Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Securitization wikipedia , lookup
Systemic risk wikipedia , lookup
Investment management wikipedia , lookup
Shadow banking system wikipedia , lookup
History of the Federal Reserve System wikipedia , lookup
Interbank lending market wikipedia , lookup
Panic of 1819 wikipedia , lookup
Corporate Governance and Risk Management at Unprotected Banks: National Banks in the 1890s CharlesW.CalomirisandMarkCarlson* Abstract Weexaminebankgovernanceandriskchoicesfromthe1890s,aperiodwithoutdistortionsfrom deposit insurance or other government assistance to banks. We link differences in managerial ownership to different corporate governance policies, risk, and methods of risk management. Formal corporate governance and high manager ownership are negatively correlated. Managerial salariesandself‐lendingaregreaterwhenmanagerialownershipishigher,andlowerwhenformal governance is employed. Banks with high managerial ownership (low formal governance) target lower default risk. High managerial ownership rather than formal governance is associated with greaterrelianceoncashratherthanequitytolimitrisk. Keywords:Managerownership,corporategovernance,rentseeking,riskpreferences,bankfailures JELCodes:G21,G32,N21 *Calomiris(correspondingauthor):ColumbiaBusinessSchoolandNBER([email protected]).Phone:212‐ 854‐8748.Fax:212‐316‐9219.MailingAddress:ColumbiaBusinessSchool,3022Broadway,UrisHall601, NewYork,NY10027. Carlson:BoardofGovernorsoftheFederalReserveandBankforInternationalSettlements ([email protected]). WethankStijnClaessens,LucLaeven,RobertaRomano,PhilStrahan,andseminarorconferenceparticipants attheNBERRiskofFinancialInstitutionSummerInstitute,theBoardofGovernors,theFederalReserve BanksofAtlanta,Cleveland,andNewYork,theInternationalMonetaryFund,ColumbiaUniversity,Tilburg University,Dartmouth,ErasmusUniversity,NewYorkUniversity,YaleUniversity,andtheUniversityof Virginiaforcomments.IanniDrivasprovidedvaluableresearchassistance.Theviewsexpressedinthis paperaresolelythoseoftheauthorsanddonotnecessarilyreflectthoseoftheFederalReserveBoard,the BankforInternationalSettlements,ortheirstaffs. 1.Introduction Incentivesofmanagersmayconflictwiththoseofoutsideshareholdersandcreditors, particularathighlyleveragedandopaqueinstitutionssuchasbanks.Agencyproblemsariseboth withrespecttotheoutrighttransferofresources(e.g.,excessivesalariesorsubsidizedaccessto credit),aswellasimplicittransfersrelatedtoriskmanagementpractices(e.g.,inadequaterisk managementeffortortransfersfromcreditorstostockholdersthrough“riskshifting”).Somerisk shiftingbenefitsmanagersattheexpenseofallclaimantsonthebank,whileotherformsofrisk shiftingbenefitstockholdersattheexpenseofcreditors.1Bankersdesigncontractingand governancestructuresthatsufficientlyresolveagencyproblemssothattheycanattractfunding fromminorityshareholdersanddepositors. Examininghowbanksresolvethoseconflictsintoday’sbankingenvironmentthroughtheir choicesofownershipstructure,governanceandrisk‐managementpractices,andportfoliostructure iscomplicatedbygovernmentregulationofcapitalstructureandcorporategovernancepractices. Also,protections,suchasdepositinsurance,too‐big‐to‐failpolicies,andalenderoflastresort,can distortincentiveswhenmakinggovernance,ownershipstructure,andrisk‐managementchoices.To investigatetheendogenousemergenceofcorporategovernancemechanismsthatlimitrentseeking andcrediblymanagerisk,welookatbanksfromthe1890s,aperiodwhentherewasnodeposit insurance,nolenderoflastresort,andvirtuallynogovernmentinterventionstosavebanks. Wedosousingnationalbanks’ExaminationReports,adetailedbutseldomusedresource thatprovidesconsiderableinformationaboutbanks’ownershipstructures,governancestructures, 1Riskshiftingor“assetsubstitution”referstochangesintheriskprofileofinvestmentsdesignedtotransfer valuefromsomeclaimantstoothers.Forexample,thedecisiontoinvestlessinriskmanagementcanresultin privatebenefitstobankersattheexpenseofoutsidesourcesoffunding(HolmstromandTirole,1997).Also, anincreaseinthevarianceofreturnstypicallytransfersvaluefromcreditorstostockholders(Jensenand Meckling,1976).Notethatthesetwokindsofriskshiftingdifferintheidentitiesofwinnersandlosers. 1 toolsformanagingriskandlevelsofrisk.Thisinformationallowsustolinkdifferencesin ownershipstructure(especiallytheextentofmanagerialownership)todifferencesincorporate governancepolicies,riskoutcomes,andbanks’approachestoriskmanagement,includingtheir portfoliostructurechoices.Wefindthatexogenousvariationinbanks’circumstancesresultin importantdifferencesinownershipstructure,risktaking,andcorporategovernancechoices,which inturnresultinimportantdifferencesacrossbanksintheextentofmanagerialrentextractionand thestructureofbanks’assetsandliabilities.Banksthateschewformalcorporategovernancein favorofconcentratedmanagerialownershiptendtoexperiencehighermanagerialrentextraction, lowerdefaultriskondebt,highercashholdingsasafractionofassets,andlowerequitycapitalasa fractionofliabilities. 2.Literaturereview Thesimplestclassofagencyproblemsrevolvesaroundthetransferofresourcestoinsiders whomaintainoperationalcontroloverthebank.2Bankmanagerswithsufficientcontrolrightsmay paythemselvesexcessivesalariesorgivethemselvesaccesstocreditonsubsidizedterms.Fora sampleofEastAsianfirmsinthemid‐1990s,Claessens,Djankov,Fan,andLang(2002)findthat concentratedmanagementownershipincreasesfirmvaluewhenownershipandcashflowrights arealigned,butincasesinwhichmanagersenjoygreatercontrolrightsthancashflowrights, managerialownershipconcentrationisvalue‐destroying. Controlrightscanalsogiverisetoagencyproblemswithrespecttoriskmanagement, involvingdifferencesinportfoliochoice,asmodeledbyJensenandMeckling(1976),Myers(1977) andMerton(1977),ordifferencesinrisk‐managementeffort,asmodeledbyHolmstromandTirole (1997).Ingeneral,managerswhohavelargestakesintheperformanceoftheirbanksmaypreferto takelessriskortoexertgreatereffortinmanagingriskinordertopreservetheirownfinancial 2SeeShleiferandVishny(1997)forageneralreviewofissuesincorporategovernance. 2 wealthortheirfirm‐specifichumancapital(seethediscussioninDemsetz,Saidenberg,andStrahan, 1997;LaevenandLevine,2009).Insomecases,particularlyinthepresenceofsafetynetsubsidies, diversifiedoutsideequityholderswithsufficientstakesinbanksmayseektoincentivizemanagers totakemoreriskandmayemploypayschemesthatrewardrisktaking(seeLaevenandLevine, 2009;Cheng,Hong,andScheinkman,2013;BaiandElyasiani,2013). Alternatively,therecouldbestatesoftheworldinwhichbankmanagersthathadbeen managingriskcarefullymaybecomeexcessivelyrisklovingwithoutbeingprovidedperformance incentives,suchasstatesinwhichtheirequityinterestinthebankbecomessufficientlysmallasthe resultofbanklosses.Inthewakeoflossesthatreducenetworthandincreaseleverage,bankers withsubstantialequitysharesandcontrolrightsmayprefertoincreaseassetrisk.3Thismoral‐ hazardprobleminriskmanagementcanbemitigatedthroughvariousmeasures,includingshort‐ termdebtcontracting;afirst‐come,first‐servedruleforbankliquidation;andactionsbybankers thatcrediblysignalgoodriskmanagement,includingthemaintenanceofaminimumamountof cashorrisk‐freeassets(Acharya,Mehran,andThakor,2013;Bhattacharya,Boot,andThakor,1998; CalomirisandKahn,1991;Calomiris,Heider,andHoerova,2014;DiamondandRajan,2001).If bankdebtholdersareprotectedbydepositinsuranceorotherguarantees,however,moralhazard canbeexacerbatedbecausebankdebtholderslosetheirincentivetomonitorandcontrolbanks’ risktaking(Bhattacharya,Boot,andThakor,1998). Avarietyofempiricalstudieshavelookedatmanagerialownership,governance,andrisk taking.Anginer,Demirguc‐Kunt,Huizinga,andMa(2013)studyaninternationalsampleofbanks for2003‐2011andfindthatstrongerformalgovernancetendstobeassociatedwithlowerbank 3Insufficientlybadstatesoftheworld,bankersmayalsochoosetocommitfraudorabscond(Calomirisand Kahn,1991).GortonandRosen(1995)arguethat,inastateoftheworldinwhichindustryprospectsare declining,managersmayboostprofitstohidethepoorprospectsfromshareholders. 3 capitalratios,andthatmanagerswithlargestakestendtochoosehigherbankcapitalizationratios.4 Forsamplesofpubliclytradedfirmsmoregenerally,Holderness,Kroszner,andSheehan(1999) findthat,in1935and1995,highermanagerialownershipisassociatedwithlowerrisk. Followingtherecentfinancialcrisis,anumberofstudieshaveexaminedtheextenttowhich corporategovernanceandmanagerincentiveschemesinfluencedhowbanksfaredduringthecrisis (Acharyaetal.2009;Berger,Imbierowicz,andRauch,2012;EllulandYeramilli,2010;Fahlenbrach andStulz,2011;andSeniorSupervisorsGroup,2008).AsMehran,Morrison,andShapiro(2011) note,thesestudiesgenerallyfindthatmanagersprovidedwithincreasedincentivestotakerisk generallydidso. Althoughsomemodelsofriskmanagementfocusonconflictsofinterestbetweenbankers andcreditors,inothermodels,managerialconflictsofinterestaffectallsourcesofoutsidefunding sources.(ExamplesoftheformerincludeJensenandMeckling,1976;Myers,1977;andMerton, 1977,whileexamplesofthelatterareAcharya,Mehran,andThakor,2013;CalomirisandKahn, 1991;HolmstromandTirole,1997;andCalomiris,Heider,andHoerova,2014.Ageneralreviewof theissueisJohnandSenbet,1998.)Underthosecircumstances,commitmentstoimprovingthe bank’scorporategovernance–forexample,thepresenceofoutsidedirectorsontheboard (directorswhoarenotmanagersofthebank)orthe“bonding”ofmanagement–shouldimprove 4Theyfind,however,thatin2006,thepayoffofriskreversedthiseffectformanagerswithsufficientlylarge stockoptionwealth.LikeAnginer,Demirguc‐Kunt,Huizinga,andMa(2013),Saunders,Strock,andTravlos (1990)findvariationovertimeintheextenttowhichhighermanagerialownershipisassociatedwithmore orlessrisktaking.Saundersetal.findthatduringtheperiod1979‐1982greatermanagerialownershipwas associatedwithhigherrisk,whichtheyattributetothederegulatoryenvironment.Consistentwiththat interpretation,BrunoandClaessens(2010)showthatlegalregimesthatareexcessivelystrictcanbevalue destroying;bettercorporategovernancecombinedwithmoreflexiblelegalenvironmentscanleadto superioroutcomesthroughtheabilitytoundertakevalue‐creatingrisk. 4 managerialeffortinmonitoringandcontrollingthebank’sborrowers,andalsomitigatetheriskof bankerdefalcation,whichbenefitsbothoutsidestockholdersanddebtholders. Corporategovernancepoliciesofbanksshouldariseendogenously,inparttoreducethe costsrelatedtothetwosetsofconflictsofinterestinrisktaking–thatis,theconflictbetween shareholdersanddebtholders,andtheconflictbetweenmanagersandoutsiders(JohnandSenbet, 1998,HermalinandWeisbach,2003).Understandinghowgovernancepoliciesrespondtosuch conflicts,andwhateffectsownershipstructureandgovernancepolicieshaveonrisktaking,is highlychallenginginthecurrentregulatoryenvironment,inwhichpoliciessuchasdeposit insurance,too‐big‐to‐fail(TBTF)bailouts,andlegalrestrictionsoncontrollingownershipinterests inbanks,removethedisciplinaryincentivesofdebtholdersandlimittheabilityofequityholdersto concentrateownership(ontheeffectsofTBTF,forexample,seeAcharya,et.al2009).5 3.Researchapproach,samplechoice,andhypotheses Tounderstandhowownershipstructureaffectscorporategovernance,andhowownership structureandcorporategovernanceaffectbanks’riskmanagement,weexaminethelinksamong ownership,governance,andriskmanagementduringaperiodpriortotheestablishmentofa regulatorysafetynetforbanks.DuringtheNationalBankingEra(1863‐1914),government protectionwasabsent,andthelatitudeforvoluntarygovernancedecisionsbybankswasgreat.We observelargecross‐sectionaldifferencesintheownershipstructurechoicesofnationalbanks,as wellasgreatvariationintheirchoicesfororganizingcorporategovernance.Banksalsostructured theirportfoliosverydifferentlyfromoneanother,anddisplayedimportantdifferencesintheir managementofrisk–indicatedbothbybalancesheetratiodifferencesandthedifferencesinbank failureexperienceduringthemostseverebankingcrisisofthisera,thePanicof1893. 5Theso‐calledseparationofbankingandcommerceplacesspecialconstraintsonwhoispermittedto exerciseacontrollinginterestinabank. 5 Differencesinownership,governancechoices,portfoliosandrisks,underacommonand relativelylaissez‐faireregulatoryenvironment,makenationalbanks’experiencesinthe1890san ideallaboratoryforexamininghowmanagerownershipandboardoversightarerelatedtorent seeking,portfoliochoice,andfailurerisk.Anotheradvantageoffocusingonnationalbanksisthat, despitetheirdifferencesinlocation,theyoperatedunderidenticallegalconstraints.Forexample, nationalbanksallissuednationalbanknotesandaccepteddeposits.Nationalbanksmadeloansbut wereprohibitedfromundertaking“guarantees”(suchasbankersacceptances).Also,unitbanking lawsrestrictedeachbanktooperatingonlyoneoffice,whichlimitedbanksize.Forallthese reasons,nationalbanks’businessmodelsweresimilar(incontrasttotoday’sbankingsystem,in whichsmallbanksfocusonlendinganddeposittaking,whileglobaluniversalbanksundertakea muchwiderrangeofactivitiesforaquitedifferentcustomerbase). CorporategovernanceinthehistoricalU.S.bankingcontexthasbeenthesubjectof numerouspriorstudies.Oneofthemostimportantthemesofthatliterature,whichisnotpresent tothesameextentinothercontexts,hasbeentheconnectionbetweenstockownershipandthe recipientsofbanklending.Intoday’sbanks,therearestrictlimitsonloanstoofficersanddirectors, anditisconsideredinappropriatetoprovidebettertermstoloansofferedtoofficers,directors,or otherlargestockholders.HistoricallyintheUnitedStates,banksoftenactedas“loanclubs”for insiders,whowereoftenlargeshareholderswithsignificantformalorinformalcontrolrights. Generally,theliteraturehastakenabenignviewofinsiderlending,arguingthatitfacilitated valuecreationandriskmanagementbecauseinsidershadstrongincentivestoscreenandmonitor oneanother(LamoreauxandGlaisek,1991;Lamoreaux,1994;Meissner,2005;HaberandMaurer, 2007;HansmannandParglender,2012;Freeman,Pearson,andTaylor,2012).Bodenhorn(2013) findsthatbankvalueincreasedwiththenumberofindividualblockholders,butdeclinedwiththe numberofinstitutionalblockholders–thatis,blockholderswhowerenotpartoftheloanclub. 