Corporate Governance, Economic Entrenchment, and Growth€¦ · Corporate Governance, Economic...

Journal of Economic LiteratureVol. XLIII (September 2005), pp. 655–720

Corporate Governance, EconomicEntrenchment, and Growth

RANDALL MORCK, DANIEL WOLFENZON, and BERNARD YEUNG∗

Outside the United States and the United Kingdom, large corporations usually havecontrolling owners, who are usually very wealthy families. Pyramidal control struc-tures, cross shareholding, and super-voting rights let such families control corporationswithout making a commensurate capital investment. In many countries, a few suchfamilies end up controlling considerable proportions of their countries’ economies.Three points emerge. First, at the firm level, these ownership structures, because theyvest dominant control rights with families who often have little real capital invested,permit a range of agency problems and hence resource misallocation. If a few familiescontrol large swaths of an economy, such corporate governance problems can attainmacroeconomic importance—affecting rates of innovation, economywide resource allo-cation, and economic growth. If political influence depends on what one controls, ratherthan what one owns, the controlling owners of pyramids have greatly amplified politi-cal influence relative to their actual wealth. This influence can distort public policyregarding property rights protection, capital markets, and other institutions. We denotethis phenomenon economic entrenchment, and posit a relationship between the distri-bution of corporate control and institutional development that generates and preserveseconomic entrenchment as one possible equilibrium. The literature suggests key deter-minants of economic entrenchment, but has many gaps where further work exploringthe political economy importance of the distribution of corporate control is needed.

655

∗ Morck: University of Alberta and National Bureau ofEconomic Research. Wolfenzon: Stern School of Business,New York University, and National Bureau of EconomicResearch. Yeung: Stern School of Business, New YorkUniversity. We gratefully acknowledge helpful suggestionsfrom Mark Casson, Ron Daniels, Mara Faccio, RogerGordon, Henry Hansmann, Campbell Harvey, DianaKniazeva, Bruce Kogut, Ross Levine, Enrico Perotti,Andrei Shleifer, René Stulz, Paul Vaaler, Belén Villalonga,Marina Whitman, and the referees.

1. Introduction

Agrowing body of work indicates thateconomic growth depends on the distri-bution of control over capital assets. Twovast literatures, already surveyed in the

Journal of Economic Literature, are relevantas points of departure. The first, reviewedby Philippe Aghion, Eve Caroli, and CeciliaGarcia-Penalosa (1999), discusses reasonswhy economic inequality might impedegrowth. The second, surveyed by RossLevine (1997), discusses the functional roleof capital markets in stimulating growth. Webuild on these insightful contributions byreviewing recent research relating the dis-tribution of corporate control with theworkings of capital markets and showinghow these interactions affect economicgrowth.

sp05_Article 1 8/18/05 9:32 AM Page 655

656 Journal of Economic Literature, Vol. XLIII (September 2005)

The key insight in Aghion, Caroli, andGarcia-Penalosa (1999) is that perfect capitalmarkets make wealth distribution irrelevantby allocating capital to each investmentopportunity until its marginal return equalsthe market clearing equilibrium interestrate. Simple capital market frictions, likemoral hazard problems, drive a wedgebetween internal and external cost of capital.This can lead the wealthy to invest morethan is first best optimal, while the impecu-nious invest less. Assuming diminishing ratesof return, redistributing wealth from the richto the poor directly raises total output.Moreover, wealth redistribution from rich topoor lessens overall borrowing in the econo-my, easing these moral hazard problems,raising overall production, and reducing thecost of debt financing on the margin(Aghion, Caroli, and Garcia-Penalosa 1999).

Levine (1997) surveys a second literatureon the functions of capital markets in eco-nomic growth—the organization of informa-tion acquisition, allocation of resources,monitoring of managers, and exercise of cor-porate control. If financial markets are largeand liquid enough, and if investors’ propertyrights are sufficiently well protected, individ-uals expend resources to acquire informationand conduct profitable informed risk arbi-trage, as modeled by Sanford J. Grossmanand Joseph E. Stiglitz (1980) and AndreiShleifer and Robert W. Vishny (1997), andthis trading capitalizes information into stockprices as in Morck, Yeung, and Wayne Yu(2000). Richard Roll (1988) argues that thismechanism is especially important for thetimely updating of stock prices with newfirm-specific information. Stock prices thatclosely track fundamental values aid outsideinvestors and corporate insiders in capitalallocation decisions, simplify monitoring, andfacilitate the design of credible incentive con-tracts that induce managers to keep faith withoutside investors and maximize firm value.

These two literatures converge on a cen-tral theme: economic growth requires thatsavings be directed into value creating

investments. Raghuram G. Rajan and LuigiZingales (2003), Art Durnev, Morck, andYeung (2001, 2004), and others argue thatcapital markets are inimitably, though cer-tainly not perfectly, effective in this regard.This is because, by creating incentives forself-interested investors to gather andprocess information, as well as to monitorand control top corporate insiders, capitalmarkets generate conditions amenable toefficient capital investment decisions.Corporate insiders needing outside capitalmust submit to this analysis, monitoring, andcontrol by outside investors. When officialregulations, laws, and enforcement makeoutside investors sufficiently confident oftheir ability to undertake these tasks, outsideinvestors hold equity. Pooling their capitalunder the management of institutionalinvestors allows small investors to attaineconomies of scale in accomplishing thesetasks. The resulting optimally efficient finan-cial market directs capital to its highest valueuses—a situation James Tobin (1982)defines as functional form stock market effi-ciency. Note that functional form efficiencydiffers from the standard usage of the termmarket efficiency in the finance literature,which describes a situation where investorscannot obtain abnormal returns after trans-actions costs. The two concepts are related,but not identical.

Economists have little difficulty listingfracture points where actual capital marketsfail to satisfy functional form efficiency. Theliterature we survey below suggests that thefunctional efficiency of capital marketsdepends on the distribution of corporatecontrol in an economy. In particular, this lit-erature views economic growth as criticallydependent on institutions that restrainentrenched elites, who could otherwisecome to dominate the capital investmentdecisions of an economy.

This literature has its recent roots in cor-porate finance, perhaps because the extremeconcentration of corporate control rights inthe hands of tiny elites observed in many


Morck, Wolfenzon, and Yeung: Corporate Governance 657

countries raises concerns about corporategovernance in those economies. The corpo-rate finance literature also considersentrenched management to be a corporategovernance issue at the firm level in waysthat are helpful in understandingentrenched control in general.

Our starting point is a curious empiricalobservation reported by Morck, DavidStangeland, and Yeung (2000). Theseauthors divide up the world’s U.S. dollar bil-lionaires into two categories—self-made bil-lionaires and billionaires who inherited theirwealth—and sum up the wealth owned byeach category of billionaire in each country.Unsurprisingly, they find that a country’s percapita GDP grows faster if its self-made bil-lionaire wealth is larger as a fraction of GDP.What is more surprising is that per capitaGDP growth is slower in countries whereinherited billionaire wealth is larger as afraction of GDP.

The literature we review below points topossible explanations. These turn on wealthyindividuals who magnify their already sub-stantial wealth into control over multiplecorporations worth vastly more. Rafael LaPorta, Florencio Lopez-de-Silanes, andShleifer (1999) show that this magnificationis most commonly achieved using controlpyramids: structures in which a family firmcontrols several listed companies, each ofwhich controls yet more listed companies,each of which controls yet more listed com-panies, and so on. Family members are usu-ally placed as executives of key firmsthroughout the structure. Other less com-mon devices also used to this end includesuperior voting shares and crossholdings.Superior voting shares, distinct classes ofstock with many more votes per share thanordinary common stock, allow insiders tocontrol the majority of votes in shareholdermeetings even though they own only a smallfraction of the firm’s equity. Crossholdingsare structures in which firms own blocks ofeach others’ stocks. Since insiders vote theseblocks, they exercise control beyond their

actual ownership. These sorts of structurescan let a few wealthy families control thegreater part of a country’s large corporatesector. They also leave the structure typicalof large U.S. firms—stand alone firms withdiffuse ownership and professional manage-ment—the rarest of curiosities in most of therest of the world.

This concentrated control can lead to cor-porate governance concerns—a range ofagency problems. But, more importantly,entrusting the governance of huge slices ofa country’s corporate sector to a tiny elitecan bias capital allocation, retard capitalmarket development, obstruct entry by out-sider entrepreneurs, and retard growth.Furthermore, to preserve their privilegedpositions under the status quo, such elitesmight invest in political connections tostymie the institutional development of cap-ital markets and to erect a variety of entrybarriers. These economywide implicationscan be most serious.

Such an outcome is a suboptimal politicaleconomy equilibrium, which we dub eco-nomic entrenchment. While wealthy estab-lished families are probably important toeconomic entrenchment in many countries,the problem is probably not restricted tocountries whose corporate sectors are con-trolled by small cliques of extremely wealthyfamilies. Other sorts of elites can alsobecome entrenched in positions of controlover corporate assets and behave much likeelite families, generating a similar out-come—functionally inefficient capital mar-kets, high barriers to entry, and a slow paceof innovation.

