Political Implications of Loss Aversion

Author(s): Robert Jervis

Source: Political Psychology, Vol. 13, No. 2, Special Issue: Prospect Theory and Political
Psychology (Jun., 1992), pp. 187-204
Published by: International Society of Political Psychology
Stable URL: http://www.jstor.org/stable/3791678
Accessed: 05-11-2015 23:58 UTC

REFERENCES
Linked references are available on JSTOR for this article:
http://www.jstor.org/stable/3791678?seq=1&cid=pdf-reference#references_tab_contents

You may need to log in to JSTOR to access the linked references.

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at http://www.jstor.org/page/
info/about/policies/terms.jsp

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content
in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship.
For more information about JSTOR, please contact support@jstor.org.

Wiley and International Society of Political Psychology are collaborating with JSTOR to digitize, preserve and extend access
to Political Psychology.

http://www.jstor.org

This content downloaded from 137.189.170.231 on Thu, 05 Nov 2015 23:58:56 UTC
All use subject to JSTOR Terms and Conditions
Political Vol.13,No. 2, 1992
Psychology,

ofLoss Aversion
PoliticalImplications
RobertJervis'

Prospecttheory offerspowerful insightsandpropositionsintopoliticaldecision-

making,especially in internationalpolitics.Evidenceindicatesthatstatesmen
are indeedrisk-acceptant for losses. Thiswouldhelpexplainobservedpatterns
in bargaining,deterrence, theoriginsofwars,as wellas suggesting whystates
are less likelyto behaveaggressively whendoingso wouldproducegainsthan
whensuchbehaviormight preventlosses.
KEY WORDS: prospect risk;crisis
theory; stability; conflict.
socialstability;

INTRODUCTION

Is prospecttheoryvalid? If it is, whatare the implications forpolitical

decision-making, international
politics,and social lifein I
general? do nothave
fullanswers,butthissketchmayprovea usefulstimulant to furtherthinking
(also see &
Quattrone Tversky,1988; Kanwisher,1989; Peterson & Lawson,
1989).
In summary, thetheoryarguesthatpeopletendto be risk-averse forgains
(this was generallyknown but
before) simultaneously to be risk-acceptantfor
losses(thiswas thesurprise).
People are loss-averse in thesense thatlossesloom
largerthanthecorresponding gains.Losingtendollars,forexample,annoysus
morethangainingtendollarsgratifies us. Whatis peculiaraboutthisis that,
contrary to most of
versions expectedutility theory, thereference point-usually
thestatusquo-is crucial(Tversky & Kahneman,1986).Morethanthehopeof
gains,the specterof losses activates,energizes,and drivesactors,producing
great(and oftenmisguided)efforts thatrisk-and frequently lead to-greater
losses. Furthermore,thechoicebetweenalternatives will be influenced by the

of PoliticalScienceandInstitute
IDepartment ofWarandPeace Studies,ColumbiaUniversity,
New
York,New York10027.
187
? 1992 International
0162-895X/92/0600-0187$06.50/1 Societyof PoliticalPsychology

This content downloaded from 137.189.170.231 on Thu, 05 Nov 2015 23:58:56 UTC
All use subject to JSTOR Terms and Conditions
188 Jervis

wayinwhichthequestionis posed.Peoplewillchoosetheriskyalternative when

thechoiceis framedin termsof avoidinglosses when,in theexactsamecase,
theywouldtaketheless riskycourseof actionif theframeof reference is the
of
possibility improving the situation. Although a setback be
might quite minor
whencomparedto theperson'stotalvalue holdings,he will see it in termsof
wherehe was shortly beforeand so maytakethegambleof an evengreater loss
in orderto gain a chanceof reestablishing his position.
Theevidenceas tothetheory's validity is farfromconclusive.Thebulkofit
comes fromtheanswerspeople give to questionnaires askingthemto choose
betweenalternatives of different risks and payoffs.The resultscould be a
hothouseartifact ofthelaboratory (Wittman, 1991;Smith,1991,pp. 890-892),
as I suspectis in partthe case forthe supposedtendencyof people to pay
insufficient attention to base rateinformation (Jervis,1986). Perhapsthemain
reasonforthinking thattheeffect is realis thatthetendency forpeopletobe risk-
acceptant for losses resonates with personalexperience. I doubt ifI am alonein
having been willing to toleratean unusuallyhigh risk of significant losses in
return forthechanceof payingno penaltyat all or in havingbeen willingto
investsignificant additionalresources in a venture in thehope-I cannotsaythe
expectation-ofrecouping a recent loss. It is not an accidentthatpeople are
warnedagainstthrowing goodmoney after bad-they oftendo. Similarly, econ-
omiststellus thatitis notrationaltobe influenced by "sunkcosts"-havingput
a lotintoa venture is nota goodreasontocontinue withit.Butthefactthatthisis
a valid prescription does notmeanthebehavioris notcommon-indeedif it
werenot,therewouldbe no needforeconomists to stressthepoint.
If loss aversionoccursin politicallife,theexplanation maynotbe entirely
psychological. Both domestic and international politicscould explainthepattern.
A leaderwho acceptsevena limiteddefeatis likelyto be punishedat thepolls.
Gamblingbyacceptinga chanceofa greater loss inreturn fora chanceofno loss
(orevena victory) mightbe irrational fromthestandpoint ofthenationalinterest,
butrationalfromthestandpoint of thepower-seeking politician.The difference
betweensmallerand largernationallosses maynottranslate intoa similardif-
ferencein theloss ofpoliticalpower;a smallloss mightbe sufficient to lose the
politicianthenextelectionand a largerloss mighteven rallysupportto him.
Similarly, DennisRoss arguesthatduringtheCold WarSovietleaderswouldrun
highrisksto avoid losses because retreating would significantly damagethe
rulingdomesticcoalition(Ross, 1984).
Losses may not scale in thieir international impacteither.That is, it is
possiblethatin termsof reputation and credibility a smallloss wouldcost the
country notmuchless thana significantly largerone. (Thiseffect wouldnothold
formoreconcretekindsof losses, such as territory and economicstrength.)
Indeed,decision-makers (Jervis,1991)-or at leastAmericandecision-makers
(Morgan,1985)-seem to believethatsmalllosseswillmultiply via thedomino

