Electoral Dividends from Programmatic Policies: A Theoretical Proposal Based on the Brazilian Case ^*

Abstract

Recent controversies in the literature on the electoral effects of cash transfer programs reveal the limitations of traditional models associating public policies to voting. Why would beneficiaries reward parties for programmatic policies when the government has no control over the distribution of the benefits? Are they guided by short-term retrospective voting in favor of the incumbent without forming durable links? By studying the Brazilian case, based on the Bolsa Escola-Program, formulated by the PSDB, and the Bolsa-Família Program, formulated by the PT, I test different theories and propose an alternative approach. My thesis is that the long-term electoral effect of programmatic policies is a consequence of credit-claiming efforts made by the competing parties, resulting in reciprocity from voters. My argument is built on the examination of parliamentary deliberations throughout the formulation of these programs and on the analysis of public opinion surveys conducted in 2005, a transitional moment when there were beneficiaries of both programs, and in 2018, the year of the first presidential election with the Bolsa-Família Program being implemented with the PT as an opposition party. The data show that party leaders came to the defense of these social programs to varying degrees and in different moments of their terms in office and that these different mobilization strategies reflected on voting behavior.

Bolsa-Família Program; elections; cash transfer programs; parliamentary debates; public policies

On a weekend in May 2013, nearly one million people, mainly from the Northeast region of Brazil, alerted by rumors that the ‘Bolsa Família Program’ (hereafter BFP) was about to be terminated, attended branches of ‘Caixa Econômica Federal’ (CEF) and other accredited facilities to withdraw their benefits from one of the largest and most notorious cash transfer programs in the world ¹ . Specific changes in the transfer schedule not previously informed by the public bank were suggested as a possible source of the rumor.

The federal government, at the time led by the PT, quickly declared that the PBF was not under threat. President Dilma Rousseff stated: “I would like to make it clear that my administration has a strong, deep, and definitive commitment to ‘Bolsa Família’” ² . The opposition, led by the PSDB, called for investigations - which had no further consequences - into the alleged crime of responsibility and misrepresentation committed by the CEF ³ .

President Michel Temer (PMDB), who took office after Rousseff's impeachment in 2016, issued decrees on two occasions authorizing adjustments above inflation to the values of the BFP benefits, which had been frozen since 2014. One of these decisions was made right before the start of the 2018 ⁴ campaign for the general elections. The following year, at the beginning of his administration, Jair Bolsonaro (then affiliated with the PSL) – who had expressed an opinion contrary to and critical of this cash transfer program several times in his career ⁵ - authorized the creation of a "13th payment of Bolsa Família" ⁶ . In 2020, as the country was grappling with the COVID-19 pandemic and emergency aid payments were being delivered ( SIMONI JR., GUICHENEY, and OLIVEIRA, 2021 ), the president expressed his desire to strengthen the BFP, albeit under a different name ⁷ .

These phenomena express the fact that the BFP is politically and electorally important. Some might say that this program has become a valence issue. However, our accumulated knowledge emphasizes that this program has benefited only presidential candidates of a specific party. Several studies show that, in the presidential elections of 2006, 2010, and 2014, the PT - the incumbent party and formulator of the BFP ⁸ - received substantial electoral rewards from the distribution of benefits of this program ⁹ ( CANÊDO-PINHEIRO, 2015 ; HUNTER and POWER, 2007 ; MAGALHÃES et al . , 2015 ; MARTINS et al., 2016 ; MARZAGÃO, 2013 ; NICOLAU and PEIXOTO, 2007 ; PEREIRA et al., 2015 ; SOARES and TERRON, 2008 ; ZUCCO JR., 2015 , 2013 , 2008 ).

With the PT out of government, will the BFP continue to have an electoral effect? And which party will benefit from it? How will the electoral competition over the BFP unfold? These questions, focused on the Brazilian case, lead to a general theoretical discussion.

According to widely adopted typologies ( DIAZ-CAYEROS et al. 2016 ; STOKES et. al, 2013 ), the BFP may be considered a programmatic public policy. In this type of policy, the rules for the distribution of benefits are public, objective, and effectively followed by government agents. This category differs from discretionary policies, whose design allows the government to strategically use the allocation of resources according to the voting behavior of citizens ( GOLDEN and MIN, 2013 ).

A report released in 2006 by the ‘Tribunal de Contas da União’ (Federal Court of Auditors), covering the period when the program most expanded, concluded that “there is no evidence (...) that the program was used for electoral purposes” ( TRIBUNAL DE CONTAS DA UNIÃO, 2006 , p. 01). Furthermore, contrary to some expectations, the majority of the Brazilian population believed that the BFP would not be terminated with the PT leaving the federal government, as shown by qualitative evidence brought by Sugiyama (2011) and quantitative evidence from the Brazilian Electoral Panel Studies of 2014 ¹⁰ ( AMES et al . , 2016 ).

According to traditional models of distributive politics, it is irrational for voters to weigh in their voting decisions a public policy of which the competing parties are not responsible for directing the benefits. Electoral effects would come only from pork-barrel projects (such as public investments for geographically concentrated public works), clientelist politics (irregular distribution of benefits during a political campaign, for example), or patronage (granting of public offices) ( STOKES et al., 2013 ).

The most recent contribution to this discussion came from the reanalysis of De La O's (2015, 2013) data conducted by Imai et al. (2020) on the Mexican conditional cash transfer program ‘Progresa’ , one of the most conspicuous examples of programmatic policy. De La O (2015, 20 , 2013 ) stated that this program had electorally rewarded the incumbent party who had formulated it. However, corrections that were later made by Imai et al . (2020) - in what is alleged to be methodological mistakes - led the authors to contest this result. These authors claim that the null effect found would confirm existing theoretical expectations about electoral outcomes from programmatic policies. Are the studies on the electoral effects of the BFP also making methodological mistakes, or is it possible to think of alternative mechanisms explaining the electoral returns to programmatic policies?

Zucco Jr. (2013) argues that the electoral dividends generated by these policies are associated with retrospective voting. By analyzing the BFP and the Bolsa-Escola Program (hereafter BEP), which was formulated by the PSDB and from which the BFP originated ¹¹ , the author points out that these rewards were short-term effects. Voters would support the incumbent who distributes the benefits from these programs, regardless of whether or not he or she is their formulator.

In this article, I propose an alternative approach based on criticisms of these propositions. I argue that the long-term electoral effect of programmatic policies is a consequence of credit claiming and efforts by competing parties to build political platforms, which result in reciprocity from voters. Depending on the features of the electoral competition, those parties formulating and defending political platforms around these public policies may create electoral links with beneficiaries beyond the short-term, even if the policy design does not allow for discretionary distribution.

The Brazilian case contributes to advancing this proposition. First, since the BFP persisted after the PT left the federal government, it is now possible to bypass the problem posed by the overlapping of the incumbent and formulator parties. Second, since the BFP replaced programmatic programs that had been previously designed by the PSDB, and because the transition between programs was gradual, a considerable share of beneficiaries was still receiving BEP benefits during the first years of the BFP.

My empirical analysis combines qualitative and quantitative methods. Legislative debates on the BEP, the BFP, and the ‘Fundo de Combate à Pobreza’ (Fund for Combating Poverty) - from which the first cash transfer programs were financed - were reconstructed and analyzed to capture how the parties came to define their positions on these policies and build their discourses. Public opinion surveys conducted in 2005 – a moment of transition between programs - and 2018 - the first presidential election after the implementation of the BFP with the PT out of government - were analyzed to assess how the benefits impacted voters' opinions in different contexts.

The findings indicate that the PSDB government was reluctant in embracing the BEP, whereas PT legislators advocated that cash transfer policies should be given emphasis. President Luís Inácio Lula da Silva made a considerable amount of effort to implement and expand the BFP during the PT government, despite him facing some resistance from his allies. The analysis of the above-mentioned surveys revealed that the beneficiaries of the two programs had different opinions in 2005: while those included in the PT program - the BFP - were more sympathetic to the PT and Lula, those who received benefits from the PSDB program - the BEP – expressed an opinion similar to that of non-beneficiaries. The 2018 survey, in turn, indicates that the PT was electorally benefited by the cash transfer program despite being out of government. Therefore, from a theoretical point of view, the results suggest that programmatic policies may reward the parties advocating for these policies, an effect that goes beyond the incumbency and the short-term; for how long this effect will last and what will be its features depend on the parameters of the political competition.

This article is divided into two sections, in addition to this introduction and the conclusion. In the first section, I provide a critical assessment of the models of distributive politics and studies on cash transfer programs in Brazil and Latin America in order to propose an alternative approach. The second section presents the empirical data, initially with an analytical narrative based on the examination of documents related to the legislative formulation processes of the ‘Fundo de Combate à Pobreza’, the BEP, and the BFP, and subsequently with data analysis of the 2005 and 2018 surveys. The conclusion summarizes the main points.

Distributive theories and electoral effects of public policies

The logic of electoral competition pervading the distribution of public resources may be analyzed within the extended framework of Downs' spatial model (1957). According to Downs (1957) , political competition occurs in a one-dimensional ideological axis in which parties move in search of votes. Voters would have a fixed ideological position, and the only strategy available to political parties is to change their platforms.

The ‘distributive politics’ models add another dimension to the Downsian equation: the distribution of goods and services by governments ( GOLDEN and MIN, 2013 ). According to these models, parties competing for votes do so not only by moving along the ideological spectrum but also by making promises and allocating specific policy benefits. A voter takes two dimensions in consideration when deciding her vote: the distance between her ideological preferences and those of each of the parties, and the promises of policy benefits or the benefits that were in fact distributed by the parties. Formally, the equation could be formulated as:

$U_i=-\frac{1}{2}{\left(v_i-x_i\right)}^2+b_i+{\epsilon }_i$ (1)

Source: Elaborated based on Stokes et al. (2013) .

Where U _i is the utility of i's vote in party A or party B ¹² , v _i = {v _A ,v _B } represents the parties' ideological position, x _i is the voter's ideological position, b _i represents the value of the resources received by the voter, which were directed or promised by the party, and ε _i is the random term.

Voters respond differently to the distribution of public policy benefits. Most studies consider that the optimal strategy for parties consists in allocating resources to voters who do not identify with any party (swing voters) ( LINDEBCK and WEIBULL, 1987 ; STOKES, 2005 ) – a proposition that has been criticized, as we will later see. Nevertheless, there is a consensus among all theoretical approaches that, for poor voters, the marginal effect of public policy resources - the term “b _i ” - is quite significant, which further justifies the analysis of the specific type of benefit directed to the most vulnerable social groups, as is the case of this article.

Much progress has been made - with case studies and comparative works - in understanding the electoral effects of conditional cash transfer policies in Latin America (cf. CORRÊA, 2015 ; CORRÊA and CHEIBUB, 2016 ; LAYTON and SMITH; 2015 ; LINOS, 2013 ; MANACORDA et al., 2011 ; NUPIA, 2011 ; WINTERS, 2010 ; ZUCCO JR., 2013, among others) ¹³ . The Mexican case studied by De La O (2015 ; 2013 ) has become a textbook example in the literature. Taking advantage of the fact that the ‘Progresa’ was initially implemented as an experiment, the author collected data showing that the distribution of benefits of this program had increased the percentage of votes for the incumbent party.

