sugiyamad69603 - University of Texas Libraries

by user






sugiyamad69603 - University of Texas Libraries
Natasha Borges Sugiyama
The Dissertation Committee for Natasha Borges Sugiyama certifies that this is the
approved version of the following dissertation:
Ideology & Social Networks: The Politics of Social Policy Diffusion in
Wendy Hunter, Co-Supervisor
Kurt Weyland, Co-Supervisor
Andrew Karch
Tse-Min Lin
Raúl Madrid
Gretchen Ritter
Robert H. Wilson
Ideology & Social Networks: The Politics of Social Policy Diffusion in
Natasha Borges Sugiyama, B.A., M.P.Aff.
Presented to the Faculty of the Graduate School of
The University of Texas at Austin
in Partial Fulfillment
of the Requirements
for the Degree of
Doctor of Philosophy
The University of Texas at Austin
August, 2007
First, I would like to thank my advisors, Kurt Weyland and Wendy Hunter, for
their invaluable guidance with this dissertation. They have been enthusiastic supporters
of my research and offered me thoughtful advice from this project’s earliest beginnings.
Along the way they reinforced their appreciation for knowledge of place, shared their
fascination with Brazil, and also pushed me to refine my theory of policy diffusion. For
their mentorship throughout my graduate studies, I extend my deepest thanks.
The five other members of my dissertation committee have also been instrumental
in various phases of the dissertation. Raúl Madrid offered early input on the research
design and persuaded me to pursue both qualitative and quantitative methods to test my
hypotheses. Andrew Karch was an early enthusiast of this study and encouraged me to
think comparatively by drawing links to the American case. Gretchen Ritter supported
my interests in drawing out the gendered implications of social policy design in Brazil.
Tse-Min Lin offered guidance on the quantitative portions of this work, from issues of
measurement to carrying out the event history analyses. Last, Robert Wilson deserves
special mention for his early influence in my career. As one of my first professors as a
Masters Student at the LBJ School of Public Affairs, he introduced me to issues of
decentralization, local governance, and “good governance” in Brazil. This dissertation is
a reflection of those first forays into Brazilian public policy analysis.
As anyone who has engaged in overseas research can attest, undertaking longterm fieldwork is only possible when a researcher receives substantial assistance. This
dissertation would not have been possible were it not for the generosity of numerous
funding agencies, institutions, and individuals.
First, several organizations provided
financial resources for this project and deserve special mention. My pre-dissertation
research trip to Brazil to explore trends in social policy development was supported by
two grants from the University of Texas at Austin, one from the Lozano Long Institute of
Latin American Studies and another through the Study Abroad Office.
Latter, a grant
from the Fulbright-IIE and a fellowship from the National Security Education Program
enabled me to conduct field research in Brazil from 2003-2004. I am also grateful to the
Spencer Foundation and the Government Department of the University of Texas at
Austin, which provided financial support during the writing phase of this dissertation.
Last, the Austin-CR chapter of P.E.O. International took great interest in me and my
project; because of their nomination I received the Esther Garrett Edgerton award.
In Brazil, I was fortunate to encounter many individuals who opened their doors
to me. I am especially grateful to Peter Spink and Marta Farah, both of the Escola de
Administração Pública of the Fundação Getúlio Vargas-São Paulo, for hosting me as a
visiting scholar while I was in São Paulo. Professor José Antonio Gomes de Pinho at the
Escola de Administração at the Universidade Federal da Bahia also welcomed me as part
of his working group at UFBA. Many other individuals shared contacts, insights, and
friendship while I was in Brazil. I would be remiss to exclude anyone from this group,
but I would like to thank, in particular: Luis Odorico de Andrade, Zuleika Arashiro,
Antônio Sergio Araújo Fernandes, Ilka Camarotti, David Fleischer, Jorge Kayano,
Anderson Lima, Diva Moreira, Paula Perry, Fabio Santos Perreira, and Brian Wampler.
Finally, I would like to offer special thanks to my research assistants, Francisco da Costa
Marques, Evelyn Chaves, Ana Paula Karruz, Lilia Asuca, and Natalia Koga. Their
interest, dedication, and diligence in collecting information for this study inspired me in
the field and enriched this study considerably.
At the University of Texas at Austin, I have also benefited from wonderful
colleagues who have read and commented on portions of this work, among them are:
Odysseas Christou, Anna Gruben, Jungkun Seo, and the participants in the studentfaculty Latin American politics working group.
Other friends have offered their
solidarity along the road to dissertation completion, including: Brian Arbour, Mijeong
Baek, Fred Cady, Carolyn Cunningham, Oya Durson, Allison Martens, Mark McKenzie,
Marcy Ribetti, Jennifer Suchland, and Michael Unger.
Most of all, I would like to give thanks to my family. My parents Maria Lúcia
Borges Sugiyama and Iutaka Sugiyama instilled in me a love of learning and appreciation
for Brazil. My brother, Alexandre Borges Sugiyama, preceded me in completing a
doctorate in 1998. Alex shared his experience and expertise, enriching this dissertation.
Along with Alex, Tamzin Sawyer Sugiyama, my sister-in-law, visited me in Brazil
during my field research and has been encouraging throughout. My mother-in-law,
Clarice Carter, warmly welcomed me into the family and has been a great champion
throughout my graduate studies. Finally, my deepest gratitude goes to Greg Carter, a
wonderful scholar and partner, who has shared all aspects of this academic life with me.
Ideology & Social Networks: The Politics of Social Policy Diffusion in
Publication No._____________
Natasha Borges Sugiyama, Ph.D.
The University of Texas at Austin, 2007
Supervisors: Wendy Hunter and Kurt Weyland
This dissertation examines the politics of local social policy making following
Brazil’s re-democratization.
Decentralization in Brazil granted municipalities
responsibility to design and tailor social policies to meet local demands. Yet instead of
developing their own programs many governments chose to adopt those made famous
What accounts for the diffusion of innovations across Brazil?
dissertation tests three approaches for understanding policy makers’ emulation decisions:
political incentives, ideology, and socialized norms. Each of these three motivations
reflects a different paradigmatic response to the question, what drives political behavior?
A conventional political incentives approach follows a rational choice framework that
incorporates neoclassical behavioral assumptions and posits people will behave
strategically to further their own self-interest. The classic assumption in this vein is that
politicians will seek to win re-election. On the other hand, scholars who adopt an
ideational approach examine the way people make choices because of their ideological
convictions. Rather than seek their own political self-interest, actors can make decisions
in spite of themselves or others because of deeply held beliefs about what is right and
how to enact social change. Lastly, a sociological approach examines how individuals
conform to shared norms and seek legitimacy in the eyes of their colleagues.
To test these motivational approaches I examine the diffusion of Bolsa Escola, an
education program, and Programa Saúde da Família, a family health program. Evidence
for my argument is based on statistical event history analysis and qualitative case study
research from four exemplary cities.
The electoral incentives approach offers a
surprisingly weak explanation for the diffusion of innovative social policies. Rather,
diffusion occurs when elected executives feel ideologically compelled to replicate
programs and when policy professionals engaged in relevant networks seek to
demonstrate their adherence to professional norms. Both ideology and social networks
can work together in mutually reinforcing ways to promote diffusion.
