when women lead, firms win - s&p global · firms that appointed a female ceo or cfo had a...

1

When Women Lead, Firms WinOctober 16, 2019

Key Takeaways

• ThestudyfindsthatfirmswithfemaleCFOsaremoreprofitableandgeneratedexcessprofits2of$1.8Toverthestudyhorizon.

• FirmswithfemaleCEOsandCFOshaveproducedsuperiorstockpriceperformance,comparedtothemarketaverage.Inthe24monthspost-appointment,femaleCEOssawa20%increaseinstockpricemomentumandfemaleCFOssawa6%increaseinprofitabilityand8%largerstockreturns.Theseresultsareeconomicallyandstatisticallysignificant.

• Firmswithahighgenderdiversityontheirboardofdirectorsweremoreprofitableandlargerthanfirmswithlowgenderdiversity.

• FirmswithfemaleCEOsandCFOshaveademonstratedcultureofDiversityandInclusion(D&I),evincedbyalargerrepresentationoffemalesonthecompany’sboardofdirectors.FirmswithfemaleCEOshavetwicethenumberoffemaleboardmembers,comparedtothemarketaverage(23%vs11%).

• Analysisofexecutivebiographiessuggeststhatonedriverofsuperiorresultsbyfemalesmaybethatfemalesareheldtoahigherstandard.Theaveragefemaleexecutivehascharacteristicsincommonwiththemostsuccessfulmaleexecutives,suggestingthatcommonattributesdrivesuccessamongmalesandfemales,alike.Overall,theattributesthatcorrelatewithsuccessamongmaleexecutiveswerefoundmoreofteninfemaleexecutives.Thisfindingrefutesthecommonlyheldbeliefin‘token’femaleexecutives.

Author

Daniel J. Sandberg,PhD,CFA,

QuantamentalResearch

www.spglobal.com

TheNewYorkTimesnotedthat“fewerlargecompaniesarerunbywomenthanbymennamedJohn”,inanarticle1publishedin2015.“TheJohns”wereinsecondplacebyyear-end2016,butnotbymuch(Figure1).AlthoughfemaleexecutivesremaingrosslyunderrepresentedintheC-suite,thissmallvictoryforgenderinclusionunderscoresachangingdynamic.Didthischangepay?

• Theanalysispresentedherein2isoneofthemostcomprehensiveexaminations,bybreadthandtimehorizon,ofgenderdiversity,todate.

• Amale-to-femaleratioof19:1forCEOand6.5:1forCFO,asofyear-end2018,exposesapersistingunderrepresentationoffemalesinkeyexecutivepositions,despiterecentadvancements.

• Evidenceoftheoutperformanceoffemaleexecutives,relativetotheirmalepeers,isoffered.FemaleCEOsdrovemorevalueappreciation3andimprovedstockpricemomentumfortheirfirms.FemaleCFOsdrovemorevalueappreciation,betterdefendedprofitabilitymoats,anddeliveredexcessrisk-adjustedreturnsfortheirfirms.

• Ananalysisofexecutives’biographiessuggeststhatthefemaleexecutiveswhohavebeenappointedtoC-suitepositionshaveattributes4consistentwiththemostsuccessfulmaleexecutives.Oneinterpretationofthisresultisthatfemaleexecutivesareheldtoahigherstandardbythecompanies’boardofdirectors,thantheirmalecounterparts.

October16,2019

2

Figure1.FemaleParticipationRateforChiefExecutiveOfficerPositions.RelativepercentageofcompaniesintheRussell3000Indexbygender.MalesaresubdividedbythosenamedJohnversusnotnamedJohn.Source:S&PGlobalMarketIntelligenceQuantamentalResearch.DataasofJune6,2019.

Introduction

In1986,CarolHymowitzandTimothySchelhardtcoinedtheterm‘GlassCeiling’asametaphorfortheforcesorcircumstanceswhichpreventfemaleprofessionalsfromreachingseniormanagementpositions.Inthethirty-threeyearssince,thetopicofgender5biashasreceivedgraduallyincreasingattention.Despitethisfocus,thefemaleparticipationrateinseniormanagementpositionsremainsfarfromparitytoday.Asofyear-end2018,thereareapproximately19maleCEOsforevery1femaleCEOand6.5maleCFOsforevery1femaleCFO,amongcompanieswithintheRussell3000Index.Theunderrepresentationoffemalesinkeyexecutivepositionshasraisedanumberofquestionsandinspiredempiricalresearchaimedatfindinganswers.