This content downloaded from 137.189.170.231 on Thu, 05 Nov 2015 23:58:56 UTC
All use subject to JSTOR Terms and Conditions
 ofLoss Aversion
PoliticalImplications 189

effect.Gainsarenotexpectedtohavesuchconsequences.So a rational statesman

would notbe willingto runhighrisksin orderto securea moderategain but
wouldacceptmuchhigherrisksto avoida short-run loss of thesamemagnitude
because it wouldlead to greaterlosses overa longerperiodof time.
To say thatthe explanationforthisodd behavioris politicalratherthan
psychological is actuallymisleading,
however.Whilewe maynotneedprospect
theoryto explainstatesmen's choicesif we can showthattheyare basedon an
analysisof how otherswillrewardor punishthem,thetheorymaytellus how
othersarereachingtheirjudgments. Thusifit is truethatthepoliticalcoststhat
leadersincurfromlosses are notproportional to themagnitude of theloss, the
reasonmaybe thatdomesticopinionoperatesaccordingto prospecttheoryin
finding evensmalllosses so painfulthatitprefers highrisksto acceptingthem
andwillpunishanyleaderwhopermits themto occur.Prospecttheory mayalso
helpaccountforthebeliefindominoeffects becauseitindicatesthatpeoplewill
focusmoreon lossesthanon gains.Thusinthelate 1970stheUnitedStateswas
deeplyconcernedthattheloss of influence in Ethiopiawouldhavewidespread
repercussionsbutpaid littleattention to thesimultaneous gain of influencein
Somalia. Similarly, earlierin theCold WartheUnitedStateswas preoccupied
withwhatit saw as losses thatmightlead to further dominoesfallingbutpaid
littleattentionto thereversalof Sovietfortunes in Egypt,theCentralAfrican
Republic,Ghana,orevenChina.Theseeventsconstituted majorgainsforAmer-
icanpowerandinterests thatdwarfedmanyofthedefeatsthatloomedso largein
Americaneyes. But theveryfactthatthiswas notunderstood was crucialto
Americanbehaviorand can be explainedat least in partby thepsychological
processesdiscussedhere.

GENERAL EFFECTS OF LOSS AVERSION

Thereseemto be manyinstanceswherepeopleandorganizations arerisk-

acceptantforlosses. One nice exampleis thatbettingis said to be particularly
heavyon thelasthorserace. People whohavelostmoneythroughout theafter-
noonplace heavybetson thefinalrace;theyare willingto riskmoremoneyin
orderto gain a chanceof recoupingtheirearlierlosses. Again,theproblemis
morecomplicated thanthoseinthestandard pencil-and-papertests.A husbandor
wifewho returns homehavinglosta partof his or herpaycheckmaysuffer a
pricethatis relativelyinsensitive
to the amountof the loss. Even withouta
spouse,thecost to theperson'sego maybe significant and again muchmore
sensitivetothefactofa gainorlossthantothemagnitude ofthegainorloss. But
thisfact,farfromundermining prospecttheory,mayconstitute oneofthereasons
it
why operates.
Perhapsthebestgenericexampleof thewillingness to takegreatrisksto

This content downloaded from 137.189.170.231 on Thu, 05 Nov 2015 23:58:56 UTC
All use subject to JSTOR Terms and Conditions
190 Jervis

avoidanyloss is cover-ups.If a personhas committed a nontrivial transgression,

shemaydevotesignificant resourcestotrying tocoveritupeventhough doingso
exposes her to much greaterpenalties if the activities are later discovered.
Knowledgeof theWatergate break-in wouldnothavecostNixonhispresidency;
evensponsorship ofitmightnothave.It was thecover-up thatdestroyed him.Of
course,the objectiveodds are not knownand so a standardexpected-utility
modelcan be builtto accountforNixon'sbehavior, although itis farfromclear
thatsucha modelwouldbe consistent withtherestofhislife.Furthermore, there
could be largenumbersof cases in whichpeopleadmitto theirtransgressions,
butsincethesedo notgiveriseto majorpoliticalscandals,we tendto overlook
them.But althoughfirmevidenceis lacking,I suspectthatcover-ups are more
than
frequent they would be if people acted on unbiased estimates of costs and
probabilitiesinstead of being driven by the need to get off without payingany
penaltyat all.
More generally, prospecttheoryleads us to expectpeopleto perseverein
losing ventures much longerthanstandard rationality wouldlead one to expect.
Vietnamis an obviouscase, butas usualthesituation is notentirely clear.First,
as I notedearlier,a personbelievingin thedominotheory wouldpayhighcosts
to avoida limiteddefeat.Second,instancesof long,losingwarsarea sampleof
cases biased in thesame way thatcover-upsare. Instancesin whichactorscut
theirlossesareless dramatic andso maybe lostsightof. Third,continuing along
or costlypathsometimes
a difficult paysoffin thelongrun.Loss aversionmay
thencontribute to usefulperseverance. AlbertHirschman (1976) speaksof the
"hiding hand." That is, sometimes we embark on a program without muchsense
of the manyobstacles.Werewe aware of themat thestart,we wouldnever
begin.But ifwe see theobstaclesonlyafterwe haveputin a greatdeal of time
and effort,we willbe reluctant to writeoffourinvestment and so willcontinue,
and perhapssucceed.The refusalto accepta loss thencan be functional, even
though thedifference in the stance toward losses than toward gains stillneeds to
be explained.Thesearguments cannotbe definitively refuted, but neither do they
denythatprospecttheory can explaincases likeVietnam:Americanleaderswere
continuallywillingto escalate even thoughtheyknewthatthe prospectsof
eventualvictorywerefarfromcertainand thattheiractionswouldgreatlyin-
creasethedomesticandinternational coststheywouldpayintheeventoffailure.
Sunkcosts,furthermore, loomedlargein thesecretdeliberations.
In otherwars as well, statesmenlose all sense of proportion aboutthe
magnitude oflosseswhensomeloss appearscertain.ThusIkl6pointsoutthatin
February1918 theGermansobjectedto theAustriansuggestion thattheyshift
theirwarobjectivesand agreethat"theirtwocountries wereobligedto fightfor
the pre-warpossessionsof Germany.But Ludendorff grantedthisconcession
onlyaftervehement opposition:'If Germanymakespeace without profit, then
Germany has lostthewar.'Whatcuriousinability to distinguish betweenloss of

This content downloaded from 137.189.170.231 on Thu, 05 Nov 2015 23:58:56 UTC
All use subject to JSTOR Terms and Conditions
 ofLoss Aversion
PoliticalImplications 191

some territoriesand loss of thenation!"(Ikle, p. 82). Cuttinglosses afterthe

expenditureofbloodandtreasure is perhapsthemostdifficult acta statesman can
take;thelureof thegamblethatpersevering will recoupthelosses is oftentoo
greatto resist.Furthermore, and whatis crucialto theclaimthatthisbehavior
cannotbe explainedbya normalexpected-utility approach,thesamepeoplewho
gamblein thiswayarecautiouswhenfacedwithchoicesthatinvolvegains:they
willtakemuchgreater risksto tryto return to wheretheywerethantheywillto
makemajorimprovements in thestatusquo.
The Fordadministrations' reactiontothepossibility ofa swinefluepidemic
someofthedynamicsofthechoicebetweena surebutsmallloss and
illustrates
thechanceof a muchlargerone, although italso bringsoutseveralambiguities.
Thegovernment launcheda program toinoculateeveryone inthebeliefthatthere
was a significantdangerthatwithoutinoculation massivedeathswouldoccur.
The politicalleadersknew,although theydid notfocuson it,thatmassinocula-
tionwas bothcostlyandwouldcertainly killa fewpeoplebecauseofsideeffects.
Whether or notthemedicalofficialsdid a goodjob of presenting theissuesto
leaders
political (Neustadt & Fineberg, 1978; Silverstein,
1981),the logicofthe
argument formass inoculations was fairlystraightforward. As Silverstein(p.
135) putit:
It mattered
a littlethattheexperts
couldnottellwhether
thechanceofpandemicinfluenza
orevenlessthan1percent.
or3 percent,
was30 percent, Whattheassistant for
secretary
thesecretary
Health, ofHEW,thepresident,
andCongress
heard
wasthattherewassome
chanceofpandemicfluandthiswasenough.No responsible atanylevelwished
politician
toputhimselfintheposition
ofopposing
theprogram,
thusrunning
theriskthat
pandemic
illnessanddeathmightprovehima villain.

