Entry and Competition in Concentrated Markets

(s):TimothyF.BresnahanandPeterC.ReissSource:TheJournalofPoliticalEconomy,Vol.99,No.5,(Oct.,1991),pp.977-1009Publishedby:TheUniversityofChicagoPressStableURL::20/06/200815:17YouruseoftheJSTORarchiveindicatesyouracceptanceofJSTOR'sTermsandConditionsofUse,availableat://=ucpress.EachcopyofanypartofaJSTORtransmissionmustcontainthesamecopyrightnoticethatappearsonthescreenorprintedpageofsuchtransmission.JSTORisanot-for-profitorganizationfoundedin1995tobuildtrusteddigitalarchivesforscholarship.Weworkwiththescholarlycommunitytopreservetheirworkandthematerialstheyrelyupon,andtobuildacommonresearchplatformthatpromotesthediscoveryanduseoftheseresources.FormoreinformationaboutJSTOR,[e-mail]hispaperproposesanempiricalframeworkformeasuringtheeffectsofentryinconcentratedmarkets.Buildingonmodelsofentryinatomisticallycompetitivemarkets,weshowhowthenumberofproducersinanoligopolisticmarketvarieswithchangesinde-mandandmarketcompetition.Theseanalyticalresultsstructureourempiricalanalysisofcompetitioninfiveretailandprofessionalindustries.Usingdataongeographicallyisolatedmonopolies,duop-olies,andoligopolies,westudytherelationshipbetweenthenumberoffirmsinamarket,marketsize,andcompetition.Ourempiricalresultssuggestthatcompetitiveconductchangesquicklyasthenum-berofincumbentsincreases.Inmarketswithfiveorfewerincum-bents,almostallvariationincompetitiveconductoccurswiththeentryofthesecondorthirdfirm.Surprisingly,oncethemarkethasbetweenthreeandfivefirms,thenextentranthaslittleeffectoncompetitiveconduct.I.IntroductionTheoreticalmodelsofimperfectcompetitionmakediversepredic-tionsaboutthecompetitiveeffectsofentry.Contestablemarkettheo-ries,forexample,arguethatthemerethreatofentrycurbsmarketpower.Bycontrast,entrybarriermodelsassignalimitedroletopo-tentialcompetitors,arguinginsteadthatonlyactualentryaffectscom-petition(cf.Baumol,Panzar,andWillig1982;Tirole1988;GeroskiWethanktheNationalScienceFoundationforfinancialsupportundergrantSES88-21543andtheAlfredP.SloanFoundationforprovidingfellowshipsupport.ClaraJongandMaryJoArboledaprovidedresearchassistance.Finally,wethankGeorgeStiglerandananonymousrefereefortheirhelpfulsuggestions.[JournalofPoliticalEconomy,1991,vol.99,no.5]?1991byTheUniversityofChicago.Allrightsreserved0022-3808/91/9905-0002$01.50977978JOURNALOFPOLITICALECONOMY1989;Schmalensee1989).Betweenthesetwoextremesliesarangeoftheorieswithvaryingquantitativepredictionsabouttheprevalenceandconsequencesofentry.Todiscriminateamongthesedifferenttheories,werequiremoredetailedempiricalmodelsofentry.Severalrecentpapershavedevelopeddetailedeconometricmodelsofentry'scompetitiveeffects.TheseincludeCarlton(1983),Bresna-hanandReiss(1988,1990),Schary(1988),Berry(1989),Geroski(1989),Lane(1989),andReissandSpiller(1989).'Weextendthisresearchbydevelopinganempiricalmodelofentryforsituationsinwhichonedoesnotobserveincumbents'orentrants'price-costmar-gins.OurempiricalmodelbuildsonChamberlin's(1933)andPanzarandRosse's(1987)theoreticaldescriptionoffree-entrycompetition.Incontrasttopreviousempiricalmodels,firmshaveU-shapedaver-agecostsandentrantsmayfaceentrybarriers.Fromthisfree-entrymodelwedeveloptheideaofademandentrythreshold,ameasureofthemarketsizerequiredtosupportagivennumberoffirms.Weshowthatratiosofentrythresholdsprovidescale-freemeasuresofentry'seffectonmarketconduct.Theempiricalsectionofourpaperestimatesentrythresholdsforfiveretailandprofessionalserviceindustries.Weobtaintheseesti-matesfromcross-sectiondataonthenumberoffirmsin202distinctgeographicmarkets.Thesemarketsdifferprimarilyinthenumberoflocalresidents,avariablewerelatetoourentrythresholdconcept.FollowingBresnahanandReiss(1990),wedeveloporderedprobitmodelsoftheequilibriumnumberofmarketentrants.Structuralshiftsinthesemodelsenableustoestimatetheeffectofentryonfirmprofits.Ourempiricalresultssuggestthatcompetitiveconductchangesquicklyasmarketsizeandthenumberofincumbentsin-crease.Inmarketswithfiveorfewerincumbents,almostallvariationincompetitiveconductoccurswiththeentryofthesecondandthirdfirms.Surprisingly,onceamarkethasbetweenthreeandfivefirms,thenextentranthaslittleeffectoncompetitiveconduct.Weusepricedatafromoneofourindustriestocross-validatethesefindings.Thesedatashowthatpricesfallwhenthesecondandthirdfirmsenterandthenleveloff.Wealsofindthatoligopolypricesleveloffabovethoseinacompetitiveurbanmarket.II.EntryandMarketSizeOurempiricalmodelprovidesinformatio