Behavioral economics

Mil'

lllli

liiilll " : ;." !": , _';: L!piliij]iijiii!L:i!i!'i

liljil

p

jjiliffipi

I

ifflfiffiffflili hSlf^.:A i{*L'i'ji --; : ij/l UHHiflira1, . -,...;. ',;':..,.:,.. . »'>•illjjlj.^M;-!,:::!:•..:;. ;.•:.. :: .-.'. ; : :;

©ss

I H ill V,;::;;:,;, iiiii

iflp

i!!i''lj!. l il, , :li!i:). ; ', :i|.';: i ! , :;iilPi|pi

li

111

fHNWMllll

*

/

_\

_•

LIBRARIES

Digitized

by

the

Internet

Archive

in

2011

with

funding

from

Boston

Library

Consortium

Member

Libraries

,M415

10f

oi'-27|

Massachusetts

Institute

of

Technology

Department

of

Economics

Working

Paper

Series

BEHAVIORAL

ECONOMICS

Sendhil Mullainathan,

MIT

&

NBER

Richard

Thaler,

UChicago

&

NBER

Working

Paper

00-27

September

2000

Room

E52-251

50 Memorial

Drive

Cambridge,

MA

02142

This

paper can

be

downloaded

without charge

from

the

Social

Science

Research

Network

Paper

Collection at

MASSACHUSETTS INSTITUTE

OFTECHNOLOGY

OCT

3 2000

Massachusetts

Institute

of

Technology

Department

of

Economics

Working

Paper

Series

BEHAVIORAL

ECONOMICS

Sendhil Mullainathan,

MIT

&

NBER

Richard

Thaler,

UChicago

&

NBER

Working

Paper

00-27

September

2000

Room

E52-251

50 Memorial

Drive

Cambridge,

MA

02142

This

paper can

be

downloaded

without

charge

from

the

Social

Science

Research

Network

Paper

Collection at

Behavioral

Economics

Sendhil Mullainathan,

MIT

and

NBER

RichardH.Thaler,UniversityofChicagoand

NBER

Preliminary: Comments welcome.

Abstract: Behavioral Economics isthecombinationofpsychology andeconomicsthat investigates what happens in marketsinwhich someoftheagents displayhumanlimitationsandcomplications.

We

begin withapreliminaryquestionaboutrelevance. Doessomecombination ofmarket forces,learningand

evolution renderthesehumanqualities irrelevant? No. Becauseoflimitsofarbitrage lessthanperfect agents surviveandinfluencemarketoutcomes.

We

then discussthreeimportantwaysin whichhumans

deviatefromthestandardeconomicmodel. Boundedrationality reflectsthe limitedcognitiveabilitiesthat constrainhumanproblemsolving. Bounded willpower capturesthefactthatpeoplesometimesmake

choicesthatarenotintheirlong-runinterest. Boundedself-interestincorporatesthecomfortingfact that

humans areoften willingtosacrificetheirown interests tohelpothers.

We

thenillustratehow these conceptscanbe appliedintwosettings: finance andsavings. Financial marketshavegreater arbitrage opportunitiesthanothermarkets,sobehavioralfactors might bethoughttobe lessimportanthere,butwe

show thatevenherethe limitsofarbitrage createanomaliesthatthepsychologyof decisionmakinghelps explain. Since savingforretirementrequiresbothcomplexcalculationsand willpower, behavioralfactors are essentialelements ofany completedescriptive theory.

This manuscriptwaswritten asanentry intheInternationalEncyclopediaoftheSocialandBehavioral Sciences.

Introduction

It says somethinginteresting about the fieldofeconomicsthatthere is asub-field called

behavioral economics. Surely all of

economics

is

meant

tobe about thebehaviorof

economic

agents, bethey firms or consumers, suppliers ordemanders, bankers or

farmers. So,

what

is behavioral economics, and

how

doesit differ

from

the rest of

economics?

Economics

traditionallyconceptualizes aworld populated

by

calculating, unemotional

maximizers thathave been

dubbed

Homo

Economicus. In a sense,neo-classical

economics has defined itselfas explicitly "anti-behavioral". Indeed, virtually all the

behavior studied

by

cognitiveand social psychologists is either ignoredor ruled outin a

standard

economic

framework. This unbehavioral

economic

agent has been defended on

numerous

grounds:

some

claimedthat the

model was

"right";

most

others simply argued

that the standard

model was

easier toformalize and practically

more

relevant. Behavioral

economics

blossomed

with the realization thatneither point ofview

was

correct.