This storydoes notfitentirely well withprospecttheory. Assumingdeci-

sion-makers realizedthatinoculations
wouldcause at leasta fewdeathsandthat
theoutbreak ofswinefluwas lessthancertain, theychosea smallbutsureloss in
orderto eliminatethe possibilityof a largercatastrophe. Threefactorshelp
unravelthis.First,thedeathsfrominoculation wereprobablyseen as an insur-
ance payment,a normalcost of doingbusiness.As Tverskyand Kahneman
(1986) have shown(but I thinknot fullyexplained),people are muchmore
willingtoacceptan outcomeifitcanbe portrayed notas a lossbutas an expected
payment for the Thus
activity. peoplebuyinsurance, whichmeanspreferring a
smallbutsureloss to theslightchanceof a muchgreater one. Second,politics
was a drivingforce(notto implythatthereis anythingwrongwiththis).Leaders
knewthatif therewas an epidemicthatcould havebeen greatlyreducedby a
programof vaccinations thattheyhad rejected,theirpoliticallosses wouldbe
enormous.Indeed,even if theepidemicdid notmaterialize, a decisionnotto
inoculatemightharmthemif thepublicbelievedthisstancewas imprudent.
Here,as in theforeignpolicycase mentioned at theoutset,we mayhaveto shift
ourattentionfromthedecision-makers tothebeliefsandvaluesofthosewhocan

This content downloaded from 137.189.170.231 on Thu, 05 Nov 2015 23:58:56 UTC
All use subject to JSTOR Terms and Conditions
192 Jervis

rewardand punishthem.Third,once deathsfrominoculations wererealrather

expectation-andoccurredin greater
thanan abstract numbers thanpredicted-
theprogram was immediately calledoff.Surelosseshad a greaterimpactwhen
theyinvolvedparticularhumanbeingsthanwhentheywereprojections.(For-
tunatelyfortheU.S., butunfortunatelyfortestsofthetheory,it simultaneously
becameclearthatlittleor no swinefluwas goingto materialize.)

EFFECTS ON BARGAINING, DETERRENCE,

AND CAUSES OF WAR

The proposition thatactorsreactverydifferently to theprospectof losses

thanto thechanceof makinggainshas important implications forinternational
bargaining and conflict.If loss aversion is widespread, states defending the
statusquo shouldhavea bigbargaining advantage. That is, a will
state be willing
to paya higherpriceandrunhigherrisksifitis facinglossesthanifitis seeking
to make gains. A relatedproposition whichis consistentwiththe literature
(Schelling, 1966; Jervis,1989) is thatcoercioncan moreeasilymaintainthe
statusquo thanalterit-deterrenceis usuallyeasierthanwhatSchellingcalls
"compellence."Thesearguments, however,assumethatit is clearto theactors
whichof themis thedefenderand whichis thechallenger.Sometimesthisis
true,butoftenit is not.This yieldsa thirdimplication: Conflictsand warsare
morelikelywheneach side believesit is defending thestatusquo.
Sincethereasonwhytheperceivedstatusquo willbe vehemently defended
is thatstateswill runhighrisksrather thansuffer evena smallbutcertainloss,
theproposition can be broadened to arguethatconflicts andwarsaremostlikely
wheneach side believesit will suffersignificant losses if it does not fight.
Perceptualbiases are likelyto enterin here.Whenstatesoverestimate others'
as
hostility, theyfrequently do, they willexpect lossesunless they takestrongif
notaggressiveaction.Bothsidescan readilycometo believethis.Notonlywill
each accepthighrisksinorderto avoida bad outcome,buteachis likelytothink
thattheotherside is merelystriving forgainsand so is likelyto be willingto
back down.Furthermore, Lebow,Stein,and othershaveshown(Lebow, 1981;
Lebow& Stein,1987;Jervis, Lebow,& Stein,1985),statesthataredesperate-
thatface severelosses-feel strongmotivated biases to believethatthepolicy
thatoffersa wayoutcan in factsucceed.Thesestatesarethenlikelyto convince
themselves thattheycan successfully coerceotherswhena disinterested perspec-
tive would revealthatthisis implausible.As Lebow and Stein(1987) note,
deterrencetheory is designedprimarily to explainhowtodeal withcountries that
areseekinggains.In thesecircumstances, deterrence maybe effective. Butifthe
otheris drivenby thefearof losses,threats andcoercionare likelyto backfire,

This content downloaded from 137.189.170.231 on Thu, 05 Nov 2015 23:58:56 UTC
All use subject to JSTOR Terms and Conditions
PoliticalImplicationsof Loss Aversion 193

producinga spiralof greaterhostility. (For deterrenceversusthespiralmodel,

see Jervis,1976, Ch. 3).
A fourth implication is thatduringtheCold Wara superpower's threatto
intervene in a local conflictwas greater themoreitslocal clientwas suffering.
Thiswas exemplified twiceinthe1973war,first increasingthecredibilityofthe
Americanintervention, thenincreasing thecredibility
ofa Sovietmovewhenthe
tideofbattleshifted againstEgypt.Superpowers areless proneto intervene,and
theirthreatstodo so arelesscredible,whensuchactionswouldbringgainstothe
client.Again,thedangeris thatbothsideswillfearlosses (and willnotunder-
standeithertheother'sperspective orthechancesofsuccess),thereby generating
not only conflictsbut undesiredand unforeseen ones. But to the extentthat
perceptions areshared,loss aversionshouldstabilizelocal conflicts bymakingit
unlikelythateitherside can "win big." Assumingthelocal clientsrealizethis,
theywill haveto moderatetheirdemandsand actions.
More generally, by givinga bargaining advantageto theside thatfearsor
sufferslosses,loss aversionsupports Thisaffects
stability. day-to-daydiplomacy
as wellas crisisbargaining. Forexample,loss aversionandtheresulting caution
probablyis partof theexplanation forwhyneither theU.S. or theSovietUnion
was anxiousto gambleon a reunited Germany. The resultmighthavebeento
greatlyimproveeitherside's position,butit wouldmeanforegoing thesureand
current advantageofcontrolling partof thatcountry.To call theresultsstability,
however,is to misstheotherhalfof thestory-becauseactorsplace a higher
valuationon whattheyhave thanon whattheymightgain,theywillrefuseto
accept bargainsthata thirdpartywouldjudge to makethembetteroff.The
generalleveloftrading withinandamongcountries willthenbe lowerthanmost
theorieswould lead one to expect.Economistshave come to arguethatthese
activitiesare less thanpreviously expectedbecausescholarshad neglectedthe
role of "transaction costs"-the costsof acquiringinformation, drawingcon-
tracts,andguarding againstcheating (Coase, 1937; Williamson, 1985). Butloss
aversionmaybe equallyimportant.