6 Loansclubsincreasedthevalueofbankstockbecauseinsidersvaluedpreferentialaccesstolending thatwasattachedtotheirblockholdingstatus.6 Asweshow,nationalbanks,likethestate‐charteredbanksstudiedbyLamoreaux, Bodenhorn,andothers,engagedinlargeamountsofinsiderlending.Ourdataonnationalbanks allowustoinvestigatehowdifferencesthataffectedownershipandgovernancerulesalsoaffected theamountandqualityofinsiderlending. Thedataweusecomeprimarilyfromnationalbanks’ExaminationReports,asourcewhich, toourknowledge,hasbeenlittleused,andneverusedforquantitativeanalysisofthequestionswe addresshere.TheseReportsprovideverydetailedpicturesofthebanksandthebankexaminers’ viewsaboutthem.TheydescribetheequityownershipofmanagersandofBoardmembers(and identifywhethermanagersareontheBoard).Thereissubstantialheterogeneitywithrespectto ownershipstructureamongthenationalbanksinoursample.Somemanagersownaconsiderable portionofthesharesofthebank,whileinothercasesthemanagersownonlyasmallfraction.We alsohaveinformationaboutavarietyofcorporategovernancemeasuresthatcouldbeusedto provideoversightofbankmanagers,suchasthefrequencyofboardmeetings,thenumberof outsidedirectorsontheboard,andwhethertherewasanindependentloanreviewcommitteethat includedoutsidedirectors.Wealsoknowwhetherthemanagerswererequiredtopostsurety bonds,whichprotectedotherequityholdersintheeventoffraud.Thereportsalsoprovideafairly detailedpictureofthebalancesheet,aswellascontainingtheexaminer’sassessmentsofvarious 6Interestingly,19thcenturycorporatecharteringrulesoftenemployedvotingrightsrulesthatreducedthe votingpoweroflargeshareholders,largelytoreduceconcentrationofcontrolovercorporations.Although thesedeparturesfromoneshare‐onevoteruleswerecommonformanyfirms,theywerelesscommonfor banks(Hilt,2008;Hilt,2013).Thismayhavereflectedthedesirabilityofencouraginginsiderblockholding,as wellastherelativeabsenceofthepoliticalconsequencesofcontroloverabankoncebanksbecamechartered freely(roughlyaroundthesecondquarterofthe19thcentury). 7 measuresofassetquality,forward‐lookingexpectationsofloss,andqualitativeevaluations.Thus, weareabletoexaminetherelationshipbetweenownershipandgovernancechoices,aswellasthe impactofbothonriskpreferencesatthebank.Therichnessofthedatapermitsustoprovidean integratedpictureofthelinkagesamongownership,governance,andfinancialstability. Forouranalysis,wegatherdatafrom206banksfrom37fairlylargecitieslocatedmainlyin theWesternandSouthernpartsoftheUnitedStates.7Thoseregionssawthegreatestfinancial turmoilandthehighestratesofbankfailureduringthePanicof1893.Byselectingallthenational banksfrom37similarcitiesintheseregionsweintendtoconstructasampleofreasonably comparablenationalbanksintermsoftheireconomicenvironmentandlendingactivities.We combinetheinformationintheExaminationReportswithstandardbalancesheetdatafromtheCall Reports(regulatoryreportswhichcontainedinformationonthebanks’balancessheetsfiledabout fivetimesayear),aswellasotherlocation‐specificcontrolsdrawnfromvariouscensusesandother sources.Weexaminethebanks’situationsintheearly1890s,justpriortothePanicof1893. Thispanicisausefulmomenttofocusuponbecauseitbroughtthemostseveredistressfor banksofanyofthecrisesduringtheNationalBankingEra.Thatepisoderesultedinthehighest numbersofbankfailuresofanyofthecrises,andwasoneofthreeepisodesduringtheNational BankingErathatwitnessedasuspensionofconvertibilityinNewYork.Althoughmostofthebanks 7ThesampleisregionallyfocusedontheWestandtheSouthduetoourinterestinobservinghowdifferent governancestructuresfaredinthepanicof1893(whichprimarilyproducedbankfailuresinthoseregions). Nevertheless,thestructuresweobserveinoursamplearesimilartowhatothersreportforthetime.For instance,BodenhornandWhite(2014)findthatcorporateboardsofstatecharteredbanksinNewYork,a quitedifferentsample,wereofsimilarsizetotheonesweobserveandusedsimilaroversighttools.Hilt’s (2014)descriptionoftheevolutionofcorporateboardsintheUnitedStatesfromtheearly1800salso suggeststhatthecorporateboardsinoursamplewereinlinewithgeneralpracticesofthetime.Giventhese similarities,webelieveourfindingsarelikelytobegenerallyapplicabletobanksintheperiod. 8 inoursampleavoidedfailure,therewasenoughfailureriskduringthisepisodetoprovide substantialobservablecross‐sectionalvariation–somethingthatisabsentduringmostofthe NationalBankingEra. Welookfirstattheinterplaybetweenownershipandgovernancebygaugingtheextentto whichthestructureofownershipaffectsbanks’choicesofcorporategovernancepolicies.We hypothesizethathighermanagerialownershipreducestheextentofformalcorporategovernance (welabelthisHypothesis1).Wetreatmanagerialownershipasanendogenousvariableand instrumentmanagerialownershipusingmanagerialturnoverevents.(SeeHermalinandWeisbach (2003)foradiscussionofendogeneityissuesrelatedtothestructureofboardsandmanagement control.)Thepatternsweobservearerobustacrossspecificationsthatweighdifferentaspectsof corporategovernancedifferently,andwereportresultsforasimplescorethatgivesequalweight toeachdimension.Allfiveoftheformalcorporategovernancepolicychoicesweconsiderare mutuallypositivelycorrelatedwithoneanother,andeachofthemisnegativelycorrelatedwiththe degreeofmanagerialownership–factsthatareconsistentwithviewingmanagerialownership concentrationasasubstituteforformalgovernancetoolsinresolvingconflictsofinterestbetween managersandthesourcesofoutsidefundingforthebank. Takingintoaccounttheendogeneityofgovernanceandownershipstructure,weexamine therentseekingbehaviorofbankersthatalternativelychoosetosolvetheirgovernancechallenges throughconcentratedownershiporformalcorporategovernancepolicies.Weexpect(Hypothesis 2)thatgreatermanagerialownershipresultsingreateropportunitiesformanagerialrent extractionthroughhigherwagesandloanstomanagersandalsohigherdividends(giventhat managers’interestsarecloselyalignedwithshareholders).Wefindthatmanagers’salariesrelative toassetstendtobehigherwhentheyownagreaterportionofoutstandingstock,whichreflects theirgreaterabilitytoextractrents.Interestingly,thetotalproportionofloansmadetoinsiders (officersandoutsidedirectors)isnotaffectedbythestructureofownershiporgovernance,but 9 ownershipandoversighthaveastrongimpactonwhichinsidersreceivethoseinsiderloans.When managershavegreaterequityownership,moreinsideloansareallocatedtothem;whenoutside directorsexercisegreateroversightincorporategovernance,agreaterproportionoftheinside loansarereceivedbythem.Greatermanagerialownershipalsoresultsinhigherdividends. Wealsoconnectownershipstructureandcorporategovernancechoicestobanks’risk preferencesandtheirbalancesheetchoices.Inparticular,weshowhowownershipand governanceaffectedbankportfoliostructure,performance,andfailureprobabilitiesduringthe Panicof1893.Wehypothesizethat,astheresultofriskaversion,greatermanagerialownership (whichreducesthediversificationofmanagerialwealth)resultsinlowerbankrisktaking (Hypothesis3). Ourresultsonrisktakingindicatethatmanagerswhoownagreaterproportionofthe bank’sstocktakelessriskaccordingtoanymeasureofriskweemploy.Forexample,withrespect toforward‐lookingmeasuresofrisk,managerswithlargeequitystakesintheirbanksarelesslikely torelyuponhigh‐cost“borrowedfunds”andarealsolesslikelytobeinvolvedinrealestate lending.Bothactivitieswereperceivedbycontemporariesasindicativeofhighriskandsuch perceptionsaregenerallyborneoutinthePanic.Moreover,wefindthatanoutsideestimateofthe riskinessofthebank’soverallportfolio,theforecastedlossesanticipatedbythebank’sexaminer, paintsasimilarpicture;greatermanagementstakesareassociatedwithlowerexpectedlosses.We viewtheseresultsasconsistentwiththeideathatmanagerswithalargershareoftheirwealth investedinthebankweremorerisk–averseintheirriskmanagementpractices.Banksthatchose lowermanagerialstakesandmoreformalgovernancepoliciestendedtoundertakegreaterlevelsof risk.Thatfindingisconsistentwiththeviewthatoutsidedirectors,whorepresenttheinterestsof diversifiedoutsideequityholders,preferaslightlyhigherlevelofrisk.Thepreferenceforlower riskappearstohavebeenbeneficialduringthePanicof1893aswefindthatbankswithhigher 10 managerownershipandlessrelianceonformalcorporategovernancewerelesslikelytofail.This effectisduelargelytohowthesebanksstructuredtheirbalancesheets. Banksseekingtoreducetheriskofdefaultontheirdebtscanusetwoalternativerisk managementtoolsincombination:ahighercash‐to‐assetratio(ontheassetsideofthebalance sheet),orahigherequity‐to‐assetratio(ontheliabilitysideofthebalancesheet).8Wehypothesize (Hypothesis4)thatbankswithgreatermanagerialownershipandlessrelianceonformalcorporate governancewillchoosetorelymoreoncashassetholdingstoreducetheirdefaultriskbecause holdingcashcompensatesfortheabsenceofoutsideoversightofriskmanagementinensuring propermanagerialriskmanagement(Calomiris,HeiderandHoerova,2014).Insomestatesofthe world–ifhiddenlossesarelarge,makingresultingmanagerialstakesunobservablysmall– managersmayprefertoundertakemoreriskthanoutsideshareholdersanddepositors(e.g.,when managerialeffortatlimitingriskiscostly).Forthatreason,theabsenceofformalcorporate governanceoversightofmanagementmaygiverisetotheneedformorecashholdings,asawayto preservethecredibilityofriskmanagementduringbadtimes.9Banksthatlackformalgovernance 8Allnationalbankswerecharteredwithdoubleliabilityrequiredforstockholders.Thatmeantthat stockholderscouldbeassessedforanadditionalamountequaltothepaidincapitalofthebankifthebank wereplacedintoreceivership.Consequently,theactualamountofequitythatactedasabufferagainstloss, fromthestandpointofbankdepositors,wasmorethantheamountofpaidincapitalandaccumulated retainedearnings.Totakeaccountofdoubleliability’seffectonbankrisk,weexperimentedwithincluding theratioofpaidincapitaltototalnetworthinourregressions.Wefindthatthisvariablewasnotstatistically significant,anditsinclusiondidnotalterourotherfindings.Intheresultsreportedhere,weexcludeit. 9Thisresultreliesonthefactthatcashdepositsinotherbanks,unlikeequity(whichisanaccounting categorysubjecttomanipulation),areobservable(andwereactuallyverifiedbyexaminers),andmost importantly,cashdepositsarenotsubjecttoriskshiftingbythebankthatpossessesthem.Therelative relianceoncashorequitymayalsoreflectadverse‐selectioncostsofissuingequity(asinMyersandMajluf 1984).CalomirisandWilson(2004)showthatbankspreferlessonequitywhentheyfacehigheradverse‐ 11 willsufferfromgreaterassetsubstitutionrisk,andgreateradverse‐selectionproblems(ifthey weretoattempttoraiseadditionalsourcesofoutsideequity).Greaterassetsubstitutionriskfrom alackofformalgovernancethereforewilltendtoleadbankstorelymoreoncashasameansof signalinggoodriskmanagementpractices.10Wefindthatbankswithhighermanagerialownership concentrationreliedmoreoncashassets,andlessonequity. Tofacilitatetheexpositionofthevarioustestsofthesefourhypotheses,Table1 summarizesthemandthetestsweperforminourempiricalwork.Theremainderofthepaperis organizedasfollows.Section4discussesthedatasourcesandthesample.Thatsectionalso explainstheconstructionofourcorporategovernancemeasuresandthevariablesweuseas indicatorsofrentseekingandriskpreferences.Thebaselineempiricalanalysisiscontainedin Section5whileavarietyofrobustnesschecksarepresentedinSection6.Section7concludes. selectioncostsofraisingequityinthemarket.Asinthecaseofriskmanagementincentiveproblems,adverse‐ selectioncostsofequityissuesalsoimplyagreaterrelianceoncash.Furthermore,weexpectproblemsof asymmetricinformationtobemitigatedbytheuseofboardoversight.Boardmeetings,aloanreview committee,andbondingshouldbeassociatedwithgreatertransparencyandhencelowercostsofequity offerings.Althoughwebelievethatasymmetricinformationmitigationwasanadditionalbenefitofformal corporategovernance,wedonotemphasizethischannelinourempiricalcross‐sectionalanalysisbecauseof theabsenceofequityofferingsforallthebanksinoursample,whichindicatesthatdifferencesinadverse‐ selectioncostsareunlikelytoexplaincross‐sectionaldifferencesinbanks’useofcash. 10Usingrulestorequirebankstoholdaportionoftheirassetsascashtocontrolassetsubstitutionriskisthe maintheoreticalconclusionofCalomiris,HeiderandHoerova(2014).Acharya,Mehran,andThakor(2013) alsosuggeststhatforcingbankstoholdastockofsafeassets(suchasTreasurysecurities)isusefulfor controllingrisk‐shiftingincentives,althoughtheysuggestthatsuchholdingsshouldbeproportionaltoequity ratherthantotalassets.Mostpreviousworktendedtofocusontheuseofcapitaltomanageincentiveswith cashandsafeassetsviewedasrelatedonlytoliquiditymanagement(see,forinstance,thediscussionin Bhattacharya,Boot,andThakor,1998). 12 4.Datasourcesandvariabledefinitions WegatheravarietyofinformationonindividualbanksusingCallReportsandExamination Reports.Herewedescribethedatasourcesandthedefinitionsofthevariablesusedinthisstudy. 