List of Tables ......................................................................................................... xi
List of Figures ....................................................................................................... xii
Abbreviations ....................................................................................................... xiii
Chapter 1: Introduction ............................................................................................1
The Puzzle: Social Policy Diffusion in Brazil ................................................3
The Argument in Brief ..................................................................................11
Research Design............................................................................................13
Scope and Organization ................................................................................20
Chapter 2: Towards a Theory of Social Policy Diffusion......................................23
Policy Outcomes: Sectoral & Institutional Approaches ...............................24
A Turn Toward Policy Diffusion ..................................................................34
Towards a Theory of Social Policy Diffusion ..............................................38
Chapter 3: Social Policy Diffusion in Brazil’s Largest Municipalities .................51
Tracing the Diffusion of Bolsa Escola and PSF ...........................................52
Understanding Policy Diffusion: Tendencies among Municipalities ...........58
An Event History Analysis ...........................................................................67
Chapter Conclusions .....................................................................................80
Chapter 4: Bolsa Escola: A Simple Idea Catches On ............................................84
National Context for Education Reform .......................................................84
Local Experimentation & Innovation ...........................................................91
Explaining the Diffusion of Municipal Bolsa Escola in Four Major Cities103
Chapter Conclusions ...................................................................................123
Chapter 5: Programa Saúde da Família: From a Small Project to Major Reform126
National Context for Health Policy Reform ...............................................127
Local Experimentation & Innovation .........................................................133
Explaining the Diffusion of PSF in Four Major Cities ...............................142
Chapter Conculsions ...................................................................................171
Chapter 6: Conclusion: Explaining Social Policy Diffusion in Brazil ................174
The Core Findings.......................................................................................175
Similar Findings Despite the Differences ...................................................189
Theoretical Implications .............................................................................200
Future Research ..........................................................................................204
Appendix A ..........................................................................................................210
Appendix B ..........................................................................................................216
Appendix C ..........................................................................................................221
Appendix E ..........................................................................................................224
Appendix F...........................................................................................................225
Bibliography ........................................................................................................232
Vita .....................................................................................................................253
Table 1.1 Adoption of BE/RM and PSF by Local Government and Administration ....... 16
Table 1.2 Local Government Partisan and Ideological Divide ......................................... 18
Table 3.1 Electoral Competition & Bolsa Escola Adoption (1998) ................................. 59
Table 3.2 Electoral Competition & PSF Adoption (1998) ............................................... 60
Table 3.3 Ideology & Bolsa Escola Adoption (1998)....................................................... 61
Table 3.4 Ideology & PSF Adoption (1998) ..................................................................... 62
Table 3.5 The Gestão Pública Network & Bolsa Escola Adoption (1998) ...................... 64
Table 3.6 Region & Bolsa Escola Adoption (1998) ......................................................... 64
Table 3.7 Summary of Dichotomous Variables in Bolsa Escola Dataset (1998) ............. 65
Table 3.8 Summary of Dichotomous Variables in PSF Dataset (1998) ........................... 66
Table 4.1 Characterics of the Municipal Bolsa Escola & Renda Mínima Programs ........ 98
Table 4.2 Ten Case Studies: Adoption & Non-Adoption ............................................... 105
Table 5.1 Case Studies: Adoption & Non-Adoption ...................................................... 144
Table 5.2 Case Studies: Mayor’s Ideology ..................................................................... 154
Table C.1 PSF coverage of State’s population over Time (%) ....................................... 221
Table C.2 PSF coverage of Region’s population over Time (%) ................................... 222
Table E.1: Structure of the Brazilian Education System ................................................ 224
Figure 1.1 Multiple Directions of Diffusion Effects ........................................................... 8
Figure 3.1 Cumulative Adoption of Bolsa Escola in the Sample ..................................... 55
Figure 3.2 Cumulative Adoption of Programa Saúde da Família in the Sample .............. 56
Figure 3.3 Annual Predicted Probability of Bolsa Escola Adoption for Modal City ....... 78
Figure 3.4 Annual Predicted Probability of PSF Adoption for Modal City ..................... 80
Figure 6.1 Explaining Social Policy Diffusion: A Motivational Approach.................... 176
Figure C.1 National Cumulative Adoption of Programa Saúde da Família ................... 222
Associação Brasileira de Pós-Graduação em Saúde Coletiva
Agentes Comunitários de Saúde
Conselho Nacional de Secretários Municipais de Saúde
Conselho de Secretários de Educação
Centro Brasileiro de Estudos de Saúde
Centro Feminista de Estudos e Assesoria
Fundação Getúlio Vargas- São Paulo
Fundo Nacional do Desenvolvimento da Educação
Fundo de Manutenção e Desenvolvimento do Ensino Fundamental e de
Valorização do Magistério
Instituto Nacional de Assistência Médica da Previdência Social
Instituto de Planejamento Econômico e Social/Instituto de Pesquisa
Econômica Aplicada
Ministério da Educação e Cultura
Ministério de Saúde
Non-Governmental Organization
Norma Operacional Básica
Programa de Agentes Comunitários de Saúde
Pan-American Health Organization
Plano de Atendimento à Saúde – São Paulo
Programa de Agentes de Saúde (PAS) - Ceará
Programa de Interiorização das Ações de Saúde e Saneamento
Programa Médico de Família
Programa Saúde da Família
Programa de Garantia de Renda Mínima
Qualidade Integral à Saúde.
Secretaria de Assistência da Saúde, Ministério da Saúde
Sistema Único de Saúde
United Nations Educational, Cultural and Scientific Organization
United Nations Childrens’ Fund
União Nacional dos Dirigentes Municipais de Educação
Political Parties
PC do B
Partido Comunista Brasileiro (Brazilian Comunist Party)
Partido Democrático Trabalhista (Democratic Labor Party)
Partido da Mobilização Nacional (Party of National Mobilization)
Partido Socialista Brasileiro (Brazilian Socialist Party)
Partido Popular Socialista (Popular Socialist Party)
Partido dos Trabalhadores (Workers’ Party)
Partido Verde (Green Party)
Center Parties
Partido da Social Democracia Brasileira (Party of Brazilian Social
Partido do Movimento Democrático Brasileiro (Party of the Brazilian
Democratic Movement)
Partido Trabalhista Brasileiro (Brazilian Labor Party)
Right Parties
Partido da Frente Liberal (Liberal Front Party)
Partido Liberal (Liberal Party)
Partido Progressista Brasileiro (Brazilian Progressive Party)
Partido da Reedificação da Ordem Nacional (Party of the Reconstruction
of National Order)
Partido Social Cristão (Social Christian Party)
Partido Social Democrático (Social Democratic Party)
Partido Social Liberal (Social Liberal Party)
Partido Trabalhista Renovador
This dissertation analyzes what motivates people to emulate “innovative” policies
that address poverty and social inequality. It is also about the politics of social policy
provision and the reason why it is difficult, or by contrast easy, for social policies to
spread. Although these concerns are relevant for many countries, they are especially
important for Brazil, which has a decentralized federal structure, where local
governments have responsibility for social policies and where the stakes are high for
addressing the needs of the poor.
The politics of social sector reform in Brazil have undergone profound
transformation since redemocratization in the late 1980s. The democratic constitution
(1988) established new social rights (Title I, Chapter 2, Article 6), including the right to
education1 and health.2 At the same time, the constitution also laid out a new set of
institutional mechanisms to carry out these social rights. Municipalities would serve as
an independent third tier of government with considerable political, fiscal, and policy
autonomy, responsible for providing social services. This newfound authority for local
1 According
to the Constitution (1988) education is the right of all and duty of the State and of the family;
free public education shall be provided with equal conditions of access (Title VIII, Chapter 3, Section 1,
Article 205 and 206).