WAR-PEACE DECISIONS

Prospecttheory producesinferences aboutboththeshort-andthelong-run

causes of wars. If it is correct,we wouldexpectthatwarsand otherconflicts
wouldbe morestrongly and morecommonlymotivated by thedesireto avoid
perceived losses than by the of
hope makinggains. Statesshouldbe moreoften
pushed intowar by the fearthat thealternative
to is
fighting a seriousdeteriora-
tionin theirpositionthanpulledinbythebeliefthatwarcan improvea situation
thatis alreadysatisfactory.As I willnotebelow,sometimes thestatusquo itselfis

This content downloaded from 137.189.170.231 on Thu, 05 Nov 2015 23:58:56 UTC
All use subject to JSTOR Terms and Conditions
194 Jervis

unsatisfactory. But moreoften,evenif it leavesa greatdeal to be desired,the

statusquo is at leasttolerablein thesensethatthestateis willingto live withit
ratherthanrunning a significant riskof suffering thegreaterlosses thana war
mightbring.Fear is usuallya morepotentmotivator thanthedesireforexpan-
sion.A statesman willruna significant riskofdestruction ofhisownpowerand
hisregimeand devastation of hiscountry ifhe thinksthealternative is a certain
and significant deterioration in his powerand security. But he would not be
willingto runsimilarrisksifhe believedthatthestatusquo couldbe maintained
by diplomacyeventhoughwar,ifsuccessful,couldbringgreatgains.Warsare
thenless frequently causedby aggressionthanby spiralsof fearand insecurity
(Jervis,1976, Ch. 3).
Thus it now appearsthatSaddam Hussein'sattackon Kuwaitcannotbe
describedentirely as an attempt to increasehis powerand influence. Rather,he
seemsto havebelievedthatifhe did notattack,hispositionwouldgetsteadily
worse.Notonlywouldtherebe littlehopeforeconomicrecovery in theabsence
ofa Kuwaitiwillingness to forgive Iraqidebt and oil
stopdriving pricesdownby
producing morethantheirOPEC quota,buthe seemsto haveseenan American-
led conspiracy totrimhispower(Viorst,1991,pp. 67-68). Thefamoushistorian
A. J. P. Taylorexaggerates as usual whenhe claimsthat"everywarbetween
GreatPowers[between1848and 1918]started as a preventive war,nota warof
conquest"(Taylor,1954,p. 166), butthepreventive motiveindeedis frequent
(Levy, 1987).
Loss aversionsimilarly impliesthattherestraints in a limitedwararemore
likely to be broken by a side that fearsthat failingto do so will resultin
significant losses than by the side thatbelieves thatexpansioncan bringit
significantgains.By contrast, that no
approaches give specialplacetothestatus-
quo point would lead us to think thatstatesshouldescalatewhendoingso is
to
expected yield more utilitythan maintaining therestraint. Since thereis no
reasonwhystatesshouldadopta different stancetowardavoidinglosses than
towardmakinggains,theyshouldbe as likelyto increasethelevelofviolencein
orderto breakstalemateand securea decisivevictory as to wardoffa defeat.In
fact,I doubtifthisis actuallytrue.The UnitedStatesthreatened escalationatthe
finalstagesoftheKoreanWar(Foot, 1990),butitis farfromclearwhatitwould
have done if the Chinesehad not agreedto a truce.The Chinese,in turn,
intervened notin thesummerof 1950,whensuchactionmighthavepushedthe
UnitedStatesoffthepeninsula,butin thefallwhenthealternative was to accept
a hostileregimeon itsborders(thatis, whenit was certainthatthestatusquo
woulddeteriorate badlyifshe did notfight).The risksChinawas willingto run
to regainQuemoyand Matsuand threaten Formosa-objectivesof presumably
greatvalue-were muchless.
Althoughit is oftensaid, especiallyof aggressors, thattheappetitegrows
withtheeating,thisdoes notseemto be as trueas it wouldbe weregainsand

This content downloaded from 137.189.170.231 on Thu, 05 Nov 2015 23:58:56 UTC
All use subject to JSTOR Terms and Conditions
PoliticalImplicationsof Loss Aversion 195

losses to be treatedsymmetrically. Althoughstatessometimesdo overreach

themselves (as theUnited States did in crossingthe38thparallelin Korea),itis
relativelyunusualforthe side withthe upperhand to increasethe level of
violenceorcontinueto fight togainadditional objectives,evenifdoingso seems
militarily feasible.ThustheChinesedid notpushdeeperintoIndianterritory in
1962whentheirinitialattackmetwithgreatsuccessand,contrary toKissinger's
argument, Indiaprobably wouldnothaveturned on WestPakistanin 1971evenif
theU.S. had nottriedto detersuch a move. Onlyin thelastof his warsdid
Bismarckyieldto thetemptation to makeextragains,andevenin thissituation
he tookAlsace and Lorraineonlybecausehe was convincedthatevenifhe did
nottheFrenchwouldbe immutably hostile.Althoughit was Hitlerwhotaught
the worldthatambitionsand risk-taking increasedwithsuccess, thispicture
actuallyis not accurate:Hitler'soutlookand stanceremainedremarkably un-
changedthroughout his monstrous career.
Thisdoes notmeanthatunprovoked aggression neveroccurs.Hitler,totake
onlythemostobviousexample,was drivenbythedesireto drastically alterthe
statusquo, notby thefearthathe was encircledby hostilepowersthatwould
destroyGermanyunlesshe tookarmsagainstthem.The case of postwarSoviet
foreign policyis ambiguousin thisregard,anditwouldnotbe surprising ifboth
motiveswereat work.Indeed,itshouldbe stressed thatprospect theory does not
denythatactorsmaywanttochangethestatusquo orthatthecombination ofthe
strength ofthemotivation to do so andtheperceivedpossibilities of successcan
lead toexpansionandwar.Actorswilltakesomeriskstoimprove theirsituations
even if theyare risk-averse forgains. But,thetheoryargues,actorsshouldbe
muchmorewillingto runriskswhentheybelievethatfailingto do so willresult
in certainlosses. Warswillthenfrequently be triggered by fearof loss; fighting
in orderto act on an opportunity will be relatively rare.
Greatmethodological problems plaguetheattempt todevelopfullyconvinc-
ingcomparisons (see below),butitappearsthatwhenstatestakeveryhighrisks
it is usuallythecase thattheybelievetheywill haveto acceptcertainlosses if
theydo notdo so. Furthermore, thisrecklessness surprisesothercountries-both
thosewho are attackedand neutrals--because theyexpectthestateto be more
cautious,whichit usuallyis becausethemorenormalcontextis one of facing
gainsrather thanlosses. ThusJapanattacked GreatBritainandtheUnitedStates
in December1941notbecauseherleadershadmuchconfidence theycouldwin,
butbecause theysaw-quite correctly-that thealternative to fighting was to
surrender muchof theirpositionin SoutheastAsia and China.This loss would
havebeena majorone. Buta military withdrawal wouldstillhavepermitted hera
greatdeal of influencein the area and would have leftthe Japaneseislands,
industry, armedforces,and citiesintact.Othersthought thechoiceforwarwas
odd because it was-and in retrospect still is-hard to empathizewiththe
pressures peoplefeelwhenfacedwithlosses.

