The Workers’ Party and Participatory Institutions: The Influence of Intra-party Dynamics in the Adoption of Participatory Budgeting^*

ABSTRACT

This article examines how the internal dynamics of the Brazilian Workers’ Party (Partido dos Trabalhadores – PT) influences the adoption of participatory public policies by municipal administrations, differentiating PT from the tradition of little social participation maintained by other political parties in Brazil. Drawing on data from the Workers’ Party direct internal elections, the article analyzes how the party’s internal factions interfere with the adoption of participatory policies in municipal administrations, especially the participatory budget (PB). The hypothesis is that the adoption of a participatory budget by a PT administration becomes more likely when the party’s left-wing factions have a stronger presence in municipalities. This hypothesis is tested by estimating different specifications of logistic regression models, including fixed effects and hierarchical models for the 2005-2012 period. Results indicate that the internal composition of the Workers’ Party – as defined by its internal elections – does influence the adoption of PB, corroborating the argument that the Workers’ Party, its internal dynamics, and the adoption of distinctive and participatory public policies are strongly linked.

Workers’ Party; participatory budget; party factions

RESUMO

O artigo argumenta que a dinâmica intrapartidária do Partido dos Trabalhadores (PT) explica a emergência de políticas públicas participativas nas administrações municipais desta organização e que acabaram diferenciando-a da tradição de pouco enraizamento social dos partidos políticos no Brasil. Utilizando de forma pioneira dados do Processo de Eleições Diretas do PT no nível local, analisamos como a dinâmica das facções do partido interfere na probabilidade de adoção, em suas administrações municipais, de políticas públicas destinadas à ampliação da participação política, a saber, as diversas versões locais do Orçamento Participativo (OP). Nossa hipótese é de que a probabilidade de adoção do OP como política pública em um município administrado pelo partido é uma função positiva da presença local de facções da esquerda do PT. Testamos nossa hipótese estimando diferentes especificações de modelos de regressão logística, incluindo uso de efeitos fixos e modelos hierárquicos, para o período de 2005 até 2012. Os resultados evidenciam a hipótese de que a composição interna do PT, medida por suas eleições internas, importa para a adoção do OP e aprofundam o nosso argumento que associa o PT e a sua dinâmica interna a políticas públicas distintivas e de perfil participativo.

Partido dos Trabalhadores; Orçamento Participativo; facções partidárias

RÉSUMÉ

L’article fait valoir que la dynamique intrapartie du Partido dos Trabalhadores (PT) explique l’émergence de politiques publiques participatives dans les administrations municipales de cette organisation et qu’elle la différencie finalement de la tradition de faible enracinement social des partis politiques au Brésil. En utilisant des données pionnières du processus électoral direct du PT au niveau local, nous analysons comment la dynamique des factions des partis interfère avec la probabilité d’adopter, dans leurs administrations municipales, des politiques publiques visant à élargir la participation politique, à savoir les divers versions locales du Budget Participatif (BP). Notre hypothèse est que la probabilité d’adopter le BP comme politique publique dans une municipalité dirigée par le parti est une fonction positive de la présence locale des factions à gauche du PT. Nous avons testé notre hypothèse en estimant différentes spécifications des modèles de régression logistique, y compris l’utilisation d’effets fixes et de modèles hiérarchiques, pour la période de 2005 à 2012. Les résultats montrent que la composition interne du PT, telle que mesurée par ses élections internes, est importante pour l’adoption du BP et approfondent notre argument qui associe le PT et sa dynamique interne à des politiques publiques distinctives et participatives.

Partido dos Trabalhadores; budgétisation participative; factions du parti

RESUMEN

El artículo argumenta que la dinámica intrapartidaria del Partido de los Trabajadores (PT) explica la emergencia de políticas públicas participativas en las administraciones municipales de esta organización y que terminaron diferenciándola de la tradición de poco arraigo social de los partidos políticos en Brasil. Utilizando de forma pionera datos del Proceso de Elecciones Directas del PT a nivel local, analizamos cómo la dinámica de las facciones del partido interfiere en la probabilidad de adopción, en sus administraciones municipales, de políticas públicas destinadas a la ampliación de la participación política, a saber, las diversas versiones locales del Presupuesto Participativo (OP). Nuestra hipótesis es que la probabilidad de adopción del OP como política pública en un municipio administrado por el partido es una función positiva de la presencia local de facciones de la izquierda do PT. Probamos nuestra hipótesis estimando diferentes especificaciones de modelos de regresión logística, incluyendo el uso de efectos fijos y modelos jerárquicos para el período de 2005 a 2012. Los resultados evidencian la hipótesis de que la composición interna del PT, medida por sus elecciones internas, importa para la adopción del OP y profundizan nuestro argumento que asocia el PT y su dinámica interna a políticas públicas distintivas y de perfil participativo.

Partido de los Trabajadores; presupuesto participativo; facciones partidarias

INTRODUCTION

One of the most debated phenomena in current Brazilian politics is what is referred to as petismo, and its opposition to alleged lulismo. The question revolves around potential explanations for the continued victories of the Workers‘ Party (Partido dos Trabalhadores – PT) in presidential elections, and its significant expansion in the Executive and Legislative powers at different Federation levels. Are these achievements due to the political qualities of its founder and main leader, Luiz Inácio Lula da Silva – in particular his charisma and close relationship with voters? Or are they derived from a more complex institutional political process, comparable to trajectories of party identity building observed in countries of longer democratic experience?

In the first hypothesis, PT’s successful trajectory would be an epiphenomenon of a more structural component of Brazilian politics: the direct relationship between charismatic leaders and the masses, especially those excluded from the economic, political and social structuring processes of democracy in Brazil. In the second case, PT would be considered a novelty in the institutional political landscape, previously lacking in solid political parties rooted and endowed with a clear identity before their electorates. In that case, Lula would be one factor among many – although a certainly significant one.

Works which explore and defend the specific nature of PT often associate it with a pioneering practice of involvement with civil society (Keck, 1992; Meneguello, 1989). They define its effectiveness – or its “efficient secret” – as the ability to voice the demands of social sectors which had been secularly excluded by the parties which previously dominated Brazil’s political landscape. The Workers’ Party would then characterize a sui generis reality in the Brazilian political framework as the first organization to be able to consolidate an image and reputation which attracts the more permanent adherence of voters and social groups through an effort to articulate civil society in a “participatory” dimension of democracy (Pogrebinschi, 2012). A sui generis reality because, even if it escaped the usual weakness of other Brazilian parties – due to either cultural, economic-structural or institutional reasons –, the party did so not because of its program or policies, as was the case in other countries. In Brazil, civil society was the main element in the transition to democracy, and it was deeply rooted in the foundation of the party (Avritzer, 2002).

This article emphasizes the importance not only of the social and organizational bases setting PT apart from other Brazilian parties, but also its political program and, especially, the public policies adopted once the party took office. It explores the complex institutional phenomenon which characterizes the close relationship between party and electorate as one that transcends Lula’s political figure – notwithstanding his ability and charisma. It also analyzes the historical contingencies of the transition to democracy embodied by the involvement of civil society in the political sphere.

