TheJournalofSocio-Economics41 (2012) 255–265
ContentslistsavailableatSciVerseScienceDirect
The Journal of Socio-Economics
jo u rn a l h om epa g e :w w w . e l s e v i e r . c o m / l o ca t e / s o c e c o
Demand cross elasticity without substitutability: An experiment
夽Luigi Luini
a,∗, Pierluigi Sabbatini
b,1aDipartimentodiEconomiaPolitica,UniversitàdiSiena,PiazzaSanFrancescoDˇıAssisi7,53100Siena,Italy
bAutoritàGarantedellaConcorrenzaedelMercato(ItalianCompetitionAuthority),PiazzaVerdi6/a,00198Roma,Italy
a r t i c l e i n f o
Articlehistory:
Received5October2010 Receivedinrevisedform 15November2011 Accepted13December2011
JELclassification:
C93 D12 K21
Keywords:
Demandcrosselasticity Pricereference Unfairness
Antitrustrelevantmarket
a b s t r a c t
Westudyamarketinwhichgoodsareproducedunderlowmarginalcostswithapoordegreeofsub- stitutabilityamongproducts.Inthis environmentwerananexperimenttoexplainwhypricesare interdependentevenwhenpreferencesareindependent.Wecompareourresultstoprevioustheoretical andlaboratoryexperimentalliteratureonpricefairness.Wefindthatevenintheabsenceofinteraction amongsubjects,pricefairness/unfairnessdoesplayamajorroleinthedecisiontoacceptorrejecta deal.Subjectstendtobemoreresistanttoapriceincreaseandrejectadealwhenthepreferredproduct isnotreferencedtopriceincreasesofnotsubstituteproducts,iftheseproductsareconsideredtobea benchmarkforfairconduct.Thusdemandcrosselasticitycanarisebetweenproductsthatarenotsub- stitutes.Thisresulthasimportantimplicationsforantitrustpolicy.Indelineatingamarketperimeter, fairnessconcernssuggestthatproductsthataresimilarbutnotinterchangeableshouldbeincludedin therelevantantitrustmarket.
© 2011 Elsevier Inc. All rights reserved.
1. Introduction
Thereareproductsorservicesthatconsumersrefertoindecid- ingwhetherornottobuyother,similarbutnotsubstituteones.
Considersoccerfanswhohavetodecidewhethertobuyaseason ticket.Naturallytheywouldneverconsiderswitchingtoanother team,yettheydoactuallytakeintoconsideration“similar”season ticketpricespostedbyotherteamsinthesameleague.2Wethere- foreexpectthepricesofseasonticketstohavereciprocaleffectson theirrespectivedemanddespitethelackofgenuinesubstitutabil- ity:inshort,wecandetectthepresenceofsomedegreeofdemand crosselasticitybetweennon-substitutes.
夽 TheviewsexpressedarethoseoftheauthorsandnotoftheItalianCompetition Authority.
∗ Correspondingauthor.Tel.:+390577232608;fax:+390577232661.
E-mailaddresses:luini@unisi.it(L.Luini),pls@agcm.it(P.Sabbatini).
1 Tel.:+390685821369;fax:+390685452369.
2 AninquirybytheItalianCompetitionAuthorityfoundthatfootballclubsmon- itorthepriceschargedbyclubsbasedinothercities,whichwouldappeartobe atoddswiththelackofsubstitutability.Theinfluenceofforeignpricesoninternal competitiveconditions(wheninternalandforeignmarketsareclearlyseparated) isanothergoodexample,alsotakenfromtheactivitiesofthesameAuthority.An inquiryintothepricingofformulababymilkbytheItalianCompetitionAuthority showedthatpricesabroadweremuchlowerthaninItaly.Consumersdonotusually goabroadtobuythistypeofproduct,butthefindingtriggeredapowerfulreaction amongconsumers,whichmadepossibletheentryofanewfirmintothemarket, withasubstantialimpactondomesticprices(Sabbatini,2008,pp.497–499).
This phenomenon cannot be plausibly explained in tradi- tional economics textbooks, accordingto which cross elasticity depends strictly onsubstitutability. Rather, it would appear to be connected withbuyers’ considerationsof fairness: fans will takeseasonticketsfortheirteamiftheyperceivethattheyare priced fairlywithreferencetootherteams’prices. Inthis case, the supporter’s utility function must include this fairness fac- tor.
Thispaperisbasedonthehypothesisthatconsumers’utility dependsnotonlyonthe“acquisition”utilityoftherelevantgood butalsoonthetermsofthetransaction,asinKalman(1968).This approach boilsdowntothechoiceof thepoint ofreference,an essentialingredientofeverytheoryinvolvingfairness.Therefer- encepricedeterminesthecontext towhichconsumers referin choosing,inparticularwhenconsideringthethresholdforaccept- ingorrejectingagiventransaction.
Theintuitionisthatinsuchcircumstances,thelogicalcandi- dateforthisroleisthepriceofsimilar,butnotsubstitute,products.
Fromtheevidence,wecaninferthatconsumersdopayattention tothesepricesinevaluatingtheproducttheyintendtobuy,and thatfirmsbehaveaccordingly.Anaturalcontinuationofthistrain ofthoughtisthatpricesofsimilargoodsthatservereciprocallyas referencepoints,shouldinfluenceeachother.Asaconsequencewe mayobservepricesofsimilargoodsconvergingtowardaunique priceeveniftheproductsarehighlydifferentiatedandthecon- sumersevaluatethemasnon-rivalgoods.Soacertaindegreeof priceuniformitycanbeexpectedevenamonghighlydifferentiated
1053-5357/$–seefrontmatter © 2011 Elsevier Inc. All rights reserved.
doi:10.1016/j.socec.2011.12.011
256 L.Luini,P.Sabbatini/TheJournalofSocio-Economics41 (2012) 255–265 products.Andthisispreciselywhatweobserveinseveralmarket
contexts.