This content downloaded from 137.189.170.231 on Thu, 05 Nov 2015 23:58:56 UTC
All use subject to JSTOR Terms and Conditions
196 Jervis

This case is farfromuniquein itsgeneraloutlines.Although theGerman

motivation in 1914was mixed,one important component was the beliefthatthe
diplomatic and military situation was deteriorating and thatGerman security
wouldbe muchlowerin a fewyearsifGermany didnotfight.Herleadersknew
thatvictory was farfromcertain;I findithardtobelievethattheywouldhaverun
thisriskis theythoughtthatthe statusquo could have been maintained by
peacefulmeansandthattheonlyreasonto fightwas to gainevenmoreterritory,
power,andprestige.The otherparticipants inWorldWarI also weremuchmore
strongly drivenby fearsthanby hopes. All wereworriedthattheirpowerand
security werebeingundermined bythecourseof events.Whiletheyhadexpan-
sionistambitions,thesewerenotsufficient to lead themto stakethefuture of
theircountrieson a warthattheyall enteredwithgreattrepidation. Similarly,
althoughIsrael's 1967 warresultedin itsgainingterritory, themotivation, as in
1956andinthewarofattrition of 1969-70, was thebeliefthatunlesstheserisky
actionsweretaken,thestatusquo woulddeteriorate badly.
The otherside ofthiscoinis harderto see becauseitconsistsof nonevents:
statesareslowto takeadvantageofopportunities toexpandat somerisk(Lebow,
1984),especiallywhencomparedwiththeirbehaviorwhentheyfeeltheirposi-
tionis underattack.We tendto takepeace and even thelack of demandsfor
changeas natural.But theyare not:giventhepostulatesof powerpoliticsand
expectedutility,we shouldsee statespushingtoalterthestatusquo intheirfavor
as oftenas theyexertthemselves to maintain it. We shouldsee moreHitlers,at
least in smallerversionsof power and ruthlessness. These are not absent:
Bismarckfoughtin orderto expand,or reallycreate,his state;the dramatic
weakeningof a country or empireoftendrawsothersto attack;aggressionof
courseoccurs.But are theseeventsas frequent as theywouldbe ifopportunity
wereas stronga motivation as thefearof loss?
In somecases, thereference pointis notthestatusquo butsomething better.
The gap betweenthisdesiredstateof affairs andthecurrent one can lead to high
risk-taking.(Of course,thedangerhereis thatthisargument willbecometaut-
ological unlesswe can a
specify priori what this referencepointwillbe andwhen
itwillbe used.Wealso needtoruleoutthepossibility thatthesameforceswhich
determine thereference pointalso directly lead totheriskybehavior.Somework
hasbeendonetoestablishwhatthelikelyreference pointswillbe, butmorework
needsto be done.) At times,thereference point will be an earlierstatusquo,
beforetheactorsuffered seriouslosses. Thusforat leasta periodaftera country
has lostsignificantterritoryorinfluence, theprevioussituation is likelytobe the
standardand thecontinuation of thecurrent situationwill be experienced as a
loss. Thiswas clearlythecase forEgyptafterthe1967war.She couldnotaccept
Israelioccupationof Sinai as normaland the 1973 war was triggered by the
belief-probably correct-that although war was risky,there was no chance that
peacefuldiplomacy could succeed (Stein,1985).

This content downloaded from 137.189.170.231 on Thu, 05 Nov 2015 23:58:56 UTC
All use subject to JSTOR Terms and Conditions
 ofLoss Aversion
PoliticalImplications 197

The factthatbothsides can have different reference pointsand so can

simultaneously be impelledbyloss aversioncan render dangerous thestrategy of
thefaitaccompli.GeorgeandSmokenotethatdeterrence can alterthestatusquo
beforethedefender has timeto react(Georgeand Smoke,1974,pp. 536-40).
Butunlessthelatterquicklyadjuststothenewsituation, itmaybe willingtorun
unusually highriskstoregainitspreviousposition.Theotherside,expecting the
firstto be rational,will be surprisedby itscontinued resistance.Because each
side will see itselfas defending the statusquo, it will therefore have strong
incentives to standfirmand-believing theadversary sees thesituation as the
statedoes-will expecttheothersideto retreat. Each willbe drivenbya strong
resistanceto acceptwhatit sees as an unfavorable change.
The belief in inevitableprogresscan producea different, and steadily
changing,reference point. If and when the Soviet leaders believed thecommu-
nistideology,theyexpecteda steadyincreasein thepowerof communism and
thenumberof countries thatwerefreedfromthegripof capitalism.The status
quo was notonlypolitically butpsychologically arbitrary and so stagnation for
themwouldhavethesamepsychological impactas lossesfor others. Thismight
explainwhatWesternobserverssaw as the unusualpropensity forthe Soviet
Unionto takerisksto increaseitsinfluence and aid leftistregimes.

Crisis Stability

Prospecttheoryalso addressestheshort-run causes of war in thenuclear

era. Almostall analystsagreeon theimportance of "crisisstability." The exis-
tenceof nuclearweaponsmeansthatstatesare deterredmoreby the fearof
retaliationthanby theexpectation thattheymightsuffer military defeat.But if
one orbothsidesbelievethata first strikecan destroytheother'sstrategic forces,
a strikemightoccur.By contrast, theconfiguration is stableifeach sidebelieves
thatstrikingfirstproduceslittleadvantageor even,in theextreme case, reduces
thestate'scapabilitymorethanitdoesthatoftheadversary. In thiscase, itis very
hardto startan all-outwarandexpectations ofthewar'sstarting willnotproduce
a self-fulfilling as
prophecy, they do under crisisinstability.
Psychological processescontribute stronglyto thepresenceor absenceof
crisisstability(Jervis,1989, Ch. 5). Prospecttheoryin particular can explain
someoftheforcesthatmightbe atwork(forfurther discussion,see Jervis,1989,
pp. 168-73). Because peopleare willingto takeunusualrisksto recouprecent
losses, even if thesesetbacksare quiteminorwhencomparedwiththeirtotal
valueholdings,a decision-maker mightriskcostlyescalationorevenworldwar
ifsucha moveheldoutthepossibility ofreversing a defeat.In cases in whicha
standardexpected-utility modelwould predictthe actorto cut his losses, he
mightup theante.Thiswouldnotleaddirectly tonuclearwar,butitcouldcreate