The layout of this article is as follows. First,we motivate this survey by discussing someempirical findings that suggest a linkbetween economic growth and, not the con-centration of wealth per se, but the “hands”in which control over corporate assets is con-centrated. Second, we review the literatureon corporate ownership, pyramidal groups,and family control. This literature shows thatmany economies entrust the governance of



large parts of their corporate sectors to tinyelites of extremely wealthy families. Third,we discuss issues associated with such con-centrated corporate governance and howthese might affect economic performance.Fourth, we evaluate the linkage betweenconcentrated corporate control on the onehand and overall capital allocation, capitalmarket development, creative destruction,and macroeconomic growth on the other.Fifth, and we think this is the most importantcomponent of the paper, we discussentrenchment as a political economy prob-lem and the determinants of economicentrenchment, which include investmentopportunities, societal tradition, and initialendowments. Also, we examine the relation-ship between economic entrenchment andopenness. Finally, we distill some generalconclusions out of the above.

2. Inherited Wealth and Growth—APreliminary Reading

Our starting point is a curious observa-tion by Morck, Stangeland, and Yeung(2000). Using Forbes’ listing of the onethousand wealthiest individuals in theworld for 1993, and the brief biographiesaccompanying each, they distinguish newmoney billionaires, entrepreneurs whomade their billions themselves, from oldmoney billionaires, who inherited theirwealth. They gauge the importance of eachtype of billionaire in a given country by thesum of the wealth of that type of billionaireas a fraction of the country’s GDP. Whenthey regress real per capita GDP growthfrom 1994 to 1996 on these measures, con-trolling for initial per capita GDP, physicalcapital accumulation, and education levels,they find new money billionaire wealth tobe associated with faster economic growth,but old money billionaire wealth to beassociated with slower growth. That is, instandard growth theory regressions of thesort used by Mankiw (1995), they report ahighly significant relationship between

growth and variables plausibly reflectingthe distribution of control over an economy’scapital assets.

Table 1 recapitulates these results. Thealternative specifications shown treat wealththat is not unambiguously controlled byeither a founder or an heir differently. H1includes only the wealth of billionaires whoare unambiguously heirs, politicians, orpoliticians’ relations. H2 also includes thewealth of billionaires who are probably heirs.H3 includes H1 plus fortunes jointly con-trolled by a founder and his heirs. H4includes all the above. H5 through H8 areanalogous to H1, H2, H3, and H4 but do notinclude politician billionaires and their rela-tions. The positive coefficients on newmoney billionaire wealth and the negativeone on old money billionaire wealth remainhighly significant across all the differentspecifications.

Previous work, surveyed by Aghion,Caroli, and Garcia-Penalosa (1999), dis-cusses the relationship between generalinequality and growth, but Morck,Stangeland, and Yeung (2000) add anothertwist—inequality involving new moneywealth seems different from inequalityinvolving old money wealth. Perhaps econ-omists need to think less about concentra-tion of wealth per se and more aboutconcentration of wealth in whose hands?Why might inequality associated withinherited wealth be fundamentally differentfrom inequality associated with entrepre-neurial wealth? To explore this, we mustfirst examine how highly concentratedwealth can translate into even more highlyconcentrated corporate governance power.

3. The Ubiquity and Purpose of the Control Pyramid

The extensive control wealthy familiesexert over the corporate sectors of most ofthe world’s economies depends on a particu-lar type of ownership structure, the controlpyramid, and other closely related structures.



TABLE 1THE CROSS-COUNTRY RELATIONSHIP BETWEEN GROWTH IN REAL PER CAPITA GDP AND CAPITAL OWNERSHIP,

CONTROLLING FOR CURRENT PER CAPITA INCOME, CAPITAL INVESTMENT RATE, AND LEVEL OF EDUCATION

1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8

Intercept 1.43 1.58 1.59 1.65 1.75 1.73 1.86 1.78

(0.32) (0.30) (0.27) (0.28) (0.22) (0.26) (0.20) (0.25)

Log of per capita –1.76 –1.77 –1.80 –1.79 –1.54 –1.66 –1.62 –1.69

GDP: ln(Y/L) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00)

Capital Accumulation 0.21 0.22 0.21 0.21 0.17 0.20 0.18 0.20

Rate: I/K (0.00) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00)

Average Total Years 00.23 00.20 00.25 00.21 00.24 00.20 00.26 00.21

of Education: ln(E) (0.27) (0.35) (0.23) (0.32) (0.24) (0.35) (0.21) (0.32)

Business Entrepreneur 0.44 0.37 0.42 0.37 0.50 0.38 0.45 0.37

Billionaire Wealth Over (0.00) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00)

GDP: B/Y

Heir Billionaire Wealth –0.29 –0.17 –0.27 –0.16 –0.41 –0.19 –0.33 –0.17

Over GDP: H/Y (0.03) (0.10) (0.03) (0.09) (0.01) (0.09) (0.01) (0.08)

Definition of “heir”1 H1 H2 H3 H4 H5 H6 H7 H8R squared 0.519 0.488 0.531 0.489 0.545 0.491 0.536 0.491

Note: Numbers in parenthesis are two tailed t-test probability levels for rejecting a zero coefficient. Coefficientsin boldface are statistically significant at 90 percent confidence or more. Sample of thirty-nine countries consistsof the countries listed in table 1 minus the United Kingdom and United States. See Morck, Stangeland, andYeung. (2000), table 2 for further details.

1 H1 includes only the wealth of billionaires known positively to be heirs, politicians or politicians= relations.H2 also includes the wealth of billionaires who are probably heirs. H3 includes H1 plus fortunes jointly controlledby a founder and his heirs. H4 includes all the above. H5 through H8 are analogous to H1, H2, H3, and H4 but donot include politician billionaires and their relations.

While such structures are commonplaceoutside of the United States, they were notwell known in the academic literature untilhighlighted in recent work by La Porta,Lopez-de-Silanes, and Shleifer (1999). Thatresearch inspired further empirical workconfirming that (1) most large corporationsaround the world have controlling ownersand (2) controlling owners use pyramidalstructures (and other mechanisms) to amasscontrol over not just a single firm, or even

just a few firms, but over large groups ofcorporations.

This section surveys empirical evidenceabout the structure of corporate controlaround the world. We first highlight theubiquity of concentrated family control out-side the United States. We then discuss con-trol pyramids in detail and show they let avery small number of wealthy individuals orfamilies leverage substantial wealth into con-trol over corporate assets worth vastly more.



1 Berle and Means (1932, chapter 5) actually describevarious forms of pyramiding, nonvoting shares, and votingtrusts. In their sample of the largest 200 firms in theUnited States in 1930, they find that 44 percent are wide-ly held (they call them “management controlled”) and 22percent are controlled through a “legal devise” (i.e., a pyra-mid). In any case, the finance literature has for the mostpart taken Berle and Means as demonstrating that largeU.S. firms are largely widely held.

While this section primarily emphasizesconcentrated control over large groups offirms using control pyramids, which is ourfocus, it also briefly mentions other relatedstructures.

3.1 Wealthy Families Control MostCompanies around the World

3.1.1 Corporations Have Dominant Owners

That large corporations have individualsand families as controlling shareholders con-tradicts the view of the firm popularly attrib-uted to Adolph A. Berle and Gardiner C.Means (1932), which implicitly or explicitlyunderlies much of modern economics andfinance.1 In this view, a corporation is wide-ly held, in that its ownership is dispersedacross a large number of small public share-holders, and freestanding, in that listed com-panies generally do not control other listedcompanies. Since there is no dominantowner, effective control resides in the handof the management team. In such corpora-tions, corporate governance is about mitigat-ing the divergence of interests, described byMichael C. Jensen and William H. Meckling(1976), between utility maximizing profes-sional managers and small public sharehold-ers who would like the value of their sharesto be maximized.

The modern U.S. large corporate sectorloosely approximates this archetype. Ofcourse, large blockholders and intercorpo-rate equity holdings occur even there.Shleifer and Vishny (1986), Morck, Shleifer,and Vishny (1988), Clifford G. Holdernessand Dennis P. Sheehan (1988), Ronald C.Anderson and David M. Reeb (2003b),Anderson, Sattar A. Mansi, and Reeb (2003),and others all find numerous instances of

large blockholders, though they are smallerand much less common in large U.S. firmsthan in most other countries. An importantdistinction, however, is that block holders inthe United States seldom control more thanone corporation. Thus, the Ford family con-trols Ford Motors, but not GM, IBM, and3M as well. Indeed, the term “family firm” inthe United States is often a synonym for“small firm.”

However, elsewhere in the world, the typ-ical large firm has a “controlling owner”—usually a wealthy family that controls it aspart of a large group of firms. The assump-tion of freestanding diffusely owned firms,typically justified with a cite to Berle andMeans (1932), is therefore of questionablegenerality. This was first pointed out in stud-ies of corporate governance in Germany andJapan, such as Stephen D. Prowse (1992),Erik Berglöf and Perotti (1994), and JeremyEdwards and Klaus Fischer (1994). In a sys-tematic investigation, La Porta, Lopez-de-Silanes, and Shleifer (1999) show that U.S.style corporate ownership is quite exception-al. In most other countries, even very largefirms have controlling shareholders, andthese are usually extremely wealthy families.

La Porta, Lopez-de-Silanes, and Shleifer(1999) investigate the ownership structuresof large corporations in twenty-sevenmainly developed economies. Their samplecontains the top twenty firms in each coun-try, ranked by market capitalization ofcommon equity at the end of 1995. Theyalso collect information on ten firms ofsimilar size in each country. These are theten smallest firms in each country withmarket capitalization of common equitygreater than U.S.$500 million at the end of1995. They allow for five types of ultimatecontrolling owners: (1) a family or an indi-vidual, (2) the state, (3) a widely heldfinancial institution such as a bank or aninsurance company, (4) a widely held cor-poration, and (5) miscellaneous, an entitysuch as a cooperative, a voting trust, or agroup with no single controlling investor.



