Business Groups in Brazil

Dante Mendes Aldrighi and Fernando A. S. Postali*

November 17th, 2007

Introduction A recent survey on business groups (Khanna and Yafeh, 2007) has pointed out that a number of unsettled issues still remain on that subject. Have they emerged as a reaction to any kind of market failures? If yes, has that reaction be propelled solely by their own initiatives or also under the aegis of government policies? Have their emergence and persistence over time been propped up by government support, achieved in turn thanks to political connections? Is the formation of business groups primarily motivated by expropriation of minority shareholders or by regulatory or tax incentives? Do they improve or reduce social welfare? Do deleterious social effects (such as rent seeking and market power) potentially resulting from their existence overwhelm purported positive effects related to attenuating financial constraints, contract imperfections, and transaction costs?

Based on economic and political considerations, Schneider (2007) proposes a three-pronged typology of business groups. “Organic groups” would evolve relying on economies of scope and/or vertical integration by exploiting some group-specific assets, generally knowledge-based assets embodied as human capital, some technological, organizational, or managerial expertise, or reputation.1 “Portfolio groups” are primarily motivated by optimizing the risk-return tradeoff, taking firms as fundamentally financial assets that should be acquired or sold depending on market circumstances. They tend to be highly diversified and management of each their subsidiaries to be significantly autonomous. The third ideal type of business group, “policy-induced groups,” emerge and grow essentially as a reaction to government incentives. As regards governance, an extensive branch of the literature about business groups, notably about those organized under pyramidal ownership structures, takes for granted that their rationale arises from expropriation of minority shareholders (or just tunneling, as dubbed by Johnson et al., 2000). Some analysts have pointed out that investors’ poor legal protection can magnify agency costs by facilitating the leverage of voting power over ownership. Pyramidal arrangements of ownership, dual-class shares, cross-shareholdings, and voting agreements are among the most used devices to separate voting rights from capital ownership. Notwithstanding the advantage of mitigating the tradeoff between monitoring and liquidity, the divergence from the “one-share-one-vote” rule would have the drawback of intensifying controlling shareholder’s incentive and power for expropriating minority shareholders.2 Concerning specifically to pyramids, this view is contested by Almeida and Wolfenzon (2006), who argue that they operate primarily as a means of ensuring financing to new enterprises. They present two reasons to uphold the claim that, when investor protection is poor and hence

* Both Professors at the University of São Paulo (Department of Economics). 1 Overall, the distinguishing feature of those assets lies in their being a non-rival good, that is, they can provide services to further activities in distinct businesses at a negligible cost, maintaining the supply of services to the original businesses. 2 Among those conceiving pyramids as a mechanism for expropriating minority shareholders, it should be mentioned: Bebchuk et al. (1999), La Porta et al. (1999), and Bertrand and Mullainathan (2002).

the latitude for diversion of cash flows is ample, the choice of the group structure is biased in favor of pyramidal arrangements. The first is “the payoff advantage”, allowing private benefits of control to be reaped by the controlling family at the expense of minority shareholders. Moreover, inasmuch as expectations of cash-flow diversion are reflected in the conditions under which outside investors are willing to supply capital, the controlling family can tap the retained profits of firms it already controls to finance the set up of new enterprises, allowing it to enjoy a “financing advantage”. Furthermore, firms with high investment requirements and/or low profitability would be more prone to pyramidal ownership arrangements. Besides providing higher payoff to the controlling family and implying lower returns to outside investors, firms with these characteristics face stronger financial constraints and, consequently, are those most benefited from intra-group transfer of resources. In light of this, the emergence of pyramids might represent an efficient response to overcome financial market failures, notably in countries with undeveloped capital markets and where the investor protection is poor (and therefore new companies’ returns to outside shareholders are low).3

Business groups’ varying observed structures and strategies around the world are likely to be connected to the circumstances under which they emerge, an issue clearly still deserving much further research. Focusing on Brazilian business groups, this paper endeavors to shed some light on that gap. To achieve this intent, the paper is structured as follows: the first section provides a brief account of the main changes underlying the Brazilian economy over the last 30 years and that shaped the background of incentives and constraints against which business groups have been operated; section 2 focuses on Brazilian groups’ structure and strategies, specifically on their ownership and control structures, the level of diversification of their businesses, the extent of internationalization of their activities, via trade and foreign direct investment, and their longevity and resilience through time; the subsequent section, following a more aggregate approach based on a large sample of public companies operating in Brazil, presents some descriptive statistics about their governance characteristics (such as their controlling shareholders’ categories and divergence between voting rights and cash-flow rights, the existence of pyramidal arrangements and other mechanisms of separating control from ownership), besides carrying out some statistical tests about possible relationships among these governance variables, and between them and firm performance; finally, the last section concludes.

1. The Sweeping Changes in the Brazilian Economy over the Last Three Decades: An Overview

Until the 70s, the industrial development in Brazil was strongly influenced by active import substitution policies, encompassing trade protection, below-market interest rate loans, tax exemptions, and other incentives (Fishlow, 1972). The economic rationale lying behind those policies was the purported insurmountable hurdles an infant industry in an underdeveloped country had to encounter to compete with long-standing foreign rivals. Besides its deliberate goal of strengthening the domestic industry, the import substitution policies aimed at reducing the dependence on imports as an endogenous reaction to adverse external shocks. These policies induced the development of Brazilian business groups, particularly over the period 1950-1979. The “Plano de 3 Khanna and Yafeh (2007) also refute that pyramids are just accounted for by expropriation, identifying potential efficiency reasons, such as mutual insurance or risk sharing. Less persuasively, they argue that part of the alleged tunneling may represent a reward to some core asset, like entrepreneurial ability or lobbying capability.

2

Metas” (Plan of Targets) and the “II Plano Nacional de Desenvolvimento” (Second National Development Plan) are remarkable instances of how those groups blossomed under the aegis of active industrial policies. Brought about between 1956 and 1960, Plano de Metas aimed to, and did, boost investment in infrastructure and some segments of the intermediate and capital good industries, such as electrical energy, petrol production, highways, steel, and cement, on top of car-making. Even though state-owned enterprises and foreign firms were at the front of this “big push”, some industries were left to Brazilian private business groups, which also benefited from demand spillovers derived from the leading firms. The Second National Development Plan, an ambitious plan stubbornly implemented by the politically-waning military administration (General Geisel) in a time of world economic slowdown and ominous balance-of-payment prospects (1974-1980), intended to overcome the structural and external constraints obstructing Brazilian economic growth. It targeted infrastructure (railways and telecommunications), energy (petrol, ethanol, and electricity), and those segments in the producer goods sector (both capital and intermediate goods, like steel, heavy chemicals, non-ferrous metals, and non-metal minerals) needed to complete an autonomous national industrial matrix (see Herman, 2005). Strong incentives were provided to Brazilian business groups with a view to inducing them to embark on this government-led catch-up, ranging from abundant, subsidized credit, to tax-exempt investments in targeted industries, to protection from foreign competition, and even to capital grants. Banco Nacional de Desenvolvimento Econômico, the Federal-Government-owned development bank, was pivotal in directing private investment decision towards the policy-targeted sectors.

The successive external shocks hitting the highly-indebted developing economies from 1979 to 1983 culminated in the debt crisis, bringing to an end the state-led development strategy followed so far by most of those economies. In Brazil, one of the main internal culprit by that crisis’ emergence and severity, the II PND, inasmuch as financing it had inflated the external debt (both public and private), also turned out to be a telling vehicle by which the dramatic dollar shortage was attenuated. Indeed, pressures on the balance of payments were significantly mitigated by virtue of either increasing exports or decreasing imports of intermediate and capital goods as II PND’s large-scale projects began to be matured. However, Brazil’ various levels of government were entangled in dire fiscal straits, partly owing to the “nationalization” of the private-sector external debt combined with the sharp currency devaluations (in 1979 and 1983), forcing the Brazilian state to bear the external debt service burden throughout the 80s and 90s and to relinquish their leadership in domestic investments. This fiscal crisis translated into high and growing inflation and interest rates, which in turn spurred firms to adopt defensive strategies, such as investing at best in enlarging their plants and too little, or not at all, in innovation (that is, in foreign technology adaptation) and exploiting the mix of high earnings and liquidity afforded by the inflation-indexed government bonds. The result can be meaningfully epitomized as “the lost decade”, during which stagnation and high inflation prevailed (artificially stopped for only ephemeral spells by successive unorthodox stabilization plans),4 large banks grew larger (cashing in on the inflationary tax), productivity levels widened their gap, instead of catching up with, developed countries’, and the fiscal crisis worsened. On the political front, Brazil began

4 The sequence of failed attempts to put an end to the high and unstable inflation process (Plano Cruzado, in 1986, Plano Bresser, in 1987, Plano Verão, in 1989, and Plano Collor I and II, respectively in 1990 and 1991) resorted to price freezing, the arrest of financial assets, and/or ad hoc measures to cope with inflation-indexed contracts.

3

its return to democracy in 1984, leaving behind 21 years of military administration, and in 1988 a new Constitution, still prevailing today, was enacted. Some contend that its populist bias added to some equivocal subsequent legislative and jurisdictional measures may have worsened the quality of the general economic regulation framework, aggravating contractual uncertainties (see Bacha and Bonelli, 2005, pp. 179-180).

The inertial inflation would be finally defeated in 1994 by “Plano Real”, which managed to cut down the annual inflation rate from 2708% in 1993 to 1093% in 1994 and to just 14.8% in 1995.5 That price stabilization was indisputably a turning point in Brazil’s recent economic history, reducing volatility and uncertainty and thus improving the business environment. Besides, it put pressure on firms to heighten productivity, since inefficiencies could no longer be concealed as they were during the inflationary times. Moreover, once the Brazilian currency stabilization plan was fundamentally underpinned by the pegged exchange rate at the same time that Brazilian inflation rate exceeded their chief trade partners’, the resulted overvalued real exchange rate further compelled firms and business groups to revise their strategies. Banks were also impacted by the dramatic plunge in inflationary tax revenues, which brought to light inefficiencies, mismanagement, and wrongdoing pervading some of them, including large ones, such as Banco Nacional, Bamerindus, and Banco Econômico.

Another watershed in business environment was privatization. Starting in the middle of the 80s, it gathered momentum mainly during Fernando Henrique Cardoso’s administration. Four stages can be identified in the privatization process conducted so far. The first one, prevailing in the 80s, focused primarily on “reprivatization,” that is, the selling of public assets that had been absorbed by the government due to financial stress or bankruptcy of private firms.6 The second phase embraced the period 1990-1994, when the “Plano Nacional de Desestatização” (PND), a formal privatization policy plan, was launched. Over that period 68 companies were sold, primarily from the steel, fertilizer and petrochemical sectors, generating US$ 11.8 billion worth in revenues and debt transfers. So far as foreign investors are concerned, their participation was trifling, inasmuch as macroeconomic instability (notably uncertainties about the value of the currency and the exchange rate) together with the absence of a neat regulatory framework threw a cloud of gloom over expected rates of return. During Cardoso’s two administration terms (1995-2002), the third stage, the privatization process was speeded up and enlarged by further including important infrastructure areas.7 This advance was prompted by important institutional changes that removed long-standing state monopolies in key sectors (e.g. energy, mining, oil and gas upstream, and

5 As measured by the IGP-DI (General Price Index - Internal Offer). See Ipeadata. 6 The unsystematic privatization from 1980 to 1990 resulted in 38 companies sold to private business groups and revenues totaling US$ 780 million to the Treasury. See BNDES (2002). 7 In terms of value, most of the deals (88.2 percent) occurred between 1995 and 2002, notably in 1997 (26.3 percent) and in 1998 (35.6 percent). With respect to the type of investor, the almost US$78.6 billion worth privatization revenue raised under Fernando Henrique Cardoso’s administration came predominantly from foreign investors (53 percent), followed by domestic firms (26 percent), domestic financial institutions (7 percent), individual investors (8 percent), and pension funds (6 percent). The great majority of the privatization revenues collected from 1995 to 2002 resulted from deals in the sectors of infrastructure and services (85 percent), extractive and manufacturing industries accounting for only 15 percent. In some sectors, debt transfers accounted for most of the price buyers paid for the firm, as in the cases of petrochemical (46.4 percent), mining (40.7 percent), and electricity (25.2 percent).