4.1.Thesample Oursamplecontains206bankinginstitutions,whichconsistsofallthenationalbanks locatedin37cities.AsweusedinformationfromboththeCallReportandexaminerreports,and willultimatelybeinterestedinhowbanksfaredduringthePanicof1893,weneedbankstohave bothreportsavailablepriortothePanic.Thus,tobeincludedinoursample,thebanksneededto haveprovidedinformationfortheSeptember1892CallReportandtohavehadatleastone ExaminationReportcompletedpriortoMay1893(theonsetofthePanic).ThosearetheReports thatprovidetheinformationusedfortheanalysis.11 The37citiesincludemanyofthelargeronesintheWesternandSouthernpartsofthe UnitedStates.12Anumberofthemweredesignatedas“reservecities”forpurposesofregulatory cashreserverequirements.Depositsheldatbanksinreservecitiescouldcountaspartofa “country”bank’slegalreserveandthesedepositsoftenservedaspartoftheregionalpayment system(seeJames1978forfurtherdetail).Someoftheothercities,eventhoughtheywerenot 11TwobanksfiletheSeptember1892CallReportbutclosepriortoMay1893.Fortheseinstitutions,weuse theexaminationreportnearestclosure,solongasitwasfiledatleastfourmonthspriortoclosure. 12Thecitiesare:Birmingham,AL;Mobile,AL;SanDiego,CA;LosAngeles,CA;Denver,CO;Pueblo,CO; Indianapolis,IN;DesMoines,IA;Dubuque,IA;Lexington,KY;Louisville,KY;NewOrleans,LA;Minneapolis, MN;Rochester,MN;St.Paul,MN;Stillwater;MN;KansasCity,MO;St.Joseph,MO;Helena,MT;Lincoln,NE; Omaha,NE;Albuquerque,NM;Fargo,ND;Cincinnati,OH;Portland,OR;Knoxville,TN;Memphis,TN; Nashville,TN;Dallas,TX;ElPaso,TX;SanAntonio,TX;SaltLakeCity,UT;Spokane,WA;Tacoma,WA; Milwaukee;WI;Racine,WI;andCheyenne,WY. 13 technicallyreservecities,wereimportantenoughregionallythatotherbankshelddepositsthere. Thus,manyofthebanksinoursampleplayedimportantrolesasintermediariesininterbank markets.Nevertheless,oursampleincludesanumberofbanksfromsmallercitiesaswell. 4.2.Primarydatasources TheExaminationReportsprovideawealthofinformation.13Themostvitalmaterialforour purposesisthedetailedinformationregardingtheextentofownershipbythebank’smanagement anditsboard,aswellastheinformationaboutcorporategovernancepractices.TheExamination Reportlistsallthebankdirectorsandmajorofficers(President,Vice‐President,Cashier),the numberofsharesheldbytheseindividuals,andanyloanstotheseindividuals.Salariesofthe officerswerenotedandwhethertheofficerswererequiredtoputupasuretybond.Theexaminer alsocommentedonwhethertheboardexercisedanyoversightoftheofficers. TheExaminationReportsalsoconsideredavarietyofaspectsofthebalancesheetbeyond whatwascoveredbytheCallReport.Thatinformationincludedadditionalquantitativedetail abouttheloanbook,suchastheamountofloansthatweredemandortimeloans,theamountof loanssecuredbyrealestate,andtheamountsecuredbyothercollateral.14Therewasalso informationonthebank’sliabilitiesincludingadditionaldetailonwhetherthebankborrowedfrom otherbanks(aformofhigher‐interest,short‐term,“hot”money). Finally,theexaminerprovidedinformationonbankperformance,whichcombinedhard factsaboutthebankwithhisownjudgments.Specifically,theexaminersevaluatedthequalityof 13CalomirisandCarlson(2014)provideadetailedsummaryofthecontentsoftheExaminationReports.See alsoRobertson(1995)formoreinformationontheexaminationprocess. 14Althoughrealestatelendingwas“prohibited”bynationalbanks,nationalbanksnonethelessfoundwaysto lendagainstrealestate.Aloanmadewithoutrealestateascollateralcouldbecomecollateralizedbyreal estateifthecreditworthinessoftheborrowerdeteriorated. 14 theloanbookbylistingthevolumeofslowandoverdueloansandprovidinganestimateof expectedlossesonthebanks’assets–whichincludedloans,aswellasotherassets.Theexaminers alsonotedtheamountanddateofthemostrecentpaymentofdividends,aswellaswhetherfunds thatwereretainedwouldcovercurrentandfuturelossesorbuildupthebank’snetworth. Formostofourbalancesheetdata,weuseinformationfromtheSeptember1892Call Report.WhilesomeadditionalinformationisavailableontheExaminationReport,theCallReport hastheadvantageofprovidingdataforallthebanksatthesamepointintime,whichreduces concernsaboutspuriousdifferencesduetoseasonalorothertime‐relatedfactors. Wealsoincludeanumberofvariablesrelatedtotheeconomicenvironmentinwhichthe bankoperated.Theseincludecountylevelvariablesfromthevariouscensuses,suchaspopulation andtheshareofincomefromagriculture. Allvariables,theirdefinitions,andtheirsourcesappearinTable2.Summarystatisticsfor thesevariablesappearinTable3. 4.3.Ownershipandgovernancevariables Theindividualsmostresponsibleforrunningthebankwereitsseniormanagers,in particularthepresident,vice‐president,andcashier(essentially,thechiefoperatingofficerofthe bank).Theyplayedalargeroleinmakingloansandarrangingthefundingofthebank.These individualstendedtoownsharesinthebankandwerefrequentlyalsoontheboardofdirectors (thePresidentofthebankwasrequiredbylawtoserveontheBoard).Akeyvariableinour analysisistheshareofthebank’sstockownedbytheofficersofthebank.Wefocus,inparticular, onthefractionofoutstandingbanksharesownedbythepresident,vice‐president,andcashier.15 15Weobtainthenumberofoutstandingbanksharesbydividingbankcapitalby100(asbankcapitalwas typicallycarriedatbookvaluebasedonsharepricesof$100pershare).Inafewcasestheexaminer indicatedthenumberofsharesoutstandingandthesereportsconfirmthatourprocedureiscorrect.Inafew 15 Theaverageportionofsharesownedbythesethreeofficers,asreportedinTable3,was25percent. ThehistograminFigure1providesabetterindicationofthedistributionofmanagerialownership. Atmanybanksinthesample,ownershipbythemanagersisfairlymodest;thethreetopmanagers ownedlessthan6percentofoutstandingsharesforabout30percentofthesample.Therearealso casesofsignificantownershipconcentration;thetopthreemanagersownedatleasthalfthe outstandingsharesinnearly10percentofthesample. ThebehaviorofmanagerscouldbeconstrainedbytheBoardofDirectors.Boardsrangedin sizefrom4membersto23members.WhilethePresidentofthebankwasalwaysontheBoardand otherbankofficersfrequentlysatontheboard,themajorityoftheboardmembersconsistedof outsidedirectors—individualsthatwerenotofficersorotheremployeesofthebank.Someoutside boardmembersownedsignificantstakesinthebank.Anumberwereprominentbusinessmenthat mightprovidebusinesstothebank.16Ahistogramofownershipbyoutsidedirectorsisshownin Figure2.Theaverageportionofsharesownedbyoutsidedirectorswas15percentbutitreached ashighas57percent.Presumably,thelargertheportionofsharesownedbytheoutsidedirectors, themoretheycouldinfluencethebehaviorofmanagers.Theownershipbyallotherindividualsis showninFigure3.Ascanbeseenfromthisfigure,individualswhoareneithermanagersnorbank managersownamajorityofthesharesinabouttwo‐thirdsofthebanksinoursample. TherewerealsootherwaysthattheBoardcouldexertcontrolovermanagers.Thebank couldmaintainanactiveindependentdiscountcommittee(definedbyregulatorsasonecontaining atleastoneoutsidedirectortoreviewandapproveloansproposedbythemanagers).Sucha committeewasmaintainedby60percentofbanks.Anotherwayofexertingcontrolwasby meetingfrequently.Boardsthatmetinfrequently,suchassemi‐annually,presumablyexertedlittle othercasestheexaminerreportedthatthevalueofcapitalhadpreviouslybeenwrittendownandshares revalued.Webelievethatwehavemadealltheappropriatecorrectionsforthesewrite‐downs. 16Forinstance,aMr.ProctorandaMr.GambleservedontheboardoftheCitizensNationalinCincinnati. 16 influence.Theboardmetmonthlyormorefrequentlyinnearlytwo‐thirdsofthebanksinour sample.Incasesinwhichthemanagerscomprisedasignificantportionoftheboard,therewas presumablylittleindependentoversight;whenoutsidedirectorsdominatedtheboard,theycould presumablyexertmorecontrol.Inoursample,themedianportionoftheboardthatconsistedof outsidedirectorswas71percent;wecreateanindicatorvariableequaltoonewhentheportionof directorsareoutsidersisabovethemedianandiszerootherwise.Ourmeasuresoftherelianceon independentdirectors,oftheexistenceofaloanreviewcommittee,andofthefrequencywith whichitmetaresimilartoothermeasuresusedtoanalyzecorporateriskmanagementinmodern financialinstitutions,suchasthe“activeboardriskcommittee”ofEllulandYerramilli(2010). Anotherwayofinfluencingbankmanagementwasrequiringmanagerstopostsurety bonds.Thesebondswouldofferthedirectors(orreceiver)awayofrecoveringfundsintheevent themanagercommittedsomespecifiedact,typicallysometypeoffraudthatcausedlossestothe bank.Bondscouldbepersonalorprovidedthroughasuretybondagency(whichoftenrequired thatthepersonbeinginsuredpostsometypeofcollateral).17Suretybondsweremostoften requiredforthecashier,whooversawthebooksandforwhomthepossibilityoffraudwas thereforehighest(nearly60percentofcashierspostedbonds).Othermanagersalsowererequired topostsuchbonds(thePresidentpostedabondin33percentofoursampleandthevice‐president didsoin12percentofthesample). InFigure4,weillustratetherelationshipbetweenmanagerownershipandoneofthe indicatorsofcorporateoversight:thefractionoftheBoardconsistingofoutsidedirectors.The negativerelationshipbetweenthesetwomeasuresindicatesthatmoremanagerownershiptends tobeassociatedwithlessformaloversight(Hypothesis1).Moreover,notonlyareeachofthe measuresofBoardoversightnegativelycorrelatedwithmanagerownership,asshowninTable4, buttheyareallpositivelycorrelatedwitheachother.Althoughweinvestigatedtheimpactofeach 17FormoreinformationonsuretybondsseeLunt(1922). 17 ofthesemeasuresofBoardcontrolonmanagerialbehavior,itisusefulforourpurposestocreate anindexthataggregatesthedifferentmeasuresintoasinglecorporategovernanceindex.Wedoso bysummingthefiveindicatorvariables.18 Thecombinedgovernancescoreisstronglynegativelycorrelatedwiththemanagement ownershipshare,andthepartialcorrelationbetweenthosevariablesislargeandstatistically significanteveninthepresenceofmanycontrolvariables,asshowninTable5,whichreportsan OLSregressionthattreatscorporategovernanceasthedependentvariable.Ofcourse,ownership shareisnotexogenousand,therefore,Table5doesnotpermitastructuralorcausalinterpretation. Wereturntoestimatethejointrelationshipamongownershipshare,governance,andother endogenousvariablesbelow. Examinersseemtohaveunderstoodthatbankscouldachievegoodmanagementofrisk withorwithoutactiveoversightofmanagementbytheBoard.Belowareexcerptsfromthe ExaminationReportsoftwobanks,onewiththeminimumcorporateoversightscoreof0andthe otherwiththemaximumscoreof5.Inneithercasedidtheexaminerhaveconcernsaboutthe managementofthebankorthesoundnessofthebank,eventhoughtheexaminerwasawareofthe cleardifferencesintheoversightbeingexercisedbytheBoard. Oversightscoreof0‐CommentontheBoard: Frequentmeetingsarenotheldbythedirectorsofthisbankandrecordsonlyshow thatformalmeetingsareheldtodeclaredividends.Nomentionbeingmadeoftheir having examined or approved loans and discounts at such times, and there is no report of discount and examining committee having acted. The management is apparentlywithMr.Gates,thepresidentofthebank. CommentontheOfficers: Officers are capable, prudent and of good reputation and their management is efficientandsuccessful,thatmanagementbeinginthehandsofMr.HenryGatesthe 18Wealsotriedaggregatingthefiveindicatorsbytakingthefirstprinciplecomponent,similartoElluland Yerramilli(2010).Allthefiveindicatorshadpositiveandroughlyequalweights.Thus,thefirstprinciple componentwasnotsodifferentthanthesimpleaveragesowestickwiththeaverageforsimplicity. 18 presidentwhohashadover30yearsexperienceinthebankingbusinessinthiscity. Nobondsrequired.19 Oversightscoreof5‐CommentontheBoard: Directors meet monthly. Minutes full and explicit. Have discount board and examiningcommittee.Discountboardpass[es]uponallloans. CommentontheOfficers: Officersarecapable,prudent,ofgoodreputation.Theirmanagementsuccessful;the bondsarefurnishedbyLouisvilleBondCo.andincustodyofLexingtonTrust.20 4.4.Financialandportfoliomeasures Anumberofmeasuresarepotentiallyofinterestascontrolsinourregressionsanalyzing ownership,governanceandriskmanagement,whileothersserveasendogenousvariablesthatwe analyze(i.e.,thecashassetsratioandtheequity‐to‐assetsratio).Twopotentialcontrolvariables arebanksizeandbankage.Smallerbanks,ceterisparibus,generallywillbemorecloselyheld eitherbecauseoflimitedbankmanagerwealthorbecauseoffixedcostsinestablishingformal governanceprocedures.Theproblemwithusingbanksizeasacontrolvariable,however,isthat sizeisitselfanendogenousvariablethatislikelytoreflectchoicesthatarecorrelatedwiththe endogenousvariablesbeinganalyzed.Forexample,asweshowintheappendix,inasimplemodel ofriskmanagementeffortchoice,foragivenlevelofmanagerialwealth,themaximumsizeofthe bankwillbeconstrainedunlessthebankadoptsformalcorporategovernancetools.Thus,rather thanincludingassetsizeinourregressions,weincludethepopulationsizeofthecityinwhichthe bankislocated,whichisanexogenousinfluenceonbankassetsizethatisnotaffectedbycorporate governancechoice. Bankriskmanagementpracticesmayalsoreflectheterogeneityinbankgoalsorexperience. Youngbanksmaylackexperience,ormayhavedifferentpreferencesormorelimitedopportunities. 19ExaminerreportofNovember14,1892fortheNebraskaNationalBankofOmaha,NE,charter2665. 20ExaminerreportofAugust18,1892fortheFayetteNationalBankofLexington,KY,charter1720. 