Unfortunately,thepaucityofdata(i.e.thelimitednumberoffemaleexecutivesandthelimitedavailabilityofstructured,historicaldata6relevanttothistopic)haslimitedthescopeofpreviousresearchuntilrecently.Earlyundertakingsattemptedtoextractinsightsbyevaluatingasfewas25diversefirms(Adler2000)orconsideringasingledatecross-sectionintheanalysis(Carter,Simkins,Simpson2002).Morerecentworkhasextendedthetimehorizon(Hunt,Layton,Prince2015)ormadeuseofamarket-representativeindexsuchastheS&P1500(Wolfers2006),withcaveatsarounddatalimitations.

TheanalyseshereinwillevaluatetheRussell3000universeovera17-yearperiod(December31,2002throughMay31,2019);including5,825newexecutiveappointments,ofwhich578werefemale;makingthisstudyoneofthemostcomprehensivecontributionstothetopicofgenderinequalityintheofficeoftheCEOandCFO.Despitethesizeofthisstudy,weadmonishthereadertointerprettheresultsasadescriptiveanalysis,relevantfromagovernancestandpoint,butnotprovidingevidenceofapredictivetradingsignal.

The Gender Effect

Amodifiedevent-study(MacKinlay1997)approachisusedthroughoutthispaperanddetailedinsection4.The“event”ofconsiderationisthebeginningofthetenureofanewexecutiveintheCEOrole(table1,figure2left)or,inaseparateanalysis,totheCFOrole(table2,figure2right).The collection of events in which the new appointee is female (male) is termed the female (male) contingent. The tables summarize the characteristics7,8 for firms on, and after, the appointment of a new executive. Averages are separately reported for the female and male contingents, as well as for the difference between the two contingents.

www.spglobal.com

When Women Lead, Firms Win

October16,2019

3

***=Significantatthe1%level;**=Significantatthe5%level;*=Significantatthe10%levelForeachvalueinthetableexceptBoardSizeandBoardFemaleParticipation,anaverageZ-scoreisreportedwithcorrespondingteststatisticinparentheses.Z-scoresarepresentedasapercentofonestandarddeviation.

www.spglobal.com October16,2019


The female contingent was associated with a greater value appreciation, defined as a declining book to market ratio, in the 24-month period after a female CEO or CFO took office. Comparatively, the male contingent was statistically indistinguishable from its sector peer group.WeakstatisticalevidencesupportsthatthisvalueappreciationwasassociatedwithanincreaseinintermediatetermpricemomentumforfemaleCEOappointments.ConsistentwithresultsreportedbyPeltomäkiandco-workers(Peltomäki,Swidler,Vähämaa2018),firmswhichappointedafemaleCFOalsohadhigherprofitability.Intheframeworkpresentedherein,wecorroboratethoseresultsandalsoshowthefemalecontingentmaintainedprofitability(average2-yearchangewasindistinguishablefrom0)whereasthemaleCFOcontingentsawaprofitability erosion.Theseobservationsareconsistentwithgreateraverageskillamongthefemale contingentthanthemalecontingent.

Thedataalsosupportculturaldifferencesbetweenfirmsinthetwocontingents,similartopreviousliterature.However,ourframeworkleadstoadifferentinterpretationthanpreviouswork.Forexample,KrishnanandParsons(2008)attributethecorrelationbetweenfirmswithhighgenderdiversityandhighearningsquality9totheways“womendifferintheirapproachtomoneyandinvesting”.Wefindthat,whilefirmsthatappointedafemaleCEOhadaboveaverageearningsquality(belowaverageaccruals)atthetimetheexecutivetookoffice,accrualsrevertedtothemean(increased)inthe24-monthperiodthereafter.Similarly,Peltomäkiandco-workers(2018)exploredthepremisethat“womentrytoavoidlossesandaremorecautious”,10showingthatfirmswithfemaleCFOsemploylowerfinancialleverage11thantheirmalecounterpartsassupport.Again,ouranalysesfindsimilarresultswithstatisticallylowerfinancialleverageforthefemalecontingentofbothCEOandCFOpositionswhentheexecutivetakesoffice.However,thefemalecontingentfirmsincreasedleverageinthe24monthsfollowingtheCEO’sstartdateandmaintainedleverageinthe24monthsfollowingtheCFO’sstartdate.Therefore,thecausalrelationshipisquestionableandpossiblyreversed.Inotherwords,ouranalysissupportsthatfirmswithhigherearningsqualityandlowerleveragearefirmswithacultureconducivetomakingafemaleappointment,ratherthanthepremisethatstereotypicaldifferencesintheactionsofthefemaleexecutives,aftertheirappointment,drivethesedifferences.