4

telecommunications).8 At the same time, a legal framework was set up with a view to regulating those sectors, including the creation of several specific regulatory agencies. Landmarks in the Brazilian privatization process occurred in that period, such as Companhia Vale do Rio Doce (CVRD), privatized in 1997, Telebras (in 1998), and Banespa (in 2000).9 The Federal Government undertook a strong and concerted effort to achieve its utmost goal, that is, to maximize privatization revenues.10 Indeed, cash payments and debt transfers related to privatization over the period 1995-2002 achieved US$ 93.4 billion (BNDES, 2002).11 Since President Lula da Silva took office in 2003, privatization has been limited to highway concession, local electric power, and energy transmission lines.

Trade liberalization also has had remarkable effects on market structures as well as on firms and business groups’ structures and strategies. Starting in 1988 with some shy measures to reduce tariff and non-tariff barriers, trade liberalization was speeded up during Collor administration (1990-1992). A tariff cut schedule was followed and most of discretionary import controls and non-tariff barriers were abolished (see Azevedo and Portugal, 1998; Secex-MDIT). The average nominal tariff fell from 41% in 1988 to 12.6% in 1995. Moreover, Mercosur, the allegedly customs union formed among Brazil, Argentina, Uruguay, and Paraguay in 1991, has deepened intra-bloc trade flows and intra- and across-bloc direct investment flows (see Bonelli, 1998). Except for a few setbacks induced by overvalued exchange rate, trade deficit and currency crises, Brazilian trade liberalization has been preserved.

Associated with trade and financial liberalization, privatization, and fierce global competition, a wave of mergers and acquisitions took place in the 90s, triggering a major restructuring in various sectors and firms. As shown in Table 1, food, beverage and tobacco were the leading sector in such deals over the period 1996-2000 (184),12

8 In 1995, a constitutional amendment released oil and gas upstream to private and foreign firms. However, the Federal Government keeps its controlling stake in Petrobras, the giant petrol firm, selling in 2000 its shares exceeding the minimum to retain control. 9 As regards the CVRD, now one of the leading mining companies in the world, the auction of its control in 1997 collected US$3.1 billion, to which should be added the US$2 billion worth global public offering in 2002. Telebras, the former Brazilian telecommunication holding, was split into 12 companies in June 1998, all of which transferred to the private sector. The Spanish group Telefonica took over through the auction the largest firm, Telesp, operating in São Paulo State. Later, the government auctioned concession rights to explore mobile phone services. Overall revenues from privatization in telecommunications services reached US$29.8 billion. Banespa, a bank originally owned by the State of São Paulo, was transferred, overburdened with debt, to the Federal Government before being privatized in 2000. 10 This ranged over many devices, including ad hoc changes in the legal and regulatory rules (such as the removal of the tag along clause comprised in the original corporate law), mobilization of the BNDES to render financially feasible some bids, inducing state-controlled firms’ pension funds to join some consortia, and allowing a minor Brazilian investment fund (Opportunity) to have an outstanding ascendancy over Brazilian pension funds in coalitions built to participate in the privatization auctions (and in the controlling of privatized firms). 11 The dramatic shift in corporate ownership at the end of the 90s can be grasped by the fraction state-controlled firms accounted for in all Bovespa-listed firms’ market capitalization value: it shrank from 61.2% in June 1997 to 22.3% in March 2007. 12 Large companies like Cargill, Parmalat, Arisco and Sadia acquired small- and middle-sized firms during this period. According to KPMG, 57% of such transactions involved foreign capital. Noteworthy among such deals was the remarkable merger of the two largest Brazilian breweries (Brahma and Antarctica), giving rise to AMBEV, the largest Latin American beverage group, and raising antitrust concerns.

5

followed by financial institutions (129),13 telecommunications (123, mostly after Telebras’ privatization), and, notably in 1999 and 2000, information technology. As a whole, about 2300 such deals occurred from 1994 to 2000, 61% of them involving foreign resources.

Table 1 Brazil: Mergers and Acquisitions

Sectors Total 2000 1999 1998 1997 1996Food, beverage and tobacco 184 36 25 36 49 38Financial institutions 129 18 16 28 36 31Telecommunications 123 26 47 31 14 5Information Technology 112 57 28 8 8 11Chemical and petrochemical 83 12 6 25 22 18Metallurgy and steel 78 11 9 23 18 17Advertising and publishing 73 23 17 19 9 5Insurance 70 6 9 15 24 16Energy 67 20 10 11 17 9Automobile parts 66 6 13 20 16 11Supermarkets 58 10 24 13 9 2Electrical equipment 53 5 5 9 19 15Oil 42 28 6 1 3 4

Source: KPMG

Great emphasis should also be laid on some material shifts in the institutional setting shaping corporate governance in Brazil since the end of the 90s. First, the Federal Government managed to put through, in 1997, a law abolishing the tag along right, so far guaranteed by the 1976 Corporate Law to minority shareholders, in the case of transfer of control. The rationale lay clearly in maximizing the revenues to be collected from the imminent privatization of some of the jewels in the crown of state-owned enterprises. Second, the approval of the new Corporate Law in 2001 and of a law (n. 10411), in 2002, reforming the law that had created (in 1976) and disciplined the “Comissão de Valores Mobiliários” (CVM), the Brazilian capital market regulator. The new law introduced only limited protection to minority shareholders, among which the right to tag along to minority shareholders owning common shares (guaranteeing at least 80 percent of the value paid to the former controlling shareholder); the right of minority shareholders and holders of preferred shares to nominate members of the company’s board of directors and fiscal committee; and a reduction in the upper limit of the fraction of non-voting shares in the firms’ capital (binding only for new issues).14

Third, the creation of differentiated levels of corporate governance by the São Paulo Stock Exchange (Bovespa), to which firms adhere voluntarily. At the end of the 90s, the Brazilian stock market stuck in appalling conditions: many public companies becoming private, a small number of IPO’s, low issues of new shares by listed companies, a shrinking volume of trading at the Bovespa, a low stock market capitalization value of the listed companies as a percentage of the GDP, and Brazilian major companies issuing

13 The banking sector was object of a deliberate restructuring policy (PROER) aiming at coping with some banks’ financial distress, mismanagement, outright fraud, and insolvency, unveiled by the sharp fall in float revenues resulting from monetary stabilization. 14 Overall, the new Corporate Law seems to take some steps in strengthening non-controlling shareholders’ rights vis-à-vis the former law. Nonetheless, it is still far from providing a legal framework hindering controlling shareholders from expropriating minority investors.

6

ADR’s on the NYSE. Probably acknowledging the insurmountable “political economy” hurdles to the adoption of legislation protecting minority shareholders’ rights, the Bovespa launched in December 2000 new listing segments wherein firms commit themselves, voluntarily and by contract, to higher governance requirements. That mechanism of governance certification allows firms’ self-selection regarding their respective governance quality, fitting the rising demand for better governance from domestic and foreign institutional investors, and facilitating Brazilian firms’ access to global capital markets, either as issuer of securities or as acquirers of foreign firms or assets. To be listed at Level 1, firms must comply with, among other requirements, disclosure of more detailed financial information as compared with that legally mandatory, of information on securities issued by the firm’s insiders, and of the terms of contract with related parties. On top of that, at least 25 percent of the firm’s shares should be free float, and the firm must ensure the capital dispersion when offering new shares.15 Listing to Level 2 requires, in addition, that firms live up to such commitments as tag along in case of firm control transfer (100 percent for holders of common stocks and 80 percent for holders of preferred stocks), disclosure of balance sheets in accordance with the U.S. GAAP or the IFRS, arbitration of any dissension between firms and their shareholders by independent specialized referees, and at least a 20% board composition of independent directors. Finally, New Market, in addition to Level 2’s obligations, precludes firms from issuing non-voting shares. It should be noted that most of large firms are listed at the Level 1, event that can be accounted for by their deliberate avoidance to migrate to the higher governance-requirements, or by specific regulation or legislation.16 This non-governmental institutional innovation is certainly connected to the boom in the stock market as well as in the number and amount of initial public offerings and new share issues taking place in Brazil since 2004.17 Most of firms going recently public have been listed at the New Market (72 out of the 102 IPO’s) and Level 2 (15). Besides, foreign investors have purchased a massive portion of the public offerings of shares, especially IPO’s,18 and a relevant number of these IPO’s were made by middle-sized and even small (net revenue below R$ 50 million) firms. Closing this overview, it should be highlighted the role that the National Economic and Social Development Bank (BNDES) and pension funds of state-owned or former state- 15 Since 2002, a Bovespa resolution has obliged that any new listing should be at least at the Level 1. 16 Around 66 percent of the 446 firms listed at the Bovespa were listed at the traditional segment on October 31st 2007, although representing just 41 percent of the value of the overall market capitalization. Level 1-listed firms showed an average market value (R$22.3 billion) well above those of any other listing segments (R$3.46 billion for the traditional segment, R$4.93 billion for the New Market, and R$4.02 billion for Level 2 (values as of 31 August, 2007). The leading Brazilian private economic groups have their main firms listed at the Level 1, such as CVRD, Votorantim Celulose e Papel, Gerdau S.A. and Metalúrgica Gerdau, Suzano Papel e Celulose, Sadia, Klabin, Iochpe, Duratex (Itaú Group), Braskem (Odebrecht), Aracruz, Alpargatas (Camargo Corrêa), Ultrapar, Brasil Telecom and Brasil Telecom Participações, Pão de Açúcar, Usiminas, the state-controlled Eletrobrás, and the largest banks (Banco Itaú Holding Financeira and Itaúsa, Bradesco and Bradespar, and Unibanco Holdings and Unibanco). TAM, Suzano Petroquímica, Sul América, Eletropaulo, Gol, Marcopolo, Net, Saraiva, Universo on Line, and the state-controlled Cemig and Cesp are listed at the Level 2, while CPFL Energia, Perdigão, Embraer, Camargo Corrêa Desenvolvimento Imobiliário and CCR, Light and the state-controlled Banco do Brasil (Federal Government) and Nossa Caixa (State of São Paulo) are listed at the New Market. 17 Between 2004 and October 2007, there were 102 IPO’s: 7 in 2004, 9 in 2005, 26 in 2006, and 60 until the end of October 2007. As a percentage of GDP, Bovespa-listed firms’ market capitalization value leapt from 36% in 2001 to 58% in 2005 to 74% in 2006 and to 130% in October 2007. In 2006, new share issues provided R$27 billion (direct financing tapped from Brazilian capital markets amounted to R$120.2 billion, the bulk of it from bonds, R$69.5 billion). 18 In 2006, they accounted for 67% of the total amount raised by new issues of shares.