Empirical andexperimental evidence

mounted

against the stark predictions of

unbounded

rationality. Further

work made

clearthat one could formalizepsychological ideas and

translate

them

into testable predictions.

The

behavioral economicsresearch

program

has consisted of

two

components: 1. Identifyingthe

ways

in

which

behaviordiffersfrom the

standard model. 2.

Showing

how

thisbehaviormatters in

economic

contexts.

This papergives aflavorofthe program.

We

begin

by

discussingthe

most

important

ways

in

which

the standard

economic

model

needsto be enriched.

We

thenillustrate

how

behavioral

economics

hasbeen fruitfully applied to

two

important fields: finance and

savings. But first,

we

discuss

why

the market forces and learning donoteliminate the

importance of

human

actions.

Is

Homo

Economicus

the

Only

One

Who

Survives?

Many

economists have argued that a combination ofmarketforces (competition and

arbitrage) plus evolution should produce a world similartothat described inan

economicstextbook:

do

onlythe rational agents survive? Or,

do

the workings ofmarkets

at least renderthe actions ofthequasi-rational irrelevant? These are questions thathave been

much

studiedin thepast

two

decades, and the early impressions of

many

economiststhat markets

would

wipe out irrationality were,well, optimistic.

Consider a specificexample:

human

capital formation.

Suppose

thata

young

economist,

call

him Sam,

decides to

become

abehavioral economist, perhaps because

Sam

mistakenlythinks this will leadtoriches, orbecause he thinks itis goingtobe thenext

fad,or because he findsit interesting and lacks thewillpower tostudy "real"economics.

Whatever

the reason forthe choice, let's

assume

forthesake ofargument that this

decision

was

a mistake for

Sam

by

any rational calculation. So, what will market forces

do?Well,

Sam may

be poorerbecause ofthis choice than ifhe had sensiblychosento

study corporate finance, buthe will not be destitute.

Sam

might even realize he could switch tocorporate finance and

make

tons

more

money

but is simply unable to resistthe

temptation to continuewasting his timeon behavioral economics. So, markets per se

do

notnecessarily solve theproblem: theyprovide an incentive to switch, but theycannot

force

Sam's

hand.

What

aboutarbitrage? In this case, like

most

we

studyin

economics

outside therealm of financial markets, there issimply

no

arbitrage opportunity available.

Suppose

awise

arbitrageuris watching

Sam's

choices,

what

bet can she place? None.

The

same

canbe

saidif

Sam

saves toolittle forretirement, picks the

wrong

wife, orbuys the

wrong

car.

None

oftheseirrational acts generatesan arbitrage opportunity for

anyone

else.

Indeed,economists

now

realizethat even in financial marketsthere are importantlimits tothe workingsofarbitrage. First, in theface ofirrational traders, the arbitrageur

may

privately benefit

more from

trading that helpspush prices in the

wrong

direction than

from

trading that pushes prices in the right direction. Put another

way,

it

may

often pay

"smart

money"

to follow

"dumb money"

ratherthan to lean against it (Haltiwanger and

Waldman,

1985; Russell and Thaler 1985). For example, an extremely smart arbitrageur

nearthe beginningofthe tulip

mania

would

have profited

more from

buyingtulips and

further destabilizing prices than

by

shortingthem. Second, and slightly related, arbitrage

is inherently risky activityand consequently the supply ofarbitrage will be inherently limited (De Long, Shleifer,

Summers

and

Waldman,

1990). Arbitrageurs

who

did decide

to short tulips early

would

probablyhave been

wiped

out

by

thetime their bets

were

provento be "right".

Add

to this the fact that in practice

most

arbitrageurs are

managing

otherpeople's

money

and,thereforejudged periodically, and one sees the short horizons

that an arbitrageur will beforced to take on. Thispoint

was

made

forcefully

by

Shleifer

and Vishny (1997)

who

essentiallyforesaw the scenario thatended

up

closing

Long

Term

Capital

Management.

So, marketsperse cannot be relied

upon

to

make

economic

agentsrational.

What

about evolution?

An

old argument that individuals

who

failedto

maximize

should have been

weeded

out

by

evolutionaryforces,

which

presumably operatedduring ancient times.

Overconfident hunters, forexample, presumably caught lessprey, ate less and died younger.

Such

reasoning, however, has turned out to be faulty. Evolutionary arguments canjust as readily explain over-confidence as theycan explain appropriatelevels of confidence. For example,consider individuals playing a

war

ofattrition (perhaps in

deciding

when

to back

down

duringcombat).

Here

overconfidence will actually help.

