PRINCIPAL-PRINCIPAL CONFLICTS LEADING TO ACTIVISM OF A LARGE GOVERNMENT-OWNED INVESTOR IN BRAZIL

How to cite this paper: Duarte, B. E., & Leal, R. P. C. (2021). Principal-principal conflicts leading to activism of a large government-owned investor in Brazil. Corporate Governance and Organizational Behavior Review, 5(1), 15-25. https://doi.org/10.22495/cgobrv5i1p2 Copyright © 2021 The Authors This work is licensed under a Creative Commons Attribution 4.0 International License (CC BY 4.0). https://creativecommons.org/licenses/by/


INTRODUCTION
In most markets, especially the emerging ones, high ownership concentration may lead to conflicts of interest between the major or controlling shareholders and minority shareholders, the so-called principal-principal conflicts, and abuse (Young, Peng, Ahlstrom, Bruton, & Jiang, 2008). Voicing discontent and shareholder activism is the set of actions that intend to influence the practices and policies of investee companies and may be a reaction to conflicts with major shareholders (Goranova & Ryan, 2014). Alternatively, selling and walking away, or exit, may be the most common reaction (Gillan & Starks, 2007). The antecedents of principal-principal conflicts are evident in Brazil, such as family ownership, high ownership concentration, less effective minority shareholder protection, low quality of corporate governance (CG) practices ratings, and majority shareholder abuse ( Activist shareholder campaigns may be a way to promote better CG practices, improve performance and protect minority shareholders from the agency and principal-principal conflicts and abuse (Young et al., 2008;Gillan & Starks, 2007). Institutional investors could have an incentive to monitor investee companies because, in general, they have greater equity ownership in companies than individual investors, can capture gains that may exceed the costs of engagement, and be more likely to succeed than other investors (Shleifer & Vishny, 1997). Yet, the evidence about the positive outcomes of engagements and activism is still not conclusive ( Gillan & Starks, 2007). This is also the case in Brazil, even after recent changes that encourage activism, such a gradual reduction in ownership concentration and new laws, regulations, and self-regulations (de Almeida & Leal, 2020; Maranho, Bortolon, & Leal, 2020; Collares, 2020; Sonza & Granzotto, 2018).
There are limitations to activism efficacy. The motivations of activist investors may be questioned when their interests do not seem aligned to those of the investee company or its minority shareholders (Inoue et al., 2013;Woidtke, 2002;Romano, 2001). The costs associated with activism may be greater than the potential benefits for its promoters due to regulatory and practical limitations, among other factors. The lack of expertise required for effective monitoring may also be one of the reasons for the apparent ineffectiveness of activism (Gillan & Starks, 2003 Gillan & Starks, 2007).
Thus, the objective of this study is to qualitatively understand and contrast the activism carried out by a large public institutional investor, as a reaction to principal-principal conflicts of interest, with the evidence found in the Brazilian and international literature about the activism of institutional investors in general. The institutional investor is BNDES Participações (BNDESPar) the equity investment arm of the Brazilian National Economic and Social Development Bank (BNDES -Banco Nacional de Desenvolvimento Econômico e Social). BNDES is one of the largest development banks in the world and is 100% owned by the Brazilian federal government. The three cases studied involve companies with different characteristics regarding their ownership and relation to the state. One of them is a family-owned company that is the largest animal protein producer in the world. The other two are large state-owned electrical utilities. Two of them were listed on the Novo Mercado, the most demanding listing segment of the Brazilian stock exchange in terms of CG and transparency requirements. Ownership concentration is high in all of them. The focus on one single very large institutional investor linked to the state is justified due to the volume of resources it controls and the potential conflict of interest between the purposes of governments and the interests of other shareholders of the investees (Girard & Gates, 2014;Shleifer & Vishny, 1997).
The cases address the antecedents of the principal-principal conflicts and of activism; the means employed by the investor; the aspects of CG or management practices that BNDESPar intended to change; and whether it achieved its goals. The results will be compared with the evidence in the literature. Goranova and Ryan (2014), Judge, Gaur, and Muller-Kahle (2010), and Young et al. (2008), in particular, will be used as a conceptual basis because they offer or review models with the background, actions, and expected results of principal-principal conflicts of interest and activist actions. The cases contribute to the scarce literature involving the activism of institutional investors in emerging markets, and in Latin America in particular (Maranho & Leal, 2018).