Consider,forexample,themusicCDmarket.Traditionalanalysis predictsthatpricesofCDsbelongingtothesamecategoryshould showhighcross-sectionvariability:withlowmarginalcostsand highoverheadcosts,andthegoodsbeingpoorsubstitutes,prices shouldmainlyreflectthespecificdemandformusicalartistsand shouldconsequentlydivergesubstantially.Instead,wefindacon- vergenceofpricesthroughsomecommonvalues.Theretailprices of recentlylaunched CDs of two artistsfar apart in popularity (sothatonemightwellexpectdifferentdemandelasticities),are almostthesame.ThispatternisnotconfinedtothemusicCDmar- ket.Ticketpricesforfootballmatchesofclubsinthesameleague butdifferentcitiesareverysimilar,independentlyofthenumber oftheirlocalviewers(andsupporters).Moregenerallywefindthe samepricepatternforfilms,CDs,DVDs,standardizedsoftware,live concertsandsportsevents,allmarketscharacterizedbygreatprod- uctdifferentiation,highfixedcostsandnegligiblemarginalcosts.
Mightconsiderationsoffairnessexplainthisevidence?
Thispaperispromptedbytheintuitionthatthepriceofagood (say,thelatestCDbyBeyoncè)canbeusedasareferencingheuris- tictobasefairnessconsiderationfortheacquisitiondecisionofa similarbutnotsubstituteproduct(say,BruceSpringsteen’slatest CD).Sothesetwoproductsshowahighdegreeofpricesimilar- itynotbecausetheyaresubstitutesbutbecausetheirpricesare reciprocallyusedasabenchmarkinassessingfairnessinpricing.
Thepeculiardemandlinksamongnon-interchangeableprod- uctsthatweexploreheremayjustifysomeconsiderationsonfirms’
strategiesandpricing.Inparticular,weexpectthatfirms,awareof theseconnections,mayconsidercooperationtorelaxreciprocal constraintsduetothereferencingroleoftheirprices.Furthermore, effectivemarketpowerneedstobeassessedcorrectly,takinginto accounttheinfluenceofproducersofsimilarproducts.Henceour analysisasksforareconsiderationofthenotionoftheantitrust relevantmarket,commonlyrestrictedtointerchangeableproducts only.Thispapersuggests,instead,broadeningthenotiontoinclude somenon-substitutegoodswhere theyarereferred to bycon- sumersasstandardsoffairness.Thatis,anantitrustmarketcannot restonlyonthenotionofsubstitutabilitybutitshouldalsoinclude productsthatserveasareferencingheuristicforconsumerchoices.
Thepaperisorganizedasfollows.InSection2wereviewthe literatureonreference-priceand priceuniformityinhighly dif- ferentiatedproductmarkets.Section3presentsamodelinwhich consumer’schoicedependsnotonlyonutilitybutalsoonperceived fairness.InSection4wefocusonthepeculiarityoftheCDmusic market,whichdespitegreatproductdifferentiationdisplaysshows aremarkabledegreeofpriceuniformity.InSection5wedescribe anexperimentaimedtobringoutthecausalrelationshipbetween transactionpricefairnessandpriceuniformity.Section6reports theresultsoftheexperiment,andtheconcludingsectiondiscusses thepaper’sfindings,includingthemainimplicationforcompetition policy.
2. Theliteratureonreference-priceandpriceuniformity Fairness intransactions is a popularthemeinthe economic andmarketingliterature.Anumberofpapersshowthatfairness doesplayarole incharacterizingeconomicbehavior. Someref- erencescanclarifyitsnature.Threequestionnaires(Thaler,1985) revealthat theperception of costinfluences theassessment of fairness,whichin turnconditionsconsumerchoices.Household surveys(Kahnemanetal.,1986a,b;FreyandPommerehne,1993) confirmtherelevanceoffairconduct:inparticular,priceincreases duetocostshocksareconsideredfairerthanpriceincreasesdueto demandshocks.
Theseempiricalfindingsthrowdoubtontherobustnessofcon- sumertheoryin which preferencesarestrictly self-centered.In fact,other-regardingpreferencescanalsoplayasignificantrole, andattitudestowardfairnesscanbeafactor.Variousanalytical modelshavebeenofferedtoexplaintheseempiricalobservations.
Kahnemanetal.(1986a) pioneeredwiththe“dualentitlement”
approach,whichhingesontheideathattransactors(consumers, workers)“haveanentitlementtothetermsofthereferencetrans- action,andfirmsareentitledtotheirreferenceprofits”.Onlywhen profits are jeopardized are firms entitled to modify prices (or wages)totransactors’detriment.Itisinterestingtonoticehowthe fairnessofatransactionisdefined:
“Themeasureoftransactionutilitydependsonthepricetheindi- vidualpayscomparedtosomereferenceprice,p*.(...)Themost importantfactorindeterminingp*isfairness.”(Thaler,1985,p.
34).
Thislineofresearchhasattractedagooddealofinterestand hasraisedanumberofissuesrelatedtotheprecisemeaningofthe senseoffairness,3 thefactorsthatinfluenceitanditsimpacton consumptionchoicesandpricingdecisions.Somearticlesreview themainanalyticalcontributionsandshowtheamountofempiri- calandexperimentalresearchonthetopic:somefocusingonprice fairness(Xiaetal.,2004)andsomeonother-regardingpreferences (CooperandKagel,2009).
Thefocusofthisapproachisonhowconsumersconductthe comparisonthattriggersthesenseofprice-unfairness.Threeref- erencingheuristicshave beenproposed:(a)previousprices,(b) seller’scosts,and(c)pricesofcompetinggoods.
Previous prices– Oneobvious waytotracebacka reference priceistolookattheearlierpriceschargedforthesameprod- uct.Experimentalstudies(Brieschetal.,1997)aswellasanalytical contributions(Kahnemanetal.,1986a;Rotemberg,2004),seemto confirmthispointofview.Inparticular,Rotemberg(2004,p.4) referstoconsumers’angryreactionstriggeredbylearning“some- thingthat makesthemwishtheyhadcarriedoutadifferentsetof transactionsatanearliertime”.Sothereferencetopreviousprices isthestartingpointofhisanalysis.Thefactthatpricechangesare noteasilyinterpretedbyconsumersdoesnotaffectthisapproach, whichisdesignedtoexploretherationaleforstickypricing.Con- sumersareassumedtobegenerallyhostiletopriceincreases,as theydonotunderstandtheunderlyingrationale.
Putler(1992)presentsananalyticalmodelofconsumerchoice based on utility functions that also encompass, as arguments, marginallossesandgains.Theselossesandgainsaredetermined bythedifferencebetweenthepriceoftheproductunderconsider- ationanditsreferenceprice,whichisanexpectedpricebasedon pastvalues.Theauthorgiveseconometricevidenceofthiseffect, whichbasicallyshowsthatdemandpriceelasticityishighwhen pricesincreaseandlowwhentheydecrease.
