lOMoARcPSD|835417 ArticleSummariesGamesandStrategy Week1:Gibbons(1997)–IntroductiontoApplicableGameTheory Game theory consist of simulations of strategic decisions that actors have to make based on their information and risk and time preferences. It is useful to analyse competition in market entry situations.Gibbonsdistinguishesfourtypesofgames,inwhichcompleteinformationmeansthatthere is no private information on timing and feasible moves. Static games refer to simultaneousmove gamesanddynamicgamesreferstosequentialmovegames. Staticgameswithcompleteinformation Dynamicgameswithcompleteinformation Staticgameswithincompleteinformation Dynamicgameswithincompleteinformation ANashequilibriumisthe courseofactionsinwhichnoplayerwantstodeviatefromtheoutcome payoff(selfenforcing).NoteverygamehasapureNashequilibrium,incasemixedstrategiesshould be applied. Games can also have multiple Nash equilibria. An application of a static game with completeinformationisCournot’sduopolymodelforcompetition.Dynamicgameswithasequential elementcan often be solved using backwardinduction,resultinginaNashequilibrium.Tofind all equilibriaalsoallnoncrediblethreatshavetobetakenintoaccount.TofindsubgameperfectNash equilibria,thesubgameshavetobeanalysed. Whengamesarerepeated,peopletakefuturecooperationorpunishmentintoaccountintheircurrent actions.Thereisadistinctionbetweenfiniteandinfinitelyrepeatedgames,focussedontheendpoint. FinitelyrepeatedgamesareplayedNtimeswithasubgameperfectNashequilibriumineverygame (doesn’thavetobeplayedthough).Infinitelyrepeatedgamesdependontriggerstrategies.Thismeans playerscanchoosetocooperateconditionallyonotherplayerscooperation.Anoncooperativeaction triggers to deviate from cooperating in future games. Also a discount factor values future payoffs accountto = # #$% .Playershaveincentivetoplaycooperateaslongas ≥ '()(*+,-../(%0+( '()(*+,123456 holds. Inrepeatedgamescoordinationfailurescanbeovercomethrough communicationandleadership. Playerscanalsolearnfrompastgames,leadingthemtochooseactionclosertoequilibriumovertime. One type of dynamic games with incomplete information are signalling games, in which private informationcanbesignalled(misleadingornot)totheuninformedparty.Theinformedplayerhasan incentive to communicate strategically and uninformed party takes this in account. In this game strategiesareasimportasbeliefs,leadingtoaperfectBayesianequilibrium.Extrarequirements: 1. Whenaplayerisuncertainabouthistoricmoves,hemusthaveabeliefoverhistoricplay 2. Strategiesmustbesequentiallyrational,giventheirbeliefsatthatpointinthegame 3. BeliefsshouldbedeterminedbyBayes’rulesfromtheplayersequilibriumstrategies Week1:Camerer(2003)–Behaviouralstudiesofstrategicthinkingingames Gametheoryisamathematicallanguagefordescribingstrategicinteractions,inwhicheachplayer’s choiceaffectsthepayoffofotherplayers.Theimpactofgametheoryinpsychologyhasbeenlimited bythelackofcognitivemechanismsunderlyinggametheoreticpredictions.Behaviouralgametheory isarecentapproachlinkinggametheorytocognitivesciencebeaddingcognitivedetailsaboutsocial utilityfunctions,theoriesoflimitsoniteratedthinkingandstatisticaltheoriesofhowplayerslearnand influenceothers.Newdirectionsincludetheeffectsofgamedescriptionsonchoice(framing),strategic