監(jiān)管沙盒與金融科技融資-Regulatory Sandboxes and Fintech Funding Evidence from the UK

DownloadedfromDownloadedfrom/rof/article/28/1/2RegulatorySandboxesandFintechFunding:EvidencefromtheUK*GiulioCornelli1,2,SebastianDoerr1,LeonardoGambacorta1,andOuardaMerrouche3AbstractOver?ftycountrieshaveintroducedregulatorysandboxestofoster?nancialinnov-ation.Thisarticleconductsthe?rstevaluationoftheirabilitytoimprove?ntechs’ac-cesstocapitalandattendantrealeffects.ExploitingthestaggeredintroductionoftheUKsandbox,weestablishthat?rmsenteringthesandboxseeanincreaseof15%incapitalraisedpost-entry.Theirprobabilityofraisingcapitalincreasesby50%.Sandboxentryalsohasasigni?cantpositiveeffectonsurvivalratesandpat-enting.Investigatingthemechanism,wepresentevidenceconsistentwithlowerasymmetricinformationandregulatorycosts.V204G.Cornellietal.DownloadedfromDownloadedfrom/rof/article/28/1/2Therapidgrowthofinnovativecompaniesinthefinancialsectorthatusenewtechnologyholdsthepromiseofspurringcompetition,leadingtoefficiencygainsandmorechoiceforconsumers.However,“fintechs”offernovelproductsinanenvironmentofhighregulatoryuncertainty,sotheyoftenstruggletoraiseenoughcapitaltodevelopproductsandexpand(HaddadandHornuf,2019).Policymakersaroundtheworldarethussteppinguptheireffortstofosterinnovationinthefinancialsector,astheyhavedonewithbusinessaccelera-torsorgrantsinotherareas(Howell,2017;Gonza′lez-UribeandLeatherbee,2018).Alandmarkinitiativetonurturefintechswasthecreationofthe“regulatorysandbox”bytheUnitedKingdom’sFinancialConductAuthority(FCA).EstablishedinNovember2015,thesandboxoffersfintechsacontrolledtestingenvironmentinwhichtheycantryouttheirproductsonalimitedsetofcustomers.Testingoccursundercloseregulatorysupervision:firmsreceiveadvicetohelpthemnavigatethecomplexitiesofregulationsandtoeasetheroutetoauthorization.Akeyobjectiveofsandboxesistofosterinnovationbyfacilitatingfintechs’accesstofinancingatearlystagesofdevelopment.1Regulatorscanusesandboxestolearnaboutnewfinancialtechnologiesandemergingtrends,aswellastoidentifyassociatedrisksbeforeproductsarelaunchedforthemassmarket.Bynow,overfiftycountrieshavefollowedtheUKandintroducedtheirownregulatorysandbox,oftenwiththegoalofnurturingthefintechsector(Wechsler,Perlman,andGurung,2018).2Andyet,despitetheirwide-spreadadoptionandsignificantattentioninthemediaandpolicycircles,littlesystematicempiricalevidenceexistsonwhethersand-boxesactuallyhelpfintechstoraisefunding,innovate,ordevelopviablebusinessmodels.Noristhereanyevidenceontheunderlyingchannelsthatcouldbeatwork.ThisarticleanalyseshowenteringtheFCA’sregulatorysandboxaffectsfintechs’abilitytoraisefundingandattendantrealeffects.WecollectdataoncapitalraisedbyfintechsintheUKsandboxfortheperiodfrom2014q1to2019q2.Detaileddeal-leveldata,brokendownbyindividualinvestor,aswellasbackgroundinformationonafirm’sage,size,indus-try,location,anditsCEOsallowustoinvestigatedifferentchannelsthroughwhichthesandboxcouldaffectafirm’saccesstocapital.Wefocusonthesampleoffirmsacceptedintothesandboxandexploitthefactthatthesefirmsenteredthesandboxinfivecohorts.Asentryisstaggeredoverroundsof6months,wecancompareafirm’scapital-raisingac-tivitybeforeandafterparticipationinthesandbox,relativetofirmsthatwillenterthesandboxlater.Entryintothesandboxisassociatedwithanincreaseintheaverageamountoffundingraisedandahigherprobabilityofraisingfunding.Infirm-levelregressions,wefindthatentryintothesandboxisfollowedbya15%increaseincapitalraised(or$700,000)overthefollowing2years,relativetofirmsthatwillenterthesandboxatalaterdate.Firms’probabilityofraisingcapitalincreasesby50%.Theincreaseincapitalraisedcorrespondstoaboutonestandarddeviation.3tofinanceforinnovatorRegulatorySandboxesandFintechFunding205Downloadedfrom/rof/article/28/1/2Weobtainsimilarfindingswhenwecomparesandboxfintechstoasetofmatchedcon-trolfirms.Specifically,usingacoarsenedexactmatching(CEM)approach,weselectasam-pleoffintechsthatarestatisticallysimilarintermsofobservablefirmcharacteristics.Weestimateadifference-in-differencesspecificationwithfirmandtimefixedeffects,compar-ingfirmsthatenterthesandboxtothosethatneverenterthesandbox,butsharesimilarcharacteristics.Inthematchedsample,wefindalmostidenticaleffectstoourbaselineDownloadedfrom/rof/article/28/1/2Facilitatingfintechs’accesstofinanceisanimportantstepinbenefittingconsumersthroughgreaterinnovationandcompetition.Whiletheshorttimespansinceitsinceptiondoesnotallowforanevaluationofthesandbox’seffectsontheconsumersurplusorcom-petitioninthefinancialsectoryet,wecananalyzesurvivalratesandpatentingactivity.Wefindthatsandboxfirmsaresignificantlymorelikelytostillbeinoperationtoday,com-paredtoasetofmatchedcontrolfirmsthatdidnotenterthesandbox(75%versus60%).4Inaddition,sandboxentryisassociatedwithanincreaseinpatentingactivity,suggestingthateasieraccesstocapitalspursfirms’innovativeactivity.Wetheninvestigatetheunderlyingmechanisms.Asymmetricinformationisparticularlyacuteinventurecapitalmarkets,becauseissuersaremainlyearly-stagefirmswithnopriortrackrecords(Trester,1998).Uncertaintyaboutthequalityofnewproductsandservicesofferedbyfintechscouldthuspresentaseriousobstacletoraisingcapital—especiallyinanenvironmentofhighregulatoryuncertainty(HaddadandHornuf,2019).Navigatingthecomplexitiesofaconstantlychangingregulatoryframeworkcouldfurtherposeasignifi-cantcosttofirms.Sandboxescouldcurbinformationalfrictionsthroughregulatoryover-sightandcontinuousdialoguebetweenfirmsandtheregulatorduringthetestingperiodthatoffersreassurancetoinvestorsthatfirmsmeettheirregulatoryobligations.Additionally,advicebytrainedcaseofficerspromisestolowerregulatorycostsforfirmsandreducetherisktofirmsofofferingproductsthatcouldbeinviolationoftheregulatoryframework.Ourevidenceisconsistentwiththenotionthatthesandboxreducesinformationasym-metriesbetweeninvestorsandfirms,aswellascostsassociatedwithregulatoryuncertainty.Wefirstshowthattheestimatedpositiveeffectofsandboxentryoncapitalraisedisstron-gerforsmallerandyoungerfirms,whicharemoreopaqueandhencesubjecttoseverein-formationalfrictions(HallandLerner,2010).Similarly,entryintothesandboxisassociatedwithgreaterincreasesindealvolumeforventurecapitaldeals,whicharegener-allymoreinformation-sensitive,comparedtoothertypesofdeals(Gompers,1995;Howell,2020).Second,dataattheinvestor-firmlevelshowthatfirmsinthesandboxraisemorecapitalespeciallyfrominvestorsbasedoutsidetheUKandinvestorsthathavenotprevious-lyinvestedintothefirm.Sincetheseinvestorslikelyfacehigherinformationasymmetries(GrinblattandKeloharju,2001;IvkoviandWeisbenner,2005),weinterpretthisfindingasevidencethatthesandboxhelpsinreducinginformationalfrictions.Finally,weshowthatfirmswithaCEOwhohasabackgroundinfinanciallawraiserelativelylesscapital206G.Cornellietal.Downloadedfrom/rof/article/28/1/2afterentryintothesandbox.ThisisinlinewithanecdotalevidencethatCEOswithoutpriorexperienceinfinancialregulationbenefitthemostfromtheguidanceprovidedbycaseofficers(Deloitte,2019),andtherebyfromthereductioninregulatorycostsDownloadedfrom/rof/article/28/1/2Akeychallengeforidentificationisthatevenamongthegroupoffirmsthatenterthesandboxatsomepoint,theentrydatecouldbecorrelatedwithunobservablefirmcharac-teristics.5Weperformanumberofexercisestoaddressthisconcern.First,weshowthatamongthegroupoffirmsthatenterthesandboxatsomepoint,thespecificentrydateisuncorrelatedwithobservablefirmcharacteristics.Second,ourmainresultsholdwhenwecomparesandboxfintechstoasetofcontrolformsthataresimilaralongobservablechar-acteristicstothesandboxfirms,selectedviaCEM.Third,weshowthatourresultsarero-busttotheinclusionoffixedeffects.Forexample,ininvestor-firmlevelregressionsweincludeinvestor*firmandfirm*timefixedeffectsfixedeffects.Thesefixedeffectsaccountforunobservableheterogeneitywithineachfirm-investorcombination,aswellasunobserv-abletime-varyingfactorsatthefirmlevel(KhwajaandMian,2008;Jime′nezetal.,2014).Whileresultsfromthesetestssuggestthatthesandboxhashelpedfintechsraisefunding,ininterpretingourfindingsitisimportanttokeepthecaveatinmindthatsandboxentryisnotrandom.Weprovideaseriesofadditionalexercises.Oneexplanationforourfindingscouldbethatinvestorssimplylearnaboutfirmsastheygraduallyrevealtheirqualitytothemarketovertime,irrespectiveofentryintothesandbox.Then,firms’abilitytoraisefundingwouldincreasegradually.Ifinsteadinvestorslearnaboutthequalityofafirmbecauseofthe“sandboxcertification”,firms’abilitytoraisefundingwillincreaseimmediatelyafterentry.Wefindthatthestrongesteffectsonfundingraisedoccurinthefirsttwotofourquartersuponentry.Fourtoeightquartersafterentry,thesandboxstillhasapositive,butsmallereffectonfundingraised.Wealsoshowthatourresultsarerobusttoalternativeestimationmethodstoaccountforthepresenceofzerosinourdependentvariable,orwhenweincludecohortorcontiguous-cohortfixedeffects.Ourfindingsdonotprecludethatthesandboxoperatesthroughadditionalchannels.Forexample,itcouldhaveageneralsignalingeffect:selectionintothesandboxcouldserveasastampofapprovalandhelpsandboxfirmsraisemorecapital.Regulatoryapprovalcouldfurtherindicatethatafirm’sproductisviableandwillfacefewerregulatoryhurdlesgoingforward.Fromthefirm’sperspective,thiswouldmeanthatitnowhastheapprovaltosellitsproducts,whichlikelyrequiresinvestmentsinsalesandscaling.Ifso,entering(andgraduatingfrom)thesandboxwouldleadtohigherdemandforcapitalamongallsandboxfirms.Thatsaid,ourfindingsonthedifferentialeffectsforsmallandyoungfirms,andespeciallythelargereffectsfoundforcapitalraisedfromforeignandfirst-timeinvest-ors,areconsistentwiththesandboxreducinginformationalfrictionsandfacilitatingfin-techs’accesstocapitalaboveandbeyonditsgeneraleffectsthroughsignalingandonthedemandforfunding.RegulatorySandboxesandFintechFunding207Downloadedfrom/rof/article/28/1/2Allinall,ourfindingssuggestthattheregulatorysandboximprovesfintechs’accesstocapital;firmsenteringthesandboxarealsomorelikelytostillbeinoperationandhaveapatent.Tothebestofourknowledge,thisarticleprovidesthefirstsystematicevidencethatsandboxeshelpfintechstoraisecapitalandinnovate—andhenceachieveoneDownloadedfrom/rof/article/28/1/2Ourarticlecontributestothedebateonhowpublicpoliciescanfosterinnovation(KerrandNanda,2015;LernerandNanda,2020).Arecentliteraturehasestablishedthatfin-techsfaceseriousobstaclestoraisingcapital(Blocketal.,2018;HaddadandHornuf,2019),despitethefactthattheirinnovationprovidesvaluetoinnovatorsandinvestors(Chen,Wu,andYang,2019).Asmarketfailurescanleadtosub-optimalprivate-sectorex-penditureonresearchanddevelopment,publicpoliciestofosterinnovation,forexamplethroughgrantsandbusinessincubatorsoraccelerators,canhavesizeablebenefits(Howell,2017;Gonza′lez-UribeandLeatherbee,2018;Yu,2020;Gonza′lez-UribeandReyes,2021).7Policymakershencewanttopromoteinnovationinthefinancialsector,andregu-latorysandboxeshaveemergedasaprominenttooltodoso.Yet,evidenceontheireffect-ivenessisscarce.Wealsorelatetoliteraturethatinvestigateshowtoregulatefintechs(Arner,Barberis,andBuckey,2017;Zetzscheetal.,2017;Magnuson,2018).Buchaketal.