Openness, international champions, and the internationalization of Multilatinas

Openness,

international

champions,

and

the

internationalization

of

Multilatinas

Jean-François

Hennart

,

Hsia

Hua

Sheng

*,

José

Marcos

Carrera

Jr.

a

CentER,TilburgUniversity,Heuvelstraat14,5131APAlphen,TheNetherlands b

SãoPauloSchoolofBusinessAdministration(EAESP),GetulioVargasFoundation(FGV),RuaItapeva,474,8thFloor,SãoPaulo,SP01332-000,Brazil c

SãoPauloSchoolofBusinessAdministration(EAESP),GetulioVargasFoundation(FGV),RuaBage139ap51,VillaMariana,SãoPaulo,SP04012-140,Brazil

ARTICLE INFO

Articlehistory: Received19May2015

Receivedinrevisedform24July2016 Accepted28August2016

Availableonline6September2016 Keywords: Multilatinas Internationalization State-ownedenterprises Nationalchampions Emergingmarkets

Newvarietiesofstatecapitalism Brazil

ABSTRACT

Inthe1990sLatinAmericancountriesabandonedtheirpoliciesofimport-substitutingindustrialization carriedoutthroughfully-ownedstateenterprises(SOEs).Theyopenedtheireconomiestointernational competitionandprivatizedtheirSOEs.Wearguethatthispragmaticadaptationdidnotnecessarily constituteafundamentalchangeinpolicies,longfollowedbysomeLatinAmericancountries,ofstate interventioninthepursuitofnationalisticobjectives,butisinsteadacontinuationofthesepoliciesby othermeans.Specifically,tosafeguardtheirautonomy,someLatinAmericanstateshaveselectedand nurtureddomesticfirmstobecomemultinationalenterprises(MNEs).Theyhavekept–andobtained– equitystakesinthesenationalMNEstoinfluencethemandtokeepthemoutofthehandsofforeigners. ThesepoliciesexplainthetimingoftheriseofMultilatinasandtheir,usuallypartial,stateownership. ã2016ElsevierInc.Allrightsreserved.

1.Introduction

Theriseofmultinationalenterprisesfromemergingcountries

(EMNEs)hasgeneratedconsiderableinterestamonginternational

business scholars (e.g. Cuervo-Cazurra & Ramamurti, 2014;

Hennart,2012;Ramamurti&Singh,2009;Williamson,Ramamurti, Fleury,&Fleury, 2013; Zeng&Williamson,2007).Muchofthe

literaturehas focused onChinese and Indian EMNEs with less

attentiondevotedtoLatinAmericanEMNEs,theMultilatinas.

Cuervo-Cazurra(2008)notesthattheinternationalexpansion

ofmanyMultilatinasdidnotstartinearnestuntilthelate1980s.He

arguesthatitcanbetracedtomajorchangesintheinstitutional

contextoftheirhomecountries,specificallytheopeningupoftheir

economiesandtheprivatizationoftheirstate-ownedenterprises

(SOEs). Until the late 1980s, Latin American countriespursued

policies of industrialization through import substitution (ISI).

UnderISI,importswerediscouragedinordertostimulatedomestic

production.Theinabilityorunwillingnessofdomesticproducers

torespondtotheseincentivesledmanyLatinAmerican

govern-ments to fill the gap with fully-owned SOEs, which came to

dominate many Latin American economies (Musacchio &

Lazzarini,2014b).Inthishighlyprotectedandregulated

environ-ment,bothprivatefirmsandSOEshadfew incentivestosellor

investabroad.Highentrybarriersprovidedjuicyopportunitiesfor

localentrepreneurswithgoodgovernmentconnections,soitmade

moresenseforthemtodiversifyintonewindustriesthantotarget

foreign customers. Governments saw the mission of SOEs as

meeting domestic demand, soSOEs had few incentives tosell

abroad.LatinAmericangovernmentsweregenerallyopposedto

foreigninvestmentsbybothprivatefirmsandSOEsbecausethey

sawsuchinvestmentsasusingupfundsthatcouldbeinvestedat

home(Wells,1971).ConsequentlybothprivatefirmsandSOEshad

lowlevelsofinternationalization.

Inthe1990sthesepoliciesbecameeconomicallyunsustainable.

Theywereabandonedandgovernmentsopenedtheireconomies

to foreign competition, opting for lower barriers to trade and

incomingforeigndirectinvestment.LatinAmericangovernments

alsostartedtoengageintheprivatizationoftheirSOEs.Theyended

up withless than full stakesin a largenumber of private and

publiclylistedfirms,resultingina“newvarietyofstatecapitalism”

(Musacchio, Lazzarini, & Aguilera, 2015): state control is now

generally exercised indirectly through national development

banksandthepensionfundsofSOEsandnewlyprivatizedSOEs

(Cuervo-Cazurra,Inkpen,Musacchio,&Ramaswamy,2014).

While some authors have attributedthe sudden rise of the

Multilatinasinthelate1980stotheopeningupoftheireconomies

*Correspondingauthor.

E-mailaddresses:j.f.hennart@uvt.nl(J.-F.Hennart),Hsia.sheng@fgv.br,

hsia_sheng@yahoo.com.br(H.H.Sheng),josemarcoscj@gmail.com(J.M.Carrera).

http://dx.doi.org/10.1016/j.jwb.2016.08.005

1090-9516/ã2016ElsevierInc.Allrightsreserved.

ContentslistsavailableatScienceDirect

Journal

of

World

Business

(Cuervo-Cazurra,2008;DelSol&Kogan,2007),lessattentionhas

beenpaidtotheimpactofthechangeinthelevelandmodalitiesof

stateownership.Yet,atleastinthecaseofBrazil,thereseemstobe

a clearlink between internationalization and state ownership.

Indeed,thelistofthe20largestBrazilianMNEsbysizeofforeign

assets shows that two-thirds of them have some degree of

governmentownership(Sheng&Carrera,2016).Isthisfortuitous,

oristhereasolidrelationshipbetweengovernmentownershipand

afirm’sdegreeofinternationalization?Andifso,whatcausesthis

relationship?AsCuervo-Cazurraetal.(2014)note,ourtheoretical

understandingoftheimpactofthenewformsofstateownership

oninternationalizationisstilllimited.

Afewstudieshavefocusedonthisrelationship,butalmostallof

themhavefocusedonChina(e.g.Li,Cui,&Lu,2015;Liang,Ren,&

Sun, 2015; Wang, Hong, Kafouros, & Wright, 2012). China is,

however,verydifferentfromLatinAmerica.TheChinese

govern-menthassetclear“goglobal”policiesforChinesefirmsandhas

unparalleledpowertoimplementthem.MostChinesefirmsare

fullyoroverwhelminglyownedbythestate.Theirmanagersare

firstandforemostpublicservantswhosecareerprospectsdepend

on their obedience to central government directives: they are

“cadresfirst andcompany mensecond”(Economist,2012).The

ChineseCommunistPartyactivelyintervenesinthemanagement

ofallChinesefirmsthroughaparallelstructureimplantedinall

firms,publicandprivate,andmakessurethatmanagementfollows

governmentalguidelines(Economist,2012).

Incontrast,LatinAmericangovernmentpoliciesofsupportto

theinternationalizationofdomestic firmsaremoreambiguous,

andthestateleverageondomesticfirmsmoretenuous.Inthecase

ofBrazil,thegovernmentstakeisonlypartial,andthegovernment

mustoftencontendwithmajorityowners,oftenfamilies.Afurther

complicatingfactoristhattheBraziliangovernment’sstakesare

generallyindirectlyheldbythestatedevelopmentbankandthe

pensionfundsofSOEsorex-SOEs.Hencetherelationshipbetween

governmentownershipandinternationalizationis,intheBrazilian

case, much less obvious, and the mechanisms by which the

governmentinfluencesfirms moredifficulttoassess(Caseiro &

Maseiro,2014).

Inthispaperwecontributetotheliteraturesontheimpactof

home-countryinstitutionsonafirm’slevelofinternationalization.

Specifically,we studytheimpactof the“new varietiesof state

capitalism”(Musacchio,Lazzarini,&Aguilera,2015)onthelevelof

internationalizationofMultilatinas.By levelof

internationaliza-tionwemeantheshareofforeignsalesintotalsales,withforeign

salesincludingbothexportsandlocalsubsidiarysales.

We provide a novel explanation for the existence of a

relationshipbetweenstateownershipandtheinternationalization

of domestic firms. We argue that sucha relationship arises in

countrieswhereanationalisticandneo-developmentalist

ideolo-gyseesthedevelopmentofinternationallycompetitivedomestic

firmsasthebestwaytosafeguardstatepower.Insuchcountries

the state keeps strategic stakes in the privatized SOEs with

internationalpotentialandexpandstherealmofstatecontrolby

takingstakesinprivate_firmswithstronginternationalprospects.

Bothgroupsoffirmsareencouragedtogainbulkbymergingwith

otherdomesticfirmssoastoincreasetheirbargainingpower,and

thenreceivestronggovernmentsupportintheir

internationaliza-tion.Asaresult,thestateendsupowningsignificantequitystakes

inthosedomesticfirmswhicharehighlyinternationalized.

Afterexplainingthelinkbetweenthelevelofstateownershipin

afirmanditsdegreeofinternationalization,wegoontoexplore

whether with whom the state partners has an impact on

internationalization.We hypothesize thathaving thestateas a

minorityownerislikelytopushfamily-managedfirms,andfirms

wherefamilies,foreigners,andthestatesharecontrol,tohavea

higher ratioof foreign salestototal sales thanwhen the state

partnerswithfirmswithdispersedownership.Ontheotherhand,

weexpectpartiallystate-ownedfirmsinwhich foreignershave

majoritystakestobelessinternationalizedthanfirmsinwhichthe

governmentsharescontrolwithfamilies.

We test our hypotheses on a sample of Brazilian listed

companies over the2002–2011 period. Brazil is an interesting

contextbecause,aswewillsee,itprovidesthebestLatinAmerican

example of liberal neo-developmentalist policies that harness

market forcestosupportstatepower.Specifically,theBrazilian

statehascarriedoutdirectpoliciesofsupporttointernationalizing

domesticfirmsbykeepingownershipstakesinprivatizedSOEsand

bytakingnewonesinprivateBrazilian_firmswithinternational

potential.Thesepoliciesofsupporttonationalchampionsgoalong

way towards explaining the internationalization of Brazilian

Multilatinas, and explain the link between internationalization

andgovernmentownership.