7

owned companies’ employees have played on the corporate scene in the last decades. Brazilian private banking system, notwithstanding its robustness and sophistication, has recurrently been loath to provide long-term financing, gap which has been traditionally filled almost exclusively by BNDES and foreign loans, Brazilian capital markets having, until recently, a tiny contribution, mainly from bond issues. BNDES has been, since its inception in the 50s, a powerful policy instrument. Up to the end of the 70s, it operated as the main financial lever of industrial policies, being critical to guide capital allocation – be it conducted by state-owned enterprises or private sector investment – to the targeted industries, by ensuring low-cost, long-term financing to firms. During the stagnant 80s, BNDES and its wholly-owned subsidiary BNDES Participações S.A. (BNDESPar) coped mainly with financially-stressed firms, bailing them out through lending, converting their debt into equity, and capitalization.19 In the following decade, BNDES and BNDESPar gave top priority to privatization and restructuring (privatized firms included), playing a key role in making viable the building of some consortia and their successful bids. Over the last 10 years, BNDES’ flows of loans have been increasing, reaching R$ 52.3 billion, of which large firms have been the chief beneficiaries, absorbing at least 3/4 of their amount. With respect to sector allocation, most of loans have been directed to services (such as electricity, railways and highways, and telecommunications), motorcars, aircrafts, agribusiness, and food and beverage manufacturing.20

BNDESPar’s asset portfolio is predominantly concentrated in shares (almost 93% in terms of market value in 2006). Of its R$ 14.93 billion worth investment in shares at the end of 2006 (estimated market value of R$ 54.7 billion), mostly were in petrol and gas (31.7%) mining (20.6%), electrical energy (17.4%), steel (6.9%), paper and cellulose (6.1%) and telecommunications (5.3%). Bonds issued by firms from the electrical energy sector accounted for by 53.5% of its bond portfolio. At the end of 2006, BNDESPar held representatives in 8 fiscal committees and in 25 boards of directors and had shareholders’ agreement in 45 of the 138 firms wherein it had an equity stake.21

As regards pension funds, they amassed assets totaling R$ 394 billion in July 2007 (17% of the GDP), value that has been growing more than 15% a year since 1999. Most of their portfolios have been directly or indirectly held in government bonds and only around 20% directly in the stock market. The largest pension funds are those related to former or ongoing state-controlled firms: PREVI (Banco do Brasil), Petros (Petrobras), Funcef (Caixa Econômica Federal), Fundação Cesp, Valia (CVRD), and Sistel (Telebrás and Embratel). PREVI held a US$ 60 billion worth asset portfolio in July

19 In addition to lending directly to firms, BNDES also operates by means of its wholly-owned subsidiaries FINAME and BNDESPAR. The latter is a holding company investing in Brazilian firms through shares subscription and convertible bonds. FINAME is an agency specialized in financing machinery and equipment acquisition and export of capital goods and services with the purpose of supporting the expansion and modernization of Brazilian industry. 20 BNDES’ flows of loans have grown 187% from 1997 to 2006. Since 1997, the distribution of loans by sectors was concentrated in electricity, gas and water services (overall accounting for very high shares, such as 32% in 1997, except in 1999, 2000, and 2001), railways and highways (with an increasing share in the last three years, achieving 14% in 2006), telecommunications (a very high share between 1999 and 2001, peaking at 20.5% in 2000), motorcars (with an increasing share since 2003, and peaking at 10% in 2006), aircrafts (very high levels since 1998, with shares of more than 17% in 2002 and 2003), agribusiness (high volumes for the whole period, but peaking at 17% in 2004), and food and beverage manufacturing. 21 At the end of 2006, BNDESPar’s equity capital embraced among others the following firms: Eletrobrás (15%), Brasiliana (11%), Valepar (10%), Copel (8%), Petrobras (7%), ALL (6%), CVRD (5%). As for its bond portfolio, its main borrowers were: Light (29%), VBC Energia (14%), Paranapanema (15%), and Cia Metro SP (12%).

8

2007, most of which in government and corporate bonds (directly or via investment funds), but a significant fraction in shares, directly and through exclusive investment funds that own holding companies that, in turn, own stakes in almost all large firms, notably firms affiliated to business groups, in addition to firms not listed in the stock exchange. For example, PREVI holds Litel Participações S.A. directly and by means of exclusive investment funds (BB Carteira and BB Renda Fixa IV). In the same vein, through exclusive investment funds that it wholly owned (“521 Participações S.A.”, a special purpose entity set up with the sole objective of join the privatization process in the electrical energy sector), PREVI is a leading shareholder in Neonergia (thanks also through direct stake), CPFL Energia, and Itapebi. The next section describes the complex arrays by which BNDES, its subsidiary BNDESPar, and pension funds, notably PREVI, through privatization bids in association with Brazilian business groups and sometimes with foreign investors, managed to become important players in Brazilian corporate landscape (in terms of value of investments and the number of stakes in firms), and even members of the controlling coalitions in business groups such as CVRD and Brasil Telecom. Section II also deals with Brazilian business groups’ diversification, internationalization, and resilience.

2. Structures and Strategies of Brazilian Business Groups: A Brief Survey This section turns on some of the largest Brazilian business groups to grasp traits characterizing their structures and strategies over the last 25 years. Focus is placed on the following set of issues: 1) business groups’ ownership and control; 2) the evolution of their degree of diversification, in particular whether banking is or was one of their chief business lines; 3) the extent of their internationalization operations, either through trade or investment; and 4) their longevity and resilience to the changing economic and political environment.

2.1 Ownership and Control From among the sixty largest business groups operating in Brazil in 2005,22 25 were controlled either by Brazilian families or by privately-owned firms, seven by Brazilian governmental entities, 27 by foreign firms, and one jointly-controlled by a Brazilian family and a foreign group (see Appendix I).23 Privatization of formerly state-owned firms and public service concessions (such as mobile telecommunications) notably contributed to this profile.24 Worthy mentioning is that 16 of those top 26 Brazilian

22 Given the lack of systematic empirical research on Brazilian business group for the more recent period, we had to follow the ranking set forth by the specialized business newspaper Valor Grandes Grupos (2006), based on gross revenue. 23 Business group has a loose meaning for that newspaper: firms under the control of the same entity, regardless of operating in related or unrelated sectors. 24 From those 25 private Brazilian groups and from the 27 groups owned by foreigners, 10 and 9, respectively, resulted from privatization or public service concessions. State-owned firms whose control was transferred to domestic private groups were the following: CVRD, Usiminas, CSN, CPFL Energia, Embraer, Copesul, Light, and Neonergia; Telemar and Brasil Telecom emerged from the break-up and privatization of Telebras, the former Federal Government-owned, monopolistic firm in telecommunications services. Neoenergia began by acquiring some state-owned electric energy distribution companies. Curiously enough, Light, purchased by the French-government-controlled EDF when privatized in 1999, was sold in 2006 to a Brazilian holding company, in which Cemig, controlled by the State of Minas Gerais, is a leading shareholder. Foreign-owned groups whose entrance in Brazil is

9

groups (including the Brazilian-French controlled Pão de Açúcar) have controlling families,25 the 10 remaining groups being predominantly controlled by those same or other family-controlled groups, usually together with pension funds of state-owned firms’ employees (such as PREVI and Petros) or state-owned firms or agencies (Petrobras, BNDESPar, or Banco do Brasil), and sometimes with foreign groups. To a great extent this pattern can be accounted for by the main driving force motivating the privatization process, namely, the Federal Government’s firm intent of maximizing the revenues to be collected from the sale of its enterprises, reason for which it pushed for the active participation in the bids of BNDESPar and pension funds over which it had some leverage. Nonetheless, the legal framework regulating then the privatization process prevented state-owned firms’ pension funds from directly controlling the firms to be privatized. Bearing in mind the intricacy of the coalitions formed among Brazilian business groups, some pension funds, governmental agencies, and foreign investors (banks and investment funds) to partake in the bids as well as in the ownership and control arrangements that subsequently prevailed, some cases of privatized companies are examined below.

CVRD, privatized in 1997, was acquired by a consortium that transferred its control to Valepar S.A., a private company owned by government-related pension funds (PREVI and others), private Brazilian groups (Bradesco, Opportunity, and Vicunha), and BNDESPar.26 After a number of ownership restructurings,27 the resulting distribution of Valepar’s voting capital is the following (as of August 2007): Litel/Litela (controlled by PREVI): 49.0%; Bradespar (Bradesco Group): 21.2%; Mitsui: 18.2%; BNDESPar: 11.6%; and Opportunity: 0.03%. Thus, government-related entities controlled more than 60% of Valepar’s voting capital, which now owns 53.3% of CVRD’s voting capital (and 32.5% of its total capital).28 Moreover, Brazil’s Federal Government holds special class preferred shares in CVRD (“golden shares”) entitling it to veto proposals related to certain issues.29

related to privatization or public service concession bids are Telefonica, Santander/Banespa, AES Eletropaulo, Embratel, Endesa, EDP Energias do Brasil, TIM Brasil, Portugal Telecom, and Claro. 25 Brazilian business groups controlled by families are: Bradesco, Itaúsa, Ipiranga, Gerdau, Odebrecht, Unibanco, Votorantim, Pão de Açúcar, CSN, Sadia, Sul América, Camargo Corrêa, Safra, TAM, Suzano, and Ultra. 26 More precisely, Valepar’s ownership structure had the following composition: a) Litel Participações S.A., a special purpose entity created in 1995 to participate in the CVRD’s privatization auction and then to manage PREVI and other pension funds’ stakes in the newly-privatized CVRD; Elétron S.A., also a private company then owned by Bradesplan Participações S.A. (a Bradesco Group’s affiliate) and Opportunity Anafi Parts S.A., holding, respectively, 85.64% and 14.36% of its capital; CSN Steel Corp., a holding controlled by Companhia Siderúrgica Nacional (CSN), itself a former state-owned steel manufacturer privatized in 1993, and controlled by Vicunha Group; Sweet River Investment; and BNDESPar. 27 Noteworthy among those ownership rearrangements were: the elimination of the cross-shareholdings between CVRD and CSN in 2001; the incorporation of a large block of CVRD’s common shares held by Litel into Valepar’s equity capital through the increase in Valepar’s capital in 2002; the sale by Sweet River Investments of its 9.3% equity stake in Valepar to Bradespar and Litel Participações in March 2003; Mitsui & Co Ltd’s acquisition of a large block of Valepar’s common shares from Bradesplan Participações in September 2003; and Bradesplan’s purchase in March 2004 of 99% of the shareholdings Opportunity Anafi Parts. S.A. held in Elétron. 28 According to BNDES’s Annual Report, BNDESPar’s portfolio of shares totaled in December 2006 R$14.93 billion, of which 10% in Valepar, and another 5% in CVRD. 29 Veto power occurs in such issues as the liquidation of the company, the disposal of strategic activities for running CVRD’s iron ore mining integrated system, and any change in the rights assigned by its bylaws to the special class preferred shares.

10

Telemar’s control involves the alliance among private business groups, governmental institutions (BNDESPar and Banco do Brasil), pension funds, and a foreign investment fund. Almost the whole voting capital of Telemar Norte Leste S.A. (97.35%) is held by Tele Norte Leste Participações S.A. and its wholly-owned subsidiary Telemar Telecomunicações. Tele Norte Leste Participações S.A., in turn, is controlled by Telemar Participações S.A., which has a 53.8% of its outstanding common shares (and just 17.9% of its total capital). Finally, as of January 2007, Telemar Participações S.A.’s voting capital was held by AG Telecom Participações S.A. (affiliated to Andrade Gutierrez Group), Asseca Participações S.A. (owned by Grupo GP Investimentos), L. F. Tel S.A. (owned by La Fonte Group) and Lexpart Participações S.A. (jointly owned by Opportunity, Citigroup, and state-owned firms’ pension funds), each holding a 10.275% stake; Alutrens Participações S.A. (owned by Banco do Brasil and private-sector insurance firms), with 10%; Fundação Atlântico de Seguridade Social (Telemar’s employees’ pension funds), 4%; BNDESPar S.A., 25%; and Fiago Participações S.A. (owned by state-owned firms’ pension funds), 19.9%.

