alternative approaches to the theory of consumer choice with

U N I V E R S I T ÀD E G L I S T U D I D I
SIENA
OUADERNI DEL DIPARTIMENTO
DI ECONOMIA POLITICA
ALESSANDROINNOCENTI
ALTERNATIVE
APPROACHES
TO THE THEORYOF
CONSUMER
CHOICEWITH ENDOGENOUS
TASTES
.s!!l
. .awsda.'
196
Alessandro
fnnocenti
ALEERNATIVE APPROACTTESTO THE THEORY OF
CONSUMER CHOICE WIEH SNDOGENOUSTASTES
Siena, aprile
1996
Abstract. According to or(hodox consulìlcr iheol')',tastes are usuaily considered ltxecl
or rlcpcndent cxogcnuusly on tirne. l{crrvcver, thc hypothesis that consullìer tastes
coLrl<ibe obtained a.s thc resuit of an endogcnous process hr,lsbcen invcstigated in
soine heterodox nrorlels. After having summariz.cclthc vuious mcaaings rvith which
tastr:s have becn used in ccononiics, a classiflcation oi the literaturc divicied in tw<.r
modeis based on a sclí-centÚredapproacli
larts is given. Thc first pcl-t dcais with the
choiccs inlluence his own futurc choilìe3.
cunent
lhe
cgnsulÌter's
acccrding io which
'l'hc
scccn{ palt is concerncd with the nlodels :isSuniingthat con5umers' chotces are
inleri.li:penclenf.lt ltìrns Oilt that a unifying approach to lhc case does not exrst :rncjCan
be provicìcd orrly by intcrdi:;ciplirra4' ar'alysis.
di EconomiaPolitica,univ*rsitàdi sienù.
Dipartimento
Inrlocenti,
Alessandro
I. TNT'IIOIjUCTION
Consunrcrtheoryhasnot jvetclearlydeiincd what is the nrcaningof tastesand
\i'hat are îhe relalionsbetrveentastesand prel-erences.
A commonmeihod to avoid this
issueis to regardboth concep'"s
as eqLrivalcnt
or, alternatively,to considefpreferences
as the onlv wa), throughrvhich tasrescan be know', a fact inplying that any attempt
to distinguishbetweenthe two is boundto fail. On the contrary,the view takenjn tlis
paperis that thcre are a numberof theoreticalreasonsfbr consideritg ta.stes
as distinct
frorn preferencesw'hich must be briefly mentionedbefcre dealing with the models
foundecj
on sucha disiinction.But in orderto avoidsemanticconlusions,the first step
tc)t;ike is to surnnrarizethe viuious meaningsof the two cenceptsas [hey have been
usedin econ(lrl.icliterature.
The theory of consurnerchoice dcfines preferencesas a binarl' reprc$entation
of rndividuai choice but emuloys this tool according to two differcnt theoretical
frlntcworks. Ttre fìrst considerspreferences
as given and takesthe choìcebehaviora-s
dcterrriinecib1' thc cunsumer'sprcferencc ordering. The second starts frorn the
consunter'schoicesand derivespreferenccsin tcrrnsof them.Both approaches
prcsent
sorne problcrn.swhich the consumer thecrist has to face. ln thc llrst approach
prcl.ciencesiu'c given independentlyfiour the siluation of choice. what tliis
assurdplioltimplics is tltat the only way to dcfine a cheice ir:;rational is ts assureits
intcrnal consistencyby inrposingsolnc abslractconditionson preferenceordering.On
thc coritrary, il a context-dependent
definition r:f rationality is acknorvledged,ii.
beconiesnecessaryto accountfor o'.herelernentsbesieles
thoseimplied by the intern;rl
featuresof prel'erence
ordering.Ta.stcsitrc oftcn invoked to play that roie perrnittingto
extcitcithe nrodcl of consumer'sLrehaviorbeyonCwhat is strictly irnputableto his
isolated action and to include the intelactiorrbetween his choices and his social
environmerrt.
For exampic,Rothenbelg(1962,p.28 l) writes that "The tastescf an
individual - even if we depreciatethe effect of advertising - are not neiuly scr
hei"oicallyhis. Ttrcy, and he himself,are only a lelatively stablestiuctureof organized
interchanges*rith a socialenvironnent".
In tlre sccondapproach,preferencesare'òJtex-postconccpt,rvhich ;ue def,rned
only alicl choicesare known. According to {his interpretation,observedbehavior is
fhe only source of information to ínfer tho consumer'sLrueprefelences.But if one
assumesthat the mereobscrvationof the eficctive choicesis insufficientto understand
the real prefercnccorderingof the consumer,it is then necessaryLoernploya different
conceptth;,rthas to be more subjectiveand introspectivethan revealedpreferences.
This role ef an cx-tutte conccpt is often playcd b1' lastes which econonristsand
decision theoristsreflerto in order to indicate the lelevant sct of values to better
his own choices.Sucli a role becomes
undcrstandlhe rvay the consumercleterrnines
clear in a dynlrnic settingwhenevertastesare employedto describethe way in which
pastconsurnptioninfluencesthe actualpreferenccordering.ln a classicalcontribution,
Gorman statesthat "it is cornmonplacethat choiccs dependon ta-stesand tasteson
are the
pastchoices"(Gorn,an1967,p.218),implyingthat tastesiurdnot preferenccs
basisfbr the.consumer'sintertemporalchoice
In oriicr to unily theseclifferentmeaningsof tastes,Gorman(1967)introduces
t h ef o l l o w i n gu t i l i t yf t r n c t i o n .
/
\
u = u\x.,a'1,
w|ere -x representsthe goods (or commodities) available for chgice and u the
generic"taste pararncters".According to this defìnition,tastesrepresentany element
clifferentfrom the objcctsof choicethat influencesthe consumer'spreferences'
givenby Pollak(1978),1thispaperintendsto survey
Followinga classification
how r.ccentcconomicliteraturehas specifiedthe tasteparametersmaking their change
enclogenousro the rnodelsof consumer'sbehavior.SectionIII deaiswith the models
that xdopt a seif-centercdapproachto the ploblem by rnaking consumer'stilstes
en his own pastchoices.SectionIV considers
cnclogenous
throughtheir dcpendence
rnoclelsassumingconsumers'choicesas interdependentby maliing taste pffameters
endogenousthrough thcir depencisnccon other consttmers'choices. But bclbre
cliscussingtheseditfcrent iipproaches,it is necessil'yto briefly ciescribein sectionlI
hcrv consrrmertheory genererliydeals with tirstes,that is by assumingthern fìxed or
dependclt exogcnouslvon tirne.
II. EXOGENOUS TASTES OR I}I' GUSTIBUS NON BST DISPUTANDUM
Usually, rhere are two ways for practically ignoring tastes.The first is to
fixed tastes.
assumethat tastcschangeexogenouslyrvith tirne,the sccondìs to a"rsume
I]oth assurnpticnsallorv one to avoid the problemsdiscussedin the introductiou;ulti
to desigrran abstractfiamework that is more convenientanalyticallybut also scarceiy
reievantÍol an empiricalanalysis.
The assugiptionol exogenolrslasteshas the r:ffcct of keepingihe validity of
preferenccs
changefor cffeqt
horvconsumcr's
the rcsuitsolrtaincdwirhoutconsirjcrrug
avr:idingany reaì;Jynamic
pastchoicesand of csls;enrially
t:,fhis or othef cotlsunìefs'
rlrawn
l'rom the iiteraîurecan
complicationin consumcrtheory. Some cxantplcs
r:xpiainthisjudgcmentl:ettcr.
1 P o l l a k ( 1 9 7 t ì ,p . 3 7 5 ) i ì i s t i n g u i s h r :tsw c ip r i n c i p a ls p e c i f ì c a t i o nosf e n d o g e n o u sl a s t c s ,î h a t
arc habit lbrrnation and interdependentproferences,ard nvo orhcr typc,s,that are tho,.einfluenceJ b;'
a r J v e r t i s i nogr b y p . r i c e s .
'l-his
usl o'i lhe conc*plof tastesjs cleal in Harnmond(1!i76)and (1977).Lî
bofh papers, flanrmcnci iìssumesthat tlie oniy u'ay to know consumer'stastes is
Lhr{r,-ish
his chcticcsand lhroLrghthis simplificaiion,hc is able to tackiethe problernoltlynaniicinconsistl:ncy.
In ilammcnd (19?6,p. 163) the char:geof îastcsis defined
rlrrougha pureiy rcvsslgdappr,-rach.
Any plan is indeedreve;rledin the sensethat ai
evcry choice nodc the cùnsurnerrcveaisit by the choiceof an imrneCiatesuccessorof
lhe node itseìi'.f)n this rccount,the necessifyof both connectlngtlie successive
chorcesin a sinele plan anclintclpretingsuctrii plan accordingio tastesdisappeais,
u'hile tastesare only invaked rvirenit is necess;rry
to explai.nan inconsistentchcice.
This also explains why l-lammonddocs not rccognize any difference berween an
exogenolls and an endogenollschangcof tastcs.Bcing preferencesthe only mea,r to
knorv tlte "tnre" tastesol u consumer,both typesof chalrgesareessentiallypreferences
getting reversedovcr time.2 ln anotherpapcr, Hammond (1977, p.338) indeecl
assuiresthat a consumertriìsconstanttastes."'rhenhis preferencesare "harrnonious"
accordingto Strotz's(1955-5ú)classicaldetlnitionand introducesa rlitferentterm to
expl:rínhis approach,namelyconsequcntia.lism
that ba-sically
concspondsto the belief
tliat a trchavioris raticnal ii and only íf it can be explainedb,vits consequences.
Sin'rilarly,
Johnsenand Donaldson(i985) definethe act of planningthrougha
siniple fì'amervork,in which a choice node is followed by a -setof chance nodes,
correspondingto thc set cf possiblestatesof Natlll€. lnci then by anotherchoice node"
l{aving l'uled out an'y'forni ol uncertaintl/or "surl'trise"frorn the model, Joh,nsenand
Donaltlsonstate thrìt if ;rn agent deviatesfiorn thc originaliy chosenplan after the
chanccnodes,this is evidenceof changingtastes.On such a basis,they anaiyzethe
condìtionsto impose on the rclation among the successivepreferenceorderiagsto
allorv consistentplanning.that.isa behaviorimplomcntingthe original plan whatever
state olltains. Therefbrc. tficir fonnal apparatusessentiaily confinns the resuits
obtainedby assumirrgrntertcmporalseparability,accordingto which the agent'sfuture
fron, and thus are separable
preferenccsare shapedbv considerationsîhat abstr-act
fronr, the backgrcundof cariierprotèrences.
If thesecxamplesshow how exogerrousiychangingtastesallow to ignore the
efttct of the parameterstastcs,the option of consideringtastesas fixed is comtnon to
most cconermicliterature. As weLl known, the standard way of rnodeiling the
and this methodpreventsany explicit an:rlysis
consur.ncr
is througli given prcf'erences
of t.astcs.
The nost quotcd fclerencìlon this issue is probably a passagctaken fronr
ìv{ilton Fricclinan'sPrice tlrcory {196?), where the author acknowledgesa ciear-cut
2 "'fh" es,sentialproblem with changing tastes, be they exogenously or endogenously
chtngrng, is that strict preferencesget reversedover time, and that, had ihe reversal been anticipated,
i l u i t e a i l i f f ' c r c n to p t i o n ( . . . ) ; n i g h th a v eb e e nc h o s e n "( l l a m r n o n d 1 9 7 6 .p . 1 6 2 ) .
division of labor betrveeneconomjcsand psychology,accorciingto which economists
qive up antrlyzingtastesanclieaveto psychologiststitc task of studyingthe processof
fbrmation anclchangeof tastes.S.Accordingto Stigler'sauthoritativereview aficle,
Slutsky rvas ihe fìrst autiior to put utihry theorl-on suclt a basis clarming that "we
assuniptionsanciphilosophical
must takc it courpletelyindcpendentof ps-vchological
"with Slutsky'sdc:velopment,
(Slutskyi915, p. 1), and as il consequence
hynotheses"
intlospectionno longelplaysa significantroie in utility theory"(Stigleri9-50,p. 3E2).