Our first goal is to illustrate this complexity by analyzing the relationship between PT and the expansion of channels for political participation in Brazil. The analysis focuses on how participatory public policies adopted by the party’s administrations allowed Brazilian voters to associate PT with a specific brand. The second specific goal consists in evaluating the role played by PT’s factions and internal divisions in defining the public policies pursued by the Executive Power and in the interaction with the participatory sphere of democracy. We argue that the presence of left-wing factions in the party’s internal ideological divisions is decisive for the adoption of participatory policies built in close dialogue with social movements, which have become a staple of the Workers’ Party along its trajectory.

The main argument here revolves around three distinct political agency vectors: i) societal coalitions of support to the party; ii) internal conflict between its factions; and iii) the participatory public policies adopted by its local administrations. Vector iii) is the main object of the present analysis. We propose that intra-party dynamics, in interaction with civil society, explains the adoption of public policies which have become an important part of the PT “brand”. Our analysis explores the way intra-party dynamics interfered in the adoption of public policies which expand political participation – especially the adoption of several local versions of participatory budget (PB). Our hypothesis states that the adoption of PB by a PT administration becomes more likely the more present the leftist factions of the party are in a given municipality.

Should our hypothesis be confirmed, our contributions concern two different areas of research. The first relates to the literature on the adoption of participatory institutions, including a new variable: the adoption of PB and intra-party dynamics (Donaghy, 2011; Pogrebinschi and Santos, 2011; Pogrebinschi and Ventura, 2017; Touchton; Wampler, 2014). The second considers the existing volume of literature about the party and proposes a reinterpretation of what has been referred to as the “PT way of governing” (Samuels, 2004; 2008). The tested hypotheses estimated different specifications of logistic regression models, including the use of fixed effects and hierarchical models for the 2005-2012 period. The adoption of PB is the dependent variable, whereas the presence of a mayor from PT and the internal composition of the party are the main independent variables in the set of models. Data about PT’s internal elections at municipal level were used to identify the composition of internal factions. To the best of the authors’ knowledge, this is the first article to use this information.

Results indicate that the survival of PT’s identity, related to the adoption of new forms of political participation, is connected to its more leftist factions rather than to the political agenda of its main faction, the Majority Camp. This interpretation may lead to a better understanding of the reasons why PB was not adopted at national level after Lula took office in 2003.

This article is divided into four sections, in addition to this Introduction and a Conclusion. The The Workers’ Party and the participatory budget section reviews the literature on the theme. Factionalism in the Workers’ Party addresses the factionalism within the Worker’s Party, introducing the Majority Camp as a potential driver of the adoption of PB in municipalities. The Data Processing and Methods section describes the data and methods used. Results presents the main results achieved. Finally, the Conclusion summarizes the main findings of the study and their implications for an ongoing research agenda.

THE WORKERS’ PARTY AND THE PARTICIPATORY BUDGET

First introduced in 1989, in the city of Porto Alegre, under the local administration of Olívio Dutra (PT), the participatory budget was widely adopted by PT administrations in Brazil up until Lula’s election in 2002. It is the main distinguishing public policy adopted by the party, and marks what is known as the “PT way of governing”. PB was also the main form of joining municipal public policies with the participatory sphere of democracy, and has always been strongly connected to PT administrations (Amaral, 2010; Goldfrank, 2011; Gurza Lavalle; Romão and Zaremberg, 2014; Samuels, 2004; Wampler, 2008b).

Both the Workers’ Party and the participatory budget have become a privileged object of investigation in national and international literature. A large volume of papers, theses, books and articles have provided explanations for the rise of partisanship in Brazil, of which PT stands out as a singular example amongst the national political system (Anderson, 2011). The adoption of participatory budget – a mix between public policy and participatory institution – was responsible, according to PT members, for achieving a true “reversal of priorities” in local government agendas through the “promotion of participation” (Amaral, 2011; Amaral and Power, 2016; Avritzer, 2009; Bezerra, 2014; Pontual and Silva, 1999; Samuels, 2004).

Until very recently, these analyses used to approach each of these phenomena in an isolated manner. The objective here is to contribute to connecting the phenomenon of partisanship in Brazil with the implementation of participatory public policies (Avritzer, 2009; Romão, 2010a; Touchton and Wampler, 2014; Wampler, 2008b). In line with Amaral and Power¹ (2016), we investigate PT from the perspective of its internal organization and structure, relating these characteristics to participatory public policies and institutions in various local contexts. This study proposes to analyze the relationship between the PT and PB based on the hypothesis that both are strongly connected. This hypothesis will be further supported by research, data and an empirical analysis aimed at understanding this phenomenon. In order to contextualize the academic debate, the present and following sections respectively make a brief commentary on the relevant literature about the adoption and dissemination of the participatory budget as a local public policy, and on structure, organization and disputes internal to the Brazilian Workers’ Party.

Wampler’s (2008a) analysis indicates the adoption of PB by municipalities with a population of over 100,000 inhabitants between 1989 and 2004 occurred in three waves. The first lasted from 1989 to 1996, when 76% of the municipalities which adopted PB were governed by PT. The second, between 1997 and 2000, when only 32% of the municipalities that adopted PB were governed by PT, recorded a significant drop, demonstrating that this participatory institution was no longer under the party’s exclusive domain. The third, between 2001 and 2004, when 36% of the municipalities that adopted PB were governed by PT, points to a small increase. What explains the adoption of PB by municipal administrations governed by the Workers’ Party and its expansion to municipalities governed by other parties?

Wampler proposes that the presence of a Workers’ Party mayor and the materialization of a party effort to adopt the program justifies the adoption of the participatory budget in the first wave. In the second wave, the adoption of PB is no longer related to the presence of the Workers’ Party in municipal administrations, since the party had a poor performance in 1996 elections, but rather due to the considerable growth of other parties to the left of the political spectrum – the Communist Party of Brazil (Partido Comunista do Brasil – PCdoB), the Brazilian Socialist Party (Partido Socialista do Brasil – PSB) and the Green Party (Partido Verde – PV) –, which then held many seats in city councils. However, it is also important to note that mayors began to seek participation in national public policy networks to improve their administration and increase their chances of being re-elected. In the third wave, the Workers’ Party’s is once again present in the municipal administration due to the success the party had in 2000 elections in comparison with 1996 (Wampler, 2008a).

Touchton and Wampler (2014) continue the research agenda initiated by Wampler (2008a), crediting participatory institutions with the merit of positively impacting relevant social indicators. The results presented by the authors demonstrate that PB has generated better social results in municipalities governed by PT. Therefore, they associate the participatory budget with the two pillars of the “PT way of governing”: not only promoting participation through the adoption of the program in Brazilian municipalities, but also promoting an inversion of priorities through the improvement of public services in the country (Touchton and Wampler, 2014:1460). For example, all municipalities with a population of over 100,000 inhabitants and under a PT administration have adopted PB between 1989 and 2004. From a macro standpoint, their research provides enough results to justify the importance of analyzing the relationship between PT and PB at the national level, while also indicating the need for more robust methodologies.

Leonardo Avritzer (2009) pioneered the analysis of the connection between the Workers’ Party and the experimentation with participatory democracy in Brazil. According to the author, the connection between civil society and political society had been underestimated in studies about the adoption of participatory institutions. He defends that the political society responsible for this connection is represented by a specific party: PT. This connection would have been established through the transformation of civil society claims for participation in participatory institutions organized by the state in the municipal administrations governed by PT. Based on the examination of participatory budget experiences, municipal health councils, and municipal master plans, the main variables presented by Avritzer (2009) were the degree of adhesion to participation shown by local political society, and the greater or lesser presence of PT in that municipality (Avritzer, 2009:173).

