Institutions and Policy Choices: The New Left in Brazil

By the late 1990s, the neo-liberal, market-friendly reforms that had swept Latin America at the beginning of that decade were increasingly disappointing. Sustained growth failed to materialize, income inequality and unemployment often worsened, and key economic policies such as privatization, trade liberalization, and market deregulation were marred by corruption and crony capitalism. Worse yet, Mexico (1995), Brazil (1998), and Argentina (2002), which had been the symbols for such reforms, experienced severe economic crises, thus casting serious doubts on the soundness of pro-market policy advice. Riding on such dissatisfaction, left-wing populist leaders running on platforms that rejected the neo-liberal agenda were elected in Venezuela (Hugo Chávez, 1998), Argentina (Nestor, Kirchner, 2003), Bolivia (Evo Morales, 2005), and Ecuador (Rafael Correa, 2007), including the return to power of Daniel Ortega in Nicaragua (2007). Similarly, critical but more moderate in approach were socialist presidents in Brazil (Luiz Inácio Lula da Silva, 2003– 2010 and Dilma Rousseff, 2010–2016), Uruguay (Tabaré Vásquez, 2005–2010), and Chile (Michelle Bachelet, 2006–2010). Among many of these governments, there was an explicit effort to return to the nationalistic model reminiscent of import substitution industrialization (ISI), which was prevalent in most of Latin America from the 1940s through the 1980s. The aim was to recapture the “commanding heights” of the economy in transportation, finance, energy, communications and heavy industry, usually through the vehicle of state-owned enterprises (SOEs), or through a heavy dose of contract renegotiations with private companies coupled with a barrage of stringent regulations across economic sectors (Koivumaeki, 2015). Moreover, this statist approach tried to revamp domestic industry and protect it against foreign competition. As a result, not only did it place great importance on direct ownership of “strategic sectors” through SOEs, but it also emphasized renewed efforts by the government toward directing investments (thus restricting the role of foreign firms and financial markets), managing wages and prices, controlling labor markets, and granting generous subsidies to both domestic producers and urban dwellers. However, there were also significant differences among the new left in Latin America. According to Flores-Macías (2010), the hallmarks of left-wing populist policies were re-nationalizations, price controls, land reform, debt defaults, revocation of Central Bank autonomy, and capital controls. By contrast, the socialist governments in Chile, Uruguay and Brazil avoided most, or all, of these measures while still opting for a statist approach to achieve similar ends. In discussing the policy differences of the new left Weyland (2010: 9) put it succinctly, stating that they all faced the same dilemma: “how to bring about change despite obstacles.” In Bolivia, Ecuador, and Venezuela, presidents used their legislative majorities to emasculate institutional checks and balances and steamrolled their socio-economic policies. Socialist administrations, facing greater constraints, worked within the system. Most of the academic debate about the rise of the new left in the early 2000s centered around social and macroeconomic policies, and their implications for democratic governance (Weyland, 2010, Levitsky & Roberts, 2011). Rufín, C and Manzetti, L. 2019. Institutions and Policy Choices: The New Left in Brazil. Iberoamericana – Nordic Journal of Latin American and Caribbean Studies, 48(1), pp. 40–52. DOI: https://doi.org/10.16993/iberoamericana.435

usually through the vehicle of state-owned enterprises (SOEs), or through a heavy dose of contract renegotiations with private companies coupled with a barrage of stringent regulations across economic sectors (Koivumaeki, 2015). Moreover, this statist approach tried to revamp domestic industry and protect it against foreign competition. As a result, not only did it place great importance on direct ownership of "strategic sectors" through SOEs, but it also emphasized renewed efforts by the government toward directing investments (thus restricting the role of foreign firms and financial markets), managing wages and prices, controlling labor markets, and granting generous subsidies to both domestic producers and urban dwellers.
However, there were also significant differences among the new left in Latin America. According to Flores-Macías (2010), the hallmarks of left-wing populist policies were re-nationalizations, price controls, land reform, debt defaults, revocation of Central Bank autonomy, and capital controls. By contrast, the socialist governments in Chile, Uruguay and Brazil avoided most, or all, of these measures while still opting for a statist approach to achieve similar ends.
In discussing the policy differences of the new left Weyland (2010: 9) put it succinctly, stating that they all faced the same dilemma: "how to bring about change despite obstacles." In Bolivia, Ecuador, and Venezuela, presidents used their legislative majorities to emasculate institutional checks and balances and steamrolled their socio-economic policies. Socialist administrations, facing greater constraints, worked within the system.
Most of the academic debate about the rise of the new left in the early 2000s centered around social and macroeconomic policies, and their implications for democratic governance (Weyland, 2010, Levitsky & Roberts, 2011. Rufín, C and Manzetti, L. 2019. Institutions and Policy Choices: The New Left in Brazil. Iberoamericana -Nordic Journal of Latin American and Caribbean Studies, 48(1), pp. 40-52. DOI: https://doi.org/10.16993/iberoamericana.435 However, these works payed scant attention to how left-wing administrations tried to reconfigure their development goals with respect to privatized industries and those (SOEs) which escaped state divestiture in the 1990s. In Argentina, Bolivia, and Venezuela, governments renationalized many companies privatized in the 1990s. In Brazil, Latin America's largest economy, this did not occur. Instead, the federal government embarked on ambitious industrial projects prioritizing selected sectors controlled by SOEs and government-financed national champions among domestic conglomerates.