Frour thcn on, this stress on quantitative variablesrather than on subjective and
psychologicalconceplshas ch:iractcrizedconsumcrtheory more and more. Savage's
Fowtdutions of Statistic:swas the most influential contributiotl to overcoming any
distinction bctween tastesaníi prefcrcncesand to abandoninga subjectiveview of
prefèrcncesand utility. In this work, preferencesltavc to be interpretedin telms of
choiceswithout using introspcctionas a sourccof information (Savagei954, p. l7)
and thereibreprefercncesare lìot distinguishablefrom tastes.This position has been
recentiy rcvised by Stigler and Bccker's paper entitled "De Gustibus N<-rnEsl
axioms
I)isputarrdum".Their initial assumptionis that "tastesare the unchallengeabic
(usefuliy)
properly
criticized
inelTiciency
be
for
in
he
may
of a man's behavior:
satisfyinghis desirc:;,but the desiresthemseivesarc datu" (Stigier and tsecker 1977,
p.'16).Therelbre,cconomjcanalysishas to consiclsrtastesas data and assumethat
they ncithel changecaprìciouslynor diffel basicall;rùmong people. In this coniext,
app:rently irrationalbehaviorsuch as addictioii,habit or fashion,that conunon sensc
attribrutesto changing tastes,can be cxpiained without any referenceto tast$s.ln
particuiar,f,heconsuniptionof a bene.ilciaiaddictivegood like music is the effect of
ihe accumulationi-rf the "consumption capita.!",which sumntarizesa process of
'fhe
growth of this capitai increasesthe productivity of thc tirne
learning by' doing.
spent listening to urusic and, consccuently,the marginai utility of timc allocatedtcr
nrusic. On the contrary, if the good is harmful, the inversion of the same effect
expiainsthe dscrcaseof its consumptionover tiurc. In order to soive the contradiction
reprcsentcdby thc addictive and harrnfui goods like heroin, Stigler and Becker
irltroducean atl hoc hypothcsisexplainingtheir addictiveelfect, namelythe rigidity of
thc demmd for thcsegoods with respectto their shadowpriccs^The conscqucnces
of
this approachfor thc econonricmeartingof tasiesare clear in the iast {.wosectionsof
Stigler and Becker's paper. Hcre, advertising iniluences ccnsumer'sciemandnot
through a changcof tastcsbut lhrough a r:henceof the stradovrprices of the goods,
while the dcmanrJfor fashionis lbrrnalizedby introducinga nelv'rarial.lc in thc
J "Ectuonric theory proceeds largely to txkr wants as fixed. This is primarily a case of
divrsion of labcur. lhc econornisthas littlc tc say abDutthc forrnationof wanis; ttris is tlie provincc of
'lhe
the psychologist.
economrst'stiLsli is to trace the consequcncesof any given set of wants"
( F r i e d r n a n1 9 i i 2 ,p . l 3 ) .
Iti:'irsehoJ,-.1
soeiairiistinctionisee also
orilLiiìctianfunction.ndlreiv tire inrJrvidu;tl's
R c c k e ri 9 7 4 ; .
rvhichha.sextendedand tested
i'his contributionhascriginaleda vastlitei:alure
the,:ri.gin:iliheory.but has alsocausedothcr works full of harshcriticirm againsrit.
'l''he
r'"rih1lrs
slipporijn!-Stigiel and Rccker'sview are princinailyinvoived with thre
iltt.rrnptti: dciciop a theùnr of rationa!aiJiiicl.ion,
that is requiredbv the great
''Pctrple
ger;eralilvcf riuch;r phr:nomerron:
get addictecl
not only to alcohoi,cocaine,
ariilcigaret(es
but aisoto rvork,eating,music,television,theirslandardof ìiring, other
pec'pli,,
rcligicn,iurclnranyothcractivities"(Beckerandh{urphy 1988,pp. 675-6).liut
in this plper, as in the r:tirersdiscusscdin the next secticn, addictive behavror is
explainedby choosinga irroxy for tasteson a priori grounds,that seern.s
arbitraryor at
leastempiricallyinadequate
like StiglerandBecker'srigid demandcun'c.
point out fhe basicmisunderstanding
r,vhichcharacterizes
Theseconsirlerations
therdcbaiefoilorving "De Gustibus".The real problem in the denial r:f the economjc
importanccof changingtastcsis that such ir ncgationdocs not matter.Actually, the
whole of ncoclassicalcorlsunlcrthcory rdopts indirectmeansof rcprcsentinga change
in tastcs.In static rnodelsthis role is irsuallyplayedby the changeof clasticity of Lhe
consulr-lcr'':;
ciemandas in Stigler iind Becker's model. In metaslatic moclels, as
Hamniond(1977)delìnesthe neociassicaltechniquesof choiceover tinrr'.,the chirnges
Bur in a
of ta:;tcsare oftcn rcpresenledby tlie changesiri the rate of time prel-erence.4
real iJyr,anric
conlext,that is one involving decisionsmade after the resolutisnof
this setof toois appearsinsufficicntto fuily capturethe cvolution of
solneuiìccìrtiìinty,
lhc c()n:,rrmer's
brharior.
Although this point has nct vet been stressedsufficiently, most critieal
coiîments on Bccker and Stigler'sp;rpcr sharethis view. Marschak {1978), Pollak
(1918), Winston (i980), Hirschman(1985), Cowen (1989) criticize Becker and
Stiglcr'sapi:roachbecauseit avoidsdeaiingwith tastesthroughnothingbut a semantic
'Ihese
authors
changethat is nct rcally useful to explain the r:onsumer'sbehavior.
basiciilly underline how consumption capital or shaclowprices have no greater
explanatcrypower than changing tasl.esand how the whole householdtechnology
approach,which consiciersthe coirsunleras a producer,canttotrepiacepsychological
inquiry.
In conclusion,the vìeu,,takenhereis thal tirstcsare implicitìy presentin a large
put o{ orthodoxconsurneít}reory,whilc Bcckcr and Stigler'stheoryis subject.to tire
sarnecritique made try Sen (197i) to tlie revealedpreferenceapproach,namely it is
not fbunded on tlìe absence of any psychological hypothesis but on ad hoc
is mearuringits
psyclrologicalassumptions.
Then tirc bcst wa1,to assessits usefìrlness
4
F , r , a n h i s t o r ì c aol v e r v i e w .s e el , o c w e n s t e i n( 1 9 9 2 )
capacityto theoreticallyrepresenta corlmon empirical phenomenonas tlie changeof
tastes.In this light, the assunlptionof a rigid demandcurve lor drugsor otiler hifmful
addictive goods is nothing but a rough method to describe a more compiex
psychologicalprocessthat may be explainedbetterby explicitly consideringtastesas
in the ncxt sectìons'
as in the modelsdiscu-sscd
cndogenous,
III. ENDOGENOUS TT\STES: THE SELF'CENTERED REPRESENTATION
Gorman (1967) statcslhat past and actualchcicesare connectedby tastes,But
wiriie in his pathbreakingpapcr Gorman studied only the conditions assuringthe
cxistcnceof a genericchoiceindicator.the utility function he proposedcan be usedto
classily the nio,Jelswhere tastesare madc cndogenousthrough the introduction of
6abit formation processes.In this light, three cli.fferentspecificationsof the t&tte
pafameterscan be recognized.The first approach can be defined as the itciie
approach,becauseit assumesthat the consunlerfails to recognizethe iull impact of
thc tasteparametersare
In tiiis ca^se,
thc habit fonning effcct on his fulure preferenccs.
r"rsuallyreprcsentedby the consunìptionveclor of the past period, xt-1, and the
by a short-runutiliiy function.The seconCand
consumer'spreferencesare describe<i
tiie thirel approachcan bc ilefinedas rational becausethe consumeris supposedto be
of the mannerin rvhich his consumptionat any
tìliy aware anil to take full accor,rnt
ntoment al-fectshis luture prcfcrences.This implies that both approachesassumethe
utility functionas tho objectof anaiysisbut they adoptdifferentmethods
interLemporal
!o maximizc it. One classof modcls maxinlizesutilily on the rvholelclevant horizon
arìoptingeither the backrvardinduction mcíhociproposedby Strotz'sclassicalpaper
(1955-56) or a solution concept cxpiicitly designedfor strategicmodeis, a.sNastr
cqLlilibriuni.In this case,the tastcparameterscorrespondtD the sct of all the short run
utility iunctions in all thc periods relevant lbr the colìslltner'schoice. The other
ratícnal approachstartsfrom a long-runutility function referredto a singleperiod but
iutrqgucesa state variable that sunrmarizesthe past eff'ect of the habit formation
process.Then it usuallyemploysstandardoptimalcontrolthcoryto obtaina solution.
by the statcvariubieantl
are iust rcpicscttred
[rr thls framcwork.tlrc taslcparrmeLcrs
utility as a functionof pasthistory.
tlìeexpccteddiscounte<i
tlie consumcrrnaxirnizes
Tl'henatlt,eapproach,accordingio the standardversion (Peston1967, Poilak
specifiesthe processof
1970, .ron Weizsàcker19?1,Itelle 1973,Gaeriner19'14,\,
vectorof the last
habitfolmationrhrougha utility function,inclrrdingtheconsumption
period,
u = ,(.t, , r, -t)
Iil anotherver:;ion'(Fnìlak1970 ancjMcCaiihy 197-1),
.r,_1is substituteci
witìr tllc
geometrically
weightedaver;lge
of consumption
in all pastperiods,
ur\ ^ i , t'y, i , t\* t ) =u( )* i ,., (t - 6. , ) E: ó f ,^r ; , r _, , _1, 1
,
\
j
=
0
)
vhcrr 0<-6 ;<i is a memorycccificient.
Poilak{1970)studlcslhe speciaicascin which tìreshori-nrnutility functionis
(or 1,1-i).Thrnugh a siniple nraximizaticn, he
a<irlitivcanr--l
elependslinearl-von -11-1
obtainsa localiy slablesystemof strort-nrndemandfirnctionsthat are locally iinear in
incoinc5 and derivesthe long run demandfunction as the stead-vstaiesolution of the
sllcìil-rundemandfunctions.This mcthod raiscsiiie problen.rof the existencesf the
ìong-run demand function. Pollak (1976) s;howsthat existenceciependson the
chaiacteristicsof ihe utiiity function. [f the short run demand function is indeed
dctcrrineei by maxirniz.ing{i@nx,-1,\subject to the budget constraint,the iong-run
demand function, that is, the steady staîe (,t,=xr-,=x) of thc short-ruri dcmand
functicrns,is not given by the rnaximizationof U(xX) but of U(Ì), rvhich takes fuil
írccouutof the effect of the habit formaticrnprocesson presentand futui'epreferences.
Vou Weizslcker(1971) doesnof acknowledgetltis problernanc reslrictshis
nttentionlo the cascof' Lwogoods, i=|,2. Fie stalts directly from the implicit short run
demanrJeqr.rations.