For this reason, Porto Alegre and Belo Horizonte – both cities with a strong civil society and a relative consensus amongst the political society concerning participation – would be ideal for the adoption of the participatory budget. São Paulo, a city which also had a strong civil society, had, however, a divided political society when it came to participation – especially due to the presence of PT factions which did not value the theme. Adopting PB in São Paulo, therefore, would be relatively weak and inefficient. In Salvador, the implementation and development of PB could find no breeding ground. The city’s civil society was fragile, while political society was hostile to participation – with a traditional leftist administration led by PCdoB and opposed to participatory institutions, following the political system implemented by the rightist Liberal Front Party (Partido da Frente Liberal – PFL) (Avritzer, 2009:173).

Avritzer argues that the failure of participatory budget experiences is inevitable if the policy is implemented in contexts which are not conducive to broad dialogue between civil and political society. The author also credits the success of Porto Alegre’s and Belo Horizonte’s experiences to the strong presence of PT in these municipalities. Porto Alegre was governed by the party between 1989 and 2004, and Belo Horizonte between 1993 and 2012. In both cases, consensus about participation amongst different party factions would have been fundamental to the success of PB.

While Avritzer (2009) is right to include political parties in his analysis, the author’s research does not sufficiently support his claims: in addition to favoring one political party –PT – in the description of “political society”, the author assumes that the dynamics of political society are the same as those of civil society, giving way to an investigation which favors the former. The author assumes that the unity of PT – i.e. intra-party harmony – is a necessary condition for the success of PB, not taking into account the party’s factions and dynamics potential interference with its adoption (Romão, 2010a).

Motivated by his criticism to Avritzer’s approach (2009), Romão (2010b) investigates the specific dynamics of political society and the way they influence the adoption of participatory budget. He uses a case study of PB in the city of Osasco (São Paulo) to indicate how PB is guided by dynamics strongly connected to the idea of political competition: instead of being a participatory institution under civil society, PB tends to be influenced by people with strong connections to political parties. The author indicates the significant participation of PT members in these participatory channels. As his main contribution, Romão demonstrates that the participatory budget has a strategic role for the Workers’ Party’s administrations and internal dynamics, especially in the selection of electoral candidates (Romão, 2010b:202).

From a micro standpoint, these works show that the party’s factionalism and internal dynamics are important for the adoption of PB, either due to the influence of intra-party harmony on the program’s success (Avritzer, 2009) or to the strategic component it represents for PT administrations (Romão, 2010a; 2010b). However, there is still a gap in the literature when it comes to the relationship between party factionalism and the development of participatory public policies at macro level, considering the national expansion of the program and the importance of producing more generalized causal inferences, without limiting the analysis to a few case studies.

The importance of identifying macro effects has marked the recent literature on the effectiveness of public policies, participatory institutions and partisanship. The present examination of the causes underlying the adoption of PB and of the way they interact with factionalism and the internal composition of PT is inserted in this debate. The next section examines existing literature on PT’s internal politics.

It is worth noting that this article does not aim to measure factors which determine the quality of the participatory process promoted by PB. This theme has been extensively investigated in comparative literature on PB, which focus on variables such as the composition of city councils, the level of decentralization of the country – and, consequently, of municipal management – and association patterns as strong explanations (Baiocchi, Heller and Silva, 2011; Dias, 2002; Goldfrank, 2011). Besides, since we opted for identifying macro effects of intra-party dynamics which interfere with the adoption of PB, to the best of the authors’ knowledge there are also no indicators on the quality of participatory processes which encompass the wide range of experiences analyzed here. In spite of this limitation, we believe the political conflict behind PB and its unfolding among actors in the electoral arena are factors which influence the quality of this participatory public policy.

FACTIONALISM IN THE WORKERS’ PARTY

Factionalism and intra-party dynamics are themes which still lack deeper conceptual and empirical development in Political Science, especially when it comes to how internal disputes interfere with the public policies adopted when certain parties hold the executive branch at different levels of government. Factionalism is a characteristic of left-wing political parties in Latin America and Europe (Boucek, 2009; Panitch and Leyes, 2001; Rand Smith, 1998), and is reflected in intra-party dynamics through the composition of party chairpersons, interfering in their more central strategies.

Analyses related to the Workers’ Party generally highlight the various social actors which gave rise to the party, focusing on the convergence between academic sectors mainly located in São Paulo, base ecclesial communities (BECs) of the Catholic Church, and a new trade unionism emerging in the ABC² region of São Paulo (Keck, 1992; Meneguello, 1989). PT is marked by strong factionalism from its outset: the party brought together a variety of the country’s left-wing organizations. These include some which had previously been a part of the Brazilian Democratic Movement³, but left the party due to their clear disagreement with its methods of opposing the military regime. The party’s composition thus is highly diverse and constantly disputing its direction (Lacerda, 2002).

According to Lacerda (2002:7), the party’s factions have diverged on two issues ever since the party was founded. The first is the party’s long-term – or ideologically foundational – objectives. The second are the public policy programs it offers the electorate. For the purposes of this article, it is worth questioning if and to what extent internal disagreements between these factions interfered in the ideals of participation expressed in the “PT way of governing” embodied by the adoption of PB in municipalities it managed. These differences were organized into two main blocks: i) factions ideologically identified as left wing; ii) what is referred to as the Majority Camp. Given its obvious leading role in the party’s historical development, the Majority Camp is at the center of the hypotheses regarding the adoption and continuity of PB in Brazilian municipalities.

This majority faction has existed ever since 1983, and was initially referred to as the “Articulation of the 113” (Articulação dos 113), later renamed “Unity in the Struggle” (Unidade na Luta), and finally consolidated as the Majority Camp (Campo Majoritário)⁴, the first dominant faction in the history of the party (Ribeiro, 2010). Except for a brief interregnum between 1993 and 1995, in which a left-wing coalition obtained a majority in the party’s internal composition, the Majority Camp has always had a majority in the Workers’ Party’s internal elections, and is primarily responsible for the political behavior adopted by the party at national level.

The Majority Camp was formed as a strategy to limit the growth of leftist factions, which had become increasingly relevant in intra-party dynamics, hindering the majority’s leaders’ strategic objectives. Thus, the Majority Camp developed an “umbrella” character, grouping party leaders and activists which had no ties the leftist factions. As a result, the party’s left-wing factions suffered systematic defeats in internal assembly elections, which favored their adoption of more flexible electoral strategies, especially with regard to the concepts of democracy and participation, and the prioritization of parliamentary activity as a method of action (Lacerda, 2002; Ribeiro, 2010).

Until 2001, political competition to elect PT’s chair used to occur at its National Meeting. The process was initiated in municipal meetings, responsible for electing representatives for state meetings, which in turn would elect representatives for the National Meeting, which elected the national chair (Lacerda, 2002). In 2001, PT became the first political party in Brazil to choose its chair through a direct election process, in which all members and affiliates who fulfilled their party obligations were called to elect the party’s chair and the ticket of their preference for leading municipal, state, and national committees. This process is part of an electoral calendar preceded by a period of campaigns marked by intense political debates, culminating in a single day of voting, along the lines of the Brazilian general elections. The choice of members for chair and committee positions is based on candidates and tickets at all three state levels.