(2018)showthattherapidgrowthoffintechlendersintheUSAismostlyexplainedbylighterregulationandbettertechnology,withbenefitstoconsumers(seealsoThakor,2020andFusteretal.,2019).Otherstudiesshowthattheuseofbigdataandmachinelearningcanleadtoalgo-rithmicdiscriminationandchangesinconsumerbehavior(Bartlettetal.,2022;Bergetal.,2020;Fusteretal.,2022),andthatthegrowthoffintechsraisesconcernsaboutdatapriv-acy(Armantieretal.,2021;Chenetal.,2023;Doerretal.,2023).Theentryoffintechsintofinancethusconstitutesadilemmaforpolicymakers:theyneedtopromoteinnovationinthefinancialsector,butwithoutcompromisingdataprivacy,financialstability,orcon-sumerwelfare(BrummerandYadav,2019).Newregulatorytoolsmightthusbeneeded,andsandboxescouldbeonesuchtool:theyprovideregulatorswiththeabilitytosupportsafeinnovationbygaugingthepotentialwelfareimplicationsofnewproductsbeforetheyarelaunched.Anassessmentoftheeffectivenessofsandboxesandanunderstandingofthechannelsthroughwhichtheyoperatearehenceindispensable.Theremainderofthearticleisorganizedasfollows.Section2providesbackgroundin-formationontheUKregulatorysandbox.Section3givesanoverviewofourdataandsam-pleoffintechs.Section4explainsourempiricalstrategy.Itthenreportsthemainresultsandprovidesevidenceonthemechanismsatwork.InSection5,wepresentrobustnesstests.Section6concludes.208G.Cornellietal.DownloadedfromDownloadedfrom/rof/article/28/1/2TheUK,andespeciallyLondon,hasbecomeaglobalfintechhub.Overthelastdecade,fin-techstart-upsraisedaroundone-fifthoftheirworldwidefundingintheUK.ThisnumberistoppedonlybytheUS,wherefintechsraisealmosthalfofglobalfunding.Notably,theUKincreaseditssharefromlessthan15%in2010–14toover20%forthe2015–19period.IttherebyovertookChinaintermsofvolume.However,despitetheseencouragingnumbers,fintechsstillfacesevereobstaclesinraisingcapital(HMTreasury,2016,2019).Sinceac-cesstocapitaliscrucialforyoungfirmswithhighgrowthpotential,itsscarcitythreatenstoslowinnovationinthefinancialsector.Partlyinresponsetotheseworries,theFCAannouncedtheworld’sfirstregulatorysandboxaspartofits“ProjectInnovate”in2015(FCA,2015).Thesandboxoperatesonacohortbasiswithtwo6-monthtestperiodsperyear.BetweenNovember2016andJuly2019(theendofoursampleperiod)fivecohortsoffirmshavebeenacceptedintothesandboxonthefollowingdates:7November2016,15June2017,5December2017,3July2018,and29April2019.8Intotal,375firmshaveappliedand118havebeenacceptedintotheFCAsandbox.Eachco-hortaveragesaroundtwentyfivefirms.Cruciallyforoursetting,theintroductionofthesand-boxwasnotaccompaniedbystricterregulationfornon-sandboxfirmsthatcouldhavediscouragedthemfromseekingfinancing,ormechanicallybenefitedsandboxfirms.NordoestheFCAprovideinsurance,inthesensethatcustomersofsandboxfirmsareprotectedfromanyrisksarisingfromusingtheirproductsandservices.TheFCApublishesthenamesofacceptedfirmsforeachroundofthesandbox;itdoesnotmakethelistofrejectedfirmsavailable.Theaveragefirminthesandboxisastart-uporsmallandmediumenterpriseinretailbanking(includingpayments),wholesalemarkets,retailandwholesaleinvestmentandlending,orinsurancepropositions.Sandboxfirmsofferawidearrayofnewproductsandservices.Forexample,firmsofferaplatformthatfacilitatesthesecuritiza-tionofSMEdebtbydigitisingcreditapplications;aninterest-freesalary-advanceandcashflowmanagementproductutilisingdistributedledgertechnology;oranaggregationplatformthatfacilitatesinvestmentinadiversifiedportfolioofP2Ploans.The“sandboxprocess”containsfourdistinctsteps:application,selection,testing,andexit.TheFCAselectsfirmsoutofthepoolofapplicantsbasedonwhetherthefirmoffersagenuineinnovationthatbenefitsUKconsumers.Theinnovationshouldconstituteanim-provementoverexistingproductsandservices,andhencepromotecompetition.9Theage,size,orprofitabilityofacompanyarenottheeligibilitycriteria;start-upsandincumbentareequallyencouragedtoapply.FirmsthatfulfilltheserequirementsandforwhichtheFCA’sinputisdeemeduseful(“needforsupportcriteria”)areselectedintothesandbox.Firmsareassignedadedicatedcaseofficerwhohelpstodesignthetestsetup,providesRegulatorySandboxesandFintechFunding209Downloadedfrom/rof/article/28/1/2guidancetocompletethenecessarypaperworkforauthorization,andhelpsDownloadedfrom/rof/article/28/1/2Oncefirmsaregrantedauthorization,theytesttheirproductsinalimitedmarketenvir-onment.Specifically,successfulapplicantssetuptheircapabilitiessubjecttoregularreport-ingrequirementstomonitorhowthetechnologyisevolvingandhowthebusinessismeetingitscompliancetargets.Inthisphase,firmshavetofamiliarizethemselveswiththeregulationwiththehelpofcaseofficers.After6months,firmssubmitafinaltestingreportandexitthesandbox.Theycanapplyforapermanentauthorizationuponcompletion.Oneofthemainobjectivesofthesandboxistoattractinvestmentstowardfintechs(FCA,2015).InthewordsoftheFCA,“thepotentialbenefitsofaregulatorysandboxcouldbesignificantfrombetteraccesstofinance.Financialinnovationreliesoninvestment,muchofitthroughequityfunding.RegulatoryuncertaintyatacrucialgrowthstagemeansthatFinTechfirmsfindithardertoraisefundsandachievelowervaluationsasinvestorstrytofactorinrisksthattheyarenotwellplacedtoassess”.11Theseargumentsareinlinewiththefindingthatasymmetricinformationisparticularlyacuteinventurecapitalmarkets,becauseissuersaremainlyearly-stagefirmswithnopriortrackrecords(Trester,1998;Howell,2020).Resolvinginformationproblemsinsuchanenvironmentrequiresthatinvestorsengageinanintensiveandcostlyup-frontscreeningandpost-investmentmonitoring.Theinherentuncertaintyaboutthequalityofnewprod-uctsandservicesofferedbyfintechsthuspresentsaseriousobstacletoraisingcapital(HaddadandHornuf,2019).Sandboxescouldfurthercurbinformationalfrictionsthroughregulatoryoversightdur-ingthetestingperiod.Continuousdialoguebetweenfirmsandtheregulatoroffersreassur-ancetoinvestorsthatfirmsmeettheirregulatoryobligations.IntheUKcontext,adverseselectionispotentiallyreducedbecausetheFCAclaimstoselectfirmsthatoffergenuinein-novationwithclearbenefitstoconsumers.MoralhazardmayalsobereducedifclosesupervisionbytheFCAspursfirmstoimprovetheirgovernanceandadoptmorerigorouspoliciesandprocesses.Besidesinformationproblems,anothercriticalobstacletocapitalaccessrelatestoregu-latorycostsanduncertainty.Regulatoryuncertaintydiscouragesinvestmentbecauseinvest-orsarehesitanttoinvestinafirmthatisofferingproductswhoseregulatoryframeworkisunclear.Eveninnovativeandsuccessfulcompaniesmightbeforcedtoaltertheirbusinessmodeltocomplywithcontinuouslyevolvingregulations.AccordingtotheWorldBank’sDoingBusinessreport,regulatoryrestraintsareakeybarriertoinnovation.12Thesandboxcouldreduceregulatoryuncertainty,asregulatorsthroughouttheprocessareabletocollectinformationonnewproducts,identifynewrisks,andaccordinglyadaptexistingorintro-ducenewregulationswiftly.Advicebytrainedcaseofficerspromisestolowerregulatory210G.Cornellietal.Downloadedfrom/rof/article/28/1/2costsforfirmsandreducetherisktofirmsofofferingproductsthatcouldbeDownloadedfrom/rof/article/28/1/2Anecdotalevidencesuggeststhatthesandboxhasdeliveredvaluetofirms(Deloitte,2019).Forexample,companiesvaluetheguidanceontheapplicationofregulationtoin-novativeproductsandservices,andhighlightthebenefitsofdetailingtherisksrelatingtotheirbusinessmodetotheregulator.Theyfurtherreportthatoperatingtheirtechnologysuccessfullyinaliveandregulatedenvironmenthelpsthemtosignalcredibilitytoinvestors.Firmsfurtherstatethattheroutetoauthorizationissignificantlysimplerandfasterwhenregulationisconsideredfromthestartandwiththehelpofcaseofficers(FCA,2019).Enhancingfintechs’accesstocapitalisanexplicitgoalofsandboxes.Acomplementarylong-termpolicyobjectiveistoincreaseconsumerwelfare,forexamplebypromotingcom-petitionandinnovationwhileensuringfinancialstability.Theshorttimespansincetheirinceptiondoesnotallowustoevaluatetheselong-termeffects.However,ourarticleempir-icallytestswhethertheUKsandboxhelpsfintechstoraisecapital,andwhetherthereareanyeffectsonsurvivalratesandpatentingactivity—therebyprovidingafirststepinassess-ingsandboxes’usefulness.3.DataDescriptionandSampleSelectionPitchBookprovidesdetaileddataoncapitalraisedatthedeallevelandisoneofthemostcomprehensivesourcesofinvestmentdataforthefintechsector.Weobtaindataonallindi-vidualdealsofthesandboxfirms,aswellasdealsforarandomsampleofaround1,400controlfirms(morebelow),overtheperiod2014q1–2019q2.Foreachdeal,PitchBookrecordsdetailedcharacteristicssuchasissuername,dealdate,dealamount,andtype/pur-poseofthedeal.Further,eachdealcontainsinformationontheindividualinvestorsandtheirlocation.Themaintypesofdealsareventurecapital(VC)deals(includingaccelera-tors,incubators,seed,andangeldeals),privateequity(PE)deals(mainlyforgrowth/expan-sion),andrestructuringdeals(includingdealsformergersandacquisitionsandbuyoutdeals).VC,PE,andrestructuringdealsrepresentaround63%,7%,and17%ofthetotalnumberofdeals.WefurthercollectanyavailableinformationonthehistoryofeachcompanyandthebiographyoftheCEO(orfounder).Weobtaintheyearthecompanywasfounded,itspri-maryindustryclassification,andthecurrentbusinessstatus(start-up,generatingrevenue/profitable,other).Wealsorecordthecitywherethecompanyisheadquartered.InformationontheCEOincludesgender,educationalbackground,andoccasionallytheyearofgraduation.Finally,wecollectinformationontheidentityofeachinvestorpartici-patinginadeal,aswellasthesizeofeachinvestor’stotalportfolio.Wealsoobservethecountrywheretheinvestorisheadquartered,whichallowsustoseparatelyinvestigatethebehaviorofforeignandUK-basedinvestors.PitchBookalsoprovidesinformationonwhetheraninvestorisanew/first-timeinvestorintheissuingfirm.3.1SandboxFirmsThemainanalysisfocusesonfirmsthatenterthesandboxinoneofthefivecohortsduringthesampleperiod.WemanuallyidentifysandboxfirmsinthePitchBookdatabase,usingthenameanddescriptionofthecompanyprovidedbytheFCA.Outofthe118firmsthathavebeenacceptedintothesandbox,weareabletoidentify106inthePitchBookdatabase.RegulatorySandboxesandFintechFunding211Downloadedfrom/rof/article/28/1/2Weperformaseriesofstepstocleanandpreparethedata.First,sevenfirmsenteredthesandboxmorethanonce.Toavoiddouble-counting,forthesefirmsweonlyconsiderthedatewhentheyenteredthesandboxforthefirsttime.Second,wedropdealsthatreportnotimeofdeal(36deals),sinceitisnotpossibletodeterminewhethertheytookplacebeforeoraftersandboxentry.Third,wedropobservationswithmissingorzerodealsize(83deals).Finally,firmsmustreporttheirprimaryindustry,location,CEOgender,Downloadedfrom/rof/article/28/1/2Wethenaggregateourdealdata,whichisatdailyfrequency,tothequarterlylevel.Sincethefocusofouranalysisisonfintechstart-ups,thebaselinespecificationexcludessandboxfirmsthatare(orbelongto)largeorlistedfirmsandthereforedonotreportaccountsordonotraisecapitalseparatelyfromtheparent(eightfirms,e.g.,LloydsorHSBC).However,wewillincludetheselargerfirmswhenweinvestigatethemechanism.Finally,wetrimlogdealvolumeatthe1stand99thpercentiletokeepoutliersfromdrivingourresults.Tocreateabalance