The role of institutional investors linked to the public sector has been widely studied in the US market as they were the pioneers of shareholder activism there (Gillan & Starks, 2007). Even though the initial evidence indicated a positive impact of their activism on the creation of shareholder value, the methods used to reach the results were questioned, leading to more recent inconclusive results about value creation (Becht et al., 2009). In Brazil, the institutional investors related to the state have also been the subject of several studies that did not find conclusive evidence that their presence as shareholders resulted in better CG practices or performance, even though these studies did not address their activism per se ( The findings of this article point out meaningful differences regarding certain antecedents of activism described in the literature. The cases did not evince a trade-off between the quality of CG practices of the investee companies and activism (Vargas et al., 2018;Gillan & Starks, 2007). These activism cases suggest that trading in the special CG segments of the stock exchange did not guarantee immunity against activist actions, particularly when performance was unsatisfactory or the institutional investor spotted possible legal violations. However, evidence of the antecedents of the principal-principal conflicts of interests are present in all cases, such as family ownership and business groups, high ownership concentration, less effective minority shareholder protection by both internal and external CG mechanisms, and controlling shareholder abuse (Young et al., 2008).
Moreover, activism was observed in high ownership concentration companies even though the evidence in the literature points out that these companies should be less frequent targets of shareholder activism (Goranova & Ryan, 2014;Judge et al., 2010). High ownership concentration, however, hindered activism through internal CG mechanisms, leading activists to resort to external mechanisms, such as using regulations in administrative proceedings or the courts, in order to uphold their interests. Even so, these external mechanisms were not completely capable of protecting the interests of shareholders and limited the effectiveness of their activism, constituting one of the possible reasons why the presence of an institutional investor related to the state did not influence the quality of CG practices or performance of the investee companies reported elsewhere (Sonza & Granzotto, 2018;De Oliveira et al., 2012).
This study proceeds with the review of the relevant literature in Section 2, the discussion of the research method in Section 3, the presentation and analysis of the cases in Section 4, and the conclusion in Section 5.

LITERATURE REVIEW
The principal-agent conflict of interest is more typical of the US and a few developed countries and is what may be one of the motivations of activists in these markets. In Brazil, as well as most other countries, however, the typical conflict is between principals in a concentrated ownership setting (Young et al., 2008;Shleifer & Vishny, 1997). This section addresses the literature about the activism of institutional investors linked to public entities, first in other countries and then in Brazil.
Institutional shareholders, in particular, pension funds sponsored by institutions controlled by the state, are among the most frequent activists (Gillan & Starks, 2007). They submit proposals as shareholders, put "behind the scenes" pressure on the company's management and use the press to coerce companies presenting unsatisfactory results in the US. Institutional investor activism gained traction with the formation of the Council of Institutional Investors (CII) in 1985. In principle, as a council to defend the rights of the beneficiaries of California's public employees (CalPERS) and teachers (CalSTRS) pension funds, the CII remains active and serves approximately 140 US asset owners. Gillan and Starks (2007) reviewed several studies about CalPERS target companies. The earlier ones displayed an increase in shareholder returns but they used very long time periods and did not exclude events that may have contaminated the result. Improved methods did not reveal the same results in more recent periods for CalPERS and other CII funds. Del Guercio and Hawkins (1999) found that proposals from shareholders moved by public pension funds were related to changes in CG and company practices, but not to share performance, whereas Prevost and Rao (2000) found a strong negative impact of public pension fund activism on companies. Song and Szewzcyk (2003) did not find evidence that the activism of institutional investors resulted in an increase in share prices. Woidtke (2002) examined the differences resulting from the presence of public and private pension funds in the value of investee companies. Her results indicated that the presence of private pension funds had a positive effect on the value of companies, which suggested efficient monitoring due to the alignment of interest between performance-compensated fund managers and other shareholders. The presence of activist public pension funds, in contrast, resulted in value loss to the investee companies, which suggested activist motivations unrelated to value creation. Shleifer and Vishny (1997) asserted that the limitations imposed on the managers of public funds to capture a portion of the wealth they generate could make them seek other forms of reward, even though their interest in social welfare potentially provide better conditions to exercise an active monitoring role. An example would be the use of public institutions for political purposes, benefiting a certain power group.
In Brazil, the presence of public institutional investors is also important. Among them, BNDES and the largest pension funds stand out because the three largest are sponsored by companies that have the Brazilian federal government as their controlling shareholder. Crisóstomo and Gonzáles (2006) stated that Brazilian pension funds have started to adopt practices that encourage activism, such as presenting proposals in shareholder meetings, exercising voting rights, and closer management supervision after the privatization wave of the 1990s.