Regardless of positive or negative opinions about the party’s internal elections, the fact is that its introduction represented a new age for PT’s composition, marked by the Majority Camp’s hegemony (Ribeiro, 2010:282), especially in the national committee. However, intra-party disputes at municipal level did not corroborate this hegemony. Leftist factions maintained an expressive presence in the composition of several municipal committees, thus leading to the question: to what extent did the rise of the Majority Camp impact the decision of the party’s municipal administrations to adopt and maintain PB as a public policy? The relationship between the relative size of the factions and the development of participatory public policies is the main object of this article, to be examined in the following section.

METHODS AND DATA

Is factions’ relative size at municipal level decisive for the adoption of a participatory budget in a given city? In order to answer this question, one must look at the basic question already asked by other studies (Abers, 2000; Avritzer, 2009; Goldfrank, 2011; Wampler, 2008a): is the Workers’ Party a necessary and sufficient element to explain the emergence of PB at municipal level and on a national scale? The empirical tests summarized below focus on these two main questions.

Research on participatory institutions tends to use case studies or comparisons between a few experiments as main empirical tools. However, three key elements have been changing this: i) the expansion of local experiences, which now cover most of the country, and the expansion of national-level participatory institutions; ii) an increase in research about the effectiveness of participatory institutions at macro level; and iii) the current popularity of studies which focus on causal effects in international political science. These pre-conditions have resulted in the current trend studying participatory institutions through quantitative techniques and databases which contain a large number of case studies in order to identify their effects and characteristics in Brazil (Lavalle, Voigt and Serafim, 2016; Pogrebinschi and Santos, 2011; Pogrebinschi and Ventura, 2017; Touchton and Wampler, 2014).

The present study is part of this recent trend, and it seeks to identify the causes and effects of the adoption of participatory institutions in Brazil at macro level. To this end, it deploys a set of econometric models⁵ to identify whether the presence of a PT mayor and the internal composition of the party’s political factions alter the probability of a municipality adopting PB. It estimates different models for both hypotheses.

First of all, our process estimates logistic models by maximum-verisimilitude estimators. These models are commonly used in Political Science for considering binary variables as dependent. The coefficients of the models must be read as logit units of each variable, and intuitive interpretation is used only when dealing with each variable’s statistical significance and effect direction. The substantive interpretation of these coefficients is based on their odds ratio, simply obtained by its exponentiated values.

A second form of control was adopted exclusively in the case of the model used to analyze the Workers’ Party’s impact, in order to make coefficients less biased. A new specification was added to this model by using the fixed effect in the states, so as to control the interference of characteristics specific to each municipality. The introduction of fixed effects allows the model to section the analysis based on characteristics considered fixed in time within each unit of analysis (Angrist and Pischke, 2009). In this case, controlling by state enables the comparison between municipalities within each state, making comparison parameters more balanced and controlling potential biases in estimators. Fixing effects by state is an attempt to compare similar municipalities by stressing differences in variables of interest. This econometric technique basically consists of including a dummy for each of the 27 Brazilian states listed in the model⁶.

Third, we estimate hierarchical logistic – or multilevel – models so as to verify the effect of PT on the adoption of PB. Hierarchical models consist in explicitly modeling estimators at different levels if pieces of data have been grouped into a given unit. When analyzing the reasons behind the implementation of PB at municipal level, data was grouped – municipalities – at a second level – states –, which explains the use of hierarchical models. In addition to the fact that they adapt to data structure, there are two other justifications for the use of hierarchical models in this article.

From a statistical point of view, hierarchical models are a more efficient remedy for balancing within and between variation (Gelman and Hill, 2007). This new specification adds statistical robustness to the results. Theoretically, hierarchical models contribute to the study by enabling the identification of the difference in the impact of each state that adopted PB as a public policy and how the presence of a PT mayor varies in these groups. These models inform an explicit estimation of parameters as random variables, thus identifying average points for each group of data. The present study also assumes a random interception for each state and the impact of the mayor’s affiliation to the PT as a random coefficient, both following normal distributions. The resulting estimation shows whether there are states are more or less prone to adopting PB, as well as whether there are states where mayors’ affiliation to PT matters more or less.

These final two families of models are not be used to measure the impact of PT factions, given the small number of cases in this model – which is restricted to municipal administrations governed by PT – and its lesser degree of freedom, which reduces its efficiency. We use a database composed of all Brazilian municipalities with over 100,000 inhabitants, which encompasses 253 municipalities representing over half the total population. These municipalities are also an appropriate representation of the Brazilian urban population, which accounts for 85% of the country’s population and is mostly found in the cities included in this group of analysis. It is also worth noting, however, that this restriction in the number of municipalities is exclusively due to the availability of data on the adoption of PB⁷. For each municipality, we use data from two local election cycles: 2005-2008 and 2009-2012. The resulting database panel includes 506 units of analysis⁸.

A) THE PARTICIPATORY BUDGETING AS A DEPENDENT VARIABLE

The adoption of PB by a municipal government i (where i represents each Brazilian municipality with over 100,000 inhabitants) in electoral cycle t is the dependent variable for the logistical models to be estimated. During the 2005-2008 cycle, 46 out of 253 municipalities adopted PB, with an increase to 54 municipalities in the 2009-2012 period.

B) THE WORKERS’ PARTY AS AN EXPLANATORY VARIABLE

The first group of models to be tested here attempt to ascertain whether the presence of a PT mayor matters for the adoption of PB. As mentioned, PB experiences had been nationally and internationally expanding throughout the 1990s. Once its adoption was no longer under the exclusive domain of PT, the discussion over whether PT was still important for the adoption of the PB at macro level became a new research agenda in Brazilian politics. Wampler (2008a) wrote the main study about the topic, using panel data for Brazilian municipalities in the 1989-1996, 1997-2000 and 2001-2004 electoral cycles. He concludes there is a positive impact of PT only during the first and third periods.

The present study seeks to expand the analysis proposed by Wampler (2008a) into the two following electoral cycles, 2005-2009 and 2010-2013. The study tests whether mayors’ affiliation to PT changes the probability of their city adopting PB. Data on mayors’ affiliation was obtained from the Superior Electoral Court (Tribunal Superior Eleitoral – TSE) and will be introduced into the econometric model as a binary variable.

C) THE INTERNAL COMPOSITION OF THE WORKERS’ PARTY AS AN INDEPENDENT VARIABLE

The second group of models to be tested here attempts to measure the impact of PT’s internal composition on the adoption of PB by using data from the party’s internal direct elections in 2005 and 2009, when its municipal, state and national committees were elected. To the best of the authors’ knowledge, this is the first article to use these data, which were obtained through a request made to Workers Party’s National Organization Secretariat (Secretaria Nacional de Organização – SORG-PT)⁹. Our main hypothesis proposes that the adoption of PB is more likely in municipalities where left-wing factions of the party are more significantly present. In the opposite case – municipalities dominated by right-wing and centrist factions of the party –, the adoption of PB would be less likely. Testing this hypothesis requires identifying two variables: the internal composition of the party at municipal level and the ideological position of the party’s internal factions.