2 Claessens, Djankov, and Lang (2000) find that inJapan a large proportion of firms are widely held. The highfigures for widely held firms in Japan are perhaps mislead-ing, in the sense that most large Japanese firms are organ-ized into keiretsu, groups of firms without controllingshareholders that collectively own control blocks in eachother. These firms are thus not really widely held in thesense of U.S. and U.K. firms, for public shareholders donot hold a majority of their stock.

La Porta, Lopez-de-Silanes, and Shleifer(1999) argue that stakes exceeding fifty per-cent are not necessary to lock in control inmost cases. This is because most small share-holders do not vote at annual meetings.Consequently, voting stakes in the ten totwenty percent range are generally sufficient.

They report widely held corporations pre-dominating among large firms only in theUnited States and United Kingdom.Elsewhere, particularly in countries withweak legal and regulatory protection forpublic shareholders, even the largest corpo-rations usually have controlling sharehold-ers—predominantly wealthy families andthe State. Using a 20 percent control thresh-old, they find that 36 percent of large corpo-rations on average are widely held. Thedistribution shifts further away from widelyheld firms when the sample is based on a tenpercent control threshold, or on a set ofsmaller firms, or on countries with poorershareholder rights protection. In a few coun-tries, notably Belgium and Germany, finan-cial institutions exercise extensive votingcontrol over other large publicly tradedcompanies.

Studies extending La Porta, Lopez-de-Silanes, and Shleifer (1999), notably StijnClaessens, Simeon Djankov, and Larry H. P.Lang (2000) on Asian countries, and MaraFaccio and Lang (2002) and Fabrizio Barcaand Marco Becht (2001) on European coun-tries, expand the number of firms examinedand sometimes the number of countries.2

They confirm the La Porta, Lopez-de-Silanes, and Shleifer (1999) findings: out-side of the United States and UnitedKingdom, a large proportion of large firmshas controlling shareholders.

3 The different studies in table 2 rely on very differentsamples of firms. For example, La Porta, Lopez-de-Silanes, and Shleifer (1999) examine the twenty largestlisted firms; Attig, Gadhoum, and Lang (2003) use the1,121 firms in the Toronto Stock Exchange; Claessens,Djankov, and Lang (2000) use 2,980 listed financial andnonfinancial firms; and Faccio and Lang (2002) study5,232 corporations in thirteen Western European coun-tries. These sample differences explain the occasionalwide differences between their estimates. Also, Claessens,Djankov, and Lang (2000, p. 95) defined control slightlydifferently by allowing a firm to have more than one sig-nificant owner. For example, if a family holds a 25 percentvoting stake and a widely held corporation holds a 10 per-cent stake, when defining control at the 10 percent thresh-old, they classify the firm as one-half controlled by eachtype of owner. At the 20 percent level, the firm is fullyfamily controlled.

3.1.2 Concentrated Control of Corporationis in the Hand of Wealthy Families

These controlling shareholders are gener-ally wealthy families. La Porta, Lopez-de-Silanes, and Shleifer (1999) report that,using a 20 percent control threshold, 30 per-cent of large firms are family-controlled inthe average country. This increases when asample of smaller firms is used or when thethreshold of control is 10 percent. Again,subsequent work, like Claessens, Djankov,and Lang (2000), Faccio and Lang (2002),and Najah Attig, Yoser Gadhoum, and Lang(2003), confirms these findings.

Diffuse corporate ownership of the sortgenerally associated with Berle and Means(1932) for the United States is thus highlyexceptional. Table 2 summarizes the findingsof these various studies, and clearly displaysthe rarity of widely held firms and the ubiq-uity of family control.3 Although table 2 con-tains a wealth of data, further work involvingadditional countries and clarifying the deter-minants of these differences would be veryuseful.

3.2 The Use of Control Pyramids

The previous subsections present evi-dence that large corporations throughoutmost of the world are not widely held andmore often than not are controlled by verywealthy families. This subsection outlines



TABLE 2WIDELY HELD FIRMS VERSUS FAMILY CONTROLLED FIRMS AS PERCENT OF

LARGE CORPORATIONS IN VARIOUS COUNTRIES1

Control inferred at Control inferred at10%2 20% Source

Widely Family Widely FamilyCountry held control held control

Argentina 0 65 0 65 La Porta, Lopez-de-Silanes, and Shleifer (1999)Australia 55 10 65 5 La Porta, Lopez-de-Silanes, and Shleifer (1999)Austria 5 15 5 15 La Porta, Lopez-de-Silanes, and Shleifer (1999)Belgium 0 50 5 50 La Porta, Lopez-de-Silanes, and Shleifer (1999)Canada 50 30 60 25 La Porta, Lopez-de-Silanes, and Shleifer (1999)

17.54 46.67 36.24 26.18 Attig, Gadhoum, and Lang (2003)Denmark 10 35 40 35 La Porta, Lopez-de-Silanes, and Shleifer (1999)Finland 15 10 35 10 La Porta, Lopez-de-Silanes, and Shleifer (1999)France 30 20 60 20 La Porta, Lopez-de-Silanes, and Shleifer (1999)

8.92 70.44 17.79 64.83 Faccio, Lang (2002)aGermany 35 10 50 10 La Porta, Lopez-de-Silanes, and Shleifer (1999)

5.61 71.64 14.02 64.62 Faccio and Lang (2002)Greece 5 65 10 50 La Porta, Lopez-de-Silanes, and Shleifer (1999)Hong Kong 10 70 10 70 La Porta, Lopez-de-Silanes, and Shleifer (1999)

0.6 64.7 7 66.7 Claessens, Djankov, and Lang (2000)Indonesia 0.6 68.6 5.1 71.5 Claessens, Djankov, and Lang (2000)Ireland 45 15 65 10 La Porta, Lopez-de-Silanes, and Shleifer (1999)Israel 5 50 5 50 La Porta, Lopez-de-Silanes, and Shleifer (1999)Italy 15 20 20 15 La Porta, Lopez-de-Silanes, and Shleifer (1999)

7.83 64.87 15.86 59.61 Faccio and Lang (2002)Japan 50 10 90 5 La Porta, Lopez-de-Silanes, and Shleifer (1999)

42 13.1 79.8 9.7 Claessens, Djankov, and Lang (2000)Korea 40 35 55 20 La Porta, Lopez-de-Silanes, and Shleifer (1999)

14.3 67.9 43.2 48.4 Claessens, Djankov, and Lang (2000)Malaysia 1 57.5 10.3 67.2 Claessens, Djankov, and Lang (2000)Mexico 0 100 0 100 La Porta, Lopez-de-Silanes, and Shleifer (1999)Netherlands 30 20 30 20 La Porta, Lopez-de-Silanes, and Shleifer (1999)New Zealand 5 45 30 25 La Porta, Lopez-de-Silanes, and Shleifer (1999)Norway 5 25 25 25 La Porta, Lopez-de-Silanes, and Shleifer (1999)Philippines 1.7 42.1 19.2 44.6 Claessens, Djankov, and Lang (2000)Portugal 0 50 10 45 La Porta, Lopez-de-Silanes, and Shleifer (1999)Singapore 5 45 15 30 La Porta, Lopez-de-Silanes, and Shleifer (1999)

1.4 52 5.4 55.4 Claessens, Djankov, and Lang (2000)Spain 15 25 35 15 La Porta, Lopez-de-Silanes, and Shleifer (1999)

12.74 67.33 28.06 55.79 Faccio and Lang (2002)Sweden 0 55 25 45 La Porta, Lopez-de-Silanes, and Shleifer (1999)Switzerland 50 40 60 30 La Porta, Lopez-de-Silanes, and Shleifer (1999)Taiwan (China) 2.9 65.6 26.2 48.2 Claessens, Djankov, and Lang (2000)Thailand 2.2 56.5 6.6 61.6 Claessens, Djankov, and Lang (2000)United Kingdom 90 5 100 0 La Porta, Lopez-de-Silanes, and Shleifer (1999)

27.06 33.75 69.08 19.88 Faccio and Lang (2002)United States 80 20 80 20 La Porta, Lopez-de-Silanes, and Shleifer (1999)

38.95 23.37 69.26 6.13 Attig, Gadhoum, and Lang (2003)1 The figures from La Porta, Lopez-de-Silanes, and Shleifer (1999) are for the twenty largest publicly listed

firms. The Attig, Gadhoum, and Lang (2003) sample is comprised of the 1,121 firms in the Toronto StockExchange for which they could find ownership data. Faccio and Lang (2002) and Claessens, Djankov, and Lang(2000) also construct samples of all publicly traded firm for which ownership information is available.

2 The published version of Faccio and Lang (2002) does not have tables with the 10 percent cutoff. We reportstatistics in their earlier unpublished draft.



Figure 1. A Stylized Control Pyramid

, . ,

. .

> 50%

> 50% < 50%

< 50% > 50%

> 50% < 50%

< 50% > 50%

> 50% < 50%

< 50% > 50%

< 50%

> 50%

> 50% < 50%

< 50% > 50%

> 50% < 50%

< 50% > 50%

> 50% < 50%

< 50% > 50%

< 50%

Firm3,8

Family Firm

Firm2,1

Firm2,2

Firm2,3

Firm2,4

Public Shareholders

Firm1,1

Firm1,2

Firm3,1

Firm3,2

Firm3,3

Firm3,4

Firm3,5

Firm3,6

Firm3,7

A family firm controls a first tier of firms with dominant voting stakes, in this casegreater than fifty percent. Each first tier firm controls several second tier firms, each ofwhich controls yet more firms. The overall effect is to extend the family’s control toencompass assets worth substantially more than its actual wealth.

how these families achieve such concentrat-ed corporate control. The impression is thatthe most important mechanism families useis the control pyramid, sometimes aug-mented by multiple voting shares andcrossholdings, although the literature hasnot generated estimates on the relative significance of these means.