Seeing theoverconfidence, a rational opponent will actuallychoose to back

down

sooner.

As

can be seen

from

this example, depending on theinitial environment (especially

when

theseenvironments have a

game

theoretic

component

to them),evolution

may

just as readily

weed

outrational behavioras itdoes

weed

outquasi-rational behavior.

The

troubling flexibilityof evolutionary

models

means

that they canjustas readily arguefor

bounds

on rationality.

The

final

argument

is that individuals

who

systematically and consistently

make

the

same

mistake will eventually learn the erroroftheirways. This kind of

argument

has alsonot stood upwell undertheoretical scrutiny. First, theoptimal experimentation literature has

intuition here is simple: as long as there are

some

opportunity costs to learning or to

experimentingwith a

new

strategy,even a completely "rational" learner will choose not

toexperiment. This playerwill get stuckin anon-optimal equilibrium, simply because

the cost of tryingsomething else istoo high. Second,

work

on learning in

games

has formallydemonstrated Keynes'

morbid

observation onthe "longrun".

The

time required

toconverge to an equilibriumstrategycan be extremely long.

Add

to this achanging environment and onecan easily bein a situation of perpetual non-convergence. In practice, for

many

ofthe important decisions

we

make,

both arguments apply with full force.

The

number

of times

we

getto learn

from

ourretirement decisions is low (and possiblyzero).

The

opportunitycost ofexperimentingwith different

ways

ofchoosing a

careercan be very high.

The

upshot ofall these theoretical innovations hasbeen clear.

One

cannot defend

unbounded

rationality on purelytheoretical grounds. Neitherarbitrage, competition,

evolution, norlearning necessarily guarantees that

unbounded

rationality

must

be an

effectivemodel. In the end, as

some

might haveexpected, it

must

ultimatelybe an

empiricalissue.

Does

"behavior" matter? Before evaluatingthis questionin

two

differentfields ofapplication,

we

explore the

ways

in

which

real behaviordiffers from

the stylized neoclassical model.

Three

Bounds

of

Human

Nature

The

standard

economic

model

of

human

behavior includes (atleast) three unrealistic

traits:

unbounded

rationality,

unbounded

willpower, and

unbounded

selfishness.

These

threetraits are

good

candidates formodification.

Herbert

Simon

(1955)

was

an early critic ofmodeling

economic

agents as having

unlimited information processingcapabilities.

He

suggested the term

"bounded

rationality" todescribe a

more

realistic conception of

human

problem

solving

capabilities.

As

stressed

by

Conlisk (1996), the failure to incorporate

bounded

rationality

into

economic

models isjust bad

economics

—

theequivalent topresumingtheexistence

ofa free lunch. Since

we

have onlyso

much

brainpower, andonly so

much

time,

we

cannot beexpectedto solve difficult problems optimally. It is eminently "rational" for

peopleto adoptrules of

thumb

as a

way

to

economize

on cognitive faculties. Yet the

standard

model

ignores these

bounds

and hencethe heuristics

commonly

used.

As

shown

by

Kahneman

and Tversky (1974), this oversightcan be important since sensible heuristics can lead to systematic errors.

Departures from rationality

emerge

both injudgments (beliefs) andin choice.

The

ways

in

which

judgment

diverges from rationality islong and extensive (see

Kahneman,

Slovic

and Tversky, 1982).

Some

illustrativeexamples include overconfidence, optimism,

anchoring, extrapolation, and

making

judgmentsoffrequencyor likelihoodbased on

salience (the availabilityheuristic) or similarity (the representativeness heuristic).

Many

ofthe departures fromrational choice arecaptured

by

prospecttheory

(Kahneman

uncertainty(see Starmer

2000

for areview ofliterature

on

non-EU

theoriesofchoice).

Prospect theoryis anexcellent

example

ofabehavioral

economic

theory in thatits key

theoretical

components

incorporate important featuresofpsychology. Considerthree features oftheprospect theory value function. 1. It is defined over changesto wealth

ratherthan levels of wealth (as in

EU)

toincorporate the concept ofadaptation. 2.

The

lossfunction is steeperthanthe gain function to incorporate the notion of"loss aversion";

thenotion thatpeople are

more

sensitive to decreases in theirwell beingthan to

increases. 3. Both the gain andloss function display diminishing sensitivity (the gain function is concave, the lossfunction convex) toreflect experimental findings.

To

fully

describe choices prospect theory often needs tobe

combined

with an understanding of

"mental accounting" (Thaler, 1985).

