Initial Public Offerings in Hot and Cold Markets

I n Piubtlic iOffaelrings in Hot and Cold Markets JeanHelwegeandNellieLiang FederalReserveBankofNewYork BoardofGovernorsoftheFederalReserveSystem Firstdraft:March1995 Thisdraft:August1996 Theanalysisandconclusionsofthispaperarethoseoftheauthorsanddonotindicate concurrencebyothermembersoftheresearchstaff,bytheBoardofGovernors,bytheFederal ReserveBankofNewYork,orbytheFederalReserveSystem. WearegratefulforhelpfulsuggestionsfromMichaelFleming,DavidIkenberry,FrankPacker, KatherineSpiess,andseminarparticipantsattheFederalReserveBankofNewYork. Kyle Nagel,PhilibertDeImus,KevinColeandKinChengprovidedexcellentresearchassistance.

Initial Public Offerings in Hot and Cold Markets ABSTRACT AsymmetricinformationmodelscharacterizehotIPOmarketsasperiodswhenbetter qualityfirmshaveanincentivetoissueequity,andcoldmarketsasperiodswhenthelemons premiumassociatedwithequityistoohightodrawinmanyissuers. Recentempiricalevidence, however,suggeststhatfirmsthatissueinhotmarketsarea majorsourceofstockprice underperformanceofequityissuers.WeinvestigatetheseopposingviewswithdataonIPOfirms thatissuedin 1983,ahotmarket,and1988,acoldmarket.Wefindthatthetwosetsoffirms havesimilaroperatingperformance,butstockreturnsareworseforfirmsthatwentpublicinthe hotmarket.Ourresultsarelargelyconsistentwithinvestoroveroptimisminhotmarkets,butnot withtheasymmetricinformationmodels. NellieLiang BoardofGovernorsoftheFederalReserveSystem DivisionofResearchandStatistics CapitalMarketsSection MailStop89 Washington,DC 20551 (202)452-2918 (202)452-3819FAX JeanHelwege CapitalMarketsDepartment 3rdFloor FederalReserveBankofNewYork 33LibertySt. NewYork,NY 10045 (212)720-5493 (212)720-1773FAX

Initial Public Offerings in Hot and Cold Markets Theinitialpublicoffering(IPO)marketfollowsacyclewithdramaticswings,often referredtoashotandcoldmarkets(e.g.,IbbotsonandJaffe(1975)andRitter(1984)).AhotIPO marketischaracterizedbyanunusuallyhighvolumeofofferings,severeunderpricing,frequent oversubscriptionofofferings,apreponderanceofsmallerissues,and,toacertainextent,by concentrationsinparticularindustries.lIncontrast,coldIPOmarketshavemuchlowerissuance, lessunderpricing,fewerinstancesofoversubscriptiona,ndlargerofferings.Someresearchers havearguedthattheseswingsintheIPOmarketreflectchangesininvestorsentiment,while othershavearguedtheyarisefromchangesinfactorsthataffectthedecisiontoissueequity,such asasymmetricinformationbetweeninvestorsandfirms. Thedecisiontoissueequityinanasymmetricinformationsettinghasbeenexaminedin signalingmodelsofIPOunderpricing(e.g.,AllenandFaulhaber(1989)),thedecisiontogo public(ChemmanurandFulghieri(1995)),andthetimingofthedecisiontocompleteaseasoned equityoffering(SEO)(e.g.,Choe,MasulisandNanda(1993),BaylessandChaplinsky(1996), andLucasandMcDonald(1990)).2 Innearlyallofthesemodels,thehotissuemarketsoccur whenbetterqualityfirmsarepulledintotheequitymarketaspricingbecomesmorefavorable. IntheIPOsignalingmodels,hotmarketstypicallyoccurwhencertainindustriesexperience positiveshockstoexpectedprofitabilitythatleadtoalowerlemonspremium. IntheSEO models,thecostsofasymmetricinformationarelowerinsomeperiodsduetobetterinvestment opportunities,lessasymmetricinformation,orrandompositivepricechanges. Theempiricalevidencedescribingissuersinhotandcoldmarketsismixed. Choeetal. andBaylessandChaplinskyprovideevidencethatannouncementeffectsarelessnegativeinhot markets,suggestingthatbetterqualityfirmsissueequityinhotmarkets.TestsoftheIPO IForexample,Ritter(1984)showsthatmostoftheunderpricinginthehotissuemarket of 1980-1981isattributabletounderpricingamongIPOSinthenaturalresourcessector. 2Therearemanymodelsofthedecisiontoissueequityortochangethedebtiequityratio. Wefocushereon modelswithdirectimplicationsforissuanceinhotandcoldmarkets. 1

signalingmodelsbyJegadeesh,Weinstein,andWelch(1993),MichaelyandShaw(1994)and SpiessandPettway(1995)castdoubtonthestrengthoftherelationshipbetweenunderpricing andfirmquality,suggestingthathotissuemarketsmaynotreflectgreaterparticipationbyhigh qualityIPOS.Morerecentevidenceonthepoorlong-termstockpriceperformanceofequity issuers(e.g.,SpiessandAffleck-Graves(1995),andhughran andRitter(1995))suggeststhat equityissuersarebelow-averagequalityfirms,especiallyiftheyissueequityinahotmarket. hughran andRitter,inparticular,arguethattheunderperformanceresultsareevidenceof investoroveroptimismandmanagers’willingnesstotakeadvantageofsuchsentiments. Inthispaper,weanalyzeIPOSthatwerecompletedintwoverydifferentmarkets,thehot IPOmarketof 1983andtherelativelycoldIPOmarketof 1988,toevaluatewhetherhotmarket issuersarehigherqualityfirmsaspredictedbytheasymmetricinformationmodels. While managers’viewsoftheirfirmsrelativetothemarket’saredifficulttomeasure,thisproblemis mitigatedbyourchoiceofdistinctmarketconditionssincemanagersrevealtheirbeliefsabout relativevaluationthroughthetimingoftheirIPOS.Thatis,asymmetricinformationmodels implythathotmarketissuersarefirmsthatconsiderthemselvestohavebetterprospectsthanthe firmsthatacceptedcoldmarketpricing.Weexaminethe 1983and1988IPOfirmsassuming theyaredrawnfromthesamevaluedistribution.Becausetheyareseparatedbyfiveyears,we alsoexaminea(small)setofIPOfirmsin 1982.Theseissuersmostcertainlyaredrawnfromthe samedistributionasthe 1983firms,differingmainlyinthatthe 1982issuersdidnotdelaytheir , offerings. WeconsiderIPOfirms’operatingperformanceandstockreturnsforuptofiveyearsafter theIPO,assumingthatfirmqualitywouldberevealedduringthisperiod. Wefindthatinthe yearoftheIPOthehotmarketfirmsof 1983arelessprofitableandsomewhatsmallerthanthe firmsthatwentpublicin 1988;nordotheyappeartohavegreatergrowthpotential. Intheyears followingtheIPO,bothsetsoffirmsexhibitsomedeteriorationinprofits,fromabnormallyhigh levelsintheyearoftheIPO,butneithersetunderperformsotherfirmsinthesameindustry. Indeed,operatingperformancesofthetwosetsoffirmsareindistinguishableN. evertheless,stock returnsforthehotmarketIPOfirmsarequitepoorforseveralyears,whereasreturnsforthecold marketIPOstocksaremorecomparabletothoseonNASDAQstocks.Wefindnoevidencethat 2

thehotmarketIPOfirmsaresuperiorfirms,astheasymmetricinformationmodelspredict. Nor dowefind,however,thatthehotmarketfirmsareinferior,astheunderperformanceliterature suggests.Wedofindevidenceconsistentwiththeviewthatexcessivelypositiveinvestor sentimentduringhotmarketsleadstolong-termunderperformance.Inparticular,thesametype offirmhaslowerstockreturnsifitcomespublicinahotmarket. Thepaperisorganizedasfollows:SectionIIisareviewofthetheoryandevidenceon IPOcycles.WedescribeourdatainSectionIIIandprovideevidencethatoursamplesare appropriateforstudyinghotandcoldIPOS.SectionW examinestheoperatingperformanceand stockreturnsofthetwosetsoffirms. BecausethetimingoftheIPOSinoursampleareseparated byfiveyears,itispossiblethatthegoodIPOfirmsin 1983aredrawnfromadifferent distributionthanthegoodIPOfirmsof 1988.Thus,inSectionVwecomparethe 1983hot marketIPOStoanothersetofcoldmarketIPOSthataremorelikelytobedrawnfromthesame distribution-firmsthatcamepublicinthecoldmarketof 1982. SectionVIisadiscussionof therelationshipbetweenstockmarketreturnsandoperatingperformanceinlightofvarying investorsentimentovertheIPOcycle. SectionVIIistheconclusion. II. IPO Cycles: Theory and Evidence Inthissectionwediscussasymmetricinformationtheoriesthatexplainwhythevolume ofequityissuancevariesovertime. ThesetheoriesincludemodelsofbothIPOSandSEOS becausetheempiricalliteratureonunderperformanceshowsthatbothtypesofequityofferings resultinpoorlong-runreturns.Wecontinuewithasummaryoftheempiricalevidencevis-a-vis thetheories’predictionsabouthotandcoldequitymarkets,aswellasempiricalresultsthat suggestmarketinefficiency. A. ModelsofEquityIssuanceinHotandColdMarkets TheIPOsignalingmodelsofunderpricing,suchasthoseofAllenandFaulhaber(1989), Welch(1989)andGrinblattandHwang(1989),showthatunderpricingequilibriacanoccur whenbetterfirmsfindunderpricingaviablewaytosignalhigherquality.Thesefirmshavean incentivetosignaltheirqualitysothattheirsubsequentequityofferingswillreceivemore favorablepricing. AllenandFaulhaberspecificallyaddresshotissuemarkets. Hotmarketsand 3