BNDESPar, Citigroup, Opportunity and pension funds are also the ultimate largest shareholders in Brasil Telecom S.A., another major telecommunication service company. Its common shares are almost entirely owned (99.07%) by Brasil Telecom Participações S.A., whose control is in the hands of Solpart Participações S.A. (which owns 51.00% of its common shares and 18.78% of its total capital).30 Solpart, in turn, is a holding company encompassing its controlling shareholder, Techold Participações S.A. (owning 61.98% of its total common shares), and Brasilco S.r.I (a trust, managed by Credit Suisse, of Telecom Italia International N.V.’s shares, representing 38.00% of Solpart’s overall common shares). Techold is fully owned by Invitel S.A., a holding company controlled by Zain Participações S.A. (with a 68.28% stake in its total common shares) and that has as minority shareholders PREVI (19.33% of its voting capital), Fundação 14 BrT (the pension fund of Brasil Telecom SA’s employees, with 6.27%), and the pension funds of Petrobras, Caixa Econômica Federal and Embratel’s employees (with, respectively, 3.78%, 2.24%, and 2.24%). Finally, Zain Participações S.A. is owned by Investidores Institucionais Fia (an investment fund comprising BNDESPar and some pension funds), with 45.85% of its voting capital, Citigroup Venture Capital International Brazil L.P., a foreign investment fund (42.47%), and Opportunity Fund (9.00%).

CPFL Energia has the idiosyncratic characteristic of being controlled by a private company (VBC Energia) owned by three family-controlled groups, an investment entity owned by PREVI (521 Participações S.A.), and a holding controlled by pension funds (Bonaire Participações S.A., involving Funcesp, Petros, Sistel and Sabesprev), on top of having BNDESPAR as a non-controlling shareholder. VBC Energia is owned by VBC Participações (99.9%), whose capital is held by Votorantim Energia Ltda., Bradesplan Participações S.A., and Camargo Corrêa Energia S.A., owned, respectively, by Votorantim Group, Bradespar S.A., and Camargo Corrêa Group.Usiminas is controlled by a voting coalition involving a foreign group (the Japanese Nippon Group, holding 24.7% of its voting capital in November 2006), a partnership of two Brazilian groups (Camargo Correa and Votorantim, each with a 11.6% stake), a pension fund (Usiminas’ employees, with 10.1%), and the privatized CVRD (5.9%). Despite not taking part in this agreement, Previ has a 10.4% interest in Usiminas.

30 Previ has a 5.14% stake in Brasil Telecom Participações S.A.’s common shares and 4.05% of its total capital.

11

Neoenergia’s voting capital is distributed among just three partners: Previ (holding directly and indirectly 49.01% of the common shares), Banco do Brasil-Banco de Investimento S/A, the giant government-controlled bank (owning directly and indirectly an 11.99% stake), and the Spanish Iberdrola (39.00%). Companhia Petroquímica do Sul (Copesul), privatized in 1992, had, until recently, its voting capital shared among two private groups (Ipiranga and Odebrecht, each with a 29.46% stake), a Petrobras’ wholly-owned subsidiary (Petrobras Química S.A. - Petroquisa, 15.63%), and banks, pension funds, and other investors (25.45%). In March 2007, Petrobras, Ultrapar Participações and Braskem (Odebrecht Group) reached an agreement by which Ipiranga Petroquímica’s stake in Copesul was divided between Braskem (60%) and Petrobras (40%).

Finally, Embraer, privatized in 1994 and now a leading global aircraft manufacturer and exporter, could be considered as an anomaly in the Brazilian corporate context, having a scattered capital ownership structure vis-à-vis other large Brazilian companies.31 Notwithstanding, government-related institutions are among its major shareholders (PREVI and BNDESPar has, respectively, 14.0% and 5.0%).32 Embraer’s by-laws restrain voting rights in any general shareholders’ meetings to the limit of 5% of the number of shares in the capital stock and the total votes by foreign shareholders, individually or collectively, to 40% of the total votes.33 In addition, no shareholder or group of shareholders may have a stake exceeding 35% of its capital stock.

2.2 Business Diversification Sector diversification varies significantly across the sample Brazilian privately-controlled groups. Some are clearly single-minded, such as Telemar and Brasil Telecom (telecommunications sector), Unibanco and Safra (finance), Sul América (insurance), Embraer (aircrafts), Neoenergia, CPFL Energia and Light (electrical energy), Sadia and Perdigão (chilled and frozen foods), Copesul (petrochemicals), TAM (civil aviation), and also Pão de Açúcar (food retailer), the French-Brazilian owned-group. Others have a core business but also joint ventures, subsidiaries or affiliates operating in vertically-related (backwards and/or forwards) industries,34 as are the cases of CVRD (specialized in minerals and certain metals, but also controlling enterprises in logistics, railways, steel, and energy), Ultra (fuel distribution, chemicals, petrochemicals, transportation and logistics), Ipiranga (the same businesses as Ultra’s plus oil refinery), Gerdau (steel, iron mining, energy, and banking), and Usiminas (steel, metal goods, mechanics, logistics and transportation). Last but not least, there are diversified economic groups conducting

31 Likewise, Perdigão displays a relatively widely-held capital relatively. Again, pension funds, in particular of state-controlled companies, hold the largest voting capital stakes: PREVI owns a 15.7% share, Petros 11.9%, Sistel 5.1%, BNDES-FAPES (BNDES’s pension fund) 3.7%, Librium Investments Fund 2.2%, Valia 4.1%, Real Grandeza 2.9%, and Previ-Banerj 1.2%. 32 A bank-based group, Bozano, is another major shareholder (8.6%), while the Brazilian Government owns “golden share” entitling it to veto power over the following corporate issues: development of third parties’ skills in technology for military programs, interruption of the supply of maintenance and replacement parts for military aircraft, and transfer of the equity control. 33 Brazilians’ control was a condition imposed by the privatization rules of Embraer. The free float shares traded at NYSE represent 55.2% of the total shares, and at Bovespa, 16.9%. 34 This may be associated with strategic reaction to uncertainties and contract imperfections. As Khanna and Yafeh (2007, 341) point out, “limited contract enforcement, weak rule of law, corruption, and an inefficient judicial system should all lead to high transaction costs between unrelated parties. Under such circumstances, intragroup trade, within the context of long run relationships supported by family and other social ties, may be relatively cheap and efficient.”

12

business in unrelated industries, like Odebrecht, Votorantim, Vicunha, Camargo Corrêa, and Suzano and the bank-centered Bradesco and Itaúsa.35 The sample state-owned Brazilian groups, excluding Petrobras, focus on single activities (either finance or electrical energy) and ancillary businesses.36 As a rule, Brazilian business groups are composed of several leading firms in the respective markets wherein they operate (de Siqueira, 2000)

Government policies have influenced, in a greater or lesser degree, the emergence, development and diversification of most Brazilian business groups.37 Besides policy incentives, those groups’ growth and diversification may also have resulted from deliberate strategies either to achieve scale and scope economies or to reduce risks (see Schneider, 2007). Probably a mix of these motivations lies behind their evolution. The account below tries to cast some light on the relative weight of these factors in the evolution of 25 leading Brazilian privately-owned groups.

To begin with, their diversification strategies have been far from convergent. Votorantim, Camargo Corrêa, Odebrecht, Bradesco, and Vicunha have clearly pursued the diversification of their businesses, even though relinquishing sometimes one of them to initiate another. Votorantim Group began in 1918 as a textile mill, and in the 70s run businesses in cement, aluminum, steel, mining, metals, pulp and paper, heavy equipments, sugar and alcohol, and textiles. Afterwards, the group disposed of its original textile business, and embarked on new activities, such as energy, concentrated orange juice, chemicals, banking, energy, and, more recently, biotechnology and information technology. In the same vein, Odebrecht, founded in 1944 as a civil construction enterprise, has later broadened its range of activities to more technologically sophisticated heavy construction projects (such as light-rail systems, nuclear power plants, airports and large bridges), chemicals and petrochemicals in the 80s (thanks to privatization, the group would be catapulted in the following years into the leadership among private-owned Brazilian petrochemical firms), and infrastructure and public services concession (highways and basic sanitation services). Vicunha Group, whose origins traces back to a weaving mill in 1946, turned into the largest Brazilian textile group (Vicunha Têxtil), which since the 80s has branched out its

35 Odebrecht carries out businesses in engineering and construction, chemicals and petrochemicals, infrastructure and public services concession (highways and basic sanitation services), mining and oil ventures, sugar and ethanol, and insurance. Votorantim is engaged in cement and concrete, pulp and paper, energy, agribusiness, chemicals, mining and metals, banking and financial services, biotechnology and information technology. Vicunha Group, owned by the Steinbruch Family, controls CSN, the former state-owned steel-maker, Vicunha Têxtil and other textile firms, the bank Fibra, and firms in the cement, energy, railway, logistics, and trading sectors. Camargo Corrêa undertakes businesses in engineering and construction (infrastructure, buildings, shipbuilding, and real estate development), cement, footwear (Alpargatas), textiles (Tavex/Santista), electrical energy (CPFL Energia), highway concessions (CCR), steel and metals, besides holding an 11.35% stake in Itaúsa’s voting capital. Suzano until recently had plants in the paper and pulp and petrochemical sectors. Bradesco, notwithstanding banking being its core business, holds, through its financial holding Bradespar, sizeable investments in other industries’ firms, like CVRD, CPFL Energia, Scopus (an information technology firm), and Globo Cabo (a cable TV service firm). Itaúsa is a holding company controlling several banking and financial institutions, and firms in segments like real estate, wood paneling (Duratex), ceramic and metal sanitary goods (Deca), chemicals (Elekeiroz), and electronics (information technology, computers and software). 36 Petrobras has undertakings in oil and gas exploration, production, and distribution, energy and petrochemicals. 37 These policies have had different manifestations along the time. For example, trade protection favoring light industries before the II World War, more systematic state-led strategies of import substitution industrialization from the 40s to 70s, government procurement, and financial support for privatization bids, restructuring, modernization, and internationalization in the last two decades.

13

interests into several industries, like mining, steel (CSN), cement, energy, railway, logistics, trading, and banking and financial services (Banco Fibra, since 1989).38 Camargo Corrêa Group, originally started as a small construction company in 1939, had in the 70s diversified its activities to heavy civil construction, real estate, agribusiness, cement, metals, mining, textiles, and finance. It has retained its businesses in engineering and construction, real estate, cement, textiles, and mining, and now runs enterprises in industries such as shipbuilding, highway concessions, environmental engineering, and footwear. In addition, it is a member of the controlling block in CPFL Energia, CCR (which operates the main Brazilian highway concessions), and Usiminas, and has a significant stake in Itaúsa.39

Bradesco likely envisioned the privatization of some major state-owned enterprises in the 90s as an opportunity for diversifying its portfolio, since the price stabilization, finally achieved thanks to Plano Real, would dramatically reduce the inflationary tax appropriated by banks. In fact, in association with other Brazilian private groups, pension funds, and BNDESPar, it was a major player in various privatization auctions, after which it became part of the controlling coalition of groups such as CVRD and CPFL Energia. Bradespar also held a stake, together with Previ, in CSN, but in 2001 both agreed to sell it to Vicunha Siderurgia S.A. in exchange for CSN selling its stake in Valepar S.A. to Bradespar S.A. and Litel Participações S.A., thus promoting the elimination of the cross-shareholding interests between CVRD and CSN.

CVRD and Gerdau have followed the strategy of marginal diversification around their major businesses. Gerdau Group’s main activity has persisted in steel production but also set up a bank in 1994 (Banco Gerdau) and, with a view to verticalization, has extended business to iron mining and manufacturing (since 1986), and more recently to electrical energy (as a reaction to the energy crisis in Brazil). CVRD Group has maintained its focus on the exploration and production of iron ore and pellets, generally preserving its enterprises in closely related sectors, like steel, railways, logistics, and energy. It has also diversified within the mining and metal sectors, beginning to operate with copper and nickel. The global acute competition in the mining sector has led CVRD to search for gains in scale and scope, prompting it to acquisitions totaling US$ 24.5 billion since 2000, out of which US$18.243 billion for the purchase of Inco in 2006. Conversely, divestitures equivalent to US$ 3.8 billion were made in steel (it sold its stakes in Açominas, CSN, CST, and a large block of shares in Usiminas, Gerdau and Siderar), pulp and paper (Bahia Sul and Cenibra), fertilizers, ships and wood.