Similar conclusions were drawn for the Brazilian case, one of the most studied and prominent cases in the literature. The BFP is considered to be one of the explanations - possibly the most important explanation - for the presidential election results from 2006 to, at least, 2014, contributing to creating PT's electoral base in the poorest regions of Brazil and among the poorest voters.

However, an intriguing theoretical question arises. The above-mentioned models of distributive politics require that public policies be discretionary so that the parties' strategies for distributing resources have an effect on voting. In other words, it is presumed that governments can direct resources as they wish (to whom, when, and where) ( DIAZ-CAYEROS et al . 2016 ; GOLDEN and MIN, 2013 ; STOKES et al., 2013 ). In the words of Dixit and Londregan (1996) , this is a ‘tactical redistribution’.

Since voters are rational individuals, they are expected to judge parties only by the actions that fall under the parties’ responsibility, such as clientelist politics, patronage, and pork-barrel politics. When making decisions, voters should thus ignore programmatic policies, in which there is no room for discretionary distribution. This mechanism, however, does not explain the electoral effects of the BFP, the Mexican program, and several other cash transfer programs in Latin America since their design, in many cases, hinders or prevents executive discretion ( DE LA O, 2015 ; FRIED, 2009 ; LAYTON and SMITH, 2015 ).

The observation of the mismatch between the theoretical expectations of traditional models of distributive politics and the empirical results motivated the reassessment of the Mexican case by Imai et al. (2020) . These authors replicated De La O's (2015 ; 2013 ) analysis, correcting some apparent methodological mistakes ¹⁴ , and analyzed an additional experimental programmatic social policy in Mexico (SPS -Seguro Popular de Salud) and, contrary to De La O's results, they did not find an electoral effect. The findings of Imai et al. (2020) thus reinforce the theoretical proposition that electoral effects should not be expected from programmatic policies. However, they do acknowledge that studies conducted in other contexts generated positive results. What is, after all, the mechanism that could explain this phenomenon?

The authors offered some tentative hypotheses to explain these results, such as the fact that most of the public policies studied in the literature - unlike the Mexican case – are not likely to have been unanimously approved by the legislature, implying that certain parties could claim electoral dividends.

As we will see, although conflicts emerged between the different political groups during the congressional proceedings related to the ‘Fundo de Combate à Pobreza’, the BEP, and the BFP, they were unanimously - or almost unanimously - approved, even during minority governments. The tentative hypothesis offered by Imai et al. (2020) is thus not applicable to the Brazilian case.

Zucco Jr. (2013) proposes an explanation for the effect of programmatic public policies, based specifically on his analysis of the Brazilian case. This author promoted an empirical innovation in the analysis of the effect of cash transfers in Brazil by incorporating not only data on the BFP for the years 2006 and 2010, as is usually done in the literature, but also data on the PBE for the year 2002. He argues that a mechanism of economic voting could explain the electoral returns brought by these social policies. However, he claims that this retrospective voting is short-lasting, that is, the benefits would entail dividends to the incumbent even if he was not the program formulator or had little or no control over its distribution. This argument could explain why both the PSDB and the PT were electorally rewarded for the cash transfer programs.

There are two implicit assumptions in Zucco Jr.’s (2013) argument. First, that the BFP and BEP may be treated as one public policy - at least from the perspective of my research question. Second, the assumption that once the PT was out of government, the BFP would bring electoral returns to the new incumbent.

Despite Zucco Jr.’s (2013) important contribution, his explanation does not take into consideration that, although the BFP was initially the result of a combination of previous PSDB programs, the PT government vastly expanded the program and changed its design, allowing the party to mobilize electorally around its own political platform ¹⁵ , potentially with long-term effects, especially in a context in which the antagonist party claimed little credit for its own cash transfer program. Moreira (2020) shows that PT and PSDB legislators diverged in their statements about the BFP in its early years, and Araújo and Flores (2019) emphasize that speeches by PSDB legislators were distancing the party from the theme of social policies.

By calling into question other two assumptions shared by the above-mentioned models and explanations, I may explore an alternative proposal. First, it is usually assumed that the two terms of the voting decision equation are independent. In other words, voters' political preferences would not be affected by the dimension of benefit distribution through public policies. Under this assumption, when distributing government benefits, a party could constantly ignore those voters who are close to it or who identify with it without altering these voters’ ideological inclination.

However, as Cox and McCubbins note, “it seems irrational in the long-run for any group to be totally unresponsive to redistributions of welfare” ( COX and McCUBBINS, 1986 , p. 382). In other words, it does not seem reasonable to assume that the ideological and distributive dimensions are orthogonal. Rather, a more realistic assumption would be that party preferences are partly shaped by the benefits received from public policies. In this regard, Diaz-Cayeros et al. (2016) offer a model called “conditional party loyalty”, in which benefits must be delivered to core voters in order to maintain their proximity. In other words, the voter’s current ideological position is in part a function of the past distribution of policy benefits. However, the scope of this model is limited because it applies exclusively to discretionary policies. Drawing on Fiorina's argument (1981) that the individuals’ contingent assessment of parties' actions is a determinant of party preferences, I propose to extend this reasoning to programmatic policies.

The second assumption found in the literature is that parties do not need to mobilize around public policies to ensure electoral returns. According to this view, the distribution of benefits would automatically lead to electoral rewards from beneficiaries who are responding to their increased well-being ¹⁶ . However, it is worth noting that in Mayhew's (1974) original theory on electoral connection, even electoral returns brought by discretionary public policies are not taken for granted: politicians need to build the perception that they are responsible for these benefits.

Corroborated by the empirical results presented in the next section, I argue that two mechanisms may explain the electoral effect of programmatic policies: first, the mobilization and credit-claiming around these policies by political parties, and second, the reciprocity from voters who change their position because of the benefits received. Even if a party has no control over the distribution of resources, it may build an agenda and a political platform in a competitive context, signaling its commitment to the public policy benefits ¹⁷ . However, as Weinschenk (2010) suggests, based on Fiorina's (1981) replication, the success of this strategy in changing voters’ preferences depends on partisan competition over social policies.

I also propose to expand the scope of the concept of reciprocity - traditionally applied only in contexts of clientelist politics ( FINAN and SCHECHTER, 2012 ) – and, drawing on Manacorda’s et al . (2011) suggestion, use it as an explanatory mechanism for electoral dividends from programmatic policies. The idea is to consider that voters who are beneficiaries may support the formulator party - although not unconditionally and permanently - as retribution for the creation of this policy even if they are aware that the distribution of benefits is not conditioned by the choices made by the formulator party and its permanence in power. In this sense, the distribution of benefits may alter ideological preferences beyond the short-term.

To empirically support this thesis, I turn to the case of conditional cash transfer policies in Brazil. The analysis of the legislative debates on public policy proposals reveals that parties behave differently: the coalition led by the PSDB was reluctant to put emphasis on this type of policy, whereas PT legislators pushed for this agenda from the outset. This process of political platform-building affected the voters’ behavior, as revealed by my analysis of the surveys of 2005 - the year of transition between the PSDB and PT policies - and of 2018 - when the PT was in opposition.

Empirical analysis

Analytical narrative on the formulation of the cash transfer programs

Conditional cash transfer programs started to be implemented in local and state governments in Brazil in 1995, usually under the name Bolsa-Escola ( SILVA et al., 2012 ). Interestingly, the first experiences with this type of policy in Brazil were led by mayors and governors of the PT and PSDB ( COELHO, 2008 ).

The visibility and rapid success in public opinion turned this type of program into a panacea. Rocha (2011) stresses that several candidates from different parties proposed to implement cash transfer programs during their mayoral campaigns in 1996 - such proposals emerged even in places where the fiscal and bureaucratic conditions are precarious.

These experiments gradually reached the federal level, controlled by the PSDB at the time ¹⁸ . In 1997, the Fernando Henrique Cardoso (hereafter FHC) government passed a law known as Programa de Garantia de Renda Mínima (Minimum Income Guarantee Program), which aimed to provide financial support to local policies by covering part of their operating costs. In the electoral campaign of 1998, the terms “hunger, poverty, and inequality” became central to PSDB's political program ( THOMÉ, 2013 ). The party had been elected in 1994 with the agenda of reducing inflation, but since 1998 it has sought to put some emphasis on social policies. In fact, according to Draibe (2003) , the major change in the party’s government program concerning social policies between the two elections was the emphasis placed on conditional cash transfer programs. The PSDB program intended to carry out a “new revolution in social and cash transfer policies” ( CARDOSO, 2008, p , p. 03). At that time, however, the predominant view was that cash transfer programs should be under the responsibility of local authorities, with the federal government participating only by providing part of the funding.

The second PSDB term was marked by competition within its coalition over the definition of the presidential candidate in the 2002 election. One of the main sources of conflict was the credit-claiming of social policies. Some analysts note that then-president Cardoso wanted the party's program for the 2002 presidential campaign to focus on this area ( ALMEIDA, 2007 ). However, Congress and its main coalition partner, the PFL, took the lead. In 1999, a Joint Committee to Combat Poverty was installed under the chairmanship of the PMDB and rapporteurship of the PFL to formulate policies to eradicate poverty and inequality. Attached to the committee's deliberations was a Constitutional Amendment Bill (PEC) proposing the creation of the Fundo de Combate à Pobreza - designed to finance social policies - that had been introduced by Antônio Carlos Magalhães (ACM) from the PFL-BA, then president of the Senate and one of the main leaders of the PFL.

The PFL sought to distance itself from the PSDB and nominate a presidential candidate in 2002. By seeking to build a political platform that included social policies, the senator came closer to the left-wing opposition. In a seminar organized by the Instituto Cidadania, an NGO led by Lula, ACM criticized the government, saying that “the steps taken so far to fight poverty have been very timid” ¹⁹ .

In the Committee on Constitution, Justice and Citizenship (CCJC), rapporteur Lucio Alcântara (PSDB-CE) mainly accepted the proposals of ACM's PEC, to the detriment of deliberations made at the Special Committee. The PT was defeated as it argued that the fund had scarce and uncertain resources that had not yet been regulated. In addition, the opposition required that 75% of the resources be earmarked for cash transfer programs, an amendment that was not accepted by the rapporteur. Eduardo Suplicy (PT-SP) also expressed concern that the executive would have the discretion to define the eligibility criteria: "An essential point to be clearly defined by the law is the way in which the transfers will be distributed to deprived populations, (...) so that the criterion for distribution is never left to the discretion of any person or entity, otherwise, there would be distortions” ( SENADO FEDERAL, 2000 , p. 02395).

In the plenary debates, opposition senators accused the government - mainly the then Minister of Finance Pedro Malan - of interfering to limit the access to financial resources for the Fundo de Combate à Pobreza. Melo (2005 , p. 866) says that the executive was initially opposed to the idea of creating a fund for fear that it would make the public budget - which was already quite constraining - even tighter. After months of negotiations - and probably fearing electoral losses - the government decided to support this bill through an agreement. The bill was passed in the second round with only the PT and PPS voting against it.

In the Chamber of Deputies, the advisory opinion of Paulo Magalhães (PFL-ES) is unanimously approved in the CCJC while, in the committee of merit, members of the PT once again voted against it. In the first round of deliberation in the plenary, Aloízio Mercadante (PT-SP), the leader of the party, requests separate voting to allocate all the Fund's financial resources to a cash transfer program. In the words of Jorge Bittar (PT-RJ): "the PT proposes that the fund have a clear focus, and the best instrument to fight poverty today, recognized nationally and internationally, is the Bolsa-Escola Program" ( CÂMARA DOS DEPUTADOS, 2000 , p. 277).