Why did the socialists in Brazil not re-nationalize? One may argue that privatization was less important to the neo-liberal agenda in Brazil, as opposed to other Latin American countries, and by 2003 it had run its course anyway. However, if this was the case re-nationalization would have been easy to accomplish, but never took place. To answer this question, we propose an alternative explanation using an analytical approach based upon the insights of new institutional economics (Henisz, 2004). Our contention, and first contribution to the current debate on the new left's policies in Latin America, is that the divergence between Brazil and other new left governments in the region rests exactly on the "obstacles" that Weyland underscored earlier. We argue that Brazil's policy makers faced significant institutional constraints on re-nationalization, while still retaining control over a significant set of SOEs. This prompted Brazilian socialist presidents to circumvent such obstacles by using SOEs to achieve similar economic development goals that their populist colleagues were pursuing during the same period. In addition, whereas the new institutional economics is typically silent about ideology and public opinion, we bring in these additional factors as necessary for a full explanation of the behavior of policy makers. Thus, our analysis also contributes to the existing literature by showing how major changes in leaders' ideological tenets and public opinion interact not only with each other but also with institutional arrangements, leading to different degrees of government opportunism, and therefore explaining more fully the degree of volatility of economic policies and reforms.
The next section of the paper explains our theoretical approach, relative to the existing literature, and our choice of Brazil as a case to examine the analytical power of the approach. This is followed by the examination of policy choices of the PT administrations under the lens of our framework. The final section concludes.

Defining and Explaining Opportunism: The Role of Political Institutions
The multidisciplinary nature of the new institutional economics (henceforth NIE) is well suited to studying socioeconomic and political changes through the examination of institutional arrangements (North, 1990). It allows us to see how economic and political incentives shaped by these arrangements affect the decisions of policymakers and their willingness to meet contractual obligations with investors (Spiller & Tommasi, 2007), thus enabling us to account for time inconsistency in policymaking. In other words, NIE provides a comprehensive analysis of the conditions that allow states to enforce, alter, or violate property rights and contracts, shaping incentives for what economists define as opportunism, which fuels government or company efforts to change agreed contractual obligations. Spiller and Tommasi (2007) define government opportunism as the expropriation of private property resulting from institutional failures, especially in regulatory governance. These authors distinguish two types of expropriation: outright confiscation of assets (nationalization) without compensation for their book or market value, or "indirect expropriation." The latter case is subtler in nature. It occurs when the government restricts a company's ability to recover its investment (including the opportunity cost of the capital) through regulatory measures that affect the size and use of company cash flows. The concept of indirect expropriation thus refers to the commitments made by governments to private investors regarding the conditions under which the investors can recover the capital invested and earn a return on this capital.
Not explicitly considered by Spiller and Tommasi, but involving the same logic, is government opportunism that affects not only companies that are fully in private hands, but also the investments made by minority private shareholders in SOEs. In these instances, minority shareholders commit their resources in the form of SOE shares, with the expectation that the SOE will use this capital to invest in projects that, on average, will return the opportunity cost of capital to these shareholders. When government interventions permanently impair the ability of the SOE to repay this capital and its opportunity cost to the minority shareholders, they are in effect suffering an indirect expropriation. In fact, this form of opportunism is another manifestation of the risks of agency in "principal-principal" agency relationships, which has received ample attention in the corporate governance literature for the case of private majority and minority shareholders (Shleifer & Vishny, 1997).
In fact, the concept of government opportunism can be applied to an SOE's "public" investors as well. The creation of an SOE entails the long-term commitment of public resources, contributed by taxpayers, for a specific purpose embodied in the decision to create the SOE, such as the exploitation of certain mineral resources, the construction and management of public infrastructure, the accumulation and investment of capital, or the delivery of public utility services, among many other possibilities. The investment of taxpayer resources into the SOE may be made in exchange for the promise to deliver socially valuable outputs like the ones previously mentioned, rather than an attractive financial reward as in the case of private investors. But the concept of opportunism, in the form of expropriation of taxpayer investments, can still be applied in this instance if the resources obtained from taxpayers are instead used for corrupt or fraudulent purposes, or for political gains at the expense of the fulfillment of the SOE's purpose (Kane & Christiansen, 2015;World Bank, 2006;Zeckhauser & Horne, 1989). Paralleling the case of minority SOE investors, this form of opportunism is in fact a specific manifestation of the risks of "principal-agent" relationships, amply researched, for the case of private shareholders, by finance and corporate governance scholars (Jensen & Meckling, 1976).
Lastly, the converse of government opportunism is company opportunism. Company opportunism occurs whenever a private-sector company is able to take advantage of institutional failures for the benefit of its managers or shareholders, at the expense of taxpayers or citizens in general-the same stakeholders who suffer under government opportunism against SOE public investors. Typically, company opportunism involves efforts by private companies to gain control over public assets for the companies' own benefit, at less than the market value of these assets, for example through corrupt privatizations; or to change the terms of contracts with the public sector, or of regulatory conditions ("regulatory capture") in ways that yield profits above opportunity costs.
Institutions play a key role in fostering or limiting opportunism: the risk of opportunism is highest when the capacity of the legislature and the courts to enforce constitutional checks and balances on the executive, is weak or ineffective. Likewise, in sectors where regulatory governance (including contract enforcement) is entrusted to specialized agencies, risk is high when such institutions lack autonomy from political authorities or are susceptible to "regulatory capture" by one or more companies under their jurisdiction.
Thus, opportunism thrives in weak institutional arrangements (i.e., ineffectual legislative-judicial and administrative oversight), which leaves the executive with large political discretion for interpretation and enforcement of contracts according to its strategic calculations. Opportunism is most likely to affect long-term investments when politicians: (i) enjoy strong executive powers that can easily overcome checks and balances (legislature, judiciary, or civil service and regulatory agencies); (ii) have short time horizons due to electoral cycles; and (iii) face low indirect institutional costs in reneging on contractual obligations (Spiller and Tommasi, 2007). Not surprisingly, then, given the prevalence of constitutions and historical traditions that favor strong executives, electoral cycles that typically do not exceed four years, and weakly institutionalized polities, extensive instances of company opportunism and of opportunistic manipulation of SOEs had already been documented throughout Latin America even before the wave of privatizations of the 1990s (Chong & López de Silanes, 2005;Savas, 1987;Tanoira, 1987).