.ri,,= r;(y,1,,,112,î,
/-t. xt,t-t, rl,t -t)
wherc 7; and } i66i.',,e prlces and incorne.lle clairns - irnproperlyin the case with
more fhan twr: goods {Poliak 19?6, F:l-Saliy 1916, HainrnunC l9"l6a) - thet
integrabiiity conditions iue fuifillecl and thercforehe considerssmooth and convex
short-rurrindifferencecurves as the ncìcessary
iurclsufficient condìtionsfor having
weli behavcdlong i-unindiffcrencecurvesas rvell. Then von Weizsàckeranalyzesîhe
propc:rties
of the stearlysîatedemandfunctionsso obtainedand shclwsthat, ciepcnding
on the diinension crf the adriictivc cffccis (òx/î)x,-), it is possibie tliat a change
evaluatcdas inferior in the short-runmay be consideiedsuperiorin the light of the
steady state indìffercnccfunction, a fact tlìat has relevant implications in terrns of
srrcial weltare irnd pubìic choice..Apart from lhese conscquencesthat do not
gcncralize beyond {he two-good case, it is interesting to i.inderlinea different
ir:terpretationgiven to the modcl. According to von Weizs;ickerhjmself {197I, pp.
358-ó0), ihe short run utility function may correspondto a singÌe trial performedin
orclerto reacha "satisl-rcing"(à ia Simon I95'i) utiiity level, while the iong nrn utility
functioncorrcsponds
to thc intrinsicleveito which the consumeris interested"
-5 Irolluk'r mo<iel adopts the Klein-Rubin iinear expenditure systcm implying demand
f i r n c t i c ' n sw i t i r l i n e a rE n g e lc u r v e s( s e eK l e i n a n i j R u b i n 1 9 4 7 ,S t o n e1 9 5 4 ) .
'I6e
introductionof the level of aspirationor satisfactionas a tool to describe
in
the consumer'schoice appears,amongothers,in Krellc (1973) and, more recently,
Gilboa ancl Schmcidler(1993). Fisher and Sheil (1968)' Houthakkerand Tayior
of leamingby
(1970),Phlips(1972):ui<tEl-Salty(1976a)int|oducesimilarprocesses
of a state
evolution
<ioing loilowed by the consumcr thai are representedby the
by the magnitude
variable.fn El-Safty'sversion,the tasÎepararletersafe represented
si the habi,.strength,à, rvhichfallows ll pfocessgiven by the differenceequation
à , = ( l - 6 ) i r r - l* . r r - i ,
of memory loss.ll the consumerdoes not anticipate
representsa úleil^surc
the fiture changesin tastesancl follows rhe ncù,-ebehavior,tire model's solution is
very similar to that of Poilak's discrctetime lbnnulation. But in his paper,El-Safty
where 6
two other results,namelythat the dcmandvectorconvergesif and oniy if the
cllscusses
arhptation process,capturedby the irabit strengthcoefhcient, convergesaxd that
conditionsassuringthe existenceof the long-rundemandfìinction have io be adapted
to ihe functionalform of the short-runutility ftrnction'
The retional approachexcludes the possibility that the consumer may be
limite6ly rationi:ilancl assumesperfect intbrmation and fuliy optimizing behaviorTheschypothesesmlke this approachtrt for a normativeiheory and allo$' us to reduce
the chiingingtasrescaseunderthe sameanaiyticalframeworkusedfor fixed tastes,but
modci'
weakerienormouslyits plausibiliryas a descriptive
that an optimal consumptionpian
assumes
A {rrst variation of this approach
can bc c;rìculatedirpder changingtastesby giving the problem a recursivestructure
and then solving it throughstandarddynamicprogfalnmingmethods.Accordingto the
developedtry Pollak (1968) and Pelegand Yaari
criginal intuitiorrof Strotz ( 195-s-56)
in orcJerto distinguishit
(1973),this methodof solutionis definedas sophisticated
from a rnyopic one. The principal problern with this appfoach is that analytical
conveniencerequiresus to adoptintertemporalseparabilityin order to apply recursive
methcrds.An cnlighteningdiscussionis cclnlainedin Yaari (1977)' lvho propose$a
<tiffcrcntsolution. He introclucesa i-rabitformation processby making the hypothesis
rlxrt a deterministicprocessgovernsthe changesin the utility function and then by
employing the rnoclclso obtainedto solve the dynamic inconsistencyproblem. Very
simply, the optimal consumptionpath is given by the standardrecursiveprogrammlng
with finite horizon I, in which the last lth choice function i.sderive<ifirst
pr.ocedure
ancl the first choice function is derived last' if x, is the consumptionvcctor to be
ccnsunredin period t, and thele is perfcct knowledgeof all prices and incomes,the
is thc solutionof thefollowingsystem,
optimaiconsumptionplan (x1+,....xrd')
+
[
*
t i = n t t L r t / t l r l , i r 2 i)x,i1 4x{ i
'i =o=('i)
:
*
+
+
, { +
X . , - . l l ' r \ . 1 ' , - f1 , . . . , I -
ì
I
irhere /r, j'; thc iunctionthat n'raximizes
ihe quantityUr, that is the utility function
plan(-r;,..."t1).
defiiredat timc I fi:r eveiy fcasiblcconsumption
'Ihr-:
existenceproblemsconnectedwith this solui.ion.generatedby the noncclnvcxfi:asiblescts{Bhckcrby,Nissen.Prirnontand Russell1973),and the fact thaî
it may be Pareto-dominated
have suggestedadoptingother solutions.An altemaîive
rurethodis to rcpresentthe intertemporal'decision
throughthe f;ction of morc selves.
(Phelpsand Poilak 1968,Ainslie 1975"Yaaril9l7,Elster 1979,
This line crfresearch
Winston i980, Schclling 1984) assumesthat dynamically incoherenthehavior
or'ig.inatcs
because;rn individual behavesas if he is composedof two or more selves
deterrnincintertenrporaldecisions.ln this analyticalframework, the
lhat alf.ernatively
processof individual choicc can be analyzedcmploying game-theoreticalsolution
cclnceptsand the solution to thc Cynarnicincoirercnceproblcm can corcspond to
diffe.rentcritcria for orderingthe successiveselvcsof the sameinclividual.Hovrever,
thìs methoci,as the otherone irrcposcdfor dealingwitli clynamicinconsistentbehavior
that is precommifrncnt,seenlsinappropriatefor model.iwiih endogenoustastesthai
of l.heintertemporalseparabilitya-ssurr,plion.
are founder,lon thc aLlancionment
lntieed, a inLìrecommon alternativeis îo incorporatethe etÎect of past anil
future prefcrcncesinto current demandsby reiaxing the assumptionof intertemporal
seLìarability.
In generaltenns, thesernoclelsintrociucein the utility funclion a staie
variable reprcsclrtingthe stock of habits. This 'rariable changes continuousiy
accordingto ii law cie-scribed
by a differenti:rlcquationthat breakstime separabilityby
aCdingin lags to the argumentsof direct utility. Threresult is that the classicalmodel
of lifc-cycle utilitv i,''itl.outhabit lbrmation can be employedto solve intertemporal
rvirh hzibitforrnation.Such an expcdicntis t'unctionalto safeguardthe validity
mr-rciels
of íhe standarcimethodsof intertcmporalmaximizationbut has seriousshortcornings
botir as a dcscriptiveurot1elatrdris a refèrencefor empiricalresearch,
between models with and wilhout habit formation is
The correspondence
e;iplicitly clairncd by Lluch (19?4), who proposcs to lrraxil'tìizethe follorving
i n l r rt c n r p o r uul t i l i t vf u n c t i o n ,
u (x( t ),s( t il = I; e-Pîu(x( t ),s(t )\1t
In this model.the vectorof the stockof habits,s, changesaccordingto
s(r)= ;t1t';-6''1t,
whcr,: ò is the constantrateof the depreciationof the stockof habits.
Ifanoptinralprogranrexistsandtheproblemsatisfiessuffìcientiystrong
by optirnal control methods' Lluch's
smoo{hnessconclitrons,the solution is given
cliffersfrorn thc solutionwithout habit
discussicnpoints out how the optimal solution
lbrmationonlybythein|rocluctionofsomeadditionalconstraintscapturingthe
prc-scntandfutureutilitycff'ectoftl.rcchangesinthestockofhabits'Forexanrple,the
price oÍ'the addictivegoodsgiven
marginaluriliry of habitshas to be equalto the tull
luture utiiity ctlstsof the addictive
by its discounteclmoncy price plus lhe tliscounted
with standardlifle-cyciemodel
stocks.Moreover, l-lrich confirrnsihe corrcsponclence
to make the solution explicit and
adopting log-lineiir anclquacraticutìlity functions
namcly that a high rate of time
cliscusscssornc empiricaliy relevantconsequenc.es,
more slowiy (seealso Phlips 1974,p' 258) afid that
prefercncemakeshabiis clisappear
theanticipationofafuturcpricegrowthlowcrsaddictedconsutt-rption'
by Spinnewyn(1981)who
.
A s*bsrantialirnprovemcntof the model is obtained
to the modelswithout haitit tormatton
makesmotlelsr.vithhabit tbrmationequivalent
of thc consumptioncostsinduccdby habit lornìationprocess
tltrouglrthc rmpututiorr
ot'a commodityis the
to the prlstclecisions.ln sucha way, tlie cost of the consumption
sumol*ìeculrentprrceanrlolihcfutureconsumptioninducedbythecurrentchoice.
Thcprincipaloutcorneisthattheconsumptiono[themoreStronglyhabitforming
in the short run ìn favour oi thc less habit fbrming ones'
corru'otlities is rlr-ccouragccì
but its effect dependson the
Such a tendencycould be reverscdin the long run
specific tì;rm of ihe utility function adopted'
solutionsare obtainedby the autliorsfollowing Beckcr and Stigler's
^*\nalogous
in thc householdproduction
suggesuotlro represeiltthe cl'rangeof tastesas a change
198ó,
ianna-ccone
possibiliticsgiven constanitastes.These morlcls (lìoyer 1983,
anctMurphy 1992)
Beckcr aitd.ivlurphy 1988, Chaloupka 1991, Beckcr' Crossman
to rationally cxplain
adopt Lluch,s tîamcwork, defining first what conclitionsallow
thc rnaximizatinn of an
;ickiistion bei.ravioranci then cleriving solutions ttuough
by the fact tlÌat
Flamiltonian.A partial cieparturefrom Lluch's niodel is represcnted
bctweenbencficial habits haviirg positive marginal utility
thesernodels<iiscnnur.iarc
on the optimal
an,Jthe harmful ones anclexplicitly treat the effect of this distinction
problem:
,tn*Î, t:}."t: the foilowing
pians. For exantple,Ianrraccone
{J\x( t t' s(t 1)= ln c-ptu\x( t I' s(t ))
.r,r. s(t) = f {x( t i, sl t ,})- Et
A=rA-pY
gives thc conditionstor hairit formatiun,
l{amiltonian
The soiution of the
that is a measureof the
on adiacentintertempolalcornplerneniarity'
which <icpencl
the same roie tlf the
inreftemporalelasticity of comrnodity substitutionwhich has
(1988)discuss
ilcnrandelasticityin Stiglerand Becker'smodel'Bcckerand ìvlurphy
tt)
the possibilit,vthat adjacentcompiementaritvcan irnply unstableanC multiple steady
stales. a result explaining sorne empirically relevant facts conceming addictive
behavior.
The principal problem posed by this approach, as Winsron (J980) or
I{errnstein and Prclec (1992) argue, is that rational addíction is a contrerdictionin
terms, in the sensethat "for a behavior to be cailed an addiction,rather than just a
personaìbent or appetite,it must be unwanîed' (Herrnstcinand Prelec L992,p. 335).
According to this view, addictioncannotbe treatcdwithin a rational ciroiceapproach
and it is rather necessatyto appeal either to behavioralmodels or to the already
mentionedmore selvesapproach.
