes

are often hard to explain in terms of fundamental news.68 Allen (2001) provides exam-

ples suggesting that bubbles have major e onomi onsequen es, and argues that agen y

problems among �nan ial institutions may ause bubbles.

Several studies explore fundamental trends and subsequent returns. Cash or earnings

surprises are followed by positive abnormal returns in the short run, and perhaps negative

abnormal returns in the long run.69 Investors also seem to extrapolate fundamentals in

options and in football betting markets (Avery and Chevalier (1999), Poteshman (2000)).

II.1.5 Predi tability Based upon Mood Proxies

Environmental fa tors that in uen e mood are orrelated with sto k pri e movements.

Kamstra, Kramer, and Levi (2000a) �nd that a deterministi variable, hanges to and

from daylight savings time, disrupts sleep patterns, and is related to sto k returns.70

A sto hasti variable, loud over in the ity of a ountry's major sto k ex hange, is

asso iated with low daily sto k index returns in a joint test of 26 national ex hanges as

well as in the U.S. (Hirshleifer and Shumway (2000), Saunders (1993)).

64Spiess and A�e k-Graves (1999), Di hev and Piotroski (2001).65Jegadeesh (1999), Denis and Sarin (2000); see also Ikenberry and Ramnath (2000).66Andrade (1999), Ashton (1976), Avery and Chevalier (1999), Cooper, Dimitrov, and Rau (2000),

Hand (1990, 1991), Ho and Mi haely (1988), Huberman and Regev (2001), Klibano�, Lamont, andWizman (1999), Rashes (2001), Rau and Vermaelen (1998).

67Cutler, Poterba, and Summers (1989), Fair (2000), Roll (1984, 1988).68See, e.g., Cutler, Poterba, and Summers (1989), Seyhun (1990), Shiller (2000b) h.4; for a mainly

rational perspe tive on the Dut h tulip bulb boom, see Garber (1989).69On short run post-earnings announ ement drift, see, e.g., Bernard and Thomas (1989, 1990). For

poor long lag performan e, see DeBondt and Thaler (1987), Lakonishok, Shleifer, and Vishny (1994),Lee and Swaminathan (2000a), but see also DeChow and Sloan (1997) and Daniel and Titman (2000).

70Kamstra, Kramer, and Levi (2000b) examine the relation of another deterministi variable, seasonalshifts in length of day, to returns in several ountries.

29

II.2 Equity Premium and Riskfree Rate Puzzles

The equity premium puzzle71 is that U.S. equity market returns are high relative to risk,

implying high levels of risk aversion and so a low elasti ity of intertemporal substitution

in onsumption. This in turn implies very high real interest rates to indu e individuals

to a ept lower onsumption now than in the future ( onsistent with histori al growth

in onsumption; see Weil (1989)).

II.3 A tions Possibly Taken in Response to Mispri ing

Corporations buy and sell shares in a way that is orrelated with possible measures of

market mispri ing.72 The amount of �nan ing and repur hase varies widely over time

in an industry-spe i� way. Mergers bids, whi h often rely on equity �nan ing, are also

prone to booms and quiet periods by industry. New losed-end funds are started in

those years when seasoned funds trade at small dis ounts or at premia relative to net

asset value (Lee, Shleifer, and Thaler (1991)), and tend to be issued at a premium (plus

ommission) before reverting to a dis ount in the aftermarket (Peavy (1990)).

II.4 A tions Possibly Taken to Create Misvaluation

Firms sometimes make a ounting adjustments (a ruals) to boost their earnings rel-

ative to a tual ash ow. These adjustments are publi ly dis losed in �rms' �nan ial

statements. When a ruals are abnormally high, sto ks on average subsequently ex-

perien e poor return performan e.73 Managers boost a ruals at the time of new IPO

and seasoned equity issues (Teoh, Wel h and Wong (1998a, 1998b)). Greater earnings

management in IPOs and in SEOs is asso iated with more optimisti errors in analyst

earnings fore asts, and with more adverse subsequent long-run abnormal sto k returns.74

Managers adjust earnings to meet threshold levels su h as zero, past levels, and

levels fore ast by analysts (DeGeorge, Patel, and Ze khauser (1999)). Possibly under

71See Hansen and Singleton (1983), Mehra and Pres ott (1985), Hansen and Jagannathan (1991),and Shiller (1982). Purely rational explanations have been o�ered based upon learning (Brennan andXia (2001)), lu k (Fama and Fren h (2000)), sele tion bias in the fo us of a ademi attention (Brown,Goetzmann, and Ross (1995)), borrowing onstraints (e.g., Constantinides, Donaldson, and Mehra(2000)), and non-sto k-market in ome sho ks (e.g., Constantinides and DuÆe (1996) and Heaton andLu as (1996)).

72See, e.g., Jindra (2000), D'Mello and Shrof (2000), Dittmar (2000); Koraj zyk, Lu as, and M Don-ald (1991) provide a possible rational explanation for this phenomenon.

73See Chan, Jegadeesh, and Lakonishok (2000), Sloan (1997), Teoh, Wel h and Wong (1998a, 1998b).74See Teoh, Wel h and Wong (1998a, 1998b), Teoh and Wong (2000).

30

the in uen e of management, sto k analysts on average `walk down' their fore asts from

overly optimisti levels at long horizons to pessimisti fore asts that �rms are likely to

beat by year-end (Ri hardson, Teoh, and Wyso ki (1999)).

II.5 Quality of Information Aggregation

In ontrast with early lassi work on experimental markets, the thrust of mu h ex-

perimental market resear h in the late 1980's and 1990's is that in only slightly more

ompli ated environments, information is not aggregated eÆ iently (see, e.g., the surveys

of Libby, Bloom�eld, and Nelson (2001), Sunder (1995)). Presumably this is be ause

onfounding e�e ts make it harder for investors to disentangle the reasons behind the

trades of others (see, e.g., Bloom�eld (1996)).

II.6 Investor Behavior

Portfolio theory suggests that (apart from transa tion osts) everyone should parti ipate

in all se urity markets. But even now, many investors negle t major asset lasses. Non-

parti ipation may derive from salien e bias, or from mere exposure (familiarity) e�e ts.

Investors are subje t to a strong bias toward investing in sto ks based in their home

ountry and in their lo al region.75 Employees invest heavily in their own �rm's sto k and

per eive it to have low risk (Huberman (1999)). The degree to whi h they invest in their

employer's sto k does not predi t the sto k's future returns (Benartzi (1997)). There is

also experimental eviden e that investors sometimes fail to form eÆ ient portfolios and

violate two-fund separation.76

Several though not all studies of investor behavior in natural and experimental mar-

kets report eviden e onsistent with a disposition e�e t|a greater readiness to realize

gains than losses.77 Certain groups of investors hange their behaviors in parallel (`herd-

ing'), in some ases engaging in momentum (or positive feedba k) trading and in other

ases in ontrarian trading.78 Similar behavior is not irrational per se, but some groups

of investors do poorly.

75See, e.g., Cooper and Kaplanis (1994), Coval and Moskowitz (1999), Grinblatt and Keloharju(2001a), Huberman (1999), Kang and Stulz (1997), Lewis (1999), and Tesar and Werner (1995).

76Bossaerts, Plott, and Zame (2000), Kroll, Levy and Rapoport (1988b, 1988a), and Kroll and Levy(1992).

77See, e.g., Ferris, Haugen, and Makhija (1988), Grinblatt and Keloharju (2001b), Odean (1998a),Shefrin and Statman (1985), Weber and Camerer (2000); but see also Ranguelova (2000).

78See Choe, Kho, and Stulz (1999), Grinblatt, Titman, and Wermers (1995), Grinblatt and Keloharju(2000), Lakonishok, Shleifer, and Vishny (1992), Nofsinger and Sias (1999), and Wermers (1999).

31

People (espe ially males) seem to trade too aggressively, in urring higher transa -

tions osts without higher returns.79 Furthermore, traders in experimental markets do

not pla e enough weight on the information and a tions of others (Bloom�eld, Libby,

and Nelson (1999)). Both �ndings are onsistent with over on�den e. In experimen-

tal markets, as in psy hologi al experiments, investors and pri es are more prone to

overrea ting to unreliable than to reliable information.80

Investors not infrequently make agrant errors, su h as failing to exer ise in-the-

money options at expiration, and apparently failing to exploit arbitrage opportunities

(Longsta�, Santa-Clara, and S hwartz (1999), Rietz (1998)). In retirement fund on-

tribution de isions, there is eviden e that people are strongly subje t to status quo bias,

diversify naively by dividing their ontributions evenly among the options o�ered, and

appear to naively extrapolate past return performan e.81

III Asset Pri ing Theories Based on Investor Psy-

hology

The eviden e in the pre eding se tion presents hallenging puzzles to be explained.

Some pioneering models aptured imperfe t rationality in asset markets by in luding

me hanisti traders who either make pure noise trades, or positive feedba k trades in

whi h new pur hases are an in reasing fun tion of past pri e moves.82 This was an

eÆ ient way to illustrate some ru ial insights about survival, arbitrage, and pri ing.

However, in full generality, the me hanisti modelling approa h is very elasti . If noise

trades an be arbitrarily orrelated with other e onomi variables, any return pattern

an be explained. The e onomi ontent of me hanisti trader models omes from the

hoi e of assumptions on trades to re e t fa ts about psy hology or trading. In the

hope of being more a urately predi tive, re ent resear h has expli itly modelled how

de isionmaking o urs in a way that re e ts psy hologi al biases.

In a spe i� investment setting, it an be hard to judge whi h do umented psy ho-

logi al bias is relevant. This reates an extra degree of freedom for model-mining not

present in the purely rational approa h. Thus, even more than for purely rational theo-

ries, a psy hologi al theory be omes more persuasive if it explains a range of empiri al

79See Odean (1999), Barber and Odean (2000b, 2000a).80See Bloom�eld, Libby, and Nelson (2000), Bloom�eld et al (2001).81See Benartzi (1997), Benartzi and Thaler (2001), and Madrian and Shea (2000).82See Cutler, Poterba, and Summers (1990), DeLong et al (1990a, 1990b), and Frankel and

Froot (1986, 1990).

32

patterns in di�erent ontexts, and generates new impli ations.

The next subse tion starts with models of the simple stati s of mispri ing and or-

re tion. Models of dynami s follow in Subse tion III.2. A stati setting an address how

risk and mispri ing determine the ross-se tion of expe ted returns. Mispri ing proxies

apture long-term misvaluation and orre tion. Models of dynami s an des ribe in-

tertemporal patterns, su h as a shift from underrea ting to overrea ting to a stream of

news, or a pattern of overrea ting and then overrea ting even more. Thus, dynami anal-

yses an address patterns of short- versus long-term return auto orrelations. Subse tion

III.3 dis usses how to empiri ally distinguish psy hology-based pri ing theories.

III.1 Stati Asset Pri ing

I onsider stati models based upon either limited attention/parti ipation, or over on-

�den e. Merton (1987) analyzed the ross-se tion of se urity returns in a stati asset

pri ing model with exogenous non-parti ipation. Su h non-parti ipation an be viewed

as re e ting limited attention, preferen e for the familiar, and salien e e�e ts. The

key impli ation of the model is that negle ted sto ks earn abnormally high expe ted

returns.

Some re ent stati analyses of psy hology and se urity returns are based on investor

over on�den e. Finan ial analysts and investors di�er in their skill at a quiring informa-

tion through means su h as interviewing management, analyzing �nan ial statements,

and internet hat. An investor who overestimates his ability to do so will underestimate

his errors in fore asting value. Thus, as in Kyle and Wang (1997), in these models an

over on�dent investor overestimates the pre ision of his information signals.

Odean (1998b) studies the stati s of over on�den e when there is a single risky

se urity. When pri e-taking investors think the signal is more a urate than it really

is, the market pri e overrea ts to the to the signal. Eventually, when the true state of

the world resolves, the pri e orre ts. This pattern of overrea tion and reversal auses

ex ess pri e volatility, and negative long run return auto orrelation.

Instead of a general tenden y to overestimate signal pre ision, in Daniel, Hirshleifer,

and Subrahmanyam (1998), investors are only over on�dent about private information

signals. This re e ts the notion that an investor's self-esteem is tied to his own ability

to a quire useful information. Individuals re eive a private signal, and subsequently

update based on an in on lusive publi signal. In the stati version of the model, investor

on�den e is �xed. Managers may sele tively undertake good news a tivities su h as

33

a sto k split or repur hase at least partly in response to market undervaluation of the

�rm, and other a tivities su h as new issue when the the �rm is overvalued.

Sin e investors overrea t to private signals, returns on private information arrival

dates tend to reverse. In ontrast, for sele tive publi events, the model implies post-

event ontinuation of sto k returns: sele tive events asso iated with positive (negative)

average event-date rea tions are also asso iated with positive (negative) average post-

event long-run abnormal returns. Intuitively, when the �rm (or another party) takes a

publi a tion in opposition to over on�dent mispri ing, the market orre ts only partially

in the short run.

In a model with multiple se urities, Daniel, Hirshleifer, and Subrahmanyam (2001a)

provide an analog to the CAPM when investors are over on�dent. Se urity terminal

ash ows satisfy a linear fa tor model, and ea h investor observes signals about the

fa tors and the idiosyn rati omponent of se urity payo�s. Risk-averse investors form

what they per eive to be mean-varian e eÆ ient portfolios. Over on�dent individuals

trade with risk averse arbitrageurs who form rational beliefs. A se urity's equilibrium

expe ted return is linearly in reasing in the se urity's beta with the market, and the

se urity's urrent mispri ing. Variables ontaining market pri e are proxies for the se-

urity's misvaluation. For example, a fundamental/pri e ratio su h as book/market is

driven down when favorable news drives a sto k up. Sin e there is overrea tion, this is

when the sto k is overvalued. Thus, a high fundamental/pri e ratio predi ts high future

returns. Aggregate value measures su h as the market dividend yield or book/market

positively predi t future market returns.

A fundamental/pri e ratio (e.g., high book/market) tends to be high if either risk is

high or if the market has overrea ted to a highly adverse signal. In either ase, pri e on

average rises. Sin e high book/market re e ts both mispri ing and risk, whereas beta

re e ts only risk, book/market tends to be a better predi tor of returns. These two

sour es of predi tive power are unequal. Beta helps disentangle these ases, so beta and

book/market are joint predi tors of future returns.

However, when over on�den e be omes very strong, and if the proxy for the un-

onditional expe ted value (e.g., book value) is perfe t, then the in remental ability of

beta to predi t future returns vanishes. The fundamental/pri e ratio dominates beta

even though risk is pri ed. This is an extreme ase, but it helps explain why empir-

i al �ndings on the in remental e�e t of beta have been weak and in onsistent. The

model also implies that in univariate regressions beta should predi t future returns. The

model further des ribes the tradeo�s in onstru ting optimal pri e-related proxies for

34

misvaluation.

Daniel, Hirshleifer, and Subrahmanyam (2001b) extend the DHS2 model to examine

regressions of future returns on both book/market and HML loadings (Subse tion II.1.2).

They �nd that in an imperfe tly rational model either hara teristi s (e.g., book/market)

or ovarian es (e.g., HML loadings) an be stronger predi tors of future returns.

III.2 Dynami Asset Pri ing

Stati models provide simple generalizations of the insights of the CAPM that an

en ompass the e�e t of risk as well as mispri ing. However, a stati approa h has

no hope of apturing the distin tion between short-term ontinuation and long-term

reversals. In both stati and dynami models, long-run reversal o urs when there is an

overrea tion to an impulse su h as the arrival of good news. In a dynami setting, short-

run positive auto orrelation is onsistent with long-run reversal so long as the pro ess of

overrea tion and orre tion is suÆ iently smooth. Su h smoothness implies that when

an impulse sets pri e rising, it will probably rise some more; that on average the last

up-move to the peak of the impulse response fun tion is not followed by a pre ipitous

drop; and when the pri e is falling, it tends to fall some more. In ontrast, a long-

lag auto orrelation tends to asso iate positive returns during the overrea tion pro ess

with negative returns arising during the orre tion pro ess. The subse tions that follow

des ribes the e�e ts of pure (independent) noise trading, me hanisti models based on

orrelated trading (positive feedba k), the e�e ts of mistaken beliefs, and the e�e ts of

alternative preferen es.

