ANDQUANTITATIVE ANALYSIS JOURNAL OF FINANCIAL VOL.38, NO. 1, MARCH 2003 WA98195 UNIVERSITY OF WASHINGTON, COPYRIGHT SEATTLE, 2003, SCHOOLOF BUSINESSADMINISTRATION, EquityOwnershipand FirmValuein Emerging Markets KarlV.Lins* Abstract This paperinvestigateswhethermanagementstock ownershipand large non-management blockholdershareownershipare relatedto firmvalue across a sample of 1433 firmsfrom 18 emerging markets. When a managementgroup's control rights exceed its cash flow rights, I find that firm values are lower. I also find that large non-managementcontrol rights blockholdings are positively relatedto firm value. Both of these effects are significantly more pronouncedin countrieswith low shareholderprotection. One interpretation of these resultsis thatexternalshareholderprotectionmechanismsplay a role in restraining managerialagency costs and that large non-managementblockholderscan act as a partial substitutefor missing institutionalgovernancemechanisms. I. Introduction Recentresearchshows thatlargeblockholdersdominatethe ownershipstructures of firms not domiciled in the U.S. or a few other developed countries (Shleifer and Vishny (1997), La Porta, Lopez-de-Silanes, Shleifer, and Vishny (LLSV) (1998), La Porta,Lopez-de-Silanes,and Shleifer(1999), Claessens, Djankov,and Lang (2000), andDenis andMcConnell(2003)). This researchsuggests thatsuch concentratedownershipcoincides with a lack of investorprotectionbecause owners who are not protectedfrom controllerswill seek to protect themselves by becoming controllers.When controlhas incrementalvalue beyond any cash flow rights associatedwith equity ownership,shareholderswill seek to obtaincontrol rights that exceed cash flow rights in a given firm. Aroundthe world, controlin *finkvl@business.utah.edu,David Eccles School of Business, University of Utah, 1645 E. Campus CenterDrive, Salt Lake City, UT 84112. I thank Stijn Claessens, JenniferConrad,Amy Dittmar, Robert Dittmar,Simeon Djankov,Mustafa Giiltekin, Mark Lang, Mike Lemmon, John McConnell (the editor), Henri Servaes, Anil Shivdasani,Marc Zenner, and Ingyu Chiou (discussant) and other participantsat the 1999 EuropeanFinance Association Conferencefor helpful remarks.I also thank seminarparticipantsat Arizona State University,Emory University, SouthernMethodist University, Universityof Georgia,Universityof NorthCarolinaat ChapelHill, Universityof Notre Dame, University of Pittsburgh,Universityof Utah, Universityof Virginia,and VanderbiltUniversity.I thankMara Faccio (the referee) for many detailed commentsand for suggesting additionalregressionmodels that have improvedthe paper substantially. I thank Stijn Claessens and Simeon Djankov for providing access to the stock markethandbooksused in the WorldBank East Asia ownershipstudies. I am also grateful to Stijn Claessens, LarryLang, and Mara Faccio for providing the ownershipdata used in their studies of East Asia and WesternEurope. 159 160 Journalof Financialand QuantitativeAnalysis excess of proportionalownershipis usually achievedthroughpyramidstructures in which one firmis controlledby anotherfirm,which may itself be controlledby some otherentity. The managementgroup(andits family members)is usuallythe largestblockholderof a firmat the top of the pyramidand thereis significantoverlapbetween the top firm'smanagementgroupand the managersof each firmdown the line in the pyramid. Thus, the controllingmanagersat the top of a pyramidare generally able to exercise effective control of all the firms in the pyramid,while they bear relatively less of the cash flow consequences of exercising their control in each firmdown the line. Finally,irrespectiveof pyramiding,managersof a given firm sometimes issue and own shareswith superiorvoting rights to achieve control rightsthatexceed theircash flow rightsin the firm (Zingales (1994), Nenova (2003)). Takentogether,the net result is that a great numberof firms aroundthe worldhave managerswho possess controlrightsthatexceed theircash flow rights in the firm, which, fundamentally,gives rise to potentially extreme managerial agency problems. The extent to which managerialagency problemsaffect firm value is likely to dependon severalfactors.If thereare cash flow incentivesthatalign managers' interestswith those of outside shareholders,this should raise firm values. Alternatively,if a managementgroupis insulatedfrom outside shareholderdemands, a situationoften referredto as managerialentrenchment,managersmight choose to use theircontrolto extractcorporateresources;this consumption(or expected consumption)of the privatebenefits of control should reduce firm values. When managershave control in excess of their proportionalownership,the consumption of privatecontrol benefits is especially likely since this type of ownership structureboth reduces cash flow incentive alignmentand increases the potential for managerialentrenchment.Conversely,if managersact in the best interestof all shareholders,then firmvalues shouldnot dependon managerialcontrolrights. Finally,to the extentthatmanagement'scontrolrightsarecorrelatedwith its cash flow rights, additionalmanagerialcontrolcould resultin higherfirmvalues. Non-managementblockholdersmight also impact firm value. If there are large non-managementshareholdersthathave both the incentiveto monitormanagementand enough controlto influence managementsuch that cash flow is increased,firmvalues shouldbe higherbecause all equityholderssharein this benefit of control. Of course, as with managers,large non-managementblockholders might choose to use their power to extractcorporateresources,which would reduce firmvalues. Finally,all of these factorsare potentiallyeven more important where externalshareholderprotectionis the weakest. This papertests the above hypothesesusing a sampleof 1433 firmsfrom 18 emergingmarkets. Emergingmarketsprovidean excellent laboratoryto study the valuationeffects of ownershipstructurefor severalreasons. First, pyramidownershipstructures are prevalentacross virtuallyall emergingmarkets.Second, emergingmarkets generallysufferfrom a lack of shareholderand creditorprotectionand have poorly developed legal systems (LLSV (1998)). Finally, marketsfor corporate control(i.e., the takeovermarket)are generallyunderdevelopedin emergingmarkets (TheEconomistIntelligence Unit (1998)). Overall,where externalcorporate governanceis weak and managerialcontroloften exceeds its proportionalowner- Lins 161 ship, extrememanagerialagency problemsmay arisebecause the privatebenefits of controlarelarge.1Non-managementblockholdersmay be especially beneficial to minorityshareholdersif they help fill the externalgovernancevoid. LLSV (2002) and Claessens, Djankov,Fan, and Lang (2002) provide some evidence on the relationbetween firmvalue, as measuredby Tobin'sQ, and ownership structureacross differenteconomies. Both papersfocus exclusively on the ownershipcharacteristicsof a firm'slargestshareholder,which is usually,but not always, the managementgroupand its family. These papersdo not explicitly test how the relationbetween management/familyownershipand firmvalue could be affectedby otherblockholdersthat are not partof the management/familygroup. LLSV study the 20 largestfirmsin each of 27 wealthyeconomies and reportthat the cash flow rights held by the largestblockholderare positively relatedto firm value. They find no relationbetween Q and a separationin the controlrights and cash flow rights held by the largest blockholder.2 Claessens et al. (2002) study a large set of firmsfrom eight East Asian emergingeconomies and also find that the cash flow rightsheld by the largestblockholderarepositivelyrelatedto value. Additionally,they find that a differencein the controlrights and cash flow rights held by the largestblockholderis negativelyrelatedto firmvalue. This paperbuildson previousworkrelatingownershipstructureto firmvalue in severalways. First,in all of my samplefirms,I explicitly accountfor the effect of managementgroup(andits family) ownershipandwhetherthereis a largenonmanagementblockholderpresentin the ownershipstructure.Since it is the management group that actually administersa firm, the reductionin value from potentiallycostly agency problemsmay be even worse when the managementgroup has sufficientcontrolto exploit minorityshareholdersand there is no large nonaffiliatedblockholderto constrainit fromdoing so. Backman(1999) details many examples of listed emerging marketfirms engaging in sometimes egregious expropriationof minorityshareholdersthroughrelated-partytransactions.3 Second, because not every emergingmarkethas identical externalcorporategovernance features,I test whetherany valuationeffects associatedwith ownershipstructure are more pronouncedwhen shareholderprotectionsare the weakest. Finally, I expandconsiderablythe numberof less developedcountriesin which ownership and valuationare studiedand use a broadcross section of firmsfrom each.4 For all of my sample firms, I trace out ultimateownership,which includes both directly and indirectlyheld control and cash flow rights. I employ a broad definitionof managementgroupownership,consisting of a firm'sofficers, directors, and top-level managers,as well as their family members. I find that management groupblockholdingsof control (i.e., voting) rights average30% across andTriantis(2000)arguethatagencyproblemsin emergingmarketsmay Kraakman, 1Bebchuk, be anorderof magnitude largerthanthosein developedeconomies. andYeung(2000)findthatfamilycontrolthroughpyramids 2In contrast,Morck,Stangeland, reducesmarketvalueforCanadian companies. 3My samplecontainsseveralof these firms--CAMInternational, CheungKong Holdings, Hyundai Corporation,Pacific Chemicals, Shangri-LaAsia, and Wembley Industries-all of which have the managementgroup as the largestblockholderand most of which also have pyramidownership structures. 4For some country-specificevidence on ownershipconcentrationand valuationin emergingmarkets, see Denis and McConnell (2003), Claessens and Djankov(1999), Claessens (1997), and Xu and Wang (1997). 