heuristicsandmentalrepresentation.Theseideaswillhelprootfametheorymoredeeplyincognitive scienceandextendthescopeofbothenterprises. lOMoARcPSD|835417 Week2:KingCasasetal(2005)–Trustintwopersoneconomicsexchange Usingamultiroundversionofaneconomicexchange(trustgame),theauthorsreportthatreciprocity expressed by one player, strongly predicts future trust expressed by their partner – a behavioural findingmirroredbyneuralresponsesinthedorsalstriatum.Here,analyseswithinandbetweenbrains revealed two signals – one encoded by response magnitude and the other by response timing. Response magnitude correlated with the ‘intention to trust’ shifted its time of occurrence by 14 secondsasplayersreputationsdeveloped.Thistemporaltransferresemblesasimilarshiftofreward predictionerrorscommontoreinforcementlearningmodels,butinthecontextofasocialexchange. These data extend previous modelbased functional magnetic resonance imaging studies into the socialdomainandbroadenourviewofthespectrumoffunctionsimplementedbythedorsalstriatum. Week2:Sanfeyetal(2003)–Neuralbasisofdecisionmakingintheultimatumgame Standardeconomicrationaleisthataproposeroffersthemarginalpriceorcontractpossibletothe responder that he will accept. Evidence shows however, that offers are around 50% of the total amountandthatrejectionoftenfollowsfromreactiontotheofferbeingperceivedasunfair.Sanfey etalinvestigateneuralandbehaviouralreactionstofairoffers(50/50)ofunfairoffers(unequalsplit inproposersadvantage).Participantsacceptedallfairoffers,butacceptanceratedropsthelessfair theoffergets,tothepointthathumansrejectionrateissignificantlyhigherthanthatofacomputer. MRIscansalsoshowbrainactivityinregionsassociatedwithnegativeemotionswhenunfairoffersare made. Week3:BertrandandSchoar(2003)–ManagingwithStyle Howmuchdoindividualmanagersmatterforfirmbehaviourandeconomicperformance?Managers areoftenperceivedashavingtheirownstyleswhenmakinginvestment,financing,andotherstrategic decisions,therebyimprintingtheirpersonalmarksonthecompaniestheymanage. We estimate the role of managers in a framework where we can control for observable and unobservabledifferencesacrossfirms. We construct a managerfirm matched panel data set, where we can track individual top managersacrossdifferentfirmsovertime. Allowstoestimatehowmuchoftheunexplainedvariationinfirmpracticescanbeattributed tomanagerfixedeffects,aftercontrollingforfirmFEandtimevaryingfirmcharacteristics. OurresultsshowthatmanagerFEareempiricallyimportantdeterminantsofawiderangeofcorporate variables.Theseresultsprovideevidencethattopexecutivesvaryconsiderablyintheirmanagement stylesandtherebysuggestarathernovelapproachforcorporatefinanceresearch: 1. Weshowthatthedifferencesinmanagerialpracticesaresystematicallyrelatedtodifferences inperformance. 2. Weshowthatmanagerswithhigherperformancefixedeffectsalsoreceivehighersalary,and thatthesemanagersaremorelikelytobefoundinbettergovernedfirms. 3. We find that older generations of CEOs appear overall more conservative in their decision making. 