III.2.1 Pure Noise Trading

Pure noise trading and positive feedba k trading ause overrea tion, and hen e negative

auto orrelations in long-run returns. When a sto k rises too high, it needs to orre t ba k

down. Equivalently, this overrea tion auses ex ess volatility in returns. Furthermore,

Campbell and Kyle (1993) showed that overrea tion an ause aggregate sto k market

value measures su h as dividend yield to predi t future market returns, so that ontrarian

investment strategies are on average pro�table.

DeLong, Shleifer, Summers, and Waldmann (1990a) (hen efore, DSSW1) model the

onsequen es of unpredi table random trades. Two se urities pay identi al, riskless

dividends. The pri e of one asset is exogenously �xed. The other asset is risky be ause

pure noise trades ause sto hasti mispri ing. Rational arbitrageurs with exogenous

35

short time horizons limit their arbitrage trades for fear that the mispri ing will get

worse before it gets better. On average the risky asset trades at a dis ount, the risk

premium demanded by the rational investors.

The noise trading approa h provides an explanation for the existen e and behavior

of losed-end fund dis ounts and their orrelations with sto k returns. A ording to

DSSW1, noise traders buy and sell losed-end funds in a orrelated fashion, ausing

dis ounts or premia relative to net asset value to u tuate. The mispri ing risk this

reates makes these funds less attra tive to rational investors, so on average there is a

dis ount. This theory implies that fund dis ounts move together based on a systemati

noise-trading fa tor; su h omovement exists (e.g., Lee, Shleifer, and Thaler (1991)).

The theory also explains why su h funds are reated: to exploit optimisti noise traders.

Lee, Shleifer, and Thaler (1991) suggest that shifts in fund dis ounts re e t shifts

in noise trader sentiment toward all small sto ks. This is onsistent with their eviden e

that narrowing of losed-end fund dis ounts is asso iated ontemporaneously with high

small sto k returns. This implies that dis ounts predi t small sto k returns (see Se tion

II). If dis ounts were a onsequen e of pure noise trading, they would be un orrelated

with future fundamentals su h as as a ounting performan e. Swaminathan (1996) �nds

that at lags of greater than one year high dis ounts predi t both low future a ounting

pro�ts and high future sto k returns. This is onsistent with fund investors overrea ting

to genuine information.

The omovement in small sto k returns do umented in Fama and Fren h (1993)

may ome from orrelated imperfe tly rational trades (see Shleifer (2000), p.20). The

DSSW1 approa h then suggests that small sto ks, in luding losed-end fund shares, will

earn high expe ted returns in ompensation for their high mispri ing risk. Alternatively,

low market-value sto ks may earn high returns be ause a sto k's low market value on

average derives partly from its being undervalued (see, e.g., Daniel, Hirshleifer, and

Subrahmanyam (2001a)). The U.S. small �rm e�e t has been weak or absent in the last

15 years, yet losed-end fund dis ounts remain.

III.2.2 Positive Feedba k Trading

Positive feedba k trading has several possible motivations, one being that investors form

expe tations of future pri es by extrapolating trends (a topi overed in the next subsub-

se tion). DeLong, Shleifer, Summers, and Waldmann (1990b) (DSSW2) o�er a model

with a risky asset and riskfree ash, in whi h information arrives sequentially. The

exogenous date 2 demand of the positive feedba k traders is linearly in reasing in the

36

pre eding pri e trend. Forseeing this, rational spe ulators buy into pri e trends, exag-

gerating trends and overshooting. As a result there is ex ess volatility, and long-term

negative auto orrelations in returns.

In Cutler, Poterba, and Summers (1990), there are two types of imperfe tly rational

traders, positive feedba k traders, and fundamental traders who ignore pri e and trade

based upon a signal about the se urity's payo�. Some fundamental traders observe

this signal with a lag. This lag reates pri e trends whi h are pro�tably exploited

by feedba k traders. The gradual pro ess of overshooting and orre tion indu es both

short-lag positive auto orrelation and long-lag negative auto orrelation.

More re ent models with endogenous de isions have found things akin to pure noise

trading| a limiting ase of over on�den e, and positive feedba k trading. But endoge-

nously derived positive feedba k is onditional and statisti al, whi h seems more realisti

than the older models. For reasons of both des riptiveness and predi tiveness, expli it

modelling of the psy hology of investors is likely to supersede the me hanisti approa h

(ex ept perhaps in otherwise-intra table appli ations).

III.2.3 Mistaken Beliefs

One explanation for return predi tability is that investors set pri es based on mistaken

expe tations.83 This subse tion �rst onsiders dynami s when irrational individuals

share the same biases (either over on�den e, or representativeness and onservatism). I

then onsider the intera tion of multiple trader types with di�erent biases.

The Dynami s of Biased Attribution and Over on�den e

Two re ent papers provide models with a single risky se urity that re e t the fa t

that people learn about their own abilities in a biased, self-promoting fashion. In these

models, investors do not know the pre ision of their private information signals, whi h

re e ts their information-gathering ability. They learn about their pre ision through

time by observing whether later publi news on�rms or dis on�rms their previous signal.

The analyses assume the dynami omplement of over on�den e, biased self-attribution.

When an investor re eives on�rming news his on�den e in his pre ision rises too mu h,

and when there is dis on�rming news his on�den e de lines too little.

In Daniel, Hirshleifer, and Subrahmanyam (1998), the impulse response fun tion to a

83Shefrin and Statman (1994) analyze the general e�e t of mistaken beliefs on equilibrium pri esin se urities markets. They predi t that when pri es are ineÆ ient, mispri ing is related to a `beta orre tion;' it has not been obvious how to test this.

37

favorable initial sho k, the private information signal, is hump-shaped. Pri e on average

rises further as publi information arrives, be ause on�den e about the private signal

on average grows. Eventually, however, a umulating eviden e for es investors ba k to

a more reasonable self-per eption. This smooth hump-shaped impulse response implies

positive short-lag and negative long-lag return auto orrelations. DHSI also numeri ally

simulate the orrelation of a publi information suprise (su h as favorable a ounting

performan e) with future returns with self-attribution bias. At short lags this orrelation

is positive, but at long lags the orrelation an be negative (see Se tion II.1.4).

Gervais and Odean (2001) provide a model that a ommodates analyti al solution

for the learning pro ess under biased self-attribution. As traders be ome over on�dent

trading volume and market return volatility in rease. Sin e equity is in positive net

supply, the model also predi ts that trading volume will be higher after market rises

than market falls, onsistent with Statman and Thorley (1998).84

The Dynami s of Representativeness and Conservatism

Barberis, Shleifer, and Vishny (1998) (BSV) o�er an explanation for under- and over-

rea tions based on a model in whi h a tual earnings for a risky asset follow a random

walk, but investors do not understand this. They mistakenly believe that the earnings

pro ess sto hasti ally u tuates between a regime with mean-reverting earnings, and a

regime with expe ted earnings growth.

If re ent earnings hanges reverse, investors erroneously believe the �rm is in a mean-

reverting state, and underrea t to re ent news, onsistent with onservatism (Se tion

I.1.4). If investors see a sequen e of growing earnings, they tend to on lude (wrongly)

that the �rm is in a growth regime, and overextrapolate trends, whi h is arguably

reminis ent of representativeness (Se tion I.1.3). Overrea tion to a long enough trend

implies subsequent low returns during the pro ess of orre tion. Thus, there an be long-

term overrea tion and orre tion, implying negative long-lag return auto orrelation. Yet

the average response to an initial impulse an be smooth, implying positive short lag

auto orrelation. Similarly, the model an a omodate a positive short-term orrelation

84The impli ation of attribution/over on�den e models for whether there should be something akinto a disposition e�e t (holding winners, selling losers) is not obvious. When a sto k is �rst be ominga winner, rational arbitrageurs who foresee further pri e rises should drive the pri e up even higherthan the over on�dent think is justi�ed. This en ourages the over on�dent to sell, onsistent with thedisposition e�e t. However, for a sto k that has been a winner for some time, the arbitrageurs will sellto the over on�dent as the pri e peaks. Other re ent models of momentum and reversal have similaropposing e�e ts.

38

between the asset return and an earnings hange, and a negative long-term orrelation. If

sporadi events su h as dividend initiations are viewed as isolated from earnings patterns,

a single-event version of the model applies implies, under appropriate parameter values,

underrea tion.

Cross-se tional e�e ts (su h as a value e�e t) are simulated with earnings that are

independently distributed a ross sto ks. This implies a nearly riskfree arbitrage oppor-

tunity for a rational investor who buys and sells sto ks based on return predi tors. Su h

arbitrage would be risky in a setting where investors update their beliefs about system-

ati fa tors in earnings trends or reversals. The psy hologi al literature on multiple ue

learning (Se tion I.1) may provide guidan e for su h a model.

Intera tions among Traders with Di�erent Biases

Hong and Stein (1999) (HS) analyze a market in whi h, as in Cutler, Poterba, and

Summers (1990), some traders rea t sluggishly, and others trade based on positive feed-

ba k. Ea h group of traders is risk averse, and is able to pro ess only a subset of available

information. Information about the liquidating dividend dribbles into the hands of dif-

ferent groups of newswat hers. Newswat hers ondition on their own private signals but

ignore market pri es, ausing underrea tion.

Momentum traders, in ontrast, ondition on the umulative pri e hange over the

last k periods. Ea h trader takes a �xed position for a given number of periods. Mo-

mentum traders exploit the underrea tion of newswat hers by buying in response to

pri e in reases. This a elerates the rea tion to news, but also auses overshooting.

The smoothness of the overrea tion pro ess auses positive short-lag and negative long-

lag auto orrelation. Slower information di�usion tends to laun h a more powerful an

overrea tion, leading to more negative long-lag auto orrelations.

Other Errors in the Dynami s of Beliefs

Although it is impossible to be omprehensive, I brie y mention some other ap-

proa hes to the dynami s of beliefs.85 Shefrin (1997) dis usses how base rate under-

weighting may shed light on the anomalous behavior of implied volatilities in options

markets. Ce hetti, Lam, and Mark (1999) model the equity premium puzzle and re-

lated issues as arising from a ombination of errors, in luding underestimation of the

85Kurz (1997) des ribes his theory of endogenous un ertainty and rational belief equilibrium, whi hfo uses on sets of beliefs that annot be reliably ontradi ted by existing data. However, Bayesianupdating has greater appeal as a theory of rational de isions.

39

persisten e of high versus low onsumption growth regimes. They des ribe a rule-of-

thumb al ulation method that lead to su h errors, but do not address whether other

rule-of-thumb methods would imply the opposite error.

Informal arguments about money illusion a�e ting pri es have been o�ered by several

authors.86 Investors subje t to money illusion may dis ount real ash ows at nominal

interest rates, ausing overdis ounting during high-in ation (growing in ation?) periods.

They also may fail to take into a ount that higher in ation redu es the real value of a

�rm's debt. Ritter and Warr (2001) provide eviden e suggesting that in ation illusion

ontributed to the 1982-99 bull market.

III.2.4 Alternative Preferen es

Psy hologi al eviden e does not support the traditional assumption of time-additive

expe ted utility. Theorists, often motivated more by puzzling se urities pri e eviden e

than by psy hologi al eviden e, have o�ered models based upon alternative preferen es.

Alternative preferen e models an address the equity premium puzzle, the interest rate

puzzle and ex ess sto k market volatility in at least two ways. First, by breaking the

link between risk aversion and the intertemporal elasti ity of substitution, a high equity

risk premium (whi h demands high aversion to risk) an be re on iled with low interest

rates (whi h demand reasonably high intertemporal elasti ity of substitution). Se ond,

by allowing risk aversion to vary sto hasti ally, sto k pri e volatility an be in reased

relative to onsumption variability.

Several papers address the equity premium and riskfree rate puzzles applying habit-

forming preferen es.87 Constantinides (1990) showed that habit formation (Subse tion

I.5) re on iles a high equity premium with realisti onsumption smoothness and growth,

and moderate levels of risk aversion. Campbell and Co hrane (1999) and Chan and Ko-

gan (2000) �nd that habit preferen es that involve a Veblen-like on ern for onsumption

of others imply sto hasti risk aversion, whi h an re on ile a variety of fa ts about �rst

and se ond moments of returns and onsumption.

Several papers apply aspe ts of prospe t theory and �rst-order risk averse prefer-

en es.88 Benartzi and Thaler (1995) onsider investors who make a sequen e of my-

86Fisher (1928), Modigliani and Cohn (1979), Ritter and Warr (2001), Sharpe (1999).87See, e.g., Abel (1990), Boldrin, Christiano, and Fisher (1997), Campbell and Co hrane (1999),

Constantinides (1990), Ferson and Constantinides (1991), Heaton (1995), and Sundaresan (1989).88See Ang, Bekaert, and Liu (2000), Barberis, Huang, and Santos (2001), Barberis and Huang (2000),

Bekaert, Hodri k, and Marshall (1997a), Benartzi and Thaler (1995), Epstein and Zin (1990, 1991, 1993),Gomes (2000), and Shumway (1998).

40

opi single-period porfolio de isions. Consistent with loss-aversion, investors are about

hanges in wealth or onsumption relative to a referen e point that shifts from de ision

to de ision, and their value fun tion is kinked at the referen e point. Investors therefore

are highly averse to risks of short term losses in sto ks relative to bonds.

Shumway (1998) extends this approa h to explain the ross-se tion of expe ted re-

turns as well as the market expe ted return. Consistent with prospe t theory, he assumes

a modi�ed power utility fun tion that implies risk aversion over gains and risk seeking

over losses. The referen e point is a zero market return. In onsequen e, small market

returns ause relatively large hanges in the sto hasti dis ount fa tor. In equilibrium

sto k pri es are a linear fun tion of the sto k's up-side beta and its down-side beta.

Empiri ally, Shumway �nds that the model does quite well in �tting both the equity

premium puzzle and the ross-se tion of se urity returns. He suggests that the high

premium on equity results from loss aversion, whi h auses marginal utility to vary

more with slightly negative market returns. This tends to magnify the e�e t of sto ks'

downside risk relative to that of bonds.

Barberis, Huang, and Santos (2001) (BHS) o�er a model based on a ombination of

loss aversion, and the `house money' e�e t of Thaler and Johnson (1990), the tenden y

for individuals who have experien ed re ent gains to be less averse to risky gambles.

To apture loss aversion, they assume a pie e-wise linear value fun tion that is steeper

among losses than among gains relative to the referen e point. After the good news

of a high dividend, individuals be ome more risk tolerant. Sto hasti variation in risk

aversion in reases the volatility of returns relative to dividends. These u tuations in

risk aversion tend to reverse, ausing predi tability in sto k returns. The high return

variability raises the equity risk premium even without high aversion to onsumption

risk, and is therefore onsistent with a reasonably low riskfree rate.

Barberis and Huang (2000) (BH), like Shumway, examine the dynami s of loss aver-

sion with many risky se urities. BH onsider two kinds of mental a ounting. Under

individual sto k a ounting, investors are about total onsumption, but are also loss

averse over individual sto k movements. In the other, portfolio a ounting, individuals

are loss averse with respe t to movements in their total sto k portfolio.

Investors are also subje t to the house money e�e t. Using plausible parameter

values, under individual sto k a ounting the typi al individual sto k has a high expe ted

ex ess return, and its returns are variable relative to dividend variability. The ross-

se tion of returns is predi table using measures of size, value, and whether the �rm was

a winner or loser over the last three years. The model implies an even higher equity

41

premium than BHS be ause investors are loss-averse with respe t to the residual risk of

individual sto k movements.

In a broadly similar spirit, Epstein and Zin (1993) examine a �rst-order risk averse

setting and report that the ase of disappointment averse preferen es �t the data well

(see also Epstein and Zin (1990)). Bekaert, Hodri k, and Marshall (1997a) �nd that

�rst-order risk aversion an explain predi tability in U.S and Japan equity, bond and

foreign ex hange markets better than the expe ted utility model, but not enough to

mat h the data. Ang, Bekaert, and Liu (2000) �nd that a high U.S. equity premium is

onsistent with reasonable parameters of disappointment averse preferen es.

A rather di�erent approa h from appli ations of loss-averse or �rst-order risk averse

preferen es fo uses on aversion to ambiguity (Subse tion I.3.1) and a onsequent taste for

robustness.89 A robust de ision rule is one that does well in the fa e of model un ertainty

when Nature hooses the most adverse possible model in response to the individual's

hoi e. Tornell (2000) provides a model based on agents who hoose robust fore asting

te hniques to explain high equity returns, predi tability and ex ess volatility.