A radicaldeparturefrom thc previousmodelsis proposeCby Browning (1991),
who develcps an intertemporal non-separable modeJ. He empioys a dual
givcn by the follGwingprofrt function,
reprcsentation
for preferences
t
\
l i | 1 ; r , .p. .r,, ^ ; ') = ^
t
[
l
l [ ' u ( x 1 , . . . , 1 2 . ] - I p ,Ir,,
L
t
J
p,,,...,p,and f, 11.Il this expressicn,
that is linearhornogenous
anddecreasingin
l, ,
is thc Lagrange mLrltipiier - or the marginal utiliry of expenditure - sf the
intertcmporalutility rnaxirnizingproblem. Since optirnal allocationrequiresthat the
cliscountedvalue of.L
be kept constantover ihe whole horizon, we can employ it to
representthe cvolutionof futurewealth.Tire functionalforrn adoptedby Browning for
the prolìt function is callcd simpìe non additivc preferences(SÌ.JAP)and dependson
onc-periodlagged,onc-pcriodleadand currentprices,
r-l
r\
-l
r
^-r\
n t p r , . ,t , r , L ;),= _ _
I o , \ n , n, t _ 1 , ))1. '
/=i
Fron: this llnctioii, Browning derivesthe 1, -ccnstant(or Frisctr) demand function
har.ingthe following form:
(p,-,,r,,.\')
x,= à<D,-1
p,rr,Lì') .
t'òa,(p,,
fan,
lèp,
This moiicl, which definescurrentdemandas the sum of a cornponentbasedonly on
past dernandand a componenttaking account of its effbct on titure preferences,
alioivs us to clearlyreject time separabilityanclto obtain csfirnateson aggregatedata.
Dcspite the progressachievcd by this paper, a reconciliation between theoretical
riynamic rnodeìling and the availablc ernpirical evidence remains the principal
problem to solve in this researcharca. Sonrc attemptsa.recontainedin Chaloupka
(1991), Becker, Grossmanand Murphy (1994) and Loewensteinand Prelec (1992),
but this recent woi:k does not prescnl an analysisbased on micro-data,which are
essentiailyto producethe interplaybetweentheoreticaiand empiricai considerations
that characterizestandardconsumertheory(for a survcy,seeBiundell i988).
l1
IV. EI{DOGENOUS
TASTES:
THE
INTERDEPENDENCE
AIVIONG
CONSUMERS
.
As the previous section has shown, economistshave paid great attention,
cspeciallyin the last two dccadcs,to the seli-centercdapproachto consumertheory.
On the contrary, thc5' have hitherto wideiy neglccted the importance of the
interdependence
bet*"veen
consumers.This fact is particularlysurprisingif we consider
the actualstresson game-thcorctical
conceptsand techniquesin economics.However,
the use of interdependentpreferencesas a way tc make ta.stesendogenoushas
characterizedin the past the work of some borderlineeconomists.Veblen'sclassicLrl
Ciscussion,Leibcnstein's1950 papcr on bandwagonand snob effect. some wc;rks
publishedin thc sevcntiesand the recentrnodclspertainingto infonnation econonlics
represeniresearchawaitinga filore thoroughstudyand extensivedcvclopment.
In order to classìfythesecontributions,it is useful to recognizethree differcrit
meansof representirigthe interdependence
amongconsurners:
a)
connectingthc consunrers'choicesthrough a medium of influence that is not
directly controliedby themselves,
i))
introclucìngbiiateral direct reiations among the demandfunctions of idcnticai
cotisiJrners.
c)
torrnallzing a socìaÌ crrdcr tbunded on unilatcral reiations among cliff,erent
consuiììers,
St.;utingfrom the finstclass,the introducîionof priccsin the utility function is
tlio most crrntmon
of rcpresentinginterdependent
pret'crences
througha variable
"';ay
diff'erent liorn the dcrnand function" T'heremay be two dift-erentreasonsfor this
inclusilin. The flrst - thrti is tlre interestingone here - correspondsto the well-known
Vcblen cffect associateC
r.vithtlie expressioncon.spir:Ltous
consurnpîion.The sccond
consistsin judging quality by price and can be viervedas a direct ccnsequence
of the
difficulties in coilecting infornrationabout the real quality of goods. Economic
analysis has usuaily ignored this distinction claimìng, for cxan:ple. that "when a
socicty begins io appraisethe value of goods ú;r iheir price, it may enclby valuing
theiny'r,"tticir price" (Scitovsky 1944-45,p. i04). Bui it is irnportantto nore ihat thc
thcoreticaiequivalenceol lhcsetwo pattemsof behavioris <.ioubtful
bccausetheyhave
quiie Lirrfereirr
chliucteristicsiind purposes.ó
E.,'enii both pattems.iinpl,va dircct
,..vccn
rcìation bct
utiiity anri pricc, only ìn the cascal snttbappt:elprice is a vlriable,
{ ) . r . l c a l yl n C K l e v o r i r : k( 1 9 7 0 ) t a k e sa c c o u n i
o f t h i s r l i s i i n r : i ì u na n d a p p l i e si t t r ) L a n c a s i e r ' s
co'rsurncr tlrcrrrv. Snob cffcr:t is repl.esentcdby a specific price attribute ìn<lependentfrom Ùthcr
characteristics,rviriie, rf lhe ctnsunrct judges quaiity b1,piice, therc mly bc complcrnentalt], ir",***n
p ; - i c ec h a n g e sa n t i t h e v l r i a t ì o n : ;o l ' r h es p e c i l ì cc h r r a c t c r i s t i jcu d g e d b y p n c c .
It
expiaininqchoicepcr sc and.thereforc,a mediumof influenceamongconsumcrs.
On
the contraia, when the consunlerjudges quaiity by price, prices are inclurtedin the
urility function as a proxy for sone featuresof the goodsthat are imperfcctly known
or complctelyignored.within this fiamework, the most appropriateway to consider
priccs as reliablejndicatorsof the nussinginformation is to considerthem as a result
ol thc interdcpendencebetrveen produccrs and custorners rather than arnong
consumers.This interpretationis cleifly pointed out by Neison's work on the
economic consequenccscf advertising(Nelson 1974, 1915) and by Milgrom and
Robcrts'signalling
nlodcl(1986),which will be discussed
belor.r,.
ln his book on lhe leisureclass,Veblendescribes
conspicuous
consumptionas
"a rneansof reputability to the gentiemanof ieisure" (Veblen 1899, p. 75) through
rvhichthe leisureclass{irnily cstablishesits upperpositionin the social iúerarchy.But
such a bchavior spreadsbeyondthe higher class becauseemulation and enry- make
this phcnomenon pclvasivc. Therctbre, the purchasc of higher price gclods"
independentlyfrom their capacityto provide satisfaction,iras to be considered,in
Veblen's approach, not a waste but a norrnal consequenceof ihe principle of
consumcr'^s
soverergnty.
A model that formalizesVcblen'ssociologicalapproachis containedin Follak
{1977). Poliak startsby claiming the independencebctweennorrnal pricesptt, which
influencepreferencesand depcn<ion both current and past prices,and market prices
7:'n,which enter the budget constraint.l'hen he defines a nomral price dependeni
plcferenceordcringand tlclives an utility function u(x, ptt)to bc na;iimized, subjecteo
the budgetconstraintl.tp,r-y.In the demandfunction so obtained,
,, =,,(.pl,Y,,pi).
the following rssumptions
apifap,>-o
perogni ^cSî
"
deseriberhe Vebten
":':.1u"
Within this mp<lel,a changcin the niarketpricesaffectsboth budgetconstraint
and utility function and makes tire demand function non-homogenous.in order to
restorehornogencity,T
Pollak assumesthat prefcrenceschangeonly wìth variationsin
relative rathel than absolute normal prices, implying that demand functions are
homogenousof deglee zero in current priccs and income. A^sidefiom the clear
destructivcconscquences
for iiny welfarejudgement,Pollak himself pornts oui thai
/ Before Pollak's paper, Kalman (1968) studiesconditions guaranteeingthe homogeneity of
a demand lunction clerivedby price dependentpreferencesbut he does not distinguish between normal
and rnarkei price. An annoying consequenceis that infiation increases utility. In Allingham and
Morisi,ima (1973). tlie rirtio betweenprices and generalprice level is substitutedtbr monetaw prices.
price dependcntpretèrencescàuseanotherproblem. If one assumesthat in steady
state, market prices are cquirl to nonnal prices, the demand luncLion can be
rationalizedby more than one price dependentpreferenceordering and this causesa
revealedpreferenceapproachto fail.
In a quite different inierpretation.Gilboa and Schqreidler(i993, 1994) have
t'ccentlysuggestedconsideringpriccsas a tool employedby consumersfor selectinga
"satisficing" behavior based cln aspirationlevels. They adopt the follorving utility
function.
b = ui\vi,Wi)
x,ith
òu;fAv;
>0
òrr'/A(upr)<
o
where v; is the "intrinsic" utility of the good i, conespondingto the consumer'stastes,
1.r;is its pricc and c ìs a "salicnceof money" coefficient measuringthe importance
attachcdLrythe consurnerto prices.Accordingto a behavioralapproach,o
increases
(decreases)
ievel, imply
if thc consumer'schoices,determinedby his actuala.spiration
an expenditurehighcr (lou,cr)than incomeand this changesttredegreeof dependcnce
between prices and prefcrcnces.But accordingto this scheme,thc interdependencc
'L.ecause
among consumers,if it erists, is only indirect,
it is an indirect eflect of the
markct activity.
The same rernark is appropriatefor desr:ribingthe w:ry in whictr ccorrornists
irave empìoyedother mediums of influenceto describethe interdependencc
among
consLrrnel:s.
tJsually these models are too closcly related to psychological and
sociologicalargumentsto be exhaustivelytreatedwithin ccononiìcs.A simple rnethod
to distinguishecoiromicrnodclsfrom thosebcìongingto other socialscierrces
is Lo
refer fircrnto th,; subjectsinvolvedin theserelations.Accordingto this convention,
advertising bclongs to thc realrn of economics becauseit concernsthe relations
betrvcenfirms and custonjers.The residual class of rnodels employs a llumtoerof
factors as pcrsonality corc, social roie identifications,moral and political sua-sion,
audicnccefl'ccts,educatir:naud cuslttm,fads and fash.ionso far as they arenot directly
createdand controlled bv producers.Even if such factorsha-r'eaiso been studieclby
economists,b
their ciirectanaivsissccnlstc,odemandingfor a quantitative
sciencelike
economics.
The judgemerìt about the ttreoreticalnalure of advertising is even more
cr-rntroversia,l.
As rnentitrnedbeforc,most literatureon the subjecttri€s to analyzewhy
firms may chooseto advertiscrather than study how advertisingii1'Íbc[sconsumers'
choices. even if consideringthe latter aspect involves necessarilyextia-econonric
qucstions. In order to :;olve the. problem, a distinction betvieen int-armativeand
Ó Trvo exantples, iiiat try to ovcrconre this perhaps t()o strict divisron ol labor, are Cintis
( i 9 7 4 ) o n e d u c a t i o na n d C o s g e l( 1 9 9 4 )o n t h c a u r l i e n c e l l e c t .
t4
persltr:;iveirJverl.rsiiig
lrrs bcerrrnatlern litcrrrture.9
[:r ihe llrst case,the effcct uf
adverrisingis to "correcrtasres"(Rothenberg1962,p.279) ralherthan io change
them,rvhile in the secondciise,consunlertastcszue"moulded" by the Droducers.