19 Tocaptureanyoralloftheseinfluences,wecontrolforbankage.Ourmeasureisthelogofthe numberofyearssincethebankwasestablished(thiscouldbethedatethebankbecamenationally chartered,orthedateitwasfounded,dependingonwhetheritwasaconversionofastatebank).21 Wehaveconsiderableinformationaboutassetportfolios.Loanswereobviouslyarelatively riskyassetbutalsoarelativelyhigh‐earningasset.Onebasicandoften‐usedassetratiothat capturesbothriskandearningpotentialistheshareofassetsconsistingofloans.TheExamination Reportsprovideadditionalinformationabouttheloanportfolio.DuringtheNationalBankingEra, realestateloanswereconsideredriskierloans.Nationalbankswerenotsupposedtooriginate mortgages;however,theywereallowedtohavemortgagesloansiftherealestatewasbeingusedto collateralizeapreviouslyexistingloan.Thus,weemployrealestateloansrelativetototalloansasa measureoflendingrisk.Wearealsointerestedinthedegreeofinsiderlending.Weconstructtwo measures:theshareofallloansthataremadetoinsiders(whetherboardmembersormanagers) andtheshareofloanstoinsidersthataremadetomanagersratherthanoutsidedirectors. Previousresearchonbankriskmanagementhasidentifiedtheproportionsofdifferent typesofbankdebtsasanimportantindicator(CalomirisandMason,1997;2003;2008;Calomiris, MasonandWheelock,2011;Carlson,2010).Abank’sliabilitystructuremayreflectexogenous liquidityrisksfacedbybanks(e.g.,ahigherproportionofcheckingdeposits).Liabilitystructure alsomaycaptureendogenouschangesinthecompositionofdebtsinreactiontochangesin unobservedcharacteristicsofbanks’assetrisks(e.g.,banksthatrelyonborrowedfundsmayfindit hardtoraisefundsfromothersources),andweincluderelianceonhigh‐interestrateborrowed 21Ageturnsouttobeahighlysignificantcontrolvariableinmanyoftheregressionsreportedbelow, althoughitsinterpretationisnotobvious.Intheregressionsreportedbelow,greaterageisassociatedwith highermanagerialownershipandlowerformalgovernance,withlowerdefaultrisk,withloweruseofequity andgreateruseofcash,andwithlowermanagerialsalariesandhigherdividends. 20 fundsasanendogenousvariableinouranalysisasanindicatorofrisk.22Insomespecifications,we includetheproportionofliabilitiesconsistingofindividualdepositsandtheproportionofdeposits incheckingdeposits,asopposedtosavingsortimedeposits,ascontrols. Wehavesomepotentiallyusefulinformationabouttheearningsandexpensesofthebanks inoursample.OneoftheexpenseslistedintheExaminationReportsisthesalariespaidto managers.Aslargerbankstendtopayhighersalaries,wescalesalariesbytheassetsofthebank. Wealsoobservedividendpayments.Weanalyzedividendsasadependentvariable;highdividend paymentsaresometimesviewedasanindicationofadisciplinedcorporategovernance environment.Ofcourse,dividendsalsoreflectdifferencesinprofitability;thatis,theymaybeused tosignalmanagement’sbeliefthatearningswillpersist.Also,dividendpaymentdifferencesmay reflectdifferentgrowthopportunities;retainingprofitsraisestheamountofequityinvestedinthe bank,whichceterisparibus,lowersthebank’sdefaultrisk,andthusincreasesthecapacityofthe banktogrowitsassets.Toanalyzedividendpayouts,weconsidertheratioofdividendpayments relativetosharesoutstandingifdividendswerepaidduringthepastsix(bankstypicallypaid dividendssemi‐annually,inJuneandDecember).Asanalternative,wealsolookedatwhetherthe bankpaidoutdividendsduringthepastsixmonthsandobtainedfairlysimilarresults. 4.5.Risk Weconsiderseveralindicatorsofcontributorstothedefaultriskofthebank.Someofthese areitemsnotedearlier,suchasmakingrealestateloansorrelyingonborrowedfunds.Wealsouse anoutsideexpert’sviewonthequalityoftheassets.Theexaminersprovidedestimatesoflikely 22Oftenthisborrowingtooktheformofrediscountingnotesorhavingbillspayable,butcouldalsotakethe formofcollateralizedcertificatesofdeposit.WhiletheformerarenotedontheCallReport,thelattertypeis notedonlyintheExaminationReports.Astheamountsarenotalwaysnoted,weinsteaduseanindicatorfor whetherornotthebankmadeofthis“hot”money. 21 lossesonassets(notjustloansbutonsecuritiesandotheritemsaswell,suchasnon‐income generatingassetssuchasfurnishings)andweusetheforecastoflossesrelatedtoassetsasa measureofrisk.Wealsomeasuredefaultriskbasedonoutcomes,inparticularwhetherthebank wasforcedtocloseitsdoorsbetweenOctober1892andDecember1893. Thetwoprimarytoolsofriskmanagementforbanksweretheequity‐to‐assetratioandthe cashassets‐to‐total‐assetsratio.Equity,ornetworth,ismeasuredasthesumofpaidincapitalplus cumulativeretainedearningsheldassurplusorundividedprofits.Therewerenoequityratio requirements,althoughbankswererequiredtomaintainminimumamountsofcapitalandsurplus. Estimatingthedemandforcashassetsiscomplicatedbythelegalminimumrequirementsofcash relativetodeposits.Cashreserverequirementsspecifiedacertainlevelofcashanddepositsin reservecitybanksrelativetodepositsandnetduetobanks.Aswefindinouranalysisofreserve holdings,however,regulatoryconstraintsonholdingsofcashreserves(discussedinmoredetail below)didnotappeartobebindingonbanks’demandsforcashassets. 4.6.Othercontrolsandinstruments Wealsoincludeanumberofvariablestocontrolforlocalconditions.Asnotedabove,we includethepopulationofthecityasreportedonthe1890census.Atthecountylevel,wegather otherinformationfromthe1890censusontheeconomicenvironment,includingforexample,the shareofcountyincomefromagriculture.Atthestatelevel,weincludeanindicatorforsignificant (greaterthan$1millionin1891)goldandsilvermininginthestateasreportedinthe1892 StatisticalAbstractoftheUnitedStates,andalsoforwhetherthestateachievedstatehoodatan earlydate.Inincorporatingsuchinfluences,weimaginethatcorporategovernancepracticesmay beinfluencedbysectoralcompositionandbythematurityofthelocaleconomy. Aswediscussfurtherbelow,weemployedtwopotentialmeasuresasinstrumentsfor ownershipstructureandgovernancechoices.Ourfirst,andprimary,instrumentisanindicator 22 variablecapturingachangeintheidentityofthepresidentofthebank,whichweidentifiedusing theExaminationReports.Weinvestigatethecausesofmanagerialturnoverbysearchingonline newspaperandothersourcesforinformationaboutthecircumstancesthatgaverisetothose changes.Thesecondinstrumentisthelogofthenumberofotherbanks(national,state,savings, andprivate)operatinginthesamecityasthesubjectbank.Thissecondinstrumentisusedonlyin somerobustnesstestsdescribedbelow.Weconsidertheexclusionrestrictionforthissecond instrumentasmorecontroversial,andthuswedonotincludeitinourbaselinespecifications. AnimportantfeatureofthebankingsystemduringtheNationalBankingErawasthe systemofinterbankdepositingofreserves.Nationalbankswererequiredtoholdcashand interbankdepositsagainsttheirowndepositliabilities.Banksoutsidemajorcitiesneedtoholda15 percentreserve,three‐fifthsofwhichcouldbeheldasdepositsatbanksinlarger“Reserve”cities or,“CentralReserve”cities—NewYork,Chicago,orSt.Louis.BanksinReservecitiesneededtohold a25percentreserve,halfofwhichcouldconsistofdepositsina“CentralReserve”city.Depositsin NewYorkplayedakeyroleinthesettlingofinterregionalpayments.Manybankshelddeposits withbanksinNewYork.Moreover,banksinNewYorkprovidedasubstantialamountofinterbank loansthroughrediscounting.TocapturethepotentialimportanceofproximitytoNewYorkin affectingbanks’risksandoperations,weincludethelogofthedistanceofbanksfromNewYorkas acontrol.Wealsoincludeanindicatorforwhetherthecityinwhichthebankislocatedisareserve city,tocapturethepossibleeffectsofdifferencesininterbankrelationshipsandreserve requirementsonbankbehavior. 5.Analysis Weareinterestedinhowexogenousbankcircumstancesinfluenceownershipand corporategovernancechoices,andrelatedbankbehavior,asdescribedinourfourHypothesesin Table1andourtheoreticaldiscussionofthosehypothesesabove.Asthesevariablesareclearly 23 inter‐related,westartbypresentingourapproachtoidentifyingthelinkagesamongownership structure,governancechoices,rentseeking,andriskmanagement.Wethenreviewourfindings. 5.1.Inter‐relatedownershipandcorporategovernancemeasures Ourinitialempiricalapproachtoidentifyingtheeffectsofownershipandgovernance choicesonbankbehavioremploysatwo‐stepprocedure,whichtreatsownershipstructureand corporategovernancepracticesasendogenous.Formostofthevariablesweanalyze,we instrumenteithermanagerialownershiporcorporategovernancescoreusingeventsassociated withmanagerialturnover.23Weexpect(andfindinTable6)thatamanagerialturnoverevent(such asthedeathofabankpresident)isassociatedwithareductioninthemanagerialownershipshare ofthebankandanincreaseincorporategovernance. Toverifythatmanagerialturnoveristraceabletoexogenousevents,weperformedweb‐ basedsearches,andalsosearchedthroughnewspapersavailablethroughthevariousdigitized searchenginesmaintainedbytheLibraryofCongress,tofindinformationaboutthechangesin bankpresidentsbetween1882and1892forbanksinoursample.Weusedboththebanknames andthepresidents’namestoobtaininformationaboutthereasonformanagerialturnover.Because thesourcescoveredbythesedigitaldatabasestendtobebiasedtowardlargercities’newspapers andnationalpublications,wewerenotabletofindinformationaboutmanyofthesemanagement changes.Forthe137relevantturnovereventsinoursample,wefindinformationexplainingthe reasonforthemanagementchangefor37oftheevents.For65oftheeventsforwhichinformation waslacking,wewereunabletolocateanynewspapersfortherelevanttimeperiodandlocation. For35oftheeventsforwhichinformationwaslacking,localnewspapersfortherelevanttime periodwereavailable,butwewereunabletofindanystoryaboutthechangesinbankpresidents. 23Asweexplainfurtherbelow,forsomevariablesofinterest,managerialturnoverdoesnotsatisfythe exclusionrestrictionsforavalidinstrument. 24 Managerialturnovergenerallywasassociatedwithclearlyidentifiableexogenousevents.In the37caseswewereabletotrace,thecausesofturnoverincludeddeathorsevereillness(23 cases),electiontopublicofficeorothernewcareeropportunity(9cases),retirement(2cases),and otherapparentlyexogenouscircumstances(onedepartureinthewakeofacashierembezzlement, onebecauseofbusinessproblemsunrelatedtothebank,andonebecausethepresidentdeclined re‐election).Wealsocheckedfornotablechangesintheconditionofthebanksasindicatedby changesinthecapitalstockaroundthetimethepresidentchanged.Wefindnoevidencethat changesincapitalsystematicallypreceded,followed,orwerecoincidentwithturnover.Inour regressions,theturnoverinstrumentismeasuredasthenumberoftimesthePresidentofthebank changedbetween1882and1892.InadditiontoreportingIVresults,wealsoreportnon‐ instrumentedOLSorprobitresultsforpurposesofcomparison,whichprovideestimates interpretableonlyundertheassumptionsthatmanagerialownershipandcorporategovernance choicesareexogenouslygivenwithrespecttotheotherendogenousvariablesweanalyze. 5.2.Corporategovernance,balancesheetcomposition,andrisktaking WebeginouranalysisofHypothesis3inTables7and8byfocusingonmeasuresofrisk fromtheliabilitysideofthebalancesheet.Withrespecttothecompositionofliabilities,we examinebankrelianceontheuseofborrowedfunds,whichpreviousresearchhasshownisa forecasterofbankdistress(CalomirisandMason,1997;2003;Carlson,2010).Borrowedfunds weremoreexpensiveandhadtobesecured;useofthesefundssuggestsagreaterlevelofrisk.Due todatalimitationsintrackingtheexactamountsofborrowedfunds,weuseaprobitspecificationto testwhetherourownershiporgovernancevariablesareassociatedwiththeuseofsuchfunds.We find,inTable7,thatbanksinwhichmanagersaremoresignificantownersarelesslikelytorelyon borrowedfundsfromotherbanks.ThatresultholdsbothforsimpleprobitandIVspecifications. Thesimpleprobitresultindicatesthatanincreaseinthemanagementownershipshareof10basis 25 points,roughlyhalfastandarddeviation,wouldhavereducedtheprobabilityofborrowingby0.05, afairlysubstantialeffect.Theresultsforcorporategovernancehaveoppositesign,asexpected; exogenousincreasesinformalcorporategovernanceareassociatedwithhigheruseofborrowed money(heretheIVresultsarestatisticallysignificantbutthesimpleprobitresultsarenot). Toeconomizeonthereportingofotherresultsrelatingtoriskchoice,oursubsequent findingsforthreeotherendogenousvariablesmeasuringriskaresummarizedinTable8,which omitsreportingthevariouscontrolvariablesandfocusesonthekeycoefficientsofinterest(the relationshipamongmanagerialownership,governancescore,ortheirinstrumentedvalues,andthe othervariablesofinterest).Theseresultscorroboratetheresultsforborrowedmoney. Withrespecttousingmeasuresofriskbasedontheassetsideofthebalancesheettotest Hypothesis3,weconsiderthecompositionofloans.Asnotedearlier,realestateloanswere generallyconsideredtoberiskierandwereforbiddenbytheNationalBankAct,butbankscould usemortgagestosecuredebtspreviouslyenteredinto.AsshowninTable8,whenmanagement ownsmoresharesinthebank,thebanktendstohavefewermortgagesonitsbooks.Thisfindingis statisticallysignificantbothinasimpleTobitregressionaswellasusingIV.TheTobitcoefficient indicatesthataonestandarddeviationincreaseinownershipwouldhavedecreasedrealestate loansasashareoftotalloansby1.2percentagepoints,asizeabledecreasegiventhat,onaverage, realestateloansonlyaccountedfor3.6percentoflending.Governancescore(whether instrumentedorotherwise)hastheoppositesign,asexpected,butisnotstatisticallysignificant. Wealsoconsidertheexaminer’sassessmentofassetproblems,measuredbytheestimated lossesonassetsrelativetototalassets.WereportinTable8thatgreaterownershipby managementisassociatedwithlowervaluesofthatmeasure.Aonestandarddeviationincreasein ownershipisassociatedwithareductioninestimatedlossestoassetsofabout0.7percentage points–aconsiderablereductiongiventhatlossesaveragedontheorderof1.2percentofassets. TheIVspecification,however,isnotstatisticallysignificant. 