Even slight sto hasti shifts in preferen es an substantially in rease the volatility

of sto k pri es relative to the variability of onsumption (Allen and Gale (1994), Kraus

and Sagi (2000), and Mehra and Sah (2000)). The psy hologi al eviden e that vis eral

fa tors a�e t de isions are onsistent with su h variability.

III.2.5 Evolving Populations

A promising �eld for exploration uses evolutionary simulation of the intera tions of

agents in �nan ial markets. In the last �ve years, physi ists have begun to do resear h on

�nan ial markets, some alling their �eld e onophysi s (see Farmer (1999)). Some of the

re ent models by physi ists make su h radi al me hanisti assumptions about investor

behavior and market stru ture that the resulting insights seem unlikely to generalize.

Fortunately, a very promising strand of evolutionary literature explores the populations

of traders who are imperfe tly rational but do learn and make endogenous de isions.

Freed from the onstraints of analyti al tra tability, modellers are able to explore a

wider spa e of e onomi settings.

An evolutionary approa h ould address the argument that even though individuals

are imperfe tly rational, as they learn from their trading out omes the system will

progress toward the fully rational equilibrium rapidly. I onje ture that a simple tropism

89See, e.g., Hansen, Sargent, and Tallarini (1999) and Maenhout (2000).

42

among traders towards a tions that generate higher investment pro�ts will not onverge

to the ICAPM qui kly. Even with a long history of eviden e, it is hard for a trader

to �gure out whether a trading strategy has done well after adjusting for risk unless he

understands risk, and the in ome and substitution e�e ts of Merton ICAPM hedging are,

I onje ture, too subtle for most individual or even sophisti ated institutional investors.

Some brief re ent surveys of the omputational �eld in lude Farmer (1999), Farmer

and Lo (1999) and LeBaron (2000a). Some re ent �nding are that long-horizon investors

frequently do not drive shorter-horizon investors out of �nan ial markets, and that pop-

ulations of long- and short- horizon agents an reate patterns of volatility and volume

similar to a tual empiri al patterns (Lebaron (2000b, 2000 )).

III.3 Empiri ally Distinguishing Pri ing Theories

The e�e ts des ribed in di�erent psy hologi al pri ing theories need not be mutually

ex lusive, but it is useful to examine how their predi tions di�er. My fo us is on value,

momentum, and event-based e�e ts.

III.3.1 Distinguishing Explanations for Size and Value E�e ts

Several past authors have pointed out that long-run overrea tion will indu e ross-

se tional value e�e ts. Two re ent models derive ross-se tional value and size e�e ts

when se urities are subje t to systemati and idiosyn rati in uen es (Barberis and

Huang (2000) (BH), and Daniel, Hirshleifer, and Subrahmanyam (2001a) (DSH2)).

DHS2 provides no help in explaining the equity premium puzzle. The BH theory does

address the equity premium and asso iated puzzles, be ause people who do individual-

sto k a ounting are averse to the high residual risk of sto ks. Thus, a further impli ation

of BH is that residual risk is ross-se tionally pri ed (Brennan (2001)). Furthermore,

in ontrast with the Merton (1987) limited parti ipation theory, the BH theory seems

to imply that, eteris paribus, greater parti ipation by individual investors will in rease

the premium for residual risk.

DHS2 o�er further impli ations, largely untested, on erning the ross-se tional dis-

persion in fundamental/pri e ratios, and the ability of urrent volume to predi t future

return volatility. Another impli ation is that as on�den e exogenously varies over time,

the dispersion in se urity fundamental/pri e ratios varies together with the ability of

su h ratios to predi t future returns. Cohen, Polk, and Vuolteenaho (2000) on�rm su h

a relationship between the book/market `value spread' and the pro�tability of value

43

trading strategies.

In DHS2 mispri ing is present in a small number of fa tors. The importan e of

idiosyn rati e�e ts in the BH theory suggests that rational arbitrageurs should take

strong ontrarian positions and earn large expe ted pro�ts. More broadly, the strong

ow of wealth in the BH theory suggests that value e�e ts should be more transitory than

in DHS2. BH also point out that the rise of mutual and pension fund sto k investment

should have led to less individual sto k a ounting, and is therefore onsistent with a

weakening in size and value e�e ts.

There is psy hologi al eviden e that over on�den e is strongest when information

signals are less pre ise and when feedba k is in on lusive (e.g., Einhorn (1980), Grif-

�n and Tversky (1992)). Thus, DHS2 predi ts that fundamental/pri e ratios should

fore ast risk-adjusted returns more strongly for businesses that are hard to value (e.g.,

R&D-intensive �rms omprised largely of intangible assets). Chan, Lakonishok, and

Sougiannis (1999) subsequently reported eviden e onsistent with su h a pattern.

Neither BH nor DHS2 apture momentum. The absen e of a uni�ed model that

dire tly aptures the two most onspi uous ross-se tional e�e ts, value and momentum,

is an obvious gap in the literature. The results of DHS1 and of Barberis, Shleifer, and

Vishny (1998) suggest that uni�ed explanations may be possible based upon either

over on�den e, or upon misper eptions of regime-shifting.

III.3.2 Distinguishing Explanations for Post-Event Continuation

The DHS1 analysis of post-event ontinuation di�ers from the BSV model in predi ting

ontinuation only for sele tive events taken by a party su h as management or an analyst

in response to market mispri ing. The support for this from one type of event (see Sub-

se tion II.1.4 at footnote 62) is intriguing. Event studies on other low-dis retion events

(su h as regulatory announ ements, input supply sho ks or output demand sho ks) pro-

vide an attra tive dire tion for further testing.

The BSV model is based on publi information. The DHS1 model implies nega-

tive long-run return auto orrelation asso iated with private information arrival. This

is onsistent with eviden e of Daniel and Titman (2000). DHS1 further predi ts that

post-event ontinuation will be strongest in sto ks about whi h investors have poor in-

formation (often illiquid or smaller sto ks). DHS1 also o�ers several untested predi tions

about about the o urren e of and pri e patterns around orporate events, and about

volatility at the time of private versus publi signals.

44

III.3.3 Distinguishing Explanations for Momentum and Reversal

Analyti ally, the three re ent models of how mistaken beliefs ause momentum and

reversals (Barberis, Shleifer, and Vishny (1998) or BSV; Daniel, Hirshleifer, and Sub-

rahmanyam (1998) or DHS1; and Hong and Stein (1999) or HS) all generate an impulse

response fun tion to a new information signal in whi h there is a gradual rise in the

average rea tion to a positive signal and a gradual average pro ess of orre tion.

In all these models, the misper eptions that drive momentum are also the drivers of

long-term reversal. These models therefore imply that those sets of sto ks with largest

momentum e�e ts should also have the largest reversal e�e ts. So it is interesting that

mu h of the empiri al eviden e of return predi tability, in luding both momentum and

reversal, is stronger in small �rms (see Fama (1998) and Loughran and Ritter (2000)).

More generally, greater un ertainty about a set of sto ks, and a la k of a urate

feedba k about their fundamentals, leaves more room for psy hologi al biases. At the

extreme, it is relatively hard to misper eive an asset that is nearly riskfree. Thus, the

misvaluation e�e ts of almost any mistaken-beliefs model should be strongest among

�rms about whi h there is high un ertainty/poor information ( ash ow varian e is

one possible proxy). Furthermore, in DHS1 and HS, greater information asymmetry

strengthens the predi ted e�e ts; the adverse sele tion omponent of the bid-ask spread

is a possible proxy. BSV does not have impli ations based on information asymmetry.

Firm size, analyst following, and dispersion in analyst fore asts are potential proxies for

information asymmetry, but they also may proxy for mere un ertainty. Thus, eviden e

that small �rms (internationally) and �rms with low analyst following have greater

momentum is onsistent with, but does not sharply distinguish, the three models.

BSV predi t overrea tion to trends, whi h an also o ur in DHS1, but it is not

obvious that the DHS1 impli ation extends to zero net supply se urities. Thus, the

eviden e of Poteshman (2000) of daily underrea tion and multiple-day overrea tion of

option pri es to shifts in volatility supports BSV (at a very di�erent time horizon).

Bloom�eld and Hales (2001) dire tly test the BSV theory that people misper eive

random walks to be shifts between ontinuation and reversal regimes by examining

predi tions by MBA-student experimental subje ts. Consistent with BSV, subje ts

overrea ted to hanges pre eded by sequen es of ontinuations, and underrea ted to

hanges pre eded by many reversals. However, people on average tended to expe t re-

versal, whereas a per eived tilt toward ontinuation is needed to obtain post-earnings

announ ement drift and post-event return ontinuation.

45

Another testing approa h is to �nd datasets in whi h the trades of irrational traders

versus rational arbitrageurs an be identi�ed.90 Coval and Shumway (2000) analyze a

ri h database to des ribe how the positions of futures market-makers hanges following

re ent trading su ess. Another suggestion has been to view market orders as irra-

tional and limit orders as rational (Hvidkjaer (2000)). However, it does not seem lear

why this would be the ase based on these theories, and empiri ally it is limit order

traders who lose money (Chordia, Roll, and Subrahmanyam (2000)). Further progress

on mi rostru ture testing of these models alls for expli it modelling of psy hology and

mi rostru ture.

IV Con lusion

Man is neither in�nite in fa ulties, nor in apprehension like a god. Nor is human fallibility

shed at the doorstep of the sto k ex hange. Psy hology-based asset pri ing theory has

promise of apturing this reality, though at this point we are at an early stage.

Finan ial e onomists have grown more re eptive to entertaining psy hologi al expla-

nations. One sign of this is the popularity of utility fun tions that seem to violate time

onsisten y or the rationality axioms of expe ted utility in re ent literature on the equity

premium and riskfree rate puzzles. Some of these preferen es ould be endogenized as

redu ed form summaries of rational settings with market fri tions, but this does not

seem to be a high resear h priority even among fans of the full-rationality approa h.

In Se tion I I tried to give some hint of the wealth of psy hologi al �ndings, many

utterly unexploited, that an inform �nan ial modelling. In Subse tion III.3.2 I o�ered

hints for empiri al work to distinguish alternative psy hology-based pri ing theories. I

now mention a few other possible theoreti al and empiri al dire tions.

1. So far few psy hology-based asset pri ing models allow for both risk aversion and

multiple risky se urities. It will be useful to explore the dynami s of mistaken

beliefs when there is a ross-se tion of se urities, to address su h issues as volume

as a predi tor of returns, and the e�e ts of di�erent rates of overrea tion and

orre tion for fa tors and residuals.

90Conje turally, the DHS approa h implies that rational arbitrageurs buy after a pri e in rease (fore-seeing further overrea tion). Over on�dent traders sell (as implied by market- learing; be ause the arbsdrive pri es up even higher than justi�ed based on urrent over on�dent beliefs). Some period of timeafter the favorable impulse, the arbs tend to sell out to the over on�dent, and to go short. Anti ipationby arbs of overrea tion should generate similar trading patterns in BSV; in the HS setting the behaviorof irrational traders is more omplex.

46

2. Pri ing models based on loss- and disappointment aversion an be viewed as re-

e ting a on ern about future feelings. But more dire tly, the e�e t of urrently-

experien ed emotions on urrent pri es merits analysis.

3. It is often not obvious how to translate pre-existing eviden e from psy hologi al

experiments into assumptions about investors in real �nan ial settings. Routine

experimental testing of the assumptions and on lusions of asset pri ing theories

is needed to guide modelling.

4. We la k a quanti�ed set of apital budgeting and risk management pro edures that

re e t mispri ing and are ready for pra titioners to apply (but see Stein (1996)).

5. The great missing hapter in asset pri ing theory, I believe, is a model of the so ial

pro ess by whi h people form and transmit ideas about markets and se urities. In

addition to studying what in uen es individuals' valuations, an appealing dire tion

is to study how attention is fo used on ertain groups of sto ks, and the e�e ts

of resulting swings in parti ipation. A di�erent empiri al dire tion is to analyze

the spe i� ontent of widespread, erroneous investor theories to identify ways of

predi ting returns. Robert Shiller has dis ussed and do umented investor theories,

belief transmission, and e�e ts on pri ing (e.g., Shiller (1984, 1990, 2000 ); see also

the analysis of DeMarzo, Vayanos, and Zwiebel (2000)). This resear h has blazed

a path upon whi h further work will follow.

My list of further dire tions is ne essarily idiosyn rati . In an area that is just

oming of age, many new prospe ts are open. This is an ex iting time for the �eld of

asset pri ing.

47

Referen es

Abarbanell, J. S., and B. J. Bushee, 1998, Abnormal returns to a fundamental strategy,A ounting Review 73, 19{45.

Abel, A. B., 1990, Asset pri es under habit formation and at hing up with the Joneses,Ameri an E onomi Review 80, 38{42.

Aggarwal, R., S. Mohanty, and F. Song, 1995, Are survey fore asts of ma roe onomi variables rational?, Journal of Business 68, 99{119.

Ahn, D.-H., J. S. Conrad, and R. F. Dittmar, 2000, Risk adjustment and trading strate-gies, Working paper, University of North Carolina at Chapel Hill.

Ainslie, G., 1975, Spe ious reward: A behavioral theory of impulsiveness and impulse ontrol, Psy hologi al Bulletin 82, 463{496.

Allais, M., 1953, La psy hologie de l'home rationnel devant le risque: ritique des pos-tulats et axiomes de l' ole amri aine, E onometri a 21, 503{46.

Allen, F., 2001, Do �nan ial institutions matter?, Journal of Finan e 56.

Allen, F., and D. Gale, 1994, Limited market parti ipation and volatility of asset pri es,Ameri an E onomi Review 84, 933{955.

Allport, G. W., and L. J. Postman, 1947, The psy hology of rumor (Holt, New York).

Alpert, M., and H. Rai�a, 1982, A progress report on the training of probability as-sessors, in Daniel Kahneman, Paul Slovi , and Amos Tversky, ed.: Judgement underUn ertainty: Heuristi s and Biases (Cambridge University Press, Cambridge).

Anderson, T. W., 1932, Remembering: A study in experimental and so ial psy hology(Cambridge University Press, London York).

Andrade, G. M., 1999, Do appearan es matter? the impa t of eps a retion and dilutionon sto k pri es, Harvard University Working Paper.

Andreassen, P. B., and S. J. Kraus, 1990, Judgemental extrapolation and the salien eof hange, Journal of Fore asting 9, 347{372.

Ang, A., G. Bekaert, and J. Liu, 2000, Why sto ks may disappoint, Working paper,Columbia Business S hool.

Argote, L., M. Turner, and M. Fi hman, 1988, To entralize or not to entralize: thee�e ts of un ertainty and threat on group stru turre and performan e, OrganizationalBehavior 42, 1{17.

Arkes, H., L. Herren, and A. Isen, 1988, The role of potential loss in the in uen e of a�e ton risk-taking behavior, Organizational Behavior and Human De ision Pro esses 66,228{236.

48

Arkes, H. R., and C. Blumer, 1985, The psy hology of sunk osts, Organizational Be-havior and Human De ision Pro esses 35, 124{140.

As h, S., 1956, Studies of independen e and onformity: A minority of one against aunanimous majority, Psy hologi al Monographs 70 Whole No. 416.

Ashton, R. H., 1976, Cognitive hanges indu ed by a ounting hanges: Experimentaleviden e on the fun tional �xation hypothesis, Journal of A ounting Resear h 14, 1.

Avery, C., and J. Chevalier, 1999, Identifying investor sentiment from pri e paths: The ase of football betting, Journal of Business 72, 493{521.

Avery, C., and P. Zemsky, 1998, Multi-dimensional un ertainty and herd behavior in�nan ial markets, Ameri an E onomi Review 88, 724{48.

Ball, R., 1978, Anomalies in relationships between se urities' yields and yield-surrogates,Journal of Finan ial E onomi s 6, 103{126.

Ball, R., and S. P. Kothari, 1989, Non-stationary expe ted returns: Impli ations for testsof market eÆ ien y and serial orrelations of returns, Journal of Finan ial E onomi s25, 51{74.

Ball, R., S. P. Kothari, and J. Shanken, 1995, Problems in measuring portfolio per-forman e: An appli ation to ontrarian investment strategies, Journal of Finan ialE onomi s 38, 79{107.

Balvers, R., Y. Wu, and E. Gilliland, 2000, Mean reversion a ross national sto k marketsand parametri ontrarian investment strategies, Journal of Finan e 55, 745 { 772.