Thc approachusually ascribedto Richard Nelson clearly belongs ro the fi.rsr
class. ln his v"'ork,advertisingis descrìbedas a signai of a high qualiry product. Ia
order to nbtain this result in the caseof experiencequalities which are qualitiesof the
brandsnot determinedprior to purchase,it is sufficient thaî the consumers'repeaied
purchasesrevcal the real quaÌity of the product becauseit is advantageousfor high
qu:rlity producersto adverlisenrore than low quaiity producers,But as Milgrorn and
Roberts(1986) and Wolinsky (1980)point out, the sameresultcan be obtainedby
employing priccs and so the theorist must explain why effort should be wasted itr
advertising"Apart fiom this problem, not t:asily solvabie in a puiely economic
approach,l0thsre are clcar reasonsto exclude this line of researchfrom consurner
thsory $ilh endogenoustastes.The simpler reasonis that thesemodeis, in line with
(1972), considertastesas entirely exogenous
thc classicalapproachof Schmalensee
and cxplicitly excludethe faci thatconsumer'stastesmay be cirangedby advertising.lI
They are built over a strict asymmefricalrelation betweenproduceisand consumers,
where producersdetcrrninequantitie:.pricesand advertisingto maxiinizc prolits and
collsl;mersdeterrnincthe only rcal quality of goods. Once they do this, their fixed
tastespermii them to maximìzcutility accordingto the standardapproach.
Turning now to pcrsuasiveadvcrtising,the best known approachis Lhe one
describedby KennethJ. Gaibraith(1958, 1970),who repeatedlyunderlinesthe great
impact of producers'persuasiorron consumefs'tastes.His position is founded on a
vast part of psychologicaland sociologicalliterature which points out. the risks of
mass rrrediainlluence on individuals. But if economistsaccept the actual growing
incidenceofthc attemptedpersuasionthroughadverlisingas self-evident,they haveto
treai it directly :ìsa determinantof the tastepararìeters.
A proposal to adopt some intermediatemodels between these twg extrenre
pcsiticns iras bccn made by Boyer, Kihlstrom and Laffont (1979) and Luinj (1994).
These rnocielstry to ciistinguishformaliy bctween informative anci persuasi'/e(or
frawlulcnt\ advertising, that is the advcrtising leaving unchangedor rnaking to
decreasethe customer'sintbrmation,by comparingthe differencesbetweenthe quality
v 'l'his classification hrs a clrssical antecedcnt in lvlarshall's (1919) distinction between
"constructivc" and "corrbrtirc'' advertising,that is however refened to thc pioducers'side.
10 1;o1un attcnlpt,see Milgrorn and Roberts (19E6).
rl "Tliere is another possible function of advertising v,'hichI have not analyzed becauseI do
not know how: advertising'simpact on a consurncr'sutility function, holding information constant.The
charrge-in-tasteidea cannot be cffectiveiy tested bccauseno real theory about raste changeshas been
developed. Fonunately, one does need such an idea to explain the major fèatwes of advertising
b e h a v i n r "( N e l s o n l 9 ' 1 1 ,p . i 5 2 ) .
l5
assessments
of customcrsthat haveor havenot receivedthc advertisingmessage.The
fundamcntalhypothesisis that the consuÍrerchangeshis probabilitydistribuúonabout
the quality of a good by cmploying advertisingaccordingto Bayes' rule. In this
framework, to receive fraudulent advertisingc,orresponds
to adopting an ìncorrect
corelation betweenthe informationprovidedby advertisingand rhereal quality of tire
goods.Therefore,even non-informaiiveadvertisingwill haveeffect on the consumer's
choices.The modei'sresultsarethal evenfrar.rdulent
advertisingincreasesdemandin a
rneasuredependingon the riifferencc betrveenthe lcvels of quaiity of the goods
presentsin the markct and on the reliabiliryof the differentsourcesof information.
The secontlclassof rnodclswith interdependent
prefercricesis foundedon thc
expiicit acknowlecigemeni
of the social nature of the prcferenceformation process.
This approachis characterized
by the presenceof a set of identicalconsunrerswhose
consumption behavjor is influenccd by external effects that are conìmcn to all of
them, bilateraiand symmetricand is discusscdfor thc first time by Iribenstein (1950)
( 1949).
urd Duesenbcrry
Leibenstein(i950) proposesa detailedclassificationof the factorsexplaining
consumerstalic demand.Aside fiom the qualitiesinherentin goods ancithe Veblen
effect, he considcrsthe bandwagonand the snob effects.Eoth efTectsare represented
bll a functional reiation between consumcr demand and market demand with a
positive derivaiive for the bandwagonel'fectand a negativederivativefor the snob
effèct^I-eibenstein'stinal aim tù oirtainaggregatedemandcurves,is achievedthrough
ihe introduction of a fictirious diminishing marginal externai consumptioneflèct,
which undoesthc effcctsof the non-additivìtyimplied by suchexternalities.
Duesenberry(1949) starts from the idea that the consumer'schoices are
connectedo'rer lime tfrlough habits and learning processes.Habit causesthe weil
known Duesenlrerrveflcct, representedby in asymmetrybctween changesof ttre
con.sumptionicvel and of the budgetconstraint.It is only atter having oriercomethis
rigidity thai consumersare able to learn which consumptionpatternis successful.But
the cnsuinglearning processis basicallya social phenomenonbecauseit dependson
the "demonstration et'fect" and, therefore, on the comparison with the other
consumers'consumption
patterns.
This processcanbe described
as a sequence
starting
with an incrcasc in the other consumers'expenditurervhich raises the fiequency
through rvhich the consumer comes in contaci with goods different from those
habitudly consunred.Thi-qccmparisonmay reveai thai nerwgoods are supcrior for
fulfilling the consumer'sneeds.Finally, thc consumcrchangcshis consumption
pattern and probably increaseshis expenditure.Duesenberryempioys the foiloiving
utility function to expressthe interdependence
amongconsumer,
IO
U ,= , r , { * r , r : , , j ) ,
where ,r; and x;e respectivc:lyrepresentthe actual and expecterJdemand of thc
consurnerI and is a weightedaverageof the consumptionexpendituresoi thc other
1
consumer.s,with weights proportional to the frequencyof sccial contacts ainong
ccnsumcri andthe othcrconsumers.
Recentwork whjch fbllows the saine approachformally developsthcse two
Lefelcnces.
Pollak(1976a)lriesto rvorkout I-eibenstein's
insightsin termsof a formal
urodel. Iie assirmestlrat c:onsuurnri's prcÍbrenccsclcpenilon consumer
/'s pasr
consumption
and vice versa,thatisi2
fi,r= ri,r(î.1*1,,-rl r)
with i,7s{t,...,,1,)
The rnodel'slogic clearly points out the similariqr to Pollak'shabit fonnation mg6el
discussedplevior'rsly.Botir rnodelsadopt a systcmof dcmanrlfunctionslocllly linear
in expcndiiLrreand obtain similar results1br short-r'unan,Jlong-i.undemurd. pollaks
original ccntributianis the substitutionof Leibenstein'ssimultaneousinterdcpendence
with laggedinterdependence.
To unCerstand
the weli knr-rwnproblemscausedby this
vafiation, we can refer to garne theory. [n an extcnsive form ganlc, thc i.empora.l
sequencecan be rcpreseniedbv the rules of the gameinducingasyruletry anong the
playcrs anci the destruction of any possible equilibrium as in the case of the
Stackelbergduopoly. On thc conlrary, the bilatelal reactiunstypicai rn the CournotNash duopoly pelrnii us to ignolc conjecturalvariatìons,by simply avoitling to cleai
with a d;'nan.ric
pr6blen-r.
'I'he
importanceof (he relative frequencyof contactsamong consumersis
stressed by Krelle (i973) and Gacrrncr (1974). Iìoth papers rlescribes social
intelaction thraugh a me;rsureof "syrnpathy"betweenthe consumers.while Krelle
does not exactly spccify how this measureis defincd, Gaerfnerassumesthat the
consuner i is rnore influcncedby thc other consumcrshaving a con.sumption
patiem
not "too rnr.rch"dilferentfron'ìhis pattern.Formalli,,
,,,,(,.,,,,rt,.
d,,r?*=,
i) =.ri,t(,
j
with ,t, (.) t O,
rvhe|cd,r'.ris a clistance^pararnetcr
betweenconsumers
r e7 ilhich is positivefor two
consumcrsthat ai'esynÌpatheticand negativein the oppositecase.The inl.uitionbehind
this is that the similarity of the consumptionplans makeslire individual "reassured
and prompterito stick lo his recenr.consumptionpattern",while iarge differencesin
ré Pollak (l97ria] proposcs
e v e n a v a r i a t i o ni n w h i c h t h c i n f l u e n c ed o e s n o t r e g a r < i : , sp a s t
c o n s u r n p r r o nF.o r r r a l i y ,
D , 1( I , D ; 1 - 1 , { )
w i t h i , j e ( t , . . . , I ) a n <jl * i .
T h ì s v a r i a t r o nd c e s t i o t i n r p l ys r g n r r i c a nct h r n g e : .i n É o l l a k ' sr c s u l t ,e x c e p tw i t h v a Ì u e so í I t o o s r n a l l .
A
; i m t l a i r n ' d e l o f i n r c r i l c p c n d c n . ies i : o n r a ì n e d
i n p r a i sa n t l H o u r h a k l : s (r 1 9 . j 5 ,p . l g ) .
t /
p' 332)'
antipathy" (Gaertner
"lead to alienarion and
"914'
to Duesenberry's
consurnptionPattern
píocessthat is similar
includesa cumulative
model
the
More;over.
lmitatcs more consumers
which the consumer
to
accorcling
demonstrarione.ff-ect,
having a lower demand'
thanhinrself'than those
having a higherclemand
t versionof thesanre
Adifferen
":i:::'1,- ," ;:]'#Tt"t:H,il:"",'l,l: ii:
g rrom
venieris(tsi1)' who'startin
i""'T'""TT'll
:
:H:
a
::fi :: ìTil:
a limited rationalitYmodel'
theory' intloduce' in
of other
f f9:fl consumption
L,ancastcr's
Thesegroupsare composed
namelyteferoncegrouFs'
concepr'
tries to
he
psychoiogic'al
"significant others" and
consumeras the
tn;
;;:
ronsurnelsconsidcrccl
e n r u l a t e t h c n r b y " o n . u n , , n , , i n L a n c a s t e r . s t e r r nanci
s , t h e s a m ederive
b u n d laetraditional
ofcharacteristics.