FirmsthatappointedafemaleCEOorCFOhadahigherfemaleparticipationrateontheirboardofdirectorscomparedtofirmsthatmademaleappointments.Empiricalevidencesupportsagrowthinthefemaleparticipationrateoftheboardoverthefirst24monthsfollowingtheappointmentofafemaleCEO.Theseobservationsfurthersupporttheideathatdiversityandinclusionarefeaturesthatgraduallyinfuseintothecultureofafirm.

Table 1: Firm Characteristics Associated with CEO Appointments by Gender(Russell3000,12/31/2002–5/31/2019)

4

Table 2: Firm Characteristics Associated with CFO Appointments by Gender(Russell3000,12/31/2002–5/31/2019)

***=Significantatthe1%level;**=Significantatthe5%level;*=Significantatthe10%LevelForeachvalueinthetableexceptBoardSizeandBoardFemaleParticipation,anaverageZ-scoreisreportedwithcorrespondingteststatisticinparentheses.Z-scoresarepresentedasapercentofonestandarddeviation.

SourceforTables1and2:S&PGlobalMarketIntelligenceQuantamentalResearch.DataasofJune6,2019.Indicesareunmanaged,statisticalcompositesandtheirreturnsdonotincludepaymentofanysaleschargesorfeesaninvestorwouldpaytopurchasethesecuritiestheyrepresent.Suchcostswouldlowerperformance.Itisnotpossibletoinvestdirectlyinanindex.Pastperformanceisnotaguaranteeoffutureresults.



5

***=Significantatthe1%level;**=Significantatthe5%level;*=Significantatthe10%LevelFigure2.Fama-French5(FF5)FactorAdjustedReturns.TheaverageFF5residualreturndemeanedatthesectorlevelisreportedforthemaleandfemalecontingentinthe36-monthsfollowingappointmentofanewCEO(left)andCFO(right).

Table 3: Adjusted Returns Following New Executive Appointments by Gender(Russell3000,12/31/2002–5/31/2019)

***=Significantatthe1%level;**=Significantatthe5%level;*=Significantatthe10%LevelSourceforFigure2andTable3:S&PGlobalMarketIntelligenceQuantamentalResearch.DataasofJune6,2019.Indicesareunmanaged,statisticalcompositesandtheirreturnsdonotincludepaymentofanysaleschargesorfeesaninvestorwouldpaytopurchasethesecuritiestheyrepresent.Suchcostswouldlowerperformance.Itisnotpossibletoinvestdirectlyinanindex.Pastperformanceisnotaguaranteeoffutureresults.

Afteradjustingfordifferencesinfirmcharacteristics(Fama,French2015)andsectorperformance,wefoundthefemalecontingentearnedlargeradjustedreturnsthanthemalecontingentamongtheCFOposition,butnottheCEOposition(Figure2).FortheCFOposition,thetestforthedifferenceoftwomeansindicatedamaximumdifferenceofgreaterthan8%betweencontingents,occurringatthe24-monthtimehorizonandstatisticallysignificantatthe1%level.ThemalecontingentofCFOappointmentsproducedreturnsthatwerestatisticallyindistinguishablefromthesectoraveragethroughoutthebacktest,whereasthefemalecontingentyieldedanaveragepremium.