One

needs to understand

when

individuals faced with separate gamblestreat

them

as separate gains and losses and

when

they treat

them

as

one, pooling

them

toproduce one gain or loss.

A

couple of

examples

canillustrate

how

these concepts areused in real

economics

contexts. Consider overconfidence. Ifinvestors are overconfidentin their abilities, they

will be willingto

make

tradeseven in the absence oftrue information. This insighthelps explain a major

anomaly

offinancial markets. In an efficient market

when

rationalityis

common

knowledge,there is virtually

no

trading, but in actual markets there are

hundreds of millions of shares tradeddaily and

most

professionally

managed

portfolios areturned overonce a year ormore. Individual investors also trade a lot: they incur

transaction costs and yet the stocks they

buy

subsequently

do worse

than the stocksthey

sell.

An

example

involving loss aversion and mental accountingis

Camerer

et al's (1997)

study of

New

York

City taxi cabdrivers.

These

cab drivers pay a fixed fee torent their

cabs fortwelvehours and thenkeep all theirrevenues.

They

must

decide

how

longto

drive each day.

A

maximizing

strategy is to

work

longerhours on

good

days (days with high earnings perhour such as rainy daysor days with abigconventionin town) andto quit early on bad days.

However,

supposecabbiesset a targetearnings level foreach

day, and treatshortfallsrelative to thattarget as aloss. Then,they will end upquitting early on

good

days and

working

longer

on

bad

days, precisely theopposite ofthe rational strategy. This is exactly

what

Camerer

et al find in theirempirical work.

Having

solvedfor the

optimum,

Homo

Economicus

is next

assumed

tochoose the

optimum.

Real

humans,

even

when

they

know

what

is best,

sometimes

fail tochooseit

for self-control reasons.

Most

of usat

some

point haveeaten, drank,or spenttoo

much,

andexercised, saved, or

worked

too little.

Such

is

human

nature, at least since

Adam

and

Eve. (Well,

make

that

Adam.)

People(even economists) also procrastinate.

We

are

writing thisentry wellafter the date on

which

it

was

due, and

we

areconfident that

we

arenot the onlyguilty parties.

Though

people havethese self-control problems, the are at least

somewhat

awareofthem: theyjoin diet plans and

buy

cigarettes

by

the pack

(because having an entirecarton around is too tempting).

They

also pay

more

withholding taxes than they need to (in 1997, nearly

90

million tax returns paid an

taxesnear midnighton April 15 (at the post office that isbeing kept open late to

accommodate

their fellow procrastinators.)

Finally, people are boundedly selfish. Although

economic

theory does not ruleout

altruism,as a practical mattereconomists stress self interestas the primary motive. For example, the free riderproblems widely discussedin economicsare predicted to occur because individualscannot be expected tocontribute to the public

good

unless their

privatewelfare isthus improved. In contrast,people often take selfless actions. In 1993,

73.4%

ofall households gave

some money

to charity, the average dollar

amount

being

2.1%

ofhousehold income. Also,

47.7%

ofthe population doesvolunteer

work

with 4.2 hours per

week

being the average hours volunteered. Similarselfless behavioris

observedin controlled laboratory experiments. Subjects systematically often cooperate

in publicgoods andprisoners

dilemma

games, andturn

down

unfair offers in

"ultimatum" games.

Finance

Ifeconomists

were

polledtwentyyears ago and askedto

name

the

domain

in

which

bounded

rationality

was

least likely to find useful applications the likely winner

would

have been finance.

The

limits ofarbitrage arguments

were

not well understood at that

time, and one leadingeconomist had called the efficient markets hypothesis the best established fact ineconomics.

Times

change.

Now,

as

we

begin the21stcentury finance

is perhaps the branch of

economics where

behavioral economicshas

made

the greatest contributions.

How

has this

happened?

Two

factors contributed tothe surprising successof behavioral finance. First, financial

economicsin general, and the efficientmarket hypothesisin particular, generated sharp, testable predictions about observable

phenomena.

Second, there are great data readily availableto testthese sharp predictions.

We

briefly

summarize

here a few examples.

The

rational efficient markets hypothesis

makes two

classes ofpredictions about stock

price behavior.

The

first is that stock prices are "correct" in thesense that asset prices

reflect the trueor rational value ofthe security. In

many

casesthis tenet oftheefficient

market hypothesis is untestable because intrinsic values are notobservable.

However,

in

some

special cases the hypothesiscan be tested

by

comparing

two

assets

whose

relative intrinsic values are

known.

One

class ofthese iscalled "Siamese Twins":

two

versions of

the

same

stock thattrade in different places.