Llakayawa venieris
assumprions,
,,*o;;;";
of
Given a;erìes
"",:'i
m a i : r o f l n d i t l e r e n c e ' o n " n o " o t " ' u ' t o ' " i * t t ;of
r r Granovstter
o p t i m a l s oand
l r r t Soong
i r : n ' T h(1986)'
esecond
threshll-::t"
the
by
of
rePrcsented
proportion
the
suggestìonis
to
pat'rernaccording
consumption
his
chrnges
wlicre any consumer
s o i r r c r e l e v a n t g r o u p o , o . " n * , c o n s t t m e r s t i i a trvhether
a r e c - , hto
a r rpurchase
g i n g t h eone
i r p aunit
t t e rof
n i nat h e s a m e
sr:pposerlto clecirle
'nO""O
U
r,vay.liach cc)llsumer
givencomtnodrtyaccorclìngtotv"odiflerentlevels'alowc'randahigherthreshold'
a s s u l n e d b y t h i s p r o p o r t i o n ' O n c e t h ' J l o r v a rctÍect
t f u e sccases
h o k i i sand
s u r pis'substitute'd
a s s e d ' t h c cby
o n as u n x : r i s
bandwagon
The
commodity'
inclucecito buy thc
thresholdhigher
snobbehaviorrvhenthehigherthresholdisrcaclreoandsurpa^ssed'Theotlrerfelevant
pricesp makesthe lower
an increaseof marker
thar
is
assunrption
aincithehiglrerthrcsholdlorvcr,whileadecreaseofphastheoppositeeffectof
dccreasingthelorvcrthreshol<iandincreasingthehigirer'*::t:i:iftheproportion
ofconsunterswlrohaveourchasedacertainconmodityattimerisp,andthc
c o n s u m e r s c u n r u l a t i v e d i s t r i b u t i o r r f u n c t i o n s o f l o w etrv
r a nthe
d urolrowing
p p e r t h r path'
esholdare
o,,,0o,,,*J,J:,]:::'.iiil:;ilHir::;í*""u
rcspectivery
r-/Îr
to
monotone wtrlb resPect P r'
and F'' not necessal'iìy
F1
betwc:cn
difference
a
Bcing
tltemoclelmiryproduceamultiplicityofequilibriaandthepresenceofchaoticand
3
unstablemarketoutcomss'l
isu'"
modeì
;-:JJ::::il.:t therast
n': 1:u::':::::''"':ilil:::T
o"''op
ffi ;:
*,,"n",i,*lnili:il'*"*
";i"';^'*;;::'::":'::::;":ffiffi''^Jll:*:
"o1,- -t:':,::::,::",:"::::
;;;;H:::T}'::j:":il"ffi
ffil:::'"".*;;;':-:::::::i::HH:?"if:ff'"::
;.Jf
canbein'[erpre'ied
il:"J'ffi
ormoders
in" crass
;'ffi*""
.i"
JJ
::::J:::::::::::
-
can
q'\^ro'c model showing that chaotic resul$
^-r
**
average
average
moving
weighted moving
J""*"0
Cranovetter
extcnds
(I e8e) extcnds
""i';;;;;;ntìally
f,
l3 ,on
n"""on"(1989ì
lonn"""on"
!'iY":Y,:*"t::1í:frH;ili,
are given tl
if ton'un'""'expectations
be oartially avoidcd
of ihe Pastconsun)Ptron
l8
as aIì outgrow{h of information econornics.The starting point is the idea rhat the
processof beìief formationltasa vety limited role in determiningthe economicagents'
trehaviorbecausethcìr decisionsare basicaily the result of a very compiex market
interaction.A way to dcfine the rationalilyof the economicagentsis then to describe
it in terms of mirnetic or imitative pattcrnsof behavicr,as arguedby orléan (1999).
Ilis rnodel of generalizedrmitation,that is foundedon the scciologicaiapproachto
opinion transmission within grcups, seems particuiarly appropriate to explain
speculativebehavior in financial markets,but in order to apply it tc consumption
theory,it is usefuito introducean asymmel-ry
betweenLheconsumerin ordertc model
uniiatera-l
reiations.
Some authors adopt an ordered array of corìsumerslhat ;rllows them to
establisha sequentialcriteriaof decisionso thateachconsurneris only concerneel
wittr
the ccnsumption of the consumer placed above hirn. C)therauthors assumethe
prc-sence
of an r,rpperclassor a distìnguishedagentthat lower classesor anonymous
agentstry to emulate.Poliak (1976a)discussesboth assumptionspointing out their
implications for aggregatedemand. Fonnally, he adopts the follorving demand
funciions,
/
.)
- f ; r = . 1. ;, ', [ - l ; , _ , ,
Jr'
'.
with
i<j
itt an ordering 1,2,..., t\i
and
^t.l -
,,,,{2.,,,,-,p)n , i t h i € ( 1 , . . .L, )
l o w e rc l a s s
i e ( 1 , . " . , U) u p p e rc l a s s
The fii'st func.tionis basedon a hiera'chybrrt it impiies resulfsthat ;ue siinilar to the
rnodel rvith sytnmetricagentsand habìt fbrmationprocesses.The only ciifferenceis
the relevanceof the distributionof piist cori$unptiouthilt influencesthe individual
cc;nsumption
pattern.ii'he sameis rrotlrue fbr per capitademandfunclions.In the twoc.iassnlodcÌ:;.the disiributiitn of expenditurebelweenclassesis thc mo-ctsignific;rnt
but i.hcdcir;auclof the lower class depcndson the averagep&stconsumptionoi the
upper class. If thc behavior of the upper class is determinedby a liabit fomration
proccss.we can easilyseethat resultsaresin:iiarlo thc hierarchymodel.
P':llitk's first modc.lcan be ncwly interpreiedemployinga new conceptdrawn
lionl iniormrtlion cconomics. According to BikhchiLndani,Hilshleifer and Welch
(1992), imitatìr'c prc)cesses
c-'anbc'represente<i
througha sequenceof actionscalled
informaiionalcascade,wherc at sornestageit is optin:al fciran indiviclualto.ignorehis
olvn private iniormartionanclto takc dccisionsr:;nlvori the basisof the pasl decisions
c'f ail thcse individiraisplacedbeforehiin il the sequeuce.
The cruciaicon4itionis
ihat errchindividual observesa conditionailyindependentsignalconcemingti-revaiue
tc ii-rÍ'cr.
Lrrthe corrsunìDtion
patternto imitate.In this context,if a .sutficienily
long
1A
seeuenceof individualsmakesthe samsdecision,the other individualsconforni to it
regardlessofthe signalreceivedandbegin an informationalcascade.
The trvo-classmodel hasbeendevelopcdby Karni and Schmeidler(1990) in a
'Ihey
introciucea utility function implying that the
game theoreticalframework.
pattemincrease
of thc agentso1'rheuppcrclassfor a cenainconsurnption
preferences
(decreasO)
with the number of agentsof upper (lower) ciass that adopt the pattern,
whiie thc prcfercncesof the agentsof the lower class increasewith the number of
that these
agentsof both classcsthat adopt the samepattern.It is ilossibieto show
with
assumprionscan imply a cyciical equilibritrm play as those usually associated
but also othcr patternsof ciemandvaiiation.f)espitethis indeterminacy,Kami
fashior.r
modsl allows us to confirm both PolÌak'sidea that in the rnodels
and Schmeidlel.'s
and
with two classesthc distribution of expendituresis relevant and Granovetter
makes
Soong'ssuggestionthat îhe introductionof externalitiesin consumptiontheory
equilibriumindefinite'
any conclusionabor-rt
But thereareat lc4srother two problemsin theseasymuretricmodels.The first
a
ìs to detìne who influenceswhon anrj to make the social order dynamic through
a
choice
of
the
tr.ansitionrule fr.om a class or a posltion to another. As regard
the ensuing
liierarchy and the linúts of classes.whrtever criteria may be chosen,
dependon
analysiswould be, to a large extsnt. arbitrarybecausethe resultsgreatly
by
soc:iologìcaland psychologicalargurnents.The second problem is discussed
agent- has
Scitovsky(1986)vrhopoints out how rhe upperciass or tire Cistinguished
argues
no superiorexanrirleto emulatc.To prcviCea solutionto this flaw,l4 Scitovsky
have thc
the lbliowing point: "To securestatus thercfore,the rlch must not only
others
that
nloney to outspcnriothers,thcy must also managetc spendit well: in ways
that
rcgardas superiorand clesirable,anciset an exampleworth foilowing. To resolve
problem requirestirat cruciai ingredientof many, perhapsmost enjoyableactivities:
raisesthe
the right clegrecof novelty" (scitovsky i986, p. 201). But this suggeslion
(i976a)
Poilak
questionas io hovrthe degreeof novelty can be meaningfulÌ1'det'ined.
l.he distinguishedagent
adopts an oppositc view assumingthat the upper ciass or
answerhas to
follows a habit formaticn plocess.Howevcr, it seemsihet any definite
generai' This
rcly on some principle or theory drawn fronl thc sociai ;ciences in
'.hewhoie literaturediscussedhere'It is clear indeed
irnpressionhasto be extendedto
the many problems- especìallythe presenceof chaoticand weak
that,notwithstancling
thesemodels
equilibria - causedby the appcaranceclf asymmetryamongconsumers'
14 A soluúon may bc given by Schurnpetcr's'ersion cl the consumer innovator niming the
not only for himseif: "But he also
enrrepreneurinnovator und u, ihis lrsi fin,ling a better solution
which they can copy- They can and
fcr
them
a
rnotlel
oreated
and
trail
thc
triurnphed for others, blaz-ecl
(schurnpeter
1934, p. 133 quoted by Jonsson
will follou,hiÙr, lirst ind!vrdualsan<ithen rvhoie orowds"
1994).
20
show that ìn orcrerto gain insigrrt
into moderswith interriependent
preferences,it is
nece'ssillJto 'introduccin ecottonúcs
stx'e explicit assirr.rptions
aboutsocraìstructure.
V. COÌVCLUDING RETÍARKS
Evcn if it is not lrre intentionof this
paperto co'er a]l theo.es
of c.nsumer
behavior incruding endogencustastes
but rather to suney a part
of theoretic;il
ccononúcswhere thi.sirJcahas
been dcveJopedand discussed,a strik.ing
impre.ssion
rccervedfrom this stud;vìs that there
is not a srructriredapproachto the
casebut onJy
aìternativeappr.achrls.not gathered
in an unified systcmyet. In the
crosingsentence
of the lbregoingsection,this fragnientariness
seemsattributableto the lack
of a more
lntenseand rigorousinterdisciplinary
approachto the problern.
A suppo't io this idea is containedin
the work of sonreauthors(for
exanrple,
sclrcrvsky1976 andyaari r977) that
cxpricitrvconsi<iertastesas ,,prefsrence
systems,,
which the consumer'.s
behaviorcan be refei-redto in order
to appraiseits motivations
and "cnds" (Erzioni l9gg, p. 13g).
Rarhera lot of u,ork has been done
recently,
startingwith Frankfurt(1971)and
Sen (1g74),onrecastingrhis meaning
of tastesin
*ays ihat enabreus to use the conce't
of ,.metapreferc.nces,,,
which reriesheavily on
tnoriri arguncnts. For exampÌe,
sen (ig77, p. 337) defines rnetaprefbrences
as
"rankings of preference
rankingsio expressour m'rar jLrrJgemcnts,,,
crearlyshowing
thiit his purpose is to make rhc
nrodcl of incrividu:richoice richer
tiran what is
permrttedby thc onl1,economic
discourse.sucir an interpretation
is conlirrnerl by
'tlirschrnanin an articrc'signi'cantly
entitrecr"Agai.r;t pzu-sirnony,,.