The internal composition of PT factions at municipal level is here identified based on the results of the party’s internal direct elections for national committees in 2005 and 2009. These voting results are the best data available to ascertain the internal composition of PT. These data show the approximate size of each faction both at national level and in each of the municipalities where internal elections took place. This information is a valuable representation of the party’s internal competition dynamics, since the votes obtained by a ticket define its influence on municipal committees.

We used municipal votes on tickets competing for the national committee as a representation of each faction’s size at municipal level. This occurred for two reasons: first, votes for national candidates and tickets are available and systematized by the party’s national committee, while votes for municipal candidates and tickets depend on the collection of data by each municipal committee of municipalities with over 100,000 inhabitants. Second, factionalism is a part of the party’s tradition and political history, and is a consistent characteristic across all jurisdiction levels. Therefore, even though these are separate votes, the vote for a given faction at national level is an excellent proxy for the presence of this group in the municipal committee, which enables its use to measure PT’s internal composition at municipal level¹⁰.

Leftist factions in the Workers’ Party’s municipal committees were identified through a relatively simple and direct solution. Instead of undertaking the herculean task involved in identifying the ideological position of each faction and defining which factions were positioned to the left of the political and ideological spectrum and their temporal variation, we used the votes for the Majority Camp ticket as proxy for the presence or absence of left-wing factions at municipal level. The reasoning behind this strategy is that, in municipalities where the Majority Camp obtained a larger vote, the presence of left-wing factions is weaker, since the camp was successful in neutralizing it (Ribeiro, 2010a; 2010b)¹¹.

The scope of analysis of this hypothesis is restricted to municipalities where mayors were affiliated to PT, which reduced the database to a hundred cases, considering the two electoral cycles. This was a theoretical choice based on the hypothesis of this article, i.e., the party’s internal composition is influenced by the presence of a PT mayor in the municipality and does not interfere by itself with the adoption of PB. In this case, since there is no conditional effect, it makes more empirical sense to reduce the database than to model the impact of this variable in interaction with the presence of PT in municipal administrations (Brambor, Clark and Golder, 2006).

D) CONTROLLING FOR ALTERNATIVE EXPLANATIONS

Several variables are added to the two aforementioned models in order to control for the value of coefficients of alternative explanations. These variables refer to the expectations expressed in the literature about the reasons behind the adoption of PB, as well as socioeconomic variables at municipal level which are able to reduce the bias of model estimators in the case of variable of interest regarding the affiliation of the mayor and the internal composition of the factions.

First, the main control variables included are: the municipal human development index (HDI) in 2000, cities’ urban population percentage and the logarithm for cities’ gross domestic product (GDP) per capita. These variables enable controlling for hypotheses in which the adoption of PB is explained by the level of local development. In order to control for the influence of the density of local civil society representation on the adoption of PB, we included the number of cities’ civil society organizations, based on data from a survey on private foundations and non-profit organizations conducted by the Brazilian Institute of Geography and Statistics (Instituto Brasileiro de Geografia e Estatística – IBGE) for the 2006-2010 period. It should be noted that this set of variables has already been used in recent literature attempting to explain the adoption of PB and its macro effects, as well as the effects of other participatory institutions on Brazilian public policy cycles (Donaghy, 2011; Touchton and Wampler, 2014; Wampler, 2008b).

RESULTS

Is the presence of the Workers’ Party important for the adoption of the participatory budget? To answer this question, two specifications were estimated according to the aforementioned model. Its coefficients and confidence intervals are shown in Figure 1. Following the suggestions of Kastellec and Leoni (2007), coefficients and confidence intervals are presented in the form of graphs, which makes their interpretation more intuitive to the reader¹².

Figure 1
Coefficients of the models representing the impact of the Workers’ Party in the adoption of the participatory budget

These models contain three important pieces of information. First, three variables have a robust impact amongst different specifications – namely municipal GDP, urban population and the affiliation of the mayor to the PT. Other variables oscillate between the models and do not indicate a consistent effect. The affiliation of the mayor to the Workers’ Party has the most statistically significant impact of the three variables, with a 95% confidence interval and absolute values of greater intensity. This confirms the study’s hypothesis. With regard to GDP and urban population variables, results do not have statistical significance and therefore exceed the vertical line at the zero point. However, their effects are robust and consistent in different specifications and the limitation of their statistical significance is explained by the number of cases and the fact that the database is limited to municipalities with over 100,000 inhabitants, reducing data variation. Thus, considering the electoral cycles of 2005-2008 and 2009-2012, municipalities which are more urbanized, have a higher GDP and mayors from the Workers’ Party are more likely to adopt PB.

What is the substantive interpretation of these results? Or, in other words, what is the marginal effect of each of these variables on the adoption of PB? The most intuitive manner to interpret the logistic models’ results is analysing the predicted probability for different scenarios. Data indicates an increase in the probability of adoption of PB under different scenarios. Figure 2 presents the effects of a PT mayor. The “observed values” method suggested by Hanmer and Ozan Kalkan¹³ (2013) is used to calculate the value of interest by comparing the probability predicted for municipalities not governed by the PT.

Figure 2
Predicted probability of impact of the Workers’ Party in the adoption of a Participatory Budget

According to Figure 2, in the first scenario, if other independent variables keep their average values, the probability of adopting PB without a PT mayor is 14%. In the second scenario, the probability of adopting PB with a PT mayor jumps to 80%. In other words, the mayor’s affiliation to the Workers’ Party increases the probability of adoption of the participatory budget by 73%. Confidence intervals are not reached, demonstrating that this difference is statistically significant at 95% confidence.

Finally, we investigate the heterogeneous effects of the states and of the Workers’ Party among the states. This is the main theoretical contribution of the use of hierarchical models in this article. Figure 3 presents random coefficients for each state, whereas 3.1 presents the impact of PT in each state. Coefficients are represented as differences based on each variable’s average (fixed component of the model). Therefore the vertical line in the center of the graph represents this average point for each parameter. This point is set at zero when the random interception by state is adopted in order to make data interpretation easier. This value was already set at zero for coefficients (Gelman and Hill, 2007). Thus, the graphs should be interpreted as answers to how PT’s presence increases the probability of adoption of PB when compared to the average coefficient of hierarchical models presented in Figure 1.

Figure 3
1 Random estimators of hierarchical models

Figures 3 and 3.1 indicate two main results. First, it reinforces the relevance of the findings, to the extent that it demonstrates a reasonable variation by state when it comes to the impact of PB’s adoption. This indicates that results are not dependent on a small number of states, or that a single state is not exclusively directing the results calculated as average impacts on the models. Results show that PB is a participatory institution and a public policy adopted throughout the national territory and that, therefore, it is perfectly appropriate to link its adoption to PT and its intra-party dynamics. It is worth noting that coefficients do not have statistical significance at 95% confidence, which is a consequence of the low number of cases when each state is analyzed in an isolated manner. Once again, this only corroborates the idea that there is no specific group of states where results are concentrated. Secondly, it expands this research agenda, especially when it comes to how factionalism relates to different regional contexts. Most studies on the Workers’ Party have focused on analyzing municipal, regional and national contexts, and also on comparing PT to other parties, especially in Latin America. However, we still lack analyses which approach the party’s strategies under a comparative lens, taking into account the different processes that may be established depending on the municipality or state in which it operates and the influence of factionalism on its intra-party dynamics – whether with regard to the adoption of a PB, the analysis of other government policies, or even the election strategies adopted by the party.