A

specific

well-known

example

is thecase ofRoyal

Dutch

Shell as

documented

in Froot

and

Dabora

(1999).

The

facts arethatRoyal

Dutch

Petroleum and ShellTransport are

independently incorporated in the Netherlands and

England

respectively.

The

current

firm

emerged

from a 1907 alliance between Royal

Dutch

and Shell Transport in

which

the

two

companies agreed to

merge

their interests on a 60:40basis. Royal Dutch trades

primarily in the

US

and the Netherlands and Shell trades primarily in London. According

toany rational model, the shares ofthese

two components

(afteradjusting forforeign

deviated

from

theexpected one

by

more

than

35%.

Simple explanations such as taxes and

transactions costscannot explain the disparity (seeFroot andDabora). This

example

illustratesthat prices candiverge

from

intrinsic value because oflimits ofarbitrage.

Some

investors

do

tryto exploit this mispricing, buyingthecheaperstock and shorting

the

more

expensive one, butthis is not a sure thing, as

many

hedge funds learned in the

Summer

of 1998 (when, atthe time hedge funds

were

trying to get

more

liquidity, the pricing disparitywidened).

The

Royal

Dutch

Shell

anomaly

is a violation ofone ofthe most basic principles of

economics: the law ofone price. Anothersimilar

example

is the caseofclosed-end mutual funds (Lee, Shleifer,and Thaler 1991).

These

funds are

much

like typical (open-end) mutual funds exceptthat tocash out ofthe fund, investors

must

sell theirshares

on

the

open

market. This

means

that closed-end funds have market prices that are

determined

by

supply and

demand,

ratherthan setequal to thevalue oftheir assets

by

the

fund

managers

as in an open-end fund. Since the holdings ofclosed-end funds are public,

market efficiency

would

leadone to expect that theprice ofthe fund should

match

the priceoftheunderlying securities theyhold (the net assetvalue or

NAV).

Instead,

closed-end funds typically trade at substantial discounts relative to their

NAV,

and occasionally

at substantial premia.

Most

interesting

from

abehavioral perspective is that closed-end

funddiscounts are correlated withone anotherand appear to reflectindividual investor

sentiment. (Closed-endfunds are primarily

owned

by

individual investors ratherthan

institutions.) Lee, Shleifer and Thalerfind that discounts shrink in

months

when

shares of small

companies

(also

owned

primarily

by

individuals)

do

well, and in

months

when

there islots of

IPO

activity, indicating a"hot" market. Since these findings

were

predicted

by

theirtheory, they

move

the research

beyond

thedemonstration of an

embarrassing fact (pricenot equal to

NAV)

toward aconstructive understandingof

how

markets work.

The

secondprinciple ofthe efficient market hypothesisis "unpredictability". In an

efficient market itis not possible to predict future stockprice

movements

basedon

publicly available information.

Many

early violations ofthis had

no

explicitlink to

behavior.

Thus

it

was

reportedthat small firms,firms with low priceearnings ratios

earned higherreturns than other stocks withthe

same

risk. Also, stocks in general, but

especially stocks of small

companies

have

done

well in January and on Fridays (but

poorly on

Mondays).

An

early study

by

De

Bondt

and Thaler (1985)

was

explicitly motivated

by

the

psychological finding that individuals tendto over-react to

new

information. For

example, experimental evidence suggestthat people tended to underweight baserate data

(orpriorinformation)in incorporating

new

data.

De

Bondt

andThalerhypothesized that ifinvestors displayed this behavior, then stocksthat had performedquite well overa

period of years will eventually haveprices that aretoo high. Individuals overreacting to

the

good news

will drive the prices ofthese stocks too high. Similarly, poor performers

will eventually have prices that are too low. Thisyields a prediction aboutfuture returns: past "winners" oughtto underperform while past "losers" ought tooutperform the

Thalerfound thatthe 35 stocks that had performed the worst overthe pastfive years (the

losers) outperformed themarket overthenext five years, while the35 biggestwinners overthe past five years subsequentlyunderperformed. Follow-up studies have

shown

thatthese earlyresults cannot be attributed to risk (by

some

measures the portfolio of

losers is actually lessriskythan theportfolio of winners), andcan be extended to other

measures of overreaction such as theratio ofmarketprice tothe

book

value ofequity.

More

recentstudies have foundother violationsofunpredictabilitythat havethe opposite

pattern

from

that foundby

DeBondt

andThaler,

namely

underreaction ratherthan

overreaction.