Control pyramids let these families con-trol not just one large firm, but many largefirms collectively worth substantially more

than the family’s actual wealth. Figure 1illustrates a simplified control pyramid. Afamily firm, which is a holding company inthis simplified example, holds a majoritystake, fifty percent plus one vote, in each ofa first tier of firms, in this case Firm1,1 andFirm1,2. The remaining fifty percent minusone vote stakes in each are held by smallpublic shareholders. Each of the first tierfirms then holds fifty percent plus one votein a set of second tier firms. In this case,



Firm1,1 holds control blocks in Firm2,1 andFirm2,2 while Firm1,2 holds control blocks inFirm2,3 and Firm2,4. Again, the remainingfifty percent, less one vote, in each tier twofirm is held by public investors. Each secondtier firm then controls a set of third tier firms,and so on, always with public shareholdersholding the remaining stakes.

Suppose each firm in figure 1 is worth onebillion dollars. The control pyramid lets thefamily, through its one billion dollar familyfirm, control fourteen other firms, also eachworth one billion dollars. There is doublecounting in this tally since, in this simplifiedexample, the assets of the higher tier firmsare shares of lower tier firms. But even ifonly the third tier firms contain actual phys-ical assets, the family has, at a minimum,leveraged one billion dollars of family wealthinto control over eight billion dollars of realcorporate assets. Contrast this with creatingeight freestanding one billion dollar firms ora single eight billion dollar firm. In eithercase, the family contributes only one-eighthof the equity and controls only 12.5 percentof the votes. In contrast, the pyramidal struc-ture in figure 1 lets the family retain absolutecontrol of the eight firms in the third layer ofthe pyramid, but hold only a 12.5 percentcash flow stake [(50%)(50%)(50%) = 12.5%].

To see this more generally, take the chainof ownership from the family apex to anyfirm in the third layer, say firm Firm3,1.Suppose the family holds α 01,1 of Firm1,1which holds α 1,12,1 of Firm2,1 which, in turn,holds α 2,13,1 of Firm3,1. The superscript, e.g.,(1,1), indicates first the “controlling” firm’slevel in the pyramid and its identity index atthe level. The subscript, e.g., (2,1), indicatesthe “controlled” firm’s level in the pyramidand its identity index at that level. If, at eachlink in the chain of control, the direct stakeis sufficiently large to control the next firm,the family effectively controls Firm3,1 whileholding stock that entitles it to onlyα 01,1 × α 1,12,1 × α 2,13,1 of the firm’s dividends. Thisformula can be straightforwardly expandedwhen there are more layers connecting the

family with the firm. Since the α ’s are all lessthan one, this product can be a very smallnumber. Nonetheless, the family controlsevery firm in the chain of firms connectingthe family firm to Firm3,1 and so controlsthat firm too. Ultimately, the family can con-trol many firms in this way even though itsactual investment in each is small.

Actual pyramids are more complicatedthan the diagram in figure 1. Firms may havevarying numbers of subsidiaries, and somefirms in intermediate tiers of the pyramidmay engage in actual production activities.Privately held firms may be interspersedamong the publicly traded firms of the pyra-mid. Firms also need not be neatly arrangedin successive tiers. Firms in lower tiers mayhave cross-shareholdings. That is, they mayown shares in firms in their own tiers or inupper tiers.

Cross shareholdings, differential votingshares, and public shareholders’ low partici-pation rates in corporate votes generally allowcontrol through chains of even smaller α ’s,resulting in even more extreme divergence ofownership from control with each additionallayer. Since family controlled boards votecross holdings, these augment direct familyvoting power. Super-voting shares also allowcontrol with smaller values of α . Returning toour example in figure 1, if super-voting shareslet the family exercise control with ten per-cent at each layer, rather than fifty percent,the family’s stake in the third layer would be(10%)(10%)(10%) = 0.1% instead of the 12.5percent calculated above. Since voting ratesby public shareholders are generally quitelow, a ten to twenty percent stake is usuallysufficient to control board elections, so thelast calculation is often relevant even in theabsence of multiple voting shares.

By way of illustration, we use an examplein La Porta, Lopez-de-Silanes, and Shleifer(1999), although other examples are amplyavailable, e.g., in Morck, Stangeland, andYeung (2000). La Porta, Lopez-de-Silanes,and Shleifer (1999, figure 8) describe theABB pyramidal structure which we replicate



Source: La Porta, Lopez-de-Silanes, and Shleifer (1999)

Figure 2. The ABB Control Pyramid of the Swedish Wallenberg Family Ownership stakes are representedwith C and voting stakes with V. Ultimate control is assigned to the Wallenberg family through a chain ofstakes of at least 20 percent.

in figure 2. The authors follow a chain ofcontrol by which ABB, Sweden’s fourthlargest firm by market capitalization, is con-trolled via a 32.8 percent stake held byIncentive, Sweden’s seventeenth largestfirm. Incentive, in turn, is controlled via a43.1 percent stake held by the WallenbergGroup, which is controlled by Investor,Sweden’s fifth largest firm, through a votingarrangement that gives it control over 35.7percent of the Wallenberg Group’s 43.1 per-cent stake in Incentive. Investor is theWallenberg’s family run closed end fund andserves as the apex firm for their family con-trol pyramid. In this way, a sharp divergencebetween control rights and cash flow rightscharacterizes many Swedish firms: Forexample, the Wallenbergs have voting con-trol over ABB, but actually have a cash flowrights stake of only about 5 percent.

As we explained above, a control pyramidallows a wealthy family or individual to con-trol a large number of firms with relativelylimited wealth. Pyramids permit this bycreating large deviations between votingrights and cash flow rights; allowing controlover many firms with a small cash flowstake in each.

Controlling families also effectuate controlby placing family members in executive posi-tions in key firms throughout a pyramidal

group. La Porta, Lopez-de-Silanes, andShleifer (1999) show that this frequentlygives controlling families direct executivedecision-making power, as well as control viapyramids. In 69 percent of their sample oflarge firms (the top twenty firms ranked bymarket capitalization of common equity atthe end of 1995 in each of twenty-sevencountries), the controlling families also par-ticipate in management. Participation isdefined as a family member (sharing thesame family last name) being the CEO, theChairman, the Honorary Chairman, or theVice-Chairman of the firm. The reportedpercentage underestimates the actualinvolvement of family members in executiveposition because executives married into thefamily may not share the family name.Claessens, Djankov, and Lang (2000) reportthat, in East Asian countries (Hong Kong,Indonesia, Japan, South Korea, Malaysia,the Philippines, Singapore, Taiwan, andThailand), professional managers are rare,and family members or trusted associates areusually in charge. Top managers are familymembers in about 60 percent of firms thatare not widely held. These studies are aboutcontrolling family participation in general,not controlling family participation in themanagement of pyramid member firms inparticular. Empirical exploration of this issue



4 Along with the managers of large insurance companiesaffiliated with the banks.

is needed. For Italy, however, Paolo F.Volpin (2002) presents data about familymembers’ participation in firms that belongto pyramidal groups. He finds that 50 per-cent of top executives in the top layer of apyramid are members of the controllingfamily. For lower layers, the figure is small-er: 26 percent for layer two firms and 7.4percent for layers three and below.

The discussion in this subsection leavesus with the following important points. Byusing pyramids and, to a lesser extent,crossholdings, super-voting shares, and theappointment of family members as execu-tives, a wealthy family can secure control ofa corporation without making a commensu-rate equity investment. Pyramids in partic-ular allow a divergence of voting rightsfrom cash flow rights much greater than istypically possible through direct owner-ship. Pyramids thus allow a family with agiven level of wealth to control corporateassets worth considerably more than directownership would permit.

3.3 Other Types of Group Control

La Porta, Lopez-de-Silanes, and Shleifer(1999) show that control pyramids are by farthe most important corporate group struc-tures, but publicly traded companies cancontrol each other in other ways. These gen-erate other group structures that are impor-tant in particular countries. For example,Japanese firms are organized into groupscalled keiretsu, in which firms with no con-trolling shareholders each hold small stakesin one another, but these stakes collectivelyamount to control blocks. Thus, each firm iscontrolled by the professional managers ofall the other firms in the group (see, e.g.,Berglöf and Perotti 1994). France, Canada,Germany, and other countries also containcontrol pyramids without family firms at theapex. In some cases, the apex firm is itselfwidely held. In France and Canada, somecontrol pyramids of publicly traded firmshave state-owned enterprises as their apexfirms. In Germany, banks sometimes serve

as de facto apex firms. Since public share-holders in German firms routinely sign overtheir voting rights to the banks that managetheir stock accounts, these control pyramidscan consist of nominally widely held firmsheld together by banks voting the holdingsof small shareholders. The large Germanbanks are all widely held, so the bank man-agers collectively vote majorities of theirown stock.4 Recent reforms require Germanbanks to advise investors of their right tovote their own shares, and this may effec-tively dismantle such structures over time.Of course family control pyramids are alsoimportant in all of these countries.