Controllingforpossibleendogeneity,wefindthatthehigher

the total government stake in a firm, thehigher its degree of

internationalization,measuredbyitsratioofforeigntototalsales.

We findalsothat family firms withstateownership and firms

wherethestatesharescontrolwithfamiliesandforeignershavea

highershareofforeigntototalsalesthanfirmsinwhichthestate

partners withdispersedowners,butthat stateownershiphasa

weakerimpactoninternationalizationinfirmswhereforeigners

aredominantshareholders.

Inthenextsection,weoutlineatheoryofstatesupportforthe

internationalization of domestic firms and explain why state

ownershipisassociated withhighforeignsalesintensitybefore

applyingthetheorytothecase ofBrazil.Wethendescribeour

sample, our methodology and our results. We conclude with

implicationsforresearchonemergingmarketmultinationalsand

onthenewvarietiesofstatecapitalism.

2.Theoreticalbackgroundandhypotheses

Thedawnofcapitalismwitnessedadebatebetweenthetenants

of economic liberalism (e.g. John Stuart Mill, 1848) and the

defendersofthenationstate,suchasList(1841).WhileList(1841)

isbetterknownforhis‘infantindustry’argumentfor

protection-ism,hiscriticismofeconomicliberalismwasmorefundamental.

Along withthemercantilists, hearguedthat“economicliberals

evaluatedeconomic policies fromthestandpoint of individuals

andthewelfareofhumanityasawhole”(Helleiner,2002:311),but

bypassedthenationstate,theunitofanalysisin-between.Forhim,

economic policiesshould not aim at increasing an individual’s

welfare,butataugmentingacountry’swealth,power,andnational

identity.

Liberals, onthe other hand, have argued that nation states

shouldnotinterferewithinternationaltradeandinvestmentflows,

becausefreetradeand investmentyield amoreefficientuseof

scarceresources,thusmaximizingwelfareatboththeglobaland

individual levels. They see economic relations as positive-sum

gamesandthegoalofeconomicactivityasthemaximizationof

globalwelfare.Incontrast,mercantilistsandeconomicnationalists

takeamorepessimisticviewofeconomicrelationsandseethemas

essentiallyconflictualand zero-sum gameswhichaimat

redis-tributingwealthandpowerbetweennations(Gilpin,1976).

In the post-WW2 era, economic nationalism led most

developingcountriestoadoptISIpolicies.Thesepoliciesadvanced

thenational interestby protectingdomestic firmsfromforeign

competition–throughhightradebarriersandrestrictionsonthe

entryofforeignfirms–untiltheycouldstandontheirowntwo

feet. ISI policies also protected domestic markets from being

colonizedbyforeignMNEs,which wereseenasinstrumentsby

whichpowerfulnationstatesprojectedtheirpowerand

Under ISI, governments actively intervened through price

settingandextensiveregulations(Cuervo-Cazurra,2008).Inthis

highlyprotectedandregulatedenvironment,domesticfirmsfaced

littlecompetitionandenjoyedsubstantialmarketpower.Theyhad

fewincentivestoundertakelarge-scaleinvestmentprojectswith

longeconomiclifeandsubstantialspillovers(Gerschenkron, 1962).

Toremedythis,governmentssetupwholly-ownedSOEs(

Musac-chio&Lazzarini,2014b).Thisresultedinanindustriallandscape

dominated by SOEs and by diversified private local business

groups,oftenunderthecontroloffamilies.

ISIpoliciesturnedouttohavetwomajordisadvantages.First,

thelackofexposuretointernationalcompetitionmadebothSOEs

andlocalfirmsincreasinglynon-competitive(Liu,1993).Domestic

privatefirmshadlittleinterestinexpandingabroad,asitmade

moresenseforthemtodiversifyintootherdomesticindustries

than to target foreign markets, since domestic diversification

allowed them to capitalize on their ties with the government

(Schneider,2008).Governmentsdiscouragedforeigninvestments

by SOEs and private firms alike because they thought such

investmentsusedupfundsthatcouldbeinvestedathomeand

replaceddomesticbyforeignjobs(Wells,1971).Domesticfirms,

both private and public, were inefficient and wellbelow their

internationalizationpotential.

Relying on SOE for national development had also major

disadvantages. SOEs suffer from a two-stage principal-agent

problembecausetheprincipalschargedtomonitorSOEmanagers

arethemselvestheagentsofthecitizens,whohavefewincentives

to get involved. This gives politicians and SOE managers

considerablelatitudetopursuetheirowngoals(Cuervo-Cazurra

etal.,2014).GovernmentsoftenrequireSOEstomaximizesocial

objectivesaswellasprofits,makingitdifficulttoevaluatetheir

performance and todraw incentivecontracts tomotivate their

managers.Mostothermechanismsusedtomonitorthemanagers

ofprivate firmsarealso absentin SOEs.Many SOEsoperate in

monopolisticindustries,makingcomparisondifficult. SOE

man-agers are often public servants, with advancement based on

seniorityratherthanperformance.Theyarenotdisciplinedbythe

threat of bankruptcy, as their firms often enjoy soft budget

constraintsandhavetheirlossescoveredbythestate(Schleifer,

1998),norbythepossibilityoflosingtheirjobduetoatakeover.

These governance problems, plus government pressures to

maintain SOE employment and prices at unsustainable levels,

led to widespread losses. These were covered by government

deficits,inlargepartfinancedthroughexternaldebt.Whenthe

level of foreign debt became excessive, these policies became

unsustainable(Musacchio&Lazzarini,2014b).

By thelate 1980s,LatinAmerican governmentsbecamealso

keenlyawarethattherateofgrowthobtainablethroughISIpolicies

wasmuchlowerthanthatofAsiancountriespursuingexport-led

growthstrategies.ThisledtoareappraisalofISIpoliciesandtotwo

major changes: First, Latin American governments dismantled

pricecontrols,loweredtariffs,andliberalizedtheentryofforeign

firms(Cuervo-Cazurra,2008).Second,theyprivatizedtheirSOEs.

Openingtheeconomytointernationalcompetitionhadmany

benefits.Italloweddomesticfirmstoobtainbetterqualityinputs;

italsoforcedthemtobecomemorecompetitive.Inmanycases,

foreignentrantsintroducedbetterproducts,services,and

mana-gerialpracticeswhichdomesticfirmshadtoadoptiftheywanted

to survive (Cuervo-Cazurra & Dau, 2009). Likewise, selling off

partialorfullstakesinSOEstotheprivatesectorhadanumberof

advantages.Wehaveseenthatmanyfactorscontributetolower

theefficiencyofSOEs.Someofthesecanbealleviatedbysellinga

majoritystaketoaprivateblockholder.Thatblockholderwillbe

motivated to maximize performance, and can be expected to

efficientlymonitorthefirm’smanagers.Thestatemaystilluseits

minoritystaketopushthefirmtomakeinefficientdecisions,for

examplehiringpoliticallyconnectedbutincompetentmanagersor

maintainingartificiallyhighemploymentlevels,butthese

prob-lemsarelikelytobelessseverethanwhenthestatehasmajorityor

fullownership.Onewouldnaturallyexpectthebenefitsofshared

ownershiptobehigher whentheother firmshareholdersown

controlling shares than when non-government ownership is

dispersed, because a controlling private owner has greater

incentivestobeefficient.Forallthesereasons,sellingamajority

orevenaminorityshareinSOEsincreasesefficiency.

Openingtheeconomytoforeigncompetitionandprivatizing

SOEs can,however, seriouslychallengea country’s sovereignty.

Imports can wreak havoc on domestic _firms that have been

shelteredfromforeigncompetition.Ifforeignfirmsareallowedto

acquireamajorityorafullstakeintheprivatizedSOEs,thenthey

will morph into subsidiaries of foreign MNEs. While national

governmentscanexertinfluenceondomesticfirms,especiallyif

theyarefullyorpartiallygovernment-owned,thisismoredifficult

withsubsidiariesofforeignMNEs,particularlyiftheirparentsare

basedinpowerfulcountries,suchastheUnitedStates.Yetstrong

domestic firms are useful to a state: they control strategic

resourcesandcanbereliedontoengageinactivitiesthatfurther

its interests, both in theshort and long runs. In the long run,

domesticfirms,especiallyiftheyhavesomelevelofgovernment

ownership,canundertakestrategicallyimportantactivities,such

as developing newproducts,both civilianand military.1 _{In the}

shortterm,theycanperformpoliticallyusefultasks(forinstance

creatingemploymentinpoliticallyimportantlocations)andcanbe

leanedontorestrainfrompoliticallydeleteriousactions(suchas

delocalizingproductiontoforeignsites).2

Hencegovernmentsinemergingmarketswereinthebeginning

ofthe1990sfacedwithadilemma:eitherkeeptheoldISIpolicies,

continuetoshutoffthecountryfrominternationalcompetition,

andacceptlowratesofeconomicgrowth,oropentheireconomies

andselltheirSOEstoforeignMNEs,butthenriskalossinnational

sovereignty.Thereis, however,especiallyfor largercountries, a

thirdalternative.Thatalternativeistoaccepttheopeningofthe

economyandaninjectionofprivatecapitalandinitiativeintotheir

SOEs,buttointervenetotransformdomesticfirmsintoMNEs,i.e.

intointernationalchampions.

InaneconomyopentothecompetitionofforeignMNEs,firms

that confine themselves to the domestic market can be at a

disadvantage if they operate in global industries and if the

domestic market is below minimum efficient size. In global

industries,theforeignMNEs’largersizeallowsthemtoamortize

thehighR&Dinvestmentsnecessarytodevelopnewproductsand

processes(Hennart,2007).Inairplanemanufacturing,forexample,

the internal market is generally too small to support the

developmentof state-of-the-artaircraft, hence theneed tosell

toforeignbuyersinordertofinancetheinvestmentsrequiredto

keepupwithcompetitors.OziresSilva,PresidentofEmbraer,putit

succinctly: _“Today there is nooption, either thecompany goes

internationaloritisdoomedtofailure”(deCarvalho,2005).MNEs

can also arbitrage between locations. They can, for example,

pursuepredatorypoliciesbycross-subsidizingbetweencountries.