4. ManagerwhoholdanMBAdegreeappearoveralltofollowmoreaggressivestrategies. Weaskhowmuchofthevarianceinaparticularcorporatepolicycanbeattributedtomanagerspecific effects. lOMoARcPSD|835417 Wearenothopingtoestimatethecausaleffectofmanagersonfirmpractices,insteadwewantto assesswhetherthereisanyevidencethatfirmpoliciessystematicallychangewiththeidentityofthe managerinthesefirms. 1. Results in Table 3+4 suggest that managerspecific effects matter both economically and statisticallyforthepolicydecisionsoffirms:managerfixedeffectsincreasetheadj.R2ofthe estimatedmodelssignificantly. 2. Thereareimportantdifferencesastowhichdecisionvariablesseemtobemostaffectedby managerdecisions. 3. Holdsforbothfinancialpoliciesandorganisationalpolicyvariables(R&D,SG&A,advertising, etc.). 4. PanelBfocusesonmeasuresofcorporateperformance,i.e.therateofreturnonassetsand operatingcashflowasaratiooftotalassets(addressesconcernof“cookingthebooks”). One might worry that the manager fixed effects identified above do not imply persistence of managerialstyleacrossjobsandfirms. Consideramanagerwhohappenstobepartofafirmduringaperiodofintenseacquisition activity;wemightestimateapositiveacquisitionfixedeffectforthatmanagereventhough thateffectdoesnotpersistinhisfuturefirm. Table5addressesthisissue.Weregressamanager’saverageresidualinhissecondfirmonhisaverage residualinthefirstfirm.Wefindapositiveandsignificantrelationshipbetweenamanager’sresidual inhislastjobandhisresidualinhisfirstjobforallpolicyvariables. Forexample,atopmanagerassociatedwith1percentextraleverageinhisfirstjobisassociatedwith about 0.5 percent extra leverage in his second. These results are consistent a persistence of the managerfixedeffectsacrossfirms. WewanttoarguethatthatthemanagerFEcapturetheactiveinfluenceofmanagersoncorporate decisions. Amanagermay,becoincidence,beinvolvedinawaveofacquisitionsinherorhispriorfirm, andmaybewronglyperceivedasanacquisitionstylebyotherboards.Ifthisleadstothehiring ofthatmanagerbyafirmthatwouldhavegoneonanexternalexpansionphaseevenwithout themanager,wemightestimateapositiveacquisitionfixedeffectforthatmanager. Toinvestigatethisalternativeinterpretation,weanalysetheprecisetimingoftheobservedchanges incorporatepolicies. Onemightexpectthatsomeofthechangesinpoliciesprecedethearrivalofthenewmanager. On the other hand, if managers do play the active role in shaping corporate policies, the changesinpolicywillonlyhappenafterthemanagerishired. We,therefore,assumethateachmanagerinourdatasetjoinshissecondfirmthreeyearspriortothe actualturnoverdateandleavesthatfirmatthetimeoftheactualturnoverdate. Wefindthatestimatedcoefficientsintheseplaceboregressionsareeconomicallyverysmall (closetozero). Theseresultsconfirmthatthebulkofthechangesincorporatepolicyhappenoncethenew managerhasjoinedthefirm,andnotpriortohisarrival. lOMoARcPSD|835417 Wewanttogoastepfurtherandinvestigatewhetherthereareoverarchingpatternsinmanagerial decisionmaking.Weanalysethecorrelationstructurebetweenthemanagerspecificfixedeffects. Managers seem to differ in their approach toward external versus internal growth: strong negativecorrelationbetweencapitalexpendituresandacquisitions. Anotherestimationsuggestthatbettergovernedfirmsselectmanagerswithperformanceenhancing styles. We see that managers with higher returns on asset fixed effects receive higher residual total