26 Interestingly,thereisanegativeassociationbetweengovernancescoreandestimatedasset lossinthenon‐instrumentedspecification.Theexaminerreportsprovideabreakdownofsources ofexpectedlossesornecessarywrite‐downs(suchasloanlosses,securitylosses,building valuation,etc.).Weanalyzedthecompositionofexpectedlossesandfindthathigherexpected losses/write‐downsrelatedtothecategory“fixturesandfurnishings”aretheprimarycontributor tothegreaterexpectedlossesofbanksthatdisplaylowmanagerialownershipandlowcorporate governancescore.Thatresultisintuitivelyappealing:excessiveexpendituresonfurnishingsarea wasteful,value‐destroyinguseoffundsthatwouldnotbechoseninadisciplinedenvironment. Withrespecttobanksurvival,Table8reportsthatincreasedownershipbymanagementis associatedwithareducedlikelihoodthatthebankclosesbetweenOctober1892andDecember 1893(thebulkoftheclosuresoccurduringthePanicof1893).Whenmanagershadagreater ownershipstake,theytooklessriskandwerethuslesslikelytosuccumb.Greateruseofformal corporategovernancecreatesmoretoleranceforfailurerisk. Wenowturntothequestionofhowownershipandgovernancestructurearerelatedto greaterorlesserrelianceonparticulartoolsofriskmanagement(Hypothesis4).Thedefaultriskof abankismainlydeterminedbythreevariables:theriskinessoftheriskyassets(loansandother riskyassets),theratioof(riskless)cashassetstototalassets,andtheratioofequitytoassets.Less riskyloans,ahigherratioofcashassets,orahigherequityratioallcontributetolowerrisk.Banks cantradeoffamongthesethreemeasurestotargetthedesiredlevelofdefaultriskontheirdebts. Ofcourse,inestimatingtherelianceoncash,otherfactorsarerelevantandmustbe controlledfor.Inparticular,thestructureofdepositshasimplicationsforliquidityrisk—abank thatismorereliantoncheckingaccountsthansavingsaccountsforitsfundingwillprobablyneed toholdmorecash,ceterisparibus.Thus,weincludeadditionalcontrolsinouranalysisofthe 27 choicesofcashandequity,inparticulartheratioofindividualdepositstototalliabilitiesandthe ratioofcheckingdepositstoallindividualdeposits.24 Theresults,presentedinTables9and10,showthatbankswithgreatermanagerial ownershipprefertomakegreateruseofcashandlessuseofequitycapitaltotargettheirdefault risk.Atbanksthathadmoreboardoversightofmanagement,capitalratiosweregenerallyhigher andcashratiostendedtobelower,consistentwiththehypothesisabove.Aonestandarddeviation increaseinmanagementownershipisassociatedwithareductioninthenetworthtoassetratioof about3.5percentagepoints(themeanratiowas33percentagepoints)andanincreasethecashto assetratioof0.5percentagepoints(themeanratiowas8percentagepoints).Theseare economicallymeaningfulmagnitudes. 5.3.Corporategovernanceandinsiderrentseeking HereweexploreHypothesis2–whethermanagerialownershipandformalgovernanceare relatedtoinsiderrentseeking.Inparticular,welookatofficersalaries,lendingtoinsiders,and (lower)dividendpaymentsaswaysthatinsidersmightseektoextractvaluefromthebank. 24Asnotedearlier,banksinReserveCitieswererequiredtoholdmorecashrelativetodepositsthanother banks.WethereforeincludeanindicatorvariableforwhetherthebankislocatedinaReserveCity.The resultsindicatethatbeinginaReserveCitydidboostcashholdingsslightly.Findingonlyamodesteffectis consistentwithCarlson(2015)whofindsthatcashholdingswerenotverydifferentbetweenbanksinlarger countrycitiesandbanksinthereservecities,asthebuffersheldbythecountrybanksweresubstantial. Moreover,hefindsthatitwasnotuncommonforbankstoholdlesscashthanrequired,whichsuggeststhat thereserverequirementswerenotstronglybinding.Forthesereasons,thesimpledummyvariablecontrolin theregressionislikelysufficient. 28 TheOLSresults,showninTable11,areconsistentwiththatideaandindicatethatwhenthe managersownmoreshares,theytendtopaythemselveshighersalaries,andwhengovernance scoreishigh,managersreceivelowersalaries. WedonotreportIVresultsformanagerialsalariesbecausewedonotbelievethatthe necessaryexclusionrestrictionsaresatisfiedforusingtheturnoverinstrumentinthesalaries regression.Turnoverofmanagementislikelytohaveadirectpositiveeffectonmanagerialsalaries, becausetheneedtohireanewbankpresidentislikelytorequireattractingcandidatestothebank quickly,sometimesfromadistance. Anotherwayofextractingrentsfromabankisfortheownerstolendtothemselvesto financetheiroutsideprojects.Therehasbeenconsiderableprioracademicanalysisofthisissue, whichindicatesthatinsiderlendingisnotalwaysvalue‐destroyingorrisky(Lamoreaux,1994; Haber,1995).Welookattwovariablesrelatedtoinsiderlending.Thefirstistheamountofloans madetoallinsiders(boardmembersandmanagement)relativetoallloans.Thesecondisthe proportionofallinsiderloansgoingtomanagers. Interestingly,inregressionsnotreportedhere,wefindthatneithermanagerialownership norBoardoversightareassociatedwiththetotalproportionofinsiderlending(definedtoinclude loanstobothmanagersanddirectors).Wedofind,however,thatownershipandgovernance structurestronglyinfluencewhoreceivesthoseinsiderloans(Table12).Atbanksinwhichthe managementownedagreaterproportionofthestock,agreaterfractionofinsiderloanswenttothe management.Aonestandarddeviationincreaseinmanagementownershipincreasedtheinsider shareofloansgoingtomanagersby7.5percentagepoints.Giventhatofficers,onaverage,received 37percentofinsiderloans,thiseffectofownershipisconsiderable.Whenthereweremore corporategovernancecontrols,moreoftheinsiderloansweremadetotheoutsidedirectors. Withrespecttodividends,wefind,intheOLSresultsreportedinTable13,thatwhenmore sharesareownedbymanagers,dividendpaymentsarehigher.Whilethisfindingisconsistentwith 29 theideathatinstitutionswithhighermanagerialownershipprovidegreaterpayoutstoowners,itis alsoconsistentwiththeideathattheseinstitutionsaremoreprofitable.WedonotreportIV resultswithrespecttodividendpaymentbecause,asinthecaseofmanagerialsalaries,wedonot believethattheexclusionrestrictionsfortheinstrumentaresatisfied.Totheextentthatmanagerial turnoverhasatemporaryeffectonbankperformance,itmayaffectdividendsdirectly.25 Takentogether,ourresultsregardingsalaries,insiderlending,anddividendpaymentsare consistentwithHypothesis2.Whenmanagersownagreaterfractionoftheequitysharesofthe bank,theyextractgreaterrentsfromthebankthroughhighersalariesandmoreloansto themselves,althoughmanagerspreferhigherdividendpayments,whichthey,asmajor stockholders,benefitfrom.Similarly,theresultsongovernancescoresshowthatstronger oversightbytheBoardofDirectorstendstobeassociatedwithlessrentextractionbythemanagers butsomewhatgreaterextractionbytheoutsidersontheBoard(insiderlendingbecamedirected moretowardtheoutsidersontheBoard). 5.4.Corroboratinganecdotalinformation Intheprevioussections,wefindthathighmanagementownershipisassociatedwithsafer assetportfoliochoices,lowmanagementownershipisassociatedwithriskierportfoliosand managerrentseeking,andstrongcorporategovernanceappearstoreducerentseeking.These 25TheNationalBankingActprohibitedbanksfrompayingdividendsiftheywereexperiencinglossesthat exceededtheirundividedprofits(i.e.lossesthatwoulderodethebank’ssurplus).Aswenotedearlierthat bothbankswithgreatermanagementownershipandwithstrongergovernancehavelowerlosses,wemight haveexpectedbothownershipandourgovernancemeasuretoincreasethelikelihoodofdividendpayments. Thatwefindgreaterinsiderownershipincreasesthelikelihoodofdividendpayoutsbutmoreformal governancedoesnotsuggeststhatourresultsmainlyreflectmanagerialbehavioraldifferencesratherthan exogenousdifferencesinprofitability. 30 findingareconsistentwithanecdotalinformationintheexaminerreports.Forexample,inone bankwithhighownershipandstronggovernance,theexaminerreportedthat:“Thisisavery conservativebankandloansanddiscountsonlywheretheybelievethattheyareperfectlysafe.I candiscernnopoorpaperinthebank.26” Moreover,wefindexamplesofexaminerexpressionsofconcernaboutbankswithlow managerownershipandlowgovernancescores: Itscapitalisbadlyimpaired…Itisshamefulandwickedthatsomuchmoneyshould befooledawayinsoshortatimeandprovethefollyofhavingrealestate speculatorsasmanagersofbankinginstitutions.27 and Thegeneralconditionofthebankisgoodexceptingthattheofficersareusingtoo muchofthebank’smoneywithoutsecurity,loaningtoomuchtotheBankofEverett andusingtoomanydevicestomakeagoodshowing.28 Theseareparticularlyaptexamplesofthesortsofbehaviorsweidentifyinthe empiricalanalysis.Moregenerally,inreviewingtheanecdotalinformation,wefindthat theretendedtobemoreconcernsaboutbankswithlowmanagementownershipandlow governanceandfewconcernsaboutbankswithhighownershiporgovernance. 6.Robustnessandextensions Herewediscussavarietyofrobustnesschecksandextensionsofthebaselineanalysis. 6.1.Consideringinteractionsbetweenownershipandgovernancechoice Ourfirstextensionistore‐runourregressions,allowingbothstockownershipstructure andcorporategovernancetoenterasendogenousvariablesinthesameregressions.TheIVresults 26ExaminerreportofJanuary7,1893fortheLumberman’sNationalBank,Stillwater,MN,charter1783. 27ExaminerreportofDecember28,1892fortheWashingtonNationalBankofTacoma,WA,charter4018. 28ExaminerreportofMarch1,1893fortheColumbiaNationalBank,Tacoma,WA,charter4623. 31 reportedthusfartreatthetwoendogenousvariablesasalternativemeasures,reflectingourview thatconcentratedownershipandformalgovernancearealternativemeansofensuringgood management,andthattheyarenegativelycorrelatedchoices.Toconsiderpossibleinteractions betweentheownershipconcentrationandformalgovernancerequirestheidentificationofa secondinstrument.Onecandidateisthedegreeofcompetitivepressurefacedbythebank,which wecapturewiththenumberofotherbanks(national,state,savings,andprivate)operatinginthe samecityasthesubjectbank.Althoughusingbothmanagerialturnoverandcompetitivepressure resultedinqualitativelyencouragingfirst‐andsecond‐stageresults,theinstrumentswerenot powerfulenoughtoyieldstatisticallysignificanteffectsinsecond‐stageregressionsthatincluded bothownershipconcentrationandformalgovernancechoiceasendogenousvariables,andwedo notreportthoseregressionresultshere. Asasimplerapproachtoinvestigatingtheinteractionsofownershipconcentrationand governancechoice,wedividebanksintofourgroups,usingatwo‐by‐twomatrixthatmeasures eachbank’scombinationofmanagerialownershipandformalcorporategovernancescore.Thefour groupsaredefinedas(1)high‐managerialownershipandhigh‐formalgovernancescorebanks,(2) high‐managerialownershipandlow‐formalgovernancescorebanks,(3)low‐managerialownership andhigh‐formalgovernancebanks,and(4)low‐managerialownershipandlow‐formalgovernance banks.Thisapproachalsohelpsassureusthatourearlierresultswerenotdrivenbyoutliersinour concentrationmeasure.OurfindingsforthesegroupsarereportedinTable14. Asexpected,thefourthgroup(whichlackseitherahighdegreeofmanagerialownershipor formalgovernance)isriskierthantheotherthree.Thisgroupwasmorelikelytouseborrowed funds,moreheavilyinvestedinrealestateloans,andhadgreaterexpectedlossesthantheother groups.Furthermore,thesebanksdisplayhigheroperatingcosts,whichamoregranularanalysis showsistheresultofunusuallyhighspendingonbankfurnishings(aformofmanagerial perquisities). 32 Severalotherfindingsarealsoconsistentwithourearlierresults.Forinstance,theratiosof managerialsalariestoassetsandofloanstomanagersrelativetoallinsiderloansaresignificantly greateratbanksinwhichmanagementownershipishigherandformalcorporategovernanceis lowercomparedtobankswithlowmanagerialownershipandhighformalgovernance.The averageratioofcashtoliabilitiesishigherwhenformalgovernanceislow.Networthtoasset ratiosarelowerforthetwogroupswithhighmanagementownershipthantheothertwogroups. 6.2.Differentcomponentsofgovernancescore Wealsoexaminedtheseparateroleofeachofthecomponentsofthegovernancescoreused above.Specifically,wetestedwhetherourresultsrelatingtothecorporategovernancescoreare drivenbyoneortwoofthefiveindividualindicatorsbyrepeatingtheregressions,replacingthe scorevariablewitheachcomponentinturn.Inmanycases,wefindthatthecoefficientsonthe individualcomponentstendtopointinthesamedirection,whichsuggeststhattheoverallresults areindeeddrivenbythesummationofthesedifferentmeasures.Forexample,wefindthatthe reductioninlossesrelativetoassetsismoststronglyassociatedwithhavinganactivediscount committeeandwithhavingabondedcashier;therelationshipwiththeothergovernancemeasures alsopointinthatdirectionbuttheeffectsaresmaller.Mostothervariablesbehavesimilarly. Forafewmeasures,therelationshipbetweenoversightandoutcomesismorecomplicated. Fortheuseofborrowedfunds,havingtheboardmeetmonthlyormorefrequently,havinga relativelyhighportionoftheBoardconsistingofoutsidedirectors,andrequiringabondfromthe cashierareallassociatedwithanincreasedlikelihoodofusingborrowedfunds.Bycontrast,having anactivediscountcommitteeandrequiringabondfromthepresidentarebothassociatedwitha lowerlikelihoodofusingborrowedmoney.Thus,thereissomeindicationthatthedifferent oversightmeasurestriggereddifferentresponsesonthepartofmanagersinsomecases. 33 6.3.Examiningexecutivecompensationschemes Managerswerepaidsalaries,andthereisnoevidenceofstock‐based,option‐based,orcash bonusesinmanagerialcompensation.Nevertheless,weareabletoconsiderhowmanagerial incentivesmayhavebeeninfluencedbytheextenttowhichtheincomeofthemanagercovaried withthebank’sincome.Themanagerreceivedasalaryaswellasdividendpayoutsbyvirtueofhis ownershipofshares.Anumberofrecentstudiesfindthatcompensationsensitivitytofirm performancemattersforrisktakingandthatwhentheexecutive’ssalaryismoresensitivetorisk– inourcase,whenitismoredependentondividends–thebank’sinvestmentstendtoberiskier(Bai andElyasiani2013;Cheng,HongandScheinkman2013).Forthisanalysis,wefocusontheincome ofthepresident. Wefindthathavingahigherproportionofthepresident’scompensationintheformof salary(ratherthandividends)isassociatedwithhavingahigherproportionofloansrelatedtoreal estateandhavinglargerforecastedlosses.Theseresultspointtogreaterrisktakingwhen compensationislessduetoprofits.Ofcourse,theseresultsaresubjecttoconcernsabout endogeneity;havinglargerexpectedlossespresumablyreducesprofitsanddividends,which increasestheproportionofcompensationduetosalary. 6.4.Alternativemeasuresofoutsidedirectorinfluence Ourmeasureofoutsidedirectorownershipconsidersalloutsidedirectorstogether. However,itispossiblethattheeffectsofboardoversightdependontheamountofsharesthat particularboardmembersown.Toinvestigatethatpossibility,wecreateadummyvariable indicatingwhenthereisanoutsidedirectorwithmoresharesthananyofthetopthreemanagers (individually,notcollectively).Suchanoutsidedirectorexistsforabout20percentofthebanksin oursample. 