The only real defense of domestic firms is then to expand

internationally so as to reach a size that makes it possible to

undertaketheR&Dinvestmentnecessarytoreachtechnological

parityandtobeabletolocateproductionfacilitiesinthecountries

1_{Amsden(2012)arguesthatforeignaffiliatesconductlittleR&Dinemerging} markets.

2

TheFrenchpresidentSarkozywasabletoleanonminoritystate-ownedRenault in2009nottoshiftcarassemblytoEasternEurope(Clift,2012)whiletheBelgian government hadearlier beenpowerless in preventingthe closureof Renault’s Vilvoordeassemblyplant.

oftheircompetitorssoastobeinapositiontoretaliateagainst

their predatory moves—the well-known exchange of hostages

strategy(Hennart&Park,1994;Yu&Ito,1988).

What does it take to develop internationally competitive

domesticfirmsifthecountryhasbeenfollowingISIpolicies?Aswe

have argued, ISI encouraged firms togrow in breadth but not

depth.Industries werepopulatedbythedivisionsofdiversified

nationalfirms,eachofthemtoosmallandtooweaktoeffectively

competewithforeignMNEs.Tobeabletodoso,thesedivisionshad

tomergeintolargerentities,and theseentitieshad toupgrade

theirproductsandprocesses.Throughthisprocessof

concentra-tion,domestic_firmscanincreasetheirbargainingpowerbecause

theycanmonopolizecomplementarylocalassetswhich foreign

MNEsneedtoaccessthemarket.Oncelarger,domesticfirmscan

trademarketaccessforforeigntechnologyordriveahardbargain

andusetheproceedstofinancetheirforeignexpansion(Hennart,

2012). Governments can help by acting on their behalf as a

monopsonist and force foreign MNEs to share technology in

exchangeformarketaccess,astheChinesegovernmenthasdone

withgreatsuccess(Hemphill&White,2013).

Itisalsonecessaryforthestatetomaintainitsinfluenceon

these new “international champions”. Having international

activitiesincreasesthebargainingpowerofthenewinternational

championsvis-à-vistheirhomegovernments,becauseitincreases

thedegreesoffreedomthesefirmsenjoy.Governmentsalsoneed

to assure that their subsidized champions are not sold to

foreigners. The most direct way to assure compliance and to

avoidlosingthechampionstoforeignersistoownsufficientequity

inthem. Thedesirethereforetosafeguardnational sovereignty

whilereducingtheinefficiencyofclosedeconomiesand

wholly-owned SOEs has led to a regime that has been called

“neo-developmentalist”. Under this regime, the state has groomed

partially-owneddomesticfirmsintointernationalchampions.

Ifwe arerightthatone responsebystates toopennessand

privatizationhasbeentobuildupdomesticfirmsintointernational

champions,andtokeeporobtainanequitypositioninthem,then

we would expect that internationalized domestic firms should

havesome degreeof government ownership. Indeed, since the

higherafirm’slevelofinternationalization,thegreateritsstrength

andpotentialusefulnesstothestate,thelargerthenshouldbethe

levelofgovernmentownershipinthem.

Which countries would we expect to have chosen

neo-developmentalistpoliciesofpromotingstate-ownedinternational

champions?Thebuildingupofinternationalchampionsiseasierif

domesticfirmsenjoyalargedomesticmarket,becausethelarger

domestic market gives domestic firms and their governments

morebargainingpowervis-à-visforeignMNEs.Building

interna-tionalchampionsisalongtermendeavorthatrequiresconsistent

policies,hencecountrieswithaprofessionalandindependentcivil

serviceshouldbemoresuccessfulatit.

Brazil is the Latin American country that best meets these

requirements.It hasthe largesteconomy. It alsohasin BNDES

(BancoNacionaldeDesenvolvimentoEconômicoeSocial)alarge

andprofessionallyrundevelopmentbankthathasplayedacrucial

role in managing Brazil’s privatizationsand in directlyhelping

Brazilianfirmstointernationalize.OtherLatinAmericancountries

do not havesimilarinstitutions. Argentina’s development bank

Banadewasclosedinthe1990s.InChile,CORFOhasonlyaided

smallandmediumfirms(Finchelstein,2016:15).PoliciesinBrazil

havealsoshownmoreintertemporalcoherencethaninArgentina

(Finchelstein,2016:Table4).Consequently,webelievethatBrazil

providesthebestcontexttostudythepotentialimpactof

neo-developmentalist policies on the internationalization of

Multi-latinas.

2.1.GovernmentownershipandinternationalizationinBrazil

Under the government of Getulio Vargas (1930–1945) the

BrazilianstatefirstsetupSOEstocarryoutISIandtoremedythe

unwillingness of the private sector to make the long-lived

investmentsnecessarytodevelopnewindustries(Trebat,1983).

Vargas,inassociationwiththeUSgovernment,createdBrazil’sfirst

integratedsteelmill,CSN,andconsolidatedanumberofsmalliron

oreminesand a railroadintotheCompanhia Valedo RioDoce

(CVRD), thepredecessor of Vale. A second wave of SOEs were

createdinthe1950s,includingPetrobras.Later,Brazil’sdesireto

haveitsownmilitaryairplanemanufacturerledtothecreationof

Embraer.The1950salsosawtheestablishmentofBNDES.Itsinitial

rolewastoprovidelongtermfinancingforinfrastructureprojects,

but under the military government it shifted its focus to the

financing of privatecompanies and took equity stakesin them

through its investment subsidiary BNDESPAR (Musacchio &

Lazzarini,2014b).BNDESisoneofthelargestdevelopmentbank

intheworld.In2013itwasresponsibleforabout21%ofthetotal

credittotheBrazilianprivatesectorandforalmostallitslong-term

credit(Musacchio&Lazzarini,2014b).

The heyday of Brazilian SOEs was during the military

dictatorship(1964–1985).Bythenthree-fourthsofthe100largest

Brazilianfirms(rankedbyassets)wereSOEs(Baer,1995:80).The

poorproductivityofthesefirmscausedlosses,whichwerecovered

throughbudgetdeficitsandforeignborrowing.Excessiveforeign

indebtednessandunsustainablehighinflationledthegovernment

tomakereforms.TheCollor(1990–92)andCardoso(1995–2002)

governmentsembarkedonamajor privatizationprogramwhich

sawSOEstakesreducedfromfulltomajorityandminorityones.In

paralleltotheseprivatizations,Brazilalsoopenedupitseconomy

to foreign competition, reducing its barriers to trade and

investment.

Inspiteofappearances,theBrazilianstatedidnotabandonits

nationalisticpoliciesanditsbeliefintheneedforstateguidanceof

theeconomy. The mainideology thatguide itspoliciesdidnot

change,butwaspragmaticallyadjustedtofitnewcircumstances.

Brazilhaspursuedsincetheturnofthenewcenturypoliciesthat

Ban (2013)hascalled“liberalneo-developmentalist”. Evansand

Sewell notethat“Brazilianelitesneverfullyembracedthe

anti-statistassumptionsofAnglo-Saxon-styleneoliberalism.

Authori-tariangeneralsanddemocraticsocialistswerebothhappytouse

the state alongside markets to make sure that the Brazilian

economy continued to diversify and that Brazil’s trajectory of

growthwasnotprimarilyshapedbytheprioritiesofcapitaland

governmentsin theNorth”(Evans&Sewell, 2013:55).3 _Brazil’s

neo-developmentalistshaveacceptedthenecessityofhavingan

open economy, but havetempered that openness byan active

industrialpolicyofsupportfortheinternationalizationofdomestic

firms.

Thus while successive Brazilian governments under ISI had

beengenerallyhostiletoforeigninvestmentsbydomesticfirms,

they startedafterliberalizationtoencourage theirinternational

expansion(Araujo,2013;Ban,2013;Kroeger,2012;Rocha,2014;

Silva, 2010). In 2004, for example, Luis Furlan, minister of

development,statedthat“thegoalofthegovernmentistokeep

atleasttenBrazilianmultinationalsinoperation”(Valor

Econom-ico,2007,citedinArbix&Caseiro,2012).Governmentsupportfor

theinternationalizationofBrazilianfirmsincreasedundertheLula

presidencywiththeimplementationoftheProduction

Develop-ment Policy (PDP) plan (2008–2010). The goal of PDP was to

establish Brazilian companies as the top five global players in

industries in which Brazil had already achieved high

3

competitiveness–aeronautics,oilandgas,petrochemicals,ethanol,

mining,steel,pulpandpaper,andmeat(Caseiro&Maseiro,2014).

In2009,thepresidentofBNDES,LucianoCoutinho,declared:

“IwouldsaythatBrazilneedstohaveworldchampions...The

countryhasalreadydeveloped verycompetitivecompanies.

.. ButBrazilhas,relativetoitssizeandpotential,fewworld

class companies.It is naturalthat in theexpansion ofthese

companies,theBNDES, undermarketconditions, isready to

support these opportunities...What exists is that companies

that have proventobehighlycompetitive aresupportedby

BNDES.Itispartofthegovernment’sindustrialpolicytoallow

the developmentofglobal Brazilianplayers withworldwide

scale”(Romero,2009;citedinAlmeida&Schneider,2012).

Whatisnotableinthisstatementisthatitindicatesthatthe

policyofthegovernment(throughBNDES)istosupport,withloans

andequityinfusions,companiesthathavealreadyshownthatthey

areinternationallycompetitive, henceestablishing a directlink

betweenthelevelofinternationalizationofBrazilianfirmsandthe

likelihoodtheywillhavesomedegreeofstateownership.

We have argued that if the Brazilian state is to have

internationalchampionswhich it caninfluence,two conditions

must be met. First, Brazilian firms must be internationally

competitive. Second, theBrazilian state (throughBNDES, other

governmententities,andSOEorex-SOEpensionfunds)musthave

sufficientequityinsuchfirmstobeabletoinfluencethemandto

preventtheirsaletoforeigners.Wefirstdiscusstheactionsthatthe

governmenttooktohelpBrazilianfirmsrealizetheir

internation-alizationpotential beforerelating howtheBrazilianstatemade

sure that it could keep control of their new international

champions.