Seller’scosts–Otherapproaches,whileintroducingtheroleof fairnessineconomicbehavior,assumethatconsumersmakesome inferenceonthecauseofpricechanges.Thatisthecaseforthe“dual entitlementtheory”(Kahnemanetal.,1986a),bywhichonlyprice changesthatarenotjustifiedbycostincreasescantriggerconsumer reaction.Thalerclarifiesthispointbyarguingthatthereference- priceisbasedoncosts(Thaler,1985,p.34).SimilarlyOkun(1981) – whois generallycredited for havingoriginatedtheliterature onfairness–notesthatonlypriceincreasesthatareunjustified bycostincreasescanprovokeconsiderationsofunfairness,which impliesthatconsumersusecostsasareferencepointinevaluat- ingprices.Interestingly,theseapproachesofferanexplanationfor
3Smith(2005)reportsalonglistofdifferentnotionsoffairnessemployedinthis branchofeconomicliterature.
L.Luini,P.Sabbatini/TheJournalofSocio-Economics41 (2012) 255–265 257 pricemark-up,apricingpolicythatanticipatesunfairnessreactions
byconsumers.
Accordingtothistheory,consumersknowthesellers’coststruc- ture,anassumptionthatcertainlycannotbetakenforgranted.For example,Boltonetal.(2003)presentexperimentsinwhichcon- sumersareunabletodistinguishthereasonsforpriceincreasesand arepuzzledbyinflation.Consumerslackaccesstocostinformation andcannotextrapolatecorrectconclusions4fromthelimiteddata theydohave(Boltonetal.,2003).Thiswouldappeartobedecisive initself.Butwecanaddafurtherargument:inourmarketoverhead costsprevailovermarginalcosts,andpricestructureshouldbeset- tledbyfirmsaccordingtothestrengthoftheseveralcomponentsof demand.Inthiscasethereisnosatisfactorywayforconsumersto relateunfairnessconcernstothecoststructure.Costs,then,seem tobeanunsuitableprice-reference.5
Pricesof substitutegoods-Anumberofworkshave explored theimpact on marketing strategies of perceptionsof prices of substitutes6inconnectionwithfairness(Xiaetal.,2004;Anderson andSimester,2008; Dholakiaet al.,2005).Whenproductslook alike,itisargued,pricecomparisoniseasyandconsumerscaneas- ilydetectunfairness.Thisthesiscannotbeendorsed:werather thinktheopposite.Thepricesofsubstitutescannotbeconsidered abenchmarkforunfairness,whichinsteadisinevitablyassociated withthelackofalternatives7:ifthesubstitutegoodsarecheaper thantheonebeingconsidered,consumerssimplychoosethelower- pricedalternative,possiblyafteralearningprocess.Soitisonly whenthedegreeofsubstitutabilityislow(orwhenmarketpower isgreat)thatunfairbehavior(andfairnessconsiderations)arises.
Fairnessshouldnotplayanyrolewhenfirmsdonothavemarket power,becauseinthiscaseconsumershavealternativesandfirms cannotexploitunfairbehavior.Itfollowsthatifourbenchmarkis pricesofcompetingproducts,therelevanceofthesenseoffairness isrestrictedmerelytoshort-termdisequilibria.
Inshort,then,alltheseapproachestoreferencepriceproveto beunsatisfactory.Ontheonehand,buyersaregenerallyunableto analyzeeitherthepreviouspricesorthecoststructuresofsellers.
Andontheother,wedonotexpecttheretobecompetingproducts whenfairnessisanissue.
Notice that the literature, although originally focused on prices,tookvariousdirections,andthereferencenotiongradually switchedfrompricestoproductcharacteristics.Thepresentlitera- tureonlossaversiontypicallyconsidersconsumerutilityfunctions conditionedby“frames”or“referencestructures”thatrefertosome standard,orexpected,orpre-existingcharacteristicbundles.8Our contributionisastepbackfromthisapproach.Thatis,werestore
4 Theretailpriceofgasolineisrelatedtothepriceinthewholesaleoilmarket, butthenumberofretailmarketsthatworklikethegasolinemarketappearstobe quitelowindeed.
5 InanempiricalsurveyconductedinSwitzerland(FreyandGygi,1988;Freyand Pommerehne,1993)anincreaseinthepriceofsnow-chainsonsnowydayswas perceivedbythegeneralpopulationashighlyunfair.Thisisalsothemajorcontri- butionbyFranciosietal.(1995);albeitfromadifferentperspective,theydescribe severalexperimentsshowingthatbuyersgetusedtoapriceincrease(thatis,the eventualinitialfeelingofregretvanishes)afterawhile,whetherornotitisjustified byhighercosts.
6 Brieschetal.(1997),Albaetal.(1994),LichtensteinandBearden(1989),and Dholakiaetal.(2005).
7 IntheRotembergmodelthesenseofunfairnesscoexistswithalternatives becauseitarisesoutofacomparisonbetweenapresentandahypotheticalpur- chase,whichcouldhaveoccurredatanearliertime.Butinthiscasethecomparison iswithpreviousprices.
8 Inthesecasespricefixingorpriceraisingbyfirmsisfullyaccepted(perceived asfair)byconsumers.Otherwiseanomalies(Kahnemanetal.,1991)emerge.With relationtothethreeclassicalanomalies:(1)endowmenteffect,(2)lossaversion, and(3)status-quobias,theendowmenteffect–atypicalseller-sideanomaly (BeggsandGraddy,2009)mustbeexcludedfromouranalysis,whichconsidersonly consumer-buyerbehavior.Underthesegeneralconditionsofprice-makingfirms andprice-takingconsumers,alltransactionsturnintoexchanges.Whilethebulk
thefocustopricecomparison(withareferenceproduct)forthe purposeofjudgingthefairnessoftheproposedtrade.
Theonlyarticle,toourknowledge,thatdealsexplicitlywith thereferencingroleofnon-substituteproductsisAndersonand Simester(2008),whichexaminesthepricesofclothesofdifferent sizes.Theirevidencesuggeststhatthedemandforclothesisinflu- encedbythepricesofclothesofdifferentsizes,inducingahigh degreeofpriceuniformity.Theauthorsfindthat thispatternis duetothesenseofunfairness;theindirecteffectoftheperception ofunfairpricingismuchstrongerinreducingconsumerdemand thanthedirecteffectofthepriceincreaseitself.Thearticlecan thusbereadasacontributiontotheanalysisofpriceuniformityin marketsforhighlydifferentiatedproducts,aphenomenonthathas beendetectedinanumberofempiricalstudies.