underpricingequilibria(whichareinfrequent)occurwhentherearepositiveshockstothe expectedprofitabilityoffirms. Sincefirmscansignaltheirquality,more highqualityfirmswill finditoptimaltoundertakeIPOS.Accordingly,hotissuemarketsareassociatedwithunusually highexpectedprofitsandtheyshouldlastonlyaslongasittakescompetitiontodrivedown profitsintheindustry. AnalternateviewofIPObehaviorisofferedbyChemmanurandFulghieri(1995),who focusonthevolumeofofferingsinhotmarketsratherthanonunderpricing.Theymodelthe decisiontogopublicorraiseprivateequityfromventurecapitalistsbasedonthedegreetowhich asymmetricinformationaffectspricingandthecostofreducinginformationproblems.Hot marketsoccurwhenproductivityshocksthatincreasethevalueofthefirminducemoreinvestors toproduceinformation.Thisinturnleadstoagreaterincentivetogopublic. Thus,they characterizehotmarketsasalargenumberofofferingsbyunusuallyprofitablefirms.3 Choe,MasulisandNanda(1993)presentamodelinwhichthedegreeofasymmetric informationassociatedwithSEOSvarieswiththebusinesscycle. Whentheeconomyisgrowing, allfirmsreceiveprojectswithhigherexpectedcashflows. Thisimpliesthatinanexpansion morefirmswillfinditoptimaltoissueequityeveninthefaceofthelemonspremium.In essence,asymmetricinformationbecomeslessofaprobleminanexpandingeconomy,giving betterqualityfirmsalargerincentivetoissueequity.4Adirectpredictionoftheirmodelisthat announcementeffectsforSEOSshouldbelessnegativeduringbusinesscycleupswingsbecause thoseperiodsleadtohigherinvestors’expectationsofthequalityoftheaverageissuer. BaylessandChaplinsky(1996)arguethathotissuemarketsneednotoccursolelybecause ofswingsinGDP. Thatis,awindowofopportunitycouldoccurinanyperiodinwhichthecost ofissuingequityislowerforallfirms,suchaswheneventsknownbybothmanagersand 3Thismodelcanalsogeneratehotmarketsthroughvariationinthecostofevaluating firms’prospects,buttheauthorsemphasizethechangeinprofitabilityasamorenatural explanationforhotmarkets. 4KorajczykL, ucasandMcDonald(1991)alsohaveamodeltoexplainthetimingof equityissuancethatisbasedontime-varyingasymmetricinformation,butitmostlyrefersto timingofindividualfirms’offeringsbasedonthetimelapsebetweenfirms’earnings announcements. 4

investorsdominatefirm-specificinformation.AsinChoeetal.,ahotmarketischaracterizedby asmallerlemonspremium,leadingtolessnegativeSEOannouncementeffectsonaverage. Thus,theytoopredictthathotmarketissuersareexpectedtobebetterqualityfirmsonaverage. FirmsintheLucasandMcDonald(1990)modelalsofaceasymmetricinformation problems,buttheymaydelaytheirequityofferingsiftheirstocksareundervalued.Incontrast, overvaluedfirmsprefertoissueimmediately,lestthestockpricefalls. LucasandMcDonald showthatalargenumberofrandompositivedrawsfromthepricingdistributionissufficientfor anincreaseinequityissuance--thereneednotbeanyfundamentalchangeinthefirms’ operationsorthedistributionofpricingerrorsforwavesofequityissuancetooccur. Their modelimpliesthatanupswinginthenumberofIPOSreflectsincentivesforagreaterfractionof betterqualityfirmstoraiseequity. B. EmpiricaEl videnceonHotandColdMarkets Theempiricalevidenceonthequalityoffirmsthatissueequityinhotandcoldmarketsis mixed. Choeetal.andBaylessandChaplinskyfindthatthestockmarketreactionsto announcementsofequityofferingsarelessnegativeinhotmarkets,astheirtheoriespredict. The twostudies’resultsdiffermainlyinhowtheydefinehotmarkets-Choeetal.analyzevariations instockreturnsaccordingtothebusinesscycle,whichisquitesimilartotheequityissuance(or IPO)cycle,whileBaylessandChaplinskydefinetheirmarketsbyvolumeofofferings.Thus, theirevidenceisconsistentwiththeasymmetricinformationtheories’viewthathotmarket issuersarebetterfirms. EmpiricaltestsofPO signalingmodelsarelesssupportiveoftheasymmetric informationview. Jegadeeshetal.findthat,whilesubsequentequityissuancepatternsare consistentwithsignalingtheories,underpricingdoesnotappeartoplayastrongsignalingrolefor IPOS.SpiessandPettwayshowthatthelossesinvolvedinunderpricingarenotjustifiedby subsequentstockpriceimprovementsforseasonedofferings.MichaelyandShawfindno evidenceinsupportofthesignalingtheoriesintheiranalysisofmasterlimitedpartnerships. InsharpcontrasttotheexaminationofannouncementeffectsofSEOS,studiesofthe long-runstockperformanceofIPOSandSEOSsuggeststhathotmarketequityissuersareinferior firms. Ritter(1991),SpiessandAffleck-Graves(1995)andLoughranandRitter(1995)findthat 5

thestocksofequityissuersseverelylagthereturnsonawidevarietyofbenchmarks,and hughran andRitternotethatreturnsaresharplyloweramongfirmsthatofferedequityinhot markets.Theyregresscross-sectionreturnsonanindicatorofahotmarketandfindanaverage returnthatis43basispointslowerthanthatforcoldmarkets.And,althoughhotmarkets typicallyinvolvegreaterinitialstockreturnsforIPOS,theyfindtheunderperformanceofIPOSis notoffsetbyhighinitialreturns.bughran andRitterconcludethathotmarketequityofferings aremadebylowerqualityfirmswhoknowinglytakeadvantageofinvestoroveroptimism. JainandKini(1994)andMikkelsonandShah(1994)donotspecificallystudyhot marketsincomparisontocoldmarkets,buttheirresultsbasedonaccountingdataindicatethat IPOfirmstimetheirofferingstocoincidewithpeakoperatingperformance.Bothstudiesfind evidenceofsharpdeclinesinoperatingperformanceoncepublic. JainandKinifindthat operatingROASandoperatingcashflowtoassetsfallbetweenthepre-IPOyearandeachofthe foursubsequentyearsexaminedthereafter.MikkelsonandShahreportsimilarresultsovera longersampleperiod-uptotenyearsaftertheIPO-andthatlowerstockreturnstendtooccurin firmswithsharperdeclinesinearnings.LoughranandRitter(1994)findsimilarlypooroperating performanceforSEOfirms. Theseresultsprovideindirectevidencethathotmarketequity issuersareworsefirmsthancoldmarketfirms. Teoh,Wong,andRao(1994)examinetheuseofdiscretionaryaccrualsinaccountingat thetimeoftheIPOandfindevidenceofmanagerialopportunism.Theyreportweakevidence thattheseaccrualsarehigheramonghotmarketIPOS. Cheng(1995)suggeststhatstockpriceunderperformanceisrelatedtothesubsequentuse ofSEOproceeds-firmsthatdonotspendthefundsoncapitalexpenditureshaveworselongtermperformance.Thisisalsosuggestiveofmanagerialopportunisminhotmarkets,although hedoesnotspecificallyexaminehotandcoldmarketissuers. Lemer(1994)alsoprovidesevidencethatfirmstakeadvantageofwindowsof opportunitytoissueequity. Inhisinvestigationofagroupofbiotechnologyfirmsthatreceived fundsfromventurecapitalists,heconcludesthattheirIPOSaretimedtocoincidewithpeaksin publicmarketprices. Althoughtheevidenceonstockunderperformancepointstohotmarketissuersasthe 6