162 Journalof Financialand QuantitativeAnalysis my sample. I also groupnon-managementblockholdersinto variouscategories. Interestingly,I findthatthe controlrightsblockholdingsof othershareholdersnot affiliatedwith managementaveragealmost 20%, which indicatesthat large nonmanagementblockholdersmay play an importantcorporategovernancerole in emergingmarketfirms. Managersand their families are the largestblockholder in two-thirdsof samplefirms,consistentwith Claessenset al. (2002) andLa Porta et al. (1999). I also find thatmanagersmake extensive use of pyramidownership structuresin all sample countriesand thatmanagersof LatinAmericanfirmsfrequentlyuse shareswith superiorvoting rightsto furtherincreasethe controlrights associatedwith theircash flow rights. My valuationanalysis containsthree sets of tests. The first uses regression models to test the relationbetween Tobin'sQ and managerialequity holdings, ignoringthe effect of the holdingsof non-managementblockholders.This approach facilitatesdirectcomparisonwith LLSV (2002) andClaessenset al. (2002). When a managementgroup'scontrolrightsexceed its cash flow rights(becauseof pyramiding and/or superiorvoting equity), I find that firm values are lower. I also conduct tests using breakpointsin the level of managerialcontrol and find that managerialcontrol between 5% and 20% is negatively related to Q, consistent with the U.S. resultsof Morck,Shleifer,andVishny (MSV) (1988). These results supportthe managerialentrenchmenthypothesisand indicatethatthe costs of the privatebenefits of control are capitalizedinto shareprices in emergingmarkets. Unlike LLSV (2002) and Claessenset al. (2002), I findno evidencethatincreases in managerialcash flow rights affect Tobin'sQ. My second set of tests provides new evidence that large non-management blockholderscan reducethe valuationdiscountassociatedwith expectedmanagerial agency problemsin emergingmarkets. I categorize firms based on whether the managementgroup is the largest blockholderof control rights and find that managementcontrolin the 5% to 20% range is associated with a substantialreductionin Q only when the managementgroupis the largestblockholderof control rights. When a largernon-managementblockholderis present,management controlin the 5% to 20%rangedoes not affect firmvalue. Regressionsalso show thatQ is positivelyrelatedto the level of non-managementcontrolandto whether a non-managemententity is the largestblockholderof controlrights. In my third set of tests, I present evidence that the valuation impact of pyramidstructuresand non-managementblockholdingsdepends on the level of shareholderprotectionin a country.When managershave controlrights that exceed theirproportionalownership,firmvalues are significantlylower in countries with low shareholderprotection.These findingssuggest thatexternalgovernance mechanismsplay a role in restrainingmanagerswho do not bear the full cash flow consequencesof exercising their privatebenefits of control. I also find that the presenceof large non-managementblockholdersis more positively relatedto value in low protectioncountries. One interpretationof this result is that nonmanagementblockholdersare a substitutefor formalexternalgovernancemechanisms. The next section of the paper describes the sample selection process and the ownershipvariablesused in the paper.Section III discusses the methodology Lins 163 and describes the results. Section IV conductstests of robustnessand Section V concludes. II. Data A. SampleSelectionandOwnership Categorization To constructmy sample, I obtainfirmlevel financialdatafor the fiscal yearend closest to December 31, 1995, from the 1997 Worldscopedatabasefor all countries consideredto have emergingmarketsby The Economist magazine. I eliminate financialfirmsfrom the analysis because Tobin's Q ratios are not suitable valuationmeasuresfor these firmsand eliminatefirmsnot listed on the stock exchange(s) of a given countryto maintainconsistencyin within-countryreporting requirements.I also eliminate 15 firms with negative book equity values to avoid capturingeffects that may be related to extreme financial distress.5 My potentialsampleincludes 2533 firmsfrom 26 emergingmarkets. I compile data on ultimate shareholdings,in which directly and indirectly held sharesand superiorvoting rights shares,if any, are takeninto consideration. I begin by collecting direct (first-level)ownershipof controlrights for all blockholders with stakes at or above a 5% thresholdfrom the most comprehensive source for each country.A detaileddescriptionof ownershipsources is provided in the Appendix. I investigatewhetherany differentialvoting rights sharesexist using Datastream,Global Data Direct, and countryhandbooks.6 Countriesare eliminatedif no primarydatasourcecan be obtainedthatreportsownershipfor at least 50% of the potentialsample firms in the country(based on marketcapitalization) or if directblockholdingsare generallyreportedas categoricaldata(e.g., "othercompanies"),which cannot be tracedbackward. These screens result in a loss of 521 firms. Because my hypotheses focus specifically on the valuation effects of differenttypes of blockholders,I remove 164 widely held firms (i.e., firmswith no blockholdersat the 5% level) from the analysis. Once the directblockholdersof my sample firmsare established,I traceout the ultimatecontrol of these directblockholders.To do so, I use countryand regional handbooksand firm-levelsearches on Lexis-Nexis. I categorize a firm's ultimateblock ownershipinto ManagementGroupownershipas well as ownership by variousnon-managemententities. I define managementgroupownership broadly,comprisingpersonslisted as: CEO, CFO, President,or any otherofficer and directorof the company;Executive, Deputy, or HonoraryChairman;Treasureror GeneralManager;and their family members(based on overlappingsurnames). Non-AffiliatedCompanyOwnershipis definedas the ownershipposition of companiesnot affiliatedwith management.GovernmentOwnershipcomprises 5Removingthese firmsis also importantbecause cross-countrydifferencesexist in severalfactors that influence the likelihood of observing firms with negative book equity, such as whetheran automatic stay on assets is allowable and whetheran equity capitalreservemust be maintained(see LLSV (1998)). I can identify the largest blockholderin eight of these firms. Ownershipis similar to the full sample-the managementcontrols six of the eight firms (four of these six have pyramids),the governmentcontrols one firm,and a bank controls one firm. 6I include non-votingstock designatedas preferredstock in this measurewhen the dividendrights and paymentsare equal to those of the common stock. 164 Journalof Financialand QuantitativeAnalysis direct and indirectownershipby all agencies and companies that I can identify as being state-controlled(e.g., TemasekHoldings in Singapore).I define InstitutionalOwnershipas ownershipby pensionfunds,mutualfunds,insurancecompanies, and directownershipby banks. I classify ownershipby personswho are not managers(or family members)as IndividualOwnership.Miscellaneousrefersto ownershipnot categorizedelsewhere (e.g., religious/educationalfoundationsand employees). If the ultimate controllerof a direct blockholderof a sample firm cannotbe clearly identifiedas being partof the managementgroupor belonging to anothercategory,thatblockholdingposition is coded as unobservable. Nominee accounts are used extensively in Asia. Fortunately,the Thornton Guideto Asian Companiesidentifiesthe ultimateownerof the nominee accounts for a large numberof Asian firms-such informationis generally not available in Worldscopeor the handbooksused in other ownershipstudies. Still, it is often impossible for me to identify the ultimateowners of some nominee accounts. Also, I am sometimesunableto find ownershipand managementdataon some of the companies that are direct blockholdersof sample firms. I eliminate firms if I cannotidentify the ultimatecontrollersof at least 90% of a firm'sdirect blockholdings. I also eliminate China and Poland because I am unable to identify the controlof at least 90% of the blockholdingsin more than half of the sample firms.7 My sample with ultimateownershipdata on controlrights contains 1433 firmsfrom 18 countries.8 Figure 1 illustrateshow I compute managementgroup controlrights using a Brazilianfirm,Acos Villares, SA. Wheneverthe managersof Acos Villares or their family membersare also the largest shareholdersof one of its blockholders, I classify these shareholdingsin the managementgroupcategory. Thus, the 50.1% of the voting shares owned by IndustriasVillares, SA are designated as managementgroup controlrights. I also classify a blockholderas controlledby the managementgroup if the managersof Acos Villares or their family members are partof the managementof the blockholder.Thus, the 5.8% of the voting sharesheld by Acesita are also assignedto the managementgroup.I define "managementgroupcontrolrights"as the sum of directblock ownershipand indirect controlblocks held by managersandtheirfamilies, which equals 55.9% for Acos Villares,SA.9 I use my direct and indirectownershipinformationto determinewhat fraction of the cash flow rights is controlledby the managementgroup. I sum the structure 7SeeTian(2001)foranownership studyon Chinausingcustomizeddata. concernthatarisesin my sampleselectionprocessis whetherthe firmsfor whichI can 80One differentfromthepotentialsample.Toassessthispossibility, dataaremeasurably gatherownership I compare,by country,summarystatisticsfor the financialvariableslistedin Table1 betweenmy potentialsampleof 2533firmsandmy finalsample.I findsignificantdifferencesonlyforArgentina I verifythatall resultsobtainedin the valuation andIndonesia(65 firmsin total). Forrobustness, sectionholdwhenthesecountriesareremoved. fromthemethodusedby LaPortaet al. 9Mymethodof assigningcontrolrightsdifferssomewhat (1999)andClaessenset al. (2000). Bothpapersassigncontrolbasedon the weakestlinkalongthe Villares controlrightsassignedto Industrias chainof control,in whichcasethefamily(management) wouldequal32.1%.Also,thesepapersdo not assigncontrolbasedon management overlapwithout cashflowownership.As such,La Portaet al. andClaessenset al. wouldclassifythe corresponding controlof Acesitaas eitherheldby a financialfirmorwidelyheld,depending upontheconcentration withinthepensionfundsthatcontrolAcesita. of ownership Lins 165 directly and indirectlyobtainedcash flow rights held by the managementgroup, including the effects of any superiorvoting rights shares, and call this measure "managementgroup cash flow rights."This ownershipis not always observable since some of my ultimateownershipsources reportthe managementstructures or beneficiariesof the blockholdersof my samplefirms,but not theircorresponding cash flow rights. When this occurs, I retain a firm only if the ultimate cash flow rightsof at least 90%of the firm'stotal blockholdingscan be observed.This reducesmy sample for tests involvingcash flow rightsto 1130 firms. I next constructa measurecalled "managementcash flow rights leverage" that identifies how much the managementgroup of a firm levers its