The text was approved by 412 legislators, with just one vote against it - which was cast by current President Jair Bolsonaro, then affiliated to the PPB-RJ. During the process of voting Mercadante's request for separate voting, one of the PMDB leaders, Mendes Ribeiro Filho (PMDB-RS), said that “we ask for the legislators’ understanding as to not say 'no' to the Fundo de Pobreza, to Bolsa-escola, but to say 'yes' to housing, health, and so many other needs and also to 'Bolsa-escola'”. Similarly, PFL leader Inocêncio Oliveira (PFL-PE) stated that “to define only the bolsa-escola is to hold this PEC back” ( CÂMARA DOS DEPUTADOS, 2000 , p. 306). An agreement was signed between the government and the opposition, the request for separate voting was withdrawn, and the bill was passed in the second round by 379-1, again with only Bolsonaro voting against it.

In the words of Melo (2005 , p. 867) “the PFL got most of the political rewards” over this matter. The PSDB, therefore, did not take a leading role in the creation of the Fundo de Combate à Pobreza - the party was rather reluctant about it - nor did it endorse the creation of national cash transfer programs from the outset, contrary to the position assumed by the PT in several instances of the processing of this proposal.

Seeking to change this image and its political platform for the 2002 election, the FHC government, at the end of its second term, consolidated a set of social policies into a brand called “ Rede de Proteção Social " (Social Protection Network). The first conditional cash transfer programs ²⁰ formulated by the federal government took a prominent position in this network: the above-mentioned BEP and the Bolsa-Alimentação Program (BAP), both created in 2001 by Provisional Presidential Decree (MP).

Overall, both programs consisted in monthly cash transfers to poor families, contingent upon children's school attendance and medical care for pregnant women and newborns. The municipalities were in charge of identifying and registering families based on quotas defined by technical criteria, while the federal government was responsible for selecting the beneficiaries and operating the payment system through the CEF. The Cadastro Único dos Programas Sociais, a database created to identify and collect information about beneficiary families, operated precariously at first, without integrating the various benefits.

The BEP was under the Education Ministry, whereas the Bolsa-Alimentação Program was under the Health Ministry. In other words, the cash transfer programs were directly linked to ministries headed by two important aspiring PSDB presidential candidates ( ALMEIDA, 2007 ). In the end, then health minister José Serra was nominated.

The MP that established the Bolsa-Alimentação Program was considered automatically approved due to the transition rules stipulated by Constitutional Amendment 32/2001. There was not much disagreement arising from the processing of the MP that created the BEP under the rapporteurship of Oswaldo Coelho (PFL-PE). PT and PDT legislators proposed amendments - all of which were rejected by the rapporteur - to increase the benefit value and restrict the federal government's discretion to define the income criteria used to determine eligibility. Valente (2003) notes that the PT criticized the low value of the transfer. The project was nevertheless approved through symbolic voting.

In 2002, close to the election, the Cartão-Cidadão was created, an electronic card for cash withdrawals associated with various federal government programs. When launching this card, the president declared that:

“The FHC government – as they call it – managed to fight inflation, it created stability. And it seems that this would have been the hallmark of this Government. I don't think that's good. If you want to know what it is that I think we were able to leave as a mark in Brazil is that, for the first time in our history, we reached the poorest, indeed the poorest, and in the furthest corners of Brazil ( CARDOSO, 2002 , p. 01)” ²¹ .

The link between the policies and the electoral calendar did not go unnoticed. Agnelo Queiroz (PCdoB-DF) stated that: “It looks more like an electoral protection network than a social protection network” ( COSTA, 2002 , p. 01).

As the government moved to build a political platform and discourse for the 2002 election, the main opposition party was also formulating its social policy proposal. In October 2001, the PT presents the Fome-Zero Project to fight hunger and poverty, which will be later described.

The connection between social policy and the electoral process is clear. Parties across the political spectrum are searching for political platforms that are electorally appealing - they distribute social policy benefits and make promises to that end. Moreover, it should be stressed that the reach of each policy is conditioned by the strategies used by the parties in their discursive competition. Although Zucco' Jr.s (2013) data suggest that voters rewarded the PSDB for the BEP and the Bolsa-Alimentação Program in 2002, the electoral importance of these policies cannot be overestimated. As shown above, the party was initially reluctant to create a public fund and, once it was created, to compulsorily direct most of its resources to cash transfers.

The Serra government plan in 2002 dedicated just a few sentences to cash transfer programs ²² . Furthermore, a study by Porto, Vasconcellos, and Bastos (2004) shows that the PSDB spent little political advertising time on these policies. Similarly, Almeida says that “in the field of social policies (...), little was said during the electoral competition” ( ALMEIDA, 2004 , p. 13). Thus, my argument here is that the PSDB's strategy in the 2002 election prevented the party from using cash transfer programs to create durable links with voters.

The PT wins the election, and the government emphasizes the social policies platform from the outset. In his inaugural speech, Lula declared that: “Among the priorities of my government, I defined a food security program named 'Fome Zero'" ²³ . This was a comprehensive project, which was divided into structural, specific, and local policies. Among the first were the cash transfer programs, notably the BEP and the Renda Mínima ²⁴ , alongside proposals for agrarian reform, family farming, and job creation. Central to the original proposal was the establishment in the municipalities of several committees designed to assist the poor and empower “civil society” with political training activities. In other words, within the PT government's original social project, cash transfer programs were not prominent nor central. In 2003, a policy named Cartão Alimentação (Food Card) was implemented; however, it was remarkably limited in scope.

Despite receiving public approval, Fome Zero experienced major operational difficulties in its beginning. The good reception would soon turn into criticism ²⁵ . The failure of the government's main program could undermine the government's chances of reelection. President Lula then decided to shift his focus on social policy ( GÓMEZ BRUERA, 2015 ). Based on diagnoses made by technical personnel outside of the traditional party ²⁶ cadres, the president reinforced the importance of cash transfer programs and gradually reduced the role of local committees and other policies under the Fome Zero umbrella. This procedure required the unification and rationalization of cash transfer programs and of the Cadastro Único since the programs in place were experiencing bureaucratic problems ( OLIVEIRA, 2018 ).

In October 2003, the government issued an MP creating the Bolsa-Família Program, initially linking it directly to the president’s office, which is an indication that this program was important in the government's agenda ²⁷ . The examination of the MP by Congress went straight to the plenary, with Deputy Odair (PT-MG) as rapporteur. At the time, the Lula government had minority support in Congress: its coalition involved 48% of the seats in the Chamber of Deputies and 28% in the Senate ( FREITAS, 2016, p , p. 149). The project was nevertheless approved by symbolic voting in both houses.

Some legislators in the opposition highlighted that the BFP was nothing more than the combination of programs that had been created by the former government under the protection of the PSDB-PFL alliance. Senator Arthur Virgílio (PSDB-AM) said that “one cannot, in good conscience, be against Bolsa-Família, which is, in fact, the realization that the Lula Government approach to welfare was at first completely messed up and wrong. Basically, he is now willing to take advantage of what he found, that is, it is the recognition, albeit implicit, that things were better before” ( SENADO FEDERAL, 2003 , p. 42663).

Agripino Maia (PFL-RN) reinforced this view, stating that

the PFL will vote in favor of this provisional presidential decree [MP] instituting Bolsa-Família because it understands that this is a defensible initiative. It is true that this is the combination of several programs, of which many were created previously, but with great social scope, [reaching] the Brazilians whom most need it. For this reason – and an agreement has been reached between the Opposition and the Government – we are going to vote in favor of this MP, which will benefit millions of Brazilians, the poorest, in all corners of the country ( SENADO FEDERAL, 2003 , p. 42663).

However, suggestions for changes in the government’s text were made through 53 amendment proposals, mostly by the opposition. Almost half of these amendments sought to expand the category of beneficiaries (incorporating the elderly, the disabled, adolescents, in addition to increasing the income limit) and/or increasing the benefit values. Six of them determined that the BFP should have the status of a law, thus exempted from budgetary constraints. The speech of Cláudio Cajado (PFL-BA) exemplifies the position taken by the opposition:

We want to know who is in favor of widows and widowers also receiving Bolsa-Família. We want to know if 16, 17, 18-year-old adolescents on the poverty line deserve or not to receive Bolsa-Família. We want to know who wants to guarantee the same benefit to equal citizens. As it is in this project, some will have it and others who are in the same level of need and poverty will not. ( CÂMARA DOS DEPUTADOS, 2003 , p. 261).

Rodrigo Maia (PFL-RJ), leader of the party, reinforces this position:

The amendment (...) seeks to increase the number of people who will be entitled to the benefit. In the previous government, that of Fernando Henrique Cardoso, the [benefit] value per capita was 60 reais. Astonishingly, honorable legislators, the PT is now proposing 50 reais, less than what was proposed by the same Government that the PT accused of being neoliberal and conservative, of not taking care of social interests. To be eligible for the benefits of the Bolsa Escola and Cartão Alimentação, the previous government stipulated a per capita income of 120 reais. Once again, the PT reduced the benefit, which now has reached 100 reais, restricting the number of families covered ( CÂMARA DOS DEPUTADOS, 2003 , p. 268).

Moroni Torgan (PFL-CE) emphasizes: “We are not against this project, although we stress that it should not be purely and simply ‘assistencialista’ in nature. It is not just social benefits that the people need, but mainly training and professional opportunities. This is the only way to avoid political dependence, an evil that existed in the time of the famous ‘coronéis da política’ (political patrons)” ( CÂMARA DOS DEPUTADOS, 2003 , p. 246).

This bet on targeted conditional cash programs was met with resistance from sections of the PT and the left ²⁸ . Almeida notes that this emphasis on cash transfer policies “points to a conception of social protection and a style of social policymaking that are distant from the diffuse expectations about the PT's reformative action” ( ALMEIDA, 2004 , p. 08).

The political-electoral benefits of the social program were therefore neither automatic nor guaranteed. Hunter and Power (2005 , p. 132) were skeptical about the success of the BFP after the first two years of government. Lula has bet on this public policy, even with the program being the target of ideological and political criticism. However, unlike the PSDB government, Lula sought to build this platform when he first entered office, long before the election. Next, I address the consequences of these strategies in terms of political behavior.

The impact of public policies on voters' behavior

In 2005, the BFP was expanding rapidly, either by relocating the beneficiaries of previous programs or incorporating new families. One year before Lula's reelection, however, a significant number of families were still receiving transfers from programs created by the FHC government, notably the BEP. Moreover, this was the year that the Lula government faced its worst political crisis. The scandal known as “mensalão” dominated the news and congressional activities. Allegations of vote-buying involving parties and legislators resulted in important PT cadres being dismissed and removed from office.

A national opinion survey with 2,537 respondents conducted by Datafolha in October 2005 (CESOP, 2019a) - when the president reached his worst approval rating - allows us to verify whether the political positions of BEP and BFP beneficiaries were different from one another, and to compare beneficiaries with non-beneficiaries in relation to three aspects: evaluation of the Lula government, party identification, and voting intention in the first round of the 2006 election.