The last factor, institutional costs, is perhaps the least obvious but just as important as the other two factors (Post and Murillo, 2016). Its importance rests on the fact that institutions not only constrain choices through the mechanics of decision-making but also through the legitimacy that they can provide to policy choices. Unilateral decisions can be constitutionally valid but still backfire if perceived by other political actors as illegitimate. Thus, to the extent that constitutional rules and their implementing laws have legitimacy for political actors, sanctioning controversial policies like re-nationalizations through a combination of congressional laws and regulatory changes can enhance the possibility that they survive an administration after it leaves office. 1 Below, we examine government and company opportunism during the administrations of Lula and Rousseff. We contend that both presidents faced institutional hurdles in Congress, in the Judiciary, and subnational entities (states and municipalities) that restrained the ability of the executive to go back on the economic reforms of the 1990s, and limited opportunistic behavior to the least constrained policy arm of the Brazilian executive-the federally-owned SOEs.

Ideology and Public Opinion
NIE does not offer a full explanation of opportunism. Although it helps us understand the conditions that facilitate or impede opportunism, it does not explain the motivations that may lead a specific government to renege on prior commitments, neglecting important contextual factors that influence those conditions in the first place, and which are crucial in swinging the balance in favor of government opportunism. To understand these motivations, we complement our institutional framework with two political variables: ideology and public opinion. For example, some analysts argue that in crisis times, the mood of a large segment of public opinion can turn negative, inducing politicians to make important policy changes (Erikson, 2015).
Let us first assess the role of ideology. When economic theory takes the form of an ideology, and becomes normative in character, it can be used to justify major socioeconomic overhauls (Yergin and Stanislaw, 1998). Ideology thus played a significant role in both privatization and re-nationalization trends in Latin America and beyond. The adoption of privatization in Latin America mirrored a worldwide debate over the role of the state, which saw advocates of economic liberalism prevailing over their Marxist and Keynesian intellectual rivals by the late 1980s. Conversely, by the early 2000s, the failure of market reforms to promote growth, often associated with corruption allegations, crony capitalism, and (in some cases) worsening income inequalities, revived the appeal of economic theories advocating economic nationalism (Kingstone, 2011). For instance, in their survey of electricity regulation in Latin America, Martínez-Gallardo and Murillo (2011: 352) contend that "the influence of ideological legacies-especially regarding the degree of state intervention-on the [regulatory] reform of public utilities" was paramount. In their view, ideology provided leaders a heuristic device that helped them to decide on different policy options and select the experts most suitable to carry out their preferred choice. Accordingly, ideology here is defined as a politician's stance regarding the degree of state intervention in the economy.
Public opinion is also a key factor in influencing policy choices. Latin American politicians are just as keen as those elsewhere to manipulate shifts in the public mood to achieve their short-term interests. Conservative and left-wing politicians alike, despite their opposing views, share a strategy of using major public opinion swings to legitimize diametrically opposed policies (Ardanaz et al. 2010). In economic policy, the poor performance of SOEs led to significant popular support for state divestiture by the end of the 1980s, a fact that conservative presidents in the region exploited to justify their privatization programs in the early 1990s. The Latin American public honeymoon with state divestiture continued well into 1995, when 75 percent of respondents still supported it (Chong & López de Silanes 2005).
However, starting in 1998, the trend reversed itself as the public's dissatisfaction with post-privatization outcomes rose in many countries (Baker, 2010). By 2003, public satisfaction with privatization in Latin America had dropped to its lowest rate, 22 percent (Latinobarómetro, 2009). Not surprisingly, anti-market reform presidents in Venezuela, Argentina, Bolivia, and Ecuador capitalized on this public opinion U-turn to launch their statist policy platforms.
Therefore, the interaction of public opinion and ideational factors can produce policy reversals at critical junctures (such as economic crises), when upstart politicians are able to exploit popular malaise and channel it toward their ideological views (Mora y Araújo, 1991), radically altering the government's willingness to manipulate state-market relations. Summing up, our analytical framework adds to Spiller and Tommasi's (2007) model two contextual factors that help to explain the causal link between political institutions and government opportunism: the role played by public opinion and the ideological inclination of the administration in charge (Figure 1).
These two additional factors make institutional explanations of privatization and re-nationalization more comprehensive. Whereas previous research has underscored the importance of these two factors in their own right in the context of contract renegotiations (Post & Murillo, 2013), we show how major changes in leaders' ideological tenets and public opinion interact not only with each other but also with institutional arrangements, leading to different degrees of government opportunism, and therefore explaining more fully the degree of volatility of economic policies and reforms.
Methodologically, we use the single case study to provide chronological and descriptive depth to our analysis (Eckstein, 2002). Our case study is nonetheless focused because we deal only with cases of government opportunism and related company opportunism to guide our data collection. Moreover, we use independent variables (institutions, ideology, and public opinion) whose explanatory power can be conducive to theory building for a broader number of cases in future research.
To establish the validity of our argument, we selected Brazil, which is not only Latin America's largest economy (and indeed one of the world's largest), and thus highly influential even beyond the region, but also offers clear examples of the policy reversals experienced throughout Latin America from market-oriented towards statist policies and back. Moreover, the unprecedented revelations unfolding in Brazil regarding the corruption of its political and business elites under the PT governments offer a rare opportunity to draw on extensive information about the use of SOEs by these governments, including their relationships with the private sector. Although we only examine the case of Brazil, our research question is formulated in relation to other new left governments in Latin America, and our analysis highlights throughout the text the temporal variation in the explanatory variables of ideology and public opinion.