AveragereturnstofirmsinthetwocontingentsfollowingtheappointmentofanewCEOwerestatisticallyindistinguishablefromeachother.Themalecontingentyieldedasmallpositivepremiumwithweakstatisticalsignificanceattimehorizonsof9-18months,whereasthefemalecontingentandthetwo-populationdifferencefailedtomeetthetestforstatisticalsignificance.AcloserinspectionofthestandarderrorsforthecontingentswithintheCEOappointmentsshowedthatoursamplemeanswouldhavehadtodifferbymorethan7%(ineitherdirection)tomeetstatisticalsignificanceatthe10%threshold,comparedtoadifferenceofjust5%forthepositionofCFO.ThedifferenceofmeansbetweencontingentsfortheCEOpositionfallswellbelow7%.Notethatthehighthresholdforsignificanceisalmostentirelyattributabletothesmallsamplesizeofonly143femaleCEOappointments.



6

Talent is Equally Distributed

Thepriorhypothesisattheoutsetofthisstudywasthattalentisequallydistributedacrossgenders.Intheprevioussection,wefindevidencethatfemaleexecutivesdrivegreatervalueappreciation,improvepricemomentum,betterdefendprofitabilitymoats,andearnexcessreturnsovertheirmalecounterparts.Dothesetwoassertionsconflict?

Wearguetheydonot.Rather,theboardofdirectorsmaybeholdingfemaleappointeestoahigherstandardthanmaleappointees,suchthatthefemalesinC-suitepositionsareconsequentlymoretalented.Thehighmale-to-femaleratioofexecutivesinC-suitepositionssupportsthispremise.Beingmoreselectivewithfemaleappointees,meansthattheboardofdirectorsmaypassoveramorequalifiedfemaleinfavorofalessqualifiedmale.Ifthisisthecase,itfollowsthattheremainingpooloffemalecontendersforC-suitepositionsremainsricherwithtalent.

Insupportoftheaforementionedpremise,weshowbelowtheresultsofanaturallanguageprocessing(NLP)analysiswhichdemonstratesthattheachievements,education,orpersonaltraitsassociatedwithsuccessoccurmoreoftenwithinthefemalecontingent.Thefeaturesassociatedwithsuccessfortheappointedexecutivesinthisstudywereextractedfromthoseexecutives’biographies,whichareincludedintheS&PCapitalIQProfessionalsdataset.First,adictionarywastrainedonthecorpusexcludingthefemalecontingent(trainingset).Thepositivityofaparticularword12wasdeterminedbytherelativeoccurrenceofthatword13amongcompaniesthatearnedpositiveexcessreturnsversusthosethatdidnot,insideofthetrainingset.Separately,therelativeoccurrenceofthesamesetofwordsinthefemalecontingent(thetestset)relativetothemalecontingentwasevaluated.Inregressionplots(Figure3),wefoundthattherelativeoccurrenceoflanguageusedtodescribeallthefemaleexecutives,versusallmaleexecutives,washighlycorrelatedwiththelanguageusedtodescribethesuccessfulmaleexecutives.

Theimplicationofthepositivecorrelationbetweenthelanguageusedtodescribeallfemaleexecutivesandsuccessfulmaleexecutivesisprofound.Unlikesomepreviousliteraturewhichattributesperformancedifferencestogender-specificbehaviorsoraversions,ouranalysissupportscommonfeaturesfavorsuccessformalesandfemalesalike,andthosefeaturesaremoreprevalentinthefemalecontingent,todate.Ourinterpretationisthatthemalecontingentisrelatively‘overfished’comparedtothefemalecontingent,asadirectresultofabiaspreventingwomenfromC-suiteappointments(theso-calledglassceiling).14



7

www.spglobal.com

Figure 3. Natural Language Processing of Executive Biographies. Foreachoftheexecutivesinourstudy,theexecutive’sbiographywasparsedbyaNaturalLanguageProcessingprocedure,whichidentifiesthepositivityandfemininityoftokenizedwords.Apositiveandsignificantcorrelationwasobservedinregressionsoffemininityscoreonpositivityscore.

Source:S&PGlobalMarketIntelligenceQuantamentalResearch.DataasofJune6,2019

Assumingourinterpretationiscorrect,theregressioncoefficientshouldapproach0asexecutiveappointmentsreachgenderparity.Inotherwords,ifC-suiteappointmentshavehistoricallybeenmadeonthebasisofmeritwithaprovisoonmalegender,wepositthatremovingthatprovisoandallowingthesystemtoequilibratewillshowthatmaleandfemaleexecutivesareequallyequippedtodrivetheirfirms’success.