cash flow rightsinto greatercontrolrights. This measureis computedas managementgroup controlrightsdividedby managementgroupcash flow rights.10 Cash flow rights leveragewill be aboveunitywhen managershold indirectstakeswith less thanfull ownershipor shareswith superiorvoting rights. To lessen the impactof outliers, I censor the values for cash flow rights leverage at the 95th percentileby setting outlying values to the 95th percentile. As a simple example of managementcash flow rights leverage,suppose the managementof Firm A owns 50% of the sharesof Firm B that owns 50% of the shares of Firm A. I compute managementcash flow rights ownershipas 25% (50% of 50%), managementcontrolrights ownershipas 50%, and management cash flow rightsleverageas two (50%/25%).A morecomplicatedexample of the computationof managementcash flow rights leverage is presentedin Figure 1. The figure details how the managementgroup of Acos Villares SA uses both a pyramidstructureand non-votingsharesto lever 2% of the cash flow rights into 56% of the controlrights for a managementcash flow rightsleveragevalue of 28 (the 95th percentilefor this measureis about 10). Forrobustness,I gatherinformationon exchange-specificregulationsregardthe ing reportingof ownershippositions and then recomputeall ownershiplevels counting only those positions that are above the requireddisclosure level (using 10%and then 20% as a cutoff when a specific level is not reported).11 I find that my results are similarand often stronger(not tabulated). B. OverallSummaryStatistics Table 1 reportssummarystatisticsfor financialvariables(panel A) and ultimate ownershipvariables(panel B). The first column of panel A lists means of my primaryvaluationmeasure,Tobin'sQ, which is computedas the marketvalue of equity plus book assets less the book value of equity,all dividedby assets. To alleviate the influence of extreme values, Tobin's Q is censored at the first and 99th percentiles by setting outlying values to the first and 99th percentiles, respectively. The second column shows that the sample is made up of relatively variable forthespread controlandcashflowrightsis different between fromthatusedin l'oMy LLSV(2002)andClaessenset al. (2002).Bothpaperscomputethemeasureas thedifference between controlandcashflowrights(ratherthanthe ratio)and,as notedin the Introduction, do so only for each firm'slargest blockholder. 1Data on requiredreportingcome from a worldwidesurveyof stock exchangeregulations(Zenner (1995)). 166 Journalof Financialand QuantitativeAnalysis 1 FIGURE Structure of AcosVillares, SA Ownership 58 6 AcosVillares, SA Pres:P.D.Villares VP:W.N.Brumer Shares Voting (36%) Shares Non-Voting (64%) 50.1 IndustriasVillares,SA Shares 31( Voting (34%) ( 20.940 BNDESPAR (Government DevelopmentBank) Pres:P.D. Villares Shares Non-Voting (66%) 32.6 20 P.D.Villares SulAmerica Group (Investment Fund) nvestment Fund) Sample Calculations Management GroupControlRights: + 0.501 0.058 = 0.559 Acesita Pres:W.N.Brumer (CiaAcosEspeciais Itabira) (nocashflowownership) Voing Shares (67%) 48.4 Non-VotingShares (33%) 28 8'(PD (Ind.Villares) (Acesita) Management Group Cash Flow Rights: (obtained through Ind. Villares only) [(0.326) * (0.34)]*[(0.501) * (0.36) = 0.02 Villares) voting (Ind. Villares) voting fraction fraction PensionFunds (e.g.,Petros,Previ,Sistel) largefirms,with mean assets of $886 million. The thirdcolumnshows that South Koreanfirmshave the most debt as measuredby total liabilities to assets. The first two columns of panel B show that about 50% of a firm's control rightsare held by 5% (or greater)blockholders,on average.Of these, about60% are held by the managementgroup. Thus, the percentageof blockholdercontrol rightsheld by entitiesotherthanmanagementis also substantial,averagingalmost 40%of totalblockholdings.The lattercolumnsof panelB list the frequencythata given type of owneris the largestultimateblockholderof controlrights. The table shows thatthe managementgroupis, by far,the dominanttype of blockholderin emergingmarkets,controlling69% of sample firms, followed by companies not affiliatedwith the managementgroup(16%)and the government(7%). Financial entities arerarelya firm'slargestultimateblockholderin emergingmarkets. Table2 reportsstatisticson the mechanismsused to achievemanagerialcontrol using the 1012 firms for which the managementgroup is the largestblockholder of the controlrights of a firm. Panel A reportsthat 66% of managementcontrolledfirms use pyramidsto increase their control rights.12 I also find that 12My results for South Korea, Taiwan, and Thailand differ somewhat from those reportedby Claessens et al. (2000) who find a higherincidence of pyramidsfor the controllingshareholder.I study only management-controlledfirms and my classification mechanismsfocus on establishingmanagement control at any level above 5%, whereas they reportpyramiddata only for firms with a 20% or largerblockholder.Firmsin which managementcontrol is less than 20%, but managementis still the Lins 167 TABLE1 SummaryStatistics PanelA. FinancialVariables PanelB. UltimateOwnershipVariables Control Rights Country N 3147 2685 1257 0.42 0.44 0.36 0.11 0.07 0.08 16 34 45 38 38 14 33 52 83 67 25 13 0 15 0 0 3 7 0 2 0 0 3 0 10 219 53 12 262 1.13 1.24 1.33 1.34 2.24 158 824 446 989 514 0.41 0.45 0.45 0.49 0.45 0.11 0.06 0.09 0.07 0.09 21 41 36 31 29 22 12 23 19 25 60 80 65 67 69 40 16 19 25 11 0 1 7 8 13 10 2 2 0 8 0 0 8 0 0 0 0 0 0 0 10 35 1.82 1.61 316 489 0.42 0.33 0.10 0.11 41 44 22 11 70 83 20 6 0 9 10 6 0 0 0 0 0.90 1.64 1.48 1.03 1.11 1.67 1.36 1.88 520 447 935 237 59 436 387 330 0.52 0.43 0.44 0.72 0.40 0.38 0.53 0.47 0.05 0.09 0.07 0.09 0.13 0.06 0.09 0.07 31 33 34 16 30 18 21 31 23 26 26 8 7 5 20 34 69 66 57 73 85 82 56 50 12 10 16 8 0 10 34 25 8 12 0 7 0 3 6 18 15 11 26 6 14 2 2 4 0 0 1 5 0 1 2 0 0 1 0 1 0 2 1 4 9 59 30 Czech Rep. Hong Kong Indonesia Israel Malaysia Portugal Singapore So. Africa So. Korea Sri Lanka Taiwan Thailand Turkey NonNonNonAffiliated Tobin's Total Debt Capex/ Mgmt. Mgmt. Affiliated 0 Assets Ratio Assets BH BH Mgmt. Co. Gov't. Inst. Indiv. Misc. 1.14 0.81 1.61 Argentina Brazil Chile Peru Philippines Frequencyof Control by BlockholderType 26 146 96 163 7 119 149 28 AllCountries 1433 1.52 886 0.48 0.08 30 19 69 7 16 7 1 1 Summarystatisticsare reportedfor financialvariablesin panel A and for ultimateownershipvariablesin panel B. In panel A, Tobin's0 is computedas the marketvalue of equityplus book assets less the book value of equityall divided by assets and is censored at the 1st and 99th percentilesby setting outlyingvalues to the firstand 99th percentiles, respectively. TotalAssets are reportedin millionsof $U.S. The Debt Ratiois computedas total liabilitiesto assets. MeanValuesare reportedin panel A. PanelB lists the percentageof totalControlRightsheld by blockholdersand the frequencythatan ownershiptype is the largestblockholderof controlrights.The Management(Mgmt.)groupcategory aggregates directownershipof votingshares and indirectcontrolobtainedthroughpyramidsforpersons listedas CEO, CFO,President,or other officersand directors,Executive,Deputy,or HonoraryChairman,and GeneralManager,as well as familymembersof these managers. Non-Affiliated CompanyOwnershipis definedas the ownershippositionof companiesnot affiliatedwithmanagement.Government(Gov't.)Ownershipcomprisesdirectand indirectownershipby all agencies and companiesthat I can identifyas being state-controlled.Institutional Ownership(Inst.)is ownershipby pension funds, mutualfunds, insurancecompanies, and directownershipby banks. I classify ownershipby persons whoare notmanagers(orfamilymembers)as Non-Affiliated Individual (Indiv.)Ownership.Miscellaneous(Misc.)refersto foundationsand employees). Non-management control ownershipnot categorizedelsewhere(e.g., religious/educational rightsaggregates the ultimatecontrolrightsheld by allentitiesotherthanmanagement.BHrefersto blockholdings. shares with superiorvoting rights are used extensively in Brazil and Peru, but rarely are used by sample firms outside of Latin America.13 Panel B of Table 2 reportsstatisticson managementcash flow rightsleverage. The panel shows that controllingmanagersare able to turnone cash flow right into 2.7 controlrights, on average. In panel C of Table 2, I dig more deeply into managementusage of shares with superiorvoting rights in the two sample countriesin which superiorvoting shares are prevalent,Brazil and Peru. Note from panel A that managersof firms from these countriesuse pyramidsfrequently.PanelC summarizesthe fractionof non-voting shares in the common equity capital structureoverall and by largest blockholdertype. I findthatthe equity structuresof management-controlled firms are heavily weighted towardnon-voting shares. The median non-voting equity percentageof 63%is very close to the legally permittedthresholdof 66% in these largestblockholder,occur frequentlyin these threecountriesand managementusually holds its shares directly in these cases. 13Nenova(2003) also finds substantialuse of non-votingequity by Brazilianfirms. 