Explaining Policy Choices under the PT Administrations (2002-2016) PT Ideology and Brazilian Public Opinion
The early 1980s saw the exhaustion of the "national developmentalist" model that began with Getúlio Vargas ' Estado Novo (1934-1945 and received particular emphasis under Vargas' second period as president (1951)(1952)(1953)(1954), his successor Juscelino Kubitschek (1956Kubitschek ( -1960, and reached its pinnacle under the military regime of 1964-1985(Musacchio and Lazzarini, 2016. A big foreign debt burden, the hyperinflation that followed, and the end of the military regime, eventually resulted in a major change of direction. In 1990, President Fernando Collor de Melo (1990Melo ( -1992 launched a limited privatization program, but it was President Fernando Henrique Cardoso (or FHC, 1995 who embarked upon an ambitious state divestiture plan. Sixty-eight SOEs were privatized and the remaining SOEs were reformed to operate like private-sector firms, with the state exercising arms' length supervision and even, in cases like Petrobras, the sale of minority blocks of shares to private investors (Hunter, 2010: 64). In addition, many of these companies were placed under the purview of newly-created regulatory agencies, which were supposed to limit the companies' freedom of action. Cardoso also introduced some legislation in Congress to prevent the use of the financial entities still in public hands, like Banco do Brasil and the Banco Nacional de Desenvolvimento Econômico e Social (BNDES), as vehicles for off-budget, discretionary spending by the federal government.
However, Brazil's privatization was largely a reaction to amend the failures of ISI. FHC embodied more than anyone else this change of heart. As a young academic he had been one of the most sophisticated proponents of economic nationalism in Latin America, but by the early 1990s he embraced neoliberalism as the only option left to revitalize the Brazilian economy. The Brazilian public was at first delighted with Cardoso's ending of hyperinflation, allowing him to win the presidency in 1994 and 1998, but privatization never gained the level of popular support it had in neighboring Argentina in the early 1990s. Consequently, FHC justified state divestiture on pragmatic grounds, primarily to modernize key sectors of the economy and use some of its proceeds to pay down federal and state debt. In the second half of the 1990s, conservative and centrist politicians endorsed FHC market-friendly policies, particularly in terms of opening trade and market deregulation. Privatization, however, did not garner the same level of congressional support nor was ideologically driven. This was underscored by São Paulo Minister of Transportation Plinio Assemann, when he told U.S. investors that " [We] are not engaged in privatization for any ideological reason; we have no other option" (D'Orto, 2000).
There is a considerable debate about the ideological orientation of political parties in Brazil, particularly in terms of policy preferences, after the return to democracy in 1985. Figueiredo and Limongi (2000) and Power and Zucco (2009) argue that a reasonable level of ideological differentiation exists on a left-right continuum, whereas Lucas and Samuels (2010) reject this thesis, contending that the only truly ideological party is the PT. In our case, the PT's ideological opposition against state divestiture was explicit in its "Programa de governo" platforms from 1989 to 2002 (Alves et al. Maciel, 2013). Its leader, Lula, charged that privatization was "an insane act that hurt national sovereignty," and Leonel Brizola (Partido Democratico Trabalhista) went as far as suggesting that if there was no civil reaction to privatization, then a military coup might be justified (Rhodes, 2005: 124). The rationale behind these statements was that state divestiture would lead to massive layoffs, increase the gap between rich and poor, and end national sovereignty over strategic sectors with the arrival of foreign investors (Alves et al. 2016). Such concerns resonated well with the public. In 1997, 68 percent of respondents in Sao Paulo opposed the sale of Telebrás (Rhodes, 2006: 124) and the PT used this data in Congress to force major concessions in high-profile privatizations affecting the oil, mining, and telecommunications (D'Orto, 2000). Moreover, during their unsuccessful 1998 presidential bid, the left-wing ticket Lula-Brizola made state divestiture a core issue of their platform, promising to halt remaining privatizations and revise existing concession contracts, possibly even nullifying the Telebrás concession contract.
During Cardoso's second term (1999)(2000)(2001)(2002), as happened in Argentina and Mexico, privatization and other neoliberal policies increasingly soured on the population.  (Latinobarómetro, 2003). In Latin America, only Argentines showed greater disapproval ratings. 2 This led some analysists to conclude that the "rejection of neoliberal economic policies by the majority (including a significant part of the elite), were evident in the opinion polls long before the 2002 elections" (Mollo and Saad-Filho 2006: 99).
Prior to assuming office in 2003, Lula decided to retain some of FHC's monetary and fiscal policies, which were popular with middle class voters and investors. 3 Noticeably absent from these concessions was the continuation of state divestiture. Indeed, Lula's ideological stance against privatization remained unmoved and, emboldened by the opinion disapproving of privatization just mentioned, he terminated plans for state divestiture for many SOEs still under federal control (Haber et al. 2003). 4 Moreover, as the commodity boom gained momentum in the mid-2000s, Lula, and his hand-picked successor Rousseff, progressively reversed course from the neoliberal policies they inherited, reviving economic nationalism through a "New Developmentalism" model (Bresser Pereira, 2004).
In the aftermath of the 2008 U.S. and European financial crisis, and of the apparent success of state-centered policies in Argentina, Bolivia, Ecuador, and Venezuela, the PT's left-wing ideologues and its core constituency of civil servants did not waste time in pushing for a much greater role of the state-which, given Brazil's federal structure, meant the federal government-in the Brazilian economy. This was particularly evident under Rousseff, who moved quickly to increase government control "over SOEs and their prices, thus undoing some of the reforms of the 1990s, which had tried to isolate SOEs from political intervention" (Musacchio & Lazzarini, 2016: 130). In her plans, SOEs played a pivotal role as they retained an enormous clout in steering the overall economy in a desired direction. In fact, in 2014 federal and state SOEs still represented 33 percent of Brazil gross domestic product compared to an average of 15 to 30 percent in advanced industrial countries (Musacchio & Lazzarini, 2016: 130).