Figure 3
Random intercepts of hierarchical models

Considering that PB is understood as a public policy, it is interesting to observe how intra-party dynamics – and the predominance of a faction in a given municipal or regional context – implies different priorities in the public policies offered by PT to its social and electoral bases. Several differences between PT administrations in the state of Rio Grande do Sul and in São Paulo may be used as examples. The state of Rio Grande do Sul – where the predominance of leftist factions is recognized and the Majority Camp is virtually non-existent – pioneered the adoption of PB at municipal and state levels during the Olívio Dutra administration (1999-2002). This state’s adoption of PB is also more impactful than most other states’. On the other hand, in the state of São Paulo – one of the few coefficients where PB has a negative impact and is statistically different from zero, also where the Majority Camp originated and where its main leaders (e.g. José Dirceu and José Genoíno) operate –, the adoption of PB was less impactful in comparison with Rio Grande do Sul and several other states, as observed in Figure 3.1.

Qualitative research is recommended for a better understanding of the reasons why the participatory budget has not been adopted at federal level, as originally proposed by the party’s government program. This might be explained by the fact that Olívio Dutra was quickly replaced as Minister for Cities by Marcio Fortes, from the Progressive Party (Partido Progressista – PP). Also, most ministers in the Federal Government were from São Paulo – a city where not only was the effect of PT weaker than on the state, but also the success of PB was only moderate (Avritzer, 2009) – and belonged to the Majority Camp, a faction that never prioritized adopting PB, as indicated by the results.

The results presented above expand the scope of research and answer the remaining questions about the Workers’ Party’s role in the national expansion of PB. As several case studies have already discussed, (Abers, 2000; Avritzer, 2009; Marquetti, Campos e Pires, 2008), having a PT mayor is fundamental for the adoption of PB. Even more importantly, the results of hypothesis 1 expand Wampler’s (2008b) findings to the most recent electoral cycles, confirming the relationship between PT and the adoption of PB at macro level. What remains to be investigated is which type – or which internal composition of PT factions – is more efficient in adopting PB policies.

Does the size of leftist factions in municipal PT committees alter the probability of a city’s adopting PB? In other words, considering the diversity of factions, are the factions located to the left of the party’s political-ideological spectrum more committed to the ideal of participation and to participatory public policies? Figure 4 presents the model estimated to answer this question, using all municipal administrations led by PT in the two electoral cycles under analysis as reference. The variable of interest for hypothesis 2 is referred to as “PT factions” and represents the vote on the Majority Camp’s national ticket at municipal level on a scale from 0 to 100.

Figure 4
Coefficients of the model of the impact of the Workers’ Party’s left-wing factions on the adoption of the participatory budget14

The coefficients of the model presented above show some interesting pieces of information. With regard to the impact of independent variables: when only municipalities led by PT are considered, the impact of GDP, HDI, and the presence of a PT mayor are all positive, the impact of the percentage of urban population is negative and the impact of the number of civil society organizations is close to zero. However, none of these variables has a 95% statistical significance in the confidence interval.

The only variable that exceeds the limit of statistical confidence – allowing for greater robustness in this coefficient’s direction since it is a reduced sample of only 100 cases – is the proxy used to measure the presence of left-wing PT factions at local level. In other words, the stronger the Majority Camp vote at the local level, the less likely is the adoption of PB by a PT municipal administration.

As done in previous models, we used the observed values method to estimate the predicted probability of the impact of the variable of interest in this model. Figure 5 presents this scenario, considering all other independent variables in their average values, and oscillating the Majority Camp vote from 0% to 100% at local level. The results presented in Figure 4 summarize the confirmation of hypothesis 2 regarding the impact of PT’s internal composition on the adoption of PB. The Majority Camp’s popularity at municipal level makes a clear difference in the probability of adoption of PB.

Figure 5
Predicted probability of the impact of the Workers Party’s left-wing factions in the adoption of the participatory budget

Municipalities where the Majority Camp had less than 10% of the votes in the party’s internal direct elections have an average probability of adopting PB above 90%. On the other hand, municipalities where this vote exceeded 80% – that is, where factions to the center and to the right of the party’s internal ideological-political spectrum had a large majority in the municipal committee –, have a probability of adopting PB below 50%.

These results corroborate the central hypothesis of this article and show that PB, widely recognized as a staple of the “PT way of governing” and as a token of the participatory profile of the public policies adopted by the party, does not represent a consensus within the Workers’ Party. Municipalities with a greater presence of leftist factions have a significantly increased probability of assigning a relevant role to PB in the municipal administration.

CONCLUSION

This study aimed to contribute to investigations that analyze the relationship between the Workers’ Party and the adoption of participatory public policies in Brazilian municipalities. Based on previous research (Avritzer, 2009; Romão, 2010; Touchton and Wampler, 2013; Wampler, 2008a), a new explanatory variable was added to this phenomenon: intra-party competition dynamics. The main hypothesis was that the probability of a PT administration’s adopting PB increases with the presence of its left-wing factions in the municipality.

This article makes two main contributions to research on participatory institutions: it expands the hypotheses presented in the comparative case studies carried out by Avritzer (2009) to the national level, and uses data from the PT’s internal elections to determine how intra-party dynamics affect the adoption of participatory institutions. To the best of the authors’ knowledge, this is a pioneering study on the theme of participatory institutions.

Two contributions were made to this analysis: Wampler’s (2008a) data were updated, demonstrating that, in election cycles subsequent to the author’s analysis – 2005-2008 and 2009-2012 –, the presence of a PT mayor was fundamental for the adoption of PB (almost 80% of correspondence). The relevance of the internal composition of the party to the adoption of PB was proven by the unprecedented use of the party’s direct internal elections database, demonstrating that the greater the presence of the Majority Camp, the less likely is the adoption of the participatory budget, with an average probability of over 90%.

This work also presents two contributions to the literature on the Workers’ Party. First, it demonstrates that the promotion of participation, one of the staples of the “PT way of governing”, was progressively more connected to the party’s left-wing factions, at least in the decades under analysis. This fact challenges the belief according to which this value represents a unified agenda inside the party. Secondly, results contribute to a better understanding of the reasons why PB was not expanded to national level when the party took office in 2003. Finally, these findings contribute to the understanding of the reasons why PB has not been locally and regionally expanded at a pace similar to the party’s growing presence in local executive powers over the past two decades. Given the fact that Majority Camp was central to the Workers’ Party federal administration at federal level, as well as to municipalities conquered recently, when the party occupied the presidency (Ribeiro, 2010; Amaral, 2011), the non-adoption and non-expansion of the participatory budget is compatible with the results found at municipal level. The evidence presented sufficiently corroborates the relationship between political parties and participatory institutions, in the hopes that the two become increasingly linked in the literature on participatory institutions in Brazil.

Lastly, considering this study as part of an ongoing research agenda, two issues deserve further analysis. The first is updating the data about the presence of the left in city councils and its influence on the adoption of participatory budgets, adding value to the analysis of the relationship between these data and the influence of the Workers’ Party’s internal dynamics on the adoption of PB. The second is the development of qualitative research to provide more elements for analyzing factionalism within the Workers’ Party and its influence on Brazilian experiences with participatory public policies in general and with the participatory budget in particular.