2 According to the Constitution (1988) “[h]ealth is a right of all and a duty of the state and shall be
guaranteed by means of social and economic policies aimed at reducing the risk of illness and other hazards
and at the universal and equal access to actions and services for its promotion, protection and recovery”
(Title VIII, Chapter 2, Section 2, Article 196).
control contributed to significant changes in Brazil’s policy landscape, as these
institutional arrangements led to the inclusion of different actors in the policy process.
In theory, decentralization could bring about greater flexibility and efficiency, and
allow local governments to tailor their social policies according to the diverse needs of
their constituents (Rueschemeyer and Evans 1985: 55-57). After all, Brazil is home to
over 5,500 municipalities that lie between the southern plains and the northern Amazon.
In principle, as municipal governments would claim their authority to develop social
policy, they would design programs according to local realities. Local governance would
also allow for better civil society participation, as cities followed constitutional mandates
for the creation of participatory mechanisms for citizens to engage in policymaking
through local councils.
Throughout much of the late 1980s and 1990s, many states and municipalities
embraced their newfound flexibilities and operated as policy “laboratories” by
experimenting with new administrative and social policies (Abers 2000; Tendler 1997;
Wampler and Avritzer 2004; Wampler 2004). Innovative programs abound as cities
instituted programs such as participatory governance (e.g. Orçamento Participativo,
Participatory Budgeting), income generating cooperatives, and recycling programs, to
name just a few (Spink et al. 2002). That a number of sub-national governments would
become the vanguard of social policy in Brazil was particularly notable given that the
federal government was home to highly specialized technocrats, and that social reforms
at the federal level took place at a snail’s pace (Ames 2001; Weyland 1996).
While sub-national governments did bring a diversity of tailored social policies to
fruition, during this time period another phenomenon was also taking place: policy
replication and diffusion. Simply put, many city governments across Brazil chose to
copy policies from elsewhere rather than customize their own programs. The extent of the
replication is surprising because municipal governments faced dramatically different
political cultures, social inequalities, and levels of poverty. Two programs that are
illustrative of the trend toward policy diffusion are Bolsa Escola/Renda Mínima,3 an
educational grant program,4 and Programa Saúde da Família (PSF), a family health
Bolsa Escola/Renda Mínima started in 1995 in two cities, Brasília (the federal
district) and Campinas (in São Paulo state). These programs provide mothers of lowincome children with cash-grants on the condition their children regularly attend school.
The goal of the program is to improve educational attainment by reducing school
absences and repetition, often caused by the high opportunity cost for poor children to
attend school. The Bolsa Escola program quickly spread across municipalities; within
two years approximately 88 cities had adopted the program (Araújo and de Souza 1998).
By 2001, over 200 cities had municipal Bolsa Escola programs (Villatoro 2004). That
same year, and on the eve of presidential elections, the federal government created a
national Bolsa Escola program that was similar in design but which by-passed
I treat both Bolsa Escola and Renda Mínima programs as comparable as they share similar policy designs
and programmatic goals.
4 This program is also known as a conditional cash transfer program.
municipalities5. This study focuses only on municipal program diffusion, which requires
budgetary and administrative obligations on the part of cities.
Programa Saúde da Família is a family healthcare program that draws on several
local initiatives, including a family doctor program in Niterói (in Rio de Janeiro state)
and other basic community-based health programs established in the rural northeast
(Viana and dal Poz 1998). The goal of the program is to improve prevention and basic
health by working directly with families through home visits. To facilitate linkages to
communities healthcare workers operate within designated territories and in teams
comprised of a doctor, nurse, nurse’s aid and several community health agents who reside
in the neighborhood they serve. This program started in 1994 with the support of the
Ministry of Health, initially in small rural towns in the northeast. Over time this program
gained wider visibility and credibility, spreading dramatically from 55 cities in 1994 to
4,944 municipalities by 2003.
Brazil, like other federal countries with decentralized systems, has experienced
significant policy diffusion. Yet, there are some unique features of Brazilian diffusion
that are worth noting.
First, it has taken place across thousands of municipal
governments. In most single-country diffusion studies of federal systems, scholars have
focused on state-level (or provisional-level) policy replication because states assume
primary jurisdictional responsibilities; a natural consequence is that those studies tend to
have a smaller number of cases, such as the fifty U.S. states. Second, policy replication
The federal government had a limited experiment with Renda Mínima in 1997, where it provided select
cities with matching grants if they established the program. This program was short-lived and is widely
viewed as a policy failure.
has occurred in a country that is well known for its regional contrasts rather than its
similarities. That a mega-city such as São Paulo with elaborate health networks would
adopt a healthcare model that owes its origin to poor small cities in the rural northeast is
remarkable. Similarly, it is surprising that a city like Salvador, with high rates of poverty
and poor educational infrastructure, would implement a school-grant program developed
for wealthy cities with some of the highest rates of human development inside Brazil
(Martins and Libânio 2005). The spread of the same policy models across such diverse
settings is puzzling and worthy of explanation. Before embarking on this study of social
policy diffusion, it is necessary to clarify some concepts and terms.
What is Diffusion?
Diffusion phenomena are all around us and have managed to capture the attention
of a wide range of scholars in the social sciences. Diffusion evokes images from the
natural sciences, such as the spread of a virus across space. But social science disciplines
as diverse as sociology, economics, political science, agriculture, and business have also
sought to explain socially created events that occur in their respective domains of inquiry.
Some occurrences seem to “make sense” as people learn about innovations and quickly
adopt them. One classic example in this vein is the spread of hybrid corn among farmers
in Iowa, who rapidly embraced the usage of hybrid seeds in the 1940s.6 Yet, there are
also trends that are harder to explain.
For instance, the QWERTY keyboard for
typewriters and computers has been widely recognized as an inefficient layout for typists.
The spread of hybrid corn in Iowa is one of the most influential studies on diffusion (see Ryan and Gross
(1943); Gross (1942) and Ryan and Gross (1951).
Yet, it became an entrenched worldwide standard even though manufacturers could have
produced a more efficient alternative.7 The realm of politics has also yielded numerous
examples of diffusion, such as the spread of pension systems across Europe, charter
schools in the United States, or even democracy around the world (for some recent
examples see Brinks & Coppedge 2006; Karch 2007; Mooney 2001; Orenstein 2003;
True & Mintrom 2001).
For social scientists interested in explaining the world around them, diffusion
events spark a plethora of research questions.
Some relate to innovations and the
attributes of innovation. What makes an idea “innovative”? Why do some ideas spread
while others do not? Why are some actors more innovative than others? Another line of
inquiry focuses on diffusion processes. How and why do innovations spread? Why are
some jurisdictions quick to adopt innovative practices while others are slow? What role
do policy entrepreneurs play in the spread of their ideas? Lastly, some scholars focus on
normative considerations such as functionality. Do actors adopt innovative practices
because they fulfill an identified need?
To what extent is replication logical or
“rational”? While each of these questions pursues distinct dimensions inherent in the
spread of innovations, altogether they contribute to a broad literature on diffusion.