Ipiranga and Suzano groups have refocused their businesses after a period of diversification. Ipiranga Group started operating in oil production, petrochemicals, and fuel distribution, but in the 70s its businesses also included hotels, agribusiness, fishing, and transportation. Sharp competition in the 90s brought about by trade liberalization led Ipiranga to dispose of its peripheral activities, and to return to its original activities. Suzano Group, in turn, producing paper and pulp since the 1920s, diversified into petrochemicals since 1974. After strengthening its petrochemical branch, notably

38 After acquiring a relevant stake in CSN, the former state-owned steel maker privatized in 1993, the group’s two controlling families (Steinbruch and Rabinovitch) became its largest ultimate shareholder. Their progressive control of CSN was reached by dint of increasing indebtedness, financed to a great extent by BNDES. 39 Some former secondary businesses (such as those in the energy, textile, cement, and footwear sectors) have turned into the group’s leading activities, notwithstanding engineering and construction still accounting for around 17% of its revenue. Privatization also influenced the group’s investment strategy, propelling it to acquire interests in energy, transportation sectors, and steel (Usiminas).

14

through privatization bids in the 90s, Suzano Petroquímica has been recently friendly taken over by Petrobras, the state owned, largest Brazilian company.40

With respect to the existence of a bank in the group structure, out of the 25 largest Brazilian private groups in 2005, four had banking as their core business, three had banks among their affiliates (Vicunha Group, Gerdau Group, and Votorantim Group), two had a bank, Bradesco, taking part in the controlling coalition (CVRD and CPFL), three had banks among their non-controlling shareholders (Telemar, Neoenergia, and Embraer), and one had a substantial, but apparently passive, stake in a large bank (Camargo Corrêa).

Overall, it can be observed an effort or willingness of most of the Brazilian business groups to strengthen their position in a few markets, to a great extent spurred by trade liberalization, deregulation, and acute global competition. Domestic and foreign acquisitions/disposals have been carried out with a view to consolidating their shares and pursuing economies of scale in a more limited range of industries. Diversification seems to be now guided primarily into technologically-related industries. Recent moves in Brazilian petrochemical sector may be illustrative of the implications unleashed by the changing competition context. Consolidation has been achieved by alliances between privately- and state-controlled groups to takeover petrochemical firms through either privatization bids or “private” acquisitions of private groups, in which BNDESPar and Petrobras have had a key role. And also by the acknowledgment of some private groups (Suzano and Ipiranga) that to not be an easy prey, they would have either to gain market share (via acquisitions or greenfield investments) or jettisoning their assets in that sector to other major players.

2.3 Internationalization: Trade and Foreign Direct Investment Firms’ export ability may be a sign of multifarious aspects: their efficiency and technological competence, their incentive to continuous process and product innovation, lower vulnerability to domestic business cycles, and easier and cheaper access to foreign financing. Indeed, Brazilian groups that have had an active presence in the international goods and service markets have tapped the international capital markets, been subject to lower volatility in earnings, and more willing to efficiency-enhancement investments. Exports from the ten Brazilian business groups that most exported in 2006

40 This takeover has been reproached by other leading Brazilian players in petrochemical industry, arguing that re-nationalization is underway. Nonetheless, Braskem’s CEO, who vehemently vocalized against this deal, also led, together with Petrobras and Ultra, the recent purchase of Ipiranga. Probably, the strong interest in the purchase of Ipiranga’s control revealed by foreign groups (such as Repsol, Agip, the French Total Fina, the British Petroleum, and the American ExxonMobil) may have facilitated the negotiation among the three Brazilian groups. Joint enterprises between privately-owned and state-owned firms are recurrent in Brazilian petrochemical industry. For example, Suzano Petroquímica participated in a joint venture in Riopol (the first Brazilian natural gas-based petrochemical complex, inaugurated in 2005), holding a 33.33% stake, while two state-controlled entities, Petroquisa and BNDESPar, held, each, a 16.67% stake.

15

amounted to almost US$ 35 billion, accounting for 26% of the country’s exports.41 Export values as a fraction of firms’ total sales vary sharply from firm to firm.42

Some large Brazilian business groups have also pursued internationalization through foreign direct investment. Main motivations have been protectionism (when FDI functions as a substitute for exports, like Gerdau’s steelworks in the USA), access to raw materials and new markets (as are the cases of many CVRD’s subsidiaries), and incentives coming from the free trade area Mercosur. CVRD and Gerdau seem to be the Brazilian groups that have mostly advanced in this front. CVRD owns joint ventures, subsidiaries and affiliates with operating plants in countries such as Canada (through the purchase of Inco in 2006), France, Chile, Colombia, South Korea, Argentine, Australia, Norway, Peru, the USA, Japan, Spain, and India. Gerdau Group began to invest abroad in steelworks in 1980, and subsequently acquired steel plants in Canada (one in 1989 and other in 1995), Chile (one in 1992 and other in 1999), Argentina (1997), and AmeriSteel, a leading steel producer in the USA. In 2003, Gerdau Group and the Canadian firm Co-Steel undertook a merger of their steel businesses in North America, involving 10 plants, by which Gerdau held a 67.5% interest in the new firm, Gerdau Ameristeel Corporation, and the plain management control. Since then, this corporation has acquired a number of other steel-makers in the USA (10 only in 2007), culminating with the acquisition of Chaparral Steel Company, a US$ 4.22 billion deal worth. Clearly this foray in the American market paralleled to protectionism against steel imports, disguised as anti-dumping measures. In 2005, Gerdau Group extended its operations to Spain, wherein, jointly with Santander, it acquired Corporación Sidenor S.A., one of the leading steel-maker in that country, and in 2007, taking its first step in Asia, it established a joint venture with Kalyani.

Camargo Corrêa, Votorantim, Odebrecht, Ultrapar Participações S.A., and Ipiranga appear as moderate foreign direct investors. Votorantim Group only began to extend production facilities overseas in 2001 and afterwards has undertaken foreign direct investments in the cement, metals, paper and pulp, agribusiness, and financial sectors. Ipiranga has one petrochemical plant in Chile and Ultrapar has three units producing chemical specialties in Mexico since 2003, one chemical plant acquired in 2007 in Venezuela, besides units in Argentina and the United States. Camargo Corrêa’s direct investments abroad began with the acquisition of a cement manufacturer in Argentina in 2005 (Loma Negra), and in the following year Santista Têxtil was merged with Tavex Algodonera, a Spanish textile firm. Odebrecht acquired, in 1988, a Portuguese firm in the heavy construction sector and incorporated in 1992 the British firm SLP Engineering, a firm specialized in providing accommodation modules for offshore platforms. The remaining groups have a limited amount of FDI, most of which represented by trading and commercial offices. Specifically, the three largest Brazilian banks have a modest international standing as compared either with its domestic operations or with foreign banks of similar scale. 41 Brazilian groups’ leading exporters were Petrobras Group (US$11.9 billion, tantamount to an 8.7% share of overall Brazilian exports), CVRD (US$11.6 billion, representing a 8.43% share), Embraer (US$ 3.3 billion and a 2.38% share), Votorantim (US$1.6 billion and a 1.1% share), Usiminas (US$1.393 billion and a 1.2% share), Sadia (US$1.316 and a 0.96% share), Odebrecht (US$1.275 billion and a 0.93% share), Gerdau (US$1.2 billion and a 0.87% share), Copersucar (US$1.068 billion and a 0.78% share), and Perdigão (US$915 million and a 0.67% share). See Análise Comércio Exterior (2007). 42 According to Exame (2006), among the top exporter firms in 2005, export as a percentage of total sales was 97.2% for Embraer, 99.5% for Samarco (which belongs to CVRD), 93.3% for Aracruz (pulp and paper), 64.6% for CVRD, 69.9% for Odebrecht Construction, 46.9% for Sadia, 44.5% for Perdigão, 42.0% for Suzano Papel e Celulose, 29.3% for Gerdau Açominas, 20.8% for CSN, 15.2% for Petrobras, and 13.6% for Usiminas.

16

Conversely, it should be noted that foreign investors have significant stakes in some Brazilian groups. The Portuguese Banco Espírito Santo and the Spanish Banco Bilbao Vizcaya Argentaria have, respectively, 5.94%, and 5.06 of Banco Bradesco’s voting capital. In the mining and steel sectors, Mitsui had 18.2% of Valepar’s voting capital, which owned 53.3% of CVRD’s voting capital in August 2007, while Nippon Group has a 24.7% stake in Usiminas’ voting capital. Casino Group co-controls Pão de Açúcar Group jointly with the Diniz Family. Iberdrola Energia S/A (wholly-controlled by Iberdrola) holds a direct 39.00% stake in Neoenergia. And the Dutch bank ING holds directly and indirectly 49% of Sul América’s total capital.

2.4 Longevity and Resilience Most of the sample business groups founded by Brazilians are long-standing: among the 26 private Brazilian groups in 2005, 19 emerged before 1950 and four between the 50s and 1976.43 A conventional, and rough, indicator of firms’ resilience in a given country is the percentage of the top largest firms in a certain year that still remain the top largest some years later. Reckoning this “survival rate” for the top 40, top 20, and top 10 Brazilian private groups (ranked by equity capital) for three periods, the following results are found:

a) From among the 40 largest Brazilian private groups in 1978, 23 (57%) remained among the top 40 in 1988;

b) From the top 40 in 1988, 19 (47%) persisted among the top 40 in 1998;

c) From the top 40 in 1998, 27 (67%) stayed in the top 40 list in 2005;

d) From the top 40 in 1978, 12 (30%) were in the top 40 list in 2005.44

As compared with the list of the top 20 and the top 10 groups, the above survival rates show major discrepancies only for the top 10 between 1978 and 1988 (40%) and between 1998 and 2005 (90%).45 Under a longer horizon perspective (from 1978 to 2005), the survival rate is not high, casting doubt on political economy explanations of business groups’ persistence. At least after the return to democracy, even being considerable the influence of private business groups on government policies and politics, those failing economically also have failed politically to eschew bankruptcy. Traditional and economically powerful business groups in the 70s, like Bonfiglioli, Matarazzo and Comind, respectively, the third, sixth, and eighteenth largest business groups in terms of net worth in 1978, had disappeared in the following decade. Likewise, Itamarati Group (the top 13 group in 1988, operating in heavy construction, agribusiness, and finance), Bamerindus (top 6 in 1988), Econômico (top 24 in 1988),

43 The remaining firms, Telemar, Brasil Telecom, and Neoenergia stemmed from the privatization in the 90s of state-owned firms that probably were created before the 1950s. The four companies created during the period 1950-1976 are Usiminas, Embraer, TAM and Copesul. 44 Another source (Balanço Anual, Gazeta Mercantil), with data available since 1978, had to be used to calculate those figures. The 12 Brazilian private groups that appear at the top 40 list in both 1978 and 2005 are Bradesco, Itaúsa, Gerdau, Unibanco, Votorantim, Pão de Açúcar, Sadia, Camargo Corrêa, Sul América, Andrade Gutierrez, Klabin, and Cosan (formerly Dedini). 45 10 from among the top 20 groups in 1978 still were among the top 20 groups in 1988 (4 among the top 10); from among the top 20 groups in 1988, 10 were among the top 20 groups in 1998 (4 for the top 10); from among the top 20 groups in 1998, 15 groups still were in the top 20 group list in 2005 (9 for the top 10); and for the top 20 groups in 1978, 6 persisted among the top 20 groups in 2005 (4 for the top 10).