Wehaveseenthatinthehighlyprotecteddomestic

environ-mentthatprevailedunderISIpolicies,itmademoresenseforlocal

incumbentsto diversify into otherdomestic industries than to

targetforeignmarkets(Schneider,2008).Furthermore,thecushy

home environment did not prepare firms for international

competition, so,while there were notable exceptions, most of

themwereunable toprofitablysellabroad. Successful

interna-tional expansion requires specialization in a smaller range of

activities and deeper skills in these activities. The Brazilian

government,through itsmajority-owned SOEs,BNDES, andthe

pensionfundsofSOEsandex-SOEs,hasfacilitatedthisprocessby

engineeringthemergersofdomesticcompaniesinsectorswhere

thegovernmentthinksBrazilhasacomparative advantage.The

goalis toreachsufficientsizetoallow Brazilianfirmstobetter

defendthemselvesagainstforeigncompetition in thedomestic

market,andtoleveragetheirdomesticmarketpowertobuildup

their foreign operations (Hennart, 2009). BNDES, for example,

facilitatedthroughanequityinjectionthemergerintoFibriaofthe

two largest Brazilian pulp producers, Aracruz and Votorantim

CeluloseePapel,resultinginafirmthatexportsmorethan90%of

its output and is the world’s largest hardwood pulp producer

(Kroeger, 2012).4 _{Petrobras, which in 2012 was 63.6% owned}

(directlyand indirectly) bythe government,5 _{acquiredIpiranga}

withthehelpofBNDES,incorporateditspetrochemicalassetsinto

Braskem(in which it held a 30% stake), acquired Suzano and

mergeditwithGrupoUltraintoQuattor,andthenmergedthetwo

backintoBraskem, making thatfirm thelargestpetrochemical

companyin theAmericas. Lessthana month afterthemerger,

BraskemacquiredaunitoftheUScompanySunocoChemicalsfor

US$350millionandmadeaUS$2.5billiongreenfieldinvestment

in a Mexican petrochemical complex (Arbix & Caseiro, 2012).

Likewise the government supported the merger of Sadia and

PerdigãowhichledtothecreationofBrasilFoods,todayoneofthe

largest foodcompanies in the world,selling tomore than 110

countriesonfivecontinents(Vargas&deMoura, 2015).Vale,a

privatized SOE in which the Brazilian government kept a

controlling stake, sold off after its privatization its paper and

pulpsubsidiariesanditsstakesinBraziliansteelcompanies,but

consolidateditsmarketshareinironorebypurchasinganumber

ofBrazilianironoremines.ThegoalwastoconsolidateVale’shome

marketshareandtokeepBHPoutoftheBrazilianironoremarket

(Chaddad,2003).By2007Valeaccountedfor85%ofBrazil_’siron

oreproductionandwasthecountry’ssoleironoreexporter.Itthen

expanded abroad,acquiring Inco in Canadaand a coalmine in

Australia, and investing in coal and copper projects in Africa

(Khanna,Musacchio,&ReisendePinho,2010).

AsecondwayinwhichtheBraziliangovernmenthashelped

Brazilianfirmstointernationalizeisbyincreasingexportfinancing.

Between2001and2012BNDESexportfinancingincreasedfromUS

$2.6 to US$11.3 billion (Araujo, 2013). BNDES offered such a

generousexportfinancingpackagetoEmbraercustomersthatit

triggeredaCanadiancomplainttotheWorldTradeOrganization

(Amann, 2009).6 _{BNDES has offered guarantees for the large}

foreigninfrastructureprojectsundertakenbyBrazilian

construc-tion companies such as Odebrecht (Finchelstein, 2012). The

BrazilianForeignOfficehasalsohelpedBraziliancompaniesmake

foreignacquisitionsandwinforeigncontracts.Itwasinstrumental

inVale’ssuccessfulbidtodeveloptheBayovarphosphatedeposits

in Peru. Two years later, Lula’s personal intervention helped

GerdauacquiretheprivatizedSiderperu(Luce,2008).

Safeguardingnationalsovereigntythroughhome-grown

inter-nationalchampionsrequiresthatthegovernmentbeabletoenlist

theeconomicpowerofsuchchampionstocarryoutitspolicies.

Hencethegovernmentmustbeabletoinfluencefirmbehavior.The

most direct way to do that is to own sufficient equity in the

champion.HowdidtheBrazilianstateachievethis?First,itmade

surethatitkeptasufficientownershipshareinthoseprivatized

SOEs which had good prospects for becoming international

champions.Second,ittookequitystakesinprivatecompaniesit

thoughtcouldbecomeinternationalchampions.

TheBraziliangovernmentmadesureitkeptcontrolofitsSOE

with international potential. BNDES, which was in charge of

privatizationstookthroughBNDESPAR directminoritystakesin

theprivatizedfirmsandhelpedformconsortiawhichbidforblocks

ofshares.IntheseconsortiaitallieditselfwithprivateBrazilian

firms,foreigninstitutionalinvestors,andSOEandex-SOEpension

fundssoastokeepcontroloutofthehandsofforeignMNEs(Inoue,

Lazzarini,&Musacchio,2013).

PensionfundsofSOEsandex-SOEsarenaturalalliesofBNDES.

TheyaremajorBrazilianinvestors.Previ,forinstance,thepension

fundoftheBancodoBrasil,agovernmentownedbank,wasin2015

thelargestLatin-Americanpension fundwithassets ofR$148.4

billion(US$38.3billion).Petros,thepension fundforPetrobras

employees, is the second largest Brazilian pension fund, with

assets ofR$68.1 billion(US$17.6 billion).Thereis considerable

evidence that theBrazilian government hasbeenable toexert

influence onthesepension funds(Costa,2012).Musacchio and

Lazzarini (2014b: 37) write that “because the government has

voice in the management of state-controlled SOEs and these

companieshaveavoiceinthemanagementoftheirpensionfunds,

governments in Brazil...have been able to strategically use

pensionfundsintheirfavor.”Hence,tomeasurethelevelofcontrol

theBraziliangovernmenthasoverfirms,itmakessensetoaddto

4

BNDESnowowns34.9%ofthecompany(Rocha,2014). 5

50.85%bytheBrazilianFederalgovernment,8.13%byBNDES,and4.62%byFundo FiscaldeInvestimentoseEstabilização,aninvestmentfundcontrolledbytheBrazilian

Federalgovernment. 6

thestakes takenby BNDESPAR those taken by SOEs and other

governmentagencies(forexampletheBrazilianstates)andbythe

pensionfundsofpresentandformerSOEs.

Through strategic allianceswith private domestic investors,

stateentities,andthepensionfundsofSOEsorprivatizedSOEs,

BNDESwas successfulin preventingforeignMNEs frombuying

controlling shares in the privatized SOEs. In the first wave of

privatizationsinvolvingindustrialandminingfirms(1991–1997),

foreignersonlygota18%stake(Goldstein,1999:691),comparedto

67%in Argentina (Finchelstein,2016).Take thecase of Vale.In

order toprevent its acquisition by RioTinto, BHP, or Japanese

tradingcompanies,BNDESarrangedtohavethesharessoldtoa

consortiumofBrazilianinstitutions(Chaddad,2003).Ittooka6.9%

directstakeinValeandan11.5%shareinValepar,whichinturn

owns53.9%ofVale.ValeparisaconsortiumofMitsui,twoBrazilian

investors(EletronandBradespar,asubsidiaryoftheBrazilianbank

Bradesco),andofLitel.Litelis,inturn,aconsortiumofPrevi,Petros,

Funcef–thepensionfundoftheCaixaEconomicaFederal(abank

SOE)–andFuncesp,whichmanagesthepensionsoftenBrazilian

utilities, all present and former SOEs (Musacchio & Lazzarini,

2014a). Because BNDES can obtain, when needed, cooperation

fromthepensionfunds,theBrazilianstatehasthroughitsdirect

and indirect stakes effective control over Vale.7 _{Similarly, the}

governmentsolda45%stakeinEmbraerto(a)aconsortiumofUS

investors assembled by Wasserstein Perella, (b) the Bozano

SimonsenBrazilianconglomerate,and(c)PreviandSistel–Sistel

isthepensionfundsof Telebras,atelecommunicationSOE.The

government got a golden share that carries veto power over

changesincontroland corporatepurposeandoverthecreation

andalterationofdefenseprograms(Goldstein,2002).

Second,BNDEShasalsotakenequitysharesinBrazilianfirms

withhighinternationalizationpotential.Between2005and2011,it

injectedUS$4.7billioninequitytofinancetheforeignacquisitions

of Brazilian companies in textiles, IT, pharmaceuticals, and

machinery(Caseiro&Masiero,2014).In2005,forexample,BNDES

tooka stakeinCoteminas tohelp itacquiretheUS textilefirm

Springs Industries (Finchelstein, 2012). Coteminas is now the

world’s largestproducerof hometextiles(daSilva,daRocha, &

Carneiro,2009).Thebulkoftheseequityinvestments,however,

has benefitted Brazilian meat processing companies. BNDES

investedUS$4billiontohelpfinanceJBS’ssuccessiveacquisitions

ofSwiftArmourinArgentina,ofUS-basedcompaniesSwiftandCo,

National Beef, Smithfield Beef, and Pilgrim’s Pride, and of the

TasmanGroupinAustralia.ThisbuyingspreemadeJBStheworld’s

largestbeefprocessorandBNDESitslargestminorityshareholder

witha30percentstake(Musacchio&Lazzarini,2014a).

Let’ssummarizeourargumentsofar.Wehaveshownthatthe

Brazilian government, while it has opened up its economy to

foreigncompetitionand reduced itsstakesin itsSOEs,hasnot

abandonedits policiesof directinterventionin theeconomyin

defense of its sovereignty. These policies, which were

inward-lookingand protectionistunderISI,arenowdirectedatmaking

international champions out of Brazilian firms which have

demonstratedapotentialforsuccessfulinternationalexpansion.

To guarantee that the international champions which have

benefittedfromstateaidremainBrazilianand collaboratewith

thegovernment,thestatehaskept–ortaken–equitypositionin

themthroughBNDES,othergovernmententities,andthepension

funds of SOEs or ex-SOEs. As a result, highly internationalized

Brazilianfirmshaveasignificantlevelofgovernmentownership.