REFERÊNCIAS

• ABERS, Rebecca Neaera. (2000), Inventing local democracry: grassroots politics in Brazil. Boulder: Lyenne Rienner.
• AMARAL, Oswaldo. (2010), “Adaptação e resistência: o PT no Governo Lula entre 2003 e 2008.” Revista Brasileira de Ciência Política, v. 4, pp. 105-134.
• AMARAL, Oswaldo. (2011). “Ainda conectado: O PT e seus vínculos com a sociedade”. Opinião Pública, v. 17, n. 1, pp. 01-44. http://dx.doi.org/10.1590/S0104-62762011000100001
  » http://dx.doi.org/10.1590/S0104-62762011000100001
• AMARAL, Oswaldo; POWER, Timothy J. (2016), “The PT at 35: revisiting scholarly interpretations of The Brazilian Workers’ Party”. Journal of Latin American Studies, v. 48, n. 1, pp. 147-71. http://www.journals.cambridge.org/abstract_S0022216x15001200
  » http://www.journals.cambridge.org/abstract_S0022216x15001200
• ANDERSON, Perry. (2011), “O Brasil de Lula”. Novos Estudos Cebrap, v. 91, pp. 23-52.
• ANGRIST, Joshua D.; PISCHKE, Jörn-Steffen. (2009), Mostly harmless econometrics: an empiricist’s companion New Jersey: The Princeton University Press.
• AVRITZER, Leonardo. (2002), Democracy and The Public Space in Latina America New Jersey, Princeton University Press.
• AVRITZER, Leonardo. (2009), Participatory institutions in democratic Brazil Baltimore: The Johns Hopkins University Press.
• BAIOCCHI, Gianpaolo; HELLER, Patrick; SILVA, Marcelo. (2011), Bootstrapping democracy. Transforming local governance and civil society in Brazil Stanford: Stanford University Press.
• BEZERRA, Carla. (2014), Do poder popular ao modo petista de governar. Dissertação (Mestrado em Ciência Política), Universidade de São Paulo , São Paulo.
• BOUCEK, Françoise. (2009), “Rethinking factionalism: typologies, intra-party dynamics and three faces of factionalism”. Party Politics, v. 15, n. 4, pp. 455-485. https://doi.org/10.1177/1354068809334553
  » https://doi.org/10.1177/1354068809334553
• BRAMBOR, Thomas; CLARK, William Roberts; GOLDER, Matt. (2006), “Understanding interaction models: improving empirical analyses”. Political Analysis, v. 14, n. 1, pp. 63-82.
• DIAS, Márcia R. (2002), Sob o signo da vontade popular: o Orçamento Participativo e o dilema da Câmara Municipal de Porto Alegre Belo Horizonte: Editora UFMG.
• DONAGHY, Maureen M. (2011), “Do participatory governance institutions matter? Municipal Councils and Social Housing Programs in Brazil”. Comparative Politics, v. 44, n. 1, pp. 83-102.
• GELMAN, Andrew; HILL, Jeniffer. (2007), Data analysis using regression and multilevel/hierarchical models Cambridge: Cambridge University Press.
• GOLDFRANK, Benjamin. (2011), Deepening local democracy in Latin America: participation, decentralization and the left. University Park - PA, Pennsylvania State University Press.
• GURZA LAVALLE, Adrián; ROMÃO, Wagner de Melo; ZAREMBERG, Gisela. (2014), “Partidos políticos e innovación democrática: más allá de purezas y perversiones”. Revista Mexicana de Ciencias Políticas y Sociales, v. 59, n. 220, pp. 21-53. https://doi.org/10.1016/S0185-1918(14)70800-1
  » https://doi.org/10.1016/S0185-1918(14)70800-1
• HANMER, Michael J.; OZAN KALKAN, Kerem. (2013), “Behind the curve: clarifying the best approach to calculating predicted probabilities and marginal effects from limited dependent variable models”. American Journal of Political Science, v. 57, n. 1, pp. 263-277.
• KASTELLEC, Jonathan P.; LEONI, Eduardo L. (2007), “Using Graphs Instead of Tables in Political Science”. Perspectives on Politics, v. 5, n. 04, 755-771.
• KECK, Margaret E. (1992), The workers’ party and democratization in Brazil New Haven: Yale University Press.
• LACERDA, Alan Daniel Freire de. (2002), “O PT e a unidade partidária como problema”. DADOS - Revista de Ciênicas Sociais, v. 45, n. 1, pp. 39-76.
• LAVALLE, Adrian Gurza; VOIGT, Jessica; SERAFIM, Lizandra. (2016), “O que fazem os conselhos e quando o fazem? Padrões decisórios e o debate dos efeitos das instituições participativas ”. Dados v. 59, n. 3, pp. 609-650.
• MARQUETTI, Ademir; CAMPOS, Geraldo; PIRES, Roberto R. Coelho. (2008), Democracia participativa e redistribuição: análise de experiências de Orçamento Participativo São Paulo: Editora Xamã.
• MENEGUELLO, Rachel. (1989), PT: a formação de um partido, 1979-1982 Rio de Janeiro: Editora Paz e Terra.
• PANITCH, Leo; LEYES, Colin. (2001), The end of parliamentary socialism : from new left to new labour London: Verso.
• POGREBINSCHI, Thamy. (2012), “Com a palavra, a sociedade”. Insight - Inteligência, n. 58, pp. 90-99.
• POGREBINSCHI, Thamy; SANTOS, Fabiano. (2011), “Participação como representação: o impacto das Conferências Nacionais de Políticas Públicas no Congresso Nacional”. DADOS - Revista de Ciências Sociais, v. 54, n. 3, pp. 259-305.
• POGREBINSCHI, Thamy; VENTURA, Tiago. (2017), “Mais participação, maior responsividade? As Conferências Nacionais de Políticas Públicas e a qualidade da democracia no Brasil”. DADOS - Revista de Ciências Sociais, v. 60, n. 1, pp. 7-43.
• PONTUAL, Pedro; SILVA, Carla. (1999), “Participação popular nos governos petistas: trajetórias, mecanismos e caráter.” In: I. Magalhães; L. Trevas; V. Barreto (orgs.), Governo e cidadania: balanço e reflexões sobre o modo petista de governar. São Paulo: Fundação Perseu Abramo.
• RAND SMITH, William. (1998), The left’s dirty job: the politics of industrial restructuring In France And Spain Pittsburgh: University of Pittsburgh Press.
• RIBEIRO, Pedro Floriano. (org.). (2010), Dos sindicatos ao governo: a organização nacional do PT de 1980 a 2005 São Carlos: Edufscar/Fapesp.
• ROMÃO, Wagner. (2010a), “Entre a sociedade civil e a sociedade política ”. Novos Estudos - Cebrap, n. 87, pp. 199-206.
• ROMÃO, Wagner. (2010b), Nas franjas da sociedade política: estudo sobre o Orçamento Participativo. Tese (Doutorado em Ciência Política), Universidade Estadual de Campinas, Campinas.
• SAMUELS, David. (2004), “From Socialism To Social Democracy: Party Organization and The Transformation of The Workers’ Party in Brazil”. Comparative Political Studies, v. 37, n. 9, pp. 999-1024.
• SAMUELS, David. (2008), “A evolução do petismo (2002-2008)”. Opinião Pública, v. 14, pp. 302-318. http://dx.doi.org/10.1590/S0104-62762008000200002
  » http://dx.doi.org/10.1590/S0104-62762008000200002
• TOUCHTON, Michael; WAMPLER, Brian. (2014), “Improving social well-being through new democratic institutions”. Comparative Political Studies, v. 47, n. 10, pp. 1442-1469. http://journals.sagepub.com/doi/10.1177/0010414013512601
  » http://journals.sagepub.com/doi/10.1177/0010414013512601
• WAMPLER, Brian. (2008a), “A difusão do Orçamento Participativo brasileiro: ‘boas práticas’ devem ser promovidas?” Opinião Pública, v. 14, n. 1, pp. 65-95.
• WAMPLER, Brian. (2008b), “When does participatory democracy deepen the quality of democracy? Lessons From Brazil”. Comparative Politics, v. 41, n^o 1, pp. 61-81.