As in its previous incarnation, this neo-developmentalist policy was focused on the strengthening of Brazil's industrial base and technological capabilities (Ban, 2013). The goal was to concentrate in Brazil the largest possible amount of industrial activities to build a strong, diversified, and technologically advanced economy (Döring et al. 2017). A second aim was to provide macroeconomic impulse by increasing government spending at a time of contraction in advanced economies resulting from their 2008 financial meltdown. To achieve the first objective, the PT governments sought a higher level of domestic investment through increased federal budget spending in key areas. They included transport infrastructure, import substitution measures to stimulate foreign direct investment into Brazil (particularly in the oil and car industries), and greater domestic investment by SOEs, and private sector "national champions" supported by the government, like conglomerates Odebrecht (construction) and JBS (meatpacking). The second objective would be partly fulfilled by the increased government spending and cheap credit from the (BNDES) (Reid, 2014: ch.8). 5 Thus, SOEs were essential in the pursuit of the industrialization and macroeconomic stimulus goals of the PT governments. 6 To increase electricity generation capacity, Eletrobras and several other public electricity companies invested in the development of massive and controversial dams in the Amazon basin as private investors were not attracted by the projects, and SOE Nuclebras undertook construction of two new nuclear plants at its Angra site, Brazil's only nuclear generation station (Bloomberg, 2015). To foster the development of a domestic equipment industry for oil exploration and production, the government required Petrobras to source domestically 70% of the equipment needed to explore the new giant offshore fields it had discovered (Financial Times, 2017a, 2017bReid, 2014: 153).
SOEs also played a major role in providing macroeconomic stimulus. 7 After the upswing of the international commodity cycle began to falter in 2012 and inflation began to rise in Brazil, Banco do Brasil cut its interest rates and expanded its domestic lending considerably to help prop up the economy (Reid, 2014: 219). Moreover, BNDES carried out a series of transactions with the Treasury to enable significant off-budget spending. Even Petrobras was enlisted to help with macroeconomic stimulus by keeping gasoline prices below costs of production (Tautz et al. 2014).
The Lula administration also relied on the privately owned national champions to pursue its neo-developmentalist policy of domestic investment. This became especially evident in the case of Vale, the world's largest iron mining company. Although Vale had been privatized in 1997, various public pension funds had bought sufficient shares of Vale to have effective control over the company if the government required them to work together on matters of Vale's governance. Vale's chief executive officer (CEO), Roger Agnelli, came into conflict with Lula when he planned for a significant expansion of the company outside Brazil, to attain the scale that would allow it to compete with its rivals in world markets, leverage its expertise, and diversify its sources of minerals. The Lula administration wanted Vale to invest instead in downstream domestic sectors, and specifically in steelmaking within Brazil, so it replaced Agnelli with a CEO who would heed the government's plans (Reid, 2014: 218).
BNDES was also enlisted in industrial policy, lending heavily to some of Brazil's largest private-sector companies to support their growth (Financial Times, 2017b;Sierra, 2015;Tautz et al. 2014), and investing in shares of former and current Brazilian SOEs through its investment banking arm, BNDESPar. 8 Importantly, these investments were intended to support the national champion leg of the PT's neo-developmentalist policy, rather than constitute a "back-door" re-nationalization, since BNDESPar does not appear to have sought managerial control over the companies it invested in (Musacchio and Lazzarini, 2016). Thus, in the case of Vale, although by 2008 BNDESPar owned 6.9% of its shares plus 11.5% of an investment fund (Valepar) that held a majority of Vale's shares, 9 when it came to shaping the decisions of Vale management, the PT governments intervened in the company via the public pension funds, rather than through BNDESPar (Aldrighi & Postali, 2010;Musacchio & Lazzarini, 2016).
Summing up, the PT's ideological aversion to privatization, reinforced by the public's opposition to this policy, put an end to state divestiture under Lula and Rousseff. In fact, the PT's neo-developmentalist ideology led PT governments to rely on SOEs as key instruments to put this ideology into action. Yet, even at the height of this ideological influence under Rousseff, the PT did not engage in re-nationalizations, unlike other left-wing governments in the region. In the next section, we will show how institutional constraints prevented the PT presidents from re-nationalizing the companies that had been privatized under FHC. This reinforced the PT presidents' reliance on the SOEs still under federal control to implement the neodevelopmentalist agenda. Abranches (1988) defined the Brazilian political system as "coalition presidentialism" noting that although the presidency has unusually broad powers to introduce legislation and administrative changes, governments must engineer multi-party Congressional alliances to pass legislation. This is because Brazil has one of the highest party fragmentations in Latin American legislatures (Jones & Hwang, 2005) since the proportional representation electoral law (without a minimum threshold) creates low barriers to entry for small parties. Moreover, whereas ideology and party discipline are hallmarks of the PT, they are largely absent in centrist and conservative parties, which tend to be mostly pork barrel, clientelistic machines. This explains why, in the latter group, party switching is common when legislators do not get what they want from their leaders or other parties promise a better deal. Indeed, congressional rules do not penalize deputies for switching parties during their incumbency (Reid, 2014: 279). This means that congressional representatives may defect any time, making presidential coalitions tenuous, particularly when controversial bills are on the floor. The difficulty that Lula and Rousseff encountered in keeping together their heterogeneous congressional coalitions explains several corruption scandals involving vote-buying. 