We use the term corporate group torefer to all of these structures collectively,and reserve the term family control pyra-mid for business groups with a basicallypyramidal structure of intercorporate own-ership and a family firm at the apex.Corporate groups other than family con-trolled pyramidal groups are economicallyimportant in some countries. La Porta,Lopez-de-Silanes, and Shleifer (1999) findthat family controlled pyramidal groupspredominate in the large corporate sectorsof most countries, so they are the focus ofsection 3.2. However, other corporategroups are lumped together with them insome studies, introducing an ambiguitythat is perhaps unavoidable at this earlystage in the literature. Further work deter-mining the economically important differ-ences, if any, between different sorts ofcorporate groups would be welcome.

3.4 A Few Wealthy Families Control aLarge Fraction of Many Economies

That family control is widespread aroundthe world is perhaps interesting, but hardlyunsettling. What makes this situationimportant to our understanding ofeconomies outside the United States andUnited Kingdom is that control pyramids



concentrate a country’s corporate decisionmaking in remarkably few hands.

The articles surveyed above are filled withrepresentative examples.

Morck, Stangeland, and Yeung (2000)describe how the Canadian Bronfman fami-ly extends its considerable wealth into con-trol over corporate assets worth vastly more.Attig, Gadhoum, and Lang (2003) adddescriptions of the Manix, McCain, and theBentley families. La Porta, Lopez-de-Silanes, and Shleifer (1999) discuss yet otherexamples of Canadian control pyramids.While these examples show how wealthyCanadian families greatly magnify their eco-nomic control, Morck, Stangeland, andYeung (2000) report that widely held firmsare also commonplace in that country. Thus,no single family controls a predominant sliceof the national economy.

Contrast this to Sweden, where JonasAgnblad, Berglöf, Peter Högfeldt, andHelena Svancar (2001) report that one fam-ily, the Wallenbergs, controls roughly half ofthe market capitalization of the StockholmStock Exchange. This is accomplishedthrough a huge and complicated controlpyramid. What distinguishes Sweden fromCanada is that control rights in Swedishfirms are substantially in the hands of a sin-gle family, rather than a handful of familiesand numerous professional managers.

Other countries typically fall between theextremes represented by Canada andSweden. For example, La Porta, Lopez-de-Silanes, and Shleifer (1999), Marco Bianchi,Magda Bianco, and Luca Enriques (2001),and Faccio and Lang (2002) describe thecontrol pyramid that lets the Agnelli familycontrol corporate assets worth vastly morethan their actual family wealth in Italy.Faccio and Lang (2002) further report thatthe Agnellis control 10.4 percent of thatcountry’s total market capitalization.Overall, across Western Europe, the value ofcorporate assets controlled by the leadingfamily, measured as a fraction of market cap-italization, ranges from 18 percent in

Switzerland to only 1.10 percent in theUnited Kingdom. For the ten largest fami-lies, the figures are 19 percent for Austria,30 percent for Belgium, 22 percent forFinland, 29 percent for France, 21 percentfor Germany, 14 percent for Ireland, 20 per-cent for Italy, 23 percent for Norway, 34 per-cent for Portugal, 11 percent for Spain, 13percent for Sweden, 29 percent forSwitzerland, and only 4 percent for theUnited Kingdom.

East Asian economies tend to resembleSweden more than Canada. Claessens,Djankov, and Lang (2000) find that the topfifteen family control pyramids in typicalEast Asian economies hold corporate assetsworth a large fraction of GDP—84 percentof GDP in Hong Kong, 76.2 percent inMalaysia, 48.3 percent in Singapore, 46.7percent in the Philippines, and 39.3 percentin Thailand (these are the largest numbersreported). They state (p. 109) that “theseresults suggest that a relatively small numberof families effectively control most EastAsian economies.” They also find that thesingle wealthiest family controls 17.1 per-cent of the market capitalization in thePhilippines, 16.6 percent in Indonesia, and11.4 percent in South Korea.

La Porta, Lopez-de-Silanes, and Shleifer(1999) report that Huchison Whampoa,the third largest listed company in HongKong, is 43.9 percent controlled byCheung Kong Holdings, the fifth largestlisted company, which is 35 percent ownedby Li Ka Shing family. Li Ka Shing,through a pyramidal arrangement, controlsthree of the twenty largest companies inHong Kong, including the eleventh largest,Hong Kong Electric Holdings. Claessens,Djankov, and Lang (2000) provide furtherdescription of the Li Ka Shing controlpyramid.

Claessens, Djankov, and Lang (2000)describe the extensive corporate controlexercised by the Ayala control pyramid inthe Philippines. The Ayala Corporation, thesecond largest listed company on the



Source: Morck, Stangeland, and Yeung (2000).

Figure 3. Billionaire Wealth in millions per Billion Dollars of GDP: Billionaire nationalities are as listed inForbes Magazine’s annual list of U.S. dollar billionaires for 1996.

Australia

Inited Kingdom

India

South Africa

Japan

Turkey

United States

Canada

Israel

Western Europe

Latin America

South East Asia

0 10 20 30 40 50 60 70 80 90 100 110 120 130 140

Manila Stock Exchange in terms of marketcapitalization, controls Ayala Land, thelargest. The Bank of the Philippine Islands,the fifth largest, also belongs to the Ayalacontrol pyramid. The principal owner of theAyala Corporation is the privately heldMermac Inc. The Tokyo Mitsubishi Bankcontrols 23 percent and others control lessthan 5 percent each. Mermac is 100 percentcontrolled by the Ayala family.

Kee-Hong Bae, Jun-Koo Kang, and Jin-Mo Kim (2002) describe the importance ofKorean family controlled pyramids, or chae-bols. Chaebols, while basically pyramidal innature, have extensive reciprocal holdingsas well as simple pyramidal intercorporateownership links. The top thirty chaebols“contribute to 62.5 percent of the totalassets and 72.6 percent of the gross sales ofall listed firms” (p. 2699) and each is diver-sified across many unrelated industries.Each is also controlled by a single wealthyfamily (p. 2702).

Figure 3 shows the wealth of billionairesas a fraction of GDP, expressed as dollars of

billionaire wealth per thousand dollars ofgross domestic product.

3.5 Summary

The findings in this section condense intothree crucial points. First, the prevalence ofwidely held firms in the large corporate sec-tors of the United States and UnitedKingdom is highly exceptional. Second, thelarge corporate sectors, excluding state-owned enterprises, of most countries arepredominantly controlled by very wealthyfamilies. Third, the number of wealthy fam-ilies exercising this control is typically quitesmall because control pyramids translatesubstantial family wealth into command ofcorporate assets worth vastly more. In somecountries, individual families use this tech-nique to control considerable, and evenpredominant, slices of the country’s marketcapitalization.

Further careful empirical work is clearlyneeded to verify the actual extents towhich these points genuinely characterize



different countries. But assuming they arecorrect, they raise a curious question: Whyare pyramidal firm groups so much rarer in the United States and UnitedKingdom? Or, are pyramidal groups aheretofore neglected aspect of corporategovernance in the United States andUnited Kingdom?

4. Potential Benefits of Control Pyramids

We now examine the impact of such dis-proportionate control over a country’s cor-porate assets by tiny elites of extremelywealthy individuals and families—both atthe firm level and at the economy level. Wefirst review work consistent with dominantshareholders exerting a positive effect onfirm performance in freestanding firms.We then show that findings from free-standing firms with dominant shareholdersneed not carry across to pyramidal groups.Pyramidal groups may have important pos-itive effects in economies with underdevel-oped factor markets and institutions.However, asking why family controlledpyramidal groups persist raises a range of possible negative effects, discussed insection 5.

4.1 The Performance of FreestandingFamily Controlled Firms

Several studies suggest that family controlper se might often have a positive effect onfirm performance. Anderson and Reeb(2003b) study 403 firms in the Standard andPoor’s 500, an index of large U.S. firms.They find that about one-third of them haveindividuals or families as substantial share-holders, holding altogether 18 percent ofequity value. Using data for 1992 through1999, excluding utilities and banks, theyregress Tobin’s average q, a standardizedmeasure of firm valuation and a well-accept-ed proxy for the quality of corporate gover-nance, on a standard set of control variables

5 Tobin’s average q is the market value of all the firm’sfinancial obligations over the replacement cost of its phys-ical and intangible assets. Note that this literature usesaverage q, not marginal q (the increase in market valueinduced by a unit increase in replacement cost of assets),which should be one absent taxes and transactions costs. Ina very long run equilibrium, average Tobin’s q should alsobe one, but need not be in the shorter run. For example, ifuncertainty about insiders’ stealing partially resolves aftershares were issued to the public, or if inframarginal andmarginal financing have different costs, an individual firm’saverage q might end up above or below one. Durnev,Morck, and Yeung (2004, table 2) show empirically thatU.S. firms’ average qs and marginal qs differ markedly. Incorporate finance, a long empirical tradition, beginningwith Morck, Shleifer, and Vishny (1988) and justified the-oretically by Stulz (1988), links average q (not marginal q)to proxies for corporate governance quality. Recent workby Paul A. Gompers, Joy L. Ishii, and Andrew Metric(2003) and others explicitly ties average q to direct meas-ures of corporate governance quality. The underlying intu-ition is that rationally expected corporate governanceproblems are revealed through time, changing individualfirms’ total market values (the numerator of average q), butonly affecting their replacement costs (the denominator ofaverage q) on the margin.

plus individual and family ownership anddummies for the presence of a founder orheir as CEO.5 They report that public share-holders assign substantially higher values tofirms with a large block held by an individualor family. They report an especially largevalue premium for firms whose foundersretain equity holdings and a smaller, but stillstatistically significant, value premium forfirms with other large individual or familyshareholders. Anderson and Reeb (2003a)use the same sample and find that familycontrolled firms are less diversified, aftercontrolling for size, outside directors, andblock shareholders. They conclude that“family firms perform as well as, if not betterthan, nonfamily firms.”