Throughout the text, I presented three hypotheses concerning the relationship between programmatic policies and political behavior. Traditional theories of distributive policy would say that, given the government's lack of control over the distribution of the benefits from these programs, the political opinion of beneficiaries should not differ from that of non-beneficiaries. Arguments based on short-term retrospective voting would say that the opinion of beneficiaries differs from non-beneficiaries, but that the effect among beneficiaries of both the PBF and PBE would be similar since both groups were receiving government benefits. The alternative hypothesis proposed in this article is that beneficiaries not only differ from non-beneficiaries but also between the two programs because Lula has mobilized around the BFP, unlike the PSDB.

Table 01
Distribution of dependent variables. Datafolha Survey, October 2005*

Approval Ratings of the Lula Government (% of valid responses)
Very Good/Good	28,54
Regular	42,94
Bad/Very bad	28,52
Party Identification (% of valid responses)
PT	17,04
Others	26,13
No identification	56,83
Voting intentions 2006 (% of valid responses)
Lula	43,49
Alckmin	21,37
Others	35,14

Cesop (2019a
• *Note: Shows the distribution of the three dependent variables considered. In all analyses, people who did not respond or were unable to respond were excluded.

In October 2005, the population was sharply divided when it comes to the government’s approval ratings. While most of the population did not express party identification, 17% of voters reported having a link with the PT, and 26% with other parties. Finally, as a projection of the 2006 election, Lula had more than 43% of the voting intentions, Geraldo Alckmin (PSDB) had 21%, and the others combined reached 35% ²⁹ .

The survey included questions about the BFP, BEP, and other existing cash transfers, such as Bolsa-Alimentação, Auxílio-Gás, Fome Zero. The questions were: ‘Are you, or someone in your family, or someone close to you participating in the program [Bolsa-Família, Bolsa-Escola, etc.]?’ With multiple responses allowed. Since the vast majority of the responses mentioned the BFP and BEP, I focused the analysis on these two programs ³⁰ .

As some studies have noted when addressing other opinion surveys ( NICOLAU, 2014 ; SIMONI JR., 2021 ), this wording may lead to measurement problems since the social benefit is granted to the family living in the same household, not the person. In addition, ‘someone in the family’ could encompass individuals beyond the nuclear family. As shown in Table 02 , when comparing to the official data, this problem, in fact, led to the underrepresentation of beneficiaries when only ‘the interviewee’ option is considered and overrepresentation when the ‘someone from the family’ option is added. In the regressions presented next, I used this last measure.

Table 02
Distribution of independent variables of interest, and comparison with official data. Datafolha Survey and MDS Official Figures - 2005

Social programmes	% of beneficiaries
Beneficiary of Bolsa-Família (interviewee)	5,03
Beneficiary of Bolsa-Escola (interviewee)	3,25
Beneficiary of Bolsa-Família (interviewee and family)	20,51
Beneficiary of Bolsa-Escola (interviewee and family)	21,01
Official figures of families receiving BFP benefits	15,66
Official figures of families receiving BEP benefits	4,19

Cesop (2019a)

It is worth noting the discrepancy between the percentage of people claiming to be a BEP beneficiary in this survey and the official data. This disparity is probably associated with the fact that some people who migrated to the BFP still say they receive the former benefit. In addition, about half of all beneficiaries of one of these programs say that they receive benefits from the other program, something not allowed in both program designs. There is thus a bias in the data structure acting against my hypothesis since the beneficiaries of the two programs are not adequately differentiated, which tends to reduce possible heterogeneous effects.

To analyze the relationship between being a beneficiary of these policies and the government’s approval rating, party identification, and voting, I used multinomial logistic models, with the following reference categories: no approval (bad and very bad), no party identification, and vote for Alckmin, respectively. In the three models, the control variables - factors that I consider to be related to being a cash transfer beneficiary and to voting - are education, income, age, region, and nature of the municipality.

For better visualization, only the estimated coefficients and the average marginal effects (AME) of the variables of interest are presented here; the other variables are kept as observed ³¹ . Full results are included in the ^Appendix .

Graph 01
Impact of being a Beneficiary on Political Opinions in 2005 – 95% CI – Multinomial logistic

The results contradict the hypothesis in traditional distributive theories and the retrospective voting theory, and they support my argument, which emphasizes the importance of mobilizing around public policy. As shown in Table 03 , being a BFP beneficiary increases the probability of approving the Lula government (from 27% to 35%), identifying with the PT (from 16% to 22%), and expressing the intention to vote for Lula (from 41% to 51%) ³² . On the other hand, being a BEP beneficiary had no impact on any of these aspects ³³ . The coefficient comparison calculation proposed by Gelman and Stern (2006) was applied to confirm that the difference between the BFP and BEP coefficients is statistically significant ³⁴ .

Table 03
Average marginal effect (AME) - 2005

	Change	From	To	p-value
Approves the Lula Government
Bolsa-Família	0,08	0,271	0,351	0,004
Bolsa-Escola	-0,008	0,29	0,282	0,749
Identification with the PT
Bolsa-Família	0,056	0,161	0,217	0,02
Bolsa-Escola	-0,018	0,176	0,158	0,368
Intention to vote for Lula
Bolsa-Família	0,093	0,416	0,509	0,005
Bolsa-Escola	0.011	0.433	0.444	0.724

Cesop (2019a)

In other words, programmatic policies do matter, but to just receive a benefit granted by the current government it is not enough. Policies created during the PSDB government did not have a long-term impact, while the BFP has had a significant effect, indicating that mobilization efforts by parties around social programs produced political-electoral outcomes.

Lula was reelected in 2006, and his preferred successor, Rousseff, was elected in 2010 and 2014. Over these years, the BFP further expanded – with an increased number of beneficiaries and resources transferred - and became an important social policy ( PAIVA et al., 2013 ).

Since the 2014 election, the Brazilian political system has confronted a serious crisis, which culminated in the impeachment of Dilma Rousseff in 2016. Micher Temer, from the PMDB, takes power and forms a broad coalition with the PSDB, DEM (former PFL), and several other parties. After nearly 15 years in power, the PT goes back to being an opposition party.

The relationship between the PMDB-led government and the BFP was dubious. Some of its members had already supported reducing spending on this social program ³⁵ . The Minister of Social and Agrarian Development, Osmar Terra (PMDB), criticized gaps in coverage and inclusion errors in the BFP ³⁶ . On the other hand, the fact that the president announced adjustments in the BFP benefits a few months after taking office and again in May 2018 - close to the beginning of the presidential campaign - reveals the importance of this social policy in the Brazilian political game.

The 2018 election was held in a very different legal and political context from previous presidential elections. Temer's government had reached its lowest approval rating and failed to secure a viable successor candidate. Lula was initially a formal candidate in the electoral process, but he was later declared ineligible by the Electoral Court and replaced by Fernando Haddad. The PSDB launched Alckmin as a candidate again, supported by the largest coalition of parties. Several other parties also presented their candidates, but, from the start, the election was polarized between the candidacy of the PT and that of Jair Bolsonaro (PSL), a ‘baixo clero’ legislator with a far-right agenda who was elected in the second round against Haddad .

With the PT out of government, one has the first opportunity to verify whether being the incumbent correlates with the party receiving electoral rewards from the distribution of BFP benefits, or whether this reward stems from reciprocity from beneficiary voters due to mobilization efforts by the party around this social policy.

For that purpose, I analyzed the Estudo Eleitoral Brasileiro (ESEB - Brazilian Electoral Study), a survey with 2,506 people that was conducted right after the second round of the 2018 presidential election. In addition to the phenomena that I examined in the 2005 Datafolha survey, with the ESEB survey I am able to analyze the impact of the BFP on anti-partisan sentiment in general and anti-PT sentiment in particular. This issue has been addressed in several studies in recent years ( BORGES and VIDIGAL, 2018 ; PAIVA, KRAUSE and LAMEIRÃO, 2016 ; RIBEIRO, CARREIRÃO and BORBA, 2016 ; SAMUELS and ZUCCO JR., 2018 ) and is considered to have played a decisive role in the 2018 election. The distribution of dependent variables is shown in Table 04:

Table 04
Distribution of dependent variable – ESEB Survey, 2018

Approval ratings of the Temer Government (% of valid responses)
Very good/Good	11,24
Regular	12,69
Bad/Very bad	76,07
Party Identification (% of valid responses)
PT	11,93
Others	13,06
No identification	75,01
Antipartisanship (% of valid responses)
PT	29,53
Others	15,65
Non-antipartisan	54,83
Declared vote in the first round of the 2018 election (% of valid responses)
Haddad	31,42
Bolsonaro	47,53
Others	21,04

Cesop (2019b)

In addition to Temer's extremely low approval rating, I stress a significant drop in the level of party identification in comparison to 2005. About half of the respondents expressed anti-partisan sentiment, most of which turned into anti-PT sentiment. On the other hand, among people who reported to have cast a valid ballot, Haddad got 31%. With respect to the main independent variable, about 31% of respondents gave a positive answer to the question ‘In the last three years, were you or someone who lives in your house a beneficiary (received) of the Bolsa-Família program?’ - numbers that are close to the official data.

The same models and controls used in the 2005 survey (except for ‘nature of the municipality’, unavailable in the ESEB) were applied to the 2018 survey. The reference categories are ‘disapproves of Temer's government’, ‘no party identification’, ‘rejects no party’, and ‘votes for Bolsonaro’ ³⁷ . The results of interest are shown in Graph 02 (all results are available in the ^Appendix ).

Graph 02
Impact of being a beneficiary on political opinions in 2018 – 95% CI – Multinomial logistic

The PT was still receiving electoral rewards from the BFP, despite being in the opposition. To be a beneficiary of this social policy strongly increased the probability of voting for Haddad (from 28% to 40%) and decreased anti-PT sentiment (from 32% to 24%). However, the program led to a slight increase in positive assessments of the Temer government and had no effect on the formation of a positive identification with the PT. The heterogeneous effects on party identification should be further examined in future studies.

Table 05
Average marginal effect (AME) - 2018

	Change	From	To	p-value
Approves the Temer Government
Bolsa-Família	0,035	0,104	0,139	0,08
Identification with the PT
Bolsa-Família	0,016	0,118	0,134	0,398
Rejection of the PT
Bolsa-Família	-0,074	0,317	0,243	0,00
Vote for Haddad
Bolsa-Família	0,122	0,279	0,4	0,00

Cesop (2019b

In short, even if the PT had no legislative nor budgetary control over the BFP, the party continued to benefit electorally from it. The results indicate that links were created between voters and the formulator party, contrary to what is expected by distributive theories and hypotheses associated with the short-term retrospective voting argument ³⁸ .

Conclusion

Traditional models of distributive politics and electoral competition cannot explain electoral rewards from programmatic policies. According to established theories, voters evaluate the parties according to their ideological position and take into account only the distribution of policy benefits in which the parties exercise discretion. However, several analyses on electoral returns to cash transfer programs in Latin America - of which many have programmatic features - contradict these expectations. How should we address this discrepancy?

Based on the Brazilian case, I offered a potential explanation. I analyzed parliamentary debates on the formulation of cash transfer programs and pointed out that the PSDB, during the FHC government, was reluctant when defending this type of policy and the public fund to finance it. PT legislators, on the other hand, presented proposals to strengthen these programs from the outset. Once in office, Lula invested heavily in the BFP at the expense of the more ambitious ‘Fome Zero’ project, even though such a move would draw criticism from his allies and create uncertainty about the policy's effectiveness. The analysis of the opinion surveys shows that the party benefited from its social policy, even in the early stages when there were still beneficiaries of the BEP, created by the PSDB, and in 2018, when the party was already out of government and therefore without legislative and budgetary control over the BFP.