While diffusion research has enjoyed great revival in the social sciences, the
diversity of questions and approaches has produced tremendous conceptual murkiness.
What is diffusion precisely? Scholars use such diverse metaphors as “contagion”,
The QWERTY keyboard was first developed for typewriters and was designed to limit jamming. It is
less efficient than an alternative keyboard designed by Professor August Dvorak in 1932. Although the
Dvorak keyboard is more efficient, it never caught on (Rogers 2003: 8-11).
“waves”, and “transfer” to describe diffusion events (see for instance, Burt 1987; Stone
1999; Walt 2000). But are they the same? This definitional question underscores the
need to clarify relationship between the actors and emulation decisions.
implies autonomous decision-making while also accounting for potential interdependence
between actors. Political actors may learn about an innovation from a neighboring
government or international meeting, and choose to copy those innovations in their own
countries. Sometimes the impetus to emulate can come from peer influence (horizontal),
reflect internal demands (bottom-up), or external pressures (vertical). These inducements
need not reflect benign forces but can include coercion or other stimuli for cooperation.
For a visual illustration of the potential influences driving diffusion effects, see Figure
Definition: This study broadly defines diffusion to include processes that affect the
likelihood that a reasonably autonomous jurisdiction will adopt an “innovative” policy
developed by another such unit, at some point in time. Various members of a social
system can trigger a unit’s emulation mechanisms and increase the probability of
adoption, including the originating jurisdiction that promotes the innovation, similar
decision-making units, as well as those who reside outside of the decision-making unit
(similar to Levi-Faur 2003: 23-26; Weyland 2007: 24-25).
Figure 1.1: Multiple Directions of Diffusion Effects
Vertical Effects
External Inducements
City A
City B
City C
City D
Horizontal Effects
Bottom Up
Source: Modified figure from Levi-Faur (2005:26)
Policy diffusion thus captures a process of decision making that involves some
degree of learning, emulation, or mimicry. If separate jurisdictions were to enact the
same policy because they share similar conditions (e.g. economic shocks, levels of
development, or institutional similarities) but were otherwise isolated from one another,
this would not constitute an instance of diffusion.
The key here is that diffusion
emphasizes the interactive process that leads to replication (i.e. the spread), not just the
outcome (i.e. policy adoption).8
The conceptualization of diffusion employed in this study is particularly relevant
for policy diffusion because the politics of policy making requires that actors advocate
for their positions when they choose to emulate an innovation within their jurisdictions.
Policy does not lend itself to “automatic” adoption because instituting new programs
generally requires the establishment of guidelines, new administrative procedures, and
budgetary commitments; all of which require decisive action.
Thus, this study
emphasizes processes as well as outcomes, including the mechanisms that drive adoption,
non-adoption, and reversal. Last, in classifying diffusion outcomes, I consider that a
policy has spread when a jurisdiction enacts a policy that shares a similar policy design as
the original. Some scholars distinguish between “emulation” and “adoption” as distinct
from one another; constructivist scholars tend to use “emulation” to emphasize the
importance of epistemic communities whereas “adoption” does not imply a similar
meaning. Since epistemic communities are central to theoretical questions on the role of
social networks, I make no prior assumptions here in selecting terminology to describe
diffusion outcomes. As such, I use emulation, adoption, and replication interchangeably.
Why would governments that face different problems, priorities, and political
conditions decide to adopt the same policy? Although policy diffusion research is new to
For a review of different treatments of the concept of diffusion – process versus outcome – see especially
Elkins and Simmons (2005:36-38).
Brazil and relatively under-explored in the developing world, scholars of American9 and
international politics have long taken up a diffusion lens to understand the spread of
public policies. Research has examined such disparate issues as abortion regulation, tax
reform, charter schools, pension systems, and women’s bureaucracies (Berry and Berry
1992; Mintrom and Vergari 1998; Mooney and Lee 1995; Orenstein 2003; True and
Mintrom 2001). In Latin America the spread of pension privatization and the shifting
economic agenda towards neoliberal economic reforms has been a prominent theme
(Brooks 2002; Madrid 2003; Weyland 2005). The popularity of diffusion as a scholarly
topic has only been matched by the potpourri of explanations, approaches, and theoretical
frameworks scholars have employed (for full review see Rogers 2003). Scholars have
attributed diffusion to such diverse factors as the nature of the policy itself, policy
entrepreneurship, geographic proximity, state competition, political opportunity, resource
competition, federal transfers, and transnational networks, to name just a few (Berry and
Berry 1992; Derthick 1970; Finnemore 1993; Gray 1973; Mintrom 1997; True and
Mintrom 2001).
In explaining their findings, scholars turn to both structural
explanations, such as federalism and geographic proximity, as well as agent-based
explanations, such as individuals’ ability to learn or “policy entrepreneurship” (Mintrom
1997; Mooney 2001; Rogers 2003; Walker 1969).
The result is that the literature
9 Institutional similarities between the United States and Brazil make it possible to draw on the American
Politics diffusion literature. Both countries share a similar federal structure. They also share presidentialism
and the executive control over the bureaucracy. The most substantial distinction between the two countries
is that Brazil, unlike the United States, explicitly recognizes the role of municipalities and creates a threetier federal government. Brazilian municipalities also have clear mandates to address social policy
provision in the constitution (1988).
provides a laundry-list of possible explanations for policy diffusion, without offering
theoretical coherency or a systematic approach for addressing the causes of diffusion.
Rather than test a host of internal prerequisites versus external pressures for
policy emulation, this dissertation takes an alternative approach to examining the
diffusion of social policy. It does so by opening up the “black box” of policymaking and
asking what motivates policymakers on the ground, to make emulation decisions.10 One
of this dissertation’s main contributions is that it offers a theoretical framework for
understanding the mechanisms that drive diffusion, by uncovering the actors’ motivations
for adopting social policies.
This dissertation contrasts and tests three approaches for understanding
individuals’ motivations for adopting innovative social policies: political incentives,
ideology, and socialized norms. Each of these three motivations reflects a different
paradigmatic response to the question, what drives diffusion? A conventional political
incentives approach typically follows a rational choice framework that incorporates
neoclassical behavioral assumptions and posits people will behave strategically to further
their own self-interest. The classic assumption in this vein is that politicians will seek to
win re-election. On the other hand, scholars who adopt an ideational approach examine
the way people make choices because of their ideological convictions. Rather than seek
As Rogers notes, most studies of diffusion have not asked such ‘why’ questions about actors’
motivations (2003: 115). Recent examples of researchers who are filling this void include (Finnemore
1998; Weyland 2007; Mossberger 2000).
their own political self-interest, actors can make decisions because of deeply held beliefs
about what is right and how to enact social change.
Sometimes these ideological
commitments can even lead actors to make decisions that counter their political selfinterest. Lastly, a sociological approach examines how individuals conform to shared
norms and seek legitimacy in the eyes of their colleagues. In this case, policy makers’
motivation to initiate change would be to demonstrate to their peers that they have keptup with their profession’s norms.