34 WhenadirectorwithalargenumberofsharesisontheBoard,wefindthatthepresenceof suchanindividualtendstomagnifythepriorresultofgreaterrisktaking.Forinstance,thebank tendstohavegreatersharesofloansrelatedtorealestate.Thetendencyforgreaterrisktaking appearstobeconsequentialasbankswithlarge‐shareholdingdirectorsalsoaremorelikelytoclose duringthepanic. 6.5.Additionalcontrolvariables Wealsotriedincludingavarietyofothervariablesascontrols.Onesuchvariablewasthe averagescoreforbanksinthesamecity,whichmightreflectthebestpracticeoftheneighboring banks.Thisvariabletendedtohavethesamecoefficientasthebank’sownscorevariable. Includingitdidnotaffecttheresultsaboutwhichwearemostinterested. Asanalternativetocontrollingforspecificfactors,wealsoreplacedourcontrolswithstate fixedeffects,whichprovideamoregeneralcontrolforthingsthatmightbelessobservable(suchas differencesintheabilityofstatebankstoofferservicesprohibitedtoNationalbanks).Usingfixed effectsalsohaslittleeffectontheownershipstructureorcorporategovernanceregressions. Wealsotriedincludingthesquareoftheownershipbythetopthreemanagersincasethere werediminishingreturnstoownershipconcentration.Thisvariablealsodidnotaffectourmain resultsandwaslargelyinsignificant. 7.Conclusion Ourresultshaveinteresting,important,andnovelimplicationsforhowgovernance differenceshelpbankstoattractoutsidefundingsourcesinanenvironmentinwhichconflictsof interestareimportant.Wefindthatmanagerialownershipandformalgovernancetoolsare alternativemeanstoresolveconflicts.Eachofthesealternativeshasimportantandsomewhat differentimplicationsforrentseeking,thetargetingofdefaultrisk,andthetoolsused(cashvs. 35 equity)toachievethetargetedlevelofdefaultrisk.Moreconcentrationofownershipleadstoless formalstructuresofgovernance,moreinsiderbenefitsthroughloansandsalaries,moredividend paying,lessrisktaking(presumablyduetoriskaversionofmanagerstockholders),andmore relianceoncash(toresolveasset‐substitutionandadverse‐selectionproblems).Ourfindingson howmanagerialownershipaffectsrisktakingareausefulcomplementtotherecentliteratureon theimpactofmanagerialincentives.Endogenouslychosenformalgovernancestructuresproduce greaterrisk,andmorerelativerelianceoncapitalforriskmanagement,butlowermanagerial salaries. Theselatterfindingsaregenerallyconsistentwiththeliteratureonthevalueofcorporate boardscurrently(JohnandSenbet,1998,HermalinandWeisbach,2003).Someofourotherresults contrastwithcurrentexperience.Theexpansionofthesafetynetandtoo‐big‐to‐failprotectionhas beenassociatedwithadramaticdeclineinbanks’holdingsofcashassetsasaproportionoftotal assets.Furthermore,recentexperiencesuggeststhatthedisciplineofoutsidestockholders (institutionalblockholders)ofbankstockhasbeenassociatedwithgreaterrisktakingbyprotected banks,whichhasbeeninterpretedasameansofmaximizingtheputoptionvalueofgovernment protection(LaevenandLevine,2009).Thedeclineofcashandincreasedtoleranceforriskby blockholdersofbanksthatenjoysafetynetprotectioncontrastwiththebehaviorofhistorical banks,whichemployedcashaspartofacrediblestrategytosignaleffectiveriskmanagement,and whichweresubjecttobothdepositordisciplineandeffectiveoversightbyoutsidestockholders whousedcorporategovernancetoolsasameansoflimitingbankrisk. Insummary,therearetwokeycorporategovernanceproblemsthatariseinbanking: managerialrentextractionthroughsimpletransfers(highsalariesandsubsidizedloansto managers)andthepossibilityofmanagers’undertakingexcessiverisk.Highmanagerialownership withoutformalcorporategovernanceaddressesthesecondoftheseproblems,butpermitsgreater managerialrentextractionthanwouldoccurundermoreformalcorporategovernancepractices. 36 Thatoutcomemaybepreferredbythemanagerswhoorganizebanks(i.e.,ifthepotentialrents fromexpandingthesizeofthebankarelimited).If,however,managerswishtoexpandtheir enterprisestoascalethatislargerelativetotheirmanagerialstakesinthebank,thenformal corporategovernanceislikelytobecomenecessary.Theformalapproachtogovernanceresultsin highertoleranceforrisk(reflectingthegreaterdiversificationofholdingsofbankstock)anda reductionintherentsthatbankmanagersareabletoextractthoughhighsalariessubsidized lending.Inthepresenceofformalgovernance,managerssharetheirprivilegedaccesstobankloans withoutsidedirectors. 37 Appendix A.1.AModelofEndogenousAsymmetricInformationandCorporateGovernance Webeginwiththesimplestpossiblemodelofcorporategovernancechoice,inwhichthe assetsofthebankconsistentirelyofloansandthefinancingofthebankconsistsonlyofstock.We relaxtheseassumptionssubsequently,andshowthatthecentralimplicationsofthemodel–thatis, thathighermanagerialwealthtendstoreducetherelianceonformalcorporategovernance–also holdwhenweallowfordepositfinancingandtheholdingofcashassets. Abankerisendowedwithwealth(E)andlendingopportunities(agivennumberof profitablepotentialloansthathemightundertake).Eachloanisnormalizedtobeofidentical, unitarysize.Thenumberandamountofloansmade,X,isbetween0andXmax.Forsimplicity,we assumethatthebankholdsonlyloansandisfinancedentirelybyequityprovidedbythebanker andoutsideinvestors(thereisnobankdebt).Themanager’sequityshareofthebank,m,is thereforeE/X.Whenweadddepositliabilitiesandcashassetstothemodel–asinCalomiris, Heider,andHoerova(2014)–themainconclusionsofthemodelarethesame,butadditional conclusionsfollowwithrespecttotheroleofcashinincentivizinggoodriskmanagement. Interestingly,inthisframework,cashplaysanimportantroleinincentivizinggoodrisk managementwhetherornotoutsiderfinancingisintheformofdebtorequity.Thiswarrants emphasis:unlikethediscussionofJensenandMeckling(1976),theproblemofriskshiftinginthis modelisaconflictbetweentheinsider/managerandalloutsidefundingsources,notjustdebt holders.29Inthesimplifiedmodel,bankmanagersfaceincentivestoincreaseriskinvalue‐ 29Inthemodelpresentedinthisappendix,theoutsideequityinvestoreitherbecomesaninsiderbybeing invitedtoparticipateingovernance,orremainsuninvolvedingovernanceknowingthatthebankerwill investinriskmanagementduetoasufficientlyhighlevelofm(thebanker’sproportionofownership). Calomiris,Heider,andHoerova(2014)showthat,intheirmodel,theoptimalcontractforinvestorswho remainoutsiders(andthereforearenotabletocontrolriskmanagement)wouldbeseniordepositsinabank 38 destroyingways(so‐called“assetsubstitution”or“riskshifting”)eventhoughdebtfinanceis absent.Minorityshareholders,likecreditors,haveaninterestinensuringproperriskmanagement bybankmanagers,whichcaneitherbeachievedthroughhighermanagerialstakesinthebankor formalcorporategovernance. Outsideequityisprovidedbyasingleoutsideinvestor.Theoutsideinvestorandthebanker arerisk‐neutralandhaveidenticalreservationreturnsofR,whichrepresentsthegrossreturnthey couldearnonanalternativetolending.Theloanopportunitiesofthebankerareworthpursuing, butonlyifthebankerinvestshisowneffortinriskmanagement.Onecanthinkofthisinvestmentin riskmanagementasthebanker’scontinuingperformanceofduediligence,monitoring,and enforcementofloancovenants.Riskmanagementeffortisprivatelycostlytothebanker;itentails disutilityequaltoBX.Withriskmanagement,loansearnacertainreturnofY>R.Withoutrisk management,loansearnYwithprobabilitypand0withprobability(1‐p).Withoutrisk management,loansarenotworthwhileinvestmentsbecausepY<R. Theobservabilityofriskmanagementdependsonthecorporategovernanceenvironment chosen.Ifthebankerchoosestoincludetheoutsideinvestorinthegovernanceofthebank,then riskmanagementisobservableandcontractible.Ifthebankerchoosesnottoincludetheoutside investorinthegovernanceofthebank,thenriskmanagementisamatterofprivateinformation onlyobservedbythebanker. withcashreservesaswellasloans.Thekeydifferencesinassumptionsbetweenthatmodelandthe simplifiedonepresentedherearetheavailabilityofasinglelargeoutsideinvestor(assumedhere)andthe possibilityofestablishingoversightofriskmanagementbythatoutsideinvestor.Calomiris,Heider,and Hoerova(2014)assumethatoutsideinvestorsarefragmented.Theirsolutiontoincentive‐compatiblerisk managemententailstheuseofdepositsandcashreserveholdings.Inasmallbank,withasinglelargeoutside investor,andthepossibilityofdirectmonitoringofthebankerbythatoutsideinvestor,depositorwithdrawal threatsandidlecashholdingsarenotnecessarytoachieveefficientriskmanagement. 39 Thebanker’s“salary”(S)isaformofrentextraction,whichisendogenoustothecorporate governancechoiceofthebanker.Ifthebankerdoesnotincludetheoutsideinvestorinthe governanceofthebank,thenhewillsethissalarysuchthattheoutsideinvestorreceivesonlythe reservationreturnR.Ifthebankerincludestheoutsideinvestorinthegovernanceofthefirm,then hemustsharetherentsfromlendingaboveRwiththeoutsideinvestor.Theprecisedegreeofthat sharingshouldreflect,inamorerealisticmodelofthemarketforoutsidefunding,thecompetition amongoutsideinvestorstosupplyfundstothebank.Inourmodel,wesimplyassume,withoutloss ofgenerality,thatthebankerandoutsideinvestorsplittherentsevenlywhentheoutsideinvestor isincludedincorporategovernance. Aswewillshow,inequilibrium,becauseriskmanagementisprivatelycostlytothebanker, withoutoutsideinvestorinvolvementincorporategovernance,thebankerwillhavetolimitthe sizeofthebanktoX*.Withoutsiderinvolvementincorporategovernance,thebankercansetthe sizeofthebanktoXmax.Thus,thebankertradesoffthebenefitofgreaterrentsthatcomewith largerbanksize(whichisonlyfeasibleifheincludestheoutsideinvestorinthegovernanceofthe bank,andsharestherentsfromlendingwiththeoutsideinvestor)againstthecostofsharingthe rentsoflendingwiththeoutsideinvestor. X*isdeterminedbytheincentive‐compatibilityconstraintforthebankertoinvestinrisk managementintheabsenceoftheinvolvementoftheoutsideinvestoringovernance.Without outsiderinvolvement,thebankerwillchoosetoinvestinriskmanagementonlyifthepayofftohim fromdoingsoexceedsthepayofffromnotdoingso.Thisiscapturedbytheexpression: SX+m(YX–SX)–BX>p[SX+m(YX–SX)]. (1) Recallthatm=E/X.X*isthemaximumfeasiblelevelofXatwhichthisequationissatisfied (thatis,wheretheequationissatisfiedasanequality,inwhichthebankerisindifferenttoinvesting inriskmanagement).X*alsoimpliesauniqueminimumvalueofm*. m*=E/X*={[B/(1‐p)]–S}/(Y–S). 40 (2) Asthisexpressionshows,thecriticalvaluesofm*andX*dependonS.Swillbechosento transferallrenttothebanker,leavingtheoutsideinvestorearningonlythereservationlevelof return,R.Inotherwords,Sischosenbythebankertosatisfythefollowingexpression,whichisthe participationconstraintfortheoutsideinvestor: (X–E)R=(1–m)(YX–SX). (3) (4) ThisexpressionreducestoS=Y–R. Thus,theconditiondeterminingthecriticalvalueofm*canberewrittenas: m*={[B/(1‐p)]–(Y–R)}/R. Thisexpressioncanbeusedtoperformcomparativestaticanalysisofm*withrespecttodifferent valuesofp,Y,andR.Inparticular,itcanbeshownthatahigherYimpliesalowervalueofm*. Intuitively,whenrentsarehigher,thebankerisabletocrediblypledgetoinvestinrisk management,withoutoversight,withalowerminimummanagerialstakem*. Whetherthebankerwillchoosenottoincludetheoutsideinvestoringovernance(and operatethebankatthelevelofX*)ortoincludetheoutsideinvestoringovernance(andoperate thebankatXmax)dependsonhowmuchthebankerreceivesundereachofthosealternatives.Recall that,iftheoutsideinvestorisincludedincorporategovernance,hewillsplittherentswiththe banker,andtherefore,boththebankerandtheoutsideinvestor/directorwilleachearnanidentical “salary”ofS=(Y–R)/2. Ifthefollowingconditionissatisfied,thebankerwillearnmorebychoosingtoincludethe outsideinvestoringovernanceandoperatethebankatXmax: ER+Xmax(Y–R)/2>ER+X*(Y–R). (5) SolongasXmax>2X*,thisconditionissatisfied.Notethat,inanycomparativestaticcalculation,Y affectsthegovernancedecisiononlyindirectlythroughthepositiveeffectofYonX*(i.e.,dX*/dY>0, implyingthat,ceterisparibus,higherYmakesitlesslikelythatoutsideinvestor’swillbeinvitedto participateingovernance). 41 Theabovemodelhasclearimplicationsforcorporategovernancedecisionsandtheir consequences.Dependingonthesizeofrentsperloan,andthenumberofloansavailabletothe banker,hewilldecidewhethertorunthebankwithnooutsideoversightortoincludetheoutside investorinoversight.Iftheoutsideinvestorisincludedinoversight,then“asymmetricinformation” and“assetsubstitutionrisk”willbeeliminated,andthebanker’ssalarywillbelower,asheisforced tosharerentwiththeoutsideinvestor. A.2.Addingcashassetsanddepositstothemodel Themodelcanbeextendedtoallowbankerstochoosetoholdcashinacredibleand observableform.Ifcashassetsareaddedtothemodelwithoutalsoallowingforseniordeposit claims,cashholdingswouldservenopurpose.Toseewhy,considertheeffectonequation(1)of bankcashholdings,C.Becausecashisriskless,thebankerwillreceive,inadditiontothepayoffs describedinequation(1),anamountmCirrespectiveofwhetherthebankerundertakesrisk management.Thus,cashhasnoeffectonthebanker’sriskmanagementeffort. AsCalomiris,Heider,andHoerova(2014)pointout,however,outsiderfinancingviaequity isnotgenerallytheoptimalcontractunderthesecircumstances.Bygivingoutsidersaseniorclaim onthecashflowsofthebank,thebankerensuresthatwhenriskisnotmanagedproperly,and whenlowpayoffsoccur,outsiderswillreceiveallofthecash,notjust(1‐m)C.Thuswhenoutsider financingispartlyintheformofdeposits,andbankersareabletoholdcash,bankersareableto committoproperriskmanagementbyholdingasufficientamountofcashassets.Depositsandcash affectriskmanagementbecause,unlikeoutsideequityfinancing,depositfinancingdoesnotdilute theupsideofthebanker’sprofit,andunlikeoutsideequityholders,depositsreceiveallofthe bank’scashassetswhenthebankerchoosesnottoinvestinriskmanagementandthebadoutcome occurs(withprobability1‐p). 