17

and Nacional (top 32 in 1978) did not survive the turbulent 90s, notwithstanding their controlling shareholders’ intimate relations with politicians.46

As regards what happened afterwards to the 40 largest Brazilian private business groups in 1978, 19 still existed as a Brazilian group in 2005, four had been acquired by other Brazilian groups, four had been acquired by foreign groups, three had gone bust and later purchased by other Brazilian groups (two) or by a foreign group (one), and 10 had bankrupted. Among those 13 that were top 40 both in 1978 and 2005, four were and are primarily engaged in finance, three in capital-intensive mining and manufactured commodities (steel, paper and pulp, metals, chemical products), two in heavy construction, two in agribusiness, and one in food retail.

In 2005, none among the 100 largest Brazilian groups operated in a significant scale in electronics, carmaker or other knowledge- or skilled-labor-intensive industries, except the aircraft manufacturer Embraer.47 Major state-owned enterprises are concentrated on the energy, gas and oil, and banking sectors. Banks stand out among the top ten Brazilian groups (Bradesco, Banco do Brasil, Itaúsa, and Caixa Econômica Federal), followed by petrol production and distribution (Petrobras and Ipiranga), mining (CVRD), steel and iron (Gerdau), telecommunications (Telemar), and the diversified group Odebrecht (mainly chemicals and petrochemicals and heavy construction).48

3. The Governance of Large Firms in Brazil This section presents some empirical evidence on the structures of public companies’ ownership and control in Brazil and their controlling shareholders’ characteristics as well as on possible relationships among those variables and between them and their performance.49 The sample covers each year over the period 1997-2002 and includes not just listed firms at the Brazilian equity market but the whole set of public companies

46 Bankruptcy was not avoided, despite political and personal pressures to bail them out, in the cases of the banks Econômico, Nacional, and Bamerindus. Econômico, a 161 year bank, was controlled by a close friend of Antonio Carlos Magalhães, a leading politician with a powerful influence on the national politics. A member of Nacional’s controlling family was the President Cardoso’s daughter-in-law while Bamerindus’ main shareholder was Cardoso’s Minister. 47 Most of them were engaged mainly in the following core businesses: banking, chemical and petrochemical, mining and metals, telecommunications, energy, pulp and paper, retail commerce, food, insurance, agribusiness, engineering and heavy construction. 48 Taking into account the top 20 Brazilian groups, the composition by industry is five for banking (adding Unibanco to the other four banking groups among the top ten groups), three for steel (adding Usiminas and CSN), three for electrical energy (adding Eletrobrás, Ceming, and CPFL Energia), two for telecommunications (adding Brasil Telecom), two for oil and gas production and distribution and petrochemical sectors, two for diversified groups (adding Votorantim), and one, respectively, to mining, aircrafts (Embraer), and food retail (Pão de Açúcar). See Valor Grandes Grupos (2006). 49 Except for the analysis of the relationships among variables, the account set forth below draws extensively on Aldrighi and Mazzer Neto (2007) who, for their research, constructed a data set out of the “Informativo Anual” (Annual Information Bulletin, IAN) and the “Demonstrações Financeiras Padronizadas” (Standardized Financial Statements, DFP) – reports that public companies in Brazil are required to file annually to the Brazilian capital market watchdog, “Comissão de Valores Imobiliários” (CVM). Among other types of data, the IAN discloses: i) the shares of capital held by the largest shareholders; ii) the identities of directors and top executives; iii) directors’ compensation and whether they have participation in the company’s capital; iv) whether there is any agreement among a group of shareholders; v) whether the company issued preferred stocks with no voting rights; and vi) general information, such as the company’s main activity. The DFP reports: i) the balance sheet; ii) the income statement; iii) the sources and uses of funds; iv) net worth changes; and v) the board’s report.

18

that furnished both these forms to the CVM.50 Even though largely relied on the corresponding literature, the key concepts and criteria to deal with the data took into account some Brazilian corporate environment peculiarities. The first step was to identify for each sample company for each year the “largest ultimate shareholder” (henceforth LUS), defined as the shareholder holding the largest voting stake, directly and indirectly, in the firm. A firm’s indirect ownership or pyramidal ownership means that there is at least one firm between the firm under scrutiny and its corresponding ultimate shareholder.51 The largest ultimate shareholders were classified into the following categories: i) an individual or a family; ii) the state or governmental agencies; iii) a foreign firm; iv) a mutual fund; v) a pension fund; vi) a foundation or a cooperative; vii) a privately-held company; and viii) a legally-registered voting agreement among shareholders.

The analysis of that data set found the following results:

(1) On average, 52.4% of the sampled firms were involved in pyramidal arrangements. Despite the shrinking number of firms under pyramidal ownership over the period, they still represented 49.3 percent of the sample in 2002. Table 2 evidences that the frequency of pyramids is higher in paper, cellulose and lumber (78.8%) and telecommunications (72.6%).

Defining degree of pyramidal structure (DPS) as the number of intermediate companies between the firm under scrutiny and the LUS, the average DPS for the sampled firms with pyramidal ownership structures (POS) is 2.06. For firms with POS in 2002, more than 50 percent presented a DPS equal to 1, 22.4 percent a DPS of 2, and 13.4 percent a DPS of 3.52 Table 3 displays the distribution of companies according to the degree of pyramidal structure from 1997 to 2002.

Almeida and Wolfenzon’s theory implies that pyramidal ownership schemes may occur even though the controlling families have not exhausted all the legal possibilities to issue non-voting shares, the simplest and straightforward way to separate voting power from ownership capital. To assess this prediction, firms involved in pyramidal

50 The number of firms that complied with the mandatory requirement of filing these forms was 670 in 1997, 836 in 1998, 807 in 1999, 772 in 2000, 727 in 2001 and 666 in 2002, totalling 4,478 IANs and the same number of DFP. 51 The calculation of the ultimate shareholder’s stake in the firm’s cash-flow rights depends on whether it has a pyramidal ownership structure. If not, it equals to the ratio between the number of shares she owns – the sum of her ON and PN shares – and the firm’s total number of shares outstanding. Having pyramidal ownership, the stake in cash-flow rights is determined by the product of the percentages of shareholdings each entity (a family or another firm) has in the next one. The calculation of voting rights depends on whether the LUS is also a controlling shareholder. For distinguishing widely-held firms from those with an ultimate controlling shareholder, it was adopted the 50 percent control cutoff. Hence a firm is said to have an ultimate controlling shareholder if it has a shareholder controlling directly and indirectly at least 50 percent of the firm’s total voting rights. If there is no controlling shareholder, the reckoning is analogous to that for cash-flow rights, that is, the product of the percentages of voting rights along the ownership chain. Contrariwise, if the ultimate shareholder is also a controlling shareholder, the percentage of voting rights is determined by the percentage of voting rights the last intermediate along the control chain has in the sample firm. For details of this methodology, see Aldrighi and Mazzer Neto (2007). For La Porta et al. (1999), a firm has a controlling shareholder if a shareholder has direct and indirect voting rights exceeding 10 percent or 20 percent. Studying East Asian companies’ ownership and voting structures, Claessens et al. (2002) use, besides those two cutoff points, the 40 percent control cutoff. Despite much higher than those assumed by the literature, the 50 percent threshold was chosen because it ensures mathematically the firm control and on the grounds of the prevalence of Brazilian public firms with an ultimate shareholder holding a more than 50 percent fraction of its voting rights. 52 In 2002, there were two firms with aDPS of 8 and one with a DPS of 9.

19

ownership arrangements were classified according to the fraction of non-voting shares in their overall capital over the period 1997-2002. Table 4 evinces that 29 percent of the sample firms with pyramidal ownership had issued non-voting shares tantamount to more than 60 percent of the overall shares and 53 percent for pyramidal firms with non-voting shares exceeding more than 40 percent of their respective total shares. Both these fractions cast doubt on Almeida and Wolfenzon’s contention. Nonetheless, the evidence that more than 27 percent of the pyramidal-owned firms issued only voting shares gives some support for their claim.

Table 2 Firms with Pyramidal Ownership by Sector (in %)

Industry % Agriculture and livestock 63.6 Food and beverage 45.5 Textile and leather 41.4 Paper, cellulose and lumber 78.8 Oil 57.9 Chemical, rubber and plastic 62.2 Non-metal minerals, pottery, cement and glass 60.5 Metallurgy, metallic minerals 47.8 Electricity, gas and sewage 47.7 Building 66.4 Transport and storage 64.9 Post office and telecommunications 72.6 Banks 55.6 Average 52.35

Table 3 Degree of Pyramidal Structure: 1997-2002 (in percentage)

DPS 1997 1998 1999 2000 2001 2002 Average 0 48.28 46.43 45.41 47.92 47.83 50.62 47.62 1 25.77 26.71 22.71 25.12 25.60 25.13 25.16 2 13.43 12.31 13.10 12.02 12.72 11.05 12.45 3 5.44 7.83 9.75 7.40 4.83 6.60 7.08 4 4.17 4.06 4.22 3.24 4.03 3.57 3.88 5 2.18 1.54 2.04 1.69 3.22 0.89 1.93 6 0.36 0.98 1.89 1.69 1.29 1.25 1.27 7 0.18 0.00 0.73 0.46 0.16 0.36 0.32 8 0.18 0.14 0.00 0.31 0.16 0.36 0.18 9 0.00 0.00 0.15 0.15 0.16 0.18 0.11

Furthermore, Almeida e Wolfenson assert that their theory, unlike the traditional view, fits recent empirical evidence showing that families controlling firms by means of pyramids hold high stakes in the firm’s capital, entailing small deviation between voting and cash-flow rights. The results found are inconclusive regarding the cash-flow rights of the largest ultimate shareholders. Although average cash-flow rights are significantly lower in pyramidal firms than in non-pyramidal firms, at least in firms with degrees of pyramidal structure ranging from 1 to 3 the LUS’s average stake in the firm’s capital is higher than 33 percent, being 47.9 percent for firms with a DPS equal to 1 (Table 5).

20

Table 4 Distribution of Pyramidal-Owned Public Companies According to the

Fraction of Non-Voting Capital 1997-2002

Fraction of Non-Voting Capital % of CompaniesEqual to 0 27.3> 0 e ≤ 20 8.8> 20 e ≤ 40 10.9> 40 e ≤ 45 4.7> 45 e ≤ 50 9.1> 50 e ≤ 55 4.7> 55 e ≤ 60 5.6> 60 e ≤ 66.67 28.2> 66.67 0.8TOTAL 100.0

Table 5

LUS’s Average Cash-Flow Rights and Degree of Pyramidal Structure (%) DPS 1997 1998 1999 2000 2001 2002 1997-02

0 59.5 60.6 63.0 62.5 65.6 66.3 62.9 1 45.1 47.4 51.0 46.5 48.7 48.9 47.9 2 37.2 38.7 38.0 41.5 40.1 41.8 39.4 3 34.6 33.7 33.7 33.9 32.2 30.8 33.3 4 24.6 31.4 24.5 25.8 23.0 35.7 27.4 5 19.4 17.9 24.7 11.8 27.9 12.2 20.6 6 25.8 34.8 29.0 22.8 31.7 23.6 27.9 7 42.1 10.7 5.8 30.2 73.7 11.2 20.6 8 24.9 4.1 - 59.3 57.8 35.6 39.4 9 - - 4.4 16.3 16.3 15.3 13.1

TOTAL 49.3 50.4 50.98 51.2 53.5 54.9 51.6

(2) The LUS holds more than 50% of the firm’s capital in 48% of the firms. More than 18% of the firms have a LUS owning a stake in the firm’s capital exceeding 90%.