168 Journalof Financialand QuantitativeAnalysis TABLE 2 Control Mechanisms to AchieveManagement PanelA. FirmswithMgmt. ControlRights ObtainedIndirectly PanelB. Mgmt.Cash Flow RightsLeverage Country N % N (Mean) Argentina Brazil Chile Czech Rep. HongKong Indonesia Israel Malaysia Peru Philippines Portugal Singapore So. Africa So. Korea SriLanka Taiwan Thailand Turkey AllCountries 3 30 25 6 179 37 8 189 8 29 18 99 55 119 6 102 83 16 1012 100 73 92 100 67 81 75 85 88 90 61 83 89 30 67 37 36 100 66 0 12 11 6 147 22 4 110 3 22 9 60 29 115 4 88 68 9 719 4.43 5.53 5.34 1.58 1.30 3.62 3.39 1.58 3.02 2.00 4.63 3.02 2.58 1.38 2.54 1.98 1.72 2.68 PanelC. ManagementUse of Differential VotingRightsShares LargestBlockholder Countries Brazil,Peru AllFirms N = 68 Mean Med Mgmt Group (N = 36) Mean Med Non-Mgmt Group (N = 32) Mean Med Difference (p-value) Mean Med 0.14 0.22 0.63 0.39 0.41 0.51 0.53 0.45 Fractionof (0.01) (0.00) non-votingshares The sample used in this table includesonly those firmsforwhichthe managementgroupis the largest blockholderof the controlrightsof a firm. Panel A reportsthe percentage of firmsfor which the managementgroup obtains some controlrightsindirectly.PanelB reportsmeanmanagementcash flowrightsleverage,whichis computedas management groupcontrolrightsdividedby managementgroupcash flowrightsand includesbothpyramidand superiorvotingequity effects. Thisvariableis coded as missingifthe fractionof unobservedmanagementgroupcash flowrightsexceeds 10% of the managementgroups'ultimateblockholdingsof controlrights.The values forcash flowrightsleveragehave been truncatedat the 95th percentile.PanelC summarizesthe fractionof non-votingshares in the equitycapitalstructurefor the two sample countries(Braziland Peru)in whichnon-votingequityshares are prevalent.The fractionof non-voting shares cannotbe determinedforthreefirms. The differencein means is computedusing a t-test and the differencein mediansis computedusing a Wilcoxonrank-sumtest. p-valuesare listedin parentheses. two countries. In contrast,firms controlledby other entities have mostly voting sharesin theirequity structure. To furtherinvestigatemanagement'susage of superior-votingequityin Latin America,I examinethe 64 Mexican firmsfor which I could not obtainconsistent ownershipdataand findthatalmosthalf of these firmshave limited or non-voting equity in their capital structures. In La Porta et al. (1999), all 20 sample firms in Mexico are controlledby management/families. It appearsthat, relative to managerselsewhere, managersof Latin Americancompaniesare unique in their propensityto use superiorvoting rightssharesto separatecontrolrightsfromcash flow rights. This observationwarrantsfurtherstudy. Overall, the ownership structuressummarizedin Tables 1 and 2 indicate that emergingmarketsprovidea rich setting to test whetherpotentialmanagerial agencyproblemsarecapitalizedin firmvalues andwhetherlargenon-management blockholdersplay a governancerole. Lins 169 Ill. ValuationMethodology and Results A. ValuationMethodology To assess the relationbetween ownershipstructureand firmvalue, I firstuse basic OLS regressionsin which Tobin's Q is the dependentvariableand ownership and controlvariablesare the independentvariables. In Section IV, I implement regressiontechniquesthat consider potential endogeneitybetween ownership and Tobin'sQ and also consideralternatefirmvalue measures. My regressionsincludea varietyof controlvariablesto ensurethatthe effects I attributeto ownershipare not due to othercorrelatedfactors. I controlfor firm size with the log of assets (in U.S. dollars). I use the ratio of capitalexpenditures to assets as a proxy for potentialinvestmentopportunities.14 I control for debt to accountfor the possibility that creditorsare able to lessen managerialagency problems(McConnelland Servaes (1995) and Harvey,Lins, and Roper (2002)). I measuredebt as the ratio of total liabilities to assets, which incorporatesstructuraldifferencesbetween countriesin the types of short-and long-termfinancing instrumentsused by firms (Demirgiig-Kuntand Maksimovic(1999)). All regressions include industrydummy variablesbased on industrygroupingsdefined in Campbell(1996). 15 To accountfor the possibilitythatinter-countryvariationin accountingtreatments affects the measurementof Q and other variables,I include countrydummies in all regressions.I considermodels in which countryeffects are allowed to be randomas my base case for all regressions,but choose a fixed effects framework for two reasons. First, a fixed effects model is designed to test for variation in the ownershipand Q relationwithin a country. Second, the Hausmantest rejects the null that country effects are randomin (unreported)regressions with managementownershipand shareholderprotectioninteractionsand in many of the basic ownershipregressionsusing alternativemeasuresof firmvalue.16 B. ValuationResults-Management GroupOwnership In this section, I constructtests thatconsideronly the relationbetween managerial ownershipcharacteristicsand firm value and do not accountfor the presence of outsideblockholders.This approachfacilitatescomparisonwith previous internationalownershipstructurestudies. The first model in panel A of Table 3 tests a simple version of the managerialentrenchmenthypothesis by regressing Tobin's Q on the percentageof control rights held by management. The model 14Analternativeproxy for investmentopportunitiesused by LLSV (2002) is annualgrowthin sales over prioryears. This proxy does not workwell for my emergingmarketssamplebecause Worldscope does not reportpre-1995 data for a significantportion of my firms. I note that annual sales growth (where available)is highly correlated(p-value < 0.000) with the capex/assetsratio of sample firms. 15Theseindustrygroupings are commonly used in internationalfirm valuation studies (see Lins and Servaes (1999), (2002) in additionto Claessens et al. (2002)). 16Thechoice between fixed and randomeffects is often subject to interpretation,even in the absence of a rejection by the Hausmantest. Greene (1997), p. 623, provides an example analogous to my frameworkin which fixed effects are chosen in an inter-countrycomparisonbecause the sample includes a nearly exhaustive set of countries (e.g., emerging markets)for which it is reasonable to assume that the model is constant. 170 Journalof Financialand QuantitativeAnalysis provides no evidence (at conventionalsignificance levels) that higher management controlrightsare linkedto lower firmvalues. I next performtests of the managerialentrenchmenthypothesisthattake into accountpotentialnon-linearitiesin the relationbetween managementownership and firm value along the lines of MSV (1988). I use control rights ownership for these tests since they will always be equal to or higher than the cash flow rights held by managementand I can observe controlrights more frequentlyfor my sample firms. MSV arguethat managemententrenchmenteffects dominate incentivealignmenteffects over a low to intermediatelevel of managementgroup ownership. They choose breakpointsin the range of managementownershipat 5% and 25%, althoughthey note that these cutoffs were chosen to fit their data. For my tests, I use a 5% to 20% range because it is likely that effective control can be obtainedat relativelylow levels in emerging markets. I create a dummy variableequal to one when the managementgrouphas between 5% and 20% of the control rights of a firm and a dummy equal to one if managementcontrols more than 20% of the firm. These dummy variableskeep the interpretationof coefficients simple-similar dummy variablesfor managementownershipcells were used in working paper predecessorsto MSV (1988). I also follow MSV (1988) and estimate a piecewise linear regressionusing a variablefor the level of managementcontrol between 5% and 20% computedas actual management control rights if they fall within this range, 0.20 if managementcontrol rights exceed 20%, and zero if no blockholdingsare held by management. Similarly, TABLE 3 Relationbetween Ownershipand FirmValuein EmergingMarkets PanelA. ManagementGroupOwnership Constant (5) (1) (2) (3) (4) 1.949 2.030 2.735 1.891 1.841 (0.00) (0.00) (0.00) (0.00) (0.00) Logof Assets -0.075 -0.070 -0.066 (0.00) (0.00) (0.00) (0.00) (0.00) LeverageRatio -0.369 (0.01) 0.256 (0.33) -0.144 (0.11) -0.357 (0.01) 0.247 (0.35) -0.363 (0.00) 0.248 (0.35) -0.379 (0.00) 0.224 (0.39) -0.301 (0.04) 0.275 (0.35) to Assets CapitalExpenditures Mgmt.GroupControlRights% -0.079 Dummyif Mgmt.Controlsbetween5%and 20% -0.176 Dummyif Mgmt.Controlsabove 20% -0.131 -0.078 (0.01) (0.01) -0.865 Levelof Mgmt.Controlbetween5%and 20% (0.03) 0.112 (0.39) Levelof Mgmt.Controlabove 20% Mgmt.IndirectControlDummy -0.090 (0.04) -0.022 Mgmt.Cash FlowRightsLeverage Adj.R2 No. of Obs. (0.02) 0.25 1433 0.25 1433 0.25 1433 0.27 1130 (continuedon nextpage) 0.25 1433 171 Lins 3 (continued) TABLE Relationbetween Ownershipand FirmValuein EmergingMarkets Blockholders PanelB. Non-Management (1) Constant Logof Assets LeverageRatio to Assets CapitalExpenditures Non-Mgmt.GroupControlRights% (2) 1.871 1.824 1.882 (0.00) -0.077 (0.00) -0.352 (0.01) 0.246 (0.35) 0.259 (0.01) (0.00) -0.077 (0.00) -0.352 (0.01) 0.242 (0.36) 0.321 (0.01) 0.083 (0.53) (0.00) -0.078 (0.00) -0.359 (0.01) 0.245 (0.35) Mgmt.GroupControlRights% Dummyif Mgmt.Groupis notthe LargestBlockholder DummyifMgmt.GroupControlsbetween5% and 20%and is the LargestBlockholder Dummyif Mgmt.GroupControlsbetween5% and 20%and is notthe LargestBlockholder DummyifMgmt.GroupControlsabove 20% and is the LargestBlockholder Adj.R2 No. of Obs. (3) (4) 2.014 (0.00) -0.078 (0.00) -0.361 (0.01) 0.225 (0.39) 0.142 (0.00) -0.264 (0.00) -0.021 (0.83) -0.127 (0.01) 0.25 1433 0.25 1433 0.25 1433 0.25 1433 Regressionanalysisof the dependentvariableTobin's0 on categoriesof managementownershipand controls.Tobin'sQ and mostof the independentvariablesare definedinTables1 and 2. Variablesnew to thistable are as follows.Dummyif Management(Mgmt.)Controlsbetween5%and 20%is an indicatorvariableset equalto one ifmanagementcontrolrights fallwithinthis rangeand zero otherwise.Dummyif ManagementControlsabove 20%is an indicatorvariableset equalto one if managementcontrolrightsexceed 20%and zero otherwise.Levelof ManagementControlbetween5%and 20% is set equal to: actualmanagementcontrolrightsif theyfallwithinthis range;0.20, if managementcontrolrightsexceed 20%;and zero if managementhas no controlrightsblockownership.Levelof ManagementControlabove 20% is set equalto: actualmanagementcontrolrightsiftheyexceed 20%;and zerootherwise.ManagementIndirectControlDummy equals one if the managementgroupobtainsat least some of its controlrightsindirectly.Dummyif ManagementGroup is not the LargestBlockholderis set equal to one if a non-managemententityis the largestcontrolrightsblockholder. PanelA regressionsfocus on managementgroupownershipvariables.Model(3) is a piecewise linearregressionsimilar to MSV(1988). Model(5) is estimatedon the subsampleof firmsforwhichmanagementcash flowrightsleveragecan be computed. The regressionsin panel B includenon-managementcontrolrightsvariables.Allregressionmodels include countryand industryfixed effects (coefficientsnot reported). Industrygroupingsare based on Campbell(1996). The p-valueof the t-testof equalityof each coefficientto zero is reportedin parentheses. a variablefor the level of managementcontrol above 20% is set equal to actual managementcontrolrightsif they exceed 20%, and zero otherwise. Model (2) of panelA shows thatthe dummyvariablesfor managementgroup controlbetween 5% and 20% and above 20% areboth significantlynegativelyrelatedto firmvalue. While the coefficienton the 5%to 20%rangeis morenegative, it is not significantlydifferentfrom the above 20% coefficient. Model (3) reports the results of the piecewise linear regressionin which the slope of managerial controlis allowed to change. The coefficient on managerialcontrolbetween 5% and 20% is -0.865, which indicates that, among firms with potential managerial entrenchmentproblems,each percentagepoint increasein managerialcontrol rightsis associatedwith