Yet, there was no evidence about the effectiveness of their monitoring. De Oliveira et al.
(2012) did not verify that the presence of the largest Brazilian pension funds as relevant shareholders (more than 5% of voting shares) was associated with the quality of CG in their investees in the years 1998, 2000, and 2002. Their finding suggests that these funds would not have taken an active stance to improve CG practices, that any activism was not effective, or that they did not aim to improve CG practices. The authors also conclude that these institutional investors did not seem to take into account the composition of the board and its processes when choosing companies to invest in, which would potentially mitigate agency problems. Da Silva, Tsai, and Gutierrez (2011) claim that Brazilian pension funds tend to invest in companies that already display better CG practices, but their presence does not improve such practices. Sonza and Granzotto (2018) claim that the presence of pension funds related to state entities is related to an increase in market value, but not of financial performance. They speculate that investors associated the presence of these funds with an improvement in company performance, even though there was no concrete evidence for it.
Some studies have found a negative relationship between the shareholding of pension funds and the quality of CG practices. Punsuvo et al. (2007) found a negative relationship regarding the presence of pension funds and the score of a CG index in 2004, even though they did not distinguish between funds sponsored by public or private entities. A possible justification for their finding was that the presence of pension funds would replace the need for good CG practices in a concentrated ownership market such as Brazil. Collares (2020) claims that there was no impact of activism on company performance and that companies with fewer independent directors were targeted more often. Maranho et al. (2020) found little impact of institutional investors on the quality of CG practices of investee companies by means of an activism index that considered each investor-investee pair. They found that holding a board seat and being a signatory of the shareholders' agreement was actually associated with worse CG practices when they examined each scoring component of their index. Only the presence of foreign institutional investors as relevant shareholders was associated to better CG practices. The case studies in de Almeida and Leal (2020) suggested that the lack of positive impacts of institutional investor activism in Brazil was due to principal-principal conflicts of interest, diversification reasons and possibly a lack of monitoring skills.
Pereira and Leal (2018) interviewed Brazilian institutional investors that were not associated with the government or large financial conglomerates. They concluded that these investors reject public activism and would solely see themselves as relationship builders through private engagements. These authors also convey that their voting policies are not consistent with the owner-fiduciary model. Investors claimed that they exercise scrutiny, including engagement, before investing, in order to avoid exit, which they prefer instead of voice. Gillan and Starks (2003) claim that large shareholders who are also credit institutions, such as banks, would be in a unique position to monitor because they would have access to better information than other shareholders. The success of BNDES as an activist, through its BNDESPar equity investment arm, is of particular interest to this article. These institutions are of great importance in the Brazilian financial market and could be promoters of better CG practices due to their size and influence.
Da Silveira (2010) analyzed the largest loans and equity holdings of BNDES in 2009. The bank could be a promoter of good CG practices but it could also be possible that it acts in the opposite way in case its CG requirements are weak or non-existent. In fact, like pension funds, BNDES did not seem to play this role at the time. The author concluded that CG did not seem to be a determining credit decision factor because borrowers were not among those that presented the best CG practices. Moreover, he contended that the loans did not appear to be conditional on any improvement in CG and presented evidence that BNDES was associated with worse CG practices when it was a relevant shareholder. Inoue et al. (2013) also investigated the loans and shareholdings of BNDES and concluded that its presence did not lead to better operational performance. The authors also gathered that companies that had political ties, measured by donations to political parties that won elections, received more resources from BNDES, although equity ownership in these companies has not necessarily been detrimental to the bank. Carvalho (2014) found that BNDES lending privileged regions and companies that generated more jobs in the electoral years. This evidence supports the arguments of Shleifer and Vishny (1997) regarding the risks of the state as a lender and shareholder.

RESEARCH METHODOLOGY
The small number of activism campaigns in Brazil imposes limitations on quantitative analyses of the impacts of shareholder activism, especially if the intention is to examine a specific type of investor, as in this article (Maranho et al., 2020;Collares, 2020;Vargas et al., 2018). Thus, the study of cases is interesting because it can complement quantitative studies and their results, helping to elucidate apparent contradictions. Specifically, this article uses multiple case studies as a method, so that investee companies with different characteristics could be explored as the unit of analysis. In each case, the antecedents to activism, the apparent motivations of activism, the means employed, and whether the shareholders were successful in their activism were the categories of analysis.