Morck, Shleifer, and Vishny (1988) reachthe same conclusion for U.S. firms in whichthe founder or a member of the foundingfamily is present in management and the firmwas incorporated within the previous thirtyyears. But for older firms with founders ortheir relatives in management, they report asignificant value discount. Holderness andSheehan (1988) report no systematic per-formance difference associated with large



U.S. shareholders. However, Francisco Perez-Gonzalez (2002), using U.S. data, divides 162CEO transitions where the retiring CEO is amember of the founding family into caseswhere the succeeding CEO is a member ofthe founding family and cases where she isnot. Stock prices, returns on assets, andmarket-to-book ratios fall sharply for firmswith inherited control, but not for firmswhere the CEO is unrelated to the foundingfamily. Moreover, these declines are partic-ularly prominent for firms that appoint fam-ily CEOs who did not attend a selectivecollege. Indeed, the differences betweenfirms with outside succession and in-familysuccession by an offspring alumnus of aselective college are not statistically differ-ent. Clearly, these results suggest that largeblock-shareholdings by a family, and familyinvolvement in management, need notdestroy value, and may even add value forpublic shareholders—especially when fam-ily control means control by an entrepre-neurial founder or a demonstrablycompetent heir. Further studies alongthese lines for other countries would behighly useful.

Similarly, in an event study usingCanadian family controlled firms, Brian F.Smith and Ben Amoako-Adu (1999) reportshare price drops when heirs take charge—especially if the heirs are young. They pro-pose that older heirs, with more businessexperience, might reassure investors.

Francesco Caselli and Nicola Gennaioli(2003) develop a model of dynastic manage-ment—the passing of control from father toson. In their model, this is a potential sourceof inefficiency because heirs may inherit lit-tle of their parents’ talent for managerialdecision making. They explore the macro-economic causes and consequences of dynastic management, and derive that theincidence of dynastic management dependson the severity of asset market imperfec-tions, the saving rate, and the inheritabilityof talent. Using numerical simulations, theyshow that dynastic management may be a

6 Jeffrey W. Allen and Gordon M. Phillips (2000) report227 instances of U.S. firms buying blocks of stock in oneanother from 1980 to 1991. These equity stakes are usual-ly small (averaging 15 percent of the seller’s equity) andare often preliminary to complete takeovers. They aresometimes referred to as toeholds. Indeed Wayne H.Mikkelson and Richard S. Ruback (1985) and DosoungChoi (1991) can explain positive announcement abnormalreturns associated with such purchases by the anticipationof subsequent takeovers. In some cases, joint venture part-ner firms may also exchange blocks of stock. Thus, U.S.intercorporate equity blocks are usually between pairs ofotherwise freestanding firms. The only extant pyramidalstructure in the United States, to our knowledge, is theventure capital organization, Thermo Electron, whichretains control blocks in its high technology public spinoffs. Thus, family control and intercorporate blockholdingsin the United States are quite atypical; for wealthy U.S.families seem virtually always to control but a single largepublicly trade firm.

substantial contributor to observed cross-country differences in productivity.

Unfortunately, the empirical results citedabove, except Smith and Amoako-Adu(1999), use U.S. data. Further studies usingdata from other countries are needed. Assection 3 shows, the United States is excep-tional for its almost complete absence ofcontrol pyramids, and for the rarity andfleeting nature of large listed firms holdingcontrol blocks in other large listed firms.6Elsewhere, family control pyramids, inwhich tiers of listed firms hold control blocksin other listed firms, are commonplace.

The U.S. results described above allowthat, in freestanding firms, family controlmight sometimes be a net advantage.However, findings regarding freestandingfirms cannot be applied hastily to pyramidfirms and empirical studies suggest impor-tant differences. For example, Morck,Stangeland, and Yeung (2000), after control-ling for firm age, size, and main industry,report that Canadian heir controlled firmsare less profitable than otherwise compara-ble firms in the United States and in Canada.One possible explanation is that typicalCanadian heirs running large firms areworse managers than their U.S. counter-parts. Another is that many large Canadianfirms belong to pyramidal groups. While we



cannot preclude the former reason, the lat-ter seems more immediately plausible. Wetherefore consider ways in which pyramidalstructures might affect firm performance—positively and negatively. We discuss thepositives in the remaining part of this sectionand the negatives in the next section.

4.2 Could Control Pyramids Have PositiveEffects?

Control pyramids must be of value tosomeone or they would not predominate asthey do in many economies. An emergingempirical literature attempts to identifyadvantages conferred by family control pyra-mids. We review some of the most importantcontributions.

Tarun Khanna and Jan W. Rivkin (2001)examine firms affiliated with businessgroups in Argentina, Brazil, Chile, India,Indonesia, Israel, Korea, Mexico, Peru, thePhilippines, South Africa, Taiwan (China),Thailand, and Turkey. They define a busi-ness group as “a set of firms which, thoughlegally independent, are bound together by aconstellation of formal and informal ties andare accustomed to taking coordinatedactions.” Thus, they are examining controlpyramids run by families as well as other cor-porate groups; however, most of the groupsthey examine appear to be family controlpyramids. They report higher averagereturns on assets (ROA) among group firmsin all the countries they study saveArgentina, Brazil, Chile, Mexico, Peru, andthe Philippines. They also find lower varia-tion in ROA for group firms, except inMexico and Turkey. From these findings,they conclude “not only that members of agroup incur costs and reap benefits, but alsothat the costs and benefits are shared.”

Khanna and Krishna Palepu (2000a) com-pare the performance of member firms ofIndian business groups, which are familycontrol pyramids, with that of freestandingfirms. Some of these business groups spanmany distinct industries, while others aremore focused. They find that accounting and

stock market measures of firm performancefor group affiliates have nonmonotonic rela-tionships with group diversification. Firmsin somewhat diversified control pyramidsperform worse than free-standing firms,while firms in highly diversified control pyr-amids perform better than their freestand-ing peers. Apparently, a high level ofdiversification ultimately benefits groupaffiliates.

Khanna and Palepu (2000a) and Khannaand Rivkin (2001) both explain their findingsby positing various arguments as to why thegroup structure might be an advantage ineconomies with poorly functioning marketsand institutions that might cause unrelatedfirms to have difficulty writing and enforcingcontracts with each other. The idea thatgroups mitigate market frictions goes back toNathaniel H. Leff (1976, 1978).

One set of arguments stems from the needto overcome capital market frictions. Ineconomies where information asymmetry issevere and institutional devices to signaltrustworthiness are inadequate and ineffec-tive, external financing is expensive and lim-ited. A group structure can enhancemonitoring and overcome liquidity con-straints by letting group firms pool resources.The pyramid’s controlling shareholder couldpotentially allocate capital more efficientlythan the weak capital market could. This isessentially the same “internal versus externalcapital market argument” that Wilbur G.Lewellen (1971), Robert H. Gertner, DavidS. Scharfstein, and Jeremy C. Stein (1994),and Stein (1997) develop in connection withfreestanding U.S. conglomerates.

In a similar vein, Mike Burkart, FaustoPanunzi, and Shleifer (2003) examine whetherentrepreneurs want to surrender control offirms they found. They accept the potentialbenefits of owner control, but pitch themagainst the utilization of outside capable pro-fessional managers. When the benefits are lessthan the forgone benefits of rendering controlto capable outside professional managers, acase may exist for surrendering control.



Outside professional managers, however, mayexpropriate firm value. Thence, the forces thatfactor in on the question of whether to sur-render control are the amenities and benefitsof running a family business verse the extrafirm value capable professional managers canbring in net of their expropriation and themonitoring costs. The net extra firm valuecapable professional managers can bring in isless in an environment where managers canmore readily disregard outside shareholder’sproperty rights. The lack of separationbetween management and ownership controlis evidence of underdevelopment in financialmarkets.

Most of the results which show that“groups” may have a positive effect on firmperformance are based on developing ornewly developed countries data. It is notclear whether the lack of reported resultsfor developed countries indicates lack ofempirical attempts using developed countrydata or that it validates the theoretical argu-ments like Burkart, Panunzi, and Shliefer(2003). (Journals seem to have a bias againstpublishing papers that report insignificantstatistical results.)

More generally, the arguments proposedby Khanna and Palepu (2000a) and othersstem from the need to overcome frictions infactor markets. In essence, markets for cap-ital, managerial talent, and intangibles areinternalized within groups because externalmarkets are poorly developed. When institu-tions are weak, doing business with strangersis dangerous and unreliable. This impedesthe operation of labor, capital, knowledge,and product markets. However, familieswith reputations for fairness and good man-agement practices are especially sought afteras business partners in such environments.Control pyramids let families with good rep-utations achieve greater economies of scale,and thus allow a higher degree of trust thanwould otherwise be possible. This, in turn,facilitates economic activity. Control pyra-mids may well be an economically rationalresponse to poorly functioning markets in

developing economies. In economies whereexternal markets for professional managersare thin and underdeveloped, a group’sinternal market structure can lead to moreinvestment in recruiting, training, andgreater incentives for employees to develop“group specific human capital.” Similar con-siderations apply to the development ofmarket skills and brand-names.