compensationaswellashighersalarycompensation. Furthermore,CEOswithanMBAappeartobemoreaggressive,choosingtoengageinahigherlevelof capitalexpenditures,holdmoredebt,andpaylessdividends. CEOsfromoldergenerationsappeartobelessaggressiveonaverage. Week3:RotembergandSaloner(2000)–Visionaries,ManagersandStrategicDirection Incentives for profitable innovation can be enhanced by employing a ‘visionary’ CEO with a vision biasedtowardscertainprojects.CEOvisionchangestheprojectsthatgetimplementedandthusaffects theincentivesofemployeeswhoareonlyrewardedforinnovativeideaswhentheyarerealised.Profits maybeenhancedfurtherbylettingobjectivemanagersdecidewhichprojectstoinvestigate,although theirdecisionscandifferfromthefirmsstrategysetoutbythebiasedCEO. AbiasedCEOinfluencesthedirectionofthefirm,wherethemiddlemanagercanstillhaveimpacton thestrategicdirectionthroughallocationofresources.Theauthorsseetheroleofstrategytoimprove the firms set of innovative projects. Employees whose plans fit the strategy are motivatedto take innovativeactions.CEObehaviourcanbemodelledwhenheisconsistentlybiasedtocertainprojects. Thebasichierarchalmodel:seniormanagement(CEO),middlemanagerandemployee.TheCEO: Decideswhichprojectsthefirmimplements Hasavisionontheindustryandmightbeoverconfidentabouttheavailableinformation Hasabiasedvisioncanbedifferentfromshareholder’sunbiasedviews(baselineviews) Themiddlemanagerisresponsibleforallocationresourcesandauthorizingemployeesinnovation Employeesgenerateinnovationsbyexertingeffortandreceiveincentivepayforsuccessfulinnovations Thebasicmodeldetails: Firmcanengageintwolinesofbusinessi{a,b}inwhichemployeecanundertakeinnovation Innovationcomesateffortcosts4 andrequiresexpendituresoffirmsresources4 Withprobability4 theinnovationissuccessful,generatingbenefits4 Atmostoneinnovationcanbeimplementedforwhichtheemployeereceivesincentivepay4 Thebasicmodeltiming: Period0Firmdecidestoinvestinbothaandborjustone.Alsothe(biased)CEOishired. Period1Firmoffersincentivecontracttoemployeesinwhichtheyreceivepayforimplementation. Period2Employeesdecidewhetherornottoexerteffort. Period3CEOobservesemployees’innovationsanddecidesonimplementation(onlyonepossible). Thebasicmodelnarrowstrategy(firstbest): Jointsurplusforinnovationofoneactivityi4 = 4 4 − 4 where4 = 4 + 4 and = , Canfirmachievefirstbest?Solvewithbackwardinduction: Period3Firmimplementsinnovationiif4 − 4 ≥ 0 lOMoARcPSD|835417 ( Period2Employeeexertseffortif4 4 − 4 ≥ 0so4 ≥ 1E E ( Period1Firmiswillingtopay4 = 1E if4 = 4 4 − 4 ≥ 0 E ( Iffirmfreelychooses4 ithasnoincentiveproblemsandanincentivepayof4 = 1E yieldsfirstbest. E Firstbest:outcomethefirmcouldattainifitpaysemployeesdirectlyfortheireffort.Expectedprofit fromhavingonlytheemployeeinactivityiattempttoinnovate:4 = 4 4 − 4 (4 = 4 + 4 , = , ) Thebasicmodelfirstbestbroadstrategy: Jointsurplusforinnovationinbothactivities) = 0 0 + I (1 − 0 )I − 0 − I Canbewrittenas) = 0 + I − 0 I I whereAischosenwithtwosuccesses(0 > I ) If) > 0, ) > 0 ) > I sohiringabiasedCEOcanhelpthefirmgetclosertofirstbest. CEObias:heseesvaluesofinnovationsdifferentfromtheirtruevalues CEOseesvalueofinnovationbas0 + 0 andofaasI + I where4 isthebias 4 canbepositiveornegativeand0 = I = 0referstoanunbiasedCEO ExogenousIncentivePayments IncentivepaykisexogenouslyfixedatN