Three BNDESPar activism cases contemplate situations of principal-principal conflicts and were selected because there was sufficient information suitable to the goals of this article. The case selection resulted from research in the specialized business press, especially the Valor Econômico newspaper, consultation of legal documents available at the Securities Commission's website (CVM or Comissão de Valores Mobiliários), and the minutes of board meetings and shareholders' meetings availed by companies. Business newspapers allow an exploratory qualitative analysis of activism cases without a survey of formal documents and may reveal possible impacts on the strategy, administrative practices, and CG mechanisms of companies.
Naturally, the analysis of activism cases has to consider the minutes of the general shareholders' meetings because it is the ultimate decision-making body of a publicly-traded company, responsible for the election of board members and top management and board compensation, which are recurring objects of activism. The minutes of shareholders' meetings reveal their deliberations with the identity of the relevant shareholders and how each one of them voted. Discussions about the issues to be put to the vote at the shareholders' meeting are discussed at the board of directors, with the justifications for certain positions in relation to the matters stated in more detail in its minutes, allowing a better analysis of the reasons for institutional investors to engage in favor of certain changes in the investee companies.
The ordinary general shareholders' meeting takes place once a year and, in general, deals with company results, profit distribution, approval of accounts, and the election of the board and fiscal board members. The fiscal board is an optional board installed upon request of shareholders and may be temporary or permanent. It is made up of outsiders and examines the company's accounts and audit process after they are publicized. It does not partake in the insider discussions that take place on the board of directors. It is easier for minority shareholders to elect representatives to this board than to the board of directors (Black et al., 2014). Extraordinary shareholders' meetings are potentially more interesting as instruments of activism because they are called to address specific issues and can be requested by shareholders that reach the minimum ownership legal limits.
The cases were selected according to their distinctive characteristics, which make it possible to explore different aspects of the activism of BNDESPar. The cases also display some of the key antecedents of principal-principal conflicts, such as family ownership, high ownership concentration, poor CG quality, performance, and use of recent institutional frameworks to try to uphold the minority shareholder views (Young et al., 2008). The first case deals with JBS, the largest animal protein producer in the world. JBS was selected because it is a family-owned company with a large ownership concentration that displayed an array of legal problems related to its involvement in politics. The second and third cases deal with BNDESPar's activism in CEMIG and Light in which both the institutional investor and the controlling shareholders of these investee companies are the state. The two companies are electric utilities. In addition, the Light case shows an application of the recent Law of State-Owned Companies, that can have a significant impact on the CG of these companies. The cases also differ in the purpose of activism, allowing a better understanding of both activism that seeks to improve CG practices, to influence the decisions of the investee, or to revert potentially damaging corporate actions.
Administrative proceedings at CVM are an important mechanism for the defense of shareholders through which activism can develop more easily. Administrative proceedings are usually initiated when argumentation at the decision-making bodies of investee companies were exhausted. The use of these proceedings is an important institutional alternative for the protection of minority shareholders' rights. It is important to highlight that there is no need to have external complaints to initiate these proceedings. The technical areas of CVM can initiate one based on an investigation that identifies possible irregularities. These proceedings may result in an indictment that requires the manifestation of federal attorney generals at CVM and are judged by the CVM collegiate body. The documents referring to the initiation of the proceedings and the decision of the collegiate are publicly available. Finally, the Notices to the Market from the investee companies targeted for activism were also consulted, particularly those indicating material facts. These communications allow the analysis of the company's perception of its own shares.
These information sources are obviously not exhaustive in terms of the possible means of activism for shareholders because they were limited to those that were publicly available. As highlighted by Becht et al. (2009), not all forms of activism are public and those that are not public may be relevant for a better understanding of the influence of certain characteristics of those involved in the process. Moreover, news sources may contain errors or biases that can only be rectified through observation of reliable non-private information or interactions with the parties involved. Naturally, there are alternative methods that could be suitable for conducting research about activism in Brazil and elsewhere. Future researchers could try to engage companies or investors involved in similar cases and perform case studies from information obtained from non-public sources, observation and interviews, in contrast to the cases addressed in this article, which were studied only from publicly available information and without interaction with the people actually involved in the situations reported. Pereira and Leal (2018) interviewed independent institutional investors about activism and CG.