a 0.0087 decline in Tobin'sQ. The coefficienton the level of managerialcontrolabove 20% is not significant.To the extent thatmanagerial controlin the 5%to 20%rangeproxiesfor potentialmanagerialentrenchment,the 172 Journalof Financialand QuantitativeAnalysis results of models (2) and (3) provide supportfor the hypothesisthat firm values are lower as the potentialfor managemententrenchmentincreases.17 The last two models of panel A investigatethe valuationimpact of mechanisms used by managementto separatecontrol rights from cash flow rights in emergingmarkets.In model (4), I regressTobin's Q on a dummyvariableequal to one if the managementgroupobtainsat least some of its controlrightsthrough pyramidsand find a negativeand significantcoefficient on this dummyvariable. The coefficient indicates that when managersuse pyramidsto obtain some of their control, Tobin's Q values are 0.09 lower. Model (5) tests the relation between managementcash flow rights leverage and firm value using the sample of 1130 firms for which managementcash flow rights, and thus managementcash flow rightsleverage,can be computed.This model shows thatfirmvalue declines as the separationof managementgroup control and cash flow rights gets larger. The coefficient of -0.022 indicates that, all else equal, a firm with an extreme cash flow rights leverage value of 10 would have a 0.198 lower Q value than a firmwith a cash flow rightsleveragevalue of one (no separation). The results from models (4) and (5) highlight the overall loss in firm value that results when the managementgroup'scontrol exceeds its proportionalownership. Thus, these results are consistent with the hypothesis that the expected privatebenefitsof controlaffect firmvalue in emergingmarkets.In an unreported model, I regressQ on the cash flow rightsheld by management(which are highly correlatedwith controlrights;p = 0.60, p-value < 0.0000), but find thatthey are not significantlyrelatedto value. These resultsprovideno supportfor the Jensen and Meckling (1976) convergence-of-interestshypothesis in emerging markets and differfrom those reportedby LLSV (2002) and Claessens et al. (2002). C. ValuationResults-Non-Management Blockholders The previoussets of tests provideevidence thatpotentialmanagerialagency problemsare relatedto the valuationof firms from emergingmarkets.However, these tests do not tell the full story of firm-level corporategovernance, since they fail to take into account any positive or negative impact that large nonmanagementblockholdersmight have on the actions of management. Panel B of Table3 containsregressionmodels that incorporatethe controlrights held by blockholdersthat are not part of the managementgroup, an approachsimilar to the one takenby McConnell and Servaes (1990) for U.S. firms. Overall, my results show that it is beneficial to separatelyinvestigate the valuationeffects of managementand non-managementblockholders,ratherthan focusing solely on the categoryof "largestblockholder"as has been done previouslyin international ownershipstudies. Model (1) of panel B shows thatthe controlrightsheld by non-management blockholdersare positively relatedto firmvalue, which is consistentwith the hypothesis that these blockholdersplay a monitoringrole of some sort in emergmodels,I testcutoffsof 15%,25%,and30%usingbothdummyandlevelvariables l7In unreported controlrightsandthesquareof management andfindsimilarresults.I alsoregressQon management controlrights(Stulz(1988)andMcConnellandServaes(1990))andfindthatthecoefficienton manat the5%level,whilethecoefficienton thesquared agementcontrolrightsis negativeandsignificant term is insignificant. Lins 173 ing markets. This conclusion is reaffirmedin regression model (2), which includes both managementand non-managementcontrol rights. The control of non-managementblockholdersremainspositively and significantlyrelatedto firm value, while the controlrightsof the managementgroupareagainnot significantly relatedto firmvalue. To isolate situationsin which large blockholdersare likely to have the greatest influence over the managementof a firm, I create a dummy variableequal to one when the managementgroupis not the largestblockholderof the control rights of a firm. This is the case for aboutone thirdof my sample firms(see Table 1). Model (3) of panel B of Table 3 shows that a controllingnon-management blockholderis associated with an increase of 0.142 in Tobin's Q. This result is again consistent with the idea that large non-managementblockholderscan provide beneficialgovernancein emergingmarkets.18 Model (4) of panel B is designedto assess whethercontrollingnon-management blockholdersmight be able to lessen the agency costs of managerialenthat can be inferredfrom models (2) and (3) of panel A. 19 I create trench:Tlent an interactionbetweenthe 5%to 20%managementcontrolrightsdummyand the dummywhen managementis the largestsingle blockholderof controlrights. This interactionvariableshouldcapturethe type of managementownershipthatis most likely to face the entrenchmentproblemsdescribedin MSV (1988). I also create a dummy equal to one if managementcontrols between 5% and 20% and is not the largest blockholder.The coefficient on this interactionvariablewill provide an indicationof whetherthe presenceof a large externalblockholderreducesthe loss in firm value associatedwith potentialagency costs of managerialentrenchment.20 Finally, I compute a dummy variableequal to one when management controlsmore than20% and is the largestblockholder.Again, the use of dummy variablesfor these rangeseases the interpretationof the interactioncoefficients. 18Tosee if specifictypesof non-management blockholders affectfirmvaluedifferently, I construct of controlrightsis a non-affiliated dummyvariablesequalto oneif thelargestblockholder company, is thegovernment, or is an institutional owner,andestimatea modelthatincludesthesethreelargest blockholder dummies.I finda significant difference(atthe 10%level)onlybetweenthedummywhen the largestblockholder is a non-affiliated is companyandthedummywhenthe largestblockholder an institution.Sincethisresultdoesnotprovidecompellingevidencethattherelationbetweenfirm valueanda largenon-management blockholder blockholder, dependson thetypeof non-management I continueto groupall non-management blockholders togetherwhenconducting myvaluationtests. of thecoefficientson managerial controlcontainedin models 19Itis possiblethatthe significance sincenon-management whichare (2) and(3) of Table3 is dueto spuriouscorrelation, blockholdings, relatedto Tobin'sQ, areomittedfromthesemodels.It is not appropriate, significantly however,to test thisconjectureby includingin thesemodelsnon-management controlrightsor a dummyif the becausebothmeasuresarehighlynegativelycormanagement groupis not thelargestblockholder, relatedwithmanagement controlabove20%. Instead,I use a dummyequalto one if thereareany as a coarse,butnothighlycorrelated, controlfor non-management non-management blockholdings effectson value.I findthatthisdummyis positivelyandsignificantly relatedto Q andthatthecoefficientsforbothdummiesandlevelsof managerial controlin the5%to 20%andabove20%ranges retaintheirmagnitude andsignificance (if any)frompriorregressions. 20Withtheexceptionof majority workbyHolderness andSheehan(1988)andDenisand ownership Denis(1994),priorresearchon managerial andvaluehasnotexplicitlystudiedwhethera ownership differential valuation relationexistswhenmanagers arethelargestcontrolling entity(seeHimmelberg, Hubbard,andPalia(1999),Holderness,Kroszner,andSheehan(1999),Cho (1998),Lodererand Martin(1997),Kole (1996),McConnellandServaes(1995),andHermalinandWeisbach(1991), amongothers,in additionto previouslyreferenced papers). 174 Journalof Financialand QuantitativeAnalysis Recall that panel A shows a negative and significantrelationbetween firm value and the dummy variablefor managementgroup control between 5% and 20%, irrespectiveof whetherthis is the largest block position. The results are much different,however, when I isolate firms in which the managementgroup controlsbetween5%and20%,butthe managementgroupis not the largestblockholder.As model (4) of panel B shows, the coefficienton this interactiondummy is insignificant. Conversely,when managementcontrols between 5% and 20% and it is also the largestblockholder,a situationin which managementmay have both the ability and desire to consume privatebenefits of control, the regression coefficient is stronglynegative. The coefficient indicates a reductionin Tobin's Q of 0.264 in this case.21 These findings demonstrateagain the governancepotentialof largeinvestorsin emergingmarkets,since managementgroupcontrolin the "entrenchment" rangedoes not correspondto a reductionin firm value when a non-managemententity controlsthe firm. D. Valuation Results-OwnershipandShareholderProtection Emergingmarketsare usually, but not always, associated with low shareholderprotection.Since thereis some dispersionin protection,one might expect that managerscan more easily consume the privatebenefits of control in countries where investors are least protectedby the law (LLSV (2000)). If this potential incrementalconsumptionof privatebenefitsis priced, one should observe lower values for firms with potentiallyextreme managerialagency problems as shareholderprotectiondeclines. To test whethershareholderprotectionmatters, I combine measures of shareholders'legal rights and the enforcementof such rights obtainedfrom LLSV (1998). The first is the "AntidirectorRights" score, which ranges from zero to five with lower scores correspondingto fewer shareholderrights. The second is the "Ruleof Law" score for a country,which ranges from zero to 10 with lower scores correspondingto less traditionfor law and order. These variablesare not reportedfor the Czech Republic so firms from this countryare excludedfrom this analysis. In my empiricaltests, I first use a randomeffects model that interactsmanagementgroupownershipvariablesanda country'sweightedaverageAntidirector Rights andRule of Law score. This type of model has the potentialto incorporate both between- and within-countryeffects of ownershipon value-a fixed effects model is poorly suited for testing between-countryeffects. Unfortunately,the Hausmantest rejectsthe null specificationthatcountryeffects arerandomin these models. Since a randomeffects model is inappropriate,I test whethermanagement agency problemsare more severe in low protectioncountriesby estimating my previouscountryfixed-effectsmodels on a subsetof firmsfromcountrieswith low AntidirectorRights and a low Rule of