andreflectstheworthoftheincentivetoemployee IncentivepaypaidbyfirmisO ) anddifferencewithN ispotentialnonpecuniarycompensation QR QS WithrelativesmallO ) firmalwaysspends4 O ) < 0 − andO ) < I − 1R (#,1S ) 1S (#,1R ) Whichcontractsareprofitable?Usebackwardinduction(unbiased) Period3CEOimplementsAif0 − O ) > I − O ) > 0 ( Period 2 A exerts effort if 0 O − 0 ≥ 0 can be written as O ≥ R and B exerts effort if ( S I (1 − 0 )O − I ≥ 0canbewrittenasO ≥ 1 (#,1 ) S 1R R ( ( S so Period 1 Unbiased CEO incentivizes only A when both successful since 1R ≤ O < 1 (#,1 ) recoursesareonlyspendonAwithoptimalcontract0 X0 − O ) Y − 0 ≥ 0 R S R CanafirmdobetterwithabiasedCEO? WhenbothactivitiesaresuccessfulCEOwillimplementAif0 − O ) + 0 ≥ I − O ) + I IfI − 0 > 0 − I > 0theCEOwillimplementBbecause0 − O ) + 0 < I − O ) + I AisonlyimplementedifinnovationBhasfailed,soAonlyexertseffortif0 (1 − I )O > 0 Situation1:0 O > 0 , I O > I , I (1 − 0 )O < I and0 (1 − I )O < 0 Neitheremployeeexertseffortifhisinnovationisonlyimplementeduponfailureoftheothers,sothe biasoftheCEOdetermineswhoexertseffort.Firmprofitswillbe0 (0 − O ) ) − 0 (narrowstrategy). AbiasedCEOtowardsBwillbeprofitableifI XI − O ) Y − I > 0 X0 − O ) Y − 0 Situation2:0 O > 0 , I O > I , I (1 − 0 )O < I and0 (1 − I )O > 0 NowemployeeAexertseffortevenifhisinnovationisonlyimplementeduponfailureofB,whoexerts effortifhisinnovationisalwaysimplementedwhensuccessful.UnbiasedchoosesAsoonlyAexerts effort.BiasedCEOtowardsBwillimplementB,orAwhenBfails.Nowbothemployeesexerteffort. UnbiasedCEOprofit) = 0 = 0 (1 − I )O ) − 0 isthenarrowstrategyprofitwhenfocussingonA BiasedCEOtowardsBprofit) = I XI − O ) Y − I + 0 (1 − I )(0 − O ) ) − 0 Q HiringCEOwithbiasBprofitableif) > 0 I − S > 0 0 + (1 − I )O ) (bias>broadstrategy). 1S Lemma:If4 O > 4 I (1 − 0 )O < I , I − 0 > 0 − I > 0 0 (1 − I )O < 0 I XI − O ) Y − I > 0 X0 − O ) Y − 0 Q 0 (1 − I )O > 0 I − S > 0 0 + (1 − I )O ) 1S lOMoARcPSD|835417 EqualIncentivePaymentAcrossActivities Incentivepayisoptimallychosenbythefirmandthesameinbothactivities0 = I = O InthisanalysisO = O ) canbeassumed ( (S ( (S firstbestwithbothexertingeffortwhenfirmssetO = 1R = 1 (#,1 Case1:1R = 1 (#,1 ) ) R S (R (S R R S R Case2:1 > 1 (#,1 )firstbestcannotbeachieved R S R (S WhenfirmsetsO = 1 (#,1 thenAhasnoincentivesinceisdoesn’tcovereffortcosts ) S R ( WhenfirmsetsO = 1R thenbothexerteffortandBearnsarentbutbiaswon’treducerent R HiringabiasedCEOtoawouldgivethesameexpectedpayoff HiringabiasedCEOtobonlyBexertseffortsonofirstbest ( (S firstbestcannotbeachieved Case3:1R < 1 (#,1 ) R S R ( CEObiashaspositiverole,unbiasedCEOsetsO = 1R (onlyAexertseffort) R (S (S ( inwhichAwillearnrentsandexpectedprofitsare) − 0 b (#,1 − R c OrO = 1 (#,1 ) ) 1 1 S (R (S R S (S ( Case3a:1 < 1 (#,1 )and0 < ) − 0 b1 (#,1 − 1R c R S R S R) R R R ( ( R FirmscanbenefitfromhiringCEOwithI − 0 > 0 − I (biasB)andsuppose1S < 1 (#,1 ) S ( BiasedCEOsetsO = 1S andonlyBwillexerteffortgeneratingI S ( R S ( S BiasisbeneficialifI > ) − 0 b1(#,1 − 1R c ) R R (R andbothAandBexerteffort BiasedCEOsetsO = 1 (#,1 ) R S R S (S[e Xfge Yhe ( [1S (#,1R )$1R ] − (0 − I ) > 0 1S 1Ri(#,1R ) ( (S (S ( Case3b:1R < 1 (#,1 and0 < ) − 0 b1 (#,1 − 1R c ) ) R S R S