Econometric studies with large samples of companies, such as those by Maranho et al. (2020), Collares (2020), and Punsuvo et al. (2007), are usually the prime choice because they can produce general conclusions. However, as discussed in the previous section, the evidence emerging from these studies in Brazil was mixed or inconclusive, and this led to the choice of studying activism cases in this article. Another alternative is to interview market practitioners, particularly those that manage pension and mutual funds and that are potential agents of activism, as in Pereira and Leal (2018).
The case method does not allow statistical generalizations from the analysis of narratives involving the activism of BNDESPar, but only analytical generalizations (Yin, 2009). Even in these cases, a detailed analysis of the consequences of activism on the value of investee companies is not intended due to the difficulties of establishing a causal relationship between activism, CG practices, and performance of invested companies. Table 1 shows a summary of the three cases, with a brief description of the company, the potential conflict situation, the BNDESPar actions, their outcome, and reasons for the outcome. The next three subsections describe each case in more detail.

JBS
BNDES aimed to increase the international Brazilian economy competitiveness. The government introduced the "national champions" policy in the late 2000s in order to transform selected large Brazilian companies into global players through subsidized credits. JBS was one of these companies and it reached the position of the world's largest producer of animal protein and one of the most important companies in the global food industry with BNDES support (BNDES, 2016). However, this period was also marked by a series of complaints and indictments involving the company and its major shareholders and by the attempts of BNDESPar to protect their interests and defend its investment from legal prosecution.
JBS was founded in 1953 and grew mainly through acquisitions. It began to stand out in the high inflation 1980s and 90s when the volatile economic environment led companies in the industry to dispose of their assets (Mendes, Adachi, Torres, & Góes, 2017). JBS needed to raise money to take advantage of these opportunities. BNDES granted it loans from the 1990s on and subsequently became a shareholder when JBS became a publicly-traded company in 2007. Family members acted as CEO and board chairs and members throughout this period.
The industry in general did not enjoy a good CG practices reputation but the initial public offer of JBS (IPO) took place on the Novo Mercado (Mendes et al., 2017). The listing of JBS in this segment would supposedly offer more protection to investors. JBS highlighted this demonstrated its commitment to effective CG. The IPO proceeds were not enough for the planned acquisitions and BNDESPar became a relevant shareholder in that same year with approximately 13% of the equity capital. BNDESPar aimed to help in the internationalization process of JBS through US acquisitions.
The purchase value of the JBS shares was based on the average prices of the last 30 days before the acquisition plus a premium of R$ 0.50 per share (about US$ 0.28), which was the subject of controversy. The Federal Court of Accounts (TCU -Tribunal de Contas da União), which supervises BNDES, claimed that the acquisition value was US$ 230 to 350 million above what was necessary . The bank refuted this accusation alleging that the value per share, including the goodwill, was below its internal assessment. The internationalization process also received support from other shareholders, who invested to maintain their equity participation, suggesting an alignment of interest between BNDESPar, JBS, and other shareholders. BNDESPar held 13% of the equity capital, other minorities 23.5%, and the controlling family 63.5% at the end of 2007. Later on, there were complaints against the controlling shareholders and public officials related to illegal political campaign donations and insider trading, and also about sanitary issues, which have resulted in indictments and plea bargains (Silva, Souza, & Lucena, 2019;Mendes et al., 2017;Mendes, 2016).
It is possible to see some antecedents described in the literature in the JBS case. Yet, the activism of BNDESPar cannot be regarded as homogeneous throughout the period. One could apprehend that its posture was passive until 2015, as defined in Elyasiani and Jia (2010), from the analysis of public documents. As far as the authors of this article could see in the surveyed documents, there was no observable active stance from BNDESPar with the objective of improving performance by upgrading the CG practices of JBS, which is a common focus of investor activism (Goranova & Ryan, 2014). If activism occurred, it was not recorded in the consulted public documents. This is not surprising considering that the empirical findings involving institutional investors linked to public entities analyzed in the literature review indicated that their presence as shareholders did not imply the improvement of CG practices or better performance (Sonza & Granzotto, 2018;De Oliveira et al., 2012;da Silveira, 2010). It is important to note, however, that BNDESPar probably considered the quality of CG practices satisfactory because JBS was listed on the Novo Mercado and this listing is used as a kind of certification. In fact, there is evidence in the literature about a trade-off between the quality of CG practices and activism (Gillan & Starks, 2007). From 2016, it is possible to perceive a more active stance by BNDESPar, motivated first by a proposed corporate reorganization and then by the successive problems involving the company and its executives (Mendes et al., 2017;BNDES, 2016).