10 This created high risks of violating anti-corruption laws, and many instances of side payments of different kinds to legislators outside the PT. 11 In sum, the Brazilian Congress is a major stumbling block for the speedy enactment of the presidential legislative agenda. During the Lula-Rousseff era the PT lacked an outright majority, and the fragmentation of the legislature created very high transaction costs to pass any kind of controversial legislation as the government would have to cobble a coalition of willing legislators. In this context, a divisive measure like re-nationalization would have been very difficult to pass, as most of the parties in the legislature, including the PT's main coalition partner, the Partido do Movimento Democrático Brasileiro (PMDB), held positions contrary to re-nationalization and general expropriation. 12 Therefore, Lula and Rousseff, in pursuit of their ideological aims, faced a much more formidable array of limitations on executive power than, for instance, left wing populist leaders in Argentina,Bolivia,and Venezuela. 13 Within this context, SOEs proved to be an expedient means to circumvent the problem and purchase political support, both directly through patronage and other SOE actions, and by facilitating company opportunism in exchange for company bribes and illegal campaign donations. This took two forms. First, PT administrations appointed representatives of friendly coalition parties to SOEs boards, which allowed them in turn to dispense company contracts and jobs to their own clienteles (Coelho, 2017). Second, SOEs engaged in outright corrupt deals to garner support. 14 The biggest scandal, still very much unfolding, implicated oil giant Petrobras' actions under direct presidential control. 15 The massive amounts of money involved in the development of new offshore fields, to be spent domestically according to the procurement rules established by the PT administrations, led to a wide-ranging scheme to rig the contract awarding process, pad the procurement contracts, and share the proceeds obtained by the winning companies, such as construction giant Odebrecht, with the PT and its allies, particularly the PMDB. The funds involved were then used to finance electoral campaigns and personally enrich legislators and other persons involved in the scheme (PR Newswire, 2017;Smith et al. 2015). Other SOEs were enlisted in similar schemes as well. For example, the BNDES played a pivotal role through its financial largesse (Musacchio & Lazzarini, 2016). "[T]he proportion of BNDES loans that a [private business] group received also often corresponded with the size of its political donations" (Financial Times, 2017b). In any case, these scandals show how costly it is for Brazilian governments to pursue policy initiatives through the federal Congress.

Institutions: Executive Power, Congress, the Judiciary, and State Government
Besides Congress, the Brazilian judiciary, including the Supreme Court, exercise an additional constraint on direct actions by the government with regard to economic policies. Indeed, in Latin America, the Brazilian judiciary ranks as one of the most independent in contrast with Argentina, Bolivia, and Venezuela (Figure 2). Even though 14 years in power gave the PT the opportunity to appoint several justices sympathetic to the PT's ideology, if anything the independence of the Supreme Court appears to have increased under Lula and Rousseff, particularly when it confirmed the sentences of several high-ranking PT officials involved in a congressional vote-buying scandal. 16 Since the Brazilian 1988 Constitution protects private property rights, attempts to renationalize or expropriate companies without full compensation to their private owners would have resulted in legal challenges that the government would not necessarily win. In addition, an attempt to circumvent this possibility by diluting the independence of the Supreme Court would have had a high indirect institutional cost, by eroding the political consensus around the desirability of judicial independence and polarizing the political system. Lastly, the highly decentralized structure of the Brazilian state, which confers substantial powers to states and municipalities, was another limitation on the federal government's direct policymaking powers. Almost half of all federal spending (42% in 2010) was earmarked for the states and municipalities through formulas that left little room for discretion (Reid, 2014: 272), and since many states and municipalities were governed by parties other than the PT, this kind of expenditure could not be easily channeled towards the goals of the federal government. Most states owned their own water and sanitation companies, and even two of Brazil's largest electricity companies, the Companhia Energética de Minas Gerais (CEMIG) in Minas Gerais and the Companhia Paranaense de Energia (COPEL) in Paraná. Decisions about these companies were subject to local politics in each state: when the Cardoso administration pursued its privatization program, it could not compel the governments of Minas Gerais and Paraná to privatize CEMIG and COPEL, respectively, since these companies were fully in the hands of each state. 17 In contrast with this situation, the use of federal SOEs for policy goals constituted an important alternative. Most SOEs are not subject to legislation limiting the appointment of directors, CEOs and top management teams. Although SOEs with publicly traded minority blocks of shares, like Petrobras, were subject to Brazilian corporate law, the law only protected minority investors from the fraudulent use of corporate resources by directors and top managers, not from the consequences of decisions about corporate strategy (Tautz et al. 2014). Even the issuance of American Depositary Receipts-a mechanism for tapping US financial markets without having to list shares in the US-imposed no limitation on SOE management, since it did not make the companies subject to US corporate law or the regulatory requirements of the Securities and Exchange Commission. Of particular value was BNDES, which could offer loans at below-market interest rates to private-sector companies seen as "national champions" by the government, in order to fund their domestic and international growth (Financial Times, 2017a).