Banerjee, A., 1992, A simple model of herd behavior, Quarterly Journal of E onomi s107, 797{817.

Banerjee, A., and D. Fudenberg, 1999, Word-of-mouth ommuni ation and so ial learn-ing, Harvard University Working Paper.

Banz, R. W., 1981, The relationship between return and the market value of ommonsto ks, Journal of Finan ial and Quantitative Analysis 14, 421{441.

Bar-Hillel, M., and E. Neter, 1996, Why are people relu tant to ex hange lottery ti kets?,Journal of Personality and So ial Psy hology 70, 17{27.

Barber, B., R. Lehavy, M. M Ni hols, and B. Trueman, 2001, Can investors pro�tfrom the prophets? onsensus analyst re ommendations and sto k returns, Journalof Finan e.

Barber, B., and T. Odean, 1999, Online investors: Do the slow die �rst?, UC Davis.

Barber, B., and T. Odean, 2000a, Boys will be boys: Gender, over on�den e, and ommon sto k investment, Quarterly Journal of E onomi s, forth oming.

49

Barber, B., and T. Odean, 2000b, Trading is hazardous to your wealth: The ommonsto k investment performan e of individual investors, Journal of Finan e 55, 773{806.

Barberis, N., and M. Huang, 2000, Mental a ounting, loss aversion, and individualsto k returns, University of Chi ago Manus ript.

Barberis, N., M. Huang, and J. Santos, 2001, Prospe t theory and asset pri es, forth- oming Quarterly Journal of E onomi s.

Barberis, N., A. Shleifer, and R. Vishny, 1998, A model of investor sentiment, Journalof Finan ial E onomi s 49, 307{343.

Barkham, R., and D. Geltner, 1995, Pri e dis overy in ameri an and british propertymarkets, Real Estate E onomi s 23, 21{44.

Barkow, J. H., L. Cosmides, and J. Tooby, 1992, The Adapted Mind: EvolutionaryPsy hology and the Generation of Culture (Oxford University Press, New York).

Barone, M., P. Miniard, and J. Romeo, 2000, The in uen e of positive mood on brandextension evaluations, Journal of Consumer Resear h 26, 386{400.

Bartov, E., F. W. Lindahl, and W. E. Ri ks, 1998, Sto k pri e behavior around an-noun ements of writeo�s, Review of A ounting Studies 3, 327{346.

Basu, S., 1983, The relationship between earnings yield, market value, and return forNYSE ommon sto ks: Further eviden e, Journal of Finan ial E onomi s 12, 126{156.

Bat helor, R., and P. Dua, 1992, Conservatism and onsensus-seeking among e onomi fore asters, Journal of Fore asting 11, 169{181.

Bekaert, G., R. J. Hodri k, and D. Marshall, 1997a, The impli ations of �rst-order riskaversion for asset market risk premiums, Journal of Monetary E onomi s 40, 3{39.

Bekaert, G., R. J. Hodri k, and D. Marshall, 1997b, `peso problem' explanations forterm stru ture anomalies, NBER Working Paper No. W6147.

Bem, D. J., 1972, Self-per eption theory, in L. Berkowitz, ed.: Advan es in ExperimentalSo ial Psy hology 6, 1{62 (A ademi Press, New York).

Benartzi, S., 1997, Ex essive extrapolation and the allo ation of 401(k) a ounts to ompany sto k, Journal of Finan e.

Benartzi, S., and R. Thaler, 1995, Myopi loss asversion and the equity premium puzzle,Quarterly Journal of E onomi s 110, 75{92.

Benartzi, S., and R. H. Thaler, 2001, Naive diversi� ation strategies in retirement savingplans, Ameri an E onomi Review.

50

Benos, A., 1998, Aggressiveness and survival of over on�dent traders, Journal of Finan- ial Markets 1, 353{83.

Berk, J., 1995, A ritique of size related anomalies, Review of Finan ial Studies 8,275{286.

Bernard, V. L., and J. K. Thomas, 1989, Post-earnings-announ ement drift: Delayedpri e response or risk premium?, Journal of A ounting Resear h, Supplement 27,1{48.

Bernard, V. L., and J. K. Thomas, 1990, Eviden e that sto k pri es do not fully re e tthe impli ations of urrent earnings for future earnings, Journal of A ounting andE onomi s 13, 305{340.

Bernardo, A., and I. Wel h, 2000, On the evolution of over on�den e and entrepreneurs,UCLA Working Paper #9-97.

Bers heid, E., and H. Reis, 1998, Attra tion and lose relationships, in S. FiskeD. Gilbert, and G. Lindzey, ed.: The Handbook of So ial Psy hology, 4rth edition2, 193{281 (M Graw Hill, New York).

Bhandari, L. C., 1988, Debt/equity ratios and expe ted ommon sto k returns: Empir-i al eviden e, Journal of Finan e 43, 507{528.

Bikh handani, S., D. Hirshleifer, and I. Wel h, 1992, A theory of fads, fashion, ustom,and ultural hange as informational as ades, Journal of Politi al E onomy 100,992{1026.

Bjerring, J. H., J. Lakonishok, and T. Vermaelen, 1983, Sto k pri es and �nan ial ana-lysts' re ommentations, Journal of Finan e 38, 187{204.

Bla k, F., M. Jensen, and M. S holes, 1972, The apital asset pri ing model: Someempiri al tests, in Mi hael C. Jensen, ed.: Studies in the Theory of Capital Marketspp. 79{121 (Praeger, New York).

Bloom�eld, R., 1996, Quotes, pri es and estimates of value in a laboratory market,Journal of Finan e 51, 1791{1808.

Bloom�eld, R., and J. Hales, 2001, Predi ting the next step of a random walk: Experi-mental eviden e of regime-shifting biases, Johnson S hool, Cornell University.

Bloom�eld, R., J. Hales, R. Libby, and M. Nelson, 2001, The e�e t of information,strength and weight on behavior in �nan ial markets, Organizational Behavior andHuman De ision Pro esses, forth oming.

Bloom�eld, R., R. Libby, and M. Nelson, 1999, Con�den e and the welfare of less-informed investors, A ounting, Organization and So iety 24, 623{647.

51

Bloom�eld, R., R. Libby, and M. Nelson, 2000, Underrea tions, overrea tions and mod-erated on�den e, Journal of Finan ial Markets 3, 113{137.

Blume, L., and D. Easley, 1982, Learning to be rational, Journal of E onomi Theory26, 340{351.

Blume, L., and D. Easley, 1990, Evolution and market behavior, Journal of E onomi Theory 58, 9{40.

Blume, L. E., and D. Easley, 2000, If you're so smart, why aren't you ri h? belief sele -tion in omplete and in omplete markets, Working paper, Department of E onomi s,Cornell University.

Bodurtha, J., D.-S. Kim, and C. M. C. Lee, 1995, Closed-end ountry funds and U.S.market sentiment, Review of Finan ial Studies 8, 879{918.

Boehme, R., and S. M. Sores u, 2000, Seven de ades of long term abnormal returnpersisten e: The ase of dividend initiations and resumptions, University of HoustonWorking Paper.

Boldrin, M., L. Christiano, and J. Fisher, 1997, Asset pri ing lessons for modelingbusiness y les, unpublished manus ript.

Bond, R., and P. Smith, 1996, Culture and onformity: A meta-analysis of studies usingas hs (1952b, 1956) line judgement task, Psy hologi al Bulletin 119, 111{137.

Bornstein, R., and P. D'Agostino, 1992, Stimulus re ognition and the mere exposuree�e t, Journal of Personality and So ial Psy hology 63, 545{552.

Bossaerts, P., 2000, Experiments with �nan ial markets: Impli ations for asset pri ingtheory, Cal Te h Working Paper.

Bossaerts, P., and P. Hillion, 1999, Implementing statisti al riteria to sele t returnfore asting models: what do we learn?, Review of Finan ial Studiess 12, 405{428.

Bossaerts, P., C. Plott, and W. Zame, 2000, Pri es and allo ations in �nan ial markets:Theory and eviden e, Cal Te h Working Paper.

Brav, A., C. Ge zy, and P. A. Gompers, 2000, Is the abnormal return following equityissuan es anomalous?, Journal of Finan ial E onomi s 56, 209{249.

Brennan, M., 2001, Mental a ounting, loss aversion, and individual sto k returns: Dis- ussion, Anderson Graduate S hool of Management, UCLA.

Brennan, M., T. Chordia, and A. Subrahmanyam, 1998, Alternative fa tor spe i� ations,se urity hara teristi s and the ross-se tion of expe ted sto k returns, Journal ofFinan ial E onomi s 49, 345{373.

Brennan, M., and Y. Xia, 2001, Sto k pri e volatility and the equity premium, Journalof Monetary E onomi s forth oming.

52

Brenner, L. A., D. J. Koehler, V. Liberman, and A. Tversky, 1996, Over on�den e inprobability and frequen y judgments: A riti al examination, Organizational Behaviorand Human De ision Pro esses 65, 212{219.

Brenner, L. A., D. J. Koehler, and A. Tversky, 1996, On the evaluation of one-sidedeviden e, Journal of Behavioral De ision Making 9, 59{70.

Brown, S. J., W. N. Goetzmann, and S. A. Ross, 1995, Survival, Journal of Finan e 50,853{873.

Bruner, J., L. Postman, and J. Rodrigues, 1951, Expe tations and the per eption of olor, Ameri an Journal of Psy hology 64, 216{227.

Camerer, C., 1995, Individual de ision making, in John H. Kagel, and Alvin E. Roth,ed.: The Handbook of Experimental E onomi s pp. 587{703 (Prin eton UniversityPress).

Camerer, C. F., 1991, The pro ess-performan e paradox in expert judgment: How anexperts know so mu h and predi t so badly?, in K. A. Eri sson, and J. Smith, ed.:Towards a general theory of expertise: Prospe ts and Limits pp. 195{217 (CambridgeUniversity Press, New York).

Camerer, C. F., 1998, Bounded rationality in individual de ision making, ExperimentalE onomi s 1, 163{183.

Camerer, C. F., G. Loewenstein, and M. Weber, 1989, The urse of knowledge in e o-nomi settings: An experimental analysis, Journal of Politi al E onomy 97, 1232{1254.

Campbell, J., 1999, Asset pri es, onsumption, and the business y le, in John B. Tay-lor, and Mi hael Woodford, ed.: Handbook of Ma roe onomi s (Amsterdam: NorthHolland).

Campbell, J. Y., 1987, Sto k returns and the term stru ture, Journal of Finan ial E o-nomi s 18, 373{399.

Campbell, J. Y., 2000, Asset pri ing at the millenium, Journal of Finan e.

Campbell, J. Y., and J. H. Co hrane, 1999, By for e of habit: A onsumption basedexplanation of aggregate sto k market behavior, Journal of Politi al E onomy 107,205{251.

Campbell, J. Y., and A. S. Kyle, 1993, Smart money, noise trading, and sto k pri ebehaviour, Review of E onomi Studies 60, 1{34.

Campbell, J. Y., and R. J. Shiller, 1988a, The dividend-pri e ratio and expe tations offuture dividends and dis ount fa tors, Review of Finan ial Studies 1, 195{228.

53

Campbell, J. Y., and R. J. Shiller, 1988b, Sto k pri es, earnings, and expe ted dividends,Journal of Finan e 43, 661{676.

Campbell, J. Y., and R. J. Shiller, 1991, Yield spreads and interest rates: A bird's eyeview, Review of E onomi Studies 58, 495{514.

Cao, H. H., and D. Hirshleifer, 2000, Conversation, learning and informational as ades,Ohio State University Fisher College of Business Working Paper.

Carmel, J. P., and M. Young, 1997, Long horizon mean reversion in the sto k market:The postwar years, Working paper, University of Mi higan Business S hool.

Case, K. E., and R. Shiller, 1990, The behavior of home buyers in boom and post-boommarkets, New England E onomi Review 80, 29{46.

Ce hetti, S. G., P.-S. Lam, and N. C. Mark, 1999, Asset pri ing with distorted beliefs:Are equity returns `too good to be true?', Ameri an E onomi Review, forth oming.

Chan, K., 1988, On the ontrarian investment strategy, Journal of Business 61, 147{163.

Chan, K., and N.-F. Chen, 1991, Stru tural and return hara teristi s of small and large�rms, Journal of Finan e 46, 1467{1484.

Chan, K., A. Hameed, and W. Tong, 2000, Pro�tability of momentum strategies inthe international equity markets, Journal of Finan ial and Quantitative Analysis 35,153{172.

Chan, K., N. Jegadeesh, and J. Lakonishok, 2000, Earnings quality and sto k returns:The eviden e from a ruals, University of Illinois.

Chan, L., Y. Hamao, and J. Lakonishok, 1991, Fundamental and sto k returns in japan,Journal of Finan e 46, 1739{1764.

Chan, L. K., N. Jegadeesh, and J. Lakonishok, 1996, Momentum strategies, Journal ofFinan e 51, 1681{1714.

Chan, L. K., J. Lakonishok, and T. Sougiannis, 1999, The sto k market valuation ofresear h and development expenditures, NBER Working Paper #7223, July.

Chan, Y. L., and L. Kogan, 2000, Cat hing up with the joneses: Heterogeneous prefer-en es and the dynami s of asset pri es, Wharton S hool #14-00.

Chang, C., Z. Chen, and M. Dong, 1999, Investing with a sto k valuation model, Workingpaper, Ohio State University and Yale University.

Chapman, G. B., 1998, Sooner or later: The psy hology of intertemporal hoi e, in Dou-glas L. Medin, ed.: The Psy hology of Learning and Motivation 38, 83{113 (A ademi Press, San Diego, CA).

54

Chapman, G. B., and E. J. Johnson, 1999, An horing, a tivation, and the onstru tionof values, Organizational Behavior and Human De ision Pro esses 79, 115{153.

Chen, J., 2000, Can the intertemporal apm explain the ross-se tion of sto k returns,Stanford University working paper.

Chen, N.-F., and F. Zhang, 1998, Risk and return of value sto ks, Journal of Business71, 501{535.

Choe, H., B.-C. Kho, and R. M. Stulz, 1999, Do foreign investors destabilize sto kmarkets? the korean experien e in 1997, Journal of Finan ial E onomi s 54, 227{264.

Chopra, N., J. Lakonishok, and J. R. Ritter, 1992, Measuring abnormal performan e:Do sto ks overrea t?, Journal of Finan ial E onomi s 31, 235{268.

Chordia, T., R. Roll, and A. Subrahmanyam, 2000, Order imbalan e, liquidity, andmarket returns, UCLA working paper.

Chordia, T., and L. Shivakumar, 2000, Momentum, business y le and time varyingexpe ted returns, Emory University Working Paper.

Clore, G. L., N. S hwarz, and M. Conway, 1994, A�e tive auses and onsequen esof so ial information pro essing, in Jr. Robert S. Wyer, and Thomas K. Srull, ed.:Handbook of So ial Cognition, 2nd edition (Lawren e Erlbaum, Hillsdale, NJ).

Clotfelter, C. T., and P. J. Cook, 1993, The `gambler's falla y' in lottery play, Manage-ment S ien e 39, 93{95.

Co hrane, J. H., 1991, Volatility tests and eÆ ient markets: Review essay, Journal ofMonetary E onomi s 27, 463{485.

Co hrane, J. H., 2000, Asset Pri ing (Prin eton University Press, NJ).

Cohen, J. B., and C. S. Areni, 1991, A�e t and onsumer behavior, in Thomas S.Robertson, and Harold H. Kassarjian, ed.: Handbook of Consumer Behavior pp. 188{240 (Prenti e-Hall, Englewood Cli�s, NJ).

Cohen, R. B., C. K. Polk, and T. Vuolteenaho, 2000, The value spread, University ofChi ago working paper.

Conlisk, J., 1996, Why bounded rationality?, Journal of E onomi Literature 34, 669{700.

Constantinides, G. M., 1990, Habit formation: A resolution of the equity premiumpuzzle, Journal of Politi al E onomy 98, 519{543.

Constantinides, G. M., J. B. Donaldson, and R. Mehra, 2000, Junior an't borrow: Anew perspe tive on the equity premium puzzle, Columbia University Working Paper.

55

Constantinides, G. M., and D. DuÆe, 1996, Asset pri ing with heterogeneous onsumers,Journal of Politi al E onomy 104, 219{240.

Cooper, I., and E. Kaplanis, 1994, Home bias in equity portfolios, in ation hedging, andinternational apital market equilibrium, Review of Finan ial Studies 7, 45{60.