NOTES

• *
  We would like to acknowledge anonymous contributions to the improvement of this study, as well as the suggestions made by Carla Bezerra, Fernando Guarnieri, Julio Canello, Thamy Pogrebinschi, Wagner Romão, Brian Wampler and André Oliveira to the different versions of this article. Translation of Tathiana Abreu and Jun Shimada.
• 1
  . Their work reviews academic analyses about the Workers’ Party in the previous 35 years, verifying their main approaches and making suggestions for future research agendas.
• 2
  . The ABC region of São Paulo, also known as ABC Paulista, is a traditionally industrial region located in the state of São Paulo, which integrates the Metropolitan Area of São Paulo.
• 3
  . The Brazilian Democratic Movement (Movimento Democrático Brasileiro – MDB) is a Brazilian centrist political party.
• 4
  . Ever since 2007, this faction has been referred to as “Building a New Brazil” (Construindo um Novo Brasil). This name change did not alter the faction’s leadership or goals, which is why this article will continue refer to it as the Majority Camp – a denomination which expresses its hegemony in the party’s internal composition.
• 5
  . Statistical analyses and graphs presented in this article used the R software, version 3.5.0. The database and code routine are available upon request to the authors.
• 6
  . Models with this specification can also be replicated using the difference between the observation and the average value in each unit of analysis, achieving numerically equivalent results.
• 7
  . The authors would like to thank to Brian Wampler for kindly providing the data regarding the adoption of the participatory budget in Brazilian municipalities.
• 8
  . Data on the adoption of the participatory budget during the 2013-2016 electoral cycle is not yet available.
• 9
  . The authors would like to thank SORG-PT for promptly providing the data and other information.
• 10
  . We recognize that this measurement strategy simplifies the diversity of PT’s internal politics, especially at local level, where trends not nationally allied to the Majority Camp have an even more conservative position on participation. However, it is argued that such strategy represents the most practical way of estimating this hypothesis, since there are no official data available to refine this information at local level. Furthermore, measurement errors added to the model by such strategy would have a negative effect on coefficients, making results more conservative and test implementation even stricter.
• 11
  . One may notice the leading role of the Socialist Democracy faction in the promotion of PB amongst the party’s left in the policy’s particular dissemination in the state of Rio Grande do Sul, where the left-wing faction is considerably stronger than the Majority Camp. See, for example, Dias (2002) and Romão (2010).
• 12
  . Regression tables with numerical values are available in the ^Annex.
• 13
  . When analyzing predictive probabilities, most empirical research builds estimates assuming that all models’ covariates are in their average values and only vary in hypothetical scenarios of interest. In the “observed values” method proposed by Hanmer and Ozan Kalkan, values present in databases are applied and only the variables of interest are changed (in this case, the mayor’s affiliation to the Workers’ Party). Then, the predicted probability is ascertained for each unit based on the de facto values present in the database. This process is repeated using a simulation of coefficients to reach an average point and confidence intervals. 1,000 simulations were used in this case.
• 14
  . In order to improve visualization, the scales of HDI, GDP and urban population variables were altered.

Annex I

MUNICIPALITIES WITH A WORKERS’ PARTY MAYOR INCLUDED IN MODEL 4

2005-2008 election cycle: Abaetetuba, Aracaju, Araraquara, Bagé, Belo Horizonte, Botucatu, Cachoeirinha, Camaçari, Cariacica, Conselheiro Lafaiete, Contagem, Coronel Fabriciano, Diadema, Dourados, Embu, Fortaleza, Gravataí, Guarulhos, Hortolândia, Itaboraí, Itajaí, Jacareí, Jandira, Lauro de Freitas, Londrina, Macapá, Mesquita, Niterói, Nova Iguaçu, Osasco, Palmas, Porto Velho, Recife, Rio Branco, Santa Maria, Santarém, Santo André, São Carlos, São Leopoldo, Sumaré, Suzano, Teófilo Otoni, Varginha, Várzea Paulista, Viamão, Vitória da Conquista.

2009-2012 election cycle: Anápolis, Araçatuba, Araras, Bagé, Belford Roxo, Bento Gonçalves, Betim, Cachoeiro de Itapemirim, Camaçari, Canoas, Carapicuíba, Cariacica, Colatina, Contagem, Coronel Fabriciano, Cubatão, Diadema, Embu, Fortaleza, Francisco Morato, Governador Valadares, Gravataí, Hortolândia, Itapevi, Jacareí, Joinville, Juazeiro do Norte, Lauro de Freitas, Mauá, Mesquita, Nova Iguaçu, Novo Hamburgo, Osasco, Petrópolis, Pinhais, Porto Velho, Pouso Alegre, Recife, Rio Branco, Santarém, Santo André, São Bernardo do Campo, São Carlos, São Leopoldo, Sapucaia do Sul, Sumaré, Suzano, Teófilo Otoni, Teresópolis, Varginha, Várzea Paulista, Viamão, Vitória da Conquista, Votorantim.

Annex II NUMERICAL RESULT OF LOGISTIC REGRESSION MODELS

	Logistic model	Logistic model with fixed effects	Hierarchical logistic models	Logistic model (PT factions)
	(1)	(2)	(3)	(4)
GDP per capita	0.287**	0.203	0.158	0.003
	(0.142)	(0.170)	(0.183)	(0.004)
Urban population (%)	2.326	3.583*	3.329	-0.094*
	(1.798)	(2.102)	(2.229)	(0.055)
HDI (2000)	-0.272	1.055**	-0.164	0.014
	(0.283)	(0.445)	(0.351)	(0.009)
PT mayor	2.372***	2.674***	3.516***
	(0.276)	(0.313)	(0.607)
Leftist factions of PT				-0.026**
				(0.012)
NGOs	-0.0001	-0.0001	0.001	-0.0001
	(0.0001)	(0.0001)	(0.0004)	(0.001)
PT governor	0.118	-1.576**	-0.498	0.385
	(0.450)	(0.711)	(0.983)	(1.211)
Constant	-5.422***	-15.284***	-5.383**	-3.856
	(2.087)	(3.432)	(2.456)	(7.472)
Observations	492	492	492	98
Logistical Probability	-266.433	-232.603	-257.907	-45.997

Notes: Standard error in brackets *p<0.1; **p<0.05; ***p<0.01

Publication Dates

• Publication in this collection
  02 Oct 2020
• Date of issue
  2020

History

• Received
  20 June 2018
• Reviewed
  10 Apr 2019
• Accepted
  8 Apr 2019