McMillan(2007)offerspartlyanecdoticevidenceofthedegree ofhomogeneityinretailers’pricingoftea,wine,clothingandbooks.
Otherpapersstudythegaininprofitsfromabandoningpriceuni- formity(orlimitedpricedifferentiation)inthemusicCDmarket (ShillerandWaldfogel,2011)andinmovietheaters(Leslie,2004;
OrbachandEinav,2007).Additionalevidencecomesfromthesticky priceliterature,whichemphasizeslowinter-temporalpricedis- crimination(BallandMankiw,1994).
Threeexplanations,allhingingonthesupplyside,havebeen suggestedtoaddresstheproblemofuniformity,purportedlynon- optimal,inpricing:(1)menucosts;(2)softeningcompetition;(3) verticalcontracts.Someauthors(Leslie,2004;BallandMankiw, 1994;McMillan,2007)andanumberofbusinessmen(interviewed byMcMillan,2007)maintainthatmenucosts,variouslydefined, canexplainfirms’pricingpractices.Thatis,theadditionalearnings thatcanbeattainedbygreaterpricedifferentiation,donotalways coverthecostoftheprecisecalculation(andupdating)ofpricestai- loredtodistinctdemandsegments.Asecondargumentseesprice uniformityastheoutcomeofstrategicbehaviorbyfirms,aimed at softeningcompetition acrossdifferentmarkets(Corts,1998).
Finally,OrbachandEinav (2007),toexplainwhattheyconsider asunprofitablepricingbehaviorbymovietheatermanagers,trace thecausetothelegalregimethatforbidsverticalarrangements betweendistributorsandthetheaters.
The authorsthemselves acknowledgethatthese supply-side approachesdonotofferaconvincingexplanationofthelowdegree ofpricedifferentiationtheyfind.Weaccordinglyturntoadifferent lineofresearch,hingingonthedemandsideandinparticularon considerationsofpricefairnessorunfairnessthatappeartomoti- vateconsumerbehavior.
3. Transactionutilityasanexplanationfordemandcross elasticity
Our basic intuition can beset out as follows.Traders natu- rallylookforareference-price,which isa syntheticindicatorof generalmarketconditions,conveyinginformationrelevantforthe exchange.Intraditionalmarkets,wherethegoodsarestandardor undifferentiated,thisfunctionisperformedbythecompetingprod- uctsthemselves.Forthisreasonthefunctioncannotbeisolated, asitisassociatedwithrelationshipsofsubstitutability.Butmar- ketswithhighlydifferentiatedgoodscanbeconsideredgenuinely monopolistic,asnoproductcompeteswithanyoftheothers.Nev- ertheless,consumersstillneedsomereference-pricetoassessthe fairnessofthedealtheyareoffered,sotheytakewhatisavail-
oftheliteratureonpricefairnessisfocusedonthesellerside,ourpaperismainly focusedonprice-unfairnessissuesontheconsumerside.Brown(2005)presentsa realcashexperimentinwhichlossaversiondoesnotdependontheendowment effectandoffersadifferentexplanationfortheWTA-WTPdisparitywithrelationto inexpensivemarketgoodswithamplesubstitutes.
258 L.Luini,P.Sabbatini/TheJournalofSocio-Economics41 (2012) 255–265 able:similarproducts.Theproductsaresimilarnotonlyinproduct
classificationbutalsoin beingsubjectedtothesame economic forces(theyallare“quasi”monopolies).Therationaleforcompar- ingquasi-monopoliescanbetracedbacktoanotionoffairprice thatmonopoliststoomustadhereto.Thepeculiarcoststructureof thesemarketsmustalsobeconsidered.Overheadcostsareoften paramount,somarginalcostsoffernoclearreferenceforfairnessin pricing,whichisonemoremotivationfortherecoursetotheprices ofsimilarproducts.
3.1. Themodel
Thecross-elasticityofdemandisusually associatedwiththe substitutabilityofproducts.Thedemandforproduct“1increases asthepriceofproduct“2rises, becausesomeconsumersshift toproduct“1:that is,theyconsiderthetwoproductstosome extentinterchangeable,sovariationsinrelativepricesmodifythe demandforthetwogoods.Thisrelationshipcanbeexemplifiedby employingtheBowleymodelofdifferentiatedproducts(Bowley, 1924)9:inthecaseoftwosubstituteproducts(x1,x2)theutility functionfortherepresentativeconsumerandthebudgetconstraint arerespectively:
U(x1,x2,m)=a(x1+x2)−1
2b(x21+x1x2+x22)+m (1a)
M=x1p1+x2p2+mpm (1b)
where is the parameter that measures the degreeof substi- tutabilitybetweenthetwogoods,misalltheothergoods(price normalizedatpm=1),aandbaretheparametersthatmeasurethe weightofproductsx1andx2respecttothecompositegoodxm,and piisthepriceofgoodi,fori=1,2.
From(1a)and(1b)wederivethetwodemandfunctions:
x1=c0−c1p1+c2p2 (2)
x2=c0+c2p1−c1p2 (3)
where c0= a(1−)
b(1−2) c1= 1
b(1−2) c2=
b(1−2)
Ourintuitionisthatwecanobservedemandcross-elasticityeven betweentwo productsthatarenotsubstitutes foroneanother.
Inotherwords,thethesisisthattheredoesexistdemandcross- elasticitynotassociatedwithadirectshiftinconsumptionbetween twoproductsandthat thisrelationshipdepends onconsumers’
perceptionofpricefairnessorunfairness.
Thisinsightcouldbeframedinthelanguageoftransactionutil- itytheory(KahnemanandTversky,1979;Thaler,1985;Kahneman etal.,1986a), whichpositsthatthetotalutilityaccruingfroma transactionisgivenbythesumoftwocomponents:acquisition utilityandtransactionutility.
Acquisitionutilityisthefactorconsideredbythetraditionalthe- ory,andisequaltothevaluederivedfromconsumingthegood lessthepricepaidforit.Transactionutilityisrelatedtoconsidera- tionsoffairnessanddependsonthecomparisonbetweentheactual pricechargedbythesellerandsomereferenceprice.Itfollowsthat transactionutilitycanalsobenegative,inthecaseofperceived unfairness.
9SeealsoMartin(2002,pp.52–54)andMotta(2004,pp.561–563).