42 Amorerealisticmodel(whichwouldhavetobemuchmorecomplicatedthateitherthe frameworkpresentedhere,orthatofCalomiris,Heider,andHoerova,2014)couldallowforboth depositsandoutsideequitysourcesoffunding.AsCalomiris,Heider,andHoerovaconjecture,such amodelcouldsolveforaninteriorsolutionfortheoptimalmixofoutsidefunding.Insuchamodel, themarginalcostofrelyingmoreonoutsideequityisthemarginalcostofcorporategovernance improvementstomakeriskmanagementdecisionsobservabletothemarginaloutsideequity holders,whilethecostofrelyingmoreonseniordepositsistheopportunitycostofcashholdings. Bankersthatchoosemoreinformalcorporategovernancewilltendtorelymoreonoutsideequity fortheiroutsidefinancing,andwillholdasmallerfractionoftheirassetsincash.Inourempirical results,wetest,andconfirm,thosepredictions. 43 References Acharya,V.,Carpenter,J.,Gabaix,X.,John,K.,Richardson,M.,Subrahmanyam,M.,Sundaram,R., ZemelE.,2009.Corporategovernanceinthemodernfinancialsector.InAcharya,V.Richardson,M., (Eds.)RestoringFinancialStability.JohnWiley&Sons,Hoboken. Acharya,V.,Mehran,H.,Thakor,A.,2013.Caughtbetweenscyllaandcharybdis?regulatingbank leveragewhenthereisrentseekingandriskshifting.FederalReserveBankofNewYorkStaff Report469. Anginer,D.,Demirguc‐Kunt,A.,Huizinga,H.,Ma,K.,2013.Howdoescorporategovernanceaffect bankcapitalizationstrategies?WorldBankPolicyResearchWorkingPaper6636. Bai,G.,Elasiani,E.,2013.Bankstabilityandmanagerialcompensation,JournalofBankingand Finance37,799‐813. Berger,A.,Imierowicz,B.,Rauch,C.,2012.Therolesofcorporategovernanceinbankfailuresduring therecentfinancialcrisis.EuropeanBankingCenterDiscussionPaper2012‐023. Bhattacharya,S.,Boot,A.,Thakor,A.,1998.Theeconomicsofbankregulation,JournalofMoney, CreditandBanking30,745‐770. Bodenhorn,H.,2013.Largeblockholders,institutionalinvestors,boardsofdirectors,andbank valueinthenineteenthcentury.Unpublishedworkingpaper.ClemsonUniversity. Bodenhorn,H.,White,E.,2014.Theevolutionofbankboardsofdirectorsinnewyork,1840‐1950. NationalBureauofEconomicResearchWorkingPaper20078. Bruno,V.,Claessens,S.,2010.Corporategovernanceandregulation:cantherebetoomuchofa goodthing?JournalofFinancialIntermediation19,461‐482. Calomiris,C.,Carlson,M.,2014.National Bank Examinations and Operations in the Early 1890s. Federal Reserve Finance and Economics Discussion Series 2014-19. 44 Calomiris,C.,Heider,F.,Hoerova,M.,2014.Atheoryofbankliquidityrequirements.Unpublished workingpaper.ColumbiaUniversity. Calomiris,C.,Kahn,C.,1991.Theroleofdemandabledebtinstructuringoptimalbanking arrangements.AmericanEconomicReview81,497‐513. Calomiris,C.,Mason,J.,1997.Contagionandbankfailuresduringthegreatdepression:thejune 1932chicagobankingpanic.AmericanEconomicReview87,863‐883. Calomiris,C.,Mason,J.,2003.Fundamentals,panics,andbankdistressduringthedepression. AmericanEconomicReview93,1615–1647. Calomiris,C.,Mason,J.,2008.Resolvingthepuzzleoftheunderissuanceofnationalbanknotes. ExplorationsinEconomicHistory45,327‐355. Calomiris,C.,Mason,J.,Wheelock,D.,2011.Diddoublingreserverequirementscausetherecession of1937‐1938?amicroeconomicapproach.NationalBureauofEconomicResearchWorkingPapers 16688. CalomirisC.,Wilson,B.,2004.Bankcapitalandportfoliomanagement:the1930s'capitalcrunch' andthescrambletoshedrisk.JournalofBusiness77,421‐455. Carlson,M.,2010.Alternativesfordistressedbanksduringthegreatdepression.JournalofMoney, CreditandBanking42,421‐442. Carlson,M.,2015.Lessonsfromthehistoricaluseofreserverequirementsintheunitedstatesto promotebankliquidity.InternationalJournalofCentralBanking11,191‐224. Cheng,I.,Hong,H.,Scheinkman,J.,2013.Yesterday’sheroes:compensationandriskatfinancial firms.Unpublishedworkingpaper.ColumbiaUniversity. Claessens,S.,Djankov,S.,Fan,J.,Lang,L.,2002.Disentanglingtheincentiveandentrenchment effectsoflargeshareholdings.JournalofFinance57,2741‐2771. 45 Demsetz,R.,Saidenberg,M.,Strahan,P.,1997.Agencyproblemsandrisktakingatbanks.Federal ReserveBankofNewYorkBankingStudiesDepartmentWorkingPaper. Diamond,D.,Rajan,R.,2001.Liquidityrisk,liquiditycreations,andfinancialfragility:atheoryof banking.JournalofPoliticalEconomy109,287‐327. Ellul,A.,Yeramilli,V.,2010.Strongerriskcontrol,lowerrisk:evidencefromUSbankholding companies.NationalBureauofEconomicResearchWorkingPaper16178. Fahlenbrach,R.,Stulz,R.,2011.BankCEOincentivesandthecreditcrisis.JournalofFinancial Economics99,11‐26 Freeman,M.,Pearson,R.,Taylor,J.,2012.ShareholderDemocracies?CorporateGovernancein Britain&Irelandbefore1850.UniversityofChicagoPress,Chicago. Gorton,G.,Rosen,R.,1995.Corporatecontrol,portfoliochoiceandthedeclineofbanking.Journalof Finance50,1377‐1420. Haber,S.,1995.Regulatoryregimes,capitalmarketsandindustrialdevelopment:acomparative studyofbrazil,mexicoandtheunitedstates,1840‐1930.In:Harriss,J.,Hunter,J.,Lewis,C.,(Eds.), TheNewInstitutionalEconomicsandThirdWorldDevelopment.Routledge,LondonandNewYork. Haber,S.,Maurer,N.,2007.Relatedlendingandeconomicperformance:evidencefrommexico. JournalofEconomicHistory67,551–581. Hansmann,H.,Parglender,M.,2012.Theevolutionofshareholdervotingrights:separationof ownershipandconsumption.YaleLawSchoolWorkingPaper. Hermalin,B.,Weisbach,M.,2003.Boardsofdirectorsasanendogenouslydeterminedinstitution:a surveyoftheeconomicliterature.FederalReserveBankofNewYorkEconomicPolicyReviewApril, 7‐26. 46 Hilt,E.,2008.Whendidownershipseparatefromcontrol?corporategovernanceintheearly nineteenthcentury.JournalofEconomicHistory68,645‐685. Hilt,E.,2013.Shareholdervotingrightsinearlyamericancorporations.BusinessHistory55,620‐ 635. Hilt,E.,2014.Historyofamericancorporategovernance:law,institutions,andpolitics.National BureauofEconomicResearchWorkingPaper20356. Holderness,C.,Kroszner,R.,Sheehan,D.,1999.Werethegoodolddaysthatgood?changesin managerialstockownershipsincethegreatdepression.JournalofFinance54,435‐469. Holmstrom,B.,Tirole,J.1997.Financialintermediation,loanablefunds,andtherealsector. QuarterlyJournalofEconomics112,663‐691. James,J.,1978.MoneyandCapitalMarketsinPostbellumAmerica.PrincetonUniversityPress, Princeton. Jensen,M.,Meckling,W.,1976.Theoryofthefirm:managerialbehavior,agencycosts,and ownershipstructure.JournalofFinancialEconomics3,305‐360. John,K.,Senbet,L.,1998.Corporategovernanceandboardeffectiveness.JournalofBankingand Finance22,371‐403. Laeven,L.,Levine,R.,2009.Bankgovernance,regulation,andrisktaking.JournalofFinancial Economics93,259‐275. Lamoreaux,N.,1994.InsiderLending:Banks,CommercialLending,andEconomicDevelopmentin IndustrialNewEngland.CambridgeUniversityPress,Cambridge. Lamoreaux,N.,Glaisek,C.,1991.Vehiclesofprivilegeormobility?banksinprovidence,rhodeisland duringtheageofJackson.BusinessHistoryReview65,537‐557. 47 Lunt,E.,1922.SuretyBonds:Nature,Functions,UnderwritingRequirements.TheRonaldPress Company,NewYork. Mehran,H.,Morrison,A.,Shapiro,J.,2011.Corporategovernanceandbanks:whathavewelearned fromthefinancialcrisis.FederalReserveBankofNewYorkStaffReport502. Meissner,C.,2005.Votingrulesandthesuccessofconnectedlendinginnineteenthcenturynew englandbanks.ExplorationsinEconomicHistory42,509‐528. Merton,R.,1977.Analyticalderivationofthecostofdepositinsuranceandloanguarantees:an applicationofmodernoptionpricingtheory.JournalofBankingandFinance1,3‐11. Myers,S.,1977.Determinantsofcorporateborrowing.JournalofFinancialEconomics5,147‐175. Myers,S.,Majluf,N.,1984.Corporatefinancingandinvestmentdecisionswhenfirmshave informationthatinvestorsdonothave.JournalofFinancialEconomics13,187‐221. Robertson,R.,1995.TheComptrollerandBankSupervision,AHistoricalAppraisal.Officeofthe ComptrolleroftheCurrency,WashingtonDC. Saunders,A.,Strock,E.,Travlos,N.,1990.Ownershipstructure,deregulation,andbankrisktaking. JournalofFinance45,643‐654. SeniorSupervisorsGroup,2008.Observationsonriskmanagementpracticesduringtherecent marketturbulence.Availableatwww.newyorkfed.org/newsevents/news/banking/2008 /SSG_Risk_Mgt_doc_final.pdf Shleifer,A.,Vishny,R.,1997.Asurveyofcorporategovernance.JournalofFinance52,737‐783. 48 Table1 Summaryofanalysis Hypothesis1 Exogenousinfluencesthatincreasemanagerialownership=>Reducerelianceonformal governance Explanation:Whenmanagerialstakesarehigher,thereislessneedforformalgovernanceto constrainmanagerialbehaviortowardrisk. Dataanalysis:Managementownershipisinstrumented(hereandthroughout)byexogenous turnoverevents.Formalgovernanceismeasuredbythefrequencyofboardmeetings,thenumber ofoutsidedirectorappointments,theexistenceofanactivediscountcommittee,andthebondingof bankofficers. Hypothesis2 More(exogenous)managerialownership=>Highermanagers’salaries,moreloanstomanagers, higherdividends Explanation:Whenmanagershavemoreownership,thereislessformalgovernanceandtherefore managersenjoygreatercontrol.Theypaythemselvesmoreandgivethemselvesgreateraccessto loans.Whentheyarelargestockholders,managersalsohavestrongincentivestopaydividends. Dataanalysis:Managers’salaries,thefractionofloansthatwerelenttomanagers,andtheratioof dividendstopaidincapitalareanalyzedasdependentvariables. Hypothesis3 More(exogenous)managerialownership=>Lessrisktakingbythebank Explanation:Whenmanagerialstakesarehigher,managerswillbemorerisk‐aversebecausethey areundiversifiedjuniorclaimantsonthebank. Dataanalysis:Risktakingismeasuredbytheuseofhigh‐cost“borrowedmoney”tofinancethe bank,aswellasbythefractionofrealestateloanstototalloans,theexaminer’sestimatesoflosses relativetoassets,andbytheriskofbankclosure. Hypothesis4 More(exogenous)managerialownership=>Moreuseofcashratherthanequitytoreducedefault risk Explanation:Cashisusefultoensuremanagerialeffortinriskmanagement.Morecashreduces managerialincentivestoshirkinriskmanagementandtherebyprotectsdepositorsandoutside shareholdersfromriskshiftinginbadstates.Thisprotection,however,islessnecessarywhenrisk managementissubjecttooversightduetothepresenceofformalgovernance. Dataanalysis:Bankschoicesofcash‐to‐assetsratioandequity‐to‐assetsratioareanalyzedas dependentvariables. 49 Variable Management ownership Source Boardmeetsmonth Examreport Examreport Outsidedirectorson Examreport board Activediscount Examreport committee Table2 Listofvariables Description Theshareofstockownedbythetop3bankmanagers –thepresident,vicepresident,andcashier Indicatorvariablefortheboardofdirectorsmeeting monthlyormorefrequently Theshareoftheboardofdirectorsthatconsistedof individualsthatwerenotmanagers Indicatorvariableforhavinganactiveindependent discountcommittee Presidentbonded Examreport Presidentpostedasuretybond Cashierbonded ExamReport Cashierpostedasuretybond Score Derived Sumofgovernanceindicators Turnover Examreports& bankersmagazine Numberofchangesinthepresidentbetween1882 callreportand1892callreport Logassets CallReport Logofassets. Logage Comptroller& RandMcNally Logofthedifferencebetween1892andthetimethe bankwasestablished. Salariestoassets Examreport Ratioofsalariesof3officerstoassets Officersloansto insiderloans Examreport Dividendstoshares Examreport Usedborrowed funds Realestateloansto totalloans Otherrealestate ownedtoassets Troubledloansto totalloans Examreport&call report Ratioofloansmadetotop3officerstoloanstoall insiders(managersandboardmembers) Ratioofdividendspaidatlastpayouttoshares outstanding(dollarspershare) Indicatorthatthebankborrowedusinginterbank certificatesofdeposit,rediscounts,orbillspayable Examreport Ratioofloanssecuredbyrealestatetototalloans Callreport Ratioofotherrealestateownedtoassets Lossestoassets Examreport Examreport Loanlossestoassets Examreport Otherlossesto Examreport assets Losseson Examreport furnishingstoassets Individualdeposits Callreport tototalliabilities Ratioof“troubled”loans– thosepastdueor suspended–tototalloans Ratiooftotallossesonallbalancesheetitemsas estimatedbytheexaminerrelativetoassets Ratiooflossesonbadloans,otheroverduepaper, otherloansandoverdraftstoassets Ratiooflossesonsecurities,bankhouse,furnitureand fixtures,otherrealestate,cash,andothertoassets Ratioofrequiredwrite‐downsonfurnitureand fixturestoassets Shareofliabilitiesconsistingofdepositsby individuals 50 Checkingdepositsto Examreport individualdeposits Shareofindividualdepositsconsistingofchecking deposits Ratioofcapital,surplus,andundividedprofitsto assets Networthtoassets Examreport Cashtoassets Examreport Cashandlegaltendertoassets Closed Comptroller reports Indicatorthatthebanksuspended,failed,voluntarily liquidatedafterfilingtheSept.1892callreportbut beforeJan1,1894. Reservecity Comptroller reports Indicatorthatthecityisareservecity Logcitypopulation 1890Census Logofcitypopulation(citypopulationisnotavailable forElPaso,TXsocountypopulationisused) LogdistanceinmilestoNY LogdistancetoNew York Fractioncounty incomefrom agriculture Mininginstate StatisticalAbstract oftheUSfor1892 Valueofagriculturalproductsinthecountydividedby thesumofthevalueofagriculturalproductsandthe valueofmanufacturing Thestateminedmorethan$1millioningoldand/or silverin1891. OldState Statehoodoccurredpriorto1851 1890Census 51 Table3 Summarystatistics Variable Mean Median Std. Dev Min 25th percentile 75th percentile Max Management ownership 0.24 0.17 0.23 0.01 0.08 0.37 0.97 Boardmeetsmonth 0.63 1 0.48 0 0 1 1 0.69 0.71 0.13 0.20 0.60 0.78 0.94 0.60 1 0.49 0 0 1 1 Presidentbonded 0.33 0 0.47 0 0 1 1 Cashierbonded 0.57 1 0.50 0 0 1 1 Score 2.69 3 1.56 0 1 4 5 Turnover 0.67 0 0.81 0 0 1 3 Logassets 14.1 14.1 0.8 12.0 13.5 14.7 15.9 Logage 2.42 2.40 0.74 0.69 1.79 3.14 3.43 0.59 0.46 0.45 0.02 0.33 0.69 3.61 36.7 34.4 29.4 0 8.1 56.2 100 Dividendstoshares 4.7 4 6.2 0 3 5 50 Usedborrowedfunds 0.31 0 0.46 0 0 1 1 3.6 1.1 6.1 0 0 1.2 11.2 0.9 .1 1.6 0 0 1.2 11.2 9.1 5.9 9.9 0 2.5 12.4 71.8 1.2 .2 3.8 0 0 1.1 32.1 .95 .10 3.05 0 0 .85 28.6 0.27 0 0.96 0 0 0.15 11.3 Outsidedirectorson board Activediscount committee Salariestoassets (percent) Officersloansto insiderloans (percent) Realestateloansto totalloans(percent) Otherrealestate ownedtoassets (percent) Troubledloanstototal loans(percent) Lossestoassets (percent) Loanlossestoassets (percent) Otherlossestoassets (percent) 52 Individualdepositsto totalliabilities Checkingdepositsto individualdeposits Networthtoassets (percent) Cashtoassets (percent) 0.70 0.72 0.17 0.20 0.57 0.85 0.97 0.74 0.77 0.20 0.18 0.61 0.91 1 32.9 30.7 12.7 8.5 24.1 39.9 76.1 7.9 7.6 3.6 0.3 5.0 9.8 20.3 Closed 0.29 0 0.45 0 0 1 1 Reservecity 0.37 0 0.48 0 0 1 1 Logcitypopulation 11.0 10.8 0.45 8.2 10.3 11.9 12.6 7.07 7.05 0.45 6.35 6.76 7.40 7.81 0.25 0.12 0.27 0.02 0.07 0.36 0.96 Mininginstate 0.21 0 0.41 0 0 1 1 Oldstate 0.55 1 0.50 0 0 1 1 LogdistancetoNew York Fractioncounty incomefrom agriculture Note.SampleincludesallNationalbanksin37citiesasdescribedinSection4.1. 