Law. This "low protection"subsample includes countriesthat score at or below four on the AntidirectorRights measure groupcontrolbetween5%and15%,25%, 21Inunreported models,I testdummiesformanagement is also the largestblockholder.Thecoefficientson thesedummiesare and30%whenmanagement always stronglynegativeand significant(p-value= 0.00), indicatingthatmy resultis robustto changes cutoffpoint. in thechoiceof anownership Lins 175 and at or below seven on the Rule of Law measure. This subsample excludes firmsfrom Chile, Hong Kong, Portugal,Singapore,SouthAfrica, andTaiwan.22 Table 4 reportsownershipstructuretests using the low protectionsubsamThe models reportedcorrespondexactly to those in Table 3. Mean Tobin's ple. in the low Q protectionsubsampleis 1.58, which is close to the mean Q value of 1.52 for the full sample. Therefore,for the purposeof assessing economic effects, the regressioncoefficientsfor the low protectionsubsamplecan essentially be directly comparedto those from Table 3 featuringthe full emergingmarkets sample. The first model of panel A in Table 4 shows that managementgroup control rights have a negative and significantrelationto firm value in emerging marketswith relativelyweak externalgovernancemechanisms. This coefficient is differentfrom the high protectionsubsamplecoefficienton managerialcontrol at the 5% level (significancebased on combined regressiontests). This finding lends some supportto the hypothesisthatthe valuationconsequencesof managerial agency problems are worse when externalgovernanceis weak. Models (2) and (3) of Table4 conductsubsampletests using dummiesandlevels for management controlin the 5% to 20% and above 20% ranges,withoutregardto whether managementis the largestblockholder.In both models, the coefficients on managerial control in the 5% to 20% range are more negative in the low protection subsample,but not significantlyso. Thus, it does not appearthatmanagemententrenchmenteffects measuredusing the 5% to 20% range of managementcontrol are any worse when shareholderprotectionis weak. Model (4) of panel A in Table4 tests whetherthe valuationsof firms with potentialmanagerialagency problemsstemmingfrom pyramidownershipstructuresarelower when shareholderprotectionis the weakest. The coefficienton the managementindirectcontroldummyof -0.19 is significantat the 1%level and is significantlydifferentfrom the high protectioncoefficientat the 1%level. This comparesto a Table3 coefficienton managementindirectcontrolof -0.09 in the full emergingmarketssample. Model (5) shows a largernegative coefficient on the cash flow rights leverage variablein low protectioncountries(-0.037 compared to -0.022). The differencein this coefficient between low and high protection subsamples,however,is not significantat conventionallevels (p-value = 0.11). Takentogether,models (4) and (5) in panel A of Table4 provide support for the hypothesis that the negativerelationbetween firm value and a separation in managementcontrol and cash flow rights is more pronouncedwhere external corporategovernancemechanismsare weakest. In panel B of Table4, I test whetherthe positive relationbetween large nonmanagementblockholdersand firm value is more pronouncedin countrieswith low externalshareholderprotection.The firstmodel shows thatnon-management controlrights are again stronglypositively relatedto firmvalue when shareholder protectionis relativelyweak. The differencein this relationbetweenlow andhigh protectionsubsamplesis significantat the 10%level. Model (2) incorporatesboth 221alsoattempta country-by-country analysisin whichtherelationbetweenownership andvalue is obtainedforeachcountryandthenthe ownership coefficientsfromeachcountryareregressedon measuresof shareholder I amunableto obtainmeaningful resultsusingthis protection. Unfortunately, becausetheownership coefficientsarerarelysignificant in thecountrieswithsmallsample procedure sizes. 176 Journalof Financialand QuantitativeAnalysis 4 TABLE inEmerging Firm Protection Markets Relation betweenOwnership, Value,andShareholder PanelA. ManagementGroupOwnership (1) Constant Logof Assets LeverageRatio to Assets CapitalExpenditures Mgmt.GroupControlRights% 2.720 (0.00) -0.120 (0.00) -0.408 (0.04) 0.058 (0.87) -0.318** (0.03) Dummyif Mgmt.Controlsbetween 5%and 20% Dummyif Mgmt.Controlsabove 20% (2) (3) (4) (5) 2.834 (0.00) -0.126 (0.00) -0.387 (0.05) 3.883 (0.00) -0.101 (0.01) -0.148 (0.61) 0.003 (0.61) -0.013 (0.06) 2.564 (0.00) -0.108 (0.00) -0.426 (0.03) 0.009 (0.97) 2.257 (0.00) -0.081 (0.01) -0.327 (0.14) -0.001 (0.77) 0.048 (0.89) -0.250 (0.01) -0.218* (0.01) -0.865 (0.03) 0.112 (0.39) Levelof Mgmt.Controlbetween 5%and 20% Levelof Mgmt.Controlabove 20% -0.190*** (0.01) Mgmt.IndirectControlDummy Mgmt.Cash FlowRights Leverage 0.29 Adj.R2 789 No. of Obs. Blockholders PanelB. Non-Management Constant Logof Assets LeverageRatio to Assets CapitalExpenditures GroupControlRights% Non-Mgmt. 0.30 789 0.37 361 0.30 789 (1) (2) (3) (4) 2.567 2.600 2.787 2.829 (0.00) (0.00) (0.00) -0.123 (0.00) -0.381 (0.05) 0.052 (0.88) 0.394* (0.01) -0.127 (0.00) -0.381 (0.05) 0.053 (0.88) 0.343 (0.08) -0.073 (0.71) -0.122 (0.00) -0.401 (0.04) 0.021 (0.95) Mgmt.GroupControlRights% Dummyif Mgmt.Groupis notthe LargestBlockholder (0.00) -0.125 (0.00) -0.401 (0.04) -0.001 (0.98) 0.221** (0.00) Dummyif Mgmt.GroupControlsbetween5%and 20%and is the LargestBlockholder Dummyif Mgmt.GroupControlsbetween5%and 20%and is notthe LargestBlockholder Dummyif Mgmt.GroupControlsabove 20%and is the LargestBlockholder Adj.R2 No. of Obs. -0.037t (0.01) 0.30 621 0.30 789 0.29 789 0.30 789 -0.356 (0.00) -0.093 (0.48) -0.218* (0.00) 0.30 789 Regressionanalysisof the dependentvariableTobin's0 on categories of managementgroupownershipand controls estimatedon a subsampleof countrieswithlowshareholderprotectionas measuredby Antidirector Rightsand Ruleof Law. Antidirector Rightsvalues rangefromzero to five and are takenfromTable2 of LLSV(1998). Ruleof Lawvalues range fromzero to 10 and are taken fromTable5 of LLSV(1998). The low protectionsubsample used in all models includescountriesthatscore at or belowfouron the Antidirector Rightsmeasureand at or belowseven on the Ruleof Law measure.Allvariableshave been definedpreviouslyinTables1-3. InpanelA, Model(5) is estimatedon the subsampleof firmsforwhichmanagementgroupcash flowrightscan be established.Allregressionsincludecountryand industryfixed effects (coefficientsnotreported).Industrygroupingsare based on Campbell(1996). Thep-valueof the t-testof equality of each coefficient to zero is reported in parentheses. ***,**, *, and t indicate that an ownership coefficient is statistically different from that for the high protection subsample at the 1%, 5%, 10%, and 15% levels, respectively. Significance levels are based on full-sample regressions with interactions for all coefficients. Lins 177 non-managementand managementcontrolrights and again shows that only nonmanagementcontrolrightsarerelatedto value. The coefficientof 0.343 in the low protectionsubsampleis slightlymorepositive thanthatfor the full sample,butthe differencebetweenhigh and low protectionsamplesis not significant.These tests provide weak evidence that increases in the percentageof control held by nonmanagementblockholdersare more positively relatedto value when shareholder protectionis low. I next test the effect of a controllingblock held by a non-managemententity and the results are much strongerthan those for the overall sample. Model (3) of panel B, Table4 reportsa positive coefficientof 0.221 on the controllingnonmanagementblockholderdummyin low shareholderprotectioncountries,which is differentfrom the high protectionsubsampleat the 3% significancelevel. This model supportsthe hypothesisthat the internalgovernanceprovidedby controlling non-managementblockholdersmattersmore when externalshareholderprotection mechanismsare relativelyweak. Finally, model (4) investigateswhetherthe interactionsbetween large nonmanagementblockholdersand managerialcontrol in the 5% to 20% and above 20% ranges have a differentimpacton firmvalue when shareholderprotectionis lower. The coefficientson managementcontrolin both the 5% to 20% and above 20% ranges when managementis also the largestblockholderare more negative in the low shareholderprotectionsubsample.However,the differencebetweenthe low and high protectionsubsamplesis significantfor only one of the coefficients, so it is hardto conclude much from this model. Summing up, the low shareholderprotectionsubsampletests conductedin Table4 containtwo importantfindings. Whereexternalshareholderprotectionis relatively weak, the relationbetween a separationof managementcontrolrights and cash flow rights and firm value is more negative and the relationbetween a controllingnon-managementblockholderand firm value is more positive. These results are consistent with the hypothesisthat potentialmanagerialagency problems arereflectedin lower firmvalues when externalgovernancemechanismsare least effective. These results are also consistent with the idea that large nonmanagementblockholdersprovide minority shareholderswith an even greater monitoringbenefit when shareholderprotectionis weak. One caveat about my conclusions on the importanceof shareholderprotectionis that I cannotrule out the possibility that these results are driven by other variables correlatedwith shareholderprotection, such as corruption,financial market development, and GDP,ratherthan shareholderprotectionitself. IV. RobustnessTests A. Endogeneity of Ownership If ownershipand value are endogenously determined,then cross-sectional regressions that indicate a relation between ownershipvariablesand firm value cannot be used to make inferences about the causality of the relation. For instance, the negativerelationbetween managementgroup control in excess of its proportionalcash flow and firm value could indicate that firm values are lower 178 Journalof Financialand QuantitativeAnalysis as a result of the market'sexpectationof costly agency problems. However,it is also plausiblethatmanagerswill increasetheir separationof cash flow rights and controlrights if they want to maintaintheir control,but have knowledge that the cash flows of their firm will be lower in the future. In this case, expected poor performancecauses a higher separationof cash flow and control rights, rather than the other way around. Similarly,regressionsthat show a positive relation between large non-managementblockholdersand firm value could indicate that monitoringof managersby large externalblockholderslessens actualor expected managerialagency