Toadaptthisconceptualizationtotheequationsabove,letus nowconsideraproductwhoseprice(py)istherepresentativecon- sumer’sreferencepriceforfairness10:
U(x1,x2,m,py)=a(x1+x2)−1
2b(x21+2x1x2+x22)
+m+(x1+x2)py (4)
Subjecttotheabovebudgetconstraint(1b).Whereistheparam- eterthatmeasuresthereferencingstrengthofthereference-price (py).
ThesecondpartofEq.(4)dealswithconsumers’perceptionof fairness:theutilityofconsumingx1orx2changesinproportion() tothereferenceprice(py).Inpractice,weassumethattheutility totherepresentativeconsumeralsodependsonthehypothetical valueofthetwoproductspricedatthereferenceprice.Noticethat theproductwhosepriceischosenasreferencedoesnotenterthe consumer’spreferenceset.11Thereforeproductsareonlyallowed tohavemutuallyexclusivesubstitutionorfairnesseffects.From thisutilityfunctionwecanderivethefollowingsystemoflinear demands:
x1=c0−c1p1+c2p2+c3py (5) x2=c0+c2p1−c1p2+c3py (6) wherec1andc2asbeforeandc3=((1−))/b(1−2).
Thisdemandsystemlooksfamiliar.Buttheinterpretationof thecoefficient(c3)forthereferenceprice(py)is differentfrom thestandardone(c2).Generally,thepricecoefficientofoneprod- uctinanotherproduct’sdemandequationimpliessomedegreeof substitution.Inourcase,thecoefficient(c3)dependsontheparam- eter,whichaccountsforthepricefairnessperception.12Italso dependsonthedegreeof“closeness”()betweenthegoodsx1and x2:because∂c3/∂<0,themorecloselytheproductsaresubstitutes, thesmalleristheroleplayedbyconsiderationsoffairness.
Thereforedemandcross-elasticitycouldarisenotonlyfromthe acquisitionutilitycomponentoftheutilityfunction,inlinewith traditionaltheory,butalsofromthepresenceofa secondcom- ponent(transactionutility)duetoareferencingsystem.Ifthisis so,thenin thepeculiarmarket contextwe considerwe should observedemandcross-elasticitywithoutashiftfromoneproduct toanother.Wealsoexpectthatpricechangesforsomeproducts doexertsomeinfluenceonthedemandforotherproductseven whentheyarenotsubstitutes.Itfollowsthatthesegoodsandser- vicesmustshowsomedegreeofpriceuniformity:firms,iftheyare awareoftheirpotentialbuyers’senseofpricefairness,willcarefully evaluatetheadverseimpactofpricedifferentiation.Thereforethe Jevonsiansingle-pricerulemayalsoapplytomarketswithhighly differentiatedproducts.
4. Thesubjectoftheexperiment:themusicCDmarket WerananexperimentonthemusicCDmarket,whichisagood candidatetotest ourmodeland ourinterpretationofdemand- drivenprice uniformity.Thismarket is highlydifferentiated, in thatconsumersexhibitstrongpreferencesforsomemusiciansand
10Thevalueofy,withrespecttom,isnegligible,sowecanstillnormalizetheprice ofpm−1tothevalueof1.
11Inthelanguageofchoicetheorywecansaythattheconsumer’s“consideration”
setdoesnotcoincidewiththe“preference”set:thisframeworkfordecisionconnects theconsiderationsetwiththeperceptionofpricefairness.Insuchacontextthe axiomofIndependenceofIrrelevantAlternativesdoesnotalwaysapply.
12One of the two dimensions of the consumer’s evaluation of fairness in Daskalopoulou(2008)ispricefarness;thesecondisservicefairness.Theauthor allowsfordifferentcombinationsofpriceandservicefairness.
L.Luini,P.Sabbatini/TheJournalofSocio-Economics41 (2012) 255–265 259
relativeindifferencetoothers.13Atthesametime,ateverypoint ofsale(physicalorvirtual)retailersdisplayaremarkabledegree ofpriceuniformity(ShillerandWaldfogel,2011),oncematurityis takenintoaccount.14
Thisfactisparticularlyclearinthecaseofdownloadeddigital music:forexample,allsongssuppliedbyAppleiTunesaresoldat thesameprice.ButitalsogoesforCDretailers,inparticularlarge distributionchains(inItaly,forexample,COOPorAuchan).Asa consequence,consumersperceivethemusicmarketwithaninner characteristicofpriceuniformity.Accordingtoarecentinvestiga- tionbytheEuropeanCommission:
“Some60%ofallthecustomerswhorespondedtothemarketinves- tigationclaimthatmajorspricetheirproductssimilarly.Notably, someclaimthatpricesaresimilarwithinpriceranges(fullprice, mid,budget),andarerathersimilarforchartalbumsbutlessfor catalogue.”(EuropeanCommission,2007,p.104)
ThepricesoftheCDsobservedforourexperiment(thetop20 oftheweekbefore)confirmedthisuniformity:most(12)havethe sameprice(D20.90), another5 wentforjust D1more,andthe remaining3weremoresharplydifferentiated.15
Thispriceuniformityistheoutcomeofacomplexprocessthat beginswithmusiclabels(majorsand independents)choosinga specificpublishedpricetodealers(PPD)foreachrecordandfixing theappropriatediscountforeachtypeofretailer.PPDsarewhole- salelistpricesappliedtoCDsaccordingtovariouscriteria(basically, typeofproductandcommercialmaturity).Firmsuseonlyalim- itednumberofPPDs:thefirst5PPDsgenerallyaccountforalarge majorityoftotalsales.Everyrecordgoesforaseriesofdifferent PPDsinthecourseofitscommerciallife,startingfromthehighest.
ThePPDsofdifferentlabelsshowminordifferencesandareusedin thesamepatternbyallofthem.Discountstoretailersdependon thetypeofretailer,volume,promotions,etc.Althoughretailprices differaccordingtotypeofretailer,thoseofCDsofthesamevintage atthesameshoparesimilar.Inpartthisreflectsadefinitepractice onthepartofretailersofequalizingthefinalpricesofCDs,which meansearningdifferentmargins.Thereforethepriceuniformityof CDsofthesamevintagesoldbythesameretailerisbasicallydueto acommonmechanismforsettingwholesaleprices(basedonPPDs) andtoretailers’tendencytoadjusttheirmark-upsforgreaterprice equality.