Cooper, J., and R. H. Fazio, 1984, A new look at dissonan e theory, in L. Berkowitz, ed.:Advan es in experimental so ial psy hology 17, 229{264 (A ademi Press, Orlando,FL).

Cooper, M. J., O. Dimitrov, and P. R. Rau, 2000, A rose. om by any other name, Journalof Finan e, forth oming.

Cornell, B., and Q. Liu, 2000, The parent ompany puzzle: When is the whole worthless than one of its parts?, Anderson S hool Working Paper #7-00, UCLA.

Cornett, M. M., H. Mehran, and H. Tehranian, 1998, Are �nan ial markets overlyoptimisti about the prospe ts of �rms that issue equity? Eviden e from voluntaryversus involuntary equity issuan e by banks, Journal of Finan e 53.

Cosmides, L., and J. Tooby, 1996, Are humans good intuitive statisti ians after all?rethinking some on lusions from the literature on judgment under un ertainty, Cog-nition 58, 1{73.

Coval, J. D., and T. J. Moskowitz, 1999, Home bias at home: Lo al equity preferen ein domesti portfolios, Journal of Finan e 54, 145{166.

Coval, J. D., and T. Shumway, 2000, Do behavioral biases a�e t pri es?, University ofMi higan Business S hool.

Cro ker, J., 1982, Biased questions in judgement of ovariation studies, Personality andSo ial Ps y hology Bulletin 8, 214{220.

Cusatis, P., J. Miles, and J. Woolridge, 1993, Restru turing through spino�s the sto kmarket eviden e, Journal of Finan ial E onomi s 33, 293{311.

Cutler, D. M., J. M. Poterba, and L. H. Summers, 1989, What moves sto k pri es?,Journal of Portfolio Management 15, 4{12.

Cutler, D. M., J. M. Poterba, and L. H. Summers, 1990, Spe ulative dynami s and therole of feedba k traders, Ameri an E onomi Review 80, 63{68.

Cutler, D. M., J. M. Poterba, and L. H. Summers, 1991, Spe ulative dynami s, Reviewof E onomi Studies 58, 529{546.

Daniel, K. D., D. Hirshleifer, and A. Subrahmanyam, 1998, Investor psy hology andse urity market under- and over-rea tions, Journal of Finan e 53, 1839{1886.

Daniel, K. D., D. Hirshleifer, and A. Subrahmanyam, 2001a, Covarian e risk, mispri ing,and the ross se tion of se urity returns, Journal of Finan e 56 forth oming.

56

Daniel, K. D., D. Hirshleifer, and A. Subrahmanyam, 2001b, Dis riminating between ef-� ient and ineÆ ient markets theories of ross-se tional return predi tability, WorkingPaper, Northwestern University.

Daniel, K. D., and S. Titman, 1997, Eviden e on the hara teristi s of ross-se tionalvariation in ommon sto k returns, Journal of Finan e 52, 1{33.

Daniel, K. D., and S. Titman, 1999, Market eÆ ien y in an irrational world, Finan ialAnalysts' Journal 55, 28{40.

Daniel, K. D., and S. Titman, 2000, Market rea tions to tangible and intangible infor-mation, Kellogg S hool.

Daniel, K. D., S. Titman, and J. Wei, 2001, Cross-se tional variation in ommon sto kreturns in Japan, Journal of Finan e.

Davis, J., E. F. Fama, and K. R. Fren h, 2000, Chara teristi s, ovarian es, and averagereturns: 1929-1997, Journal of Finan e 55, 389{406.

Davis, J. L., 1994, The ross-se tion of realized sto k returns: The pre-COMPUSTATeviden e, Journal of Finan e 50, 1579{1593.

Deaux, K., and T. Emswiller, 1974, Explanations for su essful performan e on sex-linked tasks: What is skill for the male is lu k for the female, Journal of Personalityand So ial Psy hology 29, 80{85.

DeBondt, W. F. M., 1993, Betting on trends: Intuitive fore asts of �nan ial risk andreturn, International Journal of Fore asting 9, 355{371.

DeBondt, W. F. M., and R. H. Thaler, 1985, Does the sto k market overrea t?, Journalof Finan e 40, 793{808.

DeBondt, W. F. M., and R. H. Thaler, 1987, Further eviden e on investor overrea tionand sto k market seasonality, Journal of Finan e 42, 557{581.

DeBondt, W. F. M., and R. H. Thaler, 1995, Finan ial de ision-making in marketsand �rms: A behavioral perspe tive, in Robert A. Jarrow, Voijslav Maksimovi , andWilliamT. Ziemba, ed.: Finan e, Handbooks in Operations Resear h and ManagementS ien e 9, 385{410 (North Holland, Amsterdam).

DeChow, P. M., and R. G. Sloan, 1997, Returns to ontrarian investment strategies:Tests of naive expe tations hypotheses, Journal of Finan ial E onomi s 41, 3{27.

DeGeorge, F., J. Patel, and R. Ze khauser, 1999, Earnings manipulations to ex eedthresholds, Journal of Business 71, 1{34.

DeLong, J. B., A. Shleifer, L. Summers, and R. J. Waldmann, 1990a, Noise trader riskin �nan ial markets, Journal of Politi al E onomy 98, 703{738.

57

DeLong, J. B., A. Shleifer, L. Summers, and R. J. Waldmann, 1990b, Positive feedba kinvestment strategies and destabilizing rational spe ulation, Journal of Finan e 45,375{395.

DeLong, J. B., A. Shleifer, L. Summers, and R. J. Waldmann, 1991, The survival ofnoise traders in �nan ial markets, Journal of Business 64, 1{20.

DeMarzo, P., D. Vayanos, and J. Zwiebel, 2000, A near-rational model of persuasion {with impli ations for �nan ial markets, Stanford University Business S hool, workingpaper.

Denis, D., and A. Sarin, 2000, Is the market surprised by poor earnings realizationsfollowing seasoned equity o�erings?, Journal of Finan ial E onomi s.

Desai, H., and P. C. Jain, 1997a, Firm performan e and fo us{ long-run sto k marketperforman e following spino�s, Journal of Finan ial E onomi s 54, 75{101.

Desai, H., and P. C. Jain, 1997b, Long-run ommon sto k returns following sto k splitsand reverse splits, Journal of Business 70, 409{434.

Dharan, B. G., and D. Ikenberry, 1995, The long-run negative drift of post-listing sto kreturns, Journal of Finan e 50, 1547{1574.

Di hev, I., 1998, Is the risk of bankrupt y a systemati risk?, Journal of Finan e 53,1131{1147.

Di hev, I., and J. Piotroski, 2001, The long-run sto k returns following bond ratings hanges, Journal of Finan e.

Dittmar, A. K., 2000, Why do �rms repur hase sto k, Journal of Business 73, 321{356.

D'Mello, R., and P. K. Shrof, 2000, Equity undervaluation and de isions related torepur hase tender o�ers: An empiri al investigation, Journal of Finan e 55, 2399 {2424.

E kbo, B. E., R. W. Masulis, and O. Norli, 2000, Seasoned publi o�erings: Resolutionof the `new issues puzzle', Journal of Finan ial E onomi s 56, 251{291.

E kbo, B. E., and O. Norli, 2000, Leverage, liquidity, and long-run ipo returns, Dart-mouth College Working Paper.

Edwards, W., 1968, Conservatism in human information pro essing, in B. Kleinmutz,ed.: Formal Representation of Human Judgement (Wiley, N.Y., NY).

Einhorn, H. J., 1980, Over on�den e in judgment, New Dire tions for Methodology ofSo ial and Behavioral S ien e 4, 1{16.

Einhorn, H. J., and R. Hogarth, 1978, Con�den e in judgement: Persisten e in theillusion of validity, Psy hologi al Review 85, 395{416.

58

Ellison, G., and D. Fudenberg, 1995, Word of mouth ommuni ation and so ial learning,Quarterly Journal of E onomi s 110, 93{126.

Ellsberg, D., 1961, Risk, ambiguity, and the savage axioms, Quarterly Journal of E o-nomi s 75, 643{699.

Elton, E. J., M. J. Gruber, and M. N. Gultekin, 1984, Professional expe tations: A - ura y and diagnosis of errors, Journal of Finan ial and Quantitative Analysis 19,351{363.

Engel, R., 1996, The forward dis ount anomaly and the risk premium: A survey ofre ent eviden e, Journal of Empiri al Finan e 3, 123{192.

Epstein, L., and S. Zin, 1989, Substitution, risk aversion, and the temporal behavior of onsumption growth and asset returns I: A theoreti al framework, E onometri a 57,937{969.

Epstein, L., and S. Zin, 1990, `�rst-order' risk aversion and the equity premium puzzle,Journal of Monetary E onomi s 26, 387{407.

Epstein, L., and S. Zin, 1991, Substitution, risk aversion, and the temporal behavior of onsumption growth and asset returns II: An empiri al analysis, Journal of Politi alE onomy 99, 263{286.

Epstein, L., and S. Zin, 1993, The independen e axiom and asset returns, CarnegieMellon University.

Erevelles, S., 1998, The role of a�e t in marketing, Journal of Business Resear h 42,199{215.

Fair, R., 2000, Events that shook the market, Yale University Working Paper.

Fama, E. F., 1991, EÆ ient apital markets II, Journal of Finan e 46, 1575{1643.

Fama, E. F., 1998, Market eÆ ien y, long-term returns and behavioral �nan e, Journalof Finan ial E onomi s 49.

Fama, E. F., and R. R. Bliss, 1987, The information in long-maturity forward rates, TheAmeri an E onomi Review 77, 680{692.

Fama, E. F., and K. R. Fren h, 1988a, Dividend yields and expe ted sto k returns,Journal of Finan ial E onomi s 22, 3{25.

Fama, E. F., and K. R. Fren h, 1988b, Permanent and temporary omponents of sto kpri es, Journal of Politi al E onomy 96, 246{273.

Fama, E. F., and K. R. Fren h, 1989, Business onditions and expe ted returns on sto ksand bonds, Journal of Finan ial E onomi s 25, 23{49.

59

Fama, E. F., and K. R. Fren h, 1992, The ross-se tion of expe ted sto k returns, Journalof Finan e 47, 427{465.

Fama, E. F., and K. R. Fren h, 1993, Common risk fa tors in the returns on sto ks andbonds, Journal of Finan ial E onomi s 33, 3{56.

Fama, E. F., and K. R. Fren h, 1995, Size and book-to-market fa tors in earnings andreturns, Journal of Finan e 50, 131{156.

Fama, E. F., and K. R. Fren h, 1996a, The CAPM is wanted, dead or alive, Journal ofFinan e 51, 1947{1958.

Fama, E. F., and K. R. Fren h, 1996b, Multifa tor explanations of asset pri ing anoma-lies, Journal of Finan e 51, 55{84.

Fama, E. F., and K. R. Fren h, 1998, Value versus growth: The international eviden e,Journal of Finan e 53, 1975{1999.

Fama, E. F., and K. R. Fren h, 2000, The equity premium, University of Chi ago.

Fama, E. F., and J. Ma Beth, 1973, Risk, return and equilibrium: Empiri al tests,Journal of Politi al E onomy 81, 607{636.

Farmer, J. D., 1999, Physi ists attempt to s ale the ivory towers of �nan e, Computingin S ien e and Engineering pp. 26{39.

Farmer, J. D., and A. W. Lo, 1999, Frontiers of �nan e: Evolution and eÆ ient markets,Pro eedings of the National A ademy of S ien e 96, 9991{9992.

Fernandez-Villaverde, J., and A. Mukherji, 2000, Can we really observe hyperboli dis- ounting?, University of Minnesota Working Paper.

Ferris, S. P., R. A. Haugen, and A. K. Makhija, 1988, Predi ting ontemporary volumewith histori volume at di�erential pri e levels: Eviden e supporting the dispositione�e t, Journal of Finan e 43, 677{697.

Ferson, W. E., and G. M. Constantinides, 1991, Habit persisten e and durability inaggregate onsumption; empiri al tests, Journal of Finan ial E onomi s 29, 199{240.

Festinger, L., and J. M. Carlsmith, 1959, Cognitive onsequen es of for ed omplian e,Journal of Abnormal and So ial Psy hology 58, 203{211.

Figlewski, S., 1978, Market `eÆ ien y' in a market with heterogeneous information,Journal of Politi al E onomy 86, 581{597.

Fis her, P., and R. Verre hia, 1999, Publi information and heuristi trade, Journal ofA ounting and E onomi s pp. 89{124.

Fis hho�, B., and R. Beyth-Marom, 1983, Hypothesis evaluation from a bayesian per-spe tive, Psy hologi al Review 90, 239{260.

60

Fis hho�, B., P. Slovi , and S. Li htenstein, 1977, Knowing with ertainty: the appro-priateness of extreme on�den e, Journal of Experimental Psy hology 3, 552{564.

Fis hho�, B., P. Slovi , and S. Li htenstein, 1978, Fault trees: Sensitivity of estimatedfailure probabilities to problem representation, Journal of Experimental Psy hology:Human Per eption and Performan e 4, 330{334.

Fis ho�, B., 1982, For those ondemned to study the past: Heuristi s and biases inhindsight, in Daniel Kahneman, Paul Slovi , and Amos Tversky, ed.: Judgementunder Un ertainty: Heuristi s and Biases (Cambridge University Press, Cambridge).

Fisher, I., 1928, The Money Illusion (Adelphi Company, New York).

Fisher, I., 1930, The Theory of Interest (Ma Millan, New York).

Foerster, S. R., and G. A. Karolyi, 1999, The e�e ts of market segmentation and investorre ognition on asset pri es: Eviden e from foreign sto ks listing in the united states,Journal of Finan e 54, 981 { 1013.

Foerster, S. R., and G. A. Karolyi, 2000, The long-run performan e of global equityo�erings, Journal of Finan ial and Quantitative Analysis 35, 499{528.

Forgas, J. P., 1995, Mood and judgment: The a�e t infusion model (aim), Psy hologi alBulletin 117, 39{66.

Forsythe, R., F. Nelson, G. Neumann, and J. Wright, 1992, Anatomy of an experimentalpoliti al sto k market, Ameri an E onomi Review 82, 1142{1161.

Frankel, J. A., and K. A. Froot, 1986, Understanding the us dollar in the eighties: Theexpe tations of hartists and fundamentalists, E onomi Re ord, Supp. 62, 24{38.

Frankel, J. A., and K. A. Froot, 1990, Chartists, fundamentalists, and trading in theforeign ex hange market, Ameri an E onomi Review 80, 181{185.

Frankel, R., and C. Lee, 1998, A ounting valuation, market expe tation, and the book-to-market e�e t, Journal of A ounting and E onomi s pp. 283{321.

Frankel, R., and C. Lee, 1999, A ounting diversity and international valuation, CornellUniversity Parker Center working paper.

Froot, K., and E. Dabora, 1999, How are sto k pri es a�e ted by the lo ation of trade?,Journal of Finan ial E onomi s 53, 189{216.

Garber, P., 1989, Tulipmania, Journal of Politi al E onomy 97, 535{560.

Gazzaniga, M. S., 1988, Mind Matters: How Mind and Brain Intera t to Create ourCons ious Lives (Houghton Mi�in, Boston).

Gervais, S., and T. Odean, 2001, Learning to be over on�dent, Review of Finan ialStudies, forth oming.

61

Gigerenzer, G., 1991, How to make ognitive illusions disappear: Beyond `heuristi s andbiases', in W. Stro he, and M. Hewstone, ed.: European Review of So ial Psy hology2, 83{115 (Wiley, New York).

Gigerenzer, G., and U. Ho�rage, 1995, How to improve bayesian reasoning withoutinstru tion: Frequen y formats, Psy hologi al Review 102, 684{704.

Gigerenzer, G., U. Ho�rage, and H. Kleinbolting, 1991, Probabilisti mental models: Abrunswikian theory of on�den e, Psy hologi al Review 98, 506{528.

Gigerenzer, G., U. Ho�rage, and H. Kleinbolting, 1996, On narrow norms and vagueheuristi s: A reply to kahneman and tversky, Psy hologi al Review 103, 592{596.

Gigerenzer, G., P. Todd, and ABC-Group, 1999, Simple Heuristi s that Make Us Smart(Oxford University Press, New York, N.Y.).

Gilboa, I., and D. S hmeidler, 1995, Case-based de ision theory, Quarterly Journal ofE onomi s 110, 605{639.

