the effect of advertising on brand awareness

1、QuantMarkEcon(2009)7:207–236DOI10.1007s111290099066zTheeffectofadvertisingonbrawarenessperceivedquality:AnempiricalinvestigationusingpaneldataC.RobertClarkUlrichDaszelskiMichaelaDraganskaReceived:11December2007Accepted:2

2、April2009Publishedonline:8May2009?SpringerScienceBusinessMediaLLC2009AbstractWeuseapaneldatasetthatcombinesannualbrleveladvertisingexpendituresfoverthreehundredbrswithmeasuresofbrawarenessperceivedqualityfromalargescalec

3、onsumersurveytostudytheeffectofadvertising.Advertisingismodeledasadynamicinvestmentinabr’sstocksofawarenessperceivedqualityweaskhowsuchaninvestmentchangesbrawarenessqualityperceptions.Ourpaneldataallowustocontrolfunobser

4、vedheterogeneityacrossbrstoidentifytheeffectofadvertisingfromthetimeseriesvariationwithinbrs.Theyalsoallowustoaccountftheendogeneityofadvertisingthroughrecentlydevelopeddynamicpaneldataestimationtechniques.Wefindthatadve

5、rtisinghasconsistentlyasignificantpositiveeffectonbrawarenessbutnosignificanteffectonperceivedquality.KeywdsAdvertisingBrawarenessPerceivedqualityDynamicpaneldatamethodsJELClassificationL15C23H37C.R.ClarkInstituteofAppli

6、edEconomicsHECMontrealCIRPEE3000ChemindelaCoteSainteCatherineMontrealQuebecH3T2A7Canadaemail:.DaszelskiDepartmentofEconomicsHarvardUniversity1805CambridgeStreetCambridgeMA02138USAemail:.Draganska(B)GraduateSchoolofBusine

7、ssStanfdUniversityStanfdCA943055015USAemail:ffectofadvertisingonbrawarenessperceivedquality209asaffectingthechoicesetasaffectingtheutilitythattheconsumerderivesfromabr.Iftheroleofadvertisingismistakenlyspecifiedasaffecti

8、ngqualityperceptions(i.e.preferences)ratherthanbrawarenessasitoftenisthentheestimatedparametersmaybebiased.InherstudyoftheU.S.personalcomputerindustrySovinskyGoeree(2008)findsthattraditionaldemmodelsoverstatepriceelastic

9、itiesbecausetheyassumethatconsumersareawareof—hencechooseamong—allbrsinthemarketwheninactualitymostconsumersareawareofonlyasmallfractionofbrs.Fourempiricalanalysiswedevelopadynamicestimationframewk.Brawarenessperceivedqu

10、alityarenaturallyviewedasstocksthatarebuiltupovertimeinresponsetoadvertising(NerloveArrow1962).Atthesametimethesestocksdepreciateasconsumersfgetpastadvertisingcampaignsasanoldcampaignissupersededbyanewcampaign.Advertisin

11、gcanthusbethoughtofasaninvestmentinbrawarenessperceivedquality.Thedynamicnatureofadvertisingleadsustoadynamicpaneldatamodel.Inestimatingthismodelweconfronttwoimptantproblemsnamelyunobservedheterogeneityacrossbrsthepotent

12、ialendogeneityofadvertising.Wediscussthesebelow.Whenestimatingtheeffectofadvertisingacrossbrsweneedtokeepinmindthattheyaredifferentinmanyrespects.Unobservedfactsthataffectbothadvertisingexpendituresthestocksofperceivedqu

13、alityawarenessmayleadtospuriouspositiveestimatesoftheeffectofadvertising.Putdifferentlyifwedetectaneffectofadvertisingthenwecannotbesureifthiseffectiscausalinthesensethathigheradvertisingexpendituresleadtohigherbrawarene

14、ssperceivedqualityifitisspuriousinthesensethatdifferentbrshavedifferentstocksofperceivedqualityawarenessaswellasadvertisingexpenditures.Fexamplealthoughinourdatathebrsinthefastfoodcategyonaveragehavehighadvertisinghighaw

15、arenessthebrsinthecosmeticsfragrancescategyhavelowadvertisinglowawarenesswecannotinferthatadvertisingboostsawareness.Wecanonlyconcludethattherelationshipbetweenadvertisingexpendituresperceivedqualitybrawarenessdiffersfro

16、mcategytocategyevenfrombrtobr.Muchoftheexistingliteratureusescrosssectionaldatatodiscernarelationshipbetweenadvertisingexpendituresperceivedquality(e.g.KirmaniWright1989Kirmani1990MothyZhao2000MothyHawkins2005)inanattemp

17、ttotesttheideathatconsumersdrawinferencesaboutthebr’squalityfromtheamountthatisspentonadvertisingit(Nelson1974MilgromRoberts1986TellisFnell1988).Withcrosssectionaldataitisdifficulttoaccountfunobservedheterogeneityacrossb

18、rs.Indeedifweneglectpermanentdifferencesbetweenbrsthenwefindthatbothbrawarenessperceivedqualityarepositivelycrelatedwithadvertisingexpenditurestherebyreplicatingtheearlierstudies.Oncewemakefulluseofourpaneldataaccountfun