My theoretical proposition is that the electoral return to programmatic policies is associated with mobilization efforts and credit-claiming by parties within the parameters of political competition. As political platforms are built, disputed, and mobilized, voters may support those parties formulating programmatic policies because of reciprocity mechanisms, even when they know that the distribution of benefits is not under the control of the party nor dependent upon its permanence in power.

I do not argue that this is a permanent effect of this social policy, nor that one can be certain of its electoral return. Rather, I emphasize that political competition and mobilization efforts are important factors continuously shaping the electoral effects of public policies. Furthermore, a distinction usually unnoticed in the literature is that cash transfer programs in Latin America - despite being programmatic policies - are not entitlements and, consequently, their budget must be continually approved to sustain their operation, as De La O points out (2015). As a result, the formulator parties, as well as their opponents who take power, are allowed to alter the policy design, creating opportunities for electoral rewards.

The BFP will most likely continue to have significant electoral weight in Brazil. President Bolsonaro has made several statements criticizing the program throughout his political trajectory, in addition to being the only legislator to vote against the creation of the ‘Fundo de Combate à Pobreza’. During the presidential campaign and after being elected to the presidential office, he announced that he would not only maintain the BFP but that he intended to expand it, albeit under a different name. By also considering the strategy adopted by other political actors, voters will revise their preferences and express through the ballot box which commitments they believe to be the most credible.

