JournalofFinancialMarkets2(1999)99}134Order#owcompositionandtradingcostsinadynamiclimitordermarket1ThierryFoucault*HECSchoolofManagement,1ruedelaLibe&ration,78351JouyenJosas,FranceAbstractThisarticleprovidesagametheoreticmodelofpriceformationandorderplacementdecisionsinadynamiclimitordermarket.Investorscanchoosetoeitherpostlimitordersorsubmitmarketorders.Limitordersresultinbetterexecutionpricesbutfaceariskofnon-executionandawinner'scurseproblem.Solvingfortheequilibriumofthisdynamicgame,closed-formsolutionsfortheorderplacementstrategiesareobtained.Thus,testableimplicationsforthecross-sectionalbehaviorofthemixbetweenmarketandlimitordersandtradingcostsinlimitordermarketsarederived.(1999ElsevierScienceB.V.Allrightsreserved.JELclassixcation:G19;D44;C72Keywords:Marketmicrostructure;Limitordermarkets;Limitandmarketorders;Trad-ingcosts;Order#owcomposition1.IntroductionSeveralsecuritymarkets2areorganizedaslimitordermarkets.Inthesemarkets,buyersandsellerscarrytheirtradesbysubmittingeitherlimitordersor*Correspondingauthor.Tel.:33139679411;fax:33139677085;e-mail:foucault@hec.fr.1IamgratefultoBrunoBiaisfornumerousstimulatingandinsightfuldiscussions.Ialsothanktworeferees,MatthewSpiegel(theEditor),FranklinAllen,JimAngel(theWFAdiscussant),PhillipeHenrotte,BertrandJacquillat,Jean-CharlesRochet,AilsaRoKell,PatrickSandas,DuaneSeppiandChesterSpattfortheirsuggestions.ThepaperalsobenetedfromcommentsofseminarparticipantsatCarnegieMellonUniversity,theInstituteforEconomicAnalysis,PrincetonUniversity,ToulouseUniversityandthe1998WFAMeetingsinMonterey.TheauthorgratefullyacknowledgesthenancialsupportofDGICYTgrantPB93-0388.Allerrorsaremine.2Forinstance,theNYSE,theParisBourse,theTokyoStockExchangeandtheTorontoStockExchange.Domowitz(1993)reportsthat35nancialmarketshavefeaturesoflimitordermarkets.1386-4181/99/$-seefrontmatter(1999ElsevierScienceB.V.Allrightsreserved.PII:S1386-4181(98)00012-3marketorders.3Limitordersarestoredinalimitorderbook,waitingforfutureexecution.Thisexecutionistriggeredbyincomingmarketorders,whicharematchedwiththebesto!ersinthebook.Tradersfacethefollow-ingdilemma.Withamarketorder,atraderisexecutedwithcertainty,atthepostedpricesinthemarket.Withalimitorder,atraderhasthepossibilitytoimprovehisexecutionprice.Butsherunstheriskofnotbeingexecuted.Moreoverbecausetheirpricesarexedovertime,limitorderscanbecomemispricedwhennewpublicinformationarrives.Thispossibilitycreatesawin-ner'scurseproblemforlimitordertraderssincetheyaremorelikelytobeexecuted(&pickedo!')atalosswhentheirordersbecomemispricedthanwhentheyarenot.Whatisthebehaviorofthemixbetweenmarketandlimitorders(&theorder#owcomposition')acrosssecurities?Surprisingly,thisquestionhasnotbeenaddressedyet(toourknowledge),neitherempirically,4northeoretically.Theobjectiveofthisarticleistodevelopasimplemodelinwhichthemixbetweenmarketandlimitorderscanbecharacterized,inequilibrium.Asexplainedbelow,inthisway,weobtaintestablepredictionsconcerningthecross-sectionalbehavioroftheorder#owcomposition.Furthermore,themodelhasnewtestableimplicationsforthecross-sectionalbehavioroftradingcostsinlimitordermarkets.Inordertoportray,inanaturalway,theexecutionriskandtheriskofbeingpickedo!,weconsideradynamicmodel.Tradersarrivesequentially.Uponarrival,atradercanchoosetopostquotes(placealimitorder)ortotradeatthequotespreviouslypostedbyothertraders(placeamarketorder).Executionoflimitordersisuncertainandtheassetvalue#uctuates,whichcreatesawinner'scurseproblemforlimitordertraders.Theoptimalchoicebetweenamarketandalimitorderandtheoptimalpricesforlimitordersdependontheorder3Alimitorderspeciesalimitpriceandaquantity.Forabuylimitorder,thelimitpriceisthemaximumpricethatabuyerwillpayandforaselllimitorder,thelimitpriceistheminimumpricethatasellerwillobtain.Marketordersareorderstobuyorsellagivenquantityatanyprice.Thoseordersarethemainchannelsthroughwhichliquidityissuppliedandconsumedinlimitordermarkets.Biaisetal.(1995)report(TableIII,p.1670)that,fortheParisBourse,47.2%ofallordersaremarketordersand41.3%arelimitorders.Theotherordersarecancellationsorapplications.4Biaisetal.(1995)andHedvallandNiemeyer(1996)focusonthevariationsintheorder#owduetotransientchangesinthestateofthelimitorderbook.HamaoandHasbrouck(1995)studythesupplyofliquiditywhenthereisnomarket-maker.Hamonetal.(1993),HandaandSchwartz(1996)andHarrisandHasbrouck(1996)focusonoptimalordersubmissionstrategies.DeJongetal.(1995)comparetradingcostsinalimitordermarketandadealermarket.FinallyaninterestingapproachisdevelopedbyAngel(1995)andHarris(1995)whoanalyzeoptimalorderplacementstrategiesindi!erentmarketconditions(stateofthebook,rateofarrivaloforders...)exogenouslyspecifyingtraders'beliefsontheirenvironmentandproceedingbysimulations.100T.Foucault/JournalofFinancialMarkets2(1999)99}134submissionchoicesofthefuturetraders.Solvingfortheequilibriumofthisgame,thetraders'orderplacementstrategiesarecharacterized,inclosedform,asafunctionoftraders'valuationsandthebesto!ersinthebook.Thissolutionhasmethodologicalinterest,independentoftheissuesweaddress.Actually,toourknowledge