This content downloaded from 137.189.170.231 on Thu, 05 Nov 2015 23:58:56 UTC
All use subject to JSTOR Terms and Conditions
198 Jervis

a situationthatwas extremely dangerousand one in whichthemomentum of

military moves might lead to theuse of nuclear weapons.
The dangerwouldbe especiallygreatif bothsides wereto feelthatthey
werelosing,something thatcouldeasilyhappenbecauseantagonists oftenhave
differentperspectives anduse different baselines.The MiddleEastcrisisof 1973
maybe an exampleof such a situation,withtheAmericansfeelingthatthey
couldnotallowIsraeltolose andtheRussiansfeelingthatat leasta limitedArab
victorywas necessaryto regaintheirinfluence.
A second consequenceof loss aversionforcrisisstabilityis thatif the
decision-maker thinks thata smallwar-and therefore enormous loss-is certain
if he does not strikeand thatattackingprovideseven a chance of escaping
unscathed evenifitrisksa muchlargerwar,he maydecideto strikeevenifthe
standardprobability-utility calculuscalls forrestraint.Similardynamicscould
operateon a smallerscale in less severecrises,suchas thosesetoffbya hostile
coup in an important ThirdWorldcountry or thelimiteduse of forceby the
adversary ina disputedarea.Withhisattention rivetedonthedeterioration which
will occurunlesshe acts strongly to reversethesituation, thedecision-maker
mighttakeactionswhichentailan irrationally highchanceof majorescalation.
The powerfulaversionto lossescouldlead thedecision-maker to attackin
thesituations describedin thepreviousparagraphs, butitcouldlead himto hold
backifhe thought thateventhoughstriking firstcouldbe advantageous, itwould
lead to certainretaliation whilehe mightbe able to keepthepeace ifhe did not
fire.As long as thereis any hope of avoidingtotalwar,decision-makers are
likelyto recoilfromthethought of startingit. Even GeneralCurtisLeMay's
successoras head of SAC, GeneralThomasPower,arguedagainstpreemption
"so longas thereis theslightest hopethatwe canprevent a Sovietattackthrough
diplomaticmeansof a strongpostureof deterrence" (Power& Arnhym, 1964,
pp. 80-81). Similarly, NevilleChamberlain arguedforcontinued appeasement
on thegroundsthat"we werein no positionto justifywaginga war todayin
orderto preventa warhereafter" (quotedin Mommsen,1983, p. x).
The responsecan be influencedby how the decisionis framed.If the
decision-maker takesas his baselinenottheexistingsituation, butthecasualties
thatwouldbe suffered in a war,his choicebetweenthesamealternatives might
be different. He wouldthenjudge thepoliciesaccordingto livesthatmightbe
saved,notlost,withtheresultthathe wouldchoosea courseof actionthathe
believedwould certainlysave some lives ratherthananotherthatmightsave
more,butmightnotsave any.The obviousdangeris thata firststrike,which
wouldsignificantly reducetheotherside'sstrategic forcesandone's owncasual-
ties, wouldbe preferred to restraintwhichcould notprovidethecertainty of
savingany lives. The result would notonly be unfortunate forpublicpolicy, but
is hard to square withrationalchoice theoriesif the personwould respond
differentlyifshefocusedon thepossibility ofavoidingwar,notreducing casual-

This content downloaded from 137.189.170.231 on Thu, 05 Nov 2015 23:58:56 UTC
All use subject to JSTOR Terms and Conditions
 ofLoss Aversion
PoliticalImplications 199

ties. Furthermore, as notedearlier,thereis also someevidenceindicating that

people are more to
willing acceptdisadvantages which are seen as costs(as in
insurancepremiums)ratherthanlosses. The evidenceis not strongand the
theorizing is a bitad hoc, butifthephenomenon is real it couldlead decision-
makersto takehigherriskswhentheseareseenas thecostofconducting foreign
policythanwhenrisksare conceptualized as a sacrificein thestatusquo. It is
hardto believethata firststrikecouldeverbe seen in theformer terms,buta
strategy that increases the chance of war could be.
Of course,inlaboratory experiments itis theprofessor whogivesone group
one frameandanothergroupanother. In realsituations, we needto explorewhy
people come to framethesituationin one way ratherthananother(Farnham,
1992).Alternatively, we shouldtryto see whether actorsmanipulate frames-so
thatan actorwhoknowswhatdecisionhe wantsothersto makeshouldframethe
questionin a way thatis morelikelyto elicitthisanswer.For example,in the
1930sappeasersshouldhavestressedthefactthattheirpolicyprovideda chance
of avoidingwar;anti-appeasers shouldhavestressedthatGermany wouldgrow
stronger-and Britishcasualtiesgreater-thelonger thewar was postponed and
thatappeasementwas certainto sacrificeotherBritishvalues,suchas power,
prestige,and morality. Thus as the Munichcrisisreachedits peak, the pro-
appeasement Britishambassador to Czechoslovakiastressedto theCzechsthat
they should not be lured into believingthatstandingfirmmightsucceed in
avoiding all losses- or
"refusal evasion[oftheAnglo-French plan]at thislast
moment meantthedestruction oftheirbeautiful country, while acceptancemeant
retention of mostof whatmattered to Czechs and Slovaks" (quotedin Taylor,
1979, p. 788).

RENORMALIZATION

Prospecttheoryarguesfortheimportance of thestatusquo because it is

usuallyfrom this thatgains or losses are computed.Butthereis a problemwith
theargument thatpeoplewilltakehighrisksifdoingso holdsoutsomechanceof
reversingrecentlosses.Onceone has suffered a loss,whyis notthenew-if less
desired-position consideredthe status quo, renderingthepersonrisk-aversefor
movesthatcould improveit (includingthosethatmightbringabouta return to
theold statusquo)? Thisraisesthequestionofwhatcan be called "renormaliza-
tion"--thatis, thelengthoftimeittakesforactorsto adjustto a newstatusquo.
Whenyouhavesuffered a loss thatyoucannotimmediately recoup,whendoes
thisget incorporated intoyournew sense of the statusquo? By mid-October
1962,didtheRussiansconsiderthestrategic balancewithmissilesinCuba as the
benchmark fromwhichgainsandlosseswouldbe measured? Iftheblockadeand
associatedbargaininghad not succeededin removingthemissiles,how long

This content downloaded from 137.189.170.231 on Thu, 05 Nov 2015 23:58:56 UTC
All use subject to JSTOR Terms and Conditions
200 Jervis

wouldit havetakenfortheU.S. to treatthenewsituation as thestatusquo? A

personalanecdoteillustrates some of thesequestions.On a recentvacation,I
triedtobody-surf an undisciplined waveandendedup crashing intothesand.For
a minuteor two I was paralyzed,and thethought thattheparalysiswouldbe
permanent obviouslycrossedmymind.But theparalysispassedquickly,to be
replacedbynastypainanda veryunpleasant stiff
neck.Iftheparalysishadlasted
muchlonger,I would have treatedthe pain and suffering thataccompanied
movement as a majorgainovertheotheroutcome.Butbecausemyparalysiswas
so short-lived, and I did nothavetimeto assimilateit, I comparedmycurrent
situation to thatwhichexistedbeforetheinjuryandso was annoyedandgrumpy.
In international politics-and in social lifein general-I suspectthatwe
renormalize forgainsmuchmorequicklythanwe do forlosses. We rapidly,if
noteffortlessly, adjusttogoodfortune andanyimprovement inourlives.Indeed,
we getaccustomed to itso readilythattheimmediate burstofpleasureproduced
bythechangesoondissipates.Neither individualsnornationsareso accepting of
losses, however.We remainunhappy, unreconciled,and oftenbitterfora pro-
longedperiod.Butwhatdetermines theexactlengthofthisperiodis notknown.
It is clear,however,thatit is theslow renormalization forlossesthatdrivesthe
risk-acceptance and indeed createssituations
in which bothsidesareimpelledby
loss aversion.Thus in the Cuban missilecrisisthe statusquo forthe United
Stateswas thesituation in whichtherewereno Sovietmissilesorbomberson the
island;fortheSovietUnionthestatusquo mayhavebeentheworldas it was
beforetheblockade.Of courseKhrushchev choseto withdraw (althoughKen-
nedy in turn made significant concessions).Perhaps he would have been less
accommodating if the installationhad been completed and the new statusquo
formalized by an official announcement.
A relatedproblemis thatwhena statehas made a pledge-even a secret
one-to changethesituation,it mayconsiderthatpledgedfutureas thenew
reference point.ThismayhelpexplainaspectsoftheCubanmissilecrisisandthe
Fashodacrisisof 1898.The Sovietshadstarted toputmissilesinCubabeforethe
U.S. made its formalwarningsin September1962; theFrenchmovedtoward
Fashoda beforethe Britishhad issuedformalprohibitions. In a fullyrational
world, itwould not matter thatthedefender made itscommitment-which surely
shouldhave been credible-onlyafterthechallenger had decided(butnotan-
nounced)thatit would move. But psychologically, thechallengers'decisions
perhapsled themto feelthatbackingawaywouldbe a loss rather thana forego-
ing of a gain.