This study reveals that a political incentives approach offers a surprisingly weak
explanation for the diffusion of innovative social policies. Traditional incentives such as
electoral competition rarely explain variation across municipalities and policymakers’
decisions to implement programs like Bolsa Escola and Programa Saúde da Família. Nor
do political incentives explain the variation in the adoption over time. Rather, two
different but complementary approaches explain diffusion: ideology and socialized
professional norms. First, I find that ideology does serve as a foundation for many
individuals in guiding them to action and helping them filter their policy choices. In the
Brazilian case, policymakers who self-identify as being on the left or left-of-center are
consistently more eager to adopt innovative social policies. Second, socialized norms
matter as policy professionals reveal they want to demonstrate to their peer networks they
understand and follow new professional norms. Professional associations and informal
networks play a central role in shaping norms around policies such as Bolsa Escola,
Renda Mínima and Programa Saúde da Família.
The extent of policy diffusion is
determined by the density of professional associations; the more a sector has dense and
overlapping associations, the more likely a program will diffuse. In some instances,
ideology and networks can work together in mutually reinforcing ways by convincing
actors that policies are in-line with their ideological commitments.
This project relies on a mix of methodological approaches to capture the diffusion
of social policies in Brazil: it draws on surveys, interviews, and statistical analyses to
answer the question of what motivates policymakers to emulate social policies that are
designed for other cities.
An important feature of this study is the comparison of two social policies that are
situated in distinct sectors – education and health – and that diffuse at different rates. As
Rogers notes, one of the shortcomings of diffusion research is the propensity to study
“successful” instances of diffusion (2003: 110). In other words, scholars tend to have a
“pro-innovations bias” in focusing on policies that have spread dramatically across
jurisdictions. Another component of a pro-innovations bias is that researchers tend to
focus on policies they believe are “good” and should spread. This project aims to
alleviate some of this bias by selecting two markedly different programs. Although both
Bolsa Escola and PSF won “innovations” awards,11 as this dissertation will uncover,
these policies are not perceived as universally “good” or desirable for all jurisdictions.
Since awards for these programs are based on a single city’s experience, it is entirely
possible that the same program, when adopted elsewhere, can fail to address the
11 Cities that administer Bolsa Escola and PSF have won good governance awards from the Gestão Publica
e Cidadania program in São Paulo.
emulating city’s most pressing problems. As this study uncovers, some policy makers
endorsed these policies and believed they “should” spread while others disagreed.
There are numerous benefits to conducting a large-N event history analysis. First,
increasing the number of observations provides greater leverage for causal inference
(King et al. 1994). Second, an event history model, which involves annual observations
for each jurisdiction, addresses the problem of potential interdependence among
jurisdictions and thus allows for better analyses of internal and external determinants of
diffusion (Berry and Berry 1990, 1992; Collier and Messick 1975). Advanced statistical
methods, such as event history modeling, also allow for a probabilistic interpretation of
whether cities are likely to adopt innovative social policies.
In order to understand larger trends across Brazil and map the pattern of Brazilian
social policy diffusion, I use an event history analysis to statistically test the impact of
political incentives, ideology and social networks on diffusion for Brazil’s largest cities.
To conduct such an analysis, I created a database on social policies for all 224 cities that
had populations over 100,000 in the census year 2000. This original database draws on
information related to electoral politics, socio-demographic data, and social network
connectivity. It also includes information on the adoption patterns of Bolsa Escola,
Renda Mínima and Programa Saúde da Família. Due to limited data access and in order
to access the spread of Bolsa Escola and Renda Mínima, I administered a phone survey of
education and social welfare administrators for the entire sample. For a list of cities
included in the sample, see Appendix A.
The focus of this study is to uncover the motivations behind social policy
emulation decisions in Brazil. Yet, this task is a complicated one that requires multiple
approaches. First-order analysis can start from observable data and implications, such as
information on electoral competition, politicians’ partisan affiliation, and the presence of
professional networks in a given community.
With this information we can draw
inferences and make conclusions. Also important, however, is how actors themselves
interpret these programs and explain their own role in the decision-making process. As
Taylor notes, in order to assess meaning, we must pay attention to the stories people tell:
A person’s understanding of her own life, the story she tells (constructs and
reconstructs) about herself, which itself of course becomes part of her life,
endows events with meaning, with significance for us. For most of us want to see
things we have done and events in our lives as having some meaning (2006: 33).
As such, in order to assess whether an actor’s motivations for emulating an innovative
policy are driven by political self-interest, ideology, or socialized norms, we must ask
individuals to tell a story. Their narratives will frame the way they understand the event
and the meaning it held for them.
The case study research occurred between 2003 and 2004, when I conducted
interviews of one hundred twenty Brazilian policymakers involved in health and
education policymaking at the local level, elected officials, technocrats, community
activists, and leaders in non-governmental organizations. I also interviewed a select
number of policymakers at the federal level, who are in charge of setting the federal
policy agenda. (For a full list of interviewees, see Appendix B). During face-to-face
interviews, respondents discussed their motivations for adopting or advocating for Bolsa
Escola/Renda Mínima and Programa Saúde da Família, and reflected on the policy
process in general. In instances where local governments did not have the programs or
the programs were dismantled, we discussed why this was so. I conducted interviews in
four municipalities in Brazil: Belo Horizonte (in Minas Gerais state), Brasília (in the
Federal District), Salvador (in Bahia state), and São Paulo (in São Paulo State), and
captured actors involved with three municipal administrations, from 1994 to 2003. The
qualitative evidence allowed me to identify the mechanisms by which diffusion occurred,
by including thorough process tracing.
Table 1.1:
Adoption of BE/RM and PSF by Local Government and Administration
Bolsa Escola/ Renda
Programa Saúde da
Brasília (DF)*
Belo Horizonte (MG)
Salvador (BA)
São Paulo (SP)
* The Federal District, Brasília, operates under the gubernatorial electoral calendar.
** The program was suspended or discontinued and then reintroduced under new names.
Several criteria guided selection of the research sites. First, the four cities in this
study adopted Bolsa Escola and PSF at different points in time and in a few instances
even experienced policy reversal (see Table 1.1). The variation in program adoption over
time is important, because there would otherwise be a potential for selection bias (Geddes
1990, 2003; King et al. 1994: 129-137). Second, these municipalities were selected to
allow for variation in partisan politics.
These local governments had mayors who
affiliated with eight different political parties and represented ideological leanings from
across the political spectrum, from staunch rightists to leftists; no single party dominates
and all major political parties are represented (see Table 1.2 below). Third, the case
study cities also face different levels of socio-economic development and are
geographically dispersed. Fourth, in the context of health policy, these cities had great
flexibility in determining their basic health models. Not only did they enjoy fiscal
autonomy because of their potential to generate revenues through their local tax base, but
they also had a sophisticated health infrastructure, and had a large number of health
professionals. In other words, these municipalities had the administrative flexibility to
tailor health policy and emulation of PSF was far from automatic or a forgone
conclusion. Despite these important differences, these cities share characteristics that
make comparison possible; all are state capitals and face similar institutional tensions
between local, state and national-level policymaking. The only exception is Brasília,
which has a unique status as the federal district.
Table 1.2:
Local Government Partisan and Ideological Divide
Brasília (DF)b
Belo Horizonte (MG)
Salvador (BA)
São Paulo (SP)
Executive in Office
Joaquim Roriz
Cristovam Buarque
Joaquim Roriz
Patrus Ananias
Célio de Castro
Célio de Castro
Fernando Damata Pimentel (PT)c
Lídice da Mata
Antônio José Imbassahy
Antônio José Imbassahy
Paulo Maluf
Celso Pitta
Marta Suplicy
Mayor’s partisan affiliation at the time he or she ran for office.