Howcanweexplainthispropensityforpriceuniformityonthe partofmusicfirmsandevenmoreretailers?First,wemustconsider thatfirmsarenotalwaysabletocalculatetheprecisedemandelas- ticityofanygivenCDandsoresorttosomegeneralpricingrules.
ThisisbasicallythefunctionofthePPDsystem.Insteadofinitially pricingaCDaccordingtoexpecteddemand,thelabelmayprefersto launchitatamoreorlessstandardPPDthatwillbekeptunchanged aslongastheCDreceivesattentionfromthepublic.Whendemand decreases,thecompanyrespondsbyloweringitsPPD.Thismech- anismisappealingfornicheCDs,butnotsomuchforthepricingof hitrecords.Here,abetterstrategywouldseemtobespecificcal- culationofthedemandforthatparticularrecordandthesettingof anappropriatepricebasedonthatcalculation.
Second,attheretaillevelitmightbethoughtimpracticalfor sellerstohavetodealwithmultipleprices.Oneexplanationmight bemenucosts,althoughtheseappeartobesuitableprincipallyfor explainingthepricestickinessofCDs.16Inparticular,weexpect
13 Wealsoexpectdifferentreservationpricesfordifferentartists,independently offairnessconsiderations.
14 Seller’spriceskimmingpolicy(e.g.viaintertemporalpricediscrimination)is usuallynotappliedtohitrecords.
15 PricespostedbyoneofthelargestCDsellersinItaly(Feltrinelli).
16 BrynjolssonandSmith(2000).
thaton-linesellersofMP3filesshouldnotincuranysignificant menucostsinthecaseofasharplydifferentiatedpricing.Actually, however,itispreciselyinthissectorthatwefindthegreatestuni- formityofmusicprices.Thustheuniformityofinitialretailprices musthaveotherexplanations.
Weareinclined tothinkthat thisphenomenonisassociated withthereferencefunctionofprices.Ourthesis,thatis,isthatthe pricesofCDsinanygivencategoryaretakenbyconsumersasasort ofreferencingheuristicforthepriceofanyCDinthatcategory.This wouldforcefirmstoapplythesamepriceinordernottoalienate customers.
5. Experimentdesign
Ourexperimentallowsustodetectthepresenceofreference pricingonlyinacontextofquasi-monopolies.Thatis,weemploy asimplifiedversionofthemodelindicatedinEqs.(5)and(6),asit dealswithonlyoneproduct(whichtheconsumercaneitherbuy ornot)andonereferenceprice:
x1=c0−c1p1+c3py (7)
wherec3=/b.AccordingtothefollowingmodifiedversionofBow- leyequation:
U(x1,m,py)=ax1−1
2bx21+m+x1py (8)
Theexperimentteststhefollowingprediction:
Prediction. Theexperimentalsubjectsexhibitreferencebehavior:
thereferencepriceofagoodthatisnotpartoftheirpreferenceset influencesthedecisionwhethertoacceptorrejectthedealproposed.
5.1. Thesetting
Thesubjectsoftheexperimentareundergraduatestudentsat theUniversityofSienashoppingforamusicCD.Onlystudentswho hadboughtatleastoneCDintheprevioustwelvemonths17were eligible.Theexperimentinvolvesconsumptionbehaviorconcern- ingalistof20CDs.Allarerecenthitsandallaresoldatthesame price(D18istheaveragepriceofinternetsellers).18
Theexperiment19isdesignedasfollows:
Preliminarystage–WeaskparticipantstochoosebetweenD3 incashandadiscountbonusofD9foroneCDfromaclosedlistof 20hits.Onlythosewhooptforthebonuscontinue.Thuswescreen outsubjectswithaverylowreservationprice(lessthanD9)for CDs.Atthispointthesubjectsareaskedtofilloutaquestionnaire toelicittheirown“open”list.Theopenlistisdesignedtorevealthe subject’spersonal“open”preferences,asagainstthe“closed”listof hits.Theycanlistuptotwentyoftheirownfavoritemusicians,and inthiswaytheexperimenterscancaptureheterogeneoustastesof thesubjects.
Firststage–Weaskstudents torevealtheirfirstand second preferences20ofCDs,fromtheclosedlistof20hits,atdifferentdis- counts.WestartbyaskingallparticipantswhichCDtheywould liketobuywithaninitialdiscountofD9.Thenweaskiftheywant tobuyotherCDsatthesamediscountedprice.Weprogressively increasethediscount(uptoD11)andchecktheirreactions.Theaim
17Becauseofthegrowingpopularityofdownloading,thedemandforCDshas decreasedsubstantiallyinthelastfewyears(RobandWaldfogel,2006).
18ThepricechargedbythelocalshopinSienawasapproximately15%higher.
19The experiment is conducted on computers using the “z-Tree” program (Fischbacher,2009).
20Theonlypurposeofthisfirststageistoelicitingreservationpricesfortheclosed list.WedonotcommittosellCDsatthevariouspricesindicated.Wewerecareful tokeepparticipantsfromformingtheexpectationthattheywouldbeabletobuy oneCDatthepriceindicatedinthisstage.
260 L.Luini,P.Sabbatini/TheJournalofSocio-Economics41 (2012) 255–265 ofthisstageistosoliciteachparticipanttorevealher/hisreserva-
tionpriceforthespecificlistofCDs.Thisstimulateseachsubject torevealherfirstchoice,secondchoicesand,finally,whatwecall herethe“irrelevantalternatives”,i.e.CDsshedoesnotwanttobuy evenwhentheirpriceisverylow.
Secondstage–Westartthisstagebymodifyingtheoriginallist of20 hitCDs.Foreachparticipantthenew,shorterlistconsists ofthefirstchoiceandallthe“irrelevantalternatives”.Thatis,we eliminateallsecondchoices.Wethendividetheparticipantsinto twogroups.Weofferonegroup(Sample1)thepossibilityofbuying oneCDfromthemodifiedlistatthefollowingprices:D13(aD5 discount)forthefirstchoiceandD9(theoriginalD9discount)for theirrelevantalternatives.Weofferthesecondgroup(Sample2) thepossibilityofbuyingoneCDfromthemodifiedlistforthesame priceofD13(regardlessofwhetheritisthefirstchoiceoroneof theirrelevantalternatives).StudentswhodecidenottobuyanyCD arecompensatedwithadditionalD 2.Thefunctionofthisstageisto checkwhetherthepriceoftherejected(“irrelevant”)alternatives influencesthedecisiontobuythefavoriteCD.21
Alltheparticipants,whethertheychoosecashoraCDinthe finalstage,receiveD4fortheirparticipation.