Although many federal SOEs are subject to the supervision of regulatory agencies, the latter did not constrain SOE behavior when it conflicted with regulatory mandates, despite the fact that the agencies are supposed to be independent from direct government control (Coelho, 2017). The lack of a tradition of regulatory independence in Brazil (virtually all the agencies were established in the 1990s under the Cardoso administrations) and the ability of the government to appoint their own directors has limited their independence in practice. 18 Thus, the Agência Nacional do Petróleo, Gás Natural e Biocombustíveis (ANP) did not intervene when the government forced Petrobras to sell petroleum products below cost, which normally would have been considered an instance of unfair competition for other sellers of these products; likewise, the government was able to impose a drastic cut in the price of electricity from hydroelectric sources, and to delay the passthrough of generation costs to electricity consumers, with little protest by the Agência Nacional da Energia Elétrica (ANEEL); and in the financial sector, the central bank allowed the government to weaken the balance sheet of Banco do Brasil when they cut interest rates and expanded lending under adverse economic circumstances. The deterioration of regulatory quality under the socialist administrations can be seen in the World Bank's "Regulatory Quality Index." Whereas during the FHC administrations (1995)(1996)(1997)(1998)(1999)(2000)(2001)(2002) Brazil scored an average of 0.36, during the Lula-Rousseff period (2003)(2004)(2005)(2006)(2007)(2008)(2009)(2010)(2011)(2012)(2013)(2014)(2015)(2016) the average dropped significantly to 0.07. 19 In getting SOEs to adopt these measures, the PT governments did not hesitate to opportunistically appropriate the potential earnings of private investors and taxpayers. Petrobras suffered major financial setbacks from selling gasoline at a loss, and from having to source equipment at costs that far exceeded those of foreign suppliers. Eletrobras and the other government-owned generators embarked on enormous investment efforts to build the Amazonian dams while at the same time the government was holding electricity prices down-as part of the overall policy of keeping energy prices low-and thus weakening these companies' financial positions. Employment at Petrobras rose 21% from 2007 to 2015, to 78,500 persons, at the same time as company unions achieved salary awards above inflation rates; as a result, Petrobras became the most leveraged listed energy company in the world, experiencing in 2012 its first annual loss in many years. Its shares lost most of their value, wiping out the equity held by private investors but also by public pension funds (Smith et al. 2015;Tautz et al. 2014). Major public pension funds, in fact, have been doubly affected by politicized decisions to focus their investments in SOEs and "national champions," and by the subsequent financial problems experienced by all these companies (including those caused by corruption scandals), forcing the funds to recur to extraordinary contributions from their members (SOE employees) to cover fund shortfalls. For example, the pension fund of the employees of the government-owned bank Caixa Econômica Federal, Funcef, experienced large losses from investments in companies indicted for corruption in the Lava Jato scheme (GaúzaZH, 2017). More generally, taxpayers found themselves facing weakened balance sheets in many of these SOEs, and thus the loss of financial resources contributed by them over time through their tax payments, as well as the loss of the capacity of these SOEs to discharge the duties for which they had been created, due to their financial weakness. Eletrobras, for instance, is unlikely to be able to invest in the electricity distribution companies it owns, which are already the worst in quality of service and operating performance in the country, and thus in direst need of investment, and which provide an essential service in some of the poorest parts of Brazil. Not surprisingly, investors have reacted enthusiastically to government announcements to relinquish control over SOEs. When the government recently announced plans to privatize Eletrobras, its shares rose by 40%, the largest same-day jump in their price since 1993 (Reuters, 2017).
SOEs were the vehicle with the lowest institutional costs for the PT governments to pursue their neo-developmentalist policies. SOEs offered the greatest freedom, whereas other options were precluded by significant obstacles. Federal spending was limited by a requirement that the federal budget maintain a primary surplus (that is, a surplus before payments of interest on the debt owed by the federal government). There would have been significant "indirect institutional costs" in modifying this rule, put in place in the 1990s to prevent a resurgence of hyperinflation and another debt crisis.
In addition, the PT governments had another policy instrument available at lower institutional cost than re-nationalization: the widespread opportunism of most large domestic business groups, which made them highly amenable to pursuing these governments' ambitions in exchange for protection from competition and for financial support. The institutionalized closeness between big business and the public sector in Brazil, dating back to the corporatist structures created by Vargas and reinforced by the military regime (Evans, 1979), and reinforced by the financial and contractual ties between SOEs and domestic business groups described above, made domestic big business both a willing collaborator in developmentalist policies and a formidable opponent of opportunistic re-nationalizations. Brazil had largely eschewed international investment protection treaties, 20 making domestic investors, and particularly domestic companies and business groups, highly prominent in the privatization process (Baer & Bang, n.d.). In contrast to foreign investors, lacking power other than to cut off future inflows of capital, domestic investors could impose much higher institutional costs, for example by cutting off campaign donations to major parties. Thus, even the attempt by President Temer to roll back BNDES preferential support for major business groups was successfully thwarted by them (Financial Times, 2017b).

Electoral Cycles
The elections of 2006, 2010, and 2014 put great pressure on the PT administrations of Lula and Dilma to deliver a host of benefits to many constituencies across Brazil. Not only are the Brazilian presidential elections coincidental with elections to all executive and legislative offices throughout Brazil-federal legislators, state governors and assemblies, and mayors and municipal councils-thereby raising the stakes very highly, but Brazilian electoral rules and voter behavior make redistributive pressures very strong.
To muster support in the legislature, Lula and Dilma had to maintain good relationships with key legislative allies, particularly the centrist PMDB, often at the state and local levels as well in order not to alienate important actors in the PMDB and other members of the governing coalition. Governors, in particular, could wield influence on state-level party organizations and by this means, also on the behavior of federal deputies and senators from the party elected to represent the state. In turn, helping these allies meant directing resources to fund expensive electoral campaigns and sustain party organizations. Election of federal deputies and senators by proportional representation at the state level, in a country with many states with large territories and populations, hugely increased the cost of electoral campaigns in Brazil (Ames, 1995;Reid, 2014: 279-80).
Pork-barrel politics added to redistributive pressures. According to some analysts, Brazilian voters were highly responsive to pork-barrel politics and were thus inclined to support candidates with a record of delivering local public goods, such as school lunches or paved roads, that created local jobs or transferred income to voters and their families (Ames, 1995;Alston & Mueller, 2006). Other political scientists found a more indirect connection, whereby pork-barrel spending sustained campaign contributions from grateful beneficiaries (Samuels, 2002). Programmatic promises had less appeal, although the PT's expansion of the Bolsa Família conditional cash transfer program for poor families under Lula and Dilma created an enormous groundswell of support for the PT, particularly in the poorer Northeast.