In lús worirs,,.the
ecclll.nllc approachpresentsus
rvith foo simprc;nindecr
an acco.rntof even surcli
fuudameltaìeconomicpt.ocesscs
as corrsuniptionanil proclucîion,,(Ijirschman
iggj,
rr' 7). ln order tr: makc econoinicsrnore cornplicatcd,
Flírschrnanhas recourseto
metapreferences'
He claims first that a "tiìsteìs a.lmost
defìne.las a preferenceabout
which you do not argue- de gustirtus
non es!disputandunt.A.tasteabout
which you
irrgrrc,with other-sor yourserf.ceases
ipso facto bei'g a tasie_ it iurns
lntoa varue,,
(llirschrnan i985' p' 9), thcn
he arguesfhat r*ctzrpreièrences
are
thc
usef.uiconccptto
'Jescrìber changc
in values.rn this way, rnetapref.erences
becomethe tooi through
which ecoltornicbehaviorturn.s
to cepcnclo' inoial and extra-economic
variablesin
t'trnrrr-tl
i-icwevei, in rnany r:rorlelsdisr:u.ssed
in this oapcr sur:h an intcrdiscipiinary
lippi.:-tchm'iy' be be obtaincd
withr_rtrt
ernproyingrnora! argunre'ts.concerrirng
the
sell'-centerr:d
represenrilriori,
it js nr:t satisiactor-v_
fbr exampleto cleliberarely
ignore
a
rii')rc parhoiogicalvi*v
of the acdict bcrhavrcr.accoiding
to which addiciion
2i
behavior.A pcssibleway to
inroivcs pelsonalconflict anclinconsistent
necessarily
iackle theseissuesis to employ the iìction of morc sclvesand then to apply a strategic
conceptionof ratir:nality.Generally,a more realistic view of economic addiction
seen1sto require the knowlecigeof a set of psychologicalanclsocioìogica.lconcepts
'I'hese
tools become even more necessÍuyif one
largeiy ignorcii in economics.
throughrvhichan economicagentdetermineshis levelsof
considersthat the processes
aspiration,as thoserccentlyconsideredby thc limited rationalitymodeis,are usllally
proccl:scs.
asymmetrichabitl'orntation
methodslbr correctingthc strong assumptionsthat
Devising interdisciplinar-y
uuderlieconsumerthecry scemsappropnateto the modeisassuminginterdependencc
of
arncngconsumersas well. For example,a topic to dcvelopfurther is the updating
prir:edependentprefelencesmodel in the senseof the Veblen effect. Recentresearch
it hasbeenargued(Alcaly
indeeridealswith priccsas quality signajs.on the contra-ry,
of snob appeal
and Klevorick 1970) that a clear tjistinctionamong tire phenornent-:n
For this purpose,Loibenstein'spapermay be
andjutlging qualiryby price is necessary.
other
useful again:"The conspicuousprice lthat is PoÌlak'snormal price] is tire price
peoplethink thc consumerpaid for'.hc cornmodityand which thereforedelerminesits
of
conspicuousconsufrtptlonulility" (l-eibenstcini9-50,p. 203). The close connection
more
tliis ctefinition with gamc theoreiicai reasoning requile us to employ
the
concern
rnay
psychological than purely econonric tQols. A la^rt example
ellcct oi advertising.In this regud' Akerlol:urd Dickens
explanationof tireper"suasive
(1982) havc recently proposedto recur to the theory of cognitive dissonance,a
'l'he
corollary is again to adopt into
concept widely employecl in psychology.
and
econonricrnodcJsan approíìchmore operì to Îhs contribution of other social
humansciencgs.
A final remark is about the elementsof consunlertheory with endogenous
between
tastesomitteclin this sun'e.V.The most importantis probabiythe interaction
'fhe
incrcasingavailability of individual level data
theorcticaland ernpirical work,
tastes'
allorvsreal improverrtentsin tho testingof consumeftheory with endogenous
by
But actually tiìe nost interesting area of current research is represented
making has
cxperìmentaleconomics.The laboratorysimulationof iDdividualdecision
important
rccentlyorigìnateclan impressivcquantityof rvork.l-sthat may representan
of consumer'S
source of new ideas :rud conccpts to inclease oul' undelstanding
bchavior,
l 5 F n r n s u r v e v ,s e eC a m e r e r( 1 9 9 5 ) ,
22
R};FERENCI'S
A i n s l i e ' C . I 9 ? 5 - ' ' S p e c i o u sl { e w i u r l :A I J e h a v i o r a l h e o r y o f I m p u l s n e n e s s
a n c lI n r p u l s i v eC o n t r o j , , ,
Pst'chokty,icalBulletin, 82, 463-96.
Akcrlof, C r\' antl \\r. T. I)ickcns. 198?. "The Economic Consequencesof Ccgnirive
l)issonance',,
A.rtrc
ri c<urfic cnanúc Revi ew, i 2. 30'i. 19.
Aicaìv, R. E. anii R. B, Klevorick. 1970. "Jurlging
euaÌity by Frice, snob Appeal, and thc New
Co nsumer Theory ", 7.ei tschr iJi.fùr N lttio tuzlij l:anoni e, 30, j 3-&.
Aìlinglranr' N{. G. and }f. Monshinra, 1973. "Veblen Effect-sand Portfolio Scleuion",
in Theorr-cf
Denand: Real and Monetary, editcd b-vlvi. lîorishima and others,Ciarendon press,
Oxford.242-70.
llecker, c. s. t 97,l. "A Theory of social I nteraction",JtturnaLaf palùical Ecarutnty,gz,
l 063-93.
Beckcr, G. S.' M. Grossrnanand K. Ì\4. Murpliy. 1992. "R:itionai A<tdictionand the Eff'ect price
of
on
Consurnption",in Choice over'[inre. cdited by C. Loewenjrein and J. Eister, Russell
S a g cF o u n i i a t i o nN
, ew york, 3ól-?tt.
Becker' Cì-S.' Nl. Crossrnanand K. N{. lv{urphy. i994. "An Errrpiricrl .A.na]ysis
of CigaretteAddiction",
A n r c r i c u t tE c t t n o m i cR e .t"i : w , 8 4 , 3 9 $ 4 i l .
Ilecker' C
S and K. J\'1.Murphy. i988. "A Thcory of RaLionai Ac{tliction", Jeurtal oJ palitinl
Econr.tni.\'.9ó, 675-700.
t s i k h c h a n d a n i , S . . DI.{ i r s h l e i f e r r n i J i . W e l c h . I g q z . " A T h c o r y o f F a r l s , F a s h i o n , C u s t o m . a n d C u i t u r a l
Cirangeas in1ìrrmationaiCascadcs",Journal oíPoiitic\l
Et:onnn1,,lA0,9g?-lClzÒ.
Blackorby, C., D. Nisscn, D. Prinroni anrJR. Iì. Iìussell. l973. "Consislent Intertemporal
Decision
Iviaking".RevictvoJ Econr.inúcStucjie.r,4{1,
2j9-48.
IJiundell, R. l988. "consurner Bchaviour; Thcorv and Il.rnpiricalEvidence - A survey",
Economic
Jcur;ni,98, ló-ó-5.
ÌÌoycr, lv{. 1983. "Rational Dcmand anci ExpenciituresPalreins under FIabir Formation',,
Journa! t:f
L t : o t k , n t u I. - l t c u t 13. .l , 1 . 7- 5 3 ,
Iloyer, lVI.,R. Kihlstrom and J. J. Laffont. i979. "Le calcul économiquc clc la publicité
frauduleuse".
Actualìté écottt.tnique.l?, 45-6-/.
I i r o w n i n g , M . 1 9 9 1 . ' " A S i r n p J eN c n a d d i t i ; e P n : t r i e n c eS t i l c t u l e f o r M o d e l s o f
H o u s e h o l t Ìl l c h a v i o r
ovcr'Ii nre", i o ur nal oJ f )oI i.ti ca! Ec ctwsm1,,99, 6Aj 4-Ì .
Cantcrer,C. 1995-"individuaì Decisioa Making", iri')'heÍíuncibookaf Experinentcl Etonrtr,ics,
tdíted
b y J . H , K a g c l a n t l A . E . R o l h , p r r n c e t o nU n i v e r s . i i yp r e s s ,p r r n c e t o ,n5 g 7 _ 7 2 1.
C h r r l o u p k a .F . l 9 9 i
" R e t i o n a l r \ c i i j i r : t i v eB e i l a v i r - r a n r l C i g a r e r t eS m o k i n g " , J o u r n a t o f p o l i r l c a !
iiconon\', 99, 722.42.
{-r,rsggi,p1. Ìvl. !99{. "Audiencc Effecrs in consumption", Economicsand philo-.ophy,l0.
lg-30.
i)
==-
Ccwen, T. l9ij9. "Are Aii
'l;stes
Conr,'.aritanctrldentic:ii", .laurncl of I)conomir Seheviar anC
í ) r g o r i i l a l i t ; t t ,I i , 1 2 7 - 3 5 .
und tli.e Theol, of Consu:ncr B(htvior, H:lrvaril Econc,ntic
Duesenbery, J. S. i9.19. lttutne, Sat,ittg,,
Studl,Clambridge"
El-Sairy. A. Ir. 1976, "Adaptrve BÈhavior an.J the ExlsLenceo1'Weiz.sackersLong-Run Indifference
'f
Curves", J ournal t$ Ecottontlc ítt:ory, I 3, 3 l9-28.
El SrttÌy, .{. E. i9?6a, "Adaplivc Behavror, Denrand and Prcfcrcnces",Journal of Economic Theory,
r J ,2 9 C - 3
I8.
atid.îlte 'lirens,CambridgeUniversity Prcss,Cambridge.
Elst,:r,,Ì. 1919. rJi;'s.:es
Etzioni, A. 1988. ?}r: MctraLDir,tensior,:Toward c Nctt Ecotatmics,]hc lìrec Press,Nerv York.
Fisher, I;. l\,1.and K. Shell. 1968. "'l'aste and Quality Clrangein the Pure'lheory of the True Cost-ofLrving Indcx", i,t Value, Cctpin! anJ Growth. Papers in ITonor of Sir Johtt Hicks,
edited LryJ. N. Woifc, ErJinburghUniversity Press,Edinburgh,220-35.
iìrankfurt, H. C. 1971, "Frecdom of thcrWill and the Concept crf a Person",Journal of Philosophy,6S,
5-24.
F r i e d m a n ,M . l q ó 2 . P r i c e ' ! ' h e o r ; ,A l d i n e ,C l r i c a g o .
Caertner. W. 1974. "A Dynamic lvfodci of lnterdependent Consumer Behavior", Zeiîschrtft Jit
34,321 -44'
NaîionaLókonotttie,
Galbraith.J. K. 1958. lhe AftIueri: S,rci,rry,CambrtdrlcUniversitl Picss,Carribridge
( i a l b r a i t h ,J . K . l 9 7 0 . " E c c n o m i c sr i - sa S y s t e mo f i l e l i e f ' , A m e r i u t t t E c a n o t n i cR e l i e v ' , 6 0 , 4 Ó 9 - 7 8 .
Cilbcra, t. and D. Schincrdlet. 1993. Case-BcsedCr;tsunter Tfterrry, Working papcr, Northwestem
UnìversilY
Cilb,ra" I. and D. Schmeidlcr. i99.i. RccLuionto Price Changes and. Aspiraîiost L.cve!Adjustments,
Working paper,NorthweslernL'iniversity.
Gintis, H. 1974, "Vr'elfare Cliiteria rvith Endogenous Preferences.The Economics of Education",
I nte rttutittnal EcttrttttrricRevior', 15' 4 I 5-30.
G o r n r a nW
, . M . 1 9 6 7 ." ' 1 ' a , \ 1 ehsa, b i t sa n d c i r o i c e s "I,n l e r n ( 1 î . i o nEacl o n o n t i cR c v i e w , 8 , 2 1 8 - 2 2 .
Granovctîer, M. lnd R. Scong. 1986. "Threshold Mcdeis of lntcrpersonal Eifects in Consumcr
Demancl",Journal of Econonic Belnvíor and Organigtiotr' 7, 83-99.
Ll:rmmond,P. J. 197ó."Chauging Tastesarid CoherentDynarriìcChoice", Reiew of Econamic Studies,
41, 159-7f .
Hanrmond, P. J. 197óa. "Endogcncus Tastes and Stable Long-Run Choice", Jttuntal of Econornic
Tlrory, 13' 329-40.
c h o i c c " ,E c o n o n ú c a , 4 4 , 3 3 1 - - 5 0 .
F l a m r n o n dP, . I . l g 7 7 . " D y n a n i l c R e s t r i c t r o uosr rl v l c t a s t a t iC
lìai,akarva,H. end Y. Vcnicris. i9?7. "Consumer Intcrdependencevia ReferenceCroups", Journal of
Po! it ical Ecrttttt tttt, 8.-r,599 ÓI 5.
Ilcrrnsrein. R. J.and D. Prelec. 1992. ",{ Tireorl of Addiction", ín Chttice over Time, edited by C.
i-oewensteinrrnijJ. Illster, Russcll SagcFountlation,New York, 331-60.
. A
Ilirschnian' A" O. 1985. "Agatnst Parsimony.Three Easy Ways of Complicating Some
Catcgorics of
Econornic Discoursc",Economicsand pkilosophl,, l, j -Zl.