R R Biasisbeneficialif 1R 1Si (#,1S ) S − R ( ( R FirmscanbenefitfromhiringCEOwithI − 0 > 0 − I (biasB)andsuppose1S > 1 (#,1 ) ( BiasedCEOsetsO = 1S andbothAandBexerteffortwhereAreceivesarent S S R S 1 $1 ,1 1 S R S I − (0 − I ) > 0 HiringabiasedCEOisbeneficialif R1i (#,1 ) R S ActivitySpecificIncentivePayment Incentivepayisoptimallychosenbythefirmandcanbedifferentforbothactivities ( ( Case1:0 − 1R ≥ I − 1S firstbestcanbeachieved R ( S Firmsets0 = 1R andI = 1 R (S S(fgeR) ( ( (S 1R 1S 1S(fgeR) FirmpreferstoimplementAoverB:0 − R ≥ I − S > UnbiasedCEOimplementsBifI ≥ 1 ( (S S(fgeR) ( Case2:0 − 1R < I − 1S firstbestcannotbeachieved R (R $QR S ( ( R If0 ≥ 1 (#,1 )holds,firmsets0 = 1 (#,1 andI = 1S wherebothexerteffort ) R S ( R S S Ifitdoesn’thold,firmsetsI = S andlevel0 isnotrelevantsoBexertsandAdoesn’texert 1S lOMoARcPSD|835417 Week4:GoldfarbandXiao(2011)–ManagerialAbilityandStrategicEntryinUSTelecomMarkets Empiricalmodelshavegenerallyfailedtorecognisevarianceinmanagers’abilitiestounderstandrival firms’ strategic behaviour. Telecommunications Act of 1996 opened the competitive local telecommunications industry in the US. The Act led to substantial entry; the entrants varied significantlyinsize,management,andtelecommunicationsexperience. Earlyyearsofthisindustryprovideanidealsettingforexploringheterogeneityinthestrategicability ofmanagers.Datasuggestsastrongcorrelationbetweenmanagercharacteristicsandcompetitive considerations. Modeldrawsonlaboratoryevidenceofiterateddecisionmakinginsimultaneousgames.Cognitive Hierarchy(CH)modelstheheterogeneityofplayersinsuchawaythatitincludesaparameterthat identifiesplayersasbeingbetteratplayingthegame. Highertypesarebetterabletopredictcompetitorbehaviourandconsequentlyarelesslikelytoregret theirdecisionsoncealldecisionsareobserved.Weinterpretthehierarchyasameasureofstrategic ability.ThisinterpretationallowsustoexaminewhichCEOcharacteristicsaredeterminantsofstrategic ability. 1. Experienced,bettereducatedmanagerstendtoentermarketswithfewercompetitors. 2. Measure of strategic ability predicts outcomes outside our estimation window: firms with managersofhigherestimatedabilityaremorelikelytostayinbusinessand,conditionalon survival,havehigherrevenues. 3. Comparing results across years, we find that the measured level of ability is substantially higherin2002thanin1998. Intheregressionweusetoestimatetheprobabilityofentry,ofinterestarethesignsoftheinteractions termsbetweenthenumberofcompetitorsandmanagercharacteristics. Table2Bincludesinteractiontermsbetweenmanagercharacteristicsandthedemographiccontrols relatedtodemandpotential.Weobservethatmoreexperienced,bettereducatedmanagerschoose toentermarketswithlowerpopulations;suggestingthattheysomehowentermarketsthatothers choosenottoenter. Table3:resultsareconsistentwithouridentifyingassumptionsthatmanagercharacteristicsrelateto profitsprimarilythroughentry. Whatdrivesstrategicability(strategicabilityparameter,logτ):experience,toplevelundergraduate institute,degreeinbusinessoreconomics.Theresultssuggestthathavingadegreeineconomicsor businessisastrongsubstituteforindustryexperience,i.e.aneconomics/businessdegreecanpartially