Other antecedents observed in the literature are present in the case. Large companies like JBS are more frequent targets of activism from institutional investors (Goranova & Ryan, 2014). It is possible to argue, however, that this characteristic was already present in the period when BNDESPar did not adopt an active stance, being insufficient in itself to explain its behavior.
Performance below expectations is another antecedent from the literature present in the JBS case (Judge et al., 2010;Gillan & Starks, 2007). The complaints and investigations had a negative impact on performance and motivated an activist stance with a request for compensation to shareholders. In the case of activism against the corporate reorganization, however, the expectation about the performance of JBS improved significantly after the announcement, with an immediate impact on the company's share price. It seems that performance as an antecedent only partially explains the activism of BNDESPar, similarly to JBS size.
High ownership concentration, such as in JBS, would make activism less frequent because controlling shareholders would have incentives to act in the best interest of the company in which they hold the largest share of the profits (Goranova & Ryan, 2014;Judge et al., 2010). The JBS case suggests, contrastingly, that this was not enough to align their interests with those of minority shareholders, among them BNDESPar. The series of investigations, indictments, and lawsuits against JBS, which have encouraged other actions by BNDESPar, suggest that ownership concentration was not sufficient to align the interests of shareholders.
Alternatively, high ownership concentration leads to less frequent shareholder activism because minority shareholders would have little chance of success. As a public institution with objectives that are not limited to financial gain, BNDESPar would have an incentive for an activist stance even when the chances of success are low, as in fact occurred when it opposed JBS's board members compensation. The possibility of getting the necessary votes, however, was very small because they were opposing the controlling shareholders.
Contrastingly, as a signatory of a shareholders' agreement, BNDESPar succeeded in another situation. Agreements among the largest shareholders are common in Brazil (Sternberg et al., 2011). The proposed corporate restructuring from 2016 included the change of the fiscal domicile to the United Kingdom and the creation of JBS Foods International, which would become the owner of the international businesses of JBS and list it in the US. BNDESPar vetoed the restructuring as a signatory of the shareholders' agreement. Maybe the outcome would have been different if they were not signatories of this agreement. BNDESPar's actions against the corporate reorganization, its justifications, the context, and the immediate results indicate that its main motivation for activism was not just financial. The denationalization of the company did not seem to be in line with the objectives of BNDES and, as the bank itself pointed out, it was one of the reasons for the opposition (Mendes et al., 2017). Girard and Gates (2014) highlight the role of public institutions against actions that aimed at the denationalization of companies in an analysis of French companies that were targets of activism. Thus, another possible explanation for the BNDESPar veto is that this change would not serve the political interests of the government and its national champions policy (Inoue et al., 2013;Woidtke, 2002).
The case also reveals the importance of rules and legislation for investor activism (Judge et al., 2010). The indictments against JBS motivated shareholder activism in search of redress through administrative and legal means. Naturally, there were no limitations for a minority shareholder in these venues in contrast with shareholders' meetings. However, the dispute between BNDESPar and the controlling shareholders of JBS, at CVM and in the judiciary, exposed unconsolidated jurisprudence regarding the voting rights of the controlling shareholders when the meeting agenda involved sanctions against them. CVM's collegiate decision that the existence of a conflict of interest must be declared by the shareholders themselves limited the effectiveness of BNDESPar's activism (CVM, 2017b).
The analysis of the case suggested possible explanations for previous Brazilian studies that concluded that the presence of an institutional investor linked to a public entity did not clearly result in better CG practices (Sonza & Granzotto, 2018; De Oliveira et al., 2012; da Silveira, 2010). Not many public actions resulting from activism were found in the period BNDESPar was a shareholder, in line with the evidence in the literature for Brazil (Vargas et al., 2018). Even though some actions focused on items that could have a positive impact on CG practices and performance, such as executive compensation, board member election, and the attempt to replace or interdict board members, the opposition of the controlling shareholders prevented many of them from being successful (Bessière, Kaestner, & Lafont, 2011;Crisóstomo & González, 2006).

Cemig
Cemig is the Energy Company of Minas Gerais (Companhia Energética de Minas Gerais), the secondlargest state of Brazil in population. It is a holding company formed by several energy companies. The state of Minas Gerais held 51% of its voting shares. Other relevant shareholders were BNDESPar (13% of the voting shares and 6.5% of the total equity capital) and the stock fund Dinâmica Energia (10% of the voting shares and 8% of the total equity capital).