Methodology and Data

Themethodologyandtoolsusedinthisresearcharereviewedinthissection.

Data

TheS&PCapitalIQProfessionalsDatasetprofilesprofessionalswithcurrentandpriorboard/companyaffiliations.Dataincludebiographies,standardizedjobfunctions,titles,education,compensation,optionsholdings,andfullcommitteememberships.Thisdatasetcovers4.5millionprofessionalsinternationally,withrobustcoveragefortheRussell3000startingin2002.CompanyfundamentaldatawereobtainedfromtheAlphaFactorLibrarypackage,whichprovideshundredsofpre-calculatedfactorsincludingfinancialratios,valuationmetrics,andpriceandmomentumstatistics.Allfactorsareconstructedusingpointin-timedata.AdditionalcompanyfundamentalsandpricingwereobtainedfromtheCapitalIQFinancialsDataset,whichcontainspoint-in-timeglobalcoverageofkeyfinancialmetricsandreportedfinancials.InadditiontocontentfromtheS&PGlobalMarketIntelligenceecosystem,thisstudyutilizedfreethird-partydatafromtheUnitedStatesSocialSecurityAdministration(SSA).15TheSSAmaintainsadatabaseofbabyfirstnames,babysex,yearof-birthandtotalcountforallnewbornsintheUnitedStates.Thesedatawereused,asdescribedinsection4.2.

Gender Assignments

1.IncludedwithintheProfessionalsdatabaseisafieldlabeled‘prefix’.Whentheprefixfieldwasequalto‘Mr.’,‘Sir’,‘Count’,‘Father’,‘Sheikh’,‘Bishop’,‘Lord’,‘Hafiz’,‘Baron’,or‘Janab’thentheexecutivewasassumedtobemale.Whentheprefixfieldwasequalto‘Mrs.’,‘Miss’,‘Ms.’,‘Sister’,‘Lady’,‘Madam’,‘Countess’,‘Baroness’,or‘FirstLady’thentheexecutivewasassumedtobe

October16,2019


8

female.Forallotherprefixes(suchas‘Dr.’,‘Professor’,‘Lieutenant’,etc.)thegenderwasassigned‘ambiguous’forthismethod.

2.Thebiographiesofeachexecutivewereparsedforthepresenceofgenderrelatedpronouns(“he”,“him”,“his”,“she”,“her”,“hers”).Ifaminimumof90%ofthepronounsinthebiographywerespecifictoonegender,thatgenderwasassumedfortheexecutive;otherwise,thegenderwasassigned‘ambiguous’forthismethod.

3.DatafromtheU.S.SocialSecurityAdministrationwereusedtocalculatethegendercertaintyassociatedwithafirstnameandyearofbirth.Forexample,in1975,99.3%ofbabiesnamed‘John’weremale.Ifthegendercertaintyofanexecutive’sfirstnameintheyeartheexecutivewasbornwasgreaterthan90%,thentheexecutive’sgenderwasassignedassuch;otherwisethegenderwasassigned‘ambiguous’forthismethod.

Afterthe3stepswerecompletedforeachexecutiveinthestudy,thegenderassignmentswereprogrammaticallycomparedforagreement,ignoringambiguousresults.Ambiguousrecordswereresolvedbyawebsearch.

Universe and Event Detection

TheconstituentsoftheRussell3000werefilteredtoremovepennystocksandlow-pricedstocks,duetodifficultyreliablydeterminingthestartdateoftheexecutivesformanyofthesefirms.ChangestotheuniquepersonidentifierassociatedwiththeCEOorCFOpositionoftheremainingfirmstriggeredapotentialeventforanalysis.Tominimizetheimpactofinterimexecutivesontheresults,aforwardlookinganalysiswasdoneforeachpotentialeventandiftheexecutivewasreplacedwithin24monthsofstartingthepositionthentheeventwasremovedfromtheanalysis.

Comparative Statistical Framework

Priortoaveraging,financialratioswerenormalizedbycomputingasector-relativecrosssectionalZ-scorebyusingequation1,

eqn.1

eqn.2

where∆𝑍𝑍 𝑍isthe change inthe Z-scored metric; 𝑍𝑍𝑍(24) represents the Z-scored metric24monthsaftertheexecutive’sstartdate;and𝑍𝑍 𝑍(0) represents the Z-scored metric ontheexecutive’sstartdate.