IMPLICATIONS FOR SOCIAL EFFICIENCY AND STABILITY

Experimentshave revealedan "endowment effect"relatedto loss aver-

sion-being given an object increasesone's valuationof it (Knetschet al.,

This content downloaded from 137.189.170.231 on Thu, 05 Nov 2015 23:58:56 UTC
All use subject to JSTOR Terms and Conditions
 ofLoss Aversion
PoliticalImplications 201

1987). Studentswhoweregivena mugwouldonlypartwithitfora pricemuch

higherthantheyhad earlier(beforetheyreceivedit) said itwas worth.Whilein
laboratory situations thesestriking effectsdecreasesocialefficiency byinhibiting
trades,the implications forthereal worldmaybe quitedifferent. As people
endowtheirpossessions,lives,andperhapsself-images withgreater value,their
levelof satisfaction increases.Social stability is enhancedas an addedforceof
inertiaoperates.You likewhatyouhave-be ityourjob, apartment, orspouse-
andso areunwilling to tradeitforan objectthatwouldseemto an outsider tobe
of greater value. Tradesthatarerationalfroman expected-utility standpoint will
be rejected,presumably producing considerable But
inefficiency. greater effi-
ciencymightcome at too higha price.Imaginea worldin whicheveryone was
to
willing change what he or she was doing in the of
hope makinggreater gains.
Efficient in somesensesurely;butwhatwouldbe thesocietalcostsof constant
andrapidchange?Whatwouldbe lostinthesenseofthestability, regularity, and
solidity of social life? Furthermore, envy and jealousy would be increased if
people did not a
place higher valuation on what they have than on other's
possessions.The fact-if itis a fact-thatpeoplearelessanxioustochangethan
theyotherwise wouldbe raisesthegenerallevelofcontentment withinsocieties
andhelpsmakethemwork;theknowledge thatotherswillfight veryhardtokeep
whattheyhave increasespredictability and decreasesovertsocial conflict.
Internationally, theeffects probablyare less benign.Whilevaluingwhatit
has contributes to thestate'ssenseof well-being andso counteracts thedespera-
tionthatleadsto manywars,negotiations areinhibited as eachsidefeelsthatthe
costofmakinga concessionis greater thanthegainreceivedfromtheconcession
by theother side. For example, each side mightbe morepainedby givingup
1,000 missiles in an arms control agreement thanitwouldbe gratified andmade
secureby havingthe adversary'sforcereducedby the same amount.Or two
countries mightbe better offbymostmeasuresiftheywouldexchangesomebits
of territory,butthebargainwouldbe made impossibleby thehighsubjective
evaluationseach side placedon whatit had.
Otherdifficulties are caused by the "certainty effect"-thetendencyfor
actorsto over-weight outcomeswhoselikelihoodis I or 0. Thus whilean ex-
pectedutilitymodelmightlead one to expectthata statewouldbe willingto
makean agreement thatwouldsignificantly increasethechanceof peace at the
cost of havingto fighta warat somedisadvantage if armedconflictdid break
but,loss aversionwouldinhibitthisbargain.Statesmen wouldask, "If we sign
thisagreement, will we certainly be worseoffifwarshouldcome?" and,when
theansweris yes,theywouldconcludethatthisis toohigha pricetopayforeven
a largedecreasein thechanceof war.
A secondconsequenceis thatactorsarelikelyto be willingto paymoreto
reduceuncertainty to close to zerothantheyareto reduceuncertainty thesame
amountinthemiddlerange.Thatis, a policythatreducesthechanceofwarfrom
35% to 25% wouldbe valuedmuchless thanone thatwouldreduceitfrom11%

This content downloaded from 137.189.170.231 on Thu, 05 Nov 2015 23:58:56 UTC
All use subject to JSTOR Terms and Conditions
202 Jervis

to 1%, letalone from10% to zero. Agreements andmeasuresof thelattertype

are rare; statesmay not be willingto pay as muchas theyshouldforthe
significantincreasesin thechanceofdesiredoutcomesthatleavea gooddeal of
uncertainty remaining. Logically,arrangementsthatprovidecompletecertainty
shouldbe especiallyvaluableonlyiftheyyielddisproportionate suchas
benefits,
permitting the stateto disband its But
army. the of
attraction is not
certainty
sensitiveto suchconsiderations and cannotbe explainedby them.

METHODOLOGICAL DIFFICULTIES

Getting good evidenceforanyoftheseeffects willnotbe easy.The endow-

menteffect is difficult toverify becausestatesandpeopleinsocietyarenotgiven
thingsthewaypeopleinexperiments aregivenmugsandcandybarsat random.
Furthermore, it would be difficultto specifythatthe added valuationcomes
immediately rather than as a resultof the ties thatare builtup betweenthe
country and the territoryor other value thathas been gained.But it mightbe
possible to find cases in which countries conqueredorotherwise gainedterritory
almostaccidentally-areasthey had not been seekingas primary goals-and
thensee how muchtheywerewillingto pay to keepthem.
It is at leastas difficult to gatherfirmevidenceon themaininferences from
prospecttheory. Crucialis theargument thatpeoplewillrunmuchhigherrisksto
avoidlossesthanto makegains.The experimenter givesthesubjectsthepayoffs
andprobabilities; inreallifepeopleconstruct theirsubjective
estimates ofbothof
theseelements.Similarly, theexperimenter can framethesamequestiondiffer-
entlyandsee howpeoplerespond;in socialandpoliticallifeactorsdo theirown
framing. Thuswhilewe can findcases inwhichdifferent actorsemploydifferent
frames(McDermott,1992), correlations betweendifferences in framingand
differences in choice and preferences maybe spurious-thesame factorsthat
lead an actorto framethesituationin a particular way maydriveherchoice.
Evenifwe findthatactors'preferences changeas thewaytheyframethechoice
changes,thirdvariablesmaystillbe at work.
Showingthatpeopleare loss-averseobviouslymeansdemonstrating much
morethanthattheydo notlikelosses. Rather,whatis crucialis demonstrating
thatdifferences in risk-taking propensity varyaccordingto thedirection of the
expectedchangesfromthestatusquo. The difficulties are formidable.First,we
can rarelyspecifywithgreatprecisiontherisksthatan actorperceivesinvarious
coursesof action.Thirty yearsaftertheCubanmissilecrisiswe stillargueabout
how riskyKhrushchev believedhis movewouldbe. Furthermore, showingthat
peoplearemorerisk-acceptant forlossesthanforgainsrequirescomparisons that
are quite difficult because theycall fora measureof thesubjectiveutilityof
variousoutcomes.Mostdecisionslackan objectiveyardstick of livesordollars;