Indeeditisreasonabletobelievethatthehighertheprospectofa

firm becoming aninternational champion,thehigher thestake

thatthestateislikelytotakeinthem.Thisleadsustoourfirst

hypothesis:

Hypothesis 1. The higher the total equity held by Brazilian

government institutions (BNDES,SOE pensionfunds, and other

governmententities)inBrazilianfirms,thehighertheirdegreeof

internationalization.

There is evidence that majority-ownedor fully-owned SOEs

performsignificantlyworsethanfullyprivatefirms(Megginson&

Netter,2001).WehaveseenthatSOEefficiencycanbeimprovedby

having thestatesella majoritystakeintheirSOEstoa private

blockholder. That blockholder will be motivated to maximize

performance,andcanbeexpectedtoefficientlymonitorthefirm’s

managers. Hence firms with minority state ownership should

performbetterthan SOEsmajorityorfully-ownedbythestate.

Inoueetal.(2013)findthatBNDESequityhadapositiveinfluence

ontheprofitabilityofBrazilianfirmsifitsstakewashigherthan

10%.

Onewouldexpectthenegativeconsequencesofstate

owner-shiptobelowerwhentheotherfirmshareholdersowncontrolling

shares than when non-government ownership is dispersed. A

controllingprivateownerhasincentives tobeefficient andthe

powertofightoffthenegativetendenciesofstateownership.A

striking feature of the governance of Brazilian firms is the

important role played by families (Aldrighi & Neto, 2007).

Family-ownedfirmsmakeupalmosthalfofoursampleandare

overwhelminglyfamily-managed.8 Whatistheexpectedimpact

oninternationalizationwhenfamiliesarethedominantownersof

firmspartiallyownedbythegovernment?Whenthegovernment

hasaminorityshare,ithastoworkwiththedominantowner,in

ourcasefamilies.Iffamiliesandthegovernmentdonotsharethe

same internationalizationgoals, thenhavingfamilies in control

mayleadtoalowerlevelofinternationalizationthanwhenthe

governmenthasunambiguouscontrol.Iftheydosharethesame

goals,ontheotherhand,thenhavingfamiliesinchargemayresult

in a higher level of internationalization, because families have

greaterincentivestoperformthancivilservants.

There is evidence that the Brazilian business community –

dominatedbyfamilies,aswehaveseen–sharesthegovernment

viewstowardsinternationalizationandtheneedforgovernment

support to achieve that goal. Ougaard (2014) notes that the

positionpaperspublishedbyBrazil’smainemployerassociations–

atthestateleveltheFederecãodas Industriasdo EstatodeSão

Paulo(FIESP)andatthenationalonetheConfederacãoNacionalde

Indústria (CNI)—call for “continued expansion of the already

competitiveexportindustriesthroughbettermarketaccessabroad

[and] continued growth of outward FDI by Brazilian MNEs”

(Ougaard,2014:9).Ougaard(2014)alsopointsout totheclose

linksbetweenthebusinesscommunityandthestate.Forinstance,

business representatives sit in the 19 competitiveness councils

responsible for implementing the latest development plan, the

PlanoBrasilMaior(2011-).

We would expect therefore family _firms in which the

government hastaken a minority share to be more successful

oninternational marketsthanfirms withdispersed ownership.

7

ThiswasshownwhenPresidentLulagotintoaconflictwithVale’sCEO.BNDES gotthesupportofPrevi,Petros,FuncefandFuncepsandsuccessfullyoustedtheCEO overtheobjectionoftheothershareholders.

8

Welookedatallthefirmsinoursampleforwhichthepercentageofvoting sharesheldbyfamilieswasgreaterthan10%andcollecteddataonthecomposition oftheirmanagementteamandBoardofDirectorsatend2014usingtheFormulários de Referência available at the Brazilian Securities and Exchange Commission (Comissão de Valores Mobiliários—CVM) and the Sao Paulo Stock Exchange BM&FBOVESPAwebsite.Wefoundthat114outof133firms(86%)hadatleast onefamilymemberontheBoardofDirectors,and104outof133(78%)atleastone familymemberasCEO,VicePresident,Director,orPresidentorVicePresidentofthe BoardofDirectors.

Familyfirmsmaybebetterrun.Forfamilyfirmsatearlystagesof

internationalization,aninjectionofBNDESequityisveryvaluable

becauseinBrazil,asBenjaminSteinbruch,thePresidentofFIESP,

notes“nosingleprivatebankwantstoprovidemiddleandlong

term financing, especially for investments” (Noticias Politica,

2014).Having thegovernmentasa minoritypartneralsohelps

family firms obtain subsidized loans, export financing, and

diplomatic assistance to help line up foreign contracts and to

facilitateforeignmarket entry.And indeed theexample ofJBS,

Coteminas, and Fibriashows that some of the mostsuccessful

Brazilianinternationalizersarefamilyfirms.Thisleadsustoour

secondhypothesis:

Hypothesis2.Firmsinwhichthegovernmentsharescontrolwith

familieswillhaveahigherdegreeofinternationalizationthanfirms

withdispersedownership.

In Brazil, foreigners make up a second important group of

owners(Aldrighi&Neto,2007).Inoursampletheyweredominant

shareholdersin8.3%ofallobservations.Bhaumik,Driffield,andPal

(2010) argue that foreigners usually have better knowledge of

internationalopportunitiesandbettergovernancepracticesthan

domesticfirms,andhencethatfirmsinwhichtheyhaveastake

shouldbemoreinternationalized.Theyfindthatthisisthecasein

Indianfamilyfirms.Similarly,onecouldarguethatforeignfirms

should be able to leverage government equity into superior

internationalperformance.Therearereasons,however,todoubt

thisisthecaseinoursample.InBrazil,manyofthefirmswhere

government and foreigners share control are in banking and

utilities(waterandenergy),sectorswhichwereexcludedfromour

sample.Wehavealsoexcluded firmswhicharemorethan50%

ownedbyforeigners.Inmanyofthefirmsinoursamplewhere

foreignersarethedominantshareholders,theforeignshareisheld

byprivateequityandinvestmentfunds.Forinstance,ownershipof

America Latina Logistica, a railroad company, is shared by the

Brazilian government (through BNDESPAR, and the Previ and

Funcef SOE pension funds) and foreign private equity and

investmentfundsthattogetherholdmorethan20%ofthe_firm’s

voting shares. There are reasons to believe that these foreign

ownersareprimarilyinterestedintheBrazilianmarketandnotin

internationalexpansion.Weconcludethereforethatwhilefirmsin

whichthegovernmentsharescontrolwithforeignerswillbemore

internationalizedthanfirmswithdispersedownership,theywill

be less internationalized than firms in which the government

sharescontrolwithfamilies.

Hypothesis3.BrazilianfirmsinwhichtheBraziliangovernment

shares control with foreign investors will have a degree of

internationalization that is higher than firms with dispersed

ownership,butlowerthanfirmsinwhichthegovernmentshares

controlwithfamilies.

Table1

Ownershipstructurecategories. Governmentisdominantshareholder

GOVERNMENT_D_10%(t-1) =1ifGOVERNMENT_%10%andFAM_%<10%andFOR_%<10%

BNDES_D_10%(t-1) =1ifBNDES_%10%andPFUNDS_%<10%andOTHERGOV_%<10%andFAM_%<10%andFOR_%<10% PFUNDS_D_10%(t-1) =1ifBNDES_%<10%andPFUNDS_%10%andOTHERGOV_%<10%andFAM_%<10%andFOR_%<10% OTHGOV_D_10%(t-1) =1ifBNDES_%<10%andPFUNDS_%<10%andOTHERGOV_%10%andFAM_%<10%andFOR_%<10% BNDES_PFUNDS_D_10%(t-1) =1ifBNDES_%10%andPFUNDS_%10%andOTHERGOV_%<10%andFAM_%<10%andFOR_%<10% BNDES_OTHGOV_D_10%(t-1) =1ifBNDES_%10%andPFUNDS_%<10%andOTHERGOV_%10%andFAM_%<10%andFOR_%<10% PFUNDS_OTHGOV_D_10%(t-1) =1ifBNDES_%<10%andPFUNDS_%10%andOTHERGOV_%10%andFAM_%<10%andFOR_%<10% BNDES_PFUNDS_OTHGOV_D_10%(t-1) =1ifBNDES_%10%andPFUNDS_%10%andOTHERGOV_%10%andFAM_%<10%andFOR_%<10% FamiliesandForeignersaredominantshareholders

FAMILIES_D_10%(t-1) =1ifGOVERNMENT_%<10%andFAM_%10%andFOR_%<10% FOREIGNERS_D_10%(t-1) =1ifGOVERNMENT_%<10%andFAM_%<10%andFOR_%10% FAMILIES_FOREIGNERS_D_10%(t-1) =1ifGOVERNMENT_%<10%andFAM_%10%andFOR_%10% GovernmentandFamiliesaredominantshareholders

GOVERNMENT_FAMILIES_D_10%(t-1) =1ifGOVERNMENT_%10%andFAM_%10%andFOR_%<10%

BNDES_FAM_D_10%(t-1) =1ifBNDES_%10%andPFUNDS_%<10%andOTHERGOV_%<10%andFAM_%10%andFOR_%<10% PFUNDS_FAM_D_10%(t-1) =1ifBNDES_%<10%andPFUNDS_%10%andOTHERGOV_%<10%andFAM_%10%andFOR_%<10% OTHGOV_FAM_D_10%(t-1) =1ifBNDES_%<10%andPFUNDS_%<10%andOTHERGOV_%10%andFAM_%10%andFOR_%<10% BNDES_PFUNS_FAM_D_10%(t-1) =1ifBNDES_%10%andPFUNDS_%10%andOTHERGOV_%<10%andFAM_%10%andFOR_%<10% BNDES_OTHGOV_FAM_D_10%(t-1) =1ifBNDES_%10%andPFUNDS_%<10%andOTHERGOV_%10%andFAM_%10%andFOR_%<10% PFUNDS_OTHGOV_FAM_D_10%(t-1) =1ifBNDES_%<10%andPFUNDS_%10%andOTHERGOV_%10%andFAM_%10%andFOR_%<10% BNDES_PFUNS_OTHGOV_FAM_D_10%(t-1) =1ifBNDES_%10%andPFUNDS_%10%andOTHERGOV_%10%andFAM_%10%andFOR_%<10% GovernmentandForeignersaredominantshareholders