A separate argument turns on the rela-tionships between a firm’s stakeholders—thevarious claimants to its cash flows such as theState, the public, the firm’s workers, credi-tors, suppliers, customers, and various typesof its shareholders. Mark J. Roe (2003)argues that family control pyramids serve toprotect shareholders from powerful laborunions and other social interests. Thesearguments are perhaps plausible as explana-tions for the prevalence of control pyramidsin developed welfare state economies likeSweden and Canada, but, as Roe (2003) con-cedes, they seem inadequate at explainingtheir persistence in developing economies.In support of his argument, Roe (2003)shows that countries with more socialisticpolitics and stronger labor rights have lessdispersed ownership. One difficulty we havewith the Roe (2003) argument is that causal-ity is not easily determined. Leftist politicsand strong labor movements might just aseasily form to counter the power of wealthyfamilies controlling multiple corporations.Another difficulty with Roe’s argument isthat pyramids give controlling shareholdersinterests that can sharply diverge from thoseof public shareholders. Further work isclearly needed to clarify these issues.

4.3 But, Why Pyramids?

The literature positing positive effects ofpyramidal structures, whether derived fromKhanna and Palepu (2000a) or Roe (2003),begs the question: Why do pyramidal groupsform, rather than freestanding conglomer-ates? Research on this question is verysparse, as far as we know. We can thereforeonly offer several conjectures.



First, separately registered firms mightprovide at least some limited liability pro-tection, so that a parent firm’s losses notexceed its initial investment. Limited liabili-ty, however, also reduces outside investors’ability to recover their investments in case ofa loss. In addition, limited liability need notalways provide full protection. For instance,Morck and Masao Nakamura (2004) showthat, when the Suzuki group’s Taiwan Bankfailed in 1930s Japan, all the other compa-nies in the Suzuki family’s pyramid failedsimultaneously too.

Second, a pyramid might create space forthe promotion of bright executives to topmanagement positions. Perhaps these peo-ple would be frustrated as middle managersin a big conglomerate. Also, running a legal-ly separated unit may give these executivesmore freedom and perhaps might createbetter career incentives. However, this free-dom is likely to be quite illusory because thecontrolling family often retains executivedecision power.

Third, many smaller firms, rather thanone huge one, might conceivably allow bet-ter monitoring of professional managers bythe family. Pyramids might be less opaquethan a single conglomerate—to the control-ling family, if not to public shareholders.This point and the previous one are consis-tent with George P. Baker (1992), who pro-poses that conglomerates are moreefficiently run when head office delegatesas much power as possible to operationalunits. It is possible that pyramidal groupsinsulate individual operations from the headoffice more effectively than conglomeratedivisions. However, we are unaware of anyevidence regarding this.

Fourth, many U.S. conglomerates haverecently issued tracking stocks. These arestocks whose dividends are tied to a singledivision of the conglomerate. Julia D’Souzaand John Jacob (2000) report a positiveabnormal stock return upon the creation oftracking stocks, indicating that many relatedstocks controlled by the same top managers

7 Martin Holmén and Högfeldt (2005) propose an alter-native, but closely related, model of a “pecking order the-ory of finance” induced by insiders’ extraction of privatebenefits of control.

are worth more than a single conglomerate’sstock. They propose that distinct stocks mightallow readier resolution of information asym-metries and principal agent problems, andFrank Gigler and Thomas Hemmer (2002)provide an information theory model alongthese lines. Perhaps analogous models applyto pyramidal and other corporate groups.

Fifth, Takeo Hoshi, Anil Kashyap, andScharfstein (1991) argue that Japanese firmsprovide capital to other firms in their corpo-rate groups that are experiencing financialdifficulties. They propose that this intercor-porate insurance enhances economic effi-ciency by reducing bankruptcy costs.Although Morck and Nakamura (1999)present evidence that such transactions arebetter viewed as bailouts, and are probablynot economically efficient, corporate groupsin other countries might still perform thefunction Hoshi, Kashyap, and Scharfsteinenvision.

Sixth, Heitor Almeida and Wolfenzon(forthcoming-a) argue that capital marketfrictions induce pyramids. In their model,weak investor protection keeps firms fromraising external finance unless internal fundsare available as “seed money.” If a wealthyfamily sets up a new freestanding firm, it hasonly family wealth available as seed money,which should include dividends paid fromfirms it controls. But if it uses an existingfirm to set up a new one, the accumulatedretained earnings of that firm are availableas “internal funds.” 7

Of course, this framework fails to explainwhy the wealthy family doesn’t simplyexpand an existing firm by issuing equity ordebt. However, as David Landes (1949)points out in connection with French familyfirms, equity issues dilute the family’s con-trol block and debt issues raise the risk ofbankruptcy. Both modes of expansion thusraise the risk of the family losing control.



8 See Morck and Nakamura (2004) for details.

In essence, Almeida and Wolfenzon (forth-coming-a), augmented by wealthy family’sdesires to retain control, echo the explanationof Yoshisuke Aikawa (1934), the founder ofthe Nissan pyramidal corporate group in pre-war Japan, of why he opted for a pyramidalstructure. He explains that pyramids are anideal solution to what he calls the “capitalist’squandary.” If a capitalist uses only his ownmoney or his family’s money, his scale ofoperations is too small. If he taps public equi-ty markets, he risks losing control. But, as dis-cussed above in connection with figure 1, apyramidal group provides the best of bothworlds—secure control and unlimited accessto public capital. 8

Finally, Morck, Stangeland, and Yeung(2000) and Morck and Yeung (2004) pro-pose that political influence is proportionalto what one controls, not what one actuallyowns. As we explain in more detail below,this might allow the controlling owners ofpyramidal groups unrivaled political influ-ence in many countries. Morck and Yeung(2004) argue that, by magnifying the wealthof an individual or family into control overcorporate assets worth vastly more, pyra-mids magnify political influence in the sameproportion. The wealthy individual or fami-ly can use this amplified political clout toalter the economy’s institutional frameworkto favor themselves or their firms, to cap-ture transfer payments, and so on. Thisadvantage might be considerable, especiallyin economies with readily corruptible politi-cians and officials. Indeed, it might be solarge as to explain apparently superior firmlevel performance, as found by Khanna andPalepu (2001a) and Khanna and Rivkin(2001), despite obvious evidence of poorgovernance and overall poor economy per-formance. Anne O. Krueger (1993), KevinM. Murphy, Shleifer, and Vishny (1991,1993), and many others argue thateconomies in which political rent-seeking is

highly profitable grow slowly. If pyramidalgroups prosper because they are especiallyadept rent-seekers, they perhaps ought tocontain the best performing firms in theworst performing countries. Further work isneeded to clarify these issues.

For example, the determinants of pyra-miding are not well understood. Almeidaand Wolfenzon (forthcoming-a) and Aikawa(1934) correctly emphasize that pyramidsare formed because they magnify a givenlevel of personal wealth into control overcorporations worth vastly more. However,more work is needed to clarify how the pref-erence of the controlling owner enters thepicture and why firm units are sometimestransferred between pyramidal groups andin and out of pyramids.

For example, Berle and Means (1932) andothers argue that a pyramid’s lower-tier firmsare especially inefficiently run because ofthe large gap between the controllingowner’s control rights and cashflow rights.This inefficiency should attract takeovers. Ingeneral, control should pass to the party whovalues the cashflow rights plus private bene-fits of control the most. The private benefitsinclude tangible ones, such as money fromtunneling, and intangibles, like the powerand influence derived from ruling a businessempire—see Morck, Stangeland, and Yeung(2000) and Morck and Yeung (2004).

If tunneling were costless, the controllingowner could appropriate all the benefits ofenhanced efficiency in all her pyramid’sfirms. If intangible benefits were also unim-portant, the pyramid controlling ownercapable of running firms most efficientlyshould pay the most for control blocks, andmaximally efficient management should pre-vail. Arguably, the small personal wealthneeded to acquire control via a pyramid andthe subsequent extraction of all the benefitsof enhanced efficiency via costless tunnelingshould make efficiency enhancing corporatetakeovers loom larger as attractive invest-ments than they would be in an economy offreestanding firms.



Alternatively, if tunneling were costly orintangible benefits important, controlwould pass to the controlling shareholdermost adept at tunneling or most enthusias-tic about consuming intangible privatebenefits of control. The former considera-tion leads Lucian Arye Bebchuk, ReinierKraakman, and George G. Triantis (2000)to postulate that control in an economywith pyramiding might pass to the mostefficient thief. If the controlling sharehold-er consumes intangible benefits directly fromall the firms she controls, but consumingpecuniary benefits require costly tunneling,the small personal wealth needed to acquirecontrol magnifies the importance of intangibleprivate benefits in allocating corporate con-trol. Control then might pass to the mostpower-hungry oligarch.

Much additional theoretical and empiri-cal work is needed to solidify hypotheses inthis area and to distinguish presumptionsfrom facts.

5. Control Pyramids and the Separation of Ownership from Control

The discussion above suggests that, evenwere control pyramids economically salubri-ous, for example as devices for circumvent-ing inadequate institutions, there arenonetheless reasons for taking a more skep-tical view. Certainly, Aikawa’s (1934) enthusi-asm for control pyramids demonstrates scantconcern for the rights of public shareholders.

The first reason for skepticism is that apyramidal corporate ownership structureallows the controlling shareholder tosecure control rights without commensu-rate cash flow rights. Secured control rightsprotect the controlling owner from losingpower—that is, they allow her entrench-ment. Furthermore, as we show below,once control is secured, the controllingshareholder’s low cash flow rights lead toagency problems, including non-value max-imizing investment and incentives to divertresources.