The Federal District, Brasília, operates under the gubernatorial electoral calendar.
Fernando Damata Pimentel (PT) assumed office in November 2001, after Célio de Castro suffered a
In using both large-N statistical analyses and small-N case studies, this project
bridges two research traditions on diffusion. Scholars who frame their work along the
lines of “learning” and “policy transfer” typically focus on micro-processes, actors, and
employ qualitative methods (see for example Bennett 1991; Rose 1993; 2004). By
contrast, those who employ statistical analyses of broader diffusion phenomena typically
seek the leverage a large number of cases can offer for generalizability of causal analysis
(see for example Box-Steffensmeier and Jones 2004). Employing both methods enables
comparison between this study’s findings and those from existing research.
Mainwaring, Meneguello and Power (2000: 180) inform this designation. The PTR is not classified in
Coppedge (1996).
By drawing on both qualitative and quantitative studies, this work also overcomes
some of the shortcomings that single-method studies face (see especially Brady et. al.
2004). For instance, a correlate to diffusion can include such factors as “economic
development”. Yet, this variable alone cannot explain why a jurisdiction’s level of
development matters in terms of politics, only case analysis can explain the relationship
for meaningful inference. Statistical methods can also obscure causal heterogeneity in
emulation decisions, thus limiting understanding of the complex and varied mechanisms
that lead to policy emulation (Mahoney and Goertz 2006). Large-N statistical studies can
contribute to an under-accounting of ‘causal complexity’ by making interpretation of the
underlying relationship between indicators and concepts difficult (Meseguer and Gilardi
2005). Thus, qualitative research can help uncover both the mechanisms that drive
diffusion and also clarify whether jurisdictions undergo similar causal processes. Lastly,
an important analytical advantage to using both methods includes the ability to examine
the politics of not only policy emulation, but also explore why policy reversals occur.
Decentralization in Brazil has contributed to a new era of social policy reform
where policy diffusion is increasingly prevalent. To date however, scholars of Brazilian
politics have largely ignored diffusion as either an outcome or theoretical framework for
understanding significant changes in the policymaking process. Instead, country-specific
studies of diffusion remain largely relegated to analyses of advanced industrialized
nations with federal structures, especially the U.S.
For this reason, one of the
contributions of this study includes the application of a diffusion approach to understand
the politics of a developing democratic country.
Research on Brazil also creates an opportunity to advance the existing literature
on diffusion, which has generally underspecified the motivational factors that drive
politicians to emulate innovative policies. This dissertation’s main contribution is that it
furthers previous understandings of diffusion with a new conceptual framework that
focuses on individuals’ motivations to replicate policy models developed in other
settings. Through a mixed-method approach, it is possible to identify how political
incentives, ideology, and social networks affect the decision-making process. By
exploring contrasting motivations for political action, this work assesses three theoretical
approaches that are often examined in isolation from one another.
This brief overview of the dissertation serves as a road map for the rest of this
work, and each of the subsequent chapters will address the questions first introduced
here. The next chapter will provide a more thorough discussion of public policy making
and the conventional approaches used to understand policy choices. While traditional
approaches, such as policy studies and historical institutionalism, are useful for
understanding the particularities of policy arenas and the structural constraints for social
sector reform, I argue that the diffusion lens is the most appropriate for explaining the
dynamic policy changes taking place in Brazil. As such, the chapter offers a framework
for examining social policy emulation across municipalities, which focuses on actors’
motivations to replicate policies designed for other cities.
Chapter 3 provides a bird’s-eye view of the diffusion of Bolsa Escola/Renda
Mínima and Programa Saúde da Família, though an analysis of these programs throught
the country. The chapter draws primarily on survey data collected for the country’s
largest cities. After contextualizing the trends in Bolsa Escola and PSF, it presents an
event history model, a discrete time logit model to statistically test theoretically driven
variables related to political incentives, ideology, and social networks.
One of the
benefits to the event history analysis is that it can facilitate interpretation of time and
probability of adoption.
Chapter 4 on Bolsa Escola/Renda Mínima provides an in-depth look at the politics
of education reform in Brazil. The chapter situates locally driven innovations to address
educational access and attainment in the context of stalled efforts for reform through
much of the late 1980s and 1990s at the national level.
The chapter explains the
emergence of conditional cash transfer programs for education, such as Bolsa Escola and
Renda Mínima, and the politics of their spread across the country. To uncover the
motivations that that lead to emulation decisions, this chapter draws primarily from the
in-depth qualitative analysis of policymakers in the four case study sites and uncovers the
ways in which political incentives, ideology, and professional norms influenced
policymakers’ adoption decisions.
The following chapter, on Programa Saúde da Família, discusses the emergence
of the integrated family health program within the context of the significant
reorganization of health policy in the 1990s. Unlike the education sector, national health
policy made greater advances in reforming the sector and the federal government
promoted decentralization with the municipalization of services. As local governments
took on greater administrative responsibilities, many political actors implemented the
PSF policy. Chapter 5 focuses on actors’ decision-making processes and the extent to
which they were responding to political, ideological, or socialized norms. Accounts from
key actors responsible for health care policy reveal how technocrats’ connections to
professional associations and the presence of leftist majors drove replication in this policy
Lastly, chapter 6 concludes by comparing and contrasting the diffusion of Bolsa
Escola and Programa Saúde da Família. Both the large-N and process tracing methods
offer a strikingly consistent account for what drives emulation decisions. Together, they
also provide insights into key differences between each of the policy arenas, and the
effects that the particular effects they have on diffusion. In addition to addressing
contrasts between actors’ interpretations of these policies, the chapter also underscores
how the breadth and depth of networks and informal associations influences the
socialization of professional norms. Finally, it concludes the dissertation by examining
the ways in which ideology and socialized norms work together to promote policy
emulation and the implications of these findings for understanding Brazilian
policymaking and diffusion more generally.
Brazil faces the tremendous challenge of addressing long-standing social
disparities, high levels of poverty, and economic inequality. Though many scholars have
increasingly turned their attention to determinants of social sector reform across the
region (Grindle 2000, 2004; Kaufman & Nelson 2004; Tulchin & Garland 2000), we
know very little about the politics of education and health reform at the sub-national
level, where states and municipalities have considerable authority to develop and
implement social programs (Kaufman & Nelson 2004:464). Brazilian municipalities in
particular have been at the forefront of social policy experimentation and innovation, yet
the factors that drive the spread of social programs across the country remain unknown.
The case of Brazil is especially compelling because although the federal government has
set guidelines and directives in a host of social policy arenas, there is considerable
variation in local governments’ willingness to follow them. Even more importantly,
many sub-national governments, state and municipal, have jumped to the forefront in
developing new programs when national-level policy reforms stall. Although programs
have spread across the country over the last twenty years, most research has not tackled
the spread of social policies across Brazil’s local governments.
Research on the determinants of public policy has been integral to several
subfields of political science, including American politics, policy studies, comparative
politics, and Brazilian studies.