Over

short periods oftime, e.g., six

months

toone year, stocks display

momentum

—

the stocksthat go upthe fastest forthe firstsix

months

oftheyear tend to

keep going up. Also, after

many

corporate

announcements

such as large earnings changes, dividend initiations andomissions, share repurchases, splits, and seasoned

equity offerings,there is an initial price

jump

onthe dayofthe

announcement

followed by a slow driftin the

same

direction for as long as ayearor

more

(see Shleifer 2000).

These

findingsof underreaction are afurtherchallenge tothe efficientmarkets

hypothesis,but also to behavioral finance.

Do

markets sometimes overreact, and sometimes underreact? Ifso, can anypattern beexplained, at least ex post?Is there a

unifying

framework

that can bring togetherthese apparently opposing facts andideas?

Work

is only

now

beginningto attackthisextremely importantquestion. Several explanations haverecently been offered (Shleifer

2000 summarizes

them). All rely on

psychological evidence (in one

way

oranother)in unifyingthefacts.

They

all explain the

anomalies

by

noting that underreaction appears atshort horizons while overreaction appears at longerhorizons, buteach paperprovides its

own

distinctive explanation.

Which

(ifany) ofthese isthe best one has yet tobe decided. But thefacts discovered so

far,

combined

with thesemodels, demonstratethe changing natureoffinance. Rigorous

empirical

work

building on psychological

phenomena

has given us

new

tools to unearth

interesting empirical facts. Rigorous theoretical work, on the other hand, hasbeen trying

to address thechallenge of incoiporatingthese empirical facts into apsychologically

plausiblemodel.

The

research discussed so farhas been about assetprices and thecontroversy aboutthe

efficientmarket hypothesis. Thereis another stream ofresearch thatisjust about

investorbehavior, not about prices.

One

example

ofthis stream ismotivated

by

mental

accounting and loss aversion.

The

issue is whetherinvestors are reluctant to realize

capital losses (because they

would

have to "declare" theloss to themselves). Shefrin and Statman (1985)

dubbed

this hypothesis the "disposition effect".

The

prediction that

investors will be lesswilling tosell a loserthan awinner is staking since the tax law encouragesjustthe opposite behavior. Nevertheless,

Odean

(1998) findsevidence ofjust thisbehavior. In his sample ofthecustomers ofa discount brokeragefirm, investors

were

more

likely to sell a stock thathadincreased in value than one that haddecreased in

value.

While

around

15%

ofall gains

were

realized, only

10%

ofall losses arerealized.

This hesitancyto realize gains

came

ata cost.

Odean shows

thatthe stocksthe loser

Savings

Iffinance

was

the field in

which

a behavioral approach

was

least likely, apriori,to

succeed, saving had tobe one ofthe most promising.

The

standard life-cycle

model

of savings abstracts

from

both

bounded

rationality and

bounded

willpower, yetsaving for

retirement is both adifficultcognitive

problem

and a difficult self-control problem. It is

then, perhapsless surprisingthat a behavioral approach has been fruitful here.

As

in

finance, progresshas been helped

by

thecombination ofa refined standard theorywith

testable predictions and lots of data sourceson household saving behavior.

One

crispprediction ofthe life-cycle

model

is that savingsrates are independent of

income.

Suppose

twins

Tom

and

Ray

are identical in everyrespectexcept that

Tom

earns

most

ofhis

money

early in his life (say he's a basketball player), while

Ray

earns

most

ofhis late in life(say he's a manager).

The

life-cycle

model

predicts that

Tom

the basketball playeroughtto save hisearly

income

to increase

consumption

later in life,

while

Ray

the

manager

ought toborrow

from

his future

income

to increase

consumption

earlier in life.This prediction is completelyunsupported

by

the data,

which shows

that

consumption

very closely tracks

income

overindividuals' life cycles.Furthermore, the

departures from predictedbehaviorcannot be explained merely

by

people's inability to

borrow. Banks, Blundell and

Tanner

(1998) show, forexample, that

consumption

drops sharply as individuals retireand their

incomes

drop.

They

have simply not saved

enough

forretirement. Indeed,

many

low tomiddle

income

families have essentially

no

savings whatsoever.

The

primary causeofthis lack ofsaving appears to be self-control.

One

bit

ofevidence supportingthis conclusion is that virtually all saving

done by Americans

is

accomplishedin vehicles that support

what

are often called "forced savings", e.g.,

accumulating

home

equity

by

paying the mortgage and participation in pension plans.

Coming

full circle,one "forced" savings individuals

may

choose themselves

may

behigh

tax witholdings, so that

when

therefund

comes

theycan

buy

something theymight not

have had the willpowerto save

up

for.