Houtlrak-ker,H. S. antl L. D. Taylor. 1970, Consumer Denand. in the lJnitetl Staîes. Anal),ses
and
P roj ectkt ns,I{arvard University,Press,Cambridge.
I a n n a c c o n eL. . R " 1 9 8 ó ." A d d i c t i o n a n d S a t i a t i o n " E
. c o n ú m i cL e î t e r s , 2 l , 9 5 - 9 .
lannaccone,L. R' 19 89. "ll andwagons iind the 'l hreat of Chaos", Journal of E<:onornicBe
havior tin1.
O r g n n i z . c t i o rIt .l , 4 3 1 ' 4 2 .
Johnsen' T. fl. and J. B, Donaldson. 1985. "The Structure cìf Intertemporal Preferences
Under
Unceruiinty antl'['irne Consistent plans", !] cono nret r i ca, 53, 145 l -9.
.lonsson' P. O' 199'l. "Social Influences and Individual Prefì:rences:On Schumpeter'slteoiy
of
CorrsumerChoice" . Ilei'ie++'
of Social Economy,52, 301- I 4.
Kalman, P. J. 1968. "Theory of Consurner Behrlvior When Priccs Enter tie Utilitv Functicn".
Econonrctrica, 36, 497-S 10.
Karni, Il. and D. Schmeidler. 199{)."Fixerj Prelèrencesand ClTangingfastcs", Atnerican Econpnic
Rey'it:ty,80,2{t2-l.
Klein, L. and FI. Rubin. I947. "^ C,:nstantUtility lndex of the Cost-of-Living", Relieu, of Ecotrornic
S t u d i e s , 1 5 ,E 4 - ' 7
Krclle. W. 1973. "Dynamics of the Utìlity Function" in Cart Ìt[enger antl rhe Austrian School af
Econonúcs,edited by J. lì. flicks and w. weber, clarenelonpress,orforcJ, 92-12g.
l-enc:rsrcr,K. J. l97l . ctsnsruner!)enund; A Nex,Apirrrldclr,coiurnbia univer.sity,Nelv yorkLeihenstein,H 1950. "Bandwagc.rn,
Snob, and Veblen Effects in the Theory of Corrsumers'Demanc!",
()uartt:,-lyJautnul of Economics,ó4, l83-207.
LIuch. C. l974. "Expenditure, Savingsand fìabit Formation", lnrerilotional Economic Rer,íew.
15.:'869'7.
Locwenstcin, C. 1992. "'fhc Fall and Rise of Psychoiogical Expìanations in thc Econornics gf
IntenctrpcrralChoice", in Choicc over 7ime, editcd by C. L,oewensteinaÌd J. Eister,
R u s s e l lS a g eF o u n c i a t i o nN, c w Y o r k , 3 - 3 4 .
Loervenslein, G^ and D. I'reiec. 1992. "Anonalier
in Inteltemporal Choice: Evicience an<j an
Inierprctation",tn Ohoice ovar T'irnt,edited bi, C. Loervenstcinand J. Elster. Russeil
S a g el b u n d a r i o r r N
, ew York, i tg-4_5.
I-.riini, i.. l994. "sLrl mercato ttej beni e sul n)ercato delle idee: nol:i in margine ad un mercato
ìntermcdio", in {'rospettive depli stutli di cconomia e politica industriulein Itulia, a
c u r a d r ( ì . B e c a t t i n ic S . V a c c ù ,F r a n c oA n g c l i , M i l a n o , i ( 1 5 _ 7 4 .
N{arschak,T. A. 1978, "Cn the Study of Ta.steChanging Policies", American Econamic Review.
tg.
386-9
r.
Marshall. A- 1919. lttciustrvanC Trade. 4 Studl'uJ inCustria! fechnrque and BusinessOrganipliott;
and of their lr(litence.; on the Candirìonsof Vurious Classcstnd Naîionr,lr{cmi}lan
& Cc.. London.
25
Mi:Cai!!l;1.t,{D'19;ó...i)illhj:îabilit]/cIdynimicí]ernandSySiern.,.'!níerntl|iontzlEccnom.tc
/ ? r : r ' 1 a '1r .' 5 ,2 5 6 - 9 '
'sf
"Price and Arlverlisirrg signals of Prcduct Quality"' Jaurrutl
Mii;ironr, P- antl I. Roberrs, i98ó.
Í' rt liti cnl EccnonzY'94' 79#821'
Journi':iof Political Ecor'orns"82"129'i4'
l"leiscnP
, . i 9 7 4 . "ACvertising as inlornìatloil",
'18' 213-4 I '
uencesof Advertising"' T'heJournttl of Busincss'
N c l s o n ,P . i 9 7 5 . "Tirc EconoruicCr-rnseq
''il'lrrncticCoîìîxgionanti Specuia"iveBubbles"' Tlrccr1:antl Dccisian'2'""63-97'
.}iéar, A 1989.
''vhenTastÈsAre
yaeri. l!t?1. "oÍ the Exi-.tenceof a consistcnt course of Action
P e l e g ,B . a n C N {.
39i"401'
C h a n g i n g " ,R a v i e vt'í. ' E c o t t o m i cS t u d i e s4' 0 '
in Er'ic1'sin nnthenwtical ecrsnomicsin lwnor af
Peston. ìvl. ll. 1967. "CìhsnginguÙlity firnctions"'
í)sk|lrhloisenst.,rn,editedbyM.shuhik.PrirlcetonUniversityPress,Prrnceton'?"33.
6.
Phelps,É'S-arrdll.A.Pollak.19ó.8...orlSecond-BestNationalSavingandGame.Equilibríum
102' 185-99'
Growth", Revieu"'c( Economic Stuílie's'
Phlips,L'1972'.,ADynr.mtcVersiorroftheI-inearE,xpencilure}"{odel','ReviewrfEtonomicsand
Statis tics,54, 45t)-8.
Anall'sis'North Holland' Antsterdarn'
Phlip.s.1- 19i4. Apnlier)Consttntptioa
Re"'iewoi llconanricSîudie't'35' 201-8'
Follak, R. A. l96iì. "Corrsiste!ìtPlar'nin.q,"'
Journa! of Poiitica! Economy'
poilak. R. A. 1970.,.Habit Formation anti Dynamic l)emand Funetions",
?8, 7'1'5-63
Eci'ncnicTheorv'
l
g
l
(
) . " n o , , 1 ,1 * 1 r ; a a t i o n a n d L o n g - I ì u n l ' l t i l i t y F r r n t t i c n s " ' J o u r n a l o f
l,ollak.R. A.
r1
r?)-U7
I,r:lllk,R.A.l976a...Interdependentl,re|ererrces,'.AmericanEconontic!ìeview,66,309-Za.
IbÌIrk,Il.A.19?7',.l,ricelJepencientPíefefcrìccs'',AnericatiEconanicReview,6l,64-/15'
Pcllak,R.A.l9?8...[ìnd,.igenous].astesinDeuranclrndwelfaloAnalysis''.AmerkanEconomic
/levlclt, 66' 374-80'
Prais'S.J'andli.S.Houthakker.ig55,TheAwlysisofFcrirtityBudgets,CarnbridgeUniversityPress'
Carnbridga'
..Consumcrs.SovefeigntyRevisited and the Hcspitability
Rothenberg,j, 1962.
of Freedom of Choice'',
;\nerican Econotnic Rei ietp, 52,269-90.
Wiley and Sons' Nerv Yoi:k'
Savage.L. J 1954.'l'irc!'o"sulctionsof Sratistics'john
LlniversrtyPress'Cambiidge'
arttl Consequence'Harvarrl
Schclling, T. C. 1984. L'h-otce
Sclrrnaicnsce,k.|g.72.T|te!:cotianicsofAcltlertisiltg,NorlhHolIand.Anrstet.dam'
Sclrurrpetcr'J.A.ii934.7-heTltetlryt.l.fEcononticDevelopment,HaivudUniversityPress,Cambridge.
Scitcvsky'T"1911'45...iìomcCorrsequencescîthcHabrtofJudgingQualitybyPrice,',Reviewof
EcononticSrudigs' 12' 100--5'
Scìrovsky.T.i976.7.Í|eJo|leSSEcotrcntT,.AntnquiryinroI'lumanSatisfatliottandConsumer
t io'l, Oxford University Press'Oxford'
D i ss alt sJoc
26
Scitovsky, T. 198(r.i!unuxt f)esire ortd Ecor.omicSutlsfactiott,WheatsheafBooks, Briglrton.
S e n ,A . K . 1 9 7 J . ' l l e h a v i o u ra n d t h e C o n c e p to f P r e t - e r e n c eÌ "Ì c, o n a m i c a , 4 0 , 2 4 1 - 5 9 .
Sen, A. K. 1974. "Clroicc, Orderìngs and ìvforality", ín f'rat:ticcl Reuson, edited by S. Kórner,
Il I acku'ell, Oxlìlrd,,54-ó7.
Sen, A. li. 1977. "lìational Fcols: A Critiquc of the Behavioral Iìoundations of Economic Tlreory".
Phi I osop hl ut C l' ul: ! ic tlJfrti rs, 6, 3 17- 14.
S i r n o n ,F ì . A . l q 5 ? . M o d e! s o f l ' l o t r ,J o h n\ Y i l e y a n d S o n s ,N e r v Y o r k .
S l u t s k y ,H . 1 9 1 5 ." S u l l a t c c r i ai l c l b i l a n c i oJ c l c c n s u n r a t o r e "G, i o r n a l et l e g l i E c o r u t n r i s t i , 5 l .l - 2 6 .
S p ì n n e w v nF
. . 1 9 8 l . " R a t i o n ; r l l e b i t F o r n r : r t i o n "E. u r o p c a nE t o n o m i cR e r i e r v ,1 - 5 , 9 1 - 1 0 9 .
Stigler, G. J. 1950. "'fhe Devckrpnrcntof {-itility fheory", .lournal of Poiitical Ectnomy, 58,30"i-27,
373-9ó.
Stigler, C, i. anrj C. S. Bcckcr. jtì77. "Dc Custibus Ncn Est f)isputanciurn",Anrcricat Econonic
R c v i t w . 6 7, 7 6 - 9 0 .
S t l r r c , R . l 9 5 . i . " l . i n e a r E x p c n d i t u r cS \ s t c m sa n d l ) e r n a n dA n r l y s i s : A n A p p i i c u t i o nt o t h e P a t t e r no f
llritisli Dernend",EcononrrcJournal, 64, 5Ii -27,
Súotz, lì. H. 195-5-56."Myopia:in<l Irrconsistencvin Dynamic tltility lvfaxirniz-^Liar|",
Revie\e oÍ
23. ló-i-8{i.
Econonic J-iù.1ids,
Vcblc n,
'l'.
1899. The Theon, oJ li:e Leisure CLtss,-l'heVtking Press,Nerl, York-
von Weiz.;àcker,C. C. 197l. "Notes on EndogenousChangeof Tastcs", Jountal af Econr,mic Theory,
ì 11i-71
\ \ ' ! { ì s t o n ,G . C . I 9 8 0 . " l , i i t i i c t i o n a n d B a c k s l i d i n g A
: ' I h e o r y r r i C o m p u l s i v eC o n s u m p t i o n "J, o u n t t t l c ' ! '
I'.r:ono ntìc B elut vi o r nt d O r t aniiitli on, | . 295-324.
Wolrrrsky,r\. 1983."Prjccs as Signalsol ProductQualifv", lìet,iewoi Etotroniir: J-lrrdie.r',
5A,641-58.
-I'aste:
Yaari, M. E. 1977. "IinrlogenousChangesìn
A Philosophic;rlDiscussion",Erkennrnis, i l, 15796.
)1