Second, in family-controlled pyramids,family members often retain top managementpositions. Succession by inheritance can havea negative effect because top managementpositions go, not to the most capable, but, atbest, to the most capable member of the con-trolling family, as modeled by Caselli andGennaioli (2003).

Third, even were resources efficientlymanaged and allocated within a group con-trolled by single family, corporate groupscan still be undesirable. Almeida andWolfenzon (forthcoming-b) argue that theefficient allocation of resources within agroup can actually exacerbate the ineffi-cient allocation of resources across groups.Their arguments apply to corporate groupsin general, not just to family controlledpyramids.

Finally, the concentrated control of pro-ductive assets by a few families gives themmarket power, both in the goods and capitalmarkets, leading to potentially undesirableeconomic consequences. Morck andNakamura (2004) explain how this was theexplicit rationale cited by the United StatesOccupation Force for dismantling the mainJapanese family control pyramids afterWorld War II.

Widespread problems of these types,especially those that impair capital mar-kets, can create inefficiencies that hamperoutside financing, discourage innovation,and retard economic growth. These ineffi-ciencies can arise at both the firm level andoverall economy level. In this section, wefocus on inefficiencies at the firm level.These stem primarily from the separationof ownership and control in the pyramid—the first two points enumerated above. Thenext section focuses on inefficiencies at theeconomy level which stems from the con-trol of a very large number of firms by asingle family—the last two points listedabove. This section is deliberately brief, forother surveys, notably Shleifer and Vishny(1997) and Diane K. Denis and John J.McConnell (2003), cover some of this



material, though not from the perspectivewe emphasize here.

5.1 Divergence of Interests AgencyProblems

The problems caused by the separation ofownership and control in pyramidal groupsare similar to those described in Berle andMeans (1932) and Jensen and Meckling(1976) in connection with widely held firms.Indeed, Berle and Means explicitly describepyramids as having problems like those ofwidely held firms. Jensen and Mecklingargue that the managers of firms with dis-persed shareholders have substantial discre-tion as to the actions they take becauseindividual shareholders cannot coordinateto share monitoring and control costs. Thislets managers take actions with benefitsthat, by their nature, are not shared withshareholders. In addition to indirect finan-cial benefits, like on-the-job consumption orshirking, and direct financial benefits, likeredirecting corporate assets into a personalaccount, control also provides intangiblebenefits, like status, political influence, andpower over people. The literature denotesall of these benefits as private benefits ofcontrol.

Managers pursue private benefits, butbalance this against their losses when theprice of their shares in the firm falls plus theconsequences of getting “caught and pun-ished,” such as losing their jobs. This bal-ance means managers might choose anaction with lower total value over one withhigher value, as long as the former generat-ed sufficient private benefits. This diver-gence of interests is greater the lower themanagers’ equity stake and the lower thelikelihood of her getting caught and punishedfor non-value-maximizing behavior.

As we described in section 3, firms out-side the United States and United Kingdomare owned, not solely by dispersed smallshareholders, but by wealthy families viacontrol pyramids. Pyramids allow a family toretain control of many firms while holding

only a small fraction of their cash flowrights. Indeed, we presented examples inwhich the cash flow rights of the controllingfamily in some of the pyramid memberfirms are comparable to the stakes of themanagers of the most diffusely held of U.S.corporations. By allowing cash flow rightsand voting rights to diverge, control pyra-mids permit the same divergence of interestproblems as dispersed ownership, eventhough the firms in the pyramid are notwidely held.

This divergence of interests can lead toinefficient investment in firms in which acontrolling owner has small cash flowrights. This is because the controlling fami-ly earns only a small part, corresponding toits small cash flow rights in such a firm, ofany investment’s monetary payoff but canretain all of any private benefits the invest-ment generates. Only in the special case ofan investment that maximizes the total sur-plus also maximizing the controlling owner’sprivate benefits should we observe efficientinvestment decisions.

A related reason for inefficient invest-ment arises from the high cost of capital thatthis divergence of interest entails. This highcost stems from minority shareholders’rational expectation that this divergence ofinterest distorts corporate decisions to ben-efit controlling shareholders. Such problemsare defined in the Corporations Law ofmany countries as oppression of publicshareholders by the controlling shareholder.Oppression problems are thus a variant of“perks” consumption, where the perks arediversions to benefit the controlling share-holder rather than professional managers.La Porta, Lopez-de-Silanes, and Shleifer(1999) document countries varying substan-tially regarding the protection their lawsafford external investors against oppres-sion. Shleifer and Wolfenzon (2002) showthat, when a controlling shareholder has alow cash flow stake, as with firms at thebase of a pyramid, the cost of capital to thefirm is high since other investors anticipate



expropriation and pay depressed prices forany securities that firm sells. A consequenceof this high cost of capital is severely distort-ed investment decisions because of thosefirms’ reluctance to raise external funds.

5.2 Entrenchment Agency Problems

The corporate finance literature notes asecond type of agency problem, calledentrenchment. While Jensen and Meckling(1976) argue that insiders with larger stakeshave less incentive to misallocate corporateresources, Stulz (1988) argues that higherequity stakes also give insiders more free-dom to misallocate resources. This isbecause top executives who own large blocksof equity are effectively tenured. Non-value-maximizing managers with little stock can beremoved in proxy fights by disgruntled insti-tutional investors or ousted by hostileraiders attracted by a depressed share price.Top executives with large blocks of equitycannot be cast out in these ways. Thus, adominant equity stake lets corporate insid-ers enjoy private benefits of control robustly,whereas a smaller stake means they mustlimit their inhalation to keep their shareprices high enough to ward off raiders andplacate institutional investors.

Moreover, entrenchment problems cantake on a qualitatively distinct air, forentrenchment can lock in control by honestbut inept insiders as well as clever self-serv-ing insiders. For example, entrenchmentproblems occur if a firm’s controllingfounder bequeaths her stock to an egre-giously incompetent, but power hungry son.The incompetent son cannot manage thefirm well, but cannot be removed by theother shareholders because his share ofvotes is sufficiently large to control theboard. The power hungry son gains suchutility from controlling the firm that theeffects of his blunders on the value of hisstock are an acceptable cost to him. Whilethis might even be Pareto efficient, it clearlyraises important distributive questions towhich we return below. This sort of

entrenchment is especially difficult to count-er, for if the controlling shareholder used hiscontrol rights to enrich himself pecuniarilyat the expense of public shareholders, thiscould be proscribed as “oppression.”

Of course, this picture is overly simplified.In principle, without maintaining a domi-nant voting share, the CEO of a widely heldfirm could also dominate the board throughsheer force of personality and push throughan array of antitakeover defenses, such aspoison pills, staggered boards, and the like,to deter raiders and activist shareholders.

Managerial entrenchment can occur infreestanding firms. Indeed, Almazan andSuarez (2003) argue that a degree ofentrenchment might be part of an optimalcompensation contract for top managers.Nonetheless, as Nissan founder Aikawa’s(1934) unwittingly emphasized, a fundamen-tal raison d’être of control pyramids is pre-cisely that they lock in control. That is, apyramidal structure is itself a means toentrench the controlling shareholder with-out the cost of maintaining a large equitystake nor the machinations of having toestablish poison pills or staggered boards.Control pyramids per se are simple andhighly effective antitakeover devices.

To see this, return to figure 1 and recallthe discussion in section 3.2. The familyappoints the boards of directors of compa-nies in the first tier and so can place familymembers in dominant positions on thoseboards and as top executives. The first tierfirm’s boards then appoint the boards ofcompanies in the second tier, which thenappoint the boards of companies in the thirdtier. This locks in family control over all thecompanies in the pyramid, even though pub-lic shareholders contribute most of thefinancing for most of its firms. In contrast,had the family established a single firm witheight billion dollars worth of physical assets(a union of the firms in the third layer), itsbillion dollars in family wealth would haveconstituted a 12.5 percent stake and publicshareholders could outvote the family in



9 Tunneling is of course another type of agency prob-lem, but we treat it separately here due to its importance.An agency problem is one where an agent, in this case con-trolling family, can take actions that negatively affect theutility of the principal, in this case the minority sharehold-ers. In this sense, both entrenchment and tunneling areagency problems.

shareholder meetings. This would renderfamily control contestable.

Note further that entrenchment via familycontrol pyramids is intimately linked to theappointment of family members to the boardsof the apex firm and other firms throughoutthe pyramid. The example above—allowing“an egregiously incompetent, but power hun-gry son” to retain control—underscores thepotential for damage to firm value due toentrenchment. But the potential for damagegoes beyond the controlling manager havingexcessive freedom to pursue her self-interestat the expense of firm value. It also includesthe possibility that an incompetent managermight retain her job precisely because shecan extract more personal gains from the jobthan a competent manager could. Indeed,Bebchuk, Kraakman, and Triantis (2000) sug-gest that control might naturally pass to “themost efficient thief” because he is willing topay the most for a control block.

5.3 TunnelingThe firms low in control pyramids, in which

the controlling shareholder’s cash flow rightsare small, are thus vulnerable to concurrentdivergence of interests and entrenchmentproblems. One consequence of this juxtaposi-tion is that the controlling shareholder has anincentive to tra

Date post:	14-Jun-2020
Category:	Documents
Upload:	others
View:	0 times
Download:	0 times

Corporate Governance, Economic Entrenchment, and Growth€¦ · Corporate Governance, Economic...

Documents