mainculpritinthegroupofequityissuers,theindirectevidenceontherelationshipbetweenhot marketissuersandoperatingperformanceoftenfailstoconfirmsuchapessimisticviewofhot marketIPOS.Inparticular,JainandKinifindnorelationshipbetweenoperatingperformance andIPOunderpricing,whichishigherinhotmarkets;nordoMikkelsonandShahfinda particularlystrongrelationshipbetweenoperatingperformanceandtheyearinwhichthefirm wentpublic. OurstudyprovidesdirectevidencelinkingfirmqualityandIPOcycles. III. Data Ouranalysisofhotandcoldmarketscomparesfirmsthatwentpublicin 1983,whichwas anextraordinarilyhotyear,withfirmsthatcompletedIPOSin 1988,acoldmarketyearforIPOS. Wechoosethesetwoyearsbecausetheyarebothrecentenoughthatfinancialdataarebroadly available,butnotsorecentthatwecouldnotevaluatethefirmsoverfiveyears. Basedonthe numberofnonfinancialIPOSsince1980,showninTable1,thesetwoyearsseemappropriate. Thehighestissuancesince1980wasin 1983.Issuanceremainedmoderatetoheavythrough 1987.In 1988-1990,thenumberofIPOSeachyeardroppedoffsharply.Ofthesecoldmarket years,1988wastheoneforwhichfiveyearsofdataonoperatingresultswereavailable.5 Additionally,the 1988market,whilecold,providesaslightlylargersampleforanalysisthanthe 1989market.Wealsonotethatbughran andRitter(1995)describetheperiodbetweenthe crashofOctober1987andtheGulfWarvictoryinFebruary1991asaperiodoflowequity issuance.Moreover,thesetwoyearsarecloseenoughchronologicallythattheIPOSarelikelyto bedrawnfromthesamequalitydistributions. Oursamplesof 1983and1988IPOSincludeallnonfinancialfirmsthatcompletedIPOSin thoseyearsaccordingtoSecuritiesDataCompany(SDC).Werestrictouranalysistofirmcommitmentofferingsof$2.5millionormorethatarenotreverseleveragedbuyouts,spinoffs,or 51ncontrast,BaylessandChaplinskycountApriltoSeptember1988amongtheirhot marketperiods.Theyclassifytheirperiodsaccordingtotherankingofthemonthly issuance listedintheFederalReserveBoard’sAnnualStatisticalDigest. TheDigest recordsthedollar volumeofcommonandpreferredequityraisedbyallU.S.firms. Theequityraisedin 1988 includesalargeamountofofferingsbyclosed-endinvestmentcompanies,withoutwhichthe periodwouldhavebeencharacterizedascold. 7

unitofferings.Inaddition,weonlyexaminefirmswhoseofferingsharepriceexceeded$1. FinancialdatawereobtainedfromCompustatformostfirms,supplementedbydatafrom CompactDisclosurewhenCompustatdatawereunavailable. StockpricedatawereobtainedfromtheCenterforResearchonSecurityPrices(CRSP). SeveralofthefirmsidentifiedintheSDCdatasetarenotfollowedbyCRSP. Dataforothersare questionable-thefirstpriceobservationismonthsaftertheIPOorweeksbefore;thefirsttrading priceislessthan$1eventhoughSDCindicatesanofferingpricecloserto$5. Aftereliminating theseobservations,thesamplesfor1983and1988included263and84observations, respectively. Followinghughran andRitter,wecalculatereturnsfromtwoweeksaftertheIPOto morecloselyreplicatereturnstoatypicalinvestorwhomaynotbeallocatedsharesintheIPO. Returnsarecalculatedasthepercentagedifferenceinthepriceontwodays,adjustedforsplits anddistributions.OurfocusisonreturnsfromtwoweeksaftertheIPOdateuntilone,threeand fiveyearslater. Thereturnsdistributionsarehighlyskewed,soweavoidt-testsofmeans. Instead,ouranalysisfocusesonthewealthrelative,asinLoughranandRitter. ThedescriptivestatisticsinTable2showthattraditionalindicatorsofahotmarkethold forthe 1983sample,butarenotpresentinthe1988sample.Underpricingaveraged14.6percent forthe 1983IPOSandonly6.6percentforthe 1988sample.IPOproceedsweresmallerinthe earlieryear,reflectingthesmalleraveragesizeofthe1983firms. Thesedifferencesinsizeare notattributablesolelytoinflation.GLastly,suggestiveofmorefrequentoversubscription,the 1983sampleraisedproportionatelymorefundsintheirIPOS. Table3provideslong-termstockperformancemeasuresforthe 1983and 1988samples. TocalculatethestockreturnsinTable3,wekeepfirmsthatdelistinthesampleforaslongas theytrade. Foreachstockthatdoesnotincludeafullobservationperiod(1,3or5years),we includeanobservationforthesameperiodfortheNASDAQbenchmark.Forexample,anIPO completedinJanuarythattradesuntilAprilwillhaveaNASDAQobservationfromJanuaryto April. Asinpreviousresearchonlong-termunderperforrnancet,hehotmarketIPOstockshave ‘TheCPIshowsanincreaseinthepricelevelofabout25percentoverthisperiod. 8

muchlowerwealthrelativesthanthecoldmarketIPOS,measuredbyreturnsfor 1,3,and5years. The1988groupoftendoesatleastaswellastheNASDAQ.ConsideringtheriskoftheIPOS, however,theexcessreturnovertheNASDAQreturnmaynotbesufficientforthe 1988IPOS. Wenotethatreturnsforthisgrouparequitestronginyearfive,butextremelysensitivetothe inclusionofonestock-AmericanPowerConversionCorporation(APCC).WithoutAPCC,the wealthrelativeforthe 1988sampledropsto.91. Thesedataindicate,contrarytotheasymmetric informationtheories,thathotmarketIPOfirmsareinferiorcompaniestothosethatgopublicin coldmarkets. IV. Relative Performance of Hot and Cold Market IPO Firms Webeginbydescribingthetypesoffirmsthatgopublicineachmarketusinginformation onthedistributionoffirmsbyindustryandaccountingdatafromtheyearoftheIPO. Wenext considerthepost-IPOperformanceofthehotandcoldmarket~0 firms. Ifasymmetric informationtheoryiscorrect,thefirmsthatgopublicinhotyearsshouldeventuallyprovetobe moresuccessfulfirmsthanthecoldmarketfirmsthatdidnotfeeltheywereundervalued.We alsocomparethisaccountingdatatostockreturnsoversimilartimeperiodstodetermineifstock pricesarereflectingadropinperformanceasthemanagerialopportunismtheorysuggests. Lastly,werepeattheanalysisaftermakinganadjustmentforsurvivorshipbias. A. FirmCharacteristicsintheIPOYear AsymmetricinformationtheoriesofIPOSoftendescribehotmarketsasperiodswhen productivityshocksortechnologicaladvancesoccurinaparticularindustry,suggestingagreater industryconcentrationinhotmarkets.Table4presentstheindustrydistributionsofthe 1983and 1988IPOfirmsaccordingto2-digitSICcodes.Thedistributionbyindustryofthe 1983firmsis extraordinarilyconcentrated:fourindustriesaccountformorethanhalfofthesample. The1988 sample,however,isonlyslightlymoredisperse:forexample,whilethetopfiveSICcategories represent57percentofthe 1983sample,theyaccountfor42percentofthesamplein 1988;the topeightindustriesrepresent67percentofthe 1983firms,comparedto57percentofthe 1988 sample.Whilethesedatasuggestthathotmarketsarerelatedtoindustry-specificshocks, consistentwiththeasymmetricinformationmodels,thecoldmarketindustrydistributionisalso 9

fairlyconcentrated. kss compellingforthesemodelsisthefactthatthefivemostfrequentindustriesare identicalinboththehotandcoldmarkets.Thatis,firmsincomputersoftware(SIC73)and hardware(SICS35and36),advancedmedicalequipment(SIC38),andbiotechnology(SIC28) accountedforalargefractionoftheofferingsinboth1983and1988.AX2testthattheindustry distributionsarehomogeneouisnthetwoyearscannotberejectedatstandardlevelsof significance. Thesimilarityofthetwosamples’industrydistributionsmakesiteasiertobenchmarkthe expectedperformanceofthehotmarketfirms. Ifhotmarketissuersarealikeinallrespectsto coldmarketissuersexceptinthepricingoftheirstock(suchasintheLucasandMcDonald model)andthedistributionoffirmsisunchangedinthefiveyearsbetween1983and 1988,then wecansafelyassumethatthehotmarketfirmsof 1983consideredthemselvestobebetter qualityfirmsthanthecoldmarketfirmsof 1988.Otherwise,theywouldhavebeenfoundthe offeringpricesofcoldmarketssuchas 1988and1982acceptableandwouldhavegonepublic earlier. Next,weconsiderwhetherthefinancialcharacteristicsofhotandcoldmarketIPOfirms aredifferentaroundthetimeoftheoffering.Table5presentsstatisticsforthetwosetsoffirms attheendoftheirIPOyears. Wefocusonmediansbecausethedistributionsofthesemeasures areoftenskewed.Moreover,boththeoriesofIPObehaviorrefertothetypicalfirminahot marketcomparedtothatinacoldmarket.7Tocontrolfordifferencesthatarisebecausehotand coldmarketstypicallyoccuratdifferentpointsinthebusinesscycle,wealsocomparetheIPOS tootherfirmsintheirindustries.Industry-adjustedmeasurescomparefirmvaluestothemedian valueforfirmsinthesamefour-digitSICcategoryifthefour-digitindustrysamplehadmore than15firmsduringthefiveyearsfollowingtheIPOyear. Ifthefour-digitSICcategorydidnot haveasufficientsamplesize,thefirmwasmatchedtothethree-ortwo-digitSICcategory. Firmsthatissuedin 1983weresmallerthanfirmsthatissuedin 1988,andconsiderably 71ncontrast,analysisoflong-termstockperformanceclearlyrequirestheuseofmeansto calculatewealthrelatives.Thiscreatesanunfortunatebarrierinthelinkageofoperating performancetostockperformance. 10