GOVERNMENT_FOREIGNERS_D_10%(t-1) =1ifGOVERNMENT_%10%andFAM_%<10%andFOR_%10%

BNDES_FOR_D_10%(t-1) =1ifBNDES_%10%andPFUNDS_%<10%andOTHERGOV_%<10%andFAM_%<10%andFOR_%10% PFUNDS_FOR_D_10%(t-1) =1ifBNDES_%<10%andPFUNDS_%10%andOTHERGOV_%<10%andFAM_%<10%andFOR_%10% OTHGOV_FOR_D_10%(t-1) =1ifBNDES_%<10%andPFUNDS_%<10%andOTHERGOV_%10%andFAM_%<10%andFOR_%10% BNDES_PFUNS_FOR_D(t-1) =1ifBNDES_%10%andPFUNDS_%10%andOTHERGOV_%<10%andFAM_%<10%andFOR_%10% BNDES_OTHGOV_FOR_D(t-1) =1ifBNDES_%10%andPFUNDS_%<10%andOTHERGOV_%10%andFAM_%<10%andFOR_%10% PFUNDS_OTHGOV_FOR_D(t-1) =1ifBNDES_%<10%andPFUNDS_%10%andOTHERGOV_%10%andFAM_%<10%andFOR_%10% BNDES_PFUNS_OTHGOV_FOR_D(t-1) =1ifBNDES_%10%andPFUNDS_%10%andOTHERGOV_%10%andFAM_%<10%andFOR_%10% Government,FamiliesandForeignersaredominantshareholders

GOVERNMENT_FAMILIES_FOREIGNERS_D_10%(t-1) =1ifGOVERNMENT_%10%andFAM_%10%andFOR_%10%

BNDES_FAM_FOR_D_10%(t-1) =1ifBNDES_%10%andPFUNDS_%<10%andOTHERGOV_%<10%andFAM_%10%andFOR_%10% PFUNDS_FAM_FOR_D_10%(t-1) =1ifBNDES_%<10%andPFUNDS_%10%andOTHERGOV_%<10%andFAM_%10%andFOR_%10% OTHGOV_FAM_FOR_D_10%(t-1) =1ifBNDES_%<10%andPFUNDS_%<10%andOTHERGOV_%10%andFAM_%10%andFOR_%10% BNDES_PFUNDS_FAM_FOR_D_10%(t-1) =1ifBNDES_%10%andPFUNDS_%10%andOTHERGOV_%<10%andFAM_%10%andFOR_%10% BNDES_OTHGOV_FAM_FOR_D_10%(t-1) =1ifBNDES_%10%andPFUNDS_%<10%andOTHERGOV_%10%andFAM_%10%andFOR_%10% PFUNDS_OTHGOV_FAM_FOR_D_10%(t-1) =1ifBNDES_%<10%andPFUNDS_%10%andOTHERGOV_%10%andFAM_%10%andFOR_%10% BNDES_PFUNDS_OTHGOV_FAM_FOR_D_10%(t-1) =1ifBNDES_%10%andPFUNDS_%10%andOTHERGOV_%10%andFAM_%10%andFOR_%10%

Thecasewherethegovernmentsharescontrolwithfamiliesand

foreigners is an intermediate case between that in which the

government shares control with families and that in which it

sharescontrolwithforeigners.Becausetheequityshareoffamilies

andofthegovernmentislower,wewouldexpectthesefirmsto

have a higher degree of internationalization than firms with

dispersedownership,butaloweronethanfirmsinwhichcontrolis

sharedbetweenfamiliesand thegovernment. Henceourfourth

andlasthypothesis:

Hypothesis4.BrazilianfirmsinwhichtheBraziliangovernment

sharescontrolwithfamilyfirmsandforeigninvestorswillhavea

degree of internationalization that is higher than firms with

dispersed ownershipandfirmsinwhich the governmentshares

controlwithforeignersbutlowerthanfirmsinwhichcontrolis

sharedbetweenfamiliesandgovernment.

3.Dataandmethods

3.1.Sampleanddata

WeselectedallBrazilian firmslisted ontheSãoPauloStock

Exchange(BM&FBovespa)between2002and2011thatreported

dataontheirforeignsales(zeroorpositive)andtheirownership

structure.Weexcludedfinancialfirms,publicutilities(energyand

water),andfirmsmorethan50%ownedbyforeignersinaspecific

year.Weremovedfirmsinwhichyearlyvaluesforreturnonassets

andleveragewerethreestandarddeviationsaboveandbelowthe

mean.Becauseofmissinginformationonforeignsales,especially

inearlyyears,andasaresultofmergers,acquisitions,bankruptcies

anddelisting,weendedupwithanunbalancedpanelsampleof

956firm-yearobservations.

FinancialdatawerecollectedfromtheEconomatica1database.

Informationonforeignsales(bothexportsandlocalsalesofforeign

subsidiaries)wasobtainedfromBloombergandThomsonOneand

wassupplementedbyconsultingtheannualDFP—Demonstrações

Financeiras Padronizadas and Formulários de Referência. Data on

ownership structure were obtained from the Formulários de

ReferênciaandtheInformativos AnuaisavailableattheComissão

deValoresMobiliários.

Ourdependentvariableisafirm’sratioofforeignsalestototal

netsales(FSTS).Foreignsalesconsistofexportsandthelocalsales

offoreignsubsidiaries.FSTSisthemostcommonlyusedindicator

ofinternationalization(Sullivan,1994).Oesterle,Richta,andFish

(2013) found that using FSTS rather than multidimensional

variablesofinternationalizationdidnotchangetheirresults.

Ourmainindependentvariableisthetotalgovernmentstake

taken ina firm (Total Government_ %), which isthesumof the

votingsharesheldby(1)BNDESPAR(BNDES_%);(2)thepension

fundsofBrazilianSOEsandformerSOEs(PensionFunds_%);(3)and

governmentagenciesandgovernmentwholly-ownedbanksand

enterprises(OtherGovernment_%).Wealsoinvestigatetheseparate

influence onforeign sales of these three types of government

stakes.ThelistofallthesegovernmentinvestorsisinAppendixA.

We alsoenter a firm’sfamily stake (Families_%) and foreign

stake(Foreigners_%).Families_%isthepercentageofafirm’svoting

shares owned by families and FOR_% that owned by foreign

investors.Becauseof thepyramidalstructurecommoninBrazil

(Aldrighi&Neto,2007), weconsideredboth directandindirect

stakes(i.e.stakesheldthroughotherfirmsandfunds).

We also created dummy variables torepresent instances of

sharedstockownership—seeTable1.AsinBhaumiketal.(2010),

weassumedthatowningmorethan10%ofthetotalvotingshares

gave some degree of control. In fact, Brazilian law allows

shareholderswho ownmore than10percentofa firm’svoting

sharestoaskforaseatontheboardofdirectors.Familiesarethe

T able 2 Descript iv e statistics for the v ariables and corr elation coef fi cients. Variables Mean Median S td. Dev . Max. Min. 1234 5678 9 1 0 11 1 2 1 3 1 4 1 5 1 F S T S (t) 1 4.23 1 .23 2 1 .43 9 7 .80 0.0 0 1 2 F S T S (t-1) 1 3.9 7 0.0 0 2 1 .6 7 98.39 0.0 0 0.96*** 1 3 F S T S (First) 1 3.9 1 0.0 0 2 1 .59 98.39 0.0 0 0.9*** 0.93*** 1 4 BNDES_% (t-1) 1 .23 0.0 0 4.59 3 5.77 0.0 0 0.3 5*** 0.36*** 0.3 1*** 1 5 Pension F unds_% (t-1) 3.25 0.0 0 1 0.40 79.75 0.0 0 0.23*** 0.23*** 0.25*** 0.2*** 1 6 Other Go v ernment_% (t-1) 2.7 2 0.0 0 1 3.28 1 0 0.0 0 0.0 0 0.02 0.03 0.05 0.0 1 0.03 1 7 T o tal Gov ernment_% (t-1) 7 .20 0.0 0 1 8.32 1 0 0.0 0 0.0 0 0.2*** 0.2*** 0. 1 8*** 0.3 7*** 0.64*** 0.3 5*** 1 8 Families_% (t-1) 39.05 38.65 32.29 1 0 0.0 0 0.0 0 0.04 0.05 0.02 0. 1 5*** 0.25*** 0.23*** 0.35 *** 1 9 F or eigners_% (t-1) 7 .32 0.0 0 1 2.55 49.7 4 0.0 0 0.03 0.03 0.03 0 0.04 0. 1*** 0.09*** 0.2 7*** 1 1 0 Ownership Concentra tion (t-1) 6 1 .63 62.54 25.0 0 1 0 0.0 0 5.4 7 0.04 0.04 0 0.04 0.06 0. 1 4*** 0. 1 3*** 0.45*** 0.2 4*** 1 11 Firm Size (t-1) 1 3.80 1 3.7 2 1 .75 20.0 7 8.2 7 0.43*** 0.43*** 0.36*** 0.26*** 0 0.0 7** 0. 1 2*** 0.06* 0. 1 3*** 0.08** 1 1 2 Lev er age (t-1) 26.39 24 .83 1 8.7 2 1 35 .3 5 0.0 0 0.25*** 0.2 7*** 0.26*** 0. 1 3*** 0.03 0.09*** 0.02 0. 1*** 0. 1*** 0. 11*** 0.28*** 1 1 3 Cash (t-1) 1 4 .1 6 1 0.85 1 3.20 90.02 0.0 0 0.03 0.02 0.03 0 0 0.08** 0.06* 0.06* 0. 1 7*** 0.2*** 0. 1 7*** 0. 1 4*** 1 1 4 R O A (t-1) 3.25 4.36 1 2.3 1 7 4.9 1 9 7 .69 0.06* 0.06* 0.06* 0.02 0.0 7** 0.08** 0.09*** 0.05 0.0 7** 0.08** 0.2 4*** 0.2 7*** 0.2*** 1 1 5 Firm Ag e (t-1) 5.82 6.2 1 1 .1 1 7 .2 7 0.0 0 0.2 1*** 0.22*** 0.23*** 0.05 0. 1*** 0.0 1 0.08** 0. 1*** 0.3*** 0. 1 2*** 0.04 0. 1 2*** 0. 1 8*** 0.06 1

dominantshareholdersinalmosthalfoftheobservationsinour

sample.Firmsinwhichfamiliessharecontrolwithforeignersmake

upthesecondlargestcategory(12.7%oftheobservationsinour

sample), followed by firms with dispersed ownership (10.9%),

firmscontrolledbyforeigners(8.3%),andfirmscontrolledbythe

government(7.3%).FirmswiththehighestFSTSarethoseinwhich

thefollowing shareholdersaredominant: (1) BNDESand other

government investors (FSTS=83.5%); (2) BNDES, other

govern-ment investors, and foreign investors (FSTS=80.4%); (3) SOE

pension funds and foreign investors (FSTS=70.4%; (4) BNDES,

families,andforeigninvestors(FSTS=45.2%);(5)BNDESandSOE

pension funds (FSTS=42.5%); and (6) BNDES and families

(FSTS=37.8%).