SOEs could also be enlisted to undertake investment in other programs with a redistributive and ideological bent. For example, in partnership with Venezuelan stateowned oil monopoly PDVSA, Petrobras undertook the construction of a new refinery, in a major Northeastern state, Pernambuco, in 2007, creating thousands of construction and permanent jobs in a poor region, despite the allegedly doubtful economics of the project (Tautz et al. 2014). 21

Conclusion
Following Weyland, our inquiry focused on the institutional constraints that shape left-wing governments ability to bring about change. In this article we explained the absence of a re-nationalization effort in Brazil under socialist administrations. Whereas left-wing populist governments in the region, such as Argentina, Bolivia, and Venezuela, began a process of re-nationalization of former state-owned enterprises, Brazil, however, did not. This policy divergence constitutes a puzzle that the literature on the new left in Latin America has largely neglected.
Accordingly, we have tried to explain this divergence using a NIE approach, by focusing on the institutional factors that shape governments' ability to renege on prior economic commitments (i.e. government opportunism). Our contribution has been to complement previous NIE studies centered on commitment and time inconsistency issues by showing how the ideological inclination of these administrations and the changed public mood about privatization interacted to create a window of opportunity for a major policy shift in the early 2000s. Within this context, our analysis has brought evidence that domestic institutional constraints are a key explanatory factor. Specifically, we showed how the PT governments in Brazil faced an opposition-controlled Congress and institutional limitations that made the re-nationalization policies carried out by left-wing administrations in Argentina, Bolivia, and Venezuela unviable. This left SOEs as the best available vehicles for the pursuit of ideological goals of the PT's neo-developmentalist economic policies. 22 Admittedly, our framework is not immune to criticism. Space and complexity limit us to a single case, leaving much more work to be done regarding other cases in the region, such as Venezuela or Bolivia. Moreover, single case analyses do not allow controls for alternative explanatory factors; it is possible that other elements, like the sheer size and complexity of Brazil, or the specific leadership traits of Lula and Rousseff, also played a role in the strategies of their governments. Nonetheless, we hope to have provided the foundation for a more comprehensive understanding of this complex and highly politicized policy.

Notes
1 Flores-Macías (2010) convincingly argues that the adoption of statist policies in Latin America was driven by party institutionalization, but as we are limited to a single case exhibiting little variation over the period studied, we omitted this variable in our framework. 2 These results are confirmed by other surveys that considered Brazilians' attitudes towards the role of the state in the economy at the time of the key 2002 presidential election. One survey found a clear majority of respondents in favor of public responsibility for supply across two of out three sectors of the economy, education, electricity, and automobiles: 61percent of respondents supported public responsibility in at least two of these three sectors (Balbachevsky & Holzhacker, 2006 -1998-4dbliwg8pjxuau9wapc4cdcub). 5 According to Carlos Lessa, who became BNDES president under Lula, the new objective of the bank was to turn its role as an agent of privatization into one that would play an active role in promoting, via financing, the new industrial policy of the socialist administration (Bachiller 2016). 6 Interviews with BNDES management, Rio de Janeiro, March 2010. 7 Ibid. 8 BNDESPar had also bought shares of privatized companies under the Cardoso administrations, but those purchases supported SOE privatizations by absorbing some of the shares placed in the market. 9 Valepar was created in 1997 as a vehicle for ownership of Vale's share upon the latter's privatization. Until its merger into Vale in 2017, it held 54% of Vale's common stock and 20 million preferred stock shares. U.S. Securities and Exchange Commission, Form 6-K, Vale S.A., May 2017, Exhibit 5 (https://www.sec.gov/Archives/ edgar/data/917851/000110465917032871/a17-13038_4ex99d5.htm, accessed December 9, 2018). 10 In the case of other private national champions, government support also included foreign expansion, presumably under the rationale that only international scale could guarantee their competitive viability (Sierra, 2015). 11 Congressional vote-buying in Brazil is nothing new, but the PT had always claimed to be immune from it since honesty was one of its most cherished core values. 12 Interviews with PT and PMDB congressional staffers, Brasilia, March 2010. 13 Ibid. 14 Journal O Globo, September 9, 2014 (https://oglobo. globo.com/opiniao/a-corrupcao-na-petrobras-nasestatais-13944986). 15 Veja, January 16, 2018 (https://veja.abril.com.br/ blog/sergio-praca/gilmar-mendes-e-as-origens-da-corrupcao-na-petrobras/). 16 Similar inquiries would have been unthinkable in Argentina, Bolivia, and Venezuela where left-wing administrations had a much tighter control over the judiciary during the same period. 17 Again, in this paper we limit ourselves to policies at the federal level only. 18 Pavón Mediano (2018) shows that Brazil utility regulators in Brazil are more independent than in Venezuela and Argentina, but lag behind Peru, Chile, and Colombia. 19 http://info.worldbank.org/governance/wgi/#home. 20 As of 2016, Brazil had signed 20 Bilateral Investment Treaties (BITs), but none were in force (http://investmentpolicyhub.unctad.org/IIA/CountryBits/27). 21 The refinery was supposed to be finished in 2011, but as of late 2017, it was still unfinished amid cost overruns exceeding $18bn from an original budgeted cost of $2.3bn (Reid, 2014: 197; (http://jconline. ne10.uol.com.br/canal/economia/pernambuco/noticia/2017/09/17/refinaria-abreu-e-lima-em-suape-e-amais-cara-do-mundo-307069.php). 22 It might be argued that the reason why other countries in the region renationalized more is that they had privatized more in the first place-such as, for instance, the post office, which was privatized in Argentina but not in Brazil. But even if this were true, it would be another indication of the greater institutional constraints in Brazil, which would have limited policies both ways: privatizations and subsequent re-nationalizations.