(3) 77.6 percent of the sampled firms have one controlling shareholder, that is, a shareholder holding more than 50 percent of the voting capital, while 39 percent have a LUS owning more than 90 percent of the firm’s voting capital. These percentages are very high as compared with those obtained by Claessens et al. (2002, p. 2748).53

(4) There exists a meaningful discrepancy between cash-flow rights and voting rights. While the LUS’s cash-flow rights are, on average, 51.0 percent, LUS’s voting rights is 73.3 percent, implying an average wedge between these rights of 22.3 percentage points. The highest average gap (27.1 percent) occurs for families, notably owing to those involved in pyramidal ownership (32.4 percent). Listed companies show a slightly higher average wedge of rights vis-à-vis the average public company (see Table 6).

(5) There are significant inter-industry discrepancies for the largest ultimate shareholders’ average cash-flow rights. Whereas means for transport and banking industries (respectively, 64.9 and 60.4 percent) exceed significantly the overall mean (51.0 percent), the opposite happens for telecommunications and petroleum (30.2 and 53 For a 40 percent control cut-off, only 23 percent of Claessens et al.’s sample of Asian companies had a controlling shareholder. Following the 20 percent threshold to define control adopted by La Porta et al. (1999), less than 6 percent of our sample do not have a controlling ultimate shareholder. For a 10 percent cut-off, 99.6 percent of the whole firms of the sample have a controlling shareholder.

21

33.2 percent). For the largest ultimate shareholders’ average voting rights, inter-industry disparities are lower: commerce (83.7 percent), agriculture (78.1 percent), public utilities (77.8 percent) and banking (76.3 percent) present means above the overall mean (73.3 percent), which in turn is much higher than that for petroleum (62.0 percent). As regards the LUS’s separation of rights, the leaders are telecommunications (averaging 44.0 percent, but peaking at 52.8 percent in 1999), paper and cellulose (33.4 percent) and agriculture (37.5 percent), while the smallest wedges are observed in cement and glass (13.9 percent), transport (12.8 percent) and banking (16.0 percent).

Table 6 Cash-Flow Rights, Voting Rights, and Discrepancies between Rights

Public Companies and Listed Firms Mean over 1997-2002 (Percentage)

Variable Type Mean

All Public Firms 73.3Voting Rights

Listed 70.4

All Public Firms 51.0Cash-Flow Rights

Listed 46.1

All Public Firms 22.3Divergence

Listed 24.3

(6) Families predominate among the largest ultimate shareholders, averaging 54.9 percent of them over the period 1997-2002. Foreign investors (averaging 19.0 percent), governments (8.2 percent) and mutual funds (3.7 percent) followed in importance (see Table 7). Over the same period, the share of the various levels of government and of governmental agencies in the total number of the largest ultimate shareholders loses ground to mutual funds and foreign investors.

Table 7 Identity of the Largest Ultimate Shareholders

(Percentage) Category of the LUS TOTAL

Government 8.2

Foreign Investors 19.0

Families 54.9

Foundations 1.8

Mutual Funds 3.7

Pension Funds 2.2

Privately-Held Firms 3.6

Others 6.6

Total 100.0

22

(7) In 55.9 percent of the firms, the difference between cash-flow rights and voting rights for the LUS exceeds 10 percentage points. There occurs deviation higher than 25 and 50 percentage points in, respectively, 36.2 percent and 13.2 percent of the sample firms (and in, respectively, 42.4 percent and 15.5 percent of the listed companies). This wedge is higher for firms with pyramidal ownership (mean of 30.6 percent) as compared with non-pyramidal ones (mean of 13.3 percent). The industries presenting the highest discrepancies of rights for the LUS are telecommunications (averaging 39.0 percent), paper and cellulose (33.4 percent), agriculture (31.5 percent) and petroleum (28.2 percent).

Table 6 Distribution of the Largest Ultimate Shareholders’ Wedge of Rights

Mean (percentage): 1997-2002 Percentage of the Firm’s Capital Public Companies Listed Companies

Less than 0% 3.8 4.7

Equal to 0% (no discrepancy) 20.7 10.3

> 0 and ≤ 10% 19.6 20.1

> 10% and ≤ 25% 19.7 22.5

> 25% and ≤ 50% 23.0 26.9

Total for ≤ 50% 86.8 84.5

> 50% and ≤ 75% 9.0 11.9

> 75% 4.3 3.5

Total for > 50% 13.2 15.5

Total 100.0 100.0

Aiming at investigating the relationship between ownership structure, governance, and performance, a logit model was estimated.54 The results, shown in table 9, evaluate the likelihood of observing a pyramidal ownership structure according to the following firm’s observable characteristics: (a) Size, measured as the natural logarithm of assets; (b) Performance, measured as the net profit/assets ratio; (c) Non-voting shares as a percentage of the firm’s total capital, as a proxy for the wedge between voting rights and cash-flow rights for the largest ultimate shareholder; (d) Leverage, represented by the ratio between total liabilities and total assets, as a variable to capture how easy the firm is able to access funds for financing; (e) Sectors of activity, a dummy variable; and (f) Types of the largest ultimate shareholder, a dummy variable to grasp the LUS’s identity.

54 Being y a dummy variable whose value is one if the company has a pyramidal ownership structure, the logit tries to estimate E[y = 1|x], in which x is a vector with both explanatory and control variables. By definition, E(y = 1|x) = {1 * Pr(y = 1|x) + 0 * Pr(y = 0|x)} = Pr(y = 1/x). In the logit model, probability Pr(.) has logistic distribution, such th t a

)()exp(1

)exp()1Pr()( xmx

xxyxyE ββ

β=

+===

Coefficients β describe the impact of independent variables on the probability of dummy variable to be one. They are estimated by Maximum Likelihood.

23

Table 7 Logit Results55

Dependent variable Pyramid (yit = 1)

Coefficient (Standard deviation)

Independent variables Constant -2.49 (0.420)** Percentage of non-voting shares 0.124 (0.152) Performance: Net Profit/Assets -0.388 (-2.55)* Leverage 0.028 (0.034) Size (log of assets) 0.089 (0.018)** Time dummies Yes

Sectors Agriculture and livestock -0.068 (0.381) Food and beverage -0.715 (0.186)** Textile and leather -1.035 (0.178)** Paper, cellulose and lumber 0.471 (0.248) Oil production -0.144 (0.218) Chemical, rubber and plastic 0.094 (0.189) Non-metal minerals, pottery, cement and glass 0.145 (0.275) Metallurgy, metallic minerals -0.580 (0.153)** Machinery, electrical equipment, cars -0.865 (0.225)** Technology, electronics, optics -0.515 (0.218)* Electricity, gas and sewage 0.205 (0.160) Building 0.013 (0.204) Commerce -0.580 (0.222)** Transport and storage 0.285 (0.200) Lodging and meals 0.046 (0.361) Post office and telecommunication 0.831 (0.182)** Holdings -0.928 (0.138)** Health and social service -0.336 (0.501) Education and culture -0.436 (0.233)# Banks and finance dropped

Largest ultimate shareholder profile Shareholder agreement 1.799 (0.373)** Government 0.086 (0.370) Foreign investor 1.240 (0.351)** Family 2.184 (0.348)** Foundation 3.287 (0.492)** Mutual Fund 1.486 (0.393)** Pension Fund 0.898 (0.426)* Privately-held firms 0.566 (0.428) Non-public companies dropped Number of observations 3786 LR (Chi2) 573.00** Pseudo R2 0.1093 Log likelihood -2333.5684

(#) Significant at 10%; (*) Significant at 5%; (**) Significant at 1%.

55 It should be remarked that control for possible endogeneity issues was not taken into account. This implies that unobserved factors may affect simultaneously both pyramidal structures and independent variables. Therefore, the identification hypothesis is that errors and independent variables are not correlated.

24

Since the data set is a repeated cross-section over six years (1997-2002), time dummies were also included, with the purpose of controlling for possible macroeconomic shocks in any year.

The estimation found no significant impact of the fraction of non-voting shares in the firm’s total capital on the likelihood of observing a pyramid. If the issuance of non-voting shares and pyramidal structures were substitute as regards the purpose of increasing the wedge of rights,56 the expected result would be a significantly negative coefficient for the fraction of non-voting shares. Thus, that finding apparently does not support the view that pyramids are driven primarily by expropriating purposes.

Another finding is that performance and pyramids are significantly (at a 5 percent level) and negatively correlated. That is, a higher level of net profit/asset ratio reduces the probability of observing a pyramidal ownership structure. Inasmuch as the estimation did not cope with causality issues (whether pyramids reduce performance through expropriation or instead underperformance affects the likelihood of pyramidal structure), this result only reports that, after controlling for observable characteristics, on average, companies under pyramidal ownership schemes present a worse performance than other firms.

No meaningful statistical relationship between leverage (measured as the total liability/total asset ratio) and pyramidal structures was found. The coefficient of leverage to estimate the probability of pyramidal ownership is non-significant. Hence, no evidence was raised to suggest that pyramids might be motivated with a view to overcoming credit constraints in the financial markets. With respect to size (measured by the firm’s value of assets), the model identified a significant positive relationship with the probability of observing a pyramid: companies with a high level of assets are more likely to be affiliated to pyramidal structures. Again, the direction of causality – whether pyramids enhance assets or the reverse – cannot be grasped with the available data.

Concerning the relationship between the identity of the largest ultimate shareholder and the likelihood of pyramidal structure, for all types the impact is positive (taking non-public companies as benchmark), except for government and privately-held firms, whose coefficients are non-significant. Taking banks and finance as reference, telecommunications is the only sector significantly, positively associated with pyramids, that is, the likelihood of pyramids in this sector is higher than banks’. The other sectors present either significant negative relationships (e.g. food and beverage, textile and leather, metallurgy and metals, and holdings) or non-significant ones (like agriculture and livestock, paper and cellulose, oil production, chemicals, rubber and plastic).

4. Concluding Remarks The Brazilian economy has undergone remarkable changes over the last 30 years. Firstly, the State stopped ruling the economic development strategy, leaving firms ample latitude for individually making decisions about allocation of their capital (de Castro, 2003). Trade liberalization, privatization of most of the largest state-controlled firms, and monetary stabilization have redefined the environment shaping firms’ incentives and constraints, bringing about an outstanding restructuring at the corporate level by means of mergers and acquisitions.

56 Brazilian corporate law allows firms to issue non-voting shares up to 2/3 of the firm’s capital.

25

Moreover, the new corporate law (approved in 2001) as well as the creation of differentiated levels of corporate governance by the Sao Paulo stock exchange have imparted better governance standards, contributing to strengthen Brazilian capital markets and to the massive number of IPO’s over the last three years. Also noteworthy is the striking stakes foreign investors have been acquiring in Brazilian firms. An open question is the impact of a capital market more willing to fund emerging and/or young firms on Brazilian business groups’ structures and strategies.57

As regards Brazilian business groups’ current structures and strategies, a number of characteristic features should be emphasized. Firstly, families control the majority of the Brazilian groups (16 out of 26). Secondly, state-controlled firms (such as Petrobras, BNDES, and Banco do Brasil) as well as other institutions over which the government has control (like pension funds of Banco do Brasil, Petrobras, and Caixa Econômica Federal) have significant stakes in the largest “private” business groups. This ownership structure is, to a great extent, accounted for by the privatization model followed, designed to maximizing revenues to the government.58 Third, state-controlled groups have dramatically shrunk their share among the largest groups operating in Brazil (only seven among the sixty largest business groups in 2005). Brazilian and foreign-controlled groups are almost equally represented among the largest business groups (25 by Brazilians, 27 by foreign firms, and one co-controlled by a Brazilian family and a foreign group). Forth, diversification strategies vary widely among the sampled business groups. Whereas some of them are focused on specific sectors (like Telemar in telecommunications, Unibanco in banking, Embraer in aircraft, and CPFL in energy), others have a highly diversified portfolio (such as Bradesco, Odebrecht, Votorantim and Vicunha), probably on the grounds of risk sharing or economies of scope. Other groups, despite focusing on core businesses, also have subsidiaries or affiliates operating in (vertically) related industries, as the cases of CVRD and Ultra, which have diversified their activities around their main businesses, perhaps for strategic motives, contractual shortcomings, or also scope economies. Fifth, most of the largest Brazilian groups have attempted to internationalize their operations. In many of them, foreign sales have yielded high revenues (Petrobras, CVRD, Embraer, Votorantim, Usiminas and Sadia) or represented a significant share in their total revenue (CVRD, Odebrecht, Sadia, and Embraer). The other vehicle of internationalization, foreign direct investment, has been forwarded mainly by Petrobras, Gerdau and CVRD, being followed suit more recently by Votorantim, Sadia, Camargo Corrêa, Ultra, and Perdigão. Sixth, Brazilian business groups present a considerable longevity. The majority of them were created before 1950, state support being probably fundamental to ensure their development. Notwithstanding, under a longer-run perspective, the turnover in the ranking of the largest groups is considerable (around 70% between 1978 and 2005 for the top 40).