Cemig was listed on Level 1, a special listing segment of the stock exchange that has fewer demanding requirements than Novo Mercado, where JBS was listed. In Level 1, for example, there was no requirement for a minimum proportion of independent directors in the board, an audit committee, internal audits, and the implementation of compliance functions. Companies may issue non-voting shares in Level 1.
BNDESPar's activism at Cemig addressed performance, through its opposition to proposals to exceed statutory investment limits due to the indebtedness of the company, and CG in issues related to the election and replacement of board members. BNDESPar did not succeed in its voting in shareholders' meetings (CVM, 2017a). The State of Minas Gerais ownership concentration limited BNDESPar attempts (Judge et al., 2010). Nonetheless, BNDESPar played the role of informing shareholders about good CG practices and the risks of certain decisions through several proposals, even though it did not defeat all the proposals it voted against. It adopted a monitoring stance since it became a shareholder in Cemig, as defined in Elyasiani and Jia (2010).
The difficulties encountered by an institutional investor to exert influence on a company with a controlling shareholder were evident, even though both the investee and the investor had public entities as their controlling shareholder, but not necessarily of the same political group (Bessière et al., 2011).
One possible explanation is that the interests of minority and controlling shareholders were not aligned in favor of better performance at Cemig due to the limited rights to a company's profit that public controlling shareholders have (Shleifer & Vishny, 1997). In fact, the potential breach of the investment limits in the bylaws could result in unsatisfactory performance, a frequent antecedent of activism (Denes et al., 2017).
Agency theory suggests that a more independent board tends to constrain conflicts of interest (Goranova & Ryan, 2014), which would reduce the need for activism. The 2016 company filings registered only one independent board member out of 15. It is possible, thus, that the composition of Cemig's board also contributed to BNDESPar's activist stance.

Light
Light is an energy distributor listed on the Novo Mercado. Cemig was a member of its controlling group, which together with the companies linked to the state of Minas Gerais, held approximately 51% of the voting shares. BNDESPar was a relevant shareholder of Light, with approximately 9% of the company's shares, while approximately 39% of the shares were freely traded on the market.
BNDESPar and Cemig clashed about the interpretation of the Law of State-Owned Companies regarding nominations to the board of Light. This law prevents those who acted in a "political party decision-making structure or in work linked to the organization, structuring and conducting of an electoral campaign" in the last 36 months from being eligible to the board of directors of companies with public equity capital. The first use of the law occurred in early 2017 when a lawsuit filed by CVM prevented a Cemig appointed a person to seat on the board of Light. Shortly thereafter, the call and disclosure of the proposals for voting in the 2017 shareholders' meeting included nominations to the board and fiscal board of Light. BNDESPar filed a request with CVM to postpone the meeting claiming that there was non-compliance with the Law of State-Owned Companies because four board and six fiscal board candidacies supposedly violated the law. CVM sustained the board and rejected the fiscal board nominations, conceding a partial victory to BNDESPar (CVM, 2018).
The high ownership concentration of Light did not prevent activism from BNDESPar through administrative proceedings with the market authority because voting power could not prevent it. There was no evident influence of performance as well. The antecedent for activism, in this case, were alleged illegit board nominations that would compromise the independence of Light's board and fiscal board. The regulatory environment, therefore, had an influence on the outcome of activism (Judge et al., 2010). The success of BNDESPar in using the Law of State-Owned Companies regarding nominations for the fiscal board created an important legal precedent. This outcome contributed to greater fiscal board independence and best CG practices, despite the limitations of the legal process (Romano, 2001).

CONCLUSION
The main conflicts of interest involving shareholders in Brazil are between principals, particularly opposing controlling and minority shareholders, as is also common in most countries. The three cases, thus, explore situations in which BNDESPar, the institutional investor, opposed the controlling group or family and address different antecedents of activism, including companies with high ownership concentration controlled by a family (JBS) or public entities (Cemig and Light). BNDESPar is a wholly-owned subsidiary of a federal development bank, one of the largest in the world. It was possible to observe that certain antecedents of activism were recurrent in the cases. Among them, and in line with the literature on antecedents of activism, the performance of the investee companies seemed to be the most significant (Gillan & Starks, 2007).