Duetotheirlowincomeandthepossibilityofpiracy,studentsdo notbuymanyCDsandgenerallyhaveaverylowreservationprice forrecordedmusic.Wehandledthisproblem,first,byexcluding thosewhopreferredasmallsum(D3)toasubstantialdiscountof D9ononeoftheCDsfromourlist.Wealsoofferedaverysmall amountofmoneyasanalternativetothediscountonmusicpro- posedinthelaststageoftheexperiment.
5.2. Characteristicsofthesubjects
Thepreliminaryandfirstphasesservetogeteachsubjectto revealher/hisfirstchoiceandallsuccessivechoicesbyproviding theexperimenterswithinformationaboutaseriesofvariables:
(a)numberofCDsboughtinthepreviousyear(preliminaryphase);
(b)preferencesfor“open”list(preliminaryphase);
(c)numberofsecondchoiceselicitedunderthe“closed”listoftop hits(firststage);
(d)numberofCDsselectedunderhighdiscountconditions.
Thankstothepreliminaryandfirstphases,welearnthepartici- pants’preferences,whicharethenusedtosetupthelaststage.We arealsoabletoadjustoursampleforheterogeneitybycapturing whetherthesubjectsdifferbothin“varietypreference”(variables:
(a)–(c))andin“pricesensitivity”(variable:(d)).
Becausetheexperimentisbasedonthereactionsoftwogroups facingdifferentpricestructures,itisimportantthatthetwogroups arehomogeneous.Astheaimistotesthowandwhetherreferenc- ingtothepricesof“irrelevant|”productsconditionsthebehaviorof subjectsfacedwithapriceincreasefortheirpreferredproduct,the riskisthattheresultsmaybedistortedbyotherfactors.Wedeal
21Priceunfairnesscanalsobeperceivedaslackofimpartiality,becausesubjects– byrevealingtheirpreferencesintheclosedlistinthefirststageoftheexperiment– mayconsiderthisasasortofoptiontobuyaCDatabetterpriceinthesecondstage.
Adiscussantofthispaperlabeledthisconsumer’sperceptionasasenseofdeception, notunfairnessproper.Wedonotagree,becausenoexplicitpromisewasmadein thefirststage.Inanycase,thepointiscompletelyinconsequentialwithrespectto themainresultoftheexperimentwhichinvolvesdifferentbehaviorsexplainedby thepricesofirrelevantproducts.Itmayalsobethatsubjectsreactedtotheirsense oflackofimpartialityratherthanalesssevereperceptionofpriceunfairness;the factremainsthattheydidsotodifferingdegree,andthatthisisexplainedbythe pricesoftheirrelevantproducts.Inthiswaynofeelingofexploitation(andnoteven asenseofbetrayaloftrust)canarise.Forasimilarexperimentalinterpretationof thedifferentialpreferenceforfairnessamongsubjects,seeKarnietal.(2008).
withthisbypayingattentiontosubjects’heterogeneitywithrela- tionto:(1)therangeofpreferences(“variety”effect)and(2)their sensitivitytodecreasingprice(“price-sensitivity”effect). “Price- sensitivity”isanobviousfactorthatcouldinfluenceourresults,so inpartitioningtheparticipantsintotwogroupsinthelaststage oftheexperiment,wemaintaintheoriginalproportionsofprice- sensitiveandprice-insensitivesubjects.Similarly,wepayattention tothedifferenceinattitudestovariety.Althoughwedonothavea strongaprioriontheimpactofthisfactor,wesuspectthatitcould influenceparticipants’decisionsinthelaststage.Inpresentingand discussingtheultimateresults,weshallshownotonlythatthis suspicioniswellfoundedbutalsothatthisfactorappearstointer- actwiththesenseofunfairnesswhosedetectionistheprimeaim oftheexperiment.
Inordertoconstructappropriateindicatorsofthesetwoeffects, weproceedasfollows.First,inordertocaptureonesourceofhet- erogeneity,the“varietypreferenceeffect”,thesubjectsareasked, beforefacingthe“closed”listofthetop20hits,tofillinaques- tionnairewiththeirown“open”list.Thiselicitstherevelationof personalpreferences,asthesubjectscanlistuptotwentyoftheir ownfavorite musicians.The experimenters– bycomparingthe choicesfromtheopenlist(variablesaandb)withthechoicesmade bythesamesubjectfromtheclosedlist(variablec)–areableto classifysubjectsintermsoftheirpreferences.Wehaveidentified twotypesofconsumer:(1)variety-preferenceconsumerswhoin bothopenandclosedlistsselectmanymusiciansand(2)focused consumers,whoinbothlistsselectonlyafewmusicians.22
The“pricesensitivityeffect”isdetectedbystudyingsubjects’
behaviorwhen askedtorevealtheirfirstandsecondchoicesat differentlevelsofdiscount.Theaimofthisincreasingdiscountis toelicit“pricesensitivity”.Withasuccessionofincreasesinthe discount(initiallyfromD9 toD10 andthenfromD10toD11) thesubjectsexperimentallyreveal theirswitch tocheaperCDs.
Measuringthepreferenceforpricedecrease,wegaugethe“price sensitivityeffect.”23
Bythisprocedureweclassifyourparticipantsaccordingtothe potential two sources of the heterogeneity we have identified, namelythevarietyandprice-sensitivityeffects(Table1).
The choices of 21 subjects show desire for variety and 10 subjectspresentaclearsensitivitytodecreasingprices.Thepro- portionbetween“price-sensitive”and“price-insensitive”andthat betweenpro-varietyandfocusedsubjectsaresimilarforthetwo groupsthattakepartinthetwodifferenttreatments.
6. Theresults
6.1. Themainresult
Inthesecondstageeachsubjectfacesonlyhis/heronefavorite CDandalltheirrelevantones.Thesubjectsarepartitionedintwo groups–similar,asnoted,inallrespectsexceptthatonegroup (Sample1)facesapriceincreaseforallCDsincludingthefavorite (allCDsD13),whiletheother(Sample2)facesapriceincreasefor
22InthefirststageeachsubjectisaskedwhichCD(s)hemightbuywithaninitial discountofD 9offtheInternetprice.Thenthesubjectsareaskediftheywant possiblytobuyotherCDsatthesameprice,measuringinthiswaytheir“love-of- variety”.Inthisexperimentwedefinea“love-of-variety”subjectasonewhoselects 3ormoreCDsatsameprice(no≥3)anda“focused”consumerasasubjectwho selectsonly1oratmost2(no≤2).Thisclassificationisconsistentwiththeranking obtainedbycheckingthenumberofCDsindicatedintheopenlistatthepreliminary stage.