53 Table4 Correlationofmeasuresofownershipandofcontrol Management ownership Boardmeets atleast monthly High% Outsiderson Board Active discount committee President bonded Cashier bonded ‐0.23 ‐0.44 ‐0.25 ‐0.15 ‐0.22 0.20 0.33 0.08 0.15 0.25 0.22 0.20 0.24 0.43 Boardmeetsat leastmonthly High%Outsiders onBoard Activediscount committee Presidentbonded Note.SampleincludesallNationalbanksin37citiesasdescribedinSection4.1. 54 0.50 Table5 Determinantsofthecorporategovernancescore Score ‐1.88*** Managementownership (0.44) ‐0.35*** Logage (0.13) .30 Reservecity (0.33) 0.001 Logcitypopulation (0.18) ‐1.04*** LogdistancetoNYC (0.33) ‐0.06 Fractioncountyincome fromagriculture (0.51) .35 Mininginstate (.31) .63** Oldstate (.24) 11.65*** Intercept (3.03) 206 Observations AdjR2 0.30 F‐statistic 12.2 Notes:Thesymbols(***),(**),and(*)indicatestatisticalsignificanceatthe1,5,and10percent level,respectively.Estimatedusingordinaryleastsquares.Standarderrorsinparenthesesand italics.SampleincludesallNationalbanksin37citiesasdescribedinSection4.1. 55 Table6 FirststageforIVregressions Management Score ownership 0.38*** ‐0.06*** Turnover (0.12) (0.02) ‐0.49*** 0.05*** Logage (0.14) (0.02) .20 ‐0.01 Reservecity (0.34) (0.05) 0.18 ‐0.06* Logcitypopulation (0.19) (0.03) ‐1.15*** 0.09* LogdistancetoNYC (0.33) (0.05) Fractioncounty 0.21 ‐0.11 incomefrom (0.08) (0.52) agriculture .05 .17 Mininginstate (.05) (.32) ‐.04 .71*** Oldstate .04 (.25) 9.20 .19 Intercept (3.68) (0.57) 206 206 Observations 2 .28 AdjR 0.18 10.9 F‐statistic 6.53 Notes:Thesymbols(***),(**),and(*)indicatestatisticalsignificanceatthe1,5,and10percent level,respectively.Estimatedusingordinaryleastsquares.Standarderrorsinparenthesesand italics.SampleincludesallNationalbanksin37citiesasdescribedinSection4.1. 56 Table7 Factorsassociatedwiththeuseofborrowedmoney Probit Probit IV Managementownership ‐0.51*** (0.15) ‐3.76*** 1.24 0.02 (0.2) ‐0.08* (0.04) 0.23** (0.12) ‐0.19*** (0.07) 0.15 (0.11) ‐0.58*** (0.20) 0.01 (0.10) 0.04 (0.08) 3.04 (3.78) Score Logage Reservecity Logcitypopulation LogdistancetoNY Fractioncountyincome fromagriculture Mininginstate Oldstate Intercept IV ‐0.06 (0.04) 0.23** (0.12) ‐0.21*** (0.07) 0.17* (0.10) ‐0.65*** (0.20) 0.02 (0.10) 0.03 (0.08) 4.24 (3.83) ‐.08 (0.17) 0.60 (0.40) ‐0.72*** (0.22) .74** (0.33) ‐2.10*** (0.67) ‐0.11 (0.31) ‐.01 (0.25) 3.54 (3.69) .51*** (.17) ‐0.003 (0.17) 0.43 (0.39) ‐0.50** (0.21) 0.95*** (0.34) ‐1.48** (0.65) ‐0.13 (0.30) ‐.22 (0.27) ‐2.67 (4.14) Observations 200 200 200 200 PseudoR2 0.13 0.09 LRχ2(probit)/ 30.95 21.13 39.0 42.8 Waldχ2(IV) Notes:Wereportmarginaleffectsevaluatedatthemean.Thesymbols(***),(**),and(*)indicate statisticalsignificanceatthe1,5,and10percentlevel,respectively.Specifications1and2 estimatedusingprobitanalysis;specifications3and4estimatedusinganordinaryleastsquares firststageandaprobitsecondstage.Standarderrorsinparenthesesanditalics.Standarderrors inspecifications3and4havebeenadjustedtoreflecttheuseofgeneratedregressors.Sample includesallNationalbanksin37citiesasdescribedinSection4.1exceptwheretheexaminerdid notdiscussuseofcertificatesofdepositforthepurposeofborrowingmoney. 57 Table8 Associationofothermeasuresofbankriskwithmanagementownershipandbankgovernance Realestateloansto Estimatedlossesto Bankcloseditsdoors totalloans assets Tobit IV Tobit IV Probit IV ‐3.3** ‐6.1 ‐0.27* ‐2.9** (1.6) (6.4) (.15) (1.34) ‐.49** 1.02 .03 .45** Score (.24) (1.2) (.02) (.20) Notes:Thesymbols(***),(**),and(*)indicatestatisticalsignificanceatthe1,5,and10percent level,respectively.Tobitregressionsincolumns1and3aretruncatedbelowatzero.Forcolumns 5and6,wereportmarginaleffects.IVregressionsareestimatedusingatwo‐stepprocedure. Standarderrorsinparenthesesanditalics.Standarderrorsincolumns2,4,and6havebeen adjustedtoreflecttheuseofgeneratedregressors.Allregressionsincludethecontrolsusedinthe previousregressions(suchasthoseshownintheprecedingtable).SampleincludesallNational banksin37citiesasdescribedinSection4.1. Management ownership ‐.055** (.026) .001 (0.004) ‐.19* (.11) .031 (0.020) 58 Managementownership Score Individualdepositsto totalliabilities Checkingdepositsto individualdeposits Logage Reservecity Logcitypopulation LogdistancetoNY Fractioncountyincome fromagriculture Mininginstate Oldstate Intercept Table9 Determinantsoftheratioofnetworthtoassets OLS OLS IV ‐15.6*** (3.45) ‐16.9*** (4.9) 14.9*** (4.8) ‐6.1*** (1.09) ‐7.07*** (2.63) ‐1.57 (1.61) .55 (2.55) 2.02 (4.09) ‐3.75 (2.43) ‐4.40** (2.13) 70.8** (28.5) 1.08** (0.55) ‐17.5*** (5.1) 12.2** (5.0) ‐6.38*** (1.15) ‐7.2*** (2.74) ‐.69 (1.67) 0.42 (2.72) 3.92 (4.23) ‐4.38* (2.53) ‐4.01* (2.23) 58.0* (30.3) ‐29.9** (13.9) ‐16.4*** (5.01) 17.2*** (5.31) ‐5.36*** (1.29) ‐7.31*** (2.68) ‐2.43 (1.83) 1.31 (2.88) .18 (4.50) ‐3.45 (2.48) ‐5.42** (2.37) 71.4 (28.9) IV 5.19* (2.72) ‐17.4*** (5.6) 11.4** (5.5) ‐4.6*** (1.7) ‐8.8*** (3.2) ‐.89 (1.84) 5.53 (4.45) 3.57 (4.45) ‐5.59** (2.90) ‐6.74** (3.03) 11.8 (44.7) Observations 206 206 206 206 AdjR2 0.37 0.31 F‐stat(OLS)/Waldχ2(IV) 12.9 10.4 110.0 86.2 Notes:Thesymbols(***),(**),and(*)indicatestatisticalsignificanceatthe1,5,and10percent level,respectively.Specifications1and2estimatedusingordinaryleastsquares;specifications3 and4estimatedusingtwo‐stageleastsquares.Standarderrorsinparenthesesanditalics. Standarderrorsinspecifications3and4havebeenadjustedtoreflecttheuseofgenerated regressors.SampleincludesallNationalbanksin37citiesasdescribedinSection4.1 59 Managementownership Score Individualdepositsto totalliabilities Checkingdepositsto individualdeposits Logage Reservecity Logcitypopulation LogdistancetoNY Fractioncountyincome fromagriculture Mininginstate Oldstate Intercept Table10 Determinantsoftheratioofcashtoassets OLS OLS IV 2.36** (1.07) 4.54*** (1.52) 0.45 (1.49) 1.14*** (0.34) 0.25 (0.82) 1.45*** (0.50) 1.64** (0.79) ‐0.96 (1.27) .35 (0.75) .37 (.66) ‐26.4*** (8.6) ‐0.26 (0.17) 4.62*** (1.53) .88 (1.49) 1.15*** (0.34) 0.31 (0.82) 1.32*** (0.50) 1.54* (0.81) ‐1.24 (1.27) .48 (0.76) .37 (.67) ‐23.5*** (9.08) 8.12* (4.42) 4.31*** (1.60) ‐.47 (1.69) 0.86** (0.41) 0.34 (0.85) 1.79*** (0.58) 1.09 (.92) ‐.22 (1.43) 0.24 (0.79) .78 (.75) ‐26.7 (9.2) IV ‐1.40* (.80) 4.59*** (1.67) 1.11 (1.62) .66 (0.50) 0.74 (0.94) 1.37*** (0.54) .11 (1.32) ‐1.14 (1.38) .82 (0.86) 1.14 (.90) ‐10.5 (13.2) Observations 206 206 206 206 AdjR2 0.25 0.24 F‐stat(OLS)/Waldχ2(IV) 8.75 8.36 71.5 65.04 Notes:Thesymbols(***),(**),and(*)indicatestatisticalsignificanceatthe1,5,and10percent level,respectively.Specifications1and2estimatedusingordinaryleastsquares;specifications3 and4estimatedusingtwo‐stageleastsquares.Standarderrorsinparenthesesanditalics. Standarderrorsinspecifications3and4havebeenadjustedtoreflecttheuseofgenerated regressors.SampleincludesallNationalbanksin37citiesasdescribedinSection4.1 60 Table11 Determinantsofmanagersalariesrelativetoassets OLS OLS 0.34** Managementownership (0.18) ‐0.01 Score (0.03) ‐0.20*** ‐0.18*** Logage (0.05) (0.05) 0.06 0.05 Reservecity (0.12) (0.13) ‐0.05 ‐0.06 Logcitypopulation (0.08) (0.08) 0.14 0.16 LogdistancetoNY (0.12) (0.13) 0.11 0.09 Fractioncountyincome fromagriculture (0.19) (0.20) .004 ‐0.003 Mininginstate (0.12) (0.13) 0.13 0.12 Oldstate (0.09) (0.19) .40 .53 Intercept (1.40) (1.46) Observations 172 172 AdjR2 0.12 0.10 F‐stat 4.19 3.46 Notes:Thesymbols(***),(**),and(*)indicatestatisticalsignificanceatthe1,5,and10percent level,respectively.Specifications1and2estimatedusingordinaryleastsquares;specifications3 and4estimatedusingtwo‐stageleastsquares.Standarderrorsinparenthesesanditalics. Standarderrorsinspecifications3and4havebeenadjustedtoreflecttheuseofgenerated regressors.SampleincludesallNationalbanksin37citiesasdescribedinSection4.1wherethe managersalarieswerereported. 61 Table12 Determinantsofloanstomanagementasashareofinsiderloans OLS OLS IV IV Managementownership Score Logage Reservecity Logcitypopulation LogdistancetoNY Fractioncountyincome fromagriculture Mininginstate Oldstate Intercept 33.38*** (8.89) .53 (2.69) 2.45 (6.67) ‐.62 (3.81) 8.39 (6.58) 21.1** (10.3) 1.07 (6.29) ‐2.10 (4.91) ‐31.4 (72.1) ‐5.08*** (1.38) ‐.16 (2.73) 3.18 (6.69) ‐1.55 (3.81) 5.59 (6.73) 18.56* (10.28) 3.47 (6.30) .09 (4.99) 20.6 (73.8) 62.6* (36.8) ‐.80 (3.15) 3.41 (6.80) .53 (4.09) 5.34 (7.58) 23.9** (10.9) .31 (6.38) ‐.89 (5.14) ‐28.0 (72.5) ‐10.5* (6.23) ‐2.49 (3.82) 5.12 (7.13) ‐1.15 (3.89) ‐1.08 (10.1) 19.2* (10.5) 5.08 (6.65) 3.89 (6.63) 80.1 (100.5) Observations 206 206 206 206 2 AdjR 0.18 0.14 2 F‐stat(OLS)/Waldχ (IV) 5.32 5.25 31.1 30.4 Notes:Thesymbols(***),(**),and(*)indicatestatisticalsignificanceatthe1,5,and10percent level,respectively.Specifications1and2estimatedusingordinaryleastsquares;specifications3 and4estimatedusingtwo‐stageleastsquares.Standarderrorsinparenthesesanditalics. Standarderrorsinspecifications3and4havebeenadjustedtoreflecttheuseofgenerated regressors.SampleincludesallNationalbanksin37citiesasdescribedinSection4.1. 62 Table13 Determinantsoftheratioofdividendstoshares OLS OLS Managementownership Score Logage Reservecity Logcitypopulation LogdistancetoNY Fractioncountyincome fromagriculture Mininginstate Oldstate Intercept 6.59*** (1.97) 1.85*** (0.60) ‐0.21 (1.45) 0.43 (0.83) 3.84*** (1.43) 5.21** (2.23) ‐1.73 (1.36) .75 (1.08) ‐34.5 (15.6) ‐0.55* (0.31) 1.98*** (0.62) ‐0.24 (1.48) .21 (0.84) 3.87*** (1.50) 4.72** (2.27) ‐1.40 (1.39) .93 (1.11) ‐29.6 (16.3) Observations 201 201 2 AdjR 0.15 0.11 F‐stat 5.29 4.14 Notes:Thesymbols(***),(**),and(*)indicatestatisticalsignificanceatthe1,5,and10percent level,respectively.Specifications1and2estimatedusingordinaryleastsquares;specifications3 and4estimatedusingtwo‐stageleastsquares.Standarderrorsinparenthesesanditalics. Standarderrorsinspecifications3and4havebeenadjustedtoreflecttheuseofgenerated regressors.SampleincludesallNationalbanksin37citiesasdescribedinSection4.1exceptthose toonewtobeeligibletopaydividends. 63 Table14 Meanbankcharacteristicsbymanagementownershipandgovernancescore High ownership high governance High ownership low governance Low ownership high governance Low ownership low governance [1] [2] [3] [4] 0.67 (0.55) 0.75 (0.49) 0.51 (0.45) 0.54 (0.20) 1vs 4 36.3 (26.2) 53.3 (32.2) 24.7 (22.8) 33.9 (28.3) 4.4 (3.4) 7.0 (10.4) 3.3 (1.5) 4.1 (2.9) 20.9 (41.2) 27.1 (44.8) 33.8 (47.6) 3.2 (4.4) 3.5 (4.3) 0.7 (1.5) Loanlosses toassets Otherlosses toassets Testfordifferences inmeans 2vs 4 * *** 3vs 4 * 45.8 (50.9) ** * 2.5 (4.8) 7.3 (11.7) ** ** *** 1.1 (4.1) 0.8 (1.3) 3.6 (7.6) ** ** *** 0.58 (1.28) 0.90 (3.72) 0.58 (1.00) 2.73 (5.66) ** * *** 0.10 (0.28) 0.21 (0.60) 0.21 (0.52) 0.86 (2.24) ** ** ** Closed 30.2 (46.5) 28.3 (45.4) 25.0 (43.6) 33.3 (48.0) Cashto liabilities 7.9 (4.3) 8.3 (3.8) 7.6 (3.3) 8.1 (3.1) Salaryto assets Officerloans toinsider loans Dividendsper share Used borrowed funds Realestate loanstoall loans Lossesto assets * 30.4 30.2 35.8 34.8 Networthto assets (11.2) (11.4) (13.1) (15.5) Notes:Thesymbols(***),(**),and(*)indicatestatisticalsignificanceatthe1,5,and10percent level,respectively.Standarderrorsinparenthesesanditalics.SampleincludesallNationalbanks in37citiesasdescribedinSection4.1. 64 Figure1 Distributionofownershipbytopthreemanagers Shareofsample(percent) 14 12 10 8 6 4 2 0 Ownershipshareoftopthreemanagers(percent) Note.SampleincludesallNationalbanksin37citiesasdescribedinSection4.1. Figure2 Distributionofownershipbyoutsidedirectors Shareofsample(percent) 30 25 20 15 10 5 0 Ownershipshareofoutsidedirectors(percent) Note.SampleincludesallNationalbanksin37citiesasdescribedinSection4.1. 65 Figure3 Distributionofownershipbynon‐managers,non‐boardmembers Shareofsmple(percent) 12 10 8 6 4 2 0‐2 4‐6 8‐10 12‐14 16‐18 20‐22 24‐26 28‐30 32‐34 36‐38 40‐42 44‐46 48‐50 52‐54 56‐58 60‐62 64‐66 68‐70 72‐74 76‐78 80‐82 84‐86 88‐90 92‐94 96‐98 0 Ownershipshareofnon‐managers/non‐boardmembers (percent) Portionofsharesownedbytopthree managers Note.SampleincludesallNationalbanksin37citiesasdescribedinSection4.1. Figure4 Managerownershipandboardcomposition 1 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 0 0.2 0.4 0.6 0.8 1 Portionofboardconsistingofoutsidedirectors Note.SampleincludesallNationalbanksin37citiesasdescribedinSection4.1. 66