Gilovi h, T. R., 1981, Seeing the past in the present: The e�e t of asso iations to familiarevents on judgments and de isions, Journal of Personality and So ial Psy hology 40,797{808.

Gilovi h, T. R., 1991, How We Know What Isn't So: The Fallibility of Human Reasonin Everyday Life (Free Press, New York).

Gilovi h, T. R., R. Vallone, and A. Tversky, 1985, The hot hand in basketball: On themisper eption of random sequen es, Cognitive Psy hology 17, 592{596.

Goetzmann, W. N., R. G. Ibbotson, and L. Peng, 2001, A new histori al database forthe nyse 1815 to 1925: Performan e and predi tability, Journal of Finan ial Markets4, 1{32.

Gomes, F., 2000, Loss aversion and the demand for risky assets, London Business S hool,working paper.

Gompers, P., and J. Lerner, 1998, Venture apital distributions: Short-run and long-runrea tions, Journal of Finan e.

Gompers, P., and J. Lerner, 2000, The really long-run performan e of initial publi o�erings: the pre-nasdaq eviden e, Harvard University Working Paper.

Goyal, A., and I. Wel h, 1999, The myth of predi tability: Does the dividend yieldore ast the equity premium?, Anderson S hool, UCLA.

Grether, D. M., 1980, Bayes' rule as a des riptive model: The representativeness heuris-ti , Quarterly Journal of E onomi s 95, 537{557.

Grether, D. M., 1992, Testing bayes rule and the representativeness heuristi : Someexperimental eviden e, Journal of E onomi Behavior and Organization 17, 31{57.

62

Grether, D. M., and C. R. Plott, 1979, E onomi theory of hoi e and the preferen ereversal phenomenon, Ameri an E onomi Review 69, 623{638.

GriÆn, D., and A. Tversky, 1992, The weighing of eviden e and the determinants ofover on�den e, Cognitive Psy hology 24, 411{435.

GriÆn, J. M., and M. L. Lemon, 2001, Does book-to-market equity proxy for distressor overrea ton?, University of Arizona.

Grinblatt, M., and M. Keloharju, 2000, The investment behavior and performan e ofvarious investor types: a study of �nland's unique data set, Journal of Finan ialE onomi s 55, 43{67.

Grinblatt, M., and M. Keloharju, 2001a, How distan e, language and ulture in uen esto kholdings and trades, Journal of Finan e.

Grinblatt, M., and M. Keloharju, 2001b, What makes investors trade?, Journal of Fi-nan e.

Grinblatt, M., R. W. Masulis, and S. Titman, 1984, The valuation e�e ts of sto k splitsand sto k dividends, Journal of Finan ial E onomi s 13, 97{112.

Grinblatt, M., and T. Moskowitz, 1999, The ross-se tion of expe ted returns and itsrelation to past returns: New eviden e, UCLA Anderson S hool working paper #24-99.

Grinblatt, M., S. Titman, and R. Wermers, 1995, Momentum investment strategies,portfolio performan e, and herding: A study of mutual fund behavior, Ameri anE onomi Review 85, 1088{1105.

Groth, J. C., W. G. Lewellen, G. C. S harbaum, and R. C. Lease, 1979, An analysis ofbrokerage house se urities re ommendations, Finan ial Analysts' Journal 35, 32{40.

Grundy, B., and J. S. Martin, 2001, Understanding the nature of the risks and the sour eof the rewards to momentum investing, Review of Finan ial Studies 14, 29{78.

Gul, F., 1991, A theory of disappointment aversion, E onometri a 59, 667{686.

Gyourko, J., and D. B. Keim, 1992, What does the sto k market tell us about real estatereturns?, Journal of the Ameri an Real Estate and Urban E onomi s Asso iation 20,457{485.

Hand, J. R. M., 1990, A test of the extended fun tional �xation hypothesis, A ountingReview 65, 740{63.

Hand, J. R. M., 1991, Extended fun tional �xation and se urity returns around earningsannoun ements: A reply to ball and kothari, A ounting Review 66, 739{46.

63

Handa, P., S. P. Kothari, and C. Wasley, 1993, Sensitivity of multivariate tests of the apital asset-pri ing model to the return measurement interval, Journal of Finan e48, 1543{1551.

Hansen, L., T. J. Sargent, and T. D. Tallarini, 1999, Robust permanent in ome andpri ing, Review of E onomi Studies 66, 873 { 907.

Hansen, L. P., and R. Jagannathan, 1991, Impli ations of se urity market data formodels of dynami e onomies, Journal of Politi al E onomy 99, 225{262.

Hansen, L. P., and K. J. Singleton, 1983, Sto hasti onsumption, risk aversion, and thetemporal behavior of asset returns, Journal of Politi al E onomy 91, 249{265.

Harmon-Jones, E., and J. Mills, 1999, An introdu tion to ognitive dissonan e theoryand an overview of urrent perspe tives on the theory, in Eddie Harmon-Jones, andJudson Mills, ed.: Cognitive Dissonan e: Progress on a Pivotal Theory in So ialPsy hology (Ameri an Psy hologi al Asso iation).

Harris, C., and D. Laibson, 2001, Dynami hoi es of hyperboli onsumers, E onomet-ri a, forth oming.

Harvey, C., 1989, The world pri e of ovarian e risk, Journal Finan e 46, 111{157.

Haugen, R. A., and N. L. Baker, 1996, Commonality in the determinants of expe tedsto k returns, Journal of Finan ial E onomi s 41, 401{439.

Hawawini, G., and D. B. Keim, 1995, On the predi tability of ommon sto k returns:World-Wide eviden e, in Robert A. Jarrow, Voijslav Maksimovi , and William T.Ziemba, ed.: Finan e, Handbooks in Operations Resear h and Management S ien eand Management S ien e pp. 497{544 (North Holland, Amsterdam).

Hawawini, G., and D. B. Keim, 2000, The ross se tion of ommon sto k returns: Areview of the eviden e and some new �ndings, in Gabriel Hawawini, Donald B. Keim,and William T. Ziemba, ed.: Se urity Market Imperfe tions in World Wide EquityMarkets (Cambridge University Press, Cambridge, UK).

Hawawini, G., D. B. Keim, and W. T. Ziemba, 2000, Se urity Market Imperfe tions inWorld Wide Equity Markets (Cambridge University Press, Cambridge, UK).

Hawkins, S. A., and R. Hastie, 1990, Hindsight: Biased judgements of past events afterthe out omes are known, Psy hologi al Bulletin 107, 311{327.

Heath, C., and A. Tversky, 1991, Preferen es and beliefs: Ambiguity and ompeten ein hoi e under un ertainty, Journal of Risk and Un ertainty 4, 5{28.

Heaton, J. C., 1995, An empiri al investigation of asset pri ing with temporally depen-dent preferen e spe i� ations, E onometri a 63, 681{717.

64

Heaton, J. C., and D. J. Lu as, 1996, Evaluating the e�e ts of in omplete markets onrisk-sharing and asset pri ing, Journal of Politi al E onomy 104, 443{487.

Hertwig, R., and A. Ortmann, 2001, Experimental pra ti es in e onomi s: A hallengefor psy hologists?, Behavioral and Brain S ien es, forth oming 24.

Hertzel, M., M. L. Lemmon, J. S. Lin k, and L. Rees, 1999, Long-run performan efollowing private pla ements of equity, Working Paper.

Heston, S. L., K. G. Rouwenhorst, and R. E. Wessels, 1995, The stru ture of internationalsto k returns and the integration of apital markets, Journal of Empiri al Finan e 2,173{197.

Heston, S. L., K. G. Rouwenhorst, and R. E. Wessels, 1999, The role of beta and size inthe ross-se tion of european sto k returns, European Finan ial Management 4, 4{28.

Higgins, E., 1996, Knowledge a tivation: A essibility, appli ability, and salien e, in E.T.Higgins, and A.W. Kruglanski, ed.: So ial Psy hology: Handbook of Basi Prin iplespp. 133{168 (Guilford, New York).

Hirshleifer, D., and G. Y. Luo, 2001, On the survival of over on�dent traders in a ompetitive se urity market, Journal of Finan ial Markets 4, 73{84.

Hirshleifer, D., and T. Shumway, 2000, Good day sunshine: Sto k returns and theweather, Ohio State University Working Paper.

Hirshleifer, D., and S. H. Teoh, 2001, Investor psy hology in apital markets: Eviden eand poli y impli ations, Ohio State University, in progress.

Hirshleifer, D., and I. Wel h, 2000, An e onomi approa h to the psy hology of hange:Amnesia, inertia, and impulsiveness, Ohio State University Working Paper.

Ho, T., and R. Mi haely, 1988, Information quality and market eÆ ien y, Journal ofFinan ial and Quantitative Analysis 5, 357{386.

Hodri k, R. J., 1992, Dividend yields and expe ted sto k returns: Alternative pro eduresfor inferen e and measurement, Review of Finan ial Studies 5, 357{386.

Hong, H., T. Lim, and J. Stein, 2000, Bad news travels slowly: Size, analyst overageand the pro�tability of momentum strategies, Journal of Finan e 55, 265{295.

Hong, H., and J. C. Stein, 1999, A uni�ed theory of underrea tion, momentum tradingand overrea tion in asset markets, Journal of Finan e 54, 2143{2184.

Huberman, G., 1999, Familiarity breeds investment, Columbia University Working Pa-per, 97-04.

Huberman, G., and T. Regev, 2001, Contagious spe ulation and a ure for an er,Journal of Finan e.

65

Hvidkjaer, S., 2000, A trade-based analysis of momentum, Johnson S hool of Manage-ment, Working Paper.

Ikenberry, D., J. Lakonishok, and T. Vermaelen, 1995, Market underrea tion to openmarket share repur hases, Journal of Finan ial E onomi s 39, 181{208.

Ikenberry, D., J. Lakonishok, and T. Vermaelen, 2000, Sto k repur hases in anada:Performan e and strategi trading, Journal of Finan e 55, 2373{2397.

Ikenberry, D., and S. Ramnath, 2000, Underrea tion, Ri e University Working Paper.

Ikenberry, D., G. Rankine, and E. K. Sti e, 1996, What do sto k splits really signal?,Journal of Finan ial and Quantitative Analysis 31, 357{375.

Isenberg, D. J., 1986, Group polarization: A riti al review and meta-analysis, Journalof Personality and So ial Psy hology 50, 1141{1151.

Jagannathan, R., K. Kubota, and H. Takehara, 1998, Relationship between labor-in omerisk and average return: Empiri al eviden e from the Japanese sto k market, Journalof Business 71, 319{348.

Jagannathan, R., and Z. Wang, 1996, The CAPM is alive and well, Journal of Finan e51, 3{53.

Jegadeesh, N., 1991, Seasonality in sto k pri e mean reversion: Eviden e from the U.S.and the U.K., Journal of Finan e 46, 1427{1444.

Jegadeesh, N., 1999, Long-run performan e of seasoned equity o�erings: Ben hmarkerrors and biases in expe tations, University of Illinois Working Paper.

Jegadeesh, N., and S. Titman, 1993, Returns to buying winners and selling losers: Im-pli ations for sto k market eÆ ien y, Journal of Finan e 48, 65{91.

Jegadeesh, N., and S. Titman, 1995, Overrea tion, delayed rea tion, and ontrarianpro�ts, Review of Finan ial Studies 8, 973{999.

Jenkins, H. M., and W. C. Ward, 1965, Judgement of ontingen y between responsesand out omes, Psy hologi al Monographs 79, 1{17.

Jindra, J., 2000, Seasoned equity o�erings, overvaluation and timing, Ohio State Uni-versity.

Josephs, R., R. P. Larri k, C. M. Steele, and R. E. Nisbett, 1996, Prote ting the selffrom the negative onsequen es of risky de isions, Journal of Personality and So ialPsy hology 62, 26{37.

Kahneman, D., J. L. Knets h, and R. H. Thaler, 1991, The endowment e�e t, lossaversion, and status quo bias: Anomalies, Journal of E onomi Perspe tives 5, 193{206.

66

Kahneman, D., and D. Lovallo, 1993, Timid hoi es and bold fore asts: A ognitiveperspe tive on risk taking, Management S ien e 39, 17{31.

Kahneman, D., and A. Tversky, 1973, On the psy hology of predi tion, Psy hologi alReview 80, 237{25l.

Kahneman, D., and A. Tversky, 1979, Prospe t theory: An analysis of de ision underrisk, E onometri a 47, 263{291.

Kahneman, D., and A. Tversky, 1982, The psy hology of preferen es, S ienti� Ameri an246, 160{173.

Kahneman, D., and A. Tversky, 1996, On the reality of ognitive illusions: A reply togigerenzer's ritique, Psy hologi al Review 103, 582{591.

Kamstra, M. J., L. A. Kramer, and M. D. Levi, 2000a, Losing sleep at the market: Thedaylight-savings anomaly, Ameri an E onomi Review, forth oming 12, 1005{1000.

Kamstra, M. J., L. A. Kramer, and M. D. Levi, 2000b, Winter blues: Seasonal a�e tivedisorder (sad), the january e�e t, and sto k market returns, Fa ulty of Commer e,University of British Columbia Working Paper.

Kang, J.-K., Y.-C. Kim, and R. M. Stulz, 1999, The underrea tion hypothesis and thenew issue puzzle: Eviden e from Japan, Review of Finan ial Studies 12, 519{534.

Kang, J.-K., and R. M. Stulz, 1997, The underrea tion hypothesis and the new issuepuzzle: Eviden e from Japan, Journal of Finan ial E onomi s 46, 3{28.

Keane, M. P., and D. E. Runkle, 1990, Testing the rationality of pri e fore asts: Neweviden e from panel data, Ameri an E onomi Review 80, 714{735.

Keim, D. B., 1988, Sto k market regularities: A synthesis of the eviden e and explana-tions, in Elroy Dimson, ed.: Sto k Market Anomalies (Cambridge University Press,Cambridge).

Keim, D. B., and R. F. Stambaugh, 1986, Predi ting returns in the sto k and bondmarkets, Journal of Finan ial E onomi s 17, 357{390.

Keren, G., 1991, Calibration and probability judgments: Con eptual and methodologi alissues, A ta Psy hologi a 77, 217{273.

Keynes, J. M., 1936, The General Theory of Employment, Interest and Money (Ma mil-lan, London).

Kim, D., 1997, A reexamination of �rm size, book-to-market, and earnings pri e inthe ross-se tion of expe ted sto k returns, Journal of Finan ial and QuantitativeAnalysis 32, 463{489.

Kim, M.-J., C. R. Nelson, and R. Startz, 1988, Mean reversion in sto k pri es? Areappraisal of the empiri al eviden e, Review of E onomi Studies 58, 551{528.

67

Kirby, K., and R. J. Herrnstein, 1995, Preferen e reversals due to myopi dis ounting ofdelayed reward, Psy hologi al S ien e 6, 83{89.

Klayman, J., and Y. Ha, 1987, Con�rmation, dis on�rmation and information in hy-pothesis testing, Psy hologi al Review 94, 211{228.

Kleidon, A. W., 1986, Anomalies in �nan ial e onomi s: Blueprint for hange?, Journalof Business 59, S469{S499.

Klibano�, P., O. Lamont, and T. A. Wizman, 1999, Investor rea tion to salient news in losed-end ountry funds, Journal of Finan e 53, 673{699.

Knez, P., V. L. Smith, and A. Williams, 1985, Individual rationality, market rationality,and value estimation, Ameri an E onomi Review Papers and Pro eedings 52, 397{402.

Knez, P. J., and M. J. Ready, 1997, On the robustness of size and book-to-market in ross-se tional regressions, Journal of Finan e 52, 1355{1382.

Ko herlakota, N. R., 1996, The equity premium: It's still a puzzle, Journal of E onomi Literature 34, 42{71.

Koehler, J. J., 1996, The base rate falla y re onsidered: Normative, des riptive andmethodologi al hallenges, Behavioral and Brain S ien es 19, 1{53.

Koraj zyk, R. A., D. J. Lu as, and R. L. M Donald, 1991, The e�e t of informationreleases on the pri ing and timing of equity issues, Review of Finan ial Studies 4,685{708.

Kothari, S., 2000, Capital market resear h in a ounting, MIT Sloan S hool WorkingPaper.

Kothari, S., and J. Shanken, 1997, Book-to-market, dividend yield, and expe ted marketreturns: A time-series analysis, Journal of Finan ial E onomi s 44, 169{203.