23Inthisexperimentwedefineaprice-sensitivesubjectasonewhoselectsalarger numberofCDsatthenewdiscountedprice(nv)thanattheinitialprice(no),such that(nv>no);whilewedefineaprice-insensitivesubjectasonewhoatthenew discountedpricesselectsfewerCDsthanattheinitialpriceornone(no).
L.Luini,P.Sabbatini/TheJournalofSocio-Economics41 (2012) 255–265 261
Table1
Classificationofsubjectsaccordingtosourceofheterogeneity:varietypreferenceandpricesensitivity(42subjectswhoparticipatedafterthepreliminaryphase).
Sample1(equalprices) Sample2(differentprices)
Varietypreference Varietyadverse Total Varietypreference Varietyadverse Total
Pricesensitive 2 2 4 4 2 6
Priceinsensitive 7 9 16 8 8 16
Total 9 11 20 12 10 22
Table2
Acceptanceandrejectionbyparticipantsinstagetwo.
Session Subjectsin eachsession
PreferD3cash andstop
PreferD9bonus andcontinue
AllCDsD13 FirstchoiceCDD13,othersD9
Accept(Y) Reject(N) Accept(Y) Reject(N)
1 7 1 6 2 1 0 3
2 12 1 11 4 1 2 4
3 12 3 9 4 0 2 3
4 12 5 7 2 1 1 3
5 6 1 5 2 1 0 2
6 8 4 4 0 2 1 1
Total 57 15 42 14 6 6 16
Table3
Buyers,non-buyersandthederivedarcelasticities(distributedaccordingtosamplecharacteristics).
Sample1(equalprices) Sample2(differentprices)
Varietypreference Varietyadverse Total Elasticity Varietypreference Varietyadverse Total Elasticity
Pricesensitive (0;2) (0;2) (0;4) −5.5 (0;4) (0;4) (0;6) −5.50
Priceinsensitive (5;2) (9;0) (14;2) −0.37 (0;8) (6;2) (6;10) −2.50
Total (5;4) (9;2) (14;6) −0.97 (0;12) (6;6) (6;16) −3.14
Elasticity −1.57 −0.55 −0.97 −5.50 −1.38 −3.14
ThefirstnumberwithineachsetofbracketsisthenumberofCDbuyersandthesecondisthenumberwhoprefercash.
thefavoritebutnotfortheirrelevantalternatives(firstchoiceD13, othersD9).Inthiswaywecontrolfortheeffectoftheprice-risefor theunselected(irrelevant)alternativeCDsonthedemandforthe firstchoice.Accordingtotheconventionaltheory,weshouldnot expectanydifferencebetweenthetwogroupsinthereactiontothe higherpriceofthefavoriteCD.Subjectsshouldonlybeconcerned withthepriceoftheirownchoiceandshouldignorethepriceofthe CDstheyarenotinterestedin.Butourexperimentoffersunambigu- ousevidencethatdisconfirmsthisview.Theresultofthisstagecan besummarizedbycomparingthenumbersofbuyersofthefirst choice(Y)withthenumberofnon-buyers(N)inthetwogroups (Table2):
(1)Groupwithequalprice(D13)for firstchoiceandirrelevant alternatives:Y=14,N=6;thatis,their“rejectionrate”is6/20.
(2)Groupfacingdifferentpricesforfirstchoice(D13)andirrele- vantalternatives(D9):Y=6,N=16;thatis,arejectionrateof 16/22.
Thechoicesclearlyshowtherelevanceoftheirrelevantalterna- tives.Thedataunambiguouslyindicatethatwhenthepriceofthe alternativesrisesbythesameamountasthatofthefavoriteCD, consumersarelessreluctanttobuy.Theresultisstrongerstillcon- sideringitsconsistencyin5ofthe6sectionsoftheexperiment.
Asexpected,noonebuysaCDconsidereduninteresting,evenata betterprice.
Inshort,participantsshownaclearpropensitytousetheprice ofsupposedly“irrelevant”productsasabenchmarktogaugethe priceoftheproducttheyareinterestedin.Whenthisreference pricerisesbythesameamountasthefirst-choiceCD,thisisseen aspartofageneralpricetrend,whichweakensthenaturalreaction ofrefusingtobuy.
Wearenowinapositiontomeasurethis“fairnesseffect”exper- imentallyasthedifferenceinthebehaviorof ourtwo samples.
Withoutthefairnesseffect(whenthepriceofirrelevantalterna- tivesalsorises),demandshrinksby30%,24givingaprice-elasticity ofdemandof−0.97(Table3).Withthefairnesseffect(whenonly thepriceoffirstchoiceisraised)demanddecreasesbyafurther 42%,25implyinganelasticityof−3.14.Ourresultissimilartothat ofAndersonandSimester(2008),whoalsofindthattheunfairness effect26ismorerelevantthanthestandardeffectofapriceincrease.
Theexperimentalsessionsconfirmourmainpoint:
Result1. Peopletendtoresistapriceincreaseinsofarastheyuseas areferencetheunchangedpriceforsimilarbutnon-substitutegoods.
Thisisexactlyinlinewithourpredictionthatthesubjectswillexhibit reference-price behavior:thepriceof aproductthat isnotpartof thepreference setneverthelessinfluencesthedecision torejectthe purchase.
6.2. Asecondaryresult
Weobtainadditionalinformationonoursubjects’behaviorby controllingforthesourcesofheterogeneity:varietypreferenceand pricesensitivity.Thisanalysisconfirmsouraprioriexpectations ontherelevanceofprice-sensitivity,butitalsorevealssomeunex- pectedoutcomesassociatedwithvarietypreference.
24Percentageofrejectionsinthe“sameprice”group.
25Percentageofrejectionsintheothergrouplesstheformerpercentage.
26InDaskalopoulou(2008)themeasureofoverallfairnessisdifferentfromprice fairnessandservicefairnessbecauseofthedifferentdegreesoffairnessascribedto eachspecificseller;inourwork,thepriceofalltheothernon-substitutesentersthe subject’ssenseofpriceunfairnessinthesameway.