smallerthantheirindustrycounterparts.The 1988IPOfirmsarealsosmallerthanotherfirmsin theindustry,sothatthetwosamples’sizesarenotsignificantlydifferentonanindustry-adjusted basis. ThisevidenceprovidesslightsupportforChemmanurandFulghieri’sviewthathot marketsarecharacterizedbyfirmsthatchoosetogopublicearlier,presumablywhentheyare smaller.ThedifferencesinsizedonotsupporttheLucasandMcDonaldviewthatfirmswaitfor apricerun-upbeforeissuingequity,sincefirmsinhotmarketsshouldbelargerbecausethey grewwhilewaitingtogopublic. Operatingincome(earningsbeforeinterest,taxes,depreciationandamortization),scaled eitherbyassetsorsales,issomewhatlowerforthehotmarketfirms,butonlyincomescaledby assetsdifferssignificantlyacrossthesamples.Industry-adjustedmeasuresofprofitabilityare significantlylowerforthehotmarketfirms. Thisevidencecontradictstheasymmetric informationmodels,whichpredictthathotmarketissuerswillbebetterfirms. Itispossiblethatthesehotmarketfirms,thoughsmallerandlessprofitable,aresuperior inthattheyhavegreatergrowthpotentialthancoldmarketfirms. However,investment opportunities,asmeasuredbytheratioofcapitalexpenditures-to-assetsa,resimilarinthetwo setsofIPOfirms.Industry-adjustedcapitalexpendituresaresignificantlypositive,butdonot differbetweenthehotandcoldmarketfirms.Likewise,thedegreeoffixedassets,often interpretedastheinverseofgrowthopportunities,doesnotdiffersignificantlybetweenthetwo samples.Thefractionoffirmspayingdividendsissignificantlyhigherinthe 1988sample, althoughthemedianforbothsamplesiszero. Thesimilaritiesingrowthopportunitiesforthe twosetsoffirmsarecontrarytothepredictionoftheasymmetricinformationmodels,suchas Choeetal.,inwhichthegreaterneedforinvestmentfundsspursequityissuanceinhotmarkets. TheseresultsalsofailtotieCheng’sresultsonlowcapitalexpendituresandlong-term underperformancetohotmarkets. Debt-to-assetratiosdonotindicatesubstantialdifferencesbetweenthetwosamplesof firms,asbothsetsofIPOShaveverymodestdebt-to-assetratios. B. PerformanceinSubsequentYears Asymmetricinformationtheorysuggeststhatthequalityofhotandcoldmarketfirmsis difficulttodifferentiateatthetimeoftheofferingexceptthroughtheunderpricingsignal. 11

However,acornerstoneofthistheoryisthattheinformationaboutthefirms’qualityeventually becomespublicknowledge,makingthesignalingbehaviorworthwhile.Thus,ifasymmetric informationdrivestheIPOtimingdecision,hotmarketIPOfirmsshouldeventuallyoutperform thecoldmarketfirms. ToanalyzegrowthpatternsaftertheIPOwelookattwomeasures:first, weexaminetheindustry-adjustedmediansofthehotandcoldmarketfirmsforuptofiveyears aftertheirrespectiveIPOS;second,weinvestigatethemedianindustry-adjustedyear-over-year growthratesforthetwogroups. Table6showsmediansofoperatingmeasuresforone,threeandfiveyearsaftertheIPO. RatherthangreaterdeteriorationamongthehotmarketIPOS,thedifferencesbetweenthetwo setsoffirmsarelessmarkedovertime. ProfitratiosintheyearsaftertheIPOreveallittle differencebetweentheIPOfirmsandotherfirmsinthesameindustry,andlittledifference betweenhotandcoldmarketfirms. Onlyforthe 1983IPOfirmsinyearoneisthemedianratio ofoperatingincome-to-assetssignificantlybelowthatofthetypicalfirmintheindustryand belowthatforthe 1988firms. Likewise,thefiguresforcapitalexpendituresdonotdiffersignificantlybetweenthetwo setsoffirms,asbothsamplesmostlyoutspendtheirindustrycompetitors. Onlysalesdiffersignificantlybetweenthehotandcoldmarketfirms:the 1988firms growtobelargerthantheirtypicalcompetitorsbythefifthyearaftertheIPO,makingthem significantlylargerthanthe 1983IPOSbythen. Table7presentsanalternativewayofexaminingthepost-~0 performance.The mediansofyear-over-yeargrowthratesinthistablealsoindicatethatthetwosetsofIPOfirms arequitesimilar.Changesinprofitmarginsdonotdiffersharplybetweenthem;nordoessales growthdivergesignificantly.Capitalexpendituresgrowataboutthesamepaceforhotandcold marketIPOfirms,adjustedforindustrygrowth. Overall,theseresultsonoperatingperformanceaftertheIPOcontradicttheasymmetric informationmodels:WefindlittleevidencethathotmarketIPOfirmseventuallyreveal themselvestobeofhigherqualitythancoldmarketIPOfirms. Post-IPOperformancealso allowsustoexaminethemanagerialopportunismviewputforthbyJainandKiniandLoughran andRitter,whoarguethathotmarketIPOStimetheirofferingstotakeadvantageoftheirbest 12

accountingreports,makingtheIPOyearmorelikelytobethemostprofitable.Thus,managerial opportunismimpliesthathotmarketissuerswillexperiencegreaterdeteriorationinprofitability aftertheIPO. Whilethedeclineinprofitmarginsinyearoneprovidesomesupportforthe notionthatopportunisticmanagerstaketheircompaniespublicintheirpeakyearofperformance, profitmarginsaremaintainedatlevelssimilartotheirindustrycounterpartsforfiveyearsafter theIPO. Further,thereisnoevidencethatprofitmarginsforhotmarketIPOSfallfasterthanthe coldmarketIPOS’. Howdidthestockmarketviewthesefirms’prospects?Table8showswealthrelatives forthe1983and1988firmsfortimeperiodsthataresimilartothoseshowninthetables6and7 thatuseaccountingdata. Thecalculationofthesereturnsassumesthattheportfoliosare rebalancedyearly-firmsthatdelistpriortothebeginningofayeararenotincludedinthat period’sreturns,just asfirmswithoutCompustatdatainyearsthreeandfiveareexcludedfrom theanalysisinTables6and7forthoseyears. ThewealthrelativesinTable8showthatbothhot andcoldIPOSunderperformthemarketinthefirstfullcalendaryearaftergoingpublic,andthe wealthrelativeisquiteabitlowerforthe 1983firms. BothsetsofIPOfirmsmatchthe NASDAQperformanceinsubsequentyears,butthe 1988IPOShavemuchhigherwealth relatives.Thisowespartlytotheperformanceofonefirm-APCC-thathasreturnsinexcessof 1000percentinyearsfourandfive. WithoutAPCC,thehotandcoldmarketPOS’ wealth relativesarequitesimilarinyearfive. Whileaccountingmeasuresmaynotexactlyafirm’svalue,thestockpriceresultsin Table8largelyagreewiththeoperatingresultspresentedinTable5,6 and7:thegreatersales andprofitmarginsforthe 1988firmsinyearonemightjustifysomewhatgreaterwealthrelatives inthefirstcoupleofyears. Thelackofdistinctioninprofitmarginsbetweenthehotandcold marketfirmsinlateryearsisconsistentwithreduceddifferentiationinstockreturns. Thesestock priceresults,however,donotsupporttheasymmetricinformationmodels,asthesemodels predictsuperiororequalperformance(ifsignalingiscomplete)ofhotmarketissuersrelativeto IPOfirmsincoldmarkets. C. Post-IPOperformanceadjustedforsurvivorshipbias ThestockreturnsinTable8arequitedifferentfromthosepresentedinTable3,which 13