Kothari, S., and J. Shanken, 2000, Beta and book-to-market: Is the glass half full or halfempty?, in Se urity Market Imperfe tions in World Wide Equity Markets pp. 44{64(Cambridge University Press, Cambridge).

Kothari, S., J. Shanken, and R. Sloan, 1995, Another look at the ross-se tion of expe tedreturns, Journal of Finan e 50, 185{224.

Kraus, A., and J. S. Sagi, 2000, Inter-temporal exibility preferen es, Haas S hool ofBusiness, University of California, Berkeley.

Kris he, S. D., and C. M. Lee, 2000, The information ontent of analyst sto k re om-mendations, Department of E onomi s, Simon Fraser University Working Paper.

Kroll, Y., and H. Levy, 1992, Further tests of the separation theorem and the apitalasset pri ing model, Ameri an E onomi Review 82, 664{670.

68

Kroll, Y., H. Levy, and A. Rapoport, 1988a, Experimental tests of the mean-varian emodel for portfolio sele tion, Organizational Behavior and Human De ision Pro esses42, 388{410.

Kroll, Y., H. Levy, and A. Rapoport, 1988b, Experimental tests of the separation theo-rem and the apital asset pri ing model, Ameri an E onomi Review 78, 500{519.

Krus hke, J. K., and M. K. Johansen, 1999, A model of probabilisti ategory learning,Journal of Experimental Psy hology: Learning, Memory, and Cognition 25.

Kuran, T., and C. Sunstein, 1999, Availability as ades and risk regulation, StanfordLaw Review 51, 683{768.

Kurz, M., 1997, Endogenous E onomi Flu tuations (Springer Verlag, New York, N.Y.).

Kyle, A., and F. A. Wang, 1997, Spe ulation duopoly with agreement to disagree: Canover on�den e survive the market test?, Journal of Finan e 52, 2073{2090.

Kyle, A., and W. Xiong, 2000, Contagion as a wealth e�e t, Duke University.

La Porta, R., J. Lakonishok, A. Shleifer, and R. W. Vishny, 1997, Good news for valuesto ks: Further eviden e on market eÆ ien y, Journal of Finan e 52, 859{874.

Laibson, D., 1997, Golden eggs and hyperboli dis ounting, Quarterly Journal of E o-nomi s 112, 443{477.

Lakonishok, J., A. Shleifer, and R. W. Vishny, 1992, The impa t of institutional tradingon sto k pri es, Journal of Finan ial E onomi s 32, 23{43.

Lakonishok, J., A. Shleifer, and R. W. Vishny, 1994, Contrarian investment, extrapola-tion and risk, Journal of Finan e 49, 1541{1578.

Lakonishok, J., and T. Vermaelen, 1990, Anomalous pri e behavior around repur hasetender o�ers, Journal of Finan e 45, 455{77.

Lamont, O., and R. Thaler, 2000, Can the market add and subtra t? mispri ing in te hsto k arveouts, University of Chi ago GSB Working Paper.

Langer, E. J., 1975, The illusion of ontrol, Journal of Personality and So ial Psy hology32, 311{328.

Langer, E. J., and J. Roth, 1975, Heads I win tails it's han e: The illusion of ontrol asa fun tion of the sequen e of out omes in a purely han e task, Journal of Personalityand So ial Psy hology 32, 951{955.

LeBaron, B., 2000a, Agent-based omputational �nan e: Suggested readings and earlyresear h, Journal of E onomi Dynami s and Control 24, 679{702.

LeBaron, B., 2000b, Evolution and time horizons in an agent-based sto k market, Bran-deis University and NBER.

69

LeBaron, B., 2000 , Evolving long run investors in a short run world, Brandeis Universityand NBER.

Lee, C., 2001, Market eÆ ien y and a ounting resear h, Journal of A ounting andE onomi s forth oming.

Lee, C., A. Shleifer, and R. Thaler, 1991, Investor sentiment and the losed-end fundpuzzle, Journal of Finan e 46, 75{109.

Lee, C. M., J. Myers, and B. Swaminathan, 1999, What is the intrinsi value of thedow?, Journal of Finan e 54, 1693 { 1741.

Lee, C. M., and B. Swaminathan, 2000a, Do sto k pri es overrea t to earnings news?,Parker Center Working Paper, Cornell University.

Lee, C. M., and B. Swaminathan, 2000b, Pri e momentum and trading volume, Journalof Finan e 55, 2017 { 2069.

Lee, I. H., 1998, Market rashes and informational avalan hes, Review of E onomi Studies 65, 741{759.

Lenney, E., 1977, Women's self- on�den e in a hievement settings, Psy hologi al Bul-letin 84, 1{13.

LeRoy, S. F., and R. D. Porter, 1981, Sto k pri e volatility: A test based on impliedvarian e bounds, E onometri a 49, 97{113.

Lewellen, J., 1999, The time-series relations among expe ted return, risk, and book-to-market, Journal of Finan ial E onomi s 54.

Lewellen, J., 2000, Momentum pro�ts and the auto orrelation of returns, Working Paper,University of Ro hester, Ro hester, NY.

Lewellen, J., and J. Shanken, 2000, Market eÆ ien y, rational expe tations, and esti-mation risk, Working Paper, University of Ro hester, Ro hester, NY.

Lewis, K. K., 1999, Trying to explain home bias in equities and onsumption, Journalof E onomi Literature 37, 571{608.

Libby, R., R. Bloom�eld, and M. Nelson, 2001, Experimental resear h in �nan ial a - ounting, Cornell University.

Li htenstein, S., and B. Fis ho�, 1977, Do those who know more also know more abouthow mu h they know?, Organizational Behavior and Human Performan e 20, 159{183.

Li htenstein, S., B. Fis ho�, and L. Phillips, 1982, Calibration of probabilities: The stateof the art to 1980, in Daniel Kahneman, Paul Slovi , and Amos Tversky, ed.: Judge-ment under Un ertainty: Heuristi s and Biases pp. 306{334 (Cambridge UniversityPress, Cambridge).

70

Li htenstein, S., and P. Slovi , 1971, Reversals of preferen e between bids and hoi esin gambling de isions, Journal of Experimental Ps y hology 89, 46{56.

Liew, J., and M. Vassalou, 2000, Can book-to-market, size and momentum be risk fa torsthat predi t e onomi growth?, Journal of Finan ial E onomi s 57, ???{???

Lin, P., 2000a, Analyst fore ast error and the pro�tability of earnings revisions strategies:Eviden e from u.s. and u.k. equity markets, University of Washington, Seattle WorkingPaper.

Lin, P., 2000b, Earnings revisions strategies in the u.s., u.k., and japan, University ofWashington, Seattle Working Paper.

Loewenstein, G., and D. Prele , 1992, Anomalies in intertemporal hoi e: Eviden e andan interpretation, Quarterly Journal of E onomi s 107, 573{597.

Loewenstein, G. F., 1996, Out of ontrol: Vis eral in uen es on behavior, OrganizationalBehavior and Human De ision Pro esses 65, 272{292.

Loewenstein, G. F., 2000, Emotions in e onomi theory and e onomi behavior, Amer-i an E onomi Review 65, 426{432.

Longsta�, F. A., P. Santa-Clara, and E. S hwartz, 1999, Throwing away a billion dollars:The ost of suboptimal exer ise strategies in the swaptions market, Anderson S hoolWorking Paper #8-00, UCLA.

Lord, C. G., L. Ross, and M. R. Lepper, 1979, Biased assimilation and attitude polar-ization: the e�e ts of prior theories on subsequently onsidered eviden e, Journal ofPersonality and So ial Psy hology 37, 2098{2109.

Loughran, T., and J. Ritter, 1995, The new issues puzzle, The Journal of Finan e 50,23{52.

Loughran, T., and J. Ritter, 2000, Uniformly least powerful tests of market eÆ ien y,Journal of Finan ial E onomi s 55, 361{389.

Lovell, M. C., 1986, Tests of the rational expe tations hypothesis, Ameri an E onomi Review 76, 110{124.

Lundeberg, M. A., P. W. Fox, and J. Pun o har, 1994, Highly on�dent but wrong:Gender di�eren es and similarities in on�den e judgements, Journal of Edu ationalPsy hology 86, 114{121.

Luo, G. Y., 1998, Market eÆ ien y and natural sele tion in a ommodity futures market,Review of Finan ial Studies 11, 647{674.

Luttmer, E. G. J., and T. Mariotti, 2000, Subje tive dis ounting in an ex hange e onomy,London S hool of E onomi s Working Paper.

71

Ma Kinlay, A. C., 1995, Multifa tor models do not explain deviations from the CAPM,Journal of Finan ial E onomi s 38, 3{28.

Madrian, B., and D. Shea, 2000, The power of suggestion: Inertia in 401(k) parti ipationand savings behavior, NBER working paper 7682.

Maenhout, P. J., 2000, Robust portfolio rules and asset pri ing, Harvard UniversityWorking Paper.

Mankiw, N. G., 1986, The term stru ture of interest rates revisited, Brookings PapersE on A tivity 1, 61{110.

Mankiw, N. G., and L. H. Summers, 1984, Do long-term interest rates overrea t toshort-term interest rates, Brookings Papers E onomi A tivity 1, 223{42.

Mann, L., 1992, Stress, a�e t, and risk taking, in Risk-Taking Behavior pp. 201{230(Wiley, Chi hester).

Mano, H., 1999, The in uen e of pre-existing negative a�e t on store pur hase intentions,Journal of Retailing 75, 149{172.

Markowitz, H. M., 1952, The utility of wealth, Journal Politi al E onomy 60, 151{8.

Marsh, T. A., and R. C. Merton, 1986, Dividend variability and varian e bounds testsfor the rationality of sto k market pri es, Ameri an E onomi Review 76, 483{498.

M Clelland, A., and F. Bolger, 1994, The alibration of subje tive probabilities: Theoriesand models 1980-1993, in Subje tive Probability pp. 453{482 (Wiley, Chi hester).

M Connell, J. J., and G. C. Sanger, 1987, The puzzle in post-listing ommon sto kreturns, Journal of Finan e 42, 119{140.

Mehra, R., and E. Pres ott, 1985, The equity premium: A puzzle, Journal of MonetaryE onomi s 15, 145{161.

Mehra, R., and E. Pres ott, 2001, The equity premium puzzle in retrospe t, in G.M.Constantinides, M. Harris, and R. Stulz, ed.: Handbook of the E onomi s of Finan e(North Holland).

Mehra, R., and R. Sah, 2000, Can small u tuations in investors' subje tive preferen esindu e large volatility in equity pri es?, University of California, Santa Barbara.

Merton, R. C., 1987, A simple model of apital market equilibrium with in ompleteinformation, Journal of Finan e 42, 483{510.

Mi haely, R., R. H. Thaler, and K. L. Woma k, 1995, Pri e rea tions to dividend initi-ations and omissions: Overrea tion or drift?, Journal of Finan e 50, 573{608.

Mi haely, R., and K. Woma k, 1999, Con i t of interest and the redibility of under-writer analyst re ommendations, Review of Finan ial Studies 12, 573{608.

72

Miller, D. T., and M. Ross, 1975, Self-serving bias in attribution of ausality: Fa t or� tion?, Psy hologi al Bulletin 82, 213{225.

Mit hell, M., and E. Sta�ord, 2000, Managerial de isions and long-term sto k pri eperforman e, Journal of Business 73, 287{320.

Modigliani, F., and R. Cohn, 1979, In ation, rational valuation, and the market, Finan- ial Analyst Journal pp. 24{44.

Moreland, R. L., and R. Bea h, 1992, Exposure e�e ts in the lassroom: The devel-opment of aÆnity among students, Journal of Experimental So ial Psy hology 28,255{276.

Moskowitz, T. J., and M. Grinblatt, 1999, Do industries explain momentum?, Journalof Finan e 54, 1249{1290.

Mulligan, C. B., 1996, A logi al e onomist's argument against hyperboli dis ounting,University of Chi ago.

Mussweiler, T., and F. Stra k, 1999, Hypothesis- onsistent testing and semanti primingin the an horing paradigm: A sele tive a essibility model, Journal of ExperimentalSo ial Psy hology 35, 136 { 164.

Neal, R., and S. M. Wheatley, 1998, Do measures of investor sentiment predi t returns?,Journal of Finan ial and Quantitative Analysis 33, 523{547.

Nel, E., R. Helmrei h, and E. Aronson, 1969, Opinion hange in the advo ate as a fun -tion of the persuasibility of his audien e: A lari� ation of the meaning of dissonan e,Journal of Personality and So ial Psy hology 12, 117{124.

Newman, J., W. Wol�, and E. Hearst, 1980, The feature-positive e�e t in adult humansubje ts, Journal of Experimental Psy hology: Human Learning and Memory 6, 630{650.

Ng, L. K., and Y. Fu, 2000, Market eÆ ien y and return statisti s: Eviden e from realestate and sto k markets using a present-value approa h, University of Wis onsin.

Nisbett, R., and L. Ross, 1980, Human Inferen e: Strategies and Short omings of So ialJudgment (Prenti e-Hall, Englewood Cli�s, NJ).

Nisbett, R. E., and T. D. Wilson, 1977a, The halo e�e t: Eviden e for un ons iousalteration of judgments, Journal of Personality and So ial Psy hology 35, 250{256.

Nisbett, R. E., and T. D. Wilson, 1977b, Telling more than we an know: Verbal reportson mental pro esses, Psy hologi al Review 84, 231{259.

Nofsinger, J. R., and R. W. Sias, 1999, Herding and feedba k trading by institutionaland individual investors, Journal of Finan e 54, 2263 { 2295.

73

Odean, T., 1998a, Are investors relu tant to realize their losses?, Journal of Finan e 53,1775{1798.

Odean, T., 1998b, Volume, volatility, pri e and pro�t when all traders are above average,Journal of Finan e 53, 1887{1934.

Odean, T., 1999, Do investors trade too mu h?, Ameri an E onomi Review 89, 1279{1298.

O'Donoghue, T., and M. Rabin, 1999, Doing it now or later, Ameri an E onomi s Review89, 103{124.

Palomino, F., 1996, Noise trading in small markets, Journal of Finan e 51, 1537{50.

Payne, J. W., J. R. Bettman, and E. J. Johnson, 1992, Behavioral de ision resear h: A onstru tive pro essing perspe tive, Annual Review of Psy hology 43, 87{131.

Peavy, J. W., 1990, Returns on initial publi o�erings of losed-end funds, Review ofFinan ial Studies 3, 695{708.

Pennington, N., and R. Hastie, 1988, Explanation-based de ision making: E�e ts ofmemory stru ture on judgment, Journal of Experimental Psy hology: Learning, Mem-ory, and Cognition 14, 521{533.

Pesaran, M. H., and A. Timmermann, 1995, Predi tability of sto k returns: Robustnessand e onomi signi� an e, Journal of Finan e 50, 1201{1228.

Peters, E., and P. Slovi , 1996, The role of a�e t and worldviews as orienting disposi-tions in the per eption and a eptan e of nu lear power, Journal of Applied So ialPsy hology 26, 1427{1453.

Petty, R. E., F. Glei her, and S. M. Baker, 1991, Multiple roles for a�e t in persuasion,in J. Forgas, ed.: Emotion and So ial Judgments pp. 181{200 (Pergamon, Oxford).

Ponti�, J., and L. D. S hall, 1998, Book-to-market ratios as predi tors of market returns,Journal of Finan ial E onomi s 49, 141{160.

Poterba, J. M., and L. H. Summers, 1988, Mean reversion in sto k returns: Eviden eand impli ations, Journal of Finan ial E onomi s 22, 27{59.

Poteshman, A., 2000, Underrea tion, overrea tion, and in reasing misrea tion to infor-mation in the options market, Journal of Finan e, forth oming.

Rabin, M., 1998, Psy hology and e onomi s, Journal of E onomi Literature 36, 11{46.

Rabin, M., 2000, Inferen e by believers in the law of small numbers, University of Cali-fornia, Berkeley Working Paper.

Rabin, M., and J. S hrag, 1999, First impressions matter: A model of on�rmatory bias,Quarterly Journal of E onomi s 94, 37{82.

74

Ranguelova, E., 2000, Disposition e�e t and �rm size: New eviden e on individualinvestor trading a tivity, Harvard University.

Rashes, M. S., 2001, Massively onfused investors making onspi uously ignorant hoi es(m i - m i ), Journal of Finan e forth oming.