problems. Conversely,it could be the case that high firm values lead to increasedownershipby these blockholders(Rouwenhorst(1999) and Chui, Titman,and Wei (2000)). It is difficultto disentangleendogeneity and causalityproblemsin orderto draw inferences on whethera firm's ownershipstructureaffects its value (Lemmon and Lins (2003)). Because I lack time-series data on ownershipstructures, I cannottest whetherchanges in ownershiprelate to changes in firm value.23 Instead,I model the endogeneitywithin a cross-sectionalframework(e.g., Demsetz andLehn (1985), LodererandMartin(1997), Cho (1998), Himmelberg,Hubbard, and Palia (1999), and Demsetz and Villalonga(2001)). Table5 reportsthe coefficientson two-stageleast squares(2SLS) regressions in which the structuralmodel contains the valuationequationand the first-stage model contains an ownershipequation. While a 2SLS estimationprocedureallows for endogeneitybetween Q and ownership, one shortcomingof this technique is thatit requiresthe identificationof some numberof exogenous variables thatplausiblyaffect only value or ownership,but not both. In selecting my equation specifications,I follow the models of Demsetz and Villalonga (2001) since these models incorporatea range of variablesused in prior2SLS ownershipand value studies. My models use managementcash flow rights leverageand the percentage of non-managementgroup control rights as the ownershipvariables of interest. I select these relatively simple ownershipvariablessince they need to be both related to valuation and plausibly determinedby a set of independent variables. The valuationequationhas Tobin's Q as the dependentvariable,an ownerof interest,and capitalexpendituresto assets, leverage, and country variable ship and industrydummies as controls. The specificationof my valuationequation matchesmy previouslyreportedOLS regressions,except that firm size is not included. The ownershipequationshave the ownershipvariablefrom the valuation equationas the dependentvariable,Tobin's Q as the simultaneouslydetermined variable,and controls. I include countrydummies and leverage as common controls across the valuationand ownershipequations. LLS (1999) show that ownership structuresdiffer substantiallyacross countries. Leverage is included in the ownershipequationsto reflect the possibility that creditorscan act as exter231 conductone(unreported) time-seriestestthatregressesTobin'sQ valuesforfiscalyear1996on andcontrols. variablesfor 1996ownership, variables,whichactas instrumental my 1995ownership andcontrolwhentheyexpecttheir of theirownership If managersareableto increasetheseparation firmto performpoorlyin the future,thenone wouldexpecta morenegativerelationbetweenpast andcurrentfirmvalue. I findthe oppositeresult-1996 firmvaluesare still ownershipseparation negatively and significantlyrelatedto 1995 managementcash flow rights leverage, but the magnitude of thenegativerelationis less severe. Lins 179 TABLE 5 of Least Estimation Two-Stage Ownershipand FirmValuein EmergingMarkets Squares Structural Model FirstStage Regression Structural Model FirstStage Regression DependentVariable Mgmt.Cash FlowRightsLeverage Tobin's0 Mgmt.Cash FlowRights Leverage Tobin's0 Non-Mgmt. GroupControl Rights% (1) (2) (3) (4) -0.462 (0.01) Non-Mgmt.GroupControlRights% Mgmt.GroupControlRights% to Assets CapitalExpenditures LeverageRatio CountryDummies IndustryDummies Tobin's0 0.262 (0.52) -0.466 (0.09) Yes Yes Logof Assets -0.131 (0.83) Yes No 0.005 (0.99) 0.101 (0.13) 0.245 (0.05) 0.016 (0.86) -0.015 (0.94) -0.263 (0.12) Yes Yes 0.771 (0.81) Yes No 0.404 (0.00) 0.093 (0.15) Beta -0.182 -0.090 Alpha (0.18) -0.142 (0.01) 830 (0.61) -0.058 (0.45) 1057 No. of Obs. 830 1057 Two-stageleast squares analysis of the dependent variableTobin's0 on categories of managementownership,nonmanagementownership,and controls.Alphaand Betavalues are obtainedfromWorldscope.These are computed(by Worldscope)usingbetween23 and 35 consecutivemonth-endpercentagepricechanges relativeto a localmarketindex. Allothervariablesare describedinTables1 and 2. Industrydummyvariablesare based on Campbell(1996). Thep-value of the t-testof equalityof each coefficientto zero is reportedin parentheses. nal monitors,which might affect the likelihoodof observingownershipstructures that facilitate managerialentrenchment.Firm size is in the ownershipequation to control for the possibility that managersof large firms will use pyramidsto obtain their controlrights in orderto conserve on cash or that non-management blockholderspreferownershippositions in large firms. I also include Alpha and Beta values from Worldscope,which are computed using between 23 and 35 consecutive month-endpercentageprice changes relative to a local marketindex. To the extent that Alpha, which measures past "excess"returns,is a proxy for futureexpected excess returns,higherAlpha values should increase the willingness of both managersand non-managersto hold cash flow rightsin a firm. As such, Alpha shouldbe negativelyrelatedto management cash flow rights leverage. Demsetz and Villalonga(2001) arguethathigher marketrisk (Beta) indicates betterprospects for managersto profit from inside informationand for outside shareholdersto engage in profitablemonitoringof managers. Thus, Beta should be positively relatedto the cash flow rightsblockholdings of both managersand outsidersand negatively relatedto management cash flow rightsleverage.Inclusionof Alpha andBeta reducesthe sample size by about25%. 180 Journalof Financialand QuantitativeAnalysis The Table 5 2SLS regressionsprovide mixed evidence on the causality of results. The first valuation equation (equation (1)) shows that, conOLS my for trolling simultaneity,firm values are still significantlylower when managementhas controlin excess of its proportionalownership.The ownershipequation (equation(2)) indicatesthata firm'sQ value does not influencemanagementcash flow rights leverage. Thus, these results are consistentwith an interpretationthat managementcash flow rights leverage reduces firm value. Equation(3) shows, as before, that firm values are higher as the control rights of non-management blockholdersincrease. However,equation(4) indicatesthat there is simultaneity between Q and non-managementownership,with non-managementblockholders more likely to own controlrightsin firmswith higherQ values. Forthis reason,it is best to interpretthe resultsof the OLS regressionmodels with non-management blockholdersas demonstratinga relation,but not necessarilycausation,between non-managementblockholdersand firmvalue. B. Measurement Issues The process of constructingultimate ownershipdata for emerging market firms requiresdata sources that capturethe full breadthof any overlap among family members,othercompanies,and otherinstitutions.I follow the convention of La Portaet al. (1999) by matchingmanagersand family based on family surname, but this match will obviously be imperfectwhen family membersdo not sharethe same surname.Similarly,it is plausible that I omittedfirms due to my inability to obtain data on their direct blockholdingswhen such data do, in fact, exist. It is not clear whetherthese potentialmisclassificationswill cause a bias in the sample or will simply add noise to the ownershipmeasures. Forrobustness,I comparemy datawith the ownershipstructuredatafor East Asian firms used in the Claessens et al. (2000) study and for Portuguesefirms used in the Faccio and Lang (2002) study.24I find a very strong and significant correlation,but not a perfect correlation. To see if these differences affect my results, I replace my ownershipvalues with those of Claessens et al. and Faccio and Lang where they are different, and re-estimatemy models. I find that all of my results are similarin both magnitudeand significancewhen I incorporate datacompiled by these otherauthors.As such, my results appearto be robustto measurementissues relatingto ownershipstructureclassification. I next conductrobustnesstests using two alternatemeasuresof firm performance as dependentvariables, since the suitability of Tobin's Q as a measure of firm value may be compromisedby differencesin accountingpracticesor reportingacross the countriesin my sample. I use the market-to-bookequity ratio, defined as the marketvalue of equity divided by the book value of equity, and operatingreturnon assets, definedas operatingincome deflatedby assets. These performancemeasuresare also censored at the first and 99th percentilesto alleviate the influence of outliers. Overall, I find that all of the previous ownership 24Uniquecoding strategiesused during data collection limit my ability to fully incorporatesome of these data. Specifically, I cannot use the Claessens et al. data in my tests of managementgroup control between 5% and 20% because they code the control rights of the second throughfifth largest blockholdersin incrementsof five percentagepoints, roundingdown (my variablesuse one percentage point increments). Lins 181 and value relations hold using the market-to-bookequity value measure, while some, but not all, of these relationshold when I specify operatingreturnas the performancemetric. I also controlfor the possibility that majority-controlledfirms, in which no takeoveris possible, might accountfor some of my findings since managersare, by far, the largestblockholdersin emergingmarkets. I re-estimatemy previous models, excludingfirmswhen a single blockholderholds a majorityof the firm's control rights, and find that all results continue to hold. Finally, I performall robustnesstests using the low shareholderprotectionsubsampleand find thatmy previousresults still obtain. V. Conclusion This paper investigates the relation between ownership structureand firm value across 1433 firmsfrom 18 emergingmarkets.I departfrom previouscrosscountryresearch on ownershipand valuationby explicitly examining management and family ownershipacross all of my sample firmsand whetherlargenonmanagementblockholdersprovidemonitoring.I also investigatewhetherthe relation between ownershipand value dependsuponthe level of externalshareholder protectionin a country. This papercontainsseveralinterestingresults. First,I find thatmanagement groupcontrolin excess of its proportionalownershipis negativelyrelatedto Tobin's Q in emergingmarkets.Managerialcontrolin the 5% to 20% rangeis also negatively relatedto Q. These results indicatethat investorsdiscountfirms with potentiallysevere managerialagency problemsresultingfrom misalignedincentives and managerialentrenchment.Second, I provide evidence that large nonmanagementblockholders can mitigate the valuation discount associated with these expected agency problems. Managerialcontrol in the 5% to 20% range is only associated with lower firm values when the managementgroup is also the largest blockholder.When a largernon-managementblockholderis present, managerialcontrol in the 5% to 20% range does not affect firm value. Regressions also show that large non-managementblockholdingsare positively related to Tobin'sQ values. Next, I examine whetherthe relationbetween ownershipand value depends on the level of shareholderprotectionin a country. When managershave control rights that exceed theirproportionalownership,firm values are significantly lower in countrieswith low shareholderprotection. I also find that the relation betweenlargenon-managementblockholdersandvalue is significantlymorepositive in low protectioncountries.These findingssuggest thatexternalshareholder protectionmechanismsplay a role in restrainingmanagerialagency costs. They also indicatethatlarge non-managementblockholdersmay act as a substitutefor missing institutionalgovernancemechanisms. Interestingtopics for futureownershipstructureresearchinclude identifying the factors that drive the presence of large non-managementblockholdersand studyingwhy LatinAmericanfirmsuse non-votingequity structuresmuch more frequentlythando otheremergingmarketfirms. 