showunderperformanceonaverageforbothsamplesandparticularlypoorreturnsforthehot marketIPOS.Thereareseveraldifferencesbetweenthecalculationsofthewealthrelativesinthe twotables,butthemainoneisthatthereturnsinTable8reflectonlyreturnsoffirmsthat survived. FirmsdropoutoftheCompustatandCRSPsamplesbecauseofbankruptciesand mergers.Ifthenonsurvivorsareevenlysplitbetweenunderperformersandsuccessfulfirms,the resultsshownintheprevioustablewillbeunbiased.DelistingreasonsonCompustatsuggest thatroughlyhalfoftheattritionineachsampleowestobankruptciesandhalftomergers, suggestingthedataareunbiased,butthedifferenceinstockreturnsbetweenTables3and8 suggestotherwise. Tocorrectforpossiblesurvivorshipbias,weinvestigatethetwosetsofIPOfirmswhile keepingincompletedataintheanalysis.Foroperatingperformance,weincludethelatest operatingresultsforfirmsthatexitthesampleprematurely.Thus,forexample,themediansales growthfigureforyearfivewillbebasedmainlyongrowthfromyearfourtofive,butwillalsobe basedondatafornonsurvivingfirmsfromearlieryears. Tables9and10presentmeasuresofoperatingperformanceforthetwoIPOsamples whendatafornonsurvivorsareusedinthecalculations,Boththehotandcoldmarketfirmslook marginallyworseinTable9whennonsurvivorsareusedinthecalculations.Nevertheless,most oftheprofitmeasuresremaininsignificantlydifferentfromthemedianoftheindustryandthe adjustmentfailstodifferentiatethesamples.Whileindustry-adjustedsalesfallslightlyforboth setsoffirms,thenonsurvivorsinthe 1983samplearemuchsmaller,makingthesamples significantlydifferentbysize. GrowthratesinTable10alsoaresimilartothosereportedin Table7. Thegrowthrateofcapitalexpendituresappearstobebelowtheindustryaveragelater inthesampleperiodforbothsetsoffirms. Wenote,however,thatbothsetsoffirmsinvested muchmorethantheirindustrycounterpartsintheearlyyearsoftheIPO(seeTables5and6). Stockreturnsforthecalendarperiodscomparabletotheaccountingdataperiodsforthese firmsarepresentedinTable11. Thestockreturns,similartothoseinTable3exceptforthetime periods,indicatesevereunderperforrnanceamongthe 1983IPOS,whilethe 1988IPOSneverdo asbadlyandactuallysharplyexceedthebenchmarkreturnsiftheinvestorisluckyenoughto haveAPCCinhisportfolio.Althoughadjustingforsurvivorshipbiasaffectsthewealthrelatives 14

sharply,thisadjustmenthaslittleeffectontheresultsonoperatingperformance,thussevering thelinksbetweenoperatingperformanceandstockreturns. Neitherthestockreturndatanorthe accountingdatasuggestthathotmarketIPOSaresuperiorfirms,astheasymmetricinformation modelspredict. However,thereisalsolittleinformationintheseaccountingdatathatsuggests hotmarketIPOfirmsaremuchworse,asLoughranandRitter’sstockpricedataimply.8 V. Comparison of 1983IPOSwith 1982 Cold Market IPOS Wehavecomparedthe 1983hotmarketfirmswiththe 1988coldmarketfirmsonthe assumptionthatfirmsinbothyearsaredrawnfromthesamequalitydistribution.Thatis,we assumethatthehotmarketIPOSof 1983passeduptheopportunitytogopublicin 1982(acold market)becausetheywouldhavebeenpricedascoldmarketfirms,reflectingthequality observedin 1982and1988.Itispossible,however,thatIPOfirmsin 1988maydifferfromthose intheearly1980sforreasonsunrelatedtohotandcoldmarkets.Thus,inthissection,we comparetheoperatingperformanceofthe 1983IPOfirmstothe 1982firms. Theasymmetric informationmodelswouldpredictthatfirmsthatcompletedIPOSin 1982consideredthemselves tobefairlyvaluedorovervaluedinthatmarket,whilefirmsthatwentpublicin 1983waitedfor morefavorablepricingorhadapositiveshocktoexpectedprofitability. Aswedidforthe 1983and1988samples,weidentifiedIPOScompletedin 1982using datafromSDCandeliminatedofferingsthatweretoosmallorhadofferpricesof$1orless. Reflectingthecoldmarket,the 1982samplehadonly43firms.9 Simi]artothe 1983and1988 samples,morefirmswereinthecomputersoftwareindustry(SIC73)thaninanyother,and computerhardware(SIC35)andelectroniccomponents(SIC36)eachaccountedforabout10 percentofthefirms. 8BecauseunderpricingismoresevereforhotmarketIPOS,itispossiblethattheirinitial returnsoffsettheirsubsequentpoorreturnstomakethemcomparabletocoldmarketIPOS,orthe NASDAQ.Weinvestigatedthispossibilityandfoundthatthewealthrelativesforthe 1983 basedontheofferpricewerestillbelowoneandstilllowerthanthe 1988firms’wealthrelatives basedonofferprices. ‘Compustatdatawereavailablefor60percentofthenonfinancialfirms(excluding spinoffs)thatcompletedIPOSin 1982,roughlythesameasthe62percentofIPOfirmsin 1983. 15

Table12presentsoperatingperformancemeasuresforthe 1982and1983firms. Onan industry-adjustedbasis,the 1982and1983firmshaveaboutequivalentsalesandcapital expenditures-to-assetrsatiosatthetimetheygopublic. However,aswiththe 1988firms,the 1982firmsaresubstantiallymoreprofitablethanfirmsthatwentpublicin 1983.Thisevidence isatoddswiththeasymmetricinformationviewthatbetterqualityfirmswaittoissue. NordowefindintheyearsfollowingtheIPOthatthe 1983firmseventuallyoutperform thecoldmarketfirms. InthefirstyearaftertheIPO,operatingmarginsforthe 1982firmsfall,as theydidfortheothersamples,butremainhigherthanmarginsforthe 1983firms. Bythefifth year,theymaybesomewhatmoreprofitable,butmostlyperformanceisindistinguishable. Particularly,salesgrowthandcapitalexpendituresforthe 1982and1983firmstrackrather closelyouttoyearfive. Thus,ourexaminationofasetofIPOfirmsthatarechronologically closertothehotmarketissuersin 1983indicatethatthereisnosupportfortheasymmetric informationmodelpredictionthathotmarketissuerswaitedformorefavorablepricingbecause ofsuperiorquality. VI. Investor Sentiment and Stock Returns ThewealthrelativesofhotmarketIPOSfallshortofthecomparablefiguresforthecold marketIPOSbymostmeasures,yetourevidenceonoperatingresultsdonotindicatesuchstrong differencesbetweenthetwosetsoffirms. Inthissectionweinvestigatewhetherinvestor sentimentcanexplainthisapparentdisconnectionbetweenstockreturnsandoperating performance.Todoso,weformportfoliosofstocksbasedonverysimilaroperating performance,calculatethewealthrelativesoftheseportfolios,andexaminewhethertheydiffer forthe 1983and1988samplesofIPOfirms. Inparticular,wepartitionthecombinedgroupof 1983and1988IPOfirmsintoquartiles basedonalternativeperformancemeasures--industry-adjustedsalesgrowth,profitgrowth,and growthincapitalexpenditures.Basedonthevaluesforeachmeasureineachtimeperiodthat definesthetopandbottomquartilesforthecombinedgroup,weformportfoliosofthebestand worstperformingfirmsinthe 1983and1988samplesseparately,andreporttheirwealthrelatives inTable13. Ifinvestorsentimentisthesameforgoodfirmsinbothhotandcoldmarkets,then 16

thewealthrelativesfortheseportfoliosofsimilarfirmsshouldnotvarywiththeyearinwhich theywentpublic. Ifhotmarketfirmsarevaluedmorethancoldmarketfirmsofsimilarquality atissuance,thenthelongtermperformanceshouldbeworseforthehotmarketfirmsastheir priceseventuallydeclinetothoseofcomparablefirmsthatwentpublicincoldyears. Wenote thattherearenosignificantdifferencesintheproportionsthateachIPOsamplecontributestothe bestandworstgroups-itisnotthecasethatthebestIPOScontainmore1988firmsnorthatthe worstgroupshaveadisproportionatenumberof 1983IPOfirms. ThewealthrelativesforthebestandworstperformingfirmsinTable13indicateastrong positivecorrelationbetweenstockreturnsandgrowthinprofitability.Thefirmsthatwereinthe topquartileforprofitmargingrowthendedthefirstcalendaryearwithwealthrelativesof.95and 1.17forthe 1983and1988samples,respectively,whilethoseinthelowestquartileonlyreported wealthrelativesof.56and.81,respectively.Firmswiththelowestlong-termprofitmargins (yearsthreeandfive)hadextremelylowwealthrelatives.Profitsscaledbyassetswerealsoa strongindicationofstockreturns-wealthrelativesforthefirmsinthehighestquartileofprofitsto-assetswereoftenmorethandoublethoseofthefirmsinthelowestquartile. Similarly,firms inthetopcategoriesforinvestmentgrowthhadhigherwealthrelativesthantheircounterpartsin thelowestquartile.Wedonotfind,however,thathighsalesgrowthisassociatedwithhigh stockreturns. Acomparisonofthewealthrelativesoffirmsthatwentpublicin 1983withthosethat wentpublicin 1988,showninTable13,indicatesthatinvestorsentimentdiffersgreatlyacross thetwoperiods. Portfoliosoffirmsbasedonthesameminimumlevelofprofitgrowthshowthat the1988IPOfirmshadmuchhigherwealthrelativesthanhighlyprofitablefirmsthatwent publicin 1983.Amongtheworstperformingfirms,differencesinpricingalsotendtofavorthe coldmarketfirms,althoughthedifferencesbasedonoperatingincome-to-assetsareslight. Similarly,firmswithveryrapidgrowthincapitalexpenditureshadhigherwealthrelative iftheycamepublicin 1988-IPOfirmswithequallyhighinvestmentsthatwentpublicin 1983 typicallyhad lowerstockreturns.Amongthelowinvestmentportfolios,thestockreturnsofthe 1983IPOSwereinferior. Althoughstrongsalesdoesnotappeartobehighlyvaluedinallperiods,evenamongthe 17

portfoliosbasedonsalesgrowth,thecoldmarketfirmsofsimilarcaliberalwayshavesuperior stockreturns. Theseresultssuggestthatinvestorsentimentinhotmarketsmayhaveledtopricesin 1983thatwerebiduptoohigh. Forotherwisecomparableperformanceintheyearsfollowing theIPO,higherpriceswouldleadtolowerwealthrelativesasinvestorscooledofftothehot marketinvestments.Sucharelationshipbetweenbid-uppricesandlowerreturnsisalso documentedforprivateventurecapitalinvestmentsintheearly1980sthatledtoverypoorlongrunreturns(Fenn,LiangandProwse(1995)).Theseresultsarealsoreminiscentofdifferences betweenvalueandglamourstockspointedoutbyLakonishok,ShleiferandVishney(1994). V. Conclusions InthisstudyweexaminethequalityofIPOfirmsinhotandcoldmarketstodetermineif hotequitymarketsdrawinbetterfirms,assuggestedbyIPOsignalingmodels,orifhotmarkets arecharacterizedbymanagerialopportunismcombinedwithinvestorirrationality,asLoughran andRittersuggest.Usingasampleof 1983hotmarketIPOSandasampleof 1988coldmarket IPOS,weinvestigatethecharacteristicsofthefirmsaroundthetimeoftheIPOandtheir subsequentprofitabilityandgrowth.Then,welinkstockreturnsofthesefirmstotheiroperating performance. AtthetimeoftheIPO,thehotmarketfirmsappeartobelessprofitablethancoldmarket firms,andhavelowersales,butnotwhensalesareadjustedfortypicallevelsintheindustry. Othercharacteristics,suchasthedistributionoffirmsbyindustry,investmentopportunitiesand leverage,aresimilarforthehotandcoldmarketfirmsintheIPOyear. Thisevidenceis inconsistentwiththeasymmetricinformationtheoriesasthereisnoindicationthathotmarket IPOfirmsaresuperior.Furthermore,the 1983IPOfirms’stocksunderperformtheNASDAQ overthefirstyearoftrading,althoughtheytypicallyhavehigherprofitsandmoreinvestment opportunitiesthanotherfirmsintheirindustries. DataonoperatingperformanceinthefiveyearsaftertheIPOprovidescantevidenceof differencesinfirmqualityovertheIPOcycle. Formostofthesampleperiodthereisno significantdifferenceinoperatingincometosalesorassetbetweenthehotandcoldmarket 18

issuers.Theonlysignificantdifferencebetweenthetwosetsoffirmsistheirsize-bytheendof thefifthyearthecoldmarketfirmsaresignificantlylarger. Noristheremuchevidencethatthese firmsareworsethantheirindustrycounterpartsinthepost-IPOyears. Thelackofdistinction betweenhotandcoldmarketissuersisreinforcedbyourinvestigationoffirmsthatcompleted ~Os inthecoldmarketof 1982. BecausemanyIPOfirmsdropoutofthesampleweexaminewhetherdifferencesin survivorshipratescanexplainourresults. Nonsurvivorsdonotappreciablyaffectthe distributionofprofitability,growthorinvestmentmeasures,althoughthe 1983nonsurvivorsare especiallysmall. Stockreturns,however,aresubstantiallyworseforthe 1983firmswhen nonsurvivorsareincludedinthesample. Thefactthatnonsurvivorsreducethewealthrelativesofthe 1983samplemuchmore thantheyreduceprofitabilitysuggeststhatfirmswithidenticalcharacteristicsmayhave experiencedwildlydivergentstockreturnsdependingonwhentheycamepublic. Wefurther investigatethisissuebyexaminingthebestandworstperformingfirmsinthetwosamplesand findthatsimilarfirmshavemuchhigherwealthrelativesiftheygopublicinacoldmarket. This mayreflectexcessivelyhighpricesatthetimeoftheIPO,suchthatsubsequentreturnsreflect correctionstothesebid-upprices. Ourevidenceshowsthatthelong-runoperatingperformanceofhotandcoldmarket issuersdoesnotdiffersubstantiallyinthefiveyearsfollowingtheIPO. Thus,wefindno evidenceinfavoroftheasymmetricinformationtheoriesofequityissuancecyclesthathot marketsarecharacterizedbybetterfirms. Atthesametime,ourevidencedoesnotsupportthe viewofthestockpriceunderperformanceliteraturethathotmarketissuersarelowqualityfirms withopportunisticmanagers.Wefindsomeevidencethatpricingisaffectedbyoveroptimismin hotmarkets-thewealthrelativesforfirmsofsimilarcaliberaretypicallylowerforhotmarket firms,suggestingadeclinefromunsustainablehighpricesimmediatelyfollowingtheIPO. 19

REFERENCES Allen,F.AndG.Faulhaber(1989)“SignalingbyUnderpricingintheIPOMarket,”Journal of Financial Economics, 23,303-323. Bayless,M.AndS.Chaplinsky(1996)“IsThereaWindowofOpportunityforSeasonedEquity Issuance?,”Journal ofFinance, 50,253-278. Chemmanur,T.andP.Fulghieri(1995)“ATheoryoftheGoing-PublicDecision,”unpublished paper,ColumbiaUniversityandINSEAD. Cheng,L.(1995)“EquityIssueUnder-PerformanceandtheTimingofSecurityOfferings,” unpublishedpaper,MIT. Choe,H.,R.Masulis,andV.Nanda(1993)“CommonStockOfferingsacrosstheBusiness Cycle,”Journal ofEmpirical Finance, 1,1-29. Fenn,G.,N.Liang,andS.Prowse(1995)“TheEconomicsofthePrivateEquityMarket,”Staff StudyoftheBoardofGovernorsoftheFederalReserveSystem#168. Grinblatt,M.andC.Hwang(1989)“SignalingandthePricingofNewIssues,”Journal of Finance, 44,393-420. Ibbotson,R.andJ.Jaffe(1975)“HotIssueMarkets,”Journal ofFinance, 30, 1027-1042. Jain,B.andO.Kini(1994)“ThePost-IssueOperatingPerformanceofIPOFirms,”Journal of Finance, 49, 1699-1726. Jegadeesh,N.,M.Weinstein,andI.Welch,“AnEmpiricalInvestigationofIPOReturnsand SubsequentEquityOfferings,”Journal ofFinancial Economics, 34, 153-175. Korajczyk,R., D.Lucas, andR.McDonald(1991)“TheEffectofInformationReleasesonthe PricingandTimingofSecurityIssues,”Review ofFinancial Studies, 4,685-708. Lakonishok,J.,A.ShleiferandR.Vishney(1994)“ContrarianInvestment,extrapolation,and Risk,”Journal ofFinance, 49, 1541-1578. Lemer,J.(1994)“VentureCapitalistsandtheDecisiontoGoPublic,”Journal ofFinancial Economics, 35,293-316. Loughran,T.andJ.Ritter(1994)“TheOperatingPerformanceofFirmsConductingSeasoned EquityOfferings,”unpublishedpaper,UniversityofIowaandUniversityofIllinoisat Urbana-Champaign. 20

(1995)“TheNewIssuesPuzzle,”Journal ofFinance, 50,23-51. Lucas,D.andR.McDonald(1990)“EquityIssuesandStockPriceDynamics,”Journal of Finance, 45, 1019-1043. Michaely,R.andW.Shaw(1994)“ThePricingofInitialPublicOfferings:TestsofAdverse- SelectionandSignalingTheories,”TheReview ofFinancial Studies, 7,279-319. Mikkelson,W.andK.Shah(1994)“PerformanceofCompaniesAroundInitialPublic Offerings,”unpublishedpaper,UniversityofOregonandAucklandUniversity. Ritter,J.(1984)“The‘HotIssue’Marketof 1980,”Journal ofBusiness, 57,215-240. (1991)“Thehng-Run PerformanceofInitialPublicOfferings,”Journal ofFinance, 46,3-27. Spiess,K.andJ.Affleck-Graves(1995)“Thebrig-Run PerformanceofFollowingSeasoned PublicOfferings,”Journal ofFinancial Economics. Spiess,K.andR.Pettway(1995)“TheIPOandFirstSeasonedEquitySale:IssueProceeds, Owner/ManagersW’ ealth,andtheUnderpricingSignal,”unpublishedmanuscript, UniversityofNotreDameandUniversityofMissouri. Teoh,S.,Wong,T., andG.Rao(1994)“IncentivesandOpportunitiesforEarningsManagement inInitialPublicOfferings,”unpublishedmanuscript,U.Michigan,U.Marylandand FidelityManagementandResearchCo. Welch,I(1989)“SeasonedOfferings,ImitationCosts,andtheUnderpricingofInitialPublic Offerings,”Journal ofFinance, 44,421-449. 21

I I Table1 IPO Issuance by Year* GrossProceeds NumberofIPOs ($billions) 1980 129 0.83 1981 321 2.04 1982 110 0.99 1983 575 7.19 1984 287 2.24 1985 262 3.09 1986 506 10.22 1987 400 9.73 1988 162 3.84 1989 174 3.56 1990 146 3.38 1991 322 12.84 1992 426 17.25 1993 536 19.45 1994 463 13.65 Source:SecuritiesDataCompany. *IncludesPO issuancebynonfinanciadlommticcompanies.