substituteforexperience. Wefindthatthepredictedτispositivelycorrelatedwithfourdifferentdefinitionsofsuccess. lOMoARcPSD|835417 Week6:Arcidiaconoetal(2016)–Competitiveeffectsofentry Thepaperlooksatthesupermarketindustryandcouplesweeklygrocerytransactionswiththeexact locationandopeningdatesofWalmartsenteringthemarketoveranelevenyearperiod.They examinehow their entry affects process and revenues at incumbent supermarkets and find that WalmartSupercenter entry within one mile of an incumbent causes a 16% drop in revenue, an effect thatdecreases quickly with distance. Despite large cross store differences in prices of supermarketsexposed,thefindingsindicatethatSupercenterhasnocausaleffectonincumbent prices. TheauthorsinvestigatetheimpactofWalmartSupercenterentryonpricesandrevenuesof incumbentsupermarkets.Walmarthas15%25%lowerpricesgivingacompetitiveadvantage.One shouldexpectthatthisstartsasizablereactioninpricesofincumbentsupermarkets.Empirical studiesfoundlargecompetitivepriceeffects,butaftercontrollingforfixedstoreeffect,nocausal impactonincumbentsupermarketpriceshasbeenfound. The sources of data are opening date and address of every Walmart and IRI marketing dataset containingweeklysalesdataforarepresentativesampleofsupermarketsbetween2001and2011. Focusison15foodproductcategories.Thedependentvariablesareweeklyrevenuesandpriceseries byproduct category.Thedatashows considerablevariationinrevenue andprices.Thedescriptive findingsshowsupermarketsnearWalmartSupercenterarelowerpriced.Second,withinstore prepostdifferencesinpricesarenotsignificantlydifferent(onlyiflocated<1mile). Empiricalmodel1:5+ = ∑X5tYt+ + 5+ (locationisnotrandommitigateendogeneity!) 5+ isthedependentvariable:loggedrevenueorloggedprices X5t Yisthefunctiondrivingdistance t+ istheindicatorvariablepresencejthSupercentertimet Locationisnotrandomsomitigateendogeneity5+ = 5 + + + ó(5) () + 5+ I t+ + 5 + + + ó(5) () + 5+ Empiricalmodel2:5+ = ∑()∑5t Generalizationdiffindiffapproach TreatedfirmsaresupermarketsexposedtoSupercenter Controlfirmsarefirmsthatarenotexposedorexposedatadifferentperiod Resultsonrevenue:Inthecrosssectionaleffectsonthemarketsarenotclear,onlysomesignificant. Includingfixedchaineffectswithinchainwithinmarketshownegativeeffectonrevenuesthecloser to the Supermarket. Within store effect shows an even stronger negative relation of revenues on Walmartentry.Graphshowsstablerevenuetrendbeforeentryandasteepdropatmomentof entry. Resultsonprices:Within1mileexposureofWalmartentryhasasmallsignificanteffectonprices, diminishingforfurtherlocations.Includingmarketandchainfixedeffectsstrengthen thisnegative relationonprices.ThisshowsWalmartlocatesclosetolowpricecompetitors.Exposuredoesn’thave effectonprices. lOMoARcPSD|835417 Noevidencefoundforthenullhypothesisthatpriceeffectcouldresultfromdifferentpricereactions acrossproducts.Alsopricesmightbenotunderdirectcontrolbecausemanufacturersofprivatelabels. Chainsupermarketsmighthavedifficultyadjustingprices(rigidprices). AuthorsinvestigatetheimpactofWalmartSupercenterentryonpricesandrevenuesofincumbent supermarkets.Impactonrevenuesarelarge(16%declinewithin1mile)butnoimpactonprices.Also itisfoundthatWalmartlocatesnearlowpricehighrevenuestores.Firmsshowtobelessresponsive todemandshocksthanassumedbytheoreticalmodels.