This chapter draws on these scholarly traditions to
uncover both their contributions and limitations to explaining the spread of social policies
across Brazil. The first section briefly addresses the predominant modes of analysis for
examining social policy development, including policy studies and historical
institutionalism. While these modes offer useful insights on the nature of policymaking
and the limits of decentralization, both display conceptual weaknesses for a study of
social policy in Brazil; policy studies tend to obscure the similarities across policy
domains and institutionalism under emphasizes local jurisdictional autonomy. For these
reasons, in the second section I introduce an alternative analytic lens, policy diffusion, to
understand the replications of social policies. The last section draws on the existing
diffusion literature to introduce the theoretical approach of this study. This dissertation
employs an actor-oriented framework to uncover the motivations that lead actors to
initiate social policy changes.
I do this by drawing on three analytically distinct
approaches for explaining actors’ political behavior: political self-interest, ideology, and
social networks.
Questions about what governments do are at the heart of the political science
discipline; they touch on issues such as the role of the state, the problems for collective
action, and the politics of competing groups’ claims. Given the widespread relevance of
policy outcomes to the discipline, it is not surprising that public policy topics permeate
numerous sub-fields and that scholarship reflects varied theoretical approaches. Among
scholars of American politics, policy studies emerged as its own sub-field in the 1960s
and 1970s when scholars challenged to address the pressing social problems of the day.
As such, much of theory on public policy is based on American politics, its institutions,
election cycles, and social structure. These include an emphasis on the strong influence
of interest groups and elites in the U.S. political system. Scholars of comparative politics
have been similarly pressed to explain state-society relations and redistribution; including
explanations for such phenomena as economic development, the emergence of the
welfare state, and social and economic reforms. In doing so, comparativists have also
tackled policy outcomes, albeit often from different analytic perspectives. This section
underscores both the analytic contributions and limitations of policy studies and
comparative institutionalism for the study of social policy diffusion.
Scholarship on the politics of public policies and theories of the policy process
more generally, have tended to emphasize the unique features of each policy issue rather
than draw similarities to build general theory. Policy scholars have followed the dictum
that not all policies are created equal; policy content matters in the politics of policy
making. Whether a policy issue is divisive, symbolic, or a response to a crisis can all
matter in how well it is received (Gray 1973; Sabatier 1999; Sabatier & Jenkins-Smith
1993; Stone 1989). For instance, policies that are purely symbolic, non-controversial,
and cost little to nothing (e.g. a resolution in support of “children’s day”) can be easy to
approve and implement. By contrast, those policies that are contentious (e.g. a plan for
stricter gun control) or costly (e.g. an increase in healthcare benefits) can draw out bitter
disputes that gridlock the policy process.
Lowi’s (1964) seminal classification of public policies combined an analysis of
policy content with their associated costs to explain why some are easier or harder to
pursue. The heart of his analysis lies in the tensions between long-term policy goals and
actors’ shorter-term political goals, contributes to his influential typology of public
policies. Lowi classifies policies into one of three categories: distributive, regulatory,
and redistributive.1
Social sector policies, or “welfare state” programs, fall in the
redistributive category and are most politically contentious because their impacts are
broad and so clearly distinguish between the “haves and have-nots.” Redistributive
policies open the door to interest-group politics and so the logic holds, the extent to
which goods are spread-out among the public.
Lowi’s framework not only helps
compare different types of policy, such as regulartory versus distributive, but also
assumes that in an electroral democracy with strong interest group politics, certain
polices will be difficult to pursue.
In this logic, the prospect for social policy reform
would reflect the nature of redistribution and interest group politics. Many researchers
have either explicitly or implicitly adopted Lowi’s framework when examining the
politics of social policy in Latin America and Brazil.
Although all welfare policies can be broadly classified as redistributive many
scholars have gone further to disaggregate their study of social policies by focusing on
particularities of each sector. Castro and Musgrove (1998), for instance, argue that there
1“Distributive policies are characterized by the ease with which they can be disaggregated and dispensed
unit by unit, each unit more or less in isolation from other units and from any general rule. ‘Patronage’ in
the fullest meaning of the word can be taken as a synonym for ‘distributive’…“Regulatory policies are also
specific and individual in their impact, but they are not capable of the most infinite amounts of
disaggregation…the impact of regulatory decisions is clearly one of directly raising costs and/or reducing
or expanding the alternatives of private individuals…regulatory decisions are usually disaggregable only
down to the sector level”… “Redistributive policies are like regulatory policies in the sense that relations
among broad categories of private individuals are involved and, hence, individual decisions must be
interrelated…The categories of impact are [however] much broader, approaching social classes” (Lowi,
1964 as reprinted in Theodolou and Cahn, Eds. 1995: 15-16).
is no “social sector” in Latin America and that comparisons across education and health
are problematic because of the different constituencies and politics of each sector.
Similarly, Corrales (1999) draws on a constituency-demand approach to address the
particular challenges of education reform and notes that other social reforms that have
broader constituencies are more likely to succeed when compared to the more narrow
audience for education. These underlying assumptions are among the reason that a great
deal of scholarship on social sector reform in the Americas is sector-specific, examining
issues such as social security, health, and education (primary vs. secondary) in relative
isolation (Arretche & Marques 2002; Birdsall & Sabot 1996; Corrales 1999; Mesa-Lago
In recent years the growth of practitioner-based knowledge and policy
evaluations has also contributed to the trend for sector specific inquiry in areas such as
health policy, education, and social security.3
The result is that much of the
contemporary policy scholarship on Brazil emphasizes specificity and difference rather
than identify similarities across policy domains.
Trends in sector specific approaches to policy studies however, may imbed too
many assumptions that do not translate well for a study of Brazilian public policy. First,
much of Lowi’s framework draws on underlying assumptions about the effects of elites
and interest groups on policy outcomes, but considerations may be less applicable outside
of the American context. Do elites and interest groups behave similarly across all
country contexts? Do actors, either individuals or groups, always assert their interests
2 A few notable exceptions include Weyland (1996) and Tendler (1997).
3 Numerous practitioners and policy scholars conduct specialized policy evaluations, such as Ricardo Paes
de Barros, Amélia Cohn, Sônia Draibe, Emerson Elias Merhy, and Sônia Rocha, just to name a few.
according to their economic position? Even if we assume that in principle redistributive
policies will be rarely enacted because they conflict with elite interests, this logic might
not hold for the Brazilian case. Historically, during both periods of authoritarianism and
limited democracy, politicians have relied on populist appeals and policies to garner
support from the poor and working class.
Starting in the 1930s, the Vargas
administration passed numerous laws to protect workers while at the same time bringing
union organizing into line (Fausto 1999: 200-202)4. During this period, the federal
government also introduced numerous laws and policies with mass appeal; among them
was the emergence of a corporatist welfare state that favored urban workers, including
social security and free and compulsory primary education.
Although broad
redistribution was not the “goal” of the emerging welfare state in Brazil, in effect, it did
offer wide protections for the working class that remain until today.5 Given that the most
notable social programs developed by the Brazilian government have favored the
working and middle classes over the poor, the implementation of education and health
programs might not be as divisive as Lowi would predict. Second, unlike the United
States where poor and marginalized groups tend to vote less when compared to their
middle and upper income counterparts, Brazil has mandatory voting6 for all. Thus in
4 The Vargas’ administration’s main objectives, both during the early 1930 and later under the Estado
Novo (1937), were to promote industrialization without causing major social upheavals (Fausto 1999: 217).
Vargas accomplished this by striking a balance, for instance, by legalizing unions but repressing efforts of
the urban working class to organize outside of the aegis of the state (Fausto 1999: 198-202).