Table2 Proceeds and SizeofFirms that IssuedWOSin 1983and 1988 Tablevalues,m mean(median)valuesforIPOfms in1983and1988.IPOamountisfromSecuritieDs ata Company(SDC). Salesare 1983and1988valueshornCOMPUSTAT.One-dayreturnasrecalculateadsthe percendtifferencbeetweenofferpricefromSDCandhe closingpriceonthef~stdayoftradinghornCRSP. SignificancteestsfordifferencesinmeanvaluesarebasedonF-statisticsandtestsfordifferenceisnmedian ontie Wi.lcoxontwo-samplerankedsumtest. values are based YearofIPO 1983 1988 TestStatistic One4ay Return (%) 14.6 6.6 2.86*** (3.9) (3.8) –1.00 2.00*** [POAmount ($rnill) 13.6 19.3 (7.9) (10.8) ‘2.57*** sales ($mill) 53.3 119.9 2.32*** (18.0) (37.6) 4.00*** [PO Amount–to-Sales (%) 28.4 3.0 340.lo*** (52.5) (29.4) –3.76*** Number of Firms 258 84 ‘**.**.*denotedifferencesbetween1983and 1988valuesaresignificantat 1%.5%.and10%,respectively.

Table3 The Long RunStock Returnsof 1983IPOSand 1988IPOS The sample includesfirmsthat completedIPOSin1983and 1988. Returnsare calculatedfrom CRSPdailysharepricedata. FirmswithmissingorquestionabledataonCRSPareexcludedfrom thesample. IPOreturns(r) arecalculatedusingdailydata(notcompounded)fromtwoweeksafter tjeIPOdatetothelastdayofthe1st,3rd,of5thyearaftertheIPO,whereyearsareperiodsof252 tradingdays. ForeachIPO,wecalculatethereturnontheNasdaq(rJtoforthesameperiod,which maybe lessthanayear ifthefirmdelists. The meansofeachsetofobservations(rlandrn)are usedtocomputewealthrelatives(1+ri/l+ m). 1983IPOS 1988Ipos Mean Mean Mean Mean IPO Nasdaq Wealth IPO Nasdaq Wealth Returnperiod Return Return Relative Return Return Relative Listingtoendof oneyear -24.0 -16.2 .91 20.4 16.0 1.04 Listingtoendof threeyears 4.8 22.9 .85 40.5 26.7 1.11 Listingtoendof fiveyears -7.3 24.7 .74 127.9 66.0 1.37 ExcludingAPCC - - . 51.1 66.0 .91

Table4 Industry Concentrations of Firms that Issued IPOS in 1983 and 1988 A. Percent of IPOFirmsinthe LeadingSICCategories ---- 1983 Percent 986 Percent 20 — 20 16 . 16 12 12 8 8 h 4 4 11 .7335 36 38 58 28 57 80 27 50 56 51 37 0 73 36 35 38 28 50 33 49 87 51 59 20 37 0 B. CumulativePercentofIPOFirms in the LeadingSICCategories 1983 1988 SIC Cumulative SIC Cumulative Category Percent Percent Category Percent Percent 73 16.3 16.3 73 11.8 11.8 35 12.9 29.2 36 9.4 21.2 36 11.4 40.6 35 8.2 29.4 38 11.0 51.6 38 7.1 36.5 58 5.3 56.9 28 5.9 42.4 Other 43.1 100.0 Other 57.6 100.0

TabIe5 Characteristics of Firms that Issued IPOS in 1983and 1988 T:~blcVL: ILJ;CRSthemedianvaluesforLPOfms in 1983and1988.Dataare1983and1988valuesfromthe COhfPUSTATresearchtapes. Operatingincomeisoperatingincomebeforedepreciation;capitalexpenditures tartforflxcd andequipmentandacquisitions:anddebtisshort—andlong-termdebt.Thesignificancetests Pliiilt fordifferencesinmedianvaluesarebasedontheWdcoxontw-sample tied sumtest. A. Firm Characteristics Year of IPO 1983 1988 Significance Sales(Smill) 18.0 37.6 *** Industry-Adjusted -52.8* -35.6 Operating Income–to-Sales (%) 10.5 12.5 Induscy–Adjusted 1.0+ 3.+ * Operating Income–to-Assets (%) 13.1 16.8 ** Industry-Adjusted 0.6 5.9- *** Sapital Exp.-to-Asset 8.8 8.2 (Yo) industry-Adjusted 3.3- 1.3+ >ebt-to-Assets (%) 10.7 14.3 ‘ixed Assets–to-Assets 18.3 15.0 (70) )ividends–to-hcome (%) 0.0 0.0 * **.l*.*denotedifferencesbetween1983and1988mediansaresignificantat1%,5%.and 10%,respectively, )asd onaWilcoxontw~sample rankedsumtest. ++ ‘+,+denoteindustry-adjustedmediansaresignificantlydifferenthornzeroat 1%,5%,and 10%,respecl lvely.basedonaW]lcoxonsignedranktest.

Table6 Firm Characteristics in YearsFollowing the IPO for Firms that Issued IPOSin 1983 and 1988 Tablevaluesareindustry-adjustedmtians forIPOfms in1983and1988.Medianindustry-adjustedsalesis themediandifferencebetweenfm salesandthemediansalesforfms initsindustryasapercentofmedian salesforfires initsindustty.Medianindustry-adjustedoperatingincomeratiosarethemediandifference betweenthe fim valueandthemedianvalueforfms initsindustry.Operatingincomeisoperatingincome &foredeprmiation,asdefinedonCOMPUSTATreseartcahpes.Years1-5represenctalendayrears1984-1988 forthe 1983I.POfins, andcalendaryears1989-1993fortie 1988IPOfins. Yearvaluesareb=ed onfms remaininginthesample. YearAfterIPO Year1 Year3 Year5 Saies(%) 1983PO Firms –35.& 6.3 4.0 1988IPOFirms 12.2 51.4 82.3+ Significance ** ** Op.Inc.-ttiales (%) 1983IPOFirms 0.2 -0.5 -0.1 1988IPOFirms 1.3 0.2 0.6 Significance Op. Inc.-to-Assets (%) 1983PO Firms -1.3+ 0.3 0.4 1988IPOFirms 1.9 -0.2 1.5 * Significance Cap. Exp.-to-Assets (%) 1983PO Firms 4.P 1.0+ 0.4 1988PO Firms 3.7- 1.8- 1.1+ Significance “**.l*,*denotedifferencesbetween1983and 1988mediansaresignificantat 1%,5%,and 10%,respectively, O* onaWilcoxontw-sarnple rankedsumtest. k++++ + ., denotemediansare signflcantly differentfromzero at 1%,5%,and 10%,respectively,basedon a Wilcoxonsignedranktest.

Table7 Median Annual Growth Rates ofFirm Characteristics for Firms that Issued ~Os in 1983 and 1988 Table industry-adjustedmediangrowthratesforIPOtis in1983and1988.Median vducs are industry-;idjus[edvaluesarethemedianWerence betweenthefm valueandthemedianvalueforfirmsinits industry.Operatingincomeisoperatingincomebeforedepreciation,asdefinedonCOMPUSTATresearch LIpcs.me years1-5representcalendaryears1984-1988forthe1983IPOsandcalendaryears1989-1993for the 1988IPOfirms. Yearvaluesarebasedonfms remainingintie sample. YearAfterIPO Year1 Year3 Year5 Grow~thin: Sales(%) 1983IPOFirms 21.5* 4.4* 2.9+ 1988IPOFirms 22.6* 6.4 2.2 Significance op. Inc.-ttiales (%) 1983IPOFirms -0.8* 0.5 0.5 1988IPOFirms -1.3+ -0.7 0.5 Significance op. Inc.-t&Assets (%) 1983IPOFirms –2.5* 1.4 0.8+ 1988IPOFirms –2.2 –1.0 1.0 Significance Cap. Exp.-t*Assets (%) 1983IPO Firms 1.3- -0.6 4.1 1988IPO Firms 0.7 –1.2 -0.2 Significance &** l* l denotedifferencesbetween1983and 1988mediansaresignificmtat 1%.5%.and 10%,respectively, . . basti onaWiicoxontw~sample rankedsumtest. +++ ++ + . . denotemediansare significantlydifferenthorn zero at 1%.5%.and 10%,respectively,basedon a Wilcoxonsignedrardctest.