Rau, P. R., and T. Vermaelen, 1998, Glamour, value and the post-a quisition perfor-man e of a quiring �rms, Journal of Finan ial E onomi s 49, 223{253.

Read, D., G. Loewenstein, and M. Rabin, 1999, Choi e bra keting, Journal of Risk andUn ertainty 19, 171{197.

Ri hards, A. J., 1997, Winner-loser reversals in national sto k market indi es: Can theybe explained?, Journal of Finan e 52, 2129{2144.

Ri hardson, M., and T. Smith, 1994, A uni�ed approa h to testing for serial orrelationin sto k returns, Journal of Business 67, 371{399.

Ri hardson, M., and J. H. Sto k, 1989, Drawing inferen es from statisti s based onmultiyear asset returns, Journal of Finan ial E onomi s 25, 323{348.

Ri hardson, S., S. H. Teoh, and P. D. Wyso ki, 1999, Tra king analysts' fore asts overthe annual earnings horizon: Are analysts' fore asts optimisti or pessimisti ?, OhioState University Fisher College of Business Working Paper.

Rietz, T. A., 1998, Enfor ing arbitrage restri tions in experimental asset markets, Uni-versity of Iowa Working Paper.

Ritov, I., 1996, Probability of regret: Anti ipation of un ertainty resolution in hoi e,Organizational Behavior and Human De ision Pro esses 66, 228{236.

Ritov, I., and J. Baron, 1990, Relu tan e to va inate: omission bias and ambiguity,Journal of Behavioral De ision Making 3, 263{277.

Ritter, J. R., and R. S. Warr, 2001, The de line of in ation and the bull market of 1982to 1999, University of Florida Working Paper.

Roll, R. W., 1984, Orange jui e and weather, Ameri an E onomi Review 74, 861{880.

Roll, R. W., 1988, r2, Journal of Finan e 43, 541{566.

Rosenberg, B., K. Reid, and R. Lanstein, 1985, Persuasive eviden e of market ineÆ- ien y, Journal of Portfolio Management 11, 9{17.

Rosenthal, L., and C. Young, 1990, The seemingly anomalous pri e behavior of royaldut h shell and unilever nv/pl , Journal of Finan ial E onomi s 26, 123{141.

Ross, L., 1977, The intuitive psy hologist and his short omings: Distortions in theattribution pro ess, in L. Berkowitz, ed.: Advan es in experimental so ial psy hology10, 174{220 (A ademi Press, New York, NY).

75

Ross, L., D. Green, and P. House, 1977, The "false onsensus e�e t": an ego entri bias in so ial per eption and attribution pro esses, Journal of Experimental So ialPsy hology 13, 279{301.

Ross, L., M. R. Lepper, F. Stra k, and J. Steinmetz, 1977, So ial explanation and so ialexpe tation: E�e ts of real and hypotheti al explanations on subje tive likelihood,Journal of Personality and So ial Psy hology 35, 817{829.

Rouwenhorst, K. G., 1998, International momentum strategies, Journal of Finan e 53,267{284.

Rouwenhorst, K. G., 1999, Lo al return fa tors and turnover in emerging sto k markets,Journal of Finan e 54, 1439{1464.

Roze�, M. S., and M. A. Zaman, 1988, Market eÆ ien y and insider trading: Neweviden e, Journal of Business 61, 25{44.

Rubinstein, A., 2000, Is it `e onomi s and psy hology'?: The ase of hyperboli dis- ounting, Prin eton University E onomi s Department.

Samuelson, W., and R. Ze khauser, 1988, Status-quo bias in de ision making, Journalof Risk and Un ertainty 1, 7{59.

Sandroni, A., 2000, Do markets favor agents able to make a urate predi tions?, E ono-metri a 68, 1303{1342.

Saunders, E. M. J., 1993, Sto k pri es and wall street weather, Ameri an E onomi Review 83, 1337{1345.

S hill, M. J., and C. Zhou, 1999, Pri ing an emerging industry: Eviden e from internetsubsidiary arve-outs, University of California, Riverside working paper.

S hwarz, N., and G. L. Clore, 1983, Mood, misattribution, and judgments of well-being:Informative and dire tive fun tions of a�e tive states, Journal of Personality andSo ial Psy hology 45, 513{523.

Seyhun, H. N., 1988, The information ontent of aggregate insider trading, Journal ofBusiness 61, 1{24.

Seyhun, H. N., 1990, Overrea tion or fundamentals: Some lessons from insiders' responseto the market rash of 1987, Journal of Finan e 45, 1303{1331.

Sha�r, E., P. Diamond, and A. Tversky, 1997, Money illusion, Quarterly Journal ofE onomi s 112, 341{374.

Sharpe, S. A., 1999, Sto k pri es, expe ted returns, and in ation, Board of Governorsof the Federal Reserve System, Finan e and E onomi s Dis ussion Series no. 1999-2.

Shefrin, H., 1997, Behavioral option pri ing, Working paper, Santa Clara University.

76

Shefrin, H., and M. Statman, 1984, Explaining investor preferen e for ash dividends,Journal of Finan ial E onomi s pp. 253{82.

Shefrin, H., and M. Statman, 1985, The disposition to sell winners too early and ridelosers too long: Theory and eviden e, Journal of Finan e pp. 777{790.

Shefrin, H., and M. Statman, 1994, Behavioral apital asset pri ing theory, Journal ofFinan ial and Quantitative Analysis 29, 323{349.

Shefrin, H., and M. Statman, 1995, Making sense of beta, size, and book-to-market,Journal of Portfolio Management pp. 323{349.

Shiller, R. J., 1979, The volatility of long term interest rates and expe tations modelsof the term stru ture, Journal of Politi al E onomy 87, 1190{1219.

Shiller, R. J., 1981, Do sto k pri es move too mu h to be justi�ed by subsequent hangesin dividends?, Ameri an E onomi Review 71, 421{498.

Shiller, R. J., 1982, Consumption, asset markets, and ma roe onomi u tuations,Carnegie Ro hester Conferen e Series on Publi Poli y 17, 203{238.

Shiller, R. J., 1984, Sto k pri es and so ial dynami s, Brookings Papers on E onomi A tivity Review 2, 457{498.

Shiller, R. J., 1988, Portfolio insuran e and other investor fa tors in the 1987 sto kmarket rash, NBER Ma roe onomi s Annual pp. 287{96.

Shiller, R. J., 1990, Spe ulative pri es and popular models, Journal of E onomi Per-spe tives 4, 55{65.

Shiller, R. J., 1999, Human behavior and the eÆ ien y of the �nan ial system, in J.Taylor, and M. Woodford, ed.: Handbook of Ma roe onomi s (Elsevier, Amsterdam).

Shiller, R. J., 2000a, Conversation, information, and herd behavior, Ameri an E onomi Review 85, 181{185.

Shiller, R. J., 2000b, Irrational exuberan e (Prin eton University Press).

Shiller, R. J., 2000 , Measuring bubble expe tations and investor on�den e, Journal ofPsy hology and Finan ial Markets 1, 49{60.

Shiller, R. J., J. Y. Campbell, and K. L. S hoenholtz, 1983, Forward rates and future pol-i y: Interpreting the term stru ture of interest rates, Brookings Papers on E onomi A tivity 1, 173{217.

Shiller, R. J., and J. Pound, 1989, Survey eviden e on the di�usion of interest andinformation among investors, Journal of E onomi Behavior and Organization 12,46{66.

77

Shleifer, A., 2000, IneÆ ient markets: an introdu tion to behavioral �nan e (Oxford U.Press, Oxford).

Shleifer, A., and R. Vishny, 1997, The limits to arbitrage, Journal of Finan e 52, 35{55.

Shumway, T., 1996, Size, overrea tion, and book-to-market e�e ts as default premia,University of Mi higan Business S hool.

Shumway, T., 1998, Explaining returns with loss aversion, University of Mi higan Busi-ness S hool Mimeo.

Simon, H. A., 1956, Rational hoi e and the stru ture of environments, Psy hologi alReview 63, 129{138.

Skinner, D. J., and R. G. Sloan, 2000, Earnings surprises, growth expe tations, and sto kreturns or don't let an earnings torpedo sink your portfolio, University of Mi higanBusiness S hool Working Paper.

Sloan, R., 1997, Do sto k pri es fully re e t information in a ruals and ash ows aboutfuture earnings?, A ounting Review pp. 289{315.

Soll, J. B., 1996, Determinants of over on�den e and mis alibration: The roles of ran-dom error and e ologi al stru ture, Organizational Behavior and Human De isionPro esses 65, 117{137.

Solnik, B., 1974, International pri ing of risk: An empiri al investigation of the world apital markets, Journal of Finan e 29, 48{54.

Spiess, D. K., and J. A�e k-Graves, 1995, Underperforman e in long-run sto k returnsfollowing seasoned equity o�erings, Journal of Finan ial E onomi s 38, 243{268.

Spiess, D. K., and J. A�e k-Graves, 1999, The long-run performan e of sto k returnsfollowing debt o�ers, Journal of Finan ial E onomi s 54, 45{73.

Stanovi h, K. E., and R. F. West, 2000, Individual di�eren es in reasoning: Impli ationsfor the rationality debate?, Brain and Behavioral S ien es 22.

Stasser, G., L. Taylor, and C. Hanna, 1989, Information sampling in stru tured andunstru tured dis ussions of three- and si -person groups, Journal of Personality andSo ial Psy hology 57, 67{78.

Stasser, G., and W. Titus, 1985, Pooling of unshared information in group de isionmaking: Biased information sampling during dis ussion, Journal of Personality andSo ial Psy hology 48, 1467{1478.

Statman, M., and S. Thorley, 1998, Over on�den e and trading volume, Santa ClaraUniversity.

Stattman, D., 1980, Book values and sto k returns, The Chi ago MBA: A Journal ofSele ted Papers 4, 25{45.

78

Steele, C., and T. Liu, 1983, Dissonan e pro esses as self-aÆrmation, Journal of Per-sonality and So ial Psy hology 45, 393{397.

Stein, J., 1996, Rational apital budgeting in an irrational world, Journal of Business69, 429{455.

Sundaresan, S. M., 1989, Intertemporally dependent preferen es and the volatility of onsumption and wealth, Review of Finan ial Studies 2, 73{89.

Sunder, S., 1995, Experimental asset markets: A survey, in John H. Kagel, and Alvine E.Roth, ed.: The Handbook of Experimental E onomi s (Prin eton University Press,Prin eton, NJ).

Swaminathan, B., 1996, Time-varying expe ted small �rm returns and losed-end funddis ounts, Review of Finan ial Studies 9, 845{888.

Taylor, S. E., and J. D. Brown, 1988, Illusion and well-being: A so ial psy hologi alperspe tive on mental health, Psy hologi al Bulletin 103, 193{210.

Teoh, S. H., I. Wel h, and T. J. Wong, 1998a, Earnings management and the long-termmarket performan e of initial publi o�erings, Journal of Finan e 53, 1935{1974.

Teoh, S. H., I. Wel h, and T. J. Wong, 1998b, Earnings management and the underper-forman e of seasoned equity o�erings, Journal of Finan ial E onomi s 50, 63{99.

Teoh, S. H., and T. J. Wong, 2000, Analysts' redulity about reported earnings andoveroptimism in new equity issues, Ohio State Fisher College of Business WorkingPaper.

Tesar, L., and I. M. Werner, 1995, Home bias and high turnover, Journal of InternationalMoney and Finan e.

Thaler, R., 1981, Some empiri al eviden e on dynami in onsisten y, E onomi Letter8, 201=207.

Thaler, R., and E. J. Johnson, 1990, Gambling with the house money and trying tobreak even: The e�e ts of prior out omes on risky hoi e, Management S ien e 36,643{660.

Thaler, R. H., 1985, Mental a ounting and onsumer hoi e, Marketing S ien e 4,199{214.

Thaler, R. H., and H. M. Shefrin, 1981, An e onomi theory of self- ontrol, Journal ofPoliti al E onomy 89, 392{406.

Tornell, A., 2000, Robust-h1 fore asting and asset pri ing anomalies, NBER WorkingPaper 7753.

Trivers, R., 1985, So ial Evolution (Benjamin/Cummings, Menlo Park).

79

Trivers, R., 1991, De eit and self-de eption, in R. Robinson, and L. Tiger, ed.: Man andBeast Revisited (Smithsonian Press, Washington D.C.).

Tversky, A., and D. Kahneman, 1971, Belief in the law of small numbers, Psy hologi alBulletin 76, 105{110.

Tversky, A., and D. Kahneman, 1973, Availability: A heuristi for judging frequen yand probability, Cognitive Psy hology 5, 207{232.

Tversky, A., and D. Kahneman, 1974, Judgment under un ertainty: Heuristi s andbiases, S ien e 185, 1124{1131.

Tversky, A., and D. Kahneman, 1981, The framing of de isions and the psy hology of hoi e, S ien e 211, 453{458 Jan 30.

Tversky, A., and D. Kahneman, 1983, Extensional versus intuitive reasoning: The on-jun tion falla y in probability judgment, Psy hologi al Review 90, 293{315.

Tversky, A., and D. Kahneman, 1986, Rational hoi e and the framing of de isions,Journal of Business 59, S252{S278.

Tversky, A., and D. Kahneman, 1992, Advan es in prospe t theory: Cumulative repre-sentation of un ertainty, Journal of Risk and Un ertainty 5, 297{323.

Tversky, A., P. Slovi , and D. Kahneman, 1990, The auses of preferen e reversal,Ameri an E onomi Review 80, 204{217.

Vijh, A. M., 1999, Long-term returns from equity areveouts, Journal of Finan ialE onomi s 51, 273{308.

Wang, F. A., 1998, Strategi trading, asymmetri information and heterogeneous priorbeliefs, Journal of Finan ial Markets 1, 321{352.

Wason, P. C., 1966, Reasoning, in B. M. Foss, ed.: New horizons in psy hology pp.131{151 (Penguin, Harmondsworth, Middlesex, UK).

Weber, M., and C. Camerer, 2000, The disposition e�e t in se urities trading: Anexperimental analysis, Journal of E onomi Behavior and Organization 33, 167{184.

Weil, P., 1989, The equity premium puzzle and the risk-free rate puzzle, Journal ofMonetary E onomi s 24, 401{421.

Wermers, R., 1999, Mutual fund herding and the impa t on sto k pri es, Journal ofFinan e 54, 581{622.

West, K. D., 1988, Dividend innovations and sto k pri e volatility, E onometri a 56,37{61.

Williams, A. W., 1987, The formation of pri e fore asts in experimental markets, Journalof Money, Credit, and Banking 19, 1{18.

80

Wilson, T. D., and J. S hooler, 1991, Thinking too mu h: Introspe tion an redu e thequality of preferen es and de isions, Journal of Personality and So ial Psy hology 60,181{192.

Woma k, K. L., 1996, Do brokerage analysts' re ommendations have investment value?,Journal of Finan e 51, 137{168.

Wright, W. F., and G. H. Bower, 1992, Mood e�e ts on subje tive probability assess-ment, Organizational Behavior and Human De ision Pro esses 52, 276{291.

Yariv, L., 2001, I'll see it when i believe it - a simple model of on�rmatory bias and ognitive dissonan e, Harvard University mimeo.

Yates, J. F., 1990, Judgement and De ision Making (Prenti e Hall, Englewood Cli�s,New Jersey).

81

Table 1: Common Obje tions to the Psy hologi al Approa h to Asset Pri ingand Parallel Obje tions to the Fully Rational Approa h

Obje tion to Psy hologi al Approa h Obje tion to Fully Rational Approa h

Alleged psy hologi al biases arearbitrary.

Rationality in �nan e theoryrequires impossible powers of al ulation.

Experiments that generate al-leged psy hologi al biases are notmeaningful.

The eviden e we possess does notsupport rational behavior.

It is too easy to go theory �shingfor psy hologi al biases to mat hdata ex post.

It is too easy to go theory �sh-ing for fa tor stru tures and mar-ket imperfe tions to mat h dataex post.

Rational traders should arbitrageaway mispri ing

Irrational traders should arbi-trage away eÆ ient pri ing

Rational investors will make bet-ter de isions and get ri her.

Irrational investors will bear morerisk and get ri her.

Confused investors will learn theirway to good de isions.

A urate investors will learn theirway to bad de isions.

Apparent return predi tability isspurious, so psy hologi al modelsof predi tability are misguided.

Apparent return predi tability isspurious, so rational models ofpredi tability are misguided.

82