Natural Language Processing

Thebiographyofeachnewlyappointedexecutiveinthisstudyformedthecorpusforanatural



9

languageprocessing(NLP)analysis.Thedictionaryfortheanalysiswasdefinedasthesetofuniquetokensgeneratedbyparsing,tokenizing,andstemming(Paice1990)allwordsinthecorpus.Thefollowingtokenswereremovedfromthedictionary16:1)stopwords,asdefinedbyPython’sNLTKmodule(Bird,Loper,Klein2009),2)wordsthatwereuniquetooneofthecontingentsofthecorpus,suchas‘chairwoman’,and3)numericaltokenssuchasyearsanddates.Thefinaldictionarycontainedapproximately3,000uniquetokens.

Themalecontingentofthecorpuswasusedastrainingdatatoassignapositivityscoretoeachtokeninthedictionary.First,thecontingentwassubdividedintoanoutperformsubset,containingfirmswithpositiverisk-adjustedreturns(aswereusedinFigure2),andanunderperformsubset.Thepositivityscorewascalculatedbyusingequation3,

eqn.3

Usingthefullcorpus(maleandfemalecontingents),afemininityscorewasassignedtoeachtokeninthedictionary,byusingequation4,

eqn.4

Concluding Remarks

InoneofthelargeststudiesongenderintheC-suite,todate,evidenceofunderrepresentationandoutperformanceamongfemaleexecutivesrelativetotheirmalepeershasbeenpresented.Specifically,overthetime-horizonofthestudy,femaleCEOssawmorevalueappreciationandimprovedstockpricemomentumfortheirfirms;whereasfemaleCFOsdrovemorevalueappreciation,betterdefendedprofitabilitymoats,anddeliveredexcessriskadjustedreturnsfortheirfirms.Weproposedthattheobservedoutperformancewasaresultofabove-averagetalentamongfemaleexecutives.ThefemalecontendersforC-suitepositionsrepresentarelativelyunderutilizedpooloftalent,possiblyattributabletoahigherdegreeofscrutinyfromthefirms’boardofdirectorsandconsequentlyresultinginthetendencyoffemalesinC-suitepositionstobemoretalented.Assupportforthepremise,anaturallanguageprocessing(NLP)techniquewasappliedtothebiographiesofexecutivesandtheconclusionthatfemaleexecutivesmorefrequentlypossessedtheattributesassociatedwithsuccessamongtheirmalecounterpartswasdemonstrated.Ifourpremiseiscorrect,thedifferencescitedshoulddissipatewhenfemalesareequallyrepresentedinC-suitepositions.Inotherwords,talentisequallydistributedanduntilexecutivesareselectedonthebasisoftalentwithoutotherbiases,weexpectchangepays.

Endnotes

1Wolfers,J.,2015.“FewerWomenRunBigCompaniesThanMenNamedJohn.”NewYorkTimes.2Section4providesdetailsonthedatasetcoverage,universedefinition,andmeasurementtimehorizon.3Valueappreciationisdefinedasadecreaseinthebook-to-marketmultiplerelativetothesectoraverage.Seesection4formethodologydetails.4Theprocessofdefiningthedictionaryofattributesisdetailedinsection4.5.5Ourchoiceofdictionregarding“gender”versus“sex”,usedthroughoutthework,isdiscussedinmoredetailinAppendix7.1.