Table 8 Annualreturnson 1983and 1988IPOstocks The sampleincludesfirmsthatcompletedIPOSin 1983and1988. Returnsare calculatedfrom CRSPdailysharepricedata. FirmswithmissingorquestionabledataonCRSPareexcludedfrom thesample. IPOreturns(r) arecalculatedusingdailydata(notcompounded)fromthestartofthe year tothelastdayoftheyearforthefirstthroughfifthcalendaryearsafterthe IPOdate. Firms thatdelistedpriortothe firstdayofthe periodreportedareexcludedfromthe calculationforthat period. ForeachIPO,we calculatethe returnonthe Nasdaq(rn)forthesameperiod,whichmay belessthanayearifthefirmdelists.Themeansofeachsetofobsewations(riand r.)areusedto computewealthrelatives(1+ri/l+ r~). First Second Third Foutih Fifth vear vear vear vear vear 1983IPOS -22.5 32.2 16.9 5.9 - 58.0 1983Nasdaq -11.0 30.5 7.8 -5.2 12.3 WealthRelative .87 1.01 1.09 1.12 1.41 1988IPOS 10.6 -5.0 99.2 63.2 103.4 1988Nasdaq 20.2 -17.6 55.2 13.2 15.4 WealthRelative .92 1.15 1.28 1.44 1.76 ExcludingAPCC .91 1.11 1.20 1.25 1.37 ,

Table9 FirmCharacteristicsin YearsFollowingtheIPOforFirms that issued IPOSin 1983 and 1988-Adjusted for Survivorship Bias T:lhlc industry-adjustedmediansforI-POfms in 1983and1988.Medianindustry-adjustedsalesis VaIUCS :uc themcdi~ differencebetweenfm salesandthemediansalesforfms initsindustxyasapercentofmedian S:LICfSorfirmsini~ industry.Medianindustry-adjustedoperatingincomeratiosarethemediandifference hctwecnthefim valueandthemedianvalueforfirmsinitsindustryO. peratinigncomiesoperatinigncome kfl~rcdcprcuititiona.sdefinedonCOMPUSTATresearchtapes.Years1-5representcaiendaryears1984-1988 for(heIY83IPOfires, andcalendaryears1989-1993forthe1988IPOfirms. Yearvaluesincludefms that f;iilorm ;icquiredandusethevalueofthelastyearthefm isintie sample. Year After IPO Year 1 Year 3 Year 5 Sales(%) 1983FO Finns –35.4* –8.1 –9.1 1988IPOFirms 9.7 48.0 . 82.3+ Significance * * ** op. Inc.-twSales (%) 1983IPOFirms 0.3 -0.9 –1.0 1988IPOFirms 1.3 0.2 0.5 Significance op. Inc.-t&Assets (%) 1983IPOFirms –2.1 -0.9 -0.9 1988IPOFirms 1.6 -0.8 0.8 * Significance ~ap. Exp.-t&Assets 1983FO Firms 4.1+ 1.0+ 0.2 (Yo) 1988IPOFirms 3.6* 1.8* 1.1 Significance k**** * . . denotedifferencesbetween1983and 1988mediansaresignificantat 1%,5%,and 10%,respectively, ~~sedonaWilcoxontwo-sampletied sumtest. }++++ * . . denotemediansare significantlydifferentfromzeroat 1%.5%.and 10%.respectively,bed on a Nilcoxonsignedranktest.

Table10 MedianAnnualGrowthRatesofFirmCharacteristicsfor Firms that IssuedIPOSin 1983 and 1988-Adjusted for Survivorship Bias Tablevaluesm industry-adjustedmediangrowthratesforLPOMs in1983and1988.Median industry-adjus(edvaluesarethemediandifferencebetweenthefm valuemd themedianvalueforfirmsinits industry.Operatingincomeisoperatingincomebeforedepreciation,asdefinedonCOMPUSTATresearch tapes.Theyears1-5representcalendaryears1984-1988forthe1983IPOsandcalendaryears1989-1993for the 1988IPOfires. Yearvaluesincludefms thatfailorareacquiredandusetie valueforthelastyearthe firmisintie sample. YearAfter IPO Year 1 Year3 Year 5 Growth in: Sales(%) 1983IPOFirms 21.5* 5.1* 4.0* 1988IPOFirms 22.6* 6.4 2.2 Significance Op. Inc.-t~ales (%) 1983IPO Firms -0.8++ 0.2 0.2 1988IPO Firms -1.3+ -0.9 0.5 Signflcance Op.Inc.-t&Assets (%) 1983IPOFirms –2.7- 1.1 0.6 1988IPOFirms –2.1 –1.0 1.0 Significance Cap. Exp.-t*Assets (%) 1983IPOFirms 1.3* -0.8* -0.6- 1988IPOFirms 0.6 –1.5+ -0.2 Significance b**** * . . denotedifferencebsetween1983and1988medianasresignificanattIZ.5Z,and10%.respwtively, ~asedonaWiicoxonNo-sample rankedsumtest. ++ ++ + l . . denotemediansare significantlydifferenthorn zeroat 1%,5%,and 10%,respectively,basedon a Wilcoxonsignedrardctest. I

Table 11 Long-RunStock Returnsforthe 1983IPOSand1988 IPOS Over PeriodsComparableto ThoseBasedon CompustatData The sampleincludesfirmsthat completedIPOSin 1983and 1988. Returnsare calculatedfrom CRSPdailysharepricedata. FirmswithmissingorquestionabledataonCRSPareexcludedfrom thesample. IPOreturns(r) arecalculatedusingdailydata(notcompounded)fromtwoweeksafter the IPOdatetothelastdayofthe Ist, 3rd,of5thfullcalendaryearafterthe IPO. ForeachIPO returnperiod,we calculatethe returnonthe Nasdaq(r~)forthesameperiod,whichmaybeless thanayearifthefirmdelists.Themeansofeachsetofobservations(riandrJ areusedtocompute wealthrelatives(1+ri/l+ r~). 1983IPOS 1988IPOS Mean Mean Mean Mean IPO Nasdaq Wealth IPO Nasdaq Wealth Returnperiod Return Return Relative Return Return Relative Listingtoyear-end ofnextcalendar -28.7 -15.3 .84 14.9 20.0 .96 year Listingtoyear-end ofthirdfullcalendar -4.9 16.8 .81 82.7 45.4 1.26 year Listingtoyear-end offifthfullcalendar -8.4 25.4 .73 166.8 78.1 1.50 year ExcludingAPCC - - 67.9 78.1 .94

Table12 Firm Characteristics in YearsFollowingthe IPO for Firms that Issued IPOSin 1982 and 1983 Tablevaluesareindustry~djustedmediansforIPOfms in1982and1983.Medianindustry-adjustedsalesis themediandifferencebetweenfm Wesandthemediansalesforfms initsindushyasapercentofmedian salesforfms initsindustry.Medianindustry-adjustedoperatingincomeratiosarethemediandifference betweenthefm valueandthemedianvalueforfirmsinitsindustry.Operatingincomeisoperatingincome beforedepreciation,asdefinedonCOMPUSTATresearchtapes.Years1-5representcalendary~ 1982–1987 forthe1982LFOfins, andcalendaryears1983-1998forthe1983IPOfins. Yearvaluesarebasedonfms remaininginthesample. Year YearAfterIPO ofIPo Year1 Year3 Year5 Sales(%) 1982IPOFirms –56.9+ 47.6 -11.8 37.3 1983IPOFirms –52.8+++ –35.0+++ 6.3 4.0 Significance Op.Inc.-ttiales (%) 1982IPOFirms 6.6++ 4.0 –1.3 4.6+ 1983IPOFirms 1.0+ 0.2 4.5 -0.1 ** Significance Op.Inc.-to-Assets(%) 1982IPOFirms 4.9+ 1.6 –2.0 4.0 1983IPOFirms 0.6 –1.3+ -0.3 0.4 Significance * Cap.Exp.-to-Assets(%) 1982IPOFirms 4.0++ 2.5++ 1.0 0.3 1983IPOFirms 3.3+++ 4.0+++ 1.0+ 0.4 Significance ***,**,*denotedifferencebsetween1982ad 198: mediansaresignificantat IYo,5Y0,ad 10Yor,espectively, basedonaWilcoxontwo-sampletied sumtest. ‘++,‘+, +denotemediansare significantlydifferentfromzeroat 1%,5%,and 10%,~spectively,basedon a Wilcoxonsignedranktest.

Table13 Wealth Relatives for Portfolios of Firms that Issued IPOSin 1983 and 1988 StockreturnsareadjustedfordividendsandsplitsforIFOfms in1983and1988.Years1,3,and5represent calenti years1984,1986,and1988fortie 1983IPOfins, and1989,1991,and1993forthe1988IPOfirms. ReturnsfortheIPOfirms,ri,andNASDAQ,rn,aremeansforthebestandworstperforminggroupsoverthe definedperiod,definedbyfirmsabovethe75tipercentileorbelowthe25tipercentileforthecombinedgroup offins. Wealthrelativesarecalculatedas(l+rJ/(l+rn). Returnsfrom2weeksaftertheIPOthrough Year1 Year3 Year5 Best Worst Best Worst Best Worst Growthin: Op.Inc.-to-Sales 1983 .95 .56 1.08 .31 .91 .32 1988 1.17 .81 1.86 .68 2.12 .48 Op.Inc.-to-Assets 1983 1.13 .64 1.17 .48 .87 .52 1988 1.41 .57 2.74 .52 4.91 .51 Sales 1983 .93 .61 .86 1.03 .83 .92 1988 1.71 .64 1.45 2.63 1.69 3.59 Cap.Exp.–to-Assets 1983 .91 .70 1.01 .60 1.13 .56 1988 .67 .89 2.96 .71 4.73 1.01