One

ofthe

most

interesting research areas hasbeen devoted to measuring the effectiveness of tax-subsidized savings programs such as

IRAs

and 401(k)s.

The

standardanalysis ofthese

programs

is quite simple. Considerthe original

IRA

program

from

the early 1980s. This

program

provided tax subsidies forsavings

up

to a threshold, often

$2000

peryear,there

was

no

marginal incentiveto save forany household that

was

already saving

more

than

$2000

peryear. Thus, those saving

more

thanthe threshold

should notincrease their total saving, they will merelyswitch

some

money

from

a taxable

accountto the

IRA

forthe tax gain. Moreover, everyone

who

is savinga positive

amount

should participate in this

program

forthe infra-marginal gain.

The

actual analysis of

these

programs

has

shown

that the reality isnot so clear.

By

some

accounts at least, these

programs

appear tohave generatedquite a bit of

new

savings.

Some

arguethat almost

every dollarof savings in

IRAs

appear to represent

new

savings In otherwords, people

arenot simply shiftingtheirsavings into

IRAs

and leaving their total behavior

unchanged. Similarresults arefound for401(k) plans.

The

behavioral explanation for

up

special mental accountsthat aredevoted toretirement savings. Households tend to

respect the designated use ofthese accounts, and theirself-control is helped

by

the tax

penalty that

must

be paid iffunds are

removed

prematurely.

(Some

issues remain

controversial. See thedebate in the Fall 1996 issue oftheJournal of

Economic

Perspectives.)

An

interesting flipside to

ERA/40

l(k)

programs

is thatthese programs also generatedfar less than the fullparticipation one

would

have expected.

Many

eligiblepeople

do

not

participate, foregoing ineffect acash transferfrom the

government

(and, in

some

cases,

from

theiremployer).

O'Donoghue

and Rabin (1999) present an explanation based on

procrastination and hyperbolic discounting. Individuals typically

show

very sharp impatiencefor shorthorizon decisions,but

much

more

patienceat longhorizons. This is

often referredtoas hyperbolic discounting

by

contrastwith thestandard assumptionof exponential discounting, in

which

patience is independent of horizon. Inexponential

models, peopleareequally patient at long and short horizons.

O'Donoghue

and Rabin

arguethat hyperbolic individuals will

show

exactly the

low

IRA

participation that

we

observe.

Though

hyperbolic people will

want

toeventually participate in

ERAs

(because they are patientin thelongrun), something always

comes

upin the short run (where they

arevery impatient)thatprovides greaterimmediate reward. Consequently,they

may

delay joiningthe

IRA

indefinitely.

Ifpeople procrastinate about joining the savingsplan, then it shouldbe possibleto

increase participation ratessimply

by

lowering thepsychic costs ofjoining.

One

simple

way

ofaccomplishing this is to switch the defaultoption for

new

workers. In most companies,

when

employees first

become

eligiblefor the401(k) plan theyreceive aform

inviting

them

tojoin; tojoin they haveto sendtheform back and

make some

choices.

Several firms have

made

the seemingly inconsequential changeof switchingthedefault:

employees

areenrolledinto the plan unlessthey explicitlyoptout. Thischangehas often

produced dramatic increases in savingsrates. For example, in one

company

studied by

Madrian

and

Shea

(2000) the

employees

who

joined afterswitchingthe default tobeing

in the plan were

50%

more

likely to participate than theworkers in the year priortothe

change. (They also find that thedefault asset allocation had a strongeffecton workers

choices.

The

firm had

made

thedefault asset allocation

100%

in a

money

market

account, andtheproportion ofworkers selection this allocation soared.)

Along

with these empirical facts, therehas alsobeen a bulk oftheoretical work.

To

cite a

few examples, Laibson (1997) and

O'Donoghue

and Rabin (1999) have both

examined

the effects of hyperbolic discountingon the savings decisions.

They've

highlighted that

hyperbolic individuals will

demand commitment

devices (savingsvehicles

which

are

illiquid) and generally fail to

obey

theEuler Equation.

Other

Directions

We

have concentrated on

two

fields here in orderto give a sense ofwhat behavioral

economics

can do, but

we

do

not

want

to leave theimpression that savings and financial

effective. In laboreconomics, experimental and empirical

work

has underlinedthe

importanceoffairness considerations in settingwages. For example,

wages

between

industriesdifferdramatically, even for identical workers and

most

interestingly, even

homogeneous

workers (such asjanitors)earn higher

wages

when

they

work

in industries

where

other occupations earn more. In

Law

and Economics,

we

haveseen theimportance

of"irrelevant" factors in ajury's decision to sentenceor in the magnitude ofawards they

give. In corporate finance, one can fruitfullyinterpret a manager'sto acquire or diversify as resulting

from

overconfidence.