References

• ALMEIDA, Alberto Carlos ( 2007 ), Por que Lula?: o contexto e as estratégias políticas que explicam a eleição e a crise. Rio de Janeiro: Editora Record. 336 pp..
• ALMEIDA, Maria Hermínia Tavares de ( 2004 ), A política social no governo Lula. Novos Estudos Cebrap. Nº 70, pp. 07-17.
• AMES, Barry; HUBERTS, Alyssa; MACHADO, Fabiana; RENNO, Lucio R.; SAMUELS, David; SMITH, Amy Erica, and ZUCCO, Cesar (2016), Brazilian electoral panel study: 2014 results. Technical Notes IDB-TN915 . Washington: Inter- American Development Bank. 45 pp..
• ARAÚJO, Victor e FLORES, Paulo ( 2019 ), Competição eleitoral e ideologia partidária: PT, PSDB e a agenda de proteção social no Brasil (1991-2014). In: As políticas da política: desigualdades e inclusão nos governos PSDB e PT. Edited by ARRETCHE, Marta T.; MARQUES, Eduardo, and FARIA, Carlos Aurélio Pimenta de. São Paulo: Editora Unesp. Pp. 01-22.
• BICHIR, Renata Mirandola ( 2010 ), O Bolsa-Família na berlinda? Os desafios atuais dos programas de transferência de renda. Novos Estudos Cebrap . Nº 87, pp. 115-129.
• BORGES, André and VIDIGAL, Robert ( 2018 ), Do lulismo ao antipetismo? Polarização, partidarismo e voto nas eleições presidenciais brasileiras. Opinião Pública . Vol. 24, Nº 01, pp.53-89.
• CÂMARA DOS DEPUTADOS ( 2003 ), Diários da Câmara dos Deputados de 18/12/2003 . Ano LVIII, Nº 217. Biênio 2003/2004. Brasília. 518 pp..
• CÂMARA DOS DEPUTADOS ( 2000 ), Diários da Câmara dos Deputados de 06/12/2000 . Ano LV, Nº 206, Biênio 1999/2000. Brasília. 416 pp..
• CANÊDO-PINHEIRO, Maurício (2015), Bolsa família ou desempenho da economia? Determinantes da reeleição de Lula em 2006. Economia Aplicada . Vol. 19, Nº 01, pp. 31-57.
• CARDOSO, Fernando Henrique (2002), Discurso do Presidente da República, Fernando Henrique Cardoso, na cerimônia de lançamento do ‘Cartão-Cidadão’, Rio de Janeiro. Available at ˂ https://www.psdb.org.br/acopanhe/noticias/ discurso-do-presidente-da-republica-fernando-henrique-cardoso-na-cerimonia-de-lancamento-do-%C%A5cartao-cidadao%C2%A5-rio-de-janeirorj ˃ . Accessed on October, 02, 2021 .
  » https://www.psdb.org.br/acopanhe/noticias/
• CARRARO, André; SHIKIDA, Claudio Djissey; MONASTERIO, Leonardo Monteiro; ARAUJO JR., Ari Francisco, and DAMÉ, Otávio Menezes ( 2009 ), It is the economy, companheiro! An empirical analysis of Lula's re-election based on municipal data. Economics Bulletin . Vol. 29, Nº 02, pp. 976-991.
• CARDOSO, Fernando Henrique ( 2008 ), Avança Brasil: proposta de governo. Rio de Janeiro: Centro Edelstein de Pesquisas Sociais. 179 pp..
• CESOP (2019a), Avaliação do presidente Lula/ Referendo armas de fogo/ Intenção de voto presidente/ Corrupção no governo Lula/ Lula e o PT. In: Banco de Dados do Centro de Estudos de Opinião Pública – CESOP-UNICAMP. Available at ˂ https://www.cesop.unicamp.br/por/bancodedados/v/2268 ˃ . Accessed on October, 02, 2021 .
  » https://www.cesop.unicamp.br/por/bancodedados/v/2268
• CESOP (2019b) ESEB 2018 – ESTUDO ELEITORAL BRASILEIRO. In: Banco de Dados do Centro de Estudos de Opinião Pública – CESOP-UNICAMP. Available at ˂ https://www.cesop.unicamp.br/por/bancodedados/v/4538 ˃ . Accessed on October, 02, 2021 .
  » https://www.cesop.unicamp.br/por/bancodedados/v/4538
• COELHO, Denílson Bandeira ( 2008 ), A difusão do programa bolsa escola: competição política e inovação no setor social. Paper presented at the 32º Encontro Anual da ANPOCS. Caxambu-MG.
• CORRÊA, Diego (2015), Os custos eleitorais do Bolsa Família: reavaliando seu impacto sobre a eleição presidencial de 2006. Opinião Pública . Vol. 21, Nº 03, pp. 514-534.
• CORRÊA, Diego and CHEIBUB, José Antonio ( 2016 ), The anti-incumbent effects of conditional cash transfer programs. Latin American Politics and Society . Vol. 58, Nº 01, pp. 49-71.
• COSTA, Raymundo (2002), FHC amplia ação social a 4 meses da eleição. Folha de São Paulo. 26 de maio de 2002. Available at ˂ https://www1.folha.uol.com.br/fsp/brasil/fc2605200212.htm ˃ . Accessed on 02, October, 2021 .
  » https://www1.folha.uol.com.br/fsp/brasil/fc2605200212.htm
• COX, Gary W. and McCUBBINS, Mathew D. ( 1986 ), Electoral politics as a redistributive game. Journal of Politics . Vol. 48, Nº 02, pp. 370–389.
• DE JANVRY, Alain; FINAN, Frederico, and SADOULET, Elisabeth ( 2012 ), Local electoral incentives and decentralized program performance. The Review of Economics and Statistics . Vol. 94, Nº 03, pp. 672-685.
• DE JANVRY, Alain; FINAN, Frederico; SADOULET, Elisabeth; NELSON, Donald; LINDERT, Kathy; DE LA BRIERE, Benedict, and LANJOUW, Peter ( 2005 ), Brazil’s Bolsa Escola Program: the role of local governance in decentralized implementation. Social Protection Discussion Paper . Washington: World Bank.
• DE LA O, Ana Lorena ( 2015 ), Crafting policies to end poverty in Latin America: the quiet transformation. Cambridge: Cambridge University Press. 196 pp..
• DE LA O, Ana Lorena ( 2013 ), Do conditional cash transfers affect electoral behavior? Evidence from a randomized experiment in Mexico. American Journal of Political Science . Vol. 57, Nº 01, pp. 01-14.
• DIAZ-CAYEROS, Alberto; ESTÉVEZ, Federico, and MAGALONI, Beatriz ( 2016 ), The political logic of poverty relief: electoral strategies and social policy in Mexico . New York: Cambridge University Press. 257 pp..
• DIXIT, Avinash and LONDREGAN, John ( 1996 ), The determinants of success of special interest in redistributive politics. Journal of Politics . Vol. 58, Nº 04, pp. 1132-1155.
• DOWNS, Anthony ( 1957 ), An economic theory of political action in a democracy. The Journal of Political Economy . Vol. 65, Nº 02, pp. 135-150.
• DRAIBE, Sonia Miriam ( 2003 ), A política social no período FHC e o sistema de proteção social. Tempo Social . Vol. 15, Nº 02, pp. 63-101.
• FINAN, Frederico and SCHECHTER, Laura ( 2012 ), Vote-buying and reciprocity. Econometrica . Vol. 80, Nº 02, pp. 863–881.
• FIORINA, Morris P. ( 1981 ), Retrospective voting in American National Elections . New Haven: Yale University Press. 249 pp..
• FREITAS, Andréa ( 2016 ), O presidencialismo da coalizão . Rio de Janeiro: Fundação Konrad-Adenauer. 149 pp..
• FRIED, Brian J. ( 2009 ), Distributive politics and conditional cash transfers: the case of Brazil’s Bolsa Família. World Development . Vol. 40, Nº 05, pp. 1042-1053.
• GELMAN, Andrew and STERN, Hal ( 2006 ), The difference between ‘significant’ and ‘not significant’ is not itself statistically significant. The American Statistician . Vol. 60, Nº 04, pp. 328-331.
• GOLDEN, Miriam and MIN, Brian ( 2013 ), Distributive politics around the world. Annual Review of Political Science . Vol. 16, Nº 01, pp. 73-99.
• GÓMEZ BRUERA, Hernán F. ( 2015 ), Participation under Lula: between electoral politics and governability. Latin America Politics and Society . Vol. 57, Nº 02, pp. 01-20.
• HUNTER, Wendy and POWER, Timothy J. (2007), Rewarding Lula: executive power, social policy and the Brazilian elections of 2006. Latin American politics and society . Vol. 49, Nº 01, pp. 01-30.
• HUNTER, Wendy and POWER, Timothy J. ( 2005 ), Lula’s Brazil at midterm. Journal of Democracy . Vol. 16, Nº 03, pp. 127-139.
• IMAI, Kosuke; KING, Gary and RIVERA, Carlos Velasco ( 2020 ), Do nonpartisan programmatic policies have partisan electoral effects? Evidence from two large scale experiments. Journal of Politics. Vol. 82, Nº 02, pp. 714-730.
• LAYTON, Matthew L. and SMITH, Amy Erica ( 2015 ), Incorporating marginal citizens and voters: the conditional electoral effects of targeted social assistance in Latin America. Comparative Political Studies . Vol. 48, Nº 07, pp. 854-881.
• LIMONGI, Fernando and GUARNIERI, Fernando ( 2015 ), Competição partidária e voto nas eleições presidenciais no Brasil. Opinião pública . Vol. 21, Nº 01, pp. 60-86.
• LINDBECK, Assar and WEIBULL, Jörgen W. ( 1987 ), Balanced-budget redistribution as the outcome of political competition. Public Choice . Nº 52, pp. 273-297.
• LINOS, Elizabeth ( 2013 ), Do conditional cash transfer programs shift votes? Evidence from the Honduran PRAF. Electoral Studies . Vol. 32 , Nº 04 , pp. 864 - 874 .
• MAGALHÃES, André Matos; SILVA, Marcelo Eduardo Alves, and DIAS, Fernando de Mendonça (2015), Eleição de Dilma ou segunda reeleição de Lula? Uma análise espacial do pleito de 2010. Opinião Pública . Vol. 21 , Nº 03 , pp. 535 - 573 .
• MAGALHÃES, Edgar and VEIGA, Laura da ( 2014 ), O Ministério do Desenvolvimento Social e Combate à Fome: uma nova construção institucional. In: Avalição de Políticas Públicas. Edited by MADEIRA, Lígia Mori. Porto Alegre: UFRGS/CEGOV. Pp. 144 - 178 .
• MANACORDA, Marco; MIGUEL, Edward, and VIGORITO, Andrea ( 2011 ), Government transfers and political support. American Economic Journal: Applied Economics . Vol. 03 , Nº 03 , pp. 01 – 28 .
• MARTINS, Diogo José Dalpoz; MANSANO, Fernanda Helen; PARRÉ, José Luiz, and PLASSA, Wander ( 2016 ), Fatores que contribuíram para a reeleição da presidente Dilma Rousseff. Política e Sociedade . Vol. 15 , Nº 32 , pp. 145 - 170 .
• MARZAGÃO, Thiago ( 2013 ), A dimensão geográfica das eleições brasileiras. Opinião Pública . Vol. 19 , Nº 02 , pp. 270 - 290 .
• MAYHEW, David R. ( 1974 ), Congress: the electoral connection . New Haven: Yale University Press. 216 pp..
• MELO, Marcus André ( 2005 ), O sucesso inesperado das reformas de segunda geração: federalismo, reformas constitucionais e política social. Dados . Vol. 48 , Nº 04 , pp. 845 - 889 .
• MOREIRA, Davi Cordeiro ( 2020 ), Não tomarás seu nome em vão. O Programa Bolsa Família em discursos do PT e do PSDB. In: A nova ciência da política . Edited by ARANTES, Rogério B. and PETRONI, Lucas C.. São Paulo: FFLCH/USP. pp. 125 - 154 .
• NICOLAU, Jairo ( 2014 ), Determinantes do voto no primeiro turno das eleições presidenciais brasileiras de 2010: uma análise exploratória. Opinião Pública . Vol. 20 , Nº 03, pp. 311-325 .
• NICOLAU, Jairo and PEIXOTO, Vitor (2007), Uma disputa em três tempos: uma análise das bases municipais das eleições presidenciais de 2006. Anais do XXXI Encontro da ANPOCS . Vol. 31 , pp. 22 - 26 .
• NUPIA, Oskar ( 2011 ), Anti-poverty programs and presidential election outcomes: familias en acción in Colombia. Paper CEDE Nº 2011-14. Bogotá : Universidad de los Andes .
• OLIVEIRA, Maria Clara ( 2018 ), Ideias e políticas públicas: considerações a partir da análise de programas de transferência monetária na África do Sul, no Brasil e no Chile. Doctoral thesis . Faculdade de Filosofia, Letras e Ciências Humanas. Departamento de Ciência Política. USP.
• PAIVA, Luis Henrique; FALCÃO, Tiago, and BARTHOLO, Letícia ( 2013 ), Do Bolsa-Família ao Brasil sem miséria: um resumo do percurso brasileiro recente na busca da superação da pobreza extrema. In: Programa Bolsa Família: uma década de inclusão e cidadania. Edited by CAMPELLO, Tereza and NERI, Marcelo Côrtez. Brasília : IPEA . Pp. 25 - 46 .
• PAIVA, Denise; KRAUSE, Silvana, and LAMEIRÃO, Adriana Paz ( 2016 ), O eleitor antipetista: partidarismo e avaliação retrospectiva. Opinião Pública . Vol. 22 , Nº 03 , pp. 638 - 674 .
• PEREIRA, Ana Elisa Gonçalves; SHIKIDA, Cláudio Djissey; RIBEIRO, Felipe Garcia, and NAKABASHI, Luciano (2015), A eleição de Dilma em 2010 e seus determinantes: evidências empíricas do Programa Bolsa Família. Análise Econômica . Vol. 33 , Nº 64 , pp. 111 - 142 .
• PORTO, Mauro; VASCONCELLOS, Rodrigo, and BASTOS, Bruna (2004), A televisão e o primeiro turno das eleições presidenciais de 2002: análise do Jornal Nacional e do Horário Eleitoral. In: Eleições presidenciais 2002 no Brasil . Edited by RUBIM, Antonio Albino Canelas. São Paulo : Hacker Editores . pp. 68 - 90 .
• RIBEIRO, Ednaldo; CARREIRÃO, Yan, and BORBA, Julian ( 2016 ), Sentimentos partidários e antipetismo: condicionantes e covariantes. Opinião Pública . Vol. 22 , Nº 03 , pp. 603 - 637 .
• ROCHA, Sonia ( 2011 ), O programa Bolsa Família: evolução e efeitos sobre pobreza. Economia e Sociedade . Vol. 20 , Nº 01 , pp. 113 - 139 .
• SAMUELS, David J. and ZUCCO JR., César ( 2018 ), Partisans, antipartisans, and nonpartisans: Voting behavior in Brazil . Cambridge : Cambridge University Press . 196 pp..
• SENADO FEDERAL (2003), Diários do Senado Federal de 12/2003. Available at ˂ https://legis.senado,leg.br/diarios/ver/1277?sequencia=377 ˃ . Accessed October, 02, 2021 .
  » https://legis.senado,leg.br/diarios/ver/1277?sequencia=377
• SENADO FEDERAL (2000), Diários do Senado Federal de 15/2/2000. Available at ˂ https://legis.senado,leg.br/diarios/ver/6622?sequencia=43 ˃ . Accessed October, 02, 2021 .
  » https://legis.senado,leg.br/diarios/ver/6622?sequencia=43
• SILVA, Maria Ozanira da Silva e; DI GIOVANNI, Geraldo, and YAZBEK, Maria Carmelita ( 2012 ), A política social brasileira no século XXI: a prevalência dos programas de transferência de renda. São Paulo : Ed. Cortez . 248 pp..
• SIMONI JR., Sergio ( 2021 ), Efeitos diretos e indiretos do Programa Bolsa Família nas eleições presidenciais brasileiras. Opinião Pública . Vol. 27 , Nº 01 , pp. 230 - 260 .
• SIMONI JR., Sergio ( 2017 ), Política distributiva e competição presidencial no Brasil: Programa Bolsa-Família e a tese do realinhamento eleitoral. Doctoral thesis . Faculdade de Filosofia, Letras e Ciências Humanas. Departamento de Ciência Política. USP.
• SIMONI JR., Sergio; GUICHENEY, Hellen and OLIVEIRA, João Lucas Sacchi ( 2021 ), O Congresso e a Renda Emergencial. E-LEGIS. Vol. 14 , Edição Especial, pp. 164 - 181 .
• SOARES, Sergei and SÁTYRO, Natália ( 2009 ), O Programa Bolsa Família: desenho institucional, impactos e possibilidades futuras. Paper . Texto para Discussão 1424, IPEA.
• SOARES, Gláucio Ary Dillon and TERRON, Sonia Luiza ( 2008 ), Dois Lulas: a geografia eleitoral da reeleição (explorando conceitos, métodos e técnicas de análise geoespacial). Opinião Pública . Vol. 14 , Nº 02 , pp. 269 - 301 .
• STOKES, Susan C. ( 2005 ), Perverse accountability: a formal model of machine politics with evidence from Argentina. American Political Science Review . Vol. 99 , Nº 03 , pp. 315 - 325 .
• STOKES, Susan C.; DUNNING, Thad; NAZARENO, Marcelo, and BRUSCO, Valeria ( 2013 ), Brokers, voters and clientelism: the puzzle of distributive politics. Cambridge : Cambridge University Press . 344 pp..
• SUGIYAMA, Natasha ( 2011 ), O Bolsa Família: visões desde cima e de baixo. In: Congresso por ele mesmo: autopercepções da classe política brasileira. Edited by POWER, Timothy and ZUCCO JR., Cesar. Belo Horizonte : Ed. UFMG . Pp. 273 - 304 .
• THOMÉ, Débora ( 2013 ), O Bolsa-Família e a social-democracia . Rio de Janeiro : Ed Fundação Getúlio Vargas . 160 pp..
• TRIBUNAL DE CONTAS DA UNIÃO ( 2006 ), Relatório de acompanhamento do Programa Bolsa Família. Available at ˂ https://docplayer.com.br/18341097-Relatorio-de-acompanhamento-do-programa-bolsa-familia.html ˃. Accessed on 02, October, 2021 .
  » https://docplayer.com.br/18341097-Relatorio-de-acompanhamento-do-programa-bolsa-familia.html
• VALENTE, Ana Lúcia ( 2003 ), O Programa Nacional de Bolsa Escola e as ações afirmativas no campo educacional. Revista Brasileira de Educação . Nº 24, pp. 165-182.
• WEINSCHENK, Aaron C. ( 2010 ), Revisiting the political theory of party identification. Political Behavior . Vol . 32 , pp. 473–494.
• WINTERS, Matthew (2010), Exploring changes over time in the targeting of a cash transfer program in Ecuador. Paper presented at the 2009 and 2020 Midwest Political Science Association Annual Meetings.
• ZUCCO JR., Cesar ( 2015 ), The impacts of conditional cash transfers in four presidential elections (2002-2014). Brazilian Political Science Review . Vol. 09, Nº 01, pp. 135-149.
• ZUCCO JR., Cesar ( 2013 ), When payouts pay off: conditional cash-transfer and voting behavior in Brazil 2002-2010. American Journal of Political Science . Vol. 57, Nº 04, pp. 810-822.
• ZUCCO JR., Cesar (2008), The president's ‘new’ constituency: Lula and the pragmatic vote in Brazil's 2006 presidential election. Journal of Latin American Studies . Vol. 40, Nº 01, pp. 29-49.