10

6Theinterestedreaderisreferredtosection4.1ofthispaperformoredetailontheS&PGlobalProfessionalsdataset,releasedin2012,whichmadethisresearchpossible.7Across-sectionalZ-scorewascalculatedforallcharacteristicsbeforeaveraging.Additionaldetailsareprovidedinsection4.4.8RobustnesschecksforthetabulatedcalculationscanbefoundinAppendices7.2and7.3.9Highearningsqualityisdefinedasloweraccrualsrelativetothesectoraverage,asdetailedinsection4.10Peltomäkiandcoworkerspresentevidencetothecontraryandultimatelyconcludetheirempiricalfindingsareambiguous.11Financialleverage,orleverage,isdefinedasdebttoassets.12Seeappendix7.4forexamplesofpositiveandnegativewordsobtainedfromtheCEOanalysis.13Thephrase“relativeoccurrenceofwords”isdefinedasthepercentageofbiographieswithinaparticularportionofthecorpusthatcontaintheword,lessthesamepercentageinitscounterpart.Forexample,therelativeoccurrenceofawordinthefemalecontingentwouldbeequaltothepercentageoffemalebiographiescontainingtheword,lessthepercentageofmalebiographiescontainingthesameword.Seesection4formoredetailsontheNLPprocedure.14Seeappendix7.4forexpandeddiscussionandalternativeexplanations.15Datadownloadavailableathttps://www.ssa.gov/oact/babynames/limits.html16TheremovaloftokensfromthedictionarywasperformedonthebasisofstandardNLPprotocol(suchasremovalofstopwords)andlogic(suchasremovalofgenderspecificwords).Toensurethattheremovaloftokenswasnotcreatingspuriousrelationships,robustnesscheckswereperformedandarediscussedinappendix7.4,alongwithanexpandeddiscussionontheNLPmethodology.

Thesematerialshavebeenpreparedsolelyforinformationpurposesbaseduponinformationgenerallyavailabletothepublicandfromsourcesbelievedtobereliable.Nocontent(includingindexdata,ratings,credit-relatedanalysesanddata,research,model,softwareorotherapplicationoroutputtherefrom)oranypartthereof(Content)maybemodified,reverseengineered,reproducedordistributedinanyformbyanymeans,orstoredinadatabaseorretrievalsystem,withoutthewrittenpermissionofS&PGlobaloritsaffiliates(collectively,S&PGlobal).TheContentshallnotbeusedforanyunlawfulorunauthorizedpurposes.S&PGlobalandanythird-partyproviders,(collectivelyS&PGlobalparties)donotguaranteetheaccuracy,completeness,timelinessoravailabilityoftheContent.S&PGlobalPartiesarenotresponsibleforanyerrorsoromissions,regardlessofcause,fortheresultsobtainedfromtheuseoftheContent.THECONTENTPROVIDEDON“ASIS”BASIS.S&PGLOBALPARTIESDISCLAIMANYANDALLEXPRESSORIMPLIEDWARRANTIES,INCLUDING,BUTNOTLIMITEDTO,ANYWARRANTIESOFMERCHANTABILITYORFITNESSFORAPARTICULARPURPOSEORUSE,FREEDOMFROMBUGS,SOFTWAREERRORSORDEFECTS,THATTHECONTENT’SFUNCTIONINGWILLBEUNINTERRUPTEDORTHATTHECONTENTWILLOPERATEWITHANYSOFTWAREORHARDWARECONFIGURATION.InnoevenshallS&PGlobalPartiesbeliabletoanypartyforanydirect,indirect,incidental,exemplary,compensatory,punitive,specialorconsequentialdamages,costs,expenses,legalfees,orlosses(including,withoutlimitation,lostincomeorlostprofitsandopportunitycostsorlossescausedbynegligence)inconnectionwithanyuseoftheContentevenifadvisedofthepossibilityofsuchdamages.

S&PGlobal’sopinions,quotes,andcredit-relatedandotheranalysesarestatementsofopinionasofthedatetheyareexpressedandnotstatementsoffactorrecommendationstopurchase,hold,orsellanysecuritiesortomakeanyinvestmentdecisions,anddonotaddressthesuitabilityofanysecurity.S&PGlobalassumesnoobligationtoupdatetheContentfollowingpublicationinanyformorformat.TheContentshouldnotbereliedonandisnotasubstitutefortheskill,judgementandexperienceoftheuser,itsmanagement,employees,advisorsand/orclientswhenmakinginvestmentandotherbusinessdecisions.S&PGlobalkeepscertainactivitiesofitsdivisionsseparatefromeachotherinordertopreservetheindependenceandobjectivityoftheirrespectiveactivities.Asaresult,certaindivisionsofS&PGlobalmayhaveinformationthatisnotavailabletootherS&PGlobaldivisions.S&PGlobalhasestablishedpoliciesandprocedurestomaintaintheconfidentialityofcertainnon-publicinformationreceivedinconnectionwitheachanalyticalprocess

Copyright©2019byS&PGlobalInc.Allrightsreserved.



when women lead, firms win - s&p global · firms that appointed a female ceo or cfo had a...

Documents