Finally, a statistical approach based on a data set built out of reports from hundreds of firms over the period 1997-2002 allowed to shed some light on the relevance and rationale of pyramidal ownership structures. The estimates show that pyramidal firms 57 As noted by Khana and Yafeh (2007, p. 340), should the availability of an internal capital market be the main rationale for the existence of business groups, it would be expected that they would shrink as capital market developed. 58 The role of pension funds in Brazilian corporations and capital markets is surely an important topic for future research, notably because of frequent denunciations of political and personal uses of their huge funds. On the possible governance benefits coming from pension funds for active monitoring public companies, it should be noted that pension funds’ managers are themselves “agents.” As regards mutual funds, Black and Coffee (1994) persuasively observe that “agency costs at the fund-manager level may be no less important than at the corporate-manager level.”

26

do tend to have a lower performance and a higher size. But estimations fail to find any significant impact of either leverage or the fraction of non-voting shares in the firm’s capital on the likelihood of observing a pyramidal structure. Thus, the results are ambiguous concerning the motives for the emergence of pyramidal structures of ownership.59

References ALDRIGHI, Dante and Roberto MAZZER NETO. 2007. Evidências sobre as Estruturas de Propriedade de Capital e de Voto das Empresas de Capital Aberto no Brasil. Revista Brasileira de Economia 61(2): 129-152. ALMEIDA, H. and D. WOLFENZON. 2006. “A Theory of Pyramidal Ownership and Family Business Groups”. Journal of Finance LXI (6), pp. 2637-2680. ANÁLISE COMÉRCIO EXTERIOR 2007. AZEVEDO, A. and M. PORTUGAL. 1998. “Abertura comercial brasileira e instabilidade da demanda de importações”. Nova Economia 8, pp. 37-63. BACHA, Edmar L. and Regis BONELLI. 2005. Uma interpretação das causas da desaceleração econômica do Brasil. Revista de Economia Política 25(3), pp. 163-189. BEBCHUK, L., R. KRAAKMAN and G. TRIANTIS. 1999. Stock Pyramids, Cross-Ownership and Dual Class Equity: The Mechanisms and Agency Costs of Separating Control From Cash-Flow Rights. NBER Working Paper #695. BERTRAND, M. and S. MULLAINATHAN. 2002. Pyramids. Massachusetts Institute of Technology Working Paper 02-32. BNDES. 2002. Privatização no Brasil. BONELLI, Regis. 1998. As Estratégias dos Grandes Grupos Industriais Brasileiros nos Anos 90. CLAESSENS S., S. DJANKOV, J. FAN and L. LANG. 2002. Disentangling the Incentive and Entrenchment Effects of Large Shareholdings. Journal of Finance LVII(6). De CASTRO, Antonio Barros. 2003. El Segundo Catch-Up Brasileño: Caracteríticas y Limitaciones. Revista de la Cepal 80: 73-83. DE SIQUEIRA, Tagore. 2000. Os Grandes Grupos Brasileiros: Desempenho e Estratégias na Primeira Metade dos Anos 90. Revista do BNDES 7(13): 3-32. EXAME MELHORES E MAIORES July 2006. FISHLOW, A. 1972. “Origins and Consequences of Import Substitution in Brazil”. In L. de Marco (ed.). International Economics and Development. New York: Academic Press. HERMANN, Jennifer. 2005. Auge e Declínio do Modelo de Crescimento com Endividamento: O II PND e a Crise da Dívida Externa (1974-1984). In F. GIAMBIAGI, A. VILLELA, L. DE CASTRO, and J. HERMANN, Economia Brasileira Contemporânea (1945-2004). Rio de Janeiro: Elsevier.

59 It is likely that pyramidal ownership arrangements in Brazil aims at different strategies of firms’ controlling shareholders. Some probably use them as an “internal capital market” for facilitating the financing of new enterprises, as argued by Almeida and Wolfenson (2006). For others, pyramids may operate as a device for leveraging voting power from a limited capital stake in their firms, facilitating the expropriation of minority shareholders, as the traditional view emphasizes. They may still serve as a mechanism for ensuring the family’s control over the firm, preventing other block-holders of shares from contesting it. Finally, tax avoidance should not be neglect as a motive lying behind pyramids.

27

28

JOHNSON, Simon, Rafael LA PORTA, Florencio LOPEZ-DE-SILANES, Andrei SHLEIFER e Robert VISHNY. 2000. “Tunneling.” American Economic Revue Papers Proceedings, May. KHANNA, Tarun and Yishay YAFEH. 2007. Business Groups in Emerging Markets: Paragons or Parasites? Journal of Economic Literature XLV: 331–372. La PORTA, R. F. LOPES-De-SILANES, and A. SHLEIFER. 1999. “Corporate Ownership Around the World”. Journal of Finance LIV(2). SCHNEIDER, Ben Ross. 2007. How States Organize Capitalism: A Comparative Political Economy of Diversified Business Groups. VALOR GRANDE GRUPOS (2006). WOOLDRIDGE, J.M. 2002. Econometric Analysis of Cross Section and Panel Data. MIT Press, Cambridge.

APPENDIX Ranking of the Largest Business Groups in Brazil in 2005

Rank. Business Group Origins Gross Revenue Net Worth Business Sectors 1 Petrobras BRS 179,065.30 84,964.10 exploration, production, and distribution of oil and gas, petrochemicals, energy, and transportation 2 Bradesco BRP 59,637.20 22,756.10 banking and finance, holding, electrical energy, mining, computer 3 Banco do Brasil BRS 48,520.80 16,849.80 banking and finance, computer 4 Itaúsa BRP 39,400.40 18,764.30 banking, real estate, information technology, wood paneling, ceramic and metal sanitary goods, chemicals 5 CVRD BRP 35,350.20 27,002.60 minerals and metals, logistics, railways, steel, and energy 6 Caixa BRS 34,658.70 7,951.90 banking and finance 7 Telefonica SP 30,895.10 15,101.60 telecommunications 8 AmBev BEL 28,878.70 19,990.00 beverages and food 9 Ipiranga BRP 28,870.50 2,052.20 oil refinery, lubricant and fuel distribution, chemical and petrochemical firms, transportation and logistics

10 Gerdau BRP 25,485.80 10,374.90 steel and iron mining 11 Telemar BRP 23,686.50 9,924.00 telecommunications 12 Odebrecht BRP 23,436.60 6,423.60 engin./constr., chem../petrochemicals, infrastr./public serv. concession, mining, sugar/ethanol, insurance 13 Unibanco BRP 23,300.50 10,189.90 banking and finance 14 Eletrobrás BRS 22,326.30 75,963.10 electrical energy holding 15 Votorantim BRP 22,192.60 21,068.80 cement, metals, cellul./paper, energy, agribus., chemicals, mining/metals, finance, biotech., inform. Tech. 16 Fiat IT 20,753.10 n.a. car manufacturer 17 Volkswagen GER 20,100.40 n.a. car manufacturer 18 Bunge NE 19,261.50 3,843.30 food, chemicals and petrochemicals 19 ABN AMRO NE 18,503.80 8,590.30 finance 20 Usiminas BRP 17,058.40 8,836.70 flat steel, steel goods, metals, mechanics, logistics and transportation 21 Shell Brasil NE/GB 16,828.50 2,708.10 petroleum and gas , fuel and lubricant distribution, energy, 22 Santander Banespa SP 16,605.60 7,703.60 finance 23 Arcelor Brasil LX 16,535.80 12,120.70 steel and metals 24 Pão de Açúcar BRP/FR 16,121.00 4,539.80 supermarket/food retailer 25 Brasil Telecom BRP 14,687.20 7,047.20 telecommunications 26 HSBC GB 14,317.30 3,477.60 finance 27 Cargill US 13,794.20 480.00 agribusiness and food 28 Carrefour FR 12,546.20 2,510.80 supermarket 29 CSN BRP 12,283.50 6,472.40 steel, textiles, cement, energy, railway, logistics, trading, banking and finance 30 Wal-Mart US 11,732.00 n.a. retail commerce

29

31 Cemig BRS 11,702.60 7,204.40 electric energy 32 Chevron US 11,413.80 924.90 oil distribution, chemicals, and petrochemicals 33 TIM Brasil IT 11,226.10 9,031.30 telecommunications 34 AES Eletropaulo US 11,180.30 1,955.30 electricity 35 CPFL Energia BRP 10,907.10 4,796.00 electrical energy generation and distribution 36 Embratel MX 10,178.00 7,572.20 telecommunications 37 Portugal Telecom PT 9,762.70 5,482.60 telecommunications 38 Embraer BRP 9,140.50 4,863.10 aircrafts 39 Unilever Brasil GB/NE 9,000.00 n.a. food, hygiene and cleaning products 40 Sadia BRP 8,328.00 2,225.50 chilled and frozen foods 41 Souza Cruz GB 7,965.00 1,651.40 tobacco 42 Camargo Corrêa BRP 7,958.50 6,598.30 Engine./constr., cement, footwear, textiles, el. energy, highway concessions, steel and metal, and banking 43 Siemens NE 7,686.30 n.a. electrics and electronics, information technology, telecommunications, mechanics, and car components 44 Copesul BRP 7,348.30 1,247.30 petrochemicals 45 Neoenergia BRP 7,331.50 5,959.70 electrical energy generation, transmission, distribution and trading 46 Nossa Caixa BRS 7,322.20 2,301.00 finance 47 Light BRP 7,097.00 1,699.50 electrical energy 48 Copel BRS 6,816.10 5,630.60 electrical energy and telecommunications 49 Claro MX 6,806.30 n.a. telecommunications 50 Sul América BRP 6,581.10 1,712.40 insurance 51 Brasmotor US 6,398.40 1,934.20 electrics and electronics, mechanics 52 Endesa SP 6,338.90 4,457.80 electrical energy 53 Safra BRP 6,212.70 3,665.20 banking and finance, paper and pulp, agribusiness 54 TAM BRP 5,910.10 761.90 civil aviation 55 Perdigão BRP 5,873.30 1,222.80 chilled and frozen foods 56 EDP Energias PT 5,862.80 4,088.80 electrical energy 57 Suzano BRP 5,661.80 4,115.80 paper and pulp, petrochemicals 58 Saint-Gobain FR 5,200.00 n.a. glass, ceramics, mining, and metals 59 Ultra BRP 5,158.00 1,819.80 fuel distribution, chemicals, transportation, and logistics 60 Dow Brasil US 5,146.70 1,162.90 chemicals and petrochemicals

1) BRS: Brazilian State-Controlled Group; BRP: Brazilian Privately-Controlled Group; BEL: Belgian; FR: French; GB: Britain; GER: German; IT: Italian; LX: Luxembourg; MX: Mexican; NE: Dutch; PT: Portuguese; SP: Spanish; US: American; n.a.: not available 2) Gross revenue and net worth in R$ million 3) Source: Valor Grandes Grupos (2006)

30