182 Journal of Financial and Quantitative Analysis APPENDIX Sources of ControlRightsand Cash FlowRightsOwnershipData PanelA. OverallDataSources Used FrequentlyforAllCountries UltimateOwnershipData Lexis-Nexis,Worldscope,ExtelCards,GlobalData Direct(FISonline),HoppenstedtAffiliationsand CorporateStructures,Nelson'sPublicCompanyProfiles,Hoover'sCompany German Capsules, ELCLargestCompanies,ICCFinancialAnalysisreports,Creditreform Swiss Companies,WhoOwnsWhom,company-specificwebCompanies,Creditreform sites Non-Voting EquityShares Datastream, Worldscope, GlobalDataDirect PanelB. DirectOwnershipDataSourcesand AdditionalUltimateOwnershipDataSourcesper Country UltimateOwnershipData(and non-votingsharedata) DirectOwnershipData Country ArgentinaCH; BrazilCH; LatinAmericanCompaniesHandbook, 1995, Moody'sInvestorsService, NY, NY;Hoover'sMaster List of LatinAmericanCompanies 1996-1997, The Reference Press, Austin,TX BrazilCH;ArgentinaCH; LatinAmericanCompaniesHandbook, 1995; Hoover'sMasterList of LatinAmericanCompanies 19961997 ArgentinaCH; BrazilCH; LatinAmericanCompaniesHandbook, 1995; Hoover'sMasterList of LatinAmericanCompanies 19961997 Czech StockMarketGuide;WorldBankdatafiles Argentina ArgentinaCompanyHandbook (CH),1997, Hoover'sInc.,Austin, TX Brazil BrazilCompanyHandbook,1997, Hoover'sInc.,Austin,TX Chile Worldscope Czech Rep. Czech StockMarketGuide,1996, AspektKilcullens.r.o.,Prague, Czech Republic ThorntonGuide to Asian Compa- ThorntonGuide;AsianCH;JapanCompanyHandbook,1997,Toyo nies, 1996, EdinburghFinancial KeizaiInc.,Tokyo,Japan PublishingAsia, HongKong ThorntonGuide;AsianCH;JapanCH IndonesianCapital MarketDirec- IndonesianDirectory; tory, 1993, Institutefor Economic and Financial Research, Jakarta StockExchange,Jakarta GlobalData Direct,FinancialInformationServicesOnline Thornton Guideto AsianCompanies ThorntonGuide;AsianCH;JapanCH GlobalDataDirect ArgentinaCH; BrazilCH; LatinAmericanCompaniesHandbook, 1995; Hoover'sMasterListof LatinAmericanCompanies 19961997 InvestmentsGuide,1997, Research InvestmentsGuide-PhilippineStock Exchange,ThorntonGuide; Department,PhilippineStock Ex- AsianCH;JapanCH change, Manila HongKong Indonesia Israel Malaysia Peru Philippines Portugal Singapore So. Africa So. Korea SriLanka Taiwan Thailand Worldscope Thornton Guideto AsianCompanies ThorntonGuide;AsianCH;JapanCH GlobalDataDirect ThorntonGuide;AsianCH;JapanCH Worldscope GlobalDataDirect Thornton Guideto AsianCompanies ThorntonGuide;AsianCH;JapanCH Thornton Guideto AsianCompanies ThorntonGuide;AsianCH;JapanCH;MFCInvestmentHandbook, 1996, MutualFundPublicCompanyLimited,Bangkok Turkey Worldscope References Asian CompanyHandbook.Tokyo,Japan:Toyo Keizai Inc. (1997). Backman,M. Asian Eclipse: Exposing the Dark Side of Business in Asia. Singapore:JohnWiley and Sons (Asia) Pte Ltd. (1999). Brazil CompanyHandbook.Austin, TX: Hoover's Inc. (1997) Bebchuk, L.; R. Kraakman;and G. Triantis. "Stock Pyramids, Cross-ownership,and Dual Class Equity: The Creationand Agency Costs of SeparatingControlfrom Cash Flow Rights." In ConcentratedCorporateOwnership,R. K. Morck,ed. Chicago, IL: Univ. of Chicago Press (2000). Berle, A., and G. Means. TheModem Corporationand Private Property.New York,NY: MacMillan (1932). Lins 183 Campbell,J. "UnderstandingRisk and Return."Journalof Political Economy,104 (1996), 298-345. Cho, M.-H. "OwnershipStructure,Investment,and the CorporateValue: An Empirical Analysis." Journal of Financial Economics,47 (1998), 103-121. Chui, A. C. W.; S. Titman;and K. C. J. Wei. "Momentum,Legal Systems and OwnershipStructure: An Analysis of Asian Stock Markets."WorkingPaper,Hong Kong Polytechnic Univ. (Dec. 2000). Claessens, S. "CorporateGovernanceand Equity Prices: Evidence from the Czech and Slovak Republics."Journal of Finance, 52 (1997), 1641-1658. Claessens, S., and S. Djankov,"EnterprisePerformanceand ManagementTurnoverin the Czech Republic."EuropeanEconomicReview,43 (1999), 1115-1124. Claessens, S.; S. Djankov;J. P. H. Fan;and L. H. P. Lang. "Disentanglingthe Incentiveand Entrenchment Effects of Large Shareholdings."Journalof Finance, 57 (2002), 2741-277 1. Claessens, S.; S. Djankov; and L. H. P. Lang. "The Separationof Ownershipand Control in East Asian Corporations."Journalof Financial Economics,58 (2000), 81-112. Denis, D. J., and D. K. Denis. "MajorityOwner-Managersand OrganizationalEfficiency."Journalof CorporateFinance, 1 (1994), 91-118. Denis, D. K., and J. J. McConnell. "InternationalCorporateGovernance."Journal of Financial and QuantitativeAnalysis, 38 (2003), 1-36. Demirgiiq-Kunt,A., and V. Maksimovic. "Institutions,Financial Markets,and Firm Debt Maturity." Journal of Financial Economics, 54 (1999), 295-336. Demsetz, H., and K. Lehn. "The Structureof CorporateOwnership: Causes and Consequences." Journal of Political Economy,93 (1985), 1155-1177. Demsetz, H., and B. Villalonga. "OwnershipStructureand CorporatePerformance."Journal of Corporate Finance, 7 (2001), 209-233. TheEconomistIntelligence Unit, London,England:CountryReports(1996-1998). Faccio, M., andL. H. P.Lang. "TheUltimateOwnershipof WesternEuropeanCorporations."Journal of Financial Economics 65 (2002), 365-395. Greene,W. H. EconometricAnalysis. Englewood Cliffs, NJ: Prentice-Hall,Inc. (1997). Harvey,C. R.; K. V. Lins; and A. H. Roper. "TheEffect of CapitalStructurewhen Expected Agency Costs are Extreme."WorkingPaper,Univ. of Utah (2002). Hermalin,B., and M. Weisbach. "TheEffects of Board Compensationand Direct Incentiveson Firm Performance."Financial Management,20 (1991), 101-112. Himmelberg,C. P.; R. G. Hubbard;and D. Palia. "Understandingthe Determinantsof Managerial Ownershipand the Link between Ownershipand Performance."Journal of Financial Economics, 53 (1999), 353-384. Holderness,C. G.; R. S. Kroszner;and D. P.Sheehan. "Werethe Good Old Days ThatGood? Changes in ManagerialStock Ownershipsince the GreatDepression."Journal of Finance, 54 (1999), 435470. Holderness,C. G., and D. P. Sheehan. "TheRole of MajorityShareholdersin Publicly Held Corporations: An ExploratoryAnalysis."Journalof Financial Economics, 20 (1988), 317-346. Jensen, M. C., and W. H. Meckling. "Theoryof the Firm: ManagerialBehavior, Agency Costs and OwnershipStructure."Journalof Financial Economics,3 (1976), 305-360. Kole, S. "ManagerialOwnershipand Firm Performance:Incentives or Rewards?"Advances in Financial Economics, 2 (1996), 119-149. La Porta,R.; F Lopez-de-Silanes;and A. Shleifer. "CorporateOwnershiparoundthe World."Journal of Finance, 54 (1999), 471-517. La Porta, R.; F Lopez-de-Silanes; A. Shleifer; and R. W. Vishny. "Legal Determinantsof External Finance."Journalof Finance, 52 (1997) 1131-1150. _ "Lawand Finance."Journal of Political Economy,106 (1998), 1113-1155. "InvestorProtectionand CorporateGovernance."Journal of Financial Economics, 58 (2000), 3-28. . "InvestorProtectionand CorporateValuation."Journalof Finance, 57 (2002), 1147-1170. Lemmon,M., and K. Lins. "OwnershipStructure,CorporateGovernance,and Firm Value: Evidence from the East Asian FinancialCrisis."Journal of Finance (forthcoming2003). Lins, K., and H. Servaes. "InternationalEvidence on the Value of CorporateDiversification."Journal of Finance, 54 (1999), 2215-2239. "Is CorporateDiversificationBeneficial in Emerging Markets?" Financial Management,31(2002), 1-23. Loderer,C., and K. Martin. "ExecutiveStock Ownershipand Performance:TrackingFaint Traces." Journal of Financial Economics,45 (1997), 223-255. McConnell, J. J., and H. Servaes. "AdditionalEvidence on Equity Ownershipand CorporateValue." Journal of Financial Economics, 27 (1990), 595-612. 184 Journalof Financialand QuantitativeAnalysis "EquityOwnershipand the Two Faces of Debt." Journal of Financial Economics, 39 (1995), 131-157. Morck, R.; A. Shleifer; and R. W. Vishny. "ManagementOwnership and Market Valuation: An EmpiricalAnalysis."Journalof Financial Economics,20 (1988), 293-315. Morck, R.; D. Stangeland; and B. Yeung. "InheritedWealth, CorporateControl and Economic Growth: The Canadian Disease?" In Concentrated Corporate Ownership, R. K. Morck, ed. Chicago, IL: Univ. of Chicago Press (2000). Nenova, T. "TheValue of CorporateVotes and ControlBenefits: Cross-countryAnalysis."Journal of Financial Economics (forthcoming2003). Rouwenhorst,K. G. "Local ReturnFactors and Turnoverin Emerging Stock Markets." Journal of Finance, 55 (1999), 1439-1464. Shleifer, A., and R. W. Vishny. "A Survey of CorporateGovernance."Journal of Finance, 52 (1997), 737-783. Stulz, R. M. "ManagerialControl of Voting Rights." Journal of Financial Economics, 20 (1988), 25-59. ThorntonGuide to Asian Companies.Hong Kong: EdinburghFinancialPublishingAsia (1996). Tian, G. L. "GovernmentShareholdingand the Value of China's Modern Firms." WorkingPaper, William Davidson Instituteat the Univ. of Michigan (April 2001). Xu, X., and Y. Wang,"OwnershipStructure,CorporateGovernance,and CorporatePerformance:The Case of Chinese Stock Companies."WorldBank Policy ResearchWorkingPaper 1794, Washington, D.C. (1997). Zenner,M. "Surveyof Stock Exchange Characteristicsand Requirements."Unpubl. Paper,Univ. of North Carolina(1995). Zingales, L. "The Value of the Voting Right: A Study of the Milan Stock Exchange Experience." Review of Financial Studies,7 (1994), 125-148.