This content downloaded from 137.189.170.231 on Thu, 05 Nov 2015 23:58:56 UTC
All use subject to JSTOR Terms and Conditions
PoliticalImplications
ofLossAversion 203

we cannotreadilyfindcomparisons as convincing as thosewhicharisewhen

people are faced with the possibility of a loss or a gain of a givenamountof
money or a certain number of lives. Thus even if we can showthata statesman
tookriskswhenthealternative wouldhavebeento accepta smallerbutcertain
loss, we need also to showthatin othersituations he preferredcontinuingthe
statusquo to acceptinga similargamblethatmighthaveresultedin an improve-
mentequal in magnitude totheloss he foundunacceptable. Without precise-or
at leastdecent-measuresofthemagnitude ofthegainsandlosses(whichareof
coursesubjective),greaterrisk-taking in the lattercases can be attributed to
differences in the utilitiesin thecases. Of course,confirmation of expected-
utilitytheory is plaguedbysimilardifficulties ofmeasurement, butwithout a set
of verygood comparisons therewillalwaysbe a greatdeal ofroomfordispute.
Nevertheless, itis clearboththatprospect theory yieldsimportant andcounterin-
tuitivepropositions and thatmanycases-cited hereand in theotherarticlesin
thisissue-indicatethatstatesmen areindeedmorerisk-acceptant forlossesthan
forgains. The psychological worlddoes notappearto be symmetrical, and the
implications for human and national behavior are many and important.

REFERENCES

Coase, R. (1937). The natureof thefirm.EconomicaN.S., 4, 386-405.

Farnham,B. (1992). Rooseveltand the Munichcrisis:Insightsfromprospecttheory.Political
Psychology,13, 205-235.
Foot,R. (1985). Thewrongwar:American policyand thedimensions oftheKoreanconflict, 1950-
1953. Ithaca,NY: CornellUniversity Press.
Foot,R. (1990). A substitutefor victory:Thepoliticsofpeacemaking at theKoreanarmistice talks.
Ithaca,NY: CornellUniversity Press.
George,A. and Smoke,R. (1974). Deterrencein American foreignpolicy.New York:Columbia
University Press.
Hirschman, A. (1976). Development projectsobserved.Washington, D.C.: BrookingsInstitution.
Ikle, F. (1971). Everywar mustend. New York:ColumbiaUniversity Press.
Jervis,R. (1976). Perceptionand misperception in international
politics.Princeton,NJ:Princeton
University Press.
Jervis,R. (1986). Representativeness in foreignpolicyjudgements. PoliticalPsychology,7, 483-
505.
Jervis,R. (1989). The meaningof thenuclearrevolution.Ithaca,NY: CornellUniversity Press.
Jervis,R. (1991). Dominobeliefsandstrategic behavior.In R. Jervis& J.Snyder(Eds.), Dominoes
and bandwagons:Strategicbeliefsand greatpowercompetition in theeurasianrimland(pp.
20-50). New York:Oxford.
Jervis,R., Lebow,R. N., & Stein,J.G. (1985). Psychology and deterrence. Baltimore,MD: Johns
HopkinsUniversity Press.
Kanwisher, N. (1989). Cognitiveheuristics
andAmericansecurity policy.JournalofConflictResolu-
tion,33, 652-675.
KnetschJ.,Thaler,R., & Kahneman,D. (1987). Experimental testsoftheendowment effect
andthe
Coase theorem.Unpublished manuscript.
Lebow, R. N. (1981). Betweenpeace and war. Baltimore,MD: JohnsHopkinsUniversity Press.
Lebow,R. N. (1984). Windowsofopportunity: Do statesjumpthrough them?International
Security,
9, 147-186.

This content downloaded from 137.189.170.231 on Thu, 05 Nov 2015 23:58:56 UTC
All use subject to JSTOR Terms and Conditions
204 Jervis

Lebow,R. N. & Stein,J. G. (1987). Beyonddeterrence. JournalofSocial Issues,43, 5-72.

Levy,J.(1987). Decliningpowerandthepreventive motivationforwar.WorldPolitics,40, 82-107.
McDermott, R. (1992). The failedrescuemissioninIran:Anapplication ofprospect Political
theory.
Psychology,13, 237-263.
Mommsen,W. (1983). Foreword.In W. Mommsen& L. Kettenacker, (Eds.), Thefascistchallenge
and thepolicyofappeasement(pp. ix-xii). Boston:Allen& Unwin.
Morgan,P. (1985). Savingfaceforthesakeofdeterrence. In R. Jervis,
et al. (Eds.), Psychologyand
deterrence (pp. 125-152). Baltimore:MD: JohnsHopkinsUniversity Press.
Neustadt,R. & Fineberg,H. (1978). The swineflu affair.Washington, DC: Government Printing
Office.
Peterson,S., & Lawson,R. (1989). Riskybusiness:Prospecttheory andpolitics.PoliticalPsycholo-
gy, 10, 325-340.
Power,T., & Arnhym, A. (1964). Designfor survival.New York:Coward-McCann.
Quattrone, G., & Tversky, A. (1988). Contrastingrationaland psychological analysesof political
choice.AmericanPoliticalScienceReview,82, 719-736.
Ross, D. (1984). Riskaversionin Sovietdecisionmaking. In J. Valenta& W. Potter(Eds.), Soviet
decisionmaking for nationalsecurity (pp. 237-251). Boston:Allen& Unwin.
Schelling,T. (1966). Armsand influence. New Haven,CT: Yale University Press.
A. (1981). Purepoliticsand impurescience.Baltimore,
Silverstein, MD: JohnsHopkinsUniversity
Press.
Smith,V. (1991). Rationalchoice: the contrastbetweeneconomicsand psychology. Journalof
PoliticalEconomy,99, 877-897.
Stein,J.G. (1985). Calculation,miscalculation andconventional deterrence: theviewfromCairo.In
R. Jervis,etal. (Eds.), Psychology andDeterrence (pp. 34-59). Baltimore, MD: JohnsHopkins
University Press.
Taylor,A. J.T. (1954). The struggle for mastery in europe.New York:Oxford.
Taylor,T. (1979). Munich:thepriceofpeace. GardenCity,NY: Doubleday.
Tversky,A., & Kahneman,D. (1986). Rationalchoice and theframing of decisions.Journalof
Business,59, S251-S278.
Viorst,M. (1991). ReportfromBaghdad.New Yorker,June24, 1991.
Wittman,D. (1991). Contrasting economicand psychologicalanalysesof politicalchoice: an
economist'sperspectiveon whycognitivepsychologydoes notexplaindemocratic politics.
In K. Monroe,(Ed.), The economicapproachto politics(pp. 405-432). New York:Harper
Collins.
Williamson,0. (1985). The economicinstitutions ofcapitalism.New York:FreePress.

This content downloaded from 137.189.170.231 on Thu, 05 Nov 2015 23:58:56 UTC
All use subject to JSTOR Terms and Conditions