The theoryof internationalproduction (Dunning & Lundan,

2008;Hennart,1982)tellsusthatoneofthemotivesforselling

abroadistheexploitationoffirm-specificadvantages.Tocontrol

forthis,weusefirmfixedeffects.Wealsoenteredthefollowing

controlvariables:

(1)Ownership Concentration (Ownership Concentration),

mea-suredbythepercentageofshareswithvotingrightsheldbythe

largestshareholders.Wetookintoaccountformalshareholder

agreements,usinginformationfromthefirm’sAnnualReports.

(2)Cash(Cash),measuredbytheratioofcashandequivalentsto

thebookvalueoftotalassets.

(3)Firmsize(FirmSize),measuredbythenaturallogarithmofthe

bookvalueoftotalassets.

(4) FirmAge(FirmAge),measuredbythenaturallogarithmofthe

numberofmonthssincethedateofincorporationreportedin

theCVMregistrationforms.

(5) Leverage(Leverage),measuredbythebookvalueofgrossdebt

overtotalassets.

(6) ReturnonAssets(ROA),measuredbynetincomedividedbythe

bookvalueoftotalassets.

(7) Initialdegreeofinternationalization(FSTSFirst).Toaccountfor

thepossibilitythattheBraziliangovernmentchoosestoinvest

infirmsthataremoreinternationalized,weenteredthefirm’s

firstvalidFSTSobservation(FSTSFirst).Asanotherwaytocontrol

for endogeneity, we also entered the previous year_’s FSTS

(FSTSt-1)

Table2presentssomedescriptivestatisticsforthevariables.

3.2.Estimationapproachandeconometricmodels

GovernmentagenciessuchasBNDESandSOEpensionfundsdo

notrandomlyselectthefirmsinwhichtheyinvest.Henceusing

OLS toassess the impact of governmentequity investment on

internationalization may suffer from endogeneity bias because

unobservable factors may affect both the likelihood of state

ownershipinafirmanditslevelofinternationalization,makingit

difficulttoestablishacausallink(Inoueetal.,2013).Tohandlethis

problemwefollowtheseauthorsanduseapanelregressionwith

time-invariant firm-specific effects and year-fixed effects (the

Table3

Impactofgovernmentequitystakeoninternationalization.

TobitModel—DependentVariable=FSTS(t)—MeanoftheMarginalEffectontheExpectedValueoftheTruncatedOutcome

VARIABLES Model1 Model2 Model3 Model4 Model5 Model6

TotalGovernment_%(t-1) 0.1112** 0.09084*** 0.06666** (1.99) (2.83) (2.56) BNDES_%(t-1) 0.29962*** 0.29798** 0.2085** (2.74) (2.37) (1.99) PensionFunds_%(t-1) 0.1174** _0.07286*** _0.06467*** (2.1) (4) (3.89) OtherGovernment_%(t-1) 0.00212 0.01673 0.02895 ( 0.05) ( 0.54) ( 1.36) Families_%(t-1) 0.04381 0.06952*** 0.03716** 0.02736 0.05236*** 0.02745*** ( 1.25) ( 2.7) ( 2.52) ( 0.86) ( 4.71) ( 3.42) Foreigners_%(t-1) 0.06433 0.09958 0.08528 0.04287 0.07466 0.06644 (1.16) (1.34) (1.18) (0.69) (0.89) (0.83) OwnershipConcentration(t-1) 0.01744 0.00013 0.00073 0.01233 0.01215 0.00832 ( 0.55) (0.01) (0.06) ( 0.36) (0.42) (0.46) Cash(t-1) 0.03991 0.05539 0.03445 0.03545 0.04778 0.0294 ( 1.13) ( 1.42) ( 1.39) ( 0.95) ( 1.16) ( 1.09) FirmSize(t-1) 1.40478 0.88925 0.63769 1.42609 0.90555 0.69705 (1.28) (0.83) (0.73) (1.24) (0.74) (0.71) FirmAge(t-1) 15.59408*** _14.32505*** _11.04259*** _{13.99337** 12.8461}** _9.90434* (3.35) (3.24) (3.04) (2.39) (2.07) (1.93) Leverage(t-1) 0.03266*** 0.01888** 0.01057 0.02819*** 0.01175 0.00667 ( 2.64) ( 2.53) ( 1.33) ( 2.59) ( 1.04) ( 0.76) ROA(t-1) 0.03589** 0.02785 0.00407 0.03018** 0.01607 0.00233 ( 2.13) ( 0.88) ( 0.25) ( 2.21) ( 0.6) (0.19) FSTS(First) 0.8122*** 0.73759*** (6.8) (3.75) FSTS(t-1) 0.18604*** 0.1787*** (9.4) (8.57)

Constant Yes Yes Yes Yes Yes Yes

YearFixedEffects Yes Yes Yes Yes Yes Yes

FirmFixedEffects Yes Yes Yes Yes Yes Yes

ClusteredStandardErrorsoneachIndustry Yes Yes Yes Yes Yes Yes

NumberofObservations 956 878 874 956 878 874 NumberofFirms 173 173 173 173 173 173 Logpseudolikelihood 1721.549 1495.85400 1444.14480 1713.92930 1487.71370 1438.98440 PseudoR2 0.3948 0.40940 0.42560 0.39750 0.41260 0.42770 * p<0.10. ** p<0.05. *** p<0.01.

firmsinoursampleexperiencesignificantyear-to-yearchangesin

governmentownership over our period of analysis). Because a

firm’slevelofinternationalizationislikelytobeindustry-specific

(Hennart & Park, 1994), we clustered standard errors at the

industrylevel,makingsurethattherewereatleastthreefirmsin

each industry. Because our dependent variable FSTS varies

betweenzeroandone,andbecausenearlyhalfoftheobservations

in our sample (46.7%) have zero values, we use a Tobit panel

truncatedat zero.Following Bhaumik et al. (2010), we lag our

independentandcontrolvariablesoneyear.Table2 showsthat

correlationcoefficientsbetweenthevariablesarebelowlevelsat

whichmulticollinearitywouldbeaproblem.

4.Results

In Table 3 we evaluate the influence of governmentequity

stakesoninternationalization.Thecoefficientsshowthemarginal

effectofthevariableontheexpectedvalueofFSTS.Lookingfirstat

thecontrolvariables,FirmAgeispositivelyrelatedtothedegreeof

internationalization.ThisisconsistentwiththefindingsofFloriani

andFleury(2012)forBrazilianfirms,andmoregenerallywiththe

Uppsalamodelofgradualinternationalization(Johanson&Vahlne,

1977).Thestaketakenbyfamilies,Families_%(t-1),tendstoreduce

the extent of internationalization when we control for past

internationalization (inmodels2,3, 5 and6), whilethat taken

byforeigninvestors,Foreigners_%(t-1)hasnosignificantimpacton

afirm’sFSTS.Turningnowtoourhypotheses,Models1,2and3

showapositiverelationshipbetweenafirm’sFSTSandthetotal

equitystaketakenbytheBrazilianstatethroughthedevelopment

bankBNDES,governmentagencies,SOEs,andSOEpensionfunds,

asthecoefficientsofTotalGovernment_%(t-1)areallpositiveand

statisticallysignificant.TheseresultssupportH1whichstatesthat

totalgovernmentequityparticipationhasapositiveinfluenceon

the degree of internationalization of Brazilian firms. They are

consistent with the results of Shamsuddoha, Ali, and Ndubisi

(2009),Wangetal.(2012),andZutshiandGibbons(1998)inthe

caseofAsianfirms.ThecoefficientofTotalGovernment_%(t-1)is

0.066inmodel3,whichmeansthataonepercentchangeinthe

equitytakeninafirmbyallgovernmententitieswillresultina

0.06 increase in thefirm’s percentage of foreign tototal sales.

When we split the total Brazilian government stake into its

constituents,i.e.thestaketakenbyBNDES,bySOEpensionfunds,

andbyothergovernmentinstitutions(Models4–6),wefindthat

governmentparticipationthroughSOEpensionfundsandthrough

BNDES has a positive influence on internationalization, as the

coefficients of Pension Funds_% (t-1) and BNDES_% (t-1) are all

Table4

Impactofownershiptypeoninternationalization.