• *
  This article benefited from the discussions held at the 43 Annual Meeting of Anpocs and the 1 Political Studies Conference at Universidade Federal do Rio Grande do Sul. I appreciate the comments and suggestions of the anonymous BPSR reviewers and those of Felipe Zorzi, Gabriel Madeira, Glauco Peres, Gustavo Grohmann, Jairo Pimentel, Leonardo Barone, Lilian Sendretti, Maria Clara Oliveira, Maurício Moya, Natália Bueno, Paulo Peres, Pedro Vicente de Castro, and Vitor Vasquez
• 1
  Available at ˂ https://memoria.ebc.com.br/noticias/brasil/2013/05/boato-sobre-fim-do-bolsa-familia-levou-beneficiarios-a-sacar-r-152-milhoes˃ .
• 2
  Available at ˂ https://memoria.ebc.com.br/noticias/politica/2013/05/boato-sobre-o-fim-do-bolsa-familia-e-criminoso-diz-presidenta-dilma˃ .
• 3
  Available at ˂ https://memoria.ebc.com.br/agenciabrasil/noticia/2013-05-27/psdb-pede-procuradoria-geral-da-republica-para-investigar-caixa-sobre-boatos-do-bolsa-familia˃ .
• 4
  Available at ˂ https://agenciabrasil.ebc.com.br/politica/noticia/2018-04/temer-anuncia-reajuste-do-bolsa-familia-aumento-sera-de-567˃ .
• 5
  Available at ˂ https://www.correiobraziliense.com.br/app/noticia/politica/ 2019/04/11/interna_politica,748643/bolsonaro-e-o-bolsa-familia-de-critico-feroz-a-defensor.shtml˃.
• 6
  Available at ˂ https://agenciabrasil.ebc.com.br/politica/noticia/2019-04/bolsonaro-cria-13o-salario-para-o-bolsa-familia˃ .
• 7
  Available at ˂ https://www.bbc.com/portuguese/brasil-54334283˃ .
• 8
  Initially, the program was a combination of programs that had been previously formulated by the PSDB. This is an important point, to which we will return later.
• 9
  For more nuanced arguments, see Carraro et al. (2009) ; Corrêa, (2015) , Limongi and Guarnieri (2015) ; Simoni Jr. (2017 ).
• 10
  According to this survey, around 70% of respondents thought that the BFP would not be terminated if the PT lost the 2014 presidential election.
• 11
  The implementation of the BEP at the local level was less structured than that of the BFP. Nevertheless, there are indications that the municipalities had some discretion over the program; the federal government, however, did not ( DE JANVRY et al., 2012 , 2005 ).
• 12
  The original model was formulated for two-party systems and was used here for simplification.
• 13
  Some of these studies show that cash transfer programs end up having negative electoral effects on some social groups.
• 14
  These apparent mistakes refer to the processing of geographically located data and the specification of the models applied.
• 15
  Sugiyama (2011) shows that legislators saw the program formulated by the PT as different from the cash transfer program formulated by the PSDB. By analyzing public opinion surveys, Samuels and Zucco Jr. (2018) found that support for the BFP was highly polarized between voters identified with PT and ‘antipetistas’ (anti-PT individuals).
• 16
  In Simoni Jr. (2021) , I demonstrate that social policies around which parties mobilize politically also have an impact on the voting behavior of non-beneficiaries.
• 17
  This point was also raised preliminarily by De La O (2013) .
• 18
  In fact, there was already a precedent to the national discussions on the matter, the Renda Mínima (Minimum Income) project advocated by then-senator Eduardo Suplicy (PT-SP).
• 19
  Available at ˂ http://www.senado.gov.br/noticias/OpiniaoPublica/inc/senamidia/historico/ 1999/10/zn101925.htm˃. Disagreements between the PSDB and PFL eventually led to the collapse of the PSDB-PFL coalition for the 2002 election.
• 20
  Another important program is the Auxílio-Gás (cooking gas allowances), a non-conditional cash transfer. The Benefício de Prestação Continuada (Continuous Cash Benefit Program) came into effect in 1996; this, in turn, is a non-conditional, constitutional cash transfer program.
• 21
  Available at < http://www.senado.gov.br/noticias/OpiniaoPublica/inc/senamidia/historico/ 1999/10/zn101925.htm ˃.
• 22
  Available at < https://www1.uol.com.br/fernandorodrigues/arquivos/eleicoes02/plano2002-serra.pdf> .
• 23
  Available at ˂ https://www1.folha.uol.com.br/folha/brasil/ult96u44275.shtml˃ .
• 24
  Renda Mínima would be approved in the Chamber of Deputies in 2004, but, because there was no budget allocation, it was never actually implemented.
• 25
  Available at ˂ https://www1.folha.uol.com.br/folha/brasil/ult96u47524.shtml˃ .
• 26
  Available at ˂ http://www.columbia.edu/~js3317/JASfiles/AgendaPerdida.pdf˃ .
• 27
  The program framework has changed over the years. For more details, see Soares and Sátyro (2009) , Bichir (2010) , and Magalhães and Veiga (2014) .
• 28 n
  Available at ˂ https://www1.folha.uol.com.br/fsp/brasil/fc2104200312.htm˃ .
• 29
  Table 01 excludes voters intending to cast a blank or spoiled ballot.
• 30
  The results are not significantly altered if data on Fome Zero is added to that of the BFP - as policies formulated by the PT – and data on Bolsa-Alimentação and Auxílio-Gas are both added to the BEP - as policies elaborated by the PSDB.
• 31
  The results indicate changes in the predicted values of the dependent variables obtained from the difference between a simulated situation in which no one receives the respective benefits (‘From’ column) and another in which everyone receives them (‘To’ column).
• 32
  If we consider as beneficiaries only those who have said to personally receive the benefit, these results are amplified.
• 33
  Furthermore, in comparison to other candidates, the BEP did not impact the voting for Alckmin.
• 34
  The p-values of the one-sided test of difference between the coefficients are 0.048, 0.023, and 0.059 respectively for government approval rating, party identification, and voting.
• 35
  Available at ˂ https://agenciabrasil.ebc.com.br/politica/noticia/2015-12/relator-do-orcamento-confirma-corte-de-r-10-bilhoes-no-bolsa-familia˃ .
• 36
  Available at ˂ http://g1.globo.com/politica/noticia/2016/11/pente-fino-encontra-11-milhao-de-irregularidades-no-bolsa-familia-diz-ministerio.html˃ .
• 37
  To use logistical binomial models - both in this case and for the 2005 survey - does not alter the conclusions arising from the marginal effects.
• 38
  The results are the same when I included the assessment of the economy as a control (only available in the 2018 survey), which reinforces the argument that this is not a purely retrospective vote.

Appendix

Table 06 Impact of being a beneficiary on the assessment of the Lula Government - 2005

	Dependent Variable
	Approves the Lula Government (In relation to those who do not approve)	Neutral Assessment (In relation to those who do not approve)
Bolsa-Família Beneficiary	1.514**	1.036
	[1.095,2.094]	[0.762,1.406]
Bolsa-Escola Beneficiary	0.954	0.990
	[0.693,1.313]	[0.736,1.331]
Region (Midwest and North)
Southeast	0.856	1.251
	[0.615,1.191]	[0.925,1.693]
South	0.617**	0.877
	[0.411,0.926]	[0.612,1.256]
Northeast	1.397*	1.288
	[0.977,1.995]	[0.922,1.801]
Income (Up to 2 MW)
From 2 to 3 MW	0.741*	1.088
	[0.541,1.016]	[0.818,1.445]
From 3 to 5 MW	0.651**	0.939
	[0.461,0.918]	[0.696,1.266]
Over 5 MW	0.629**	0.812
	[0.437,0.904]	[0.588,1.123]
Sex (male)
Female	0.586***	0.829*
	[0.467,0.735]	[0.676,1.016]
Education Level (Incomplete Elementary School)
Incomplete Middle School	0.648**	1.094
	[0.451,0.930]	[0.765,1.567]
Middle School	0.678*	1.369
	[0.454,1.012]	[0.926,2.025]
High School	0.700*	1.124
	[0.469,1.044]	[0.756,1.671]
Higher Education	0.662	0.725
	[0.367,1.194]	[0.415,1.267]
Age (16 to 24 years)
25 to 34 years	1.211	1.144
	[0.868,1.688]	[0.858,1.526]
35 to 44 years	1.449**	1.003
	[1.026,2.048]	[0.739,1.361]
45 to 59 years	1.260	1.005
	[0.878,1.806]	[0.731,1.383]
60 years or more	1.024	0.781
	[0.681,1.538]	[0.531,1.147]
Municipality (Capital)
Metropolitan Region	1.359	1.418**
	[0.930,1.984]	[1.016,1.979]
Interior	1.217	1.325**
	[0.940,1.574]	[1.049,1.672]
Constant	1.579*	1.076
	[0.917,2.719]	[0.645,1.797]
N	2444
Log pseudolikelihood	-2551.374
AIC	5182.7

Source: Elaborated by the author based on Cesop (2019a) .

Notes: Relative risk ratios are presented; the 95% confidence intervals are in square brackets. Base categories are in bold.^* p < 0.1,^** p < 0.05,^*** p < 0.01

Table 07 Impact of being a beneficiary on Party Identification - 2005

	Dependent Variable
	Identification with the PT (In relation to those who have no identification)	Identification with another party (In relation to those who have no identification)
Bolsa-Família Beneficiary	1.483**	1.040
	[1.080,2.034]	[0.777,1.391]
Bolsa-Escola Beneficiary	0.868	0.975
	[0.638,1.182]	[0.736,1.290]
Region (Midwest and North)
Southeast	0.832	0.538***
	[0.583,1.188]	[0.399,0.725]
South	0.963	0.836
	[0.615,1.508]	[0.588,1.188]
Northeast	1.356	0.984
	[0.933,1.971]	[0.720,1.347]
Income (Up to 02 MW)
From 02 to 03 MW	1.306*	1.413**
	[0.951,1.794]	[1.066,1.872]
From 03 to 05 MW	1.332	1.196
	[0.943,1.883]	[0.886,1.615]
Over 05 MW	1.136	1.239
	[0.765,1.688]	[0.901,1.704]
Sex (male)
	0.483***	0.514***
Female	[0.382,0.610]	[0.421,0.627]
Education Level (Incomplete Elementary School)
Incomplete Middle School	1.228	1.165
	[0.828,1.821]	[0.821,1.654]
Middle School	1.294	1.418*
	[0.853,1.962]	[0.973,2.067]
High School	1.001	1.736***
	[0.646,1.551]	[1.187,2.538]
Higher Education	1.821*	2.700***
	[0.957,3.464]	[1.583,4.605]
Age (16 to 24 years)
25 to 34 years	0.703**	0.892
	[0.511,0.967]	[0.673,1.182]
35 to 44 years	0.803	0.889
	[0.570,1.133]	[0.658,1.202]
45 to 59 years	0.701*	0.847
	[0.485,1.013]	[0.613,1.169]
60 years or more	0.462***	0.923
	[0.289,0.737]	[0.635,1.340]
Municipality (Capital)
Metropolitan Region	0.829	0.997
	[0.573,1.198]	[0.707,1.406]
Interior	0.644***	1.183
	[0.493,0.841]	[0.937,1.495]
Constant	0.544**	0.516***
	[0.305;0.969]	[0.313;0.849]
N	2476
Log pseudolikelihood	-2321.7608
AIC	4723.5

Source: Elaborated by the author based on Cesop (2019a) .