One

could goon. There is

much

tobe done.

References

Banks, James; Blundell, Richard;Tanner, Sarah, 1998, "Is There aRetirement-Savings Puzzle?"

American

Economic

Review. Vol. 88 (4). p 769-88.

September

1998.

Camerer, Colin, et al.,

"Labor

Supply of

New

York

City Cabdrivers:

One

Day

at a Time,"

The

Quarterly Journal of

Economics,

vol. 112 (2),pp. 407-41,

May

1997.

Conlisk, John,

"Why

Bounded

Rationality?" Journal of

Economic

Literature. Vol. 34 (2).

p 669-700. June 1996.

De

Bondt,

Werner

F

M

andThaler, Richard, "

Does

the Stock

Market

Overreact?"

Journal ofFinance. Vol.

40

(3). p 793-805. July 1985.

DeLong

B. Shleifer, A.,

Summers,

L., and

Waldman,

R., 1990, "Noise Trader Risk in

Financial Markets," Journal ofPolitical

Economy.

Vol. 98 (4). p703-38.

August

1990.

Froot,

Kenneth

A; Dabora,

Emil

M.,

"How

Are

StockPrices Affected

by

theLocation of

Trade?" Journal of FinancialEconomics. Vol. 53 (2). p 189-216.

August

1999. Haltiwanger, John and

Waldman,

Michael, "Rational Expectations and the Limitsof

Rationality:

An

Analysis of Heterogeneity,"

American

Economic

Review. Vol. 75 (3). _p

326-40. June 1985.

Kahneman,

Daniel and Tversky,

Amos,

"Judgement

Under

Uncertainty: Heuristics and

Biases,"Science, v. 185, 1974, 1124-31.

Kahneman,

Daniel and Tversky,

Amos,

"ProspectTheory:

An

Analysis ofDecision

underRisk," Econometrica. Vol.

47

(2). p 263-91.

March

1979

Kahneman,

Daniel, Slovic, Paul and Tversky,

Amos.

1982,

Judgement

under

Uncertainty: Heuristics andBiases,

Cambridge

and

New

York:

Cambridge

University

Press.

Laibson, David,

"Golden

Eggs

and Hyperbolic Discounting,"

The

Quarterly Journal of

Lee, Charles

M

C, Shleifer, Andrei andThaler, Richard H., "InvestorSentiment and the

Closed-End

Fund

Puzzle," Journal of Finance. Vol.

46

(1). p 75-109.

March

1991.

Madnan,

Bridgitte and Dennis Shea, "The

Power

ofSuggestion: Inertia in 401(k) Savings Behavior." University of

Chicago

mimeo,

February 2000.

Odean, Terrance, "Are Investors Reluctantto RealizeTheirLosses?" Journal of Finance, 1998, vol. 53 (5). p 1775-98.

O'Donoghue,

Ted

andRabin, Matthew, 1999b, "Procrastination inPreparingfor

Retirement," inBehavioral

Dimensions

ofRetirement Economics,

Henry

Aaron,editor,

The

Brookings Institution, 1999.

Russell,

Thomas

andThaler, Richard,

"The

RelevanceofQuasi Rationality in

Competitive Markets,"

American

Economic

Review. Vol. 75 (5). p 1071-82.

December

1985.

Shefrin, Hersh andStatman, Meir,

"The

Disposition to Sell

Winners

Too

Early and Ride

Losers

Too

Long:

Theory

and Evidence," Journal of Finance. Vol.

40

(3). p 777-90. July 1985.

Shleifer, Andrei, InefficientMarkets:

An

Introduction to BehavioralFinance. Clarendon

Lectures, Oxford UniversityPress, 2000.

Shleifer, Andrei and Robert Vishny,

"The

Limits of Arbitrage," Journal of Finance. Vol. 52(1). p 35-55.

March

1997.

Simon, HerbertA.,

"A

Behavioral

Model

of Rational Choice," Quarterly Journal of

Economics, 69 (February 1955): 99-118.

Starmer, Chris,

"Developments

in non-expected utility: the huntfor a descriptivetheory of choice underrisk" Journal of

Economic

Literature, 2000.

Thaler, Richard. Mental Accounting and

Consumer

Choice," MarketingScience. Vol. 4

MIT LIBRARIES