TobitModeltruncatedat0—DependentVariable=FSTS(t)—MeanoftheMarginalEffectontheExpectedValueoftheTruncatedOutcome

>10% >0%

VARIABLES Obs. Model1 Model2 Model3 Obs. Model4 Model5 Model6

GOVERNMENT_D(t-1) 70 1.4732 2.03477 4.14117*** _{57 1.31179}** _2.60331*** _2.90773*** (0.86) (1.18) (2.87) (2.31) (3.14) (3.83) GOVERNMENT_FAMILIES_D(t-1) 64 3.36419*** _2.81862*** _4.51762*** _{92 5.27575}*** _5.93005*** _5.53048*** (3.25) (2.57) (3.35) (7.31) (5.6) (8.8) GOVERNMENT_FOREIGNERS_D(t-1) 24 1.19967 0.76894 2.95523* 19 0.53196 0.97698 0.34487 (0.71) (0.44) (1.72) ( 0.21) ( 0.47) ( 0.32) GOVERNMENT_FAMILIES_FOREIGNERS_D(t-1) 28 5.00788** 4.92482* 6.56277*** 82 4.57184** 6.28511* 6.14614** (2.24) (1.83) (2.97) (2.07) (1.93) (2.54) FAMILIES_D(t-1) 466 0.35316 0.5322 2.56525 365 2.09189** _3.02592*** _3.78709*** ( 0.1) ( 0.15) (0.83) (2.18) (2.58) (3.16) FOREIGNERS_D(t-1) 79 1.64257* 2.8171*** 2.46809*** 76 0.32083 0.76648 1.86175* ( 1.78) ( 5) ( 4.33) (0.16) ( 0.47) ( 1.82) FAMILIES_FOREIGNERS_D(t-1) 121 0.81391 2.20295 4.57328 191 3.30148*** 6.6766*** 6.41212*** (0.18) (0.48) (1.19) (5.19) (4.16) (3.92) OwnershipConcentration(t-1) 0.0214 0.01569 0.00188 0.02033 0.01586* _0.00495 ( 1.27) ( 1.03) ( 0.1) ( 1.54) ( 1.66) ( 0.43) Cash(t-1) 0.04237 0.05766 0.03375 0.03606 0.05858 0.03654 ( 1.18) ( 1.52) ( 1.48) ( 0.83) ( 1.21) ( 1.14) FirmSize(t-1) 1.57662 1.09426 0.72801 1.93992 1.3589 0.90831 (1.14) (0.78) (0.66) (1.56) (1.1) (0.93) FirmAge(t-1) 14.67363** 12.88133** 10.17314** 13.35125* 12.45735* 10.5105* (2.52) (2.38) (2.26) (1.91) (1.78) (1.95) Leverage(t-1) 0.02507** _{0.01406 0.00709 0.03593}*** _0.02721*** _0.01646* ( 2.01) ( 1.38) ( 1.09) ( 3.49) ( 3.13) ( 1.88) ROA(t-1) 0.03262 0.02388 0.00075 0.04256** 0.03356 0.00633 ( 1.62) 0.69) ( 0.05) ( 2.2) ( 0.93) ( 0.39) FSTS(First) 0.62309*** 0.48535** (3.43) (2.2) FSTS(t-1) 0.19997*** _0.19883*** (8.9) (9.93)

Constant Yes Yes Yes Yes Yes Yes

YearFixedEffects Yes Yes Yes Yes Yes Yes

FirmFixedEffects Yes Yes Yes Yes Yes Yes

ClusteredStandardErrorsoneachIndustry Yes Yes Yes Yes Yes Yes

Observations 956 878 874 956 878 874 Firms 173 173 173 173 173 173 Logpseudolikelihood 1728.9861 1505.9179 1447.2989 1729.3877 1504.3377 1445.7549 PseudoR2 0.3922 0.4054 0.4244 0.3921 0.406 0.425 * p<0.10. ** p<0.05. *** p<0.01.

positiveandstatisticallysignificant.Thestaketakenbyother

state-owned agencies (Other Government_% (t-1)) is not statistically

significant.

InTable4weinvestigatetheimpactofgovernmentstakeson

internationalizationinfirmsinwhichthesestakesarecombined

withthoseoffamiliesandforeigners.Todothisweusethedummy

variablesdescribedinTable1,whichhavea10%cutoff.Hence,as

showninTable1,GOVERNMENT_Dcorrespondstothecaseoffirms

wherethetotalgovernmentstakeisgreaterthan10% whilethe

stakesoffamiliesandforeignersarebothlessthan10%.Likewise,

GOVERNMENT_FAMILIES_D corresponds tothe case of a firm in

whichthetotalgovernmentstakeandthestakeheldbyfamilies

arebothgreaterthan10%,andthatheldbyforeignersislessthan

10%,whileGOVERNMENT_FOREIGNERS_Dcorrespondstothecase

wherethetotalgovernmentstakeandthestakeheldbyforeigners

arebothgreaterthan10%andthatheldbyfamilieslessthan10%.

Theomittedgrouparefirmsinwhichthegovernment,thefamilies

andforeigninvestorsownlessthan10%ofthevotingshares,i.e.

firmswithdispersedownership.Models4–6ofTable4alsoshows

the results with a zero cutoff. In that case, GOVERNMENT_D

corresponds to the case where the total government stake is

greaterthanzero,whilethestakesoffamiliesandforeignersare

zero.LikewiseGOVERNMENT_FAMILIES_Dthendescribesthecase

wherethetotalgovernmentstakeandthestakeheldbyfamilies

areboth greaterthanzero,and that heldby foreignersiszero,

while GOVERNMENT_FOREIGNERS_D corresponds to the case

wherethetotalgovernmentstakeandthestakeheldbyforeigners

arebothgreaterthanzerowiththatheldbyfamiliesbeingequalto

zero.Theomittedgroupinthiscaseiscomposedoffirmsinwhich

thegovernment, familiesand foreigninvestorsdo nothaveany

directorindirectvotingshares.Thecoefficientsshowthemeansof

themarginaleffectsofthevariablesontheexpectedvalueofFSTS.

Model3ofTable4showsthatthedegreeofinternationalization

is higher for firms in which the government is the dominant

shareholder,thatiswhenitistheonlypartywithmorethan10%of

theshares,thanforfirmswithdispersedownership.Theimpactof

governmentownershipis particularly strongwhenthe

govern-ment allies itself with family firms: the coefficients of the

GOVERNMENT_FAMILIES_D dummy are positive and statistically

significantat1%inallmodels.Theyarealsolargerthanthosefor

theGOVERNMENT_Ddummyatbotha10%andazerocutoff.This

supportsH2inwhichwehypothesizedthatfirmsunderthejoint

influence of families and the government would be strong

internationalizers.Thesizeoftheeffectiseconomicallysignificant,

sincethecoefficientofGOVERNMENT_FAMILIES_Dis4.52inModel

3,indicatingthatFSTSis4.5percentagepointshigherinfirmsin

whichthegovernmentandfamiliesowneachmorethan10%ofthe

voting shares(andforeign investorsless than 10%)than in the

omittedcategory.FSTSisalsohigherinfirmswheretheBrazilian

government,families,andforeignerseachownmorethan10%of

the shares than in the omitted category, as the coefficient of

GOVERNMENT_FAMILIES_FOREIGNERS_Dispositiveandsignificant

inModel3.TheFSTSforthesefirms(6.56inmodel3)isalsohigher

than that for firms where the government sharescontrol with

foreigners(2.95inmodel3),butalsohigherthanthatforfirms

wherethegovernmentsharescontrolwithfamilies(4.52inmodel

3).ThisprovidesonlypartialsupportforH4.

ThecoefficientofGOVERNMENT_FOREIGNERS_Dispositive,but

onlymildlysigni_ficantinModel3,showingthat_firmsinwhichthe

governmentsharescontrolwithforeigners arenot significantly

moreinternationalizedthanfirmswithdispersedownership.Thus

ourresultsonlyprovideweaksupportforH3.Aswementioned,

theforeignownersofmostofthesefirmsareprivateequityand

investmentfirmswho maypreferthat thefirmsinwhich they

investfocusentirelyonBrazil.Foreignownersarealsolesslikelyto

beinfluencedbygovernmentequityastheyarelessdependenton

governmentresources.AsTable4shows,alltheprecedingresults

areunchangedwhenweuseazeropercentcutoff(models4–6).

4.1.Robustnesstests

We have seen that endogeneity poses a potential problem

becausetheBraziliangovernmentdoesnotchooserandomlythe

firmsinwhichitwantstotakeequity.Ifitchoosestoinvestin

companiesthatalreadyhaveahighdegreeofinternationalization,

this maylead ustoerroneouslyconcludethatit isgovernment

equitythatleadstointernationalization.Tocontrolforthis,weuse

Stata’spropensityscorematchingprogram.Thisprogramcreates

comparablecontrolgroupsoffirmswithcharacteristicssimilarto

thoseinwhichthegovernmenthastakenastake(eitherthrough

BNDES,SOE pension funds,orothergovernmentagencies).The

firm-levelcharacteristicsconsideredaretheindustrytowhichthe

firm belongs, and firm size, ownership concentration, cash

holdings, leverage, age and profitability. We then compare the

levelofinternationalizationinthesetwogroupsofsimilarfirms

whentheydifferinthelevelofgovernmentequity.

TheresultsarepresentedinTable5.Theyshowthatfirmsin

which thegovernment has an equity stake have on average a

higherdegree ofinternationalizationthan firmswhich haveno

governmentequity stakebut aresimilar interms of size,cash,

leverage,andprofitability,andbelongtothesameindustry.This

resultisstatisticallysignificantatthe5%confidencelevelwhen

using0%ownershipasthresholdfortotalgovernmentstakeandat

the 1% confidence level using 10% as threshold for total

Table5

Impactofgovernmentequitystakeoninternationalization. PropensityScoreMatching

Output—FSTS(t) N.Obs. N.ObsTreated N.ObsUntreated Treated Controls Difference StandardError T-stat

TotalGovernment>0% 941 242 699 26.149 19.332 6.817**

3.4647 1.97

TotalGovernment>10% 948 185 763 28.107 17.581 10.526***

3.8804 2.71

TotalGovernment<10%and>0% 948 63 885 17.992 18.094 0.103 4.0396 0.03

BNDES>0% 948 106 842 38.759 28.417 10.342** _{5.1418 2.01} BNDES>10% 688 48 640 44.725 37.029 7.696 6.3316 1.22 BNDES<10%and>0% 793 58 735 33.822 31.181 2.64 7.6669 0.34 PensionFunds>0% 941 163 778 24.156 17.389 6.767** 3.2146 2.11 PensionFunds>10% 924 117 807 27.592 19.673 7.919** 3.6035 2.2

PensionFunds<10%and>0% 931 52 879 13.735 12.839 0.895 3.5397 0.25

OtherGovernment>0% 776 74 702 24.951 15.540 9.411 6.4446 1.46

OtherGovernment>10% 769 48 721 23.269 16.090 7.179 7.0343 1.02

OtherGovernment<10%and>0% 752 26 726 28.057 29.513 1.456 7.8922 0.18

* p<0.10. ** p<0.05. *** p<0.01.