Notes: Relative risk ratios are presented; the 95% confidence intervals are in square brackets. Base categories are in bold.^* p < 0.1,^** p < 0.05,^*** p < 0.01

Table 08 Impact of being a beneficiary on voting intention - 2005

	Dependent Variable
	Intention to vote for Lula (In relation to Alckmin)	Intention to vote for other candidates (In relation to Alckmin)
Bolsa-Família Beneficiary	1.671**	1.197
	[1.104,2.529]	[0.769,1.863]
Bolsa-Escola Beneficiary	0.964	0.877
	[0.643,1.445]	[0.570,1.350]
Region (Midwest and North)
Southeast	0.255***	0.185***
	[0.163,0.398]	[0.118,0.290]
South	0.548**	0.473***
	[0.318,0.946]	[0.273,0.819]
Northeast	1.427	1.039
	[0.821,2.478]	[0.596,1.810]
Income (Up to 2 MW)
From 2 to 3 MW	0.646**	0.552***
	[0.438,0.952]	[0.370,0.823]
From 3 to 5 MW	0.488***	0.408***
	[0.325,0.733]	[0.270,0.615]
Over 5 MW	0.391***	0.356***
	[0.255,0.599]	[0.229,0.552]
Sex (male)
Female	0.788*	1.157
	[0.600,1.035]	[0.877,1.526]
Education Level (Incomplete Elementary School)
Incomplete Middle School	0.738	1.308
	[0.450,1.212]	[0.772,2.216]
Middle School	0.782	1.379
	[0.452,1.352]	[0.771,2.465]
High School	0.601*	1.326
	[0.348,1.037]	[0.747,2.353]
Higher Education	0.498*	1.535
	[0.237,1.049]	[0.738,3.195]
Age (16 to 24 years)
25 to 34 years	0.907	0.850
	[0.613,1.340]	[0.573,1.261]
35 to 44 years	1.176	0.759
	[0.777,1.780]	[0.497,1.159]
45 to 59 years	1.288	1.149
	[0.837,1.984]	[0.739,1.786]
60 years or more	1.156	1.317
	[0.663,2.017]	[0.755,2.297]
Municipality (Capital)
Metropolitan Region	1.337	1.477*
	[0.852,2.099]	[0.931,2.344]
Interior	0.666**	0.619***
	[0.486,0.914]	[0.448,0.854]
Constant	10.914***	7.183***
	[5.125;23.24]	[3.313;15.574]
N	1854
Log pseudolikelihood	-1821.9312
AIC	3723.9

Source: Elaborated by the author based on Cesop (2019a) .

Notes: Relative risk ratios are presented; the 95% confidence intervals are in square brackets. Base categories are in bold.^* p < 0.1,^** p < 0.05,^*** p < 0.01

Table 09 Impact of being a beneficiary on the assessment of the Temer Government - 2018

	Dependent Variable
	Approves the Temer Government (In relation to those who do not approve)	Neutral Assessment (In relation to those who do not approve)
Bolsa-Família Beneficiary	1.343	0.753
	[0.912,1.978]	[0.517,1.096]
Income (Up to 01 MW)
From 01 to 02 MW	0.795	0.769
	[0.570,1.110]	[0.557,1.060]
From 02 to 05 MW	0.599*	0.728**
	[0.342,1.052]	[0.532,0.998]
Over 05 MW	1.371	1.143
	[0.722,2.604]	[0.656,1.993]
Education Level (Incomplete Elementary School)
Incomplete Middle School	1.33	1.202
	[0.895,1.978]	[0.602,2.402]
Middle School	0.599*	1.056
	[0.357,1.006]	[0.665,1.678]
High school	0.858	0.977
	[0.550,1.338]	[0.682,1.400]
Higher Education	0.615	0.898
	[0.271,1.396]	[0.429,1.878]
Sex (male)
Female	0.562***	0.680**
	[0.402,0.786]	[0.467,0.989]
Region (North)
Northeast	1.359	2.339*
	[0.761,2.427]	[0.982,5.573]
Southeast	2.502***	3.421***
	[1.535,4.08]	[1.463,7.999]
South	1.962*	4.984***
	[0.879,4.378]	[1.833,13.56]
Midwest	0.799	3.767***
	[0.376,1.698]	[1.444,9.825]
Age (16 to 24 years)
25 to 34 years	0.921	0.897
	[0.511,1.662]	[0.642,1.252]
35 to 44 years	0.758	0.731
	[0.463,1.240]	[0.446,1.200]
45 to 54 years	1.229	0.861
	[0.718,2.104]	[0.593,1.251]
Over 55 years	1.173	1.025
	[0.756,1.818]	[0.612,1.719]
Constant	0.135***	0.0880***
	[0.0639,0.284]	[0.0355,0.218]
N	1750
Log pseudolikelihood	-1198.869
AIC	2449.7

Source: Elaborated by the author based on Cesop (2019b) .

Notes: Relative risk ratios are presented; the 95% confidence intervals are in square brackets. Base categories are in bold. Standard errors are clustered by federation unit.^* p < 0.1,^** p < 0.05,^*** p < 0.01

Table 10 Impact of being a beneficiary on Party Identification - 2018

	Dependent Variable
	Identification with the PT (In relation to those who have no identification)	Identification with another party (In relation to those who have no identification)
Bolsa-Família Beneficiary	1.156		0.947
	[0.825,1.620]		[0.744,1.207]
	Income (Up to 01 MW)
	0.722**		1.136
From 01 to 02 MW	[0.547,0.953]		[0.721,1.791]
From 02 to 05 MW	0.589***		1.528*
	[0.441,0.787]		[0.926,2.522]
Over 05 MW	0.515**		1.649**
	[0.286,0.928]		[1.030,2.639]
Education Level (Incomplete Elementary School)
Incomplete Middle School	0.973		1.446
	[0.660,1.435]		[0.901,2.321]
Middle School	0.874		1.414
	[0.588,1.298]		[0.679,2.942]
High school	0.587*		1.984**
	[0.322,1.073]		[1.013,3.885]
Higher Education	0.949		1.842*
	[0.481,1.872]		[0.890,3.810]
Sex (Male)
Female	0.818*		0.416***
	[0.646,1.036]		[0.318,0.545]
Region (North)
Northeast	2.081**		1.133
	[1.111,3.901]		[0.679,1.892]
Southeast	1.024		0.957
	[0.612,1.713]		[0.588,1.558]
South	0.992		1.104
	[0.381,2.581]		[0.668,1.825]
Midwest	0.777		0.706
	[0.370,1.632]		[0.381,1.308]
Age (16 to 24 years)
25 to 34 years	1.262		1.065
	[0.823,1.936]		[0.716,1.586]
35 to 44 years	1.283		0.840
	[0.773,2.128]		[0.548,1.288]
45 to 54 years	1.154		0.524***
	[0.704,1.892]		[0.328,0.839]
Over 55 years	0.976		0.428***
	[0.565,1.684]		[0.282,0.649]
Constant	0.212***		0.152***
	[0.107,0.422]		[0.0617,0.376]
N	2076
Log pseudolikelihood	-1447.6294
AIC	2947.3

Source: Elaborated by the author based on Cesop (2019b) .

Notes: Relative risk ratios are presented; the 95% confidence intervals are in square brackets. Base categories are in bold. Standard errors are clustered by federation unit.^* p < 0.1,^** p < 0.05,^*** p < 0.01

Table 11 Impact of being a beneficiary on Party Rejection - 2018

	Dependent Variable
	Rejects the PT (In relation to those who do not reject any party)	Rejects another party (In relation to those who do not reject any party)
Bolsa-Família Beneficiary	0.701***	1.127
	[0.556,0.883]	[0.837,1.516]
Income (Up to 01 MW)
From 01 to 02 MW	1.505***	1.207
	[1.116,2.030]	[0.942,1.547]
From 02 to 05 MW	2.028***	1.306
	[1.440,2.855]	[0.933,1.827]
Over 05 MW	3.081***	2.532***
	[2.072,4.581]	[1.604,3.995]
Education Level (Incomplete Elementary School)
Incomplete Middle School	1.084	1.556**
	[0.755,1.557]	[1.082,2.239]
Middle School	1.410	2.362***
	[0.820,2.423]	[1.449,3.852]
High school	1.930***	2.216***
	[1.337,2.788]	[1.506,3.259]
Higher Education	2.040**	3.753***
	[1.177,3.537]	[2.541,5.545]
Sex (Male)
Female	0.670***	0.499***
	[0.560,0.800]	[0.393,0.634]
Region (North)
Northeast	0.533**	1.373
	[0.288,0.987]	[0.938,2.008]
Southeast	0.820	0.675***
	[0.473,1.419]	[0.511,0.891]
South	0.684	0.744
	[0.391,1.196]	[0.385,1.437]
Midwest	0.527*	0.524***
	[0.274,1.014]	[0.391,0.703]
Age (16 to 24 years)
25 to 34 years	1.051	0.923
	[0.741,1.490]	[0.644,1.324]
35 to 44 years	1.113	1.340
	[0.632,1.962]	[0.897,2.004]
45 to 54 years	1.382	1.737**
	[0.884,2.161]	[1.124,2.683]
Over 55 years	1.342	1.788**
	[0.789,2.281]	[1.095,2.919]
Constant	0.334***	0.135***
	[0.162,0.692]	[0.0722,0.252]
N	2100
Log pseudolikelihood	-1956.3514
AIC	3964.7

Source: Elaborated by the author based on Cesop (2019b) .

Notes: Relative risk ratios are presented; the 95% confidence intervals are in square rackets. Base categories are in bold. Standard errors are clustered by federation unit.^* p < 0.1,^** p < 0.05,^*** p < 0.01

Table 12 Impact of being a beneficiary on voting - 2018

	Dependent Variable
	Vote for Haddad (in relation to Bolsonaro)	Vote for other candidates (in relation to Bolsonaro)
Bolsa-Família Beneficiary	1.744***	0.835
	[1.329,2.289]	[0.638,1.094]
Income (Up to 1 MW)
From 01 to 02 MW	0.693**	1.114
	[0.494,0.973]	[0.735,1.690]
From 02 to 05 MW	0.542***	1.186
	[0.414,0.711]	[0.745,1.887]
Over 05 MW	0.494**	1.051
	[0.282,0.865]	[0.501,2.205]
Education Level (Incomplete Elementary School)
Incomplete Middle School	0.621**	0.953
	[0.420,0.917]	[0.504,1.800]
Middle School	0.938	1.121
	[0.590,1.491]	[0.688,1.826]
High school	0.410***	1.227
	[0.260,0.649]	[0.772,1.952]
Higher Education	0.790	2.193***
	[0.455,1.371]	[1.318,3.646]
Sex (Male)
Female	1.582***	1.685***
	[1.236,2.024]	[1.300,2.184]
Region (North)
Northeast	2.840***	3.119***
	[1.685,4.787]	[1.579,6.163]
Southeast	0.968	1.679
	[0.553,1.693]	[0.883,3.191]
South	1.185	1.813
	[0.519,2.708]	[0.711,4.625]
Midwest	1.145	0.976
	[0.507,2.585]	[0.507,1.879]
Age (16 to 24 years)
25 to 34 years	0.842	0.625**
	[0.546,1.300]	[0.411,0.951]
35 to 44 years	1.270	0.708
	[0.738,2.185]	[0.440,1.138]
45 to 54 years	0.955	0.787
	[0.530,1.721]	[0.511,1.211]
Over 55 years	0.944	0.490***
	[0.516,1.727]	[0.332,0.722]
Constant	0.762	0.203***
	[0.416,1.397]	[0.0955,0.430]
N	1610
Log pseudolikelihood	-1530.5384
AIC	3113.1

Source: Elaborated by the auyhor based on Cesop (2019b) .

Notes: Relative risk ratios are presented; the 95% confidence intervals are in square brackets. Base categories are in bold. Standard errors are clustered by federation unit.^* p < 0.1,^** p < 0.05,^*** p < 0.01

Edited by

Translated by Karin Blikstad

Publication Dates

• Publication in this collection
  11 Mar 2022
• Date of issue
  2022

History

• Received
  02 Feb 2020
• Accepted
  26 Apr 2021