The conflicts and the resulting activism of shareholders involved attempts to change CG. The CG mechanisms, however, were not able to protect the interests of shareholders even in companies belonging to the Novo Mercado. Consequently, the trade-off between activism and CG suggested in the literature was not observed (Vargas et al., 2018;Gillan & Starks, 2007). It was possible to observe the prevalence of reactive activism by the shareholders caused by principalprincipal conflicts of interest (Young et al., 2008), either due to alleged irregularities, as in the cases of JBS and Light, or in reaction to an already established unfavorable financial situation, as in the case of Cemig and JBS. Aspects related to the institutional shareholders also seem to have significantly influenced their position of activism. In the case of JBS, it was found that BNDESPar's national policy interests conflicted with those of the controlling group and even with those of some large minority shareholders and prevented the investee's corporate reorganization and transfer of the fiscal domicile abroad, possibly perceived as a positive change by others in the market.
As for ownership concentration as an antecedent of principal-principal conflicts of interest and activism, the analysis of the cases suggests that high ownership concentration did not align the interests of large shareholders for a better financial performance of the invested company ( Consequently, there would be less possibility for the success of minority activism through internal CG mechanisms, such as shareholders' meetings. In all three companies with high ownership concentration, however, it was possible to observe an active stance by BNDESPar. One possible explanation is that the immediate interest of the controlling shareholders may not be better financial performance, as suggested by the case of Cemig, and that the ownership concentration does not guarantee protection to minority shareholders, or even if the minority shareholder has interests that overlap with the company's better financial performance, as in the case of BNDESPar's activism against the corporate reorganization at JBS. The fact that it was a minority did not prevent this shareholder from using administrative and judicial processes in its activism, although it hindered its actions at the shareholders' meetings.
It was possible to verify, therefore, that ownership concentration was very important for the definition of the means used by the activist shareholder, as well as for the means of reacting to principal-principal conflicts of interest and the outcome of these actions. BNDESPar was not successful in its actions against the controlling groups through internal CG mechanisms. At JBS, however, BNDESPar was successful in opposing the restructuring proposed by the controlling shareholders because it used a veto clause as a signatory of the shareholders' agreement.
BNDESPar had to frequently resort to external mechanisms through regulatory authorities and the Judiciary to protect its interests. The cases reinforce the importance of the institutional framework to mediate principal-principal conflicts of interest, especially in the case of companies with a high ownership concentration. The application of the recent Law of State-Owned Companies favored BNDESPar in its action at Light, consolidating important jurisprudence towards more independent boards. The JBS case, however, reveals the need for an evolution in the regulatory framework in the identification of cases when major shareholders may be conflicted. Specifically, the analysis of the cases suggests the influence of antecedents of the principal-principal conflicts of interest and activism related to the investee company expected in the literature, such as high ownership concentration, family ownership, size, unsatisfactory performance, and clauses for protection against acquisitions.
Future research could contrast situations of principal-principal conflicts of interest that led to the activism of institutional investors related to public entities with that of other institutional investors, particularly in reference to the background of the conflicts, the motivations for activism, and immediate results. This might indicate whether these findings are due to case selection, if the activism of institutional investors related to public entities is common even in companies with a high ownership concentration, or if the activism is frequent in companies with a high ownership concentration regardless of the kind of institutional investor. In this last situation, the reason for activism would not be related to the public nature of the investor.
It remains evident, however, the difficulty for shareholders to achieve success in their opposition to the majority through the mechanisms of CG under high ownership concentration. In these cases, the regulatory framework was decisive for the outcome of activism, despite of some problems discussed above. Thus, it is possible that the Brazilian CG and regulatory frameworks are still insufficient to protect the interests of minority shareholders, promote better CG practices or affect the performance of companies with high ownership concentration even for large institutional investors related to public entities. This would possibly explain the low number of activist campaigns in Brazil and why previous Brazilian research did not find positive impacts for the presence of institutional investors linked to public entities It can also be seen in the selected cases, as mentioned earlier, that the activism of institutional investors linked to public entities had an influence on performance as a more frequent objective, possibly due to the unsatisfactory results presented by the investee companies. Consequently, it is possible that, in addition to the low number of actions of institutional investors, the focus of this activism on issues outside CG practices is one of the reasons for the difficulty in establishing a relationship between activism and CG practices. In this regard, the analysis corroborates the literature that emphasizes engagement aimed at improving performance.
As a final word, as has already been discussed in the methodology section, case study research can only offer analytical but no statistical generalizations (Yin, 2009). Nevertheless, it can shed light on the potential reasons for the inconclusive evidence of econometric studies that used large samples. The case studies in this article were solely based on publicly available information. Future research can address this limitation by studying cases in which non-public information can be examined and direct observation exerted and interaction with key agents is possible.