5 The Brazilian welfare system offers the greatest social protections to the middle and working class. The
poor, who tend to work in the informal sector or in rural production, were left out of the social security
system for decades.
6 All Brazilians over eighteen years are required to register and appear at the polls on Election Day.
However, voters need not actually cast a valid ballot; they have the option of nullifying their ballot or
theory, elected officials would be more responsive and electorally accountable to Brazil’s
sizable poor population.7 Populist pro-poor appeals remain an important feature of local
politics as some elected officials utilize social programs to garner political support for
their electoral campaign. A few other politicians openly embrace patronage politics and
vote-buying approaches through pre-election “giveaways” of land, refrigerators, kitchen
appliances, and foodstuffs.8 These dimensions of local politics together with the historic
patterns of social welfare policy making, raise doubts to the applicability of Lowi’s
assumptions about the politics of redistribution for Brazil.
Unlike the interest based framework for understanding policy outcomes,
comparative scholars and Braziliansists have framed their understanding of policy
outcomes within the context of institutions,9 both formal organizations and informal rules
and procedures. “[I]n general, institutionalists are interested in the whole range of state
and societal institutions that shape how political actors define their interests and that
structure their relations of power to other groups” (Thelen & Steinmo 1992: 2).
Generally, institutions are not part of the “cause” of different outcomes, but researchers
leaving their ballots “blank”. Registration and voting is optional for those who are illiterate and between
sixteen to eighteen years of age (Constitution, II, IV, 14). Citizens incur a small fine for failing to appear at
the polls.
7 In 2005, 30.6 percent of the population feel had family incomes per capita that feel below the poverty line
8 Accounts of this type were frequently brought up by political opponents of Governor Joaquim Roriz (DF)
during interviews in Brasília.
9 As Thelen and Steinmo acknowledge, the study of institutions has been a longstanding enterprise with
contributions from such scholars as Karl Polanyi, Thorstein Veblen, Max Weber in fields such as political
science, economics, and sociology (1992: 3). The distinction is that “new” institutionalism is a response to
the behavioral revolution in the 1950s and 60s in political science, and draws attention to the “enduring
socioeconomic and political structures that mold behavior in distinctive ways in different national contexts”
(1992: 1). This category does not include the rational choice variant, exemplified by the works of Douglass
North and Robert Bates.
look to them to see how they constrain and shape political outcomes. In doing so, this
approach seeks a mid-range analysis that can bridge macro- and micro-level variables.10
As institutions are normally stable over time, this approach is useful in clarifying the
differences across countries (e.g. democracy, emergence of welfare states, etc.).
Historical institutionalist approaches are particularly well-represented in Brazilian
scholarship as scholars draw on such diverse factors as Brazil’s social structure, historical
legacies, the country’s political development and various policy outcomes (see for
example, Abrucio 2002; Affonso & Silva 1996; Arretche 2000; Draibe 1985, 1994;
Fernandes 2004).
This approach has led scholars to identify the ways that critical
junctures, such as Getúlio Vargas’ Estado Novo (1930s) or the transition to democracy
(1970s and 80s), have led the way to explain numerous features of Brazilian politics,
including decentralization, the limits for social security reform, and the emergence of
participatory governance practices.
The transition to democracy in Brazil created new institutional challenges for
shared governance in the federal system. Many scholars have undertaken research that
explores the successes and failures of decentralization; Marta Arretche’s (2000) work is
exemplary of this scholarly agenda. She examines the extent to which decentralization of
social programs has taken place following the 1988 constitution and focuses on the extent
to which state and local capacity, such as administrative and fiscal capacity, institutional
rules, and inter-governmental relations, have shaped decentralization. Her work draws on
10 Marx, Weber, and Moore are exemplars of the macro-structural approach. For more recent examples,
see (Collier & Collier 2002; Skocpol 1979).
cases from various sectors, including urban development, primary education, health, and
social assistance with measures of decentralization based on compliance with federal
mandates for municipal adoption. In education, she examines the extent to which state
and local governments were able to follow with the requirement they “supply” primary
education and adopt a school lunch program; in the health sector, she examines whether
municipalities qualified for the Sistema Único de Saúde (Unified Health System) and
medical services. Ultimately, she finds that local policymakers11 assess the costs and
benefits of taking on new responsibilities and only do so when they have the
administrative and financial capacity.
Institutionalist approaches to decentralization in Brazil, such as Arretche (2000),
have certainly provided valuable insights on the politics of reform and policy
administration. Nevertheless, there are important differences of viewpoint between those
works and the present thesis’ research focus. First, decentralization topics have tended to
focus on inter-governmental relations, state-level politics, and the extent to which subnational governments comply with federal mandates. While the federal government
pushed for certain institutional reforms and programs, it is also the case that several
“bottom-up” initiatives from municipalities competed for attention in the local policy
arena, and these initiatives were often dismissed as illustrative of decentralization.
Second, the focus on following “mandates” misses an important feature of
decentralization being that local governments also have the flexibility to experiment,
11 Arretche examines six states in her work, Rio Grande do Sul, Paraná, São Paulo, Bahia, Pernambuco
and Ceará.
tailor, and occasionally deviate from federal policy. For instance, in the 1990s several
local governments pursued alternative healthcare models that diverge from the policy
approach of the federal government: Niterói (in Rio de Janeiro state) and Londrina in
(Paraná state) both undertook a Programa Médico de Família (Family Doctor Program)
imported from Cuba and São Paulo adopted a semi-privatized model called Plano de
Atendimento à Saúde (PAS).12 Similarly, Bolsa Escola and Renda Mínima programs
were initiated at the local level and began spreading without much attention or support
from the federal government. These innovations and their spread are important features
of Brazil’s new policy making environment. In sum, a focus on intergovernmental
relationships tends to emphasize the role of the federal government and vertical pressures
for reform, while potentially under estimating the importance of horizontal processes
such as learning from peer networks at the municipal level.
The relationship between federal support for social policy enactment and
municipal decision making opens up questions that have both empirical and theoretical
On the one hand, the country’s strong tradition of intergovernmental
cooperation (Samuels 2000) suggests that the spread of these programs may simply
reflect top-down directives from the federal government. On the other hand, Brazilian
fiscal and political federalism allows for municipal policy making independence; thus
diffusion decisions may reflect “horizontal” peer emulation, rather than “vertical”
12 From 1995-2000, São Paulo designed and implemented the PAS, a semi-privatized system of healthcare
for its municipal residents. This system was a radical departure from the municipal decentralization plans
promoted by the federal Ministry of Health. As a result, the city eschewed federal funds and approval.
Although numerous health policy experts asserted that São Paulo was failing to “decentralize” at that time,
Paulo Elias argues that the PAS was indeed illustrative of decentralization and municipal control over
health services (Interview 2003).
Researchers often privilege either “vertical” or “horizontal” explanations
when accounting for diffusion (for discussion on vertical versus horizontal diffusion, see
Elkins & Simmons 2005: 35; Levi-Faur 2005: 25-27; Weyland 2005: 268-271). In the
case of Brazilian social policy making, the relationship between federal inducements for
policy adoption and local policy making is far from clear, particularly for the largest
cities which can raise revenue and resist federal pressures. This study does not assume
that the replication of social policies across Brazilian municipalities is a simple function
of bottom-up pressures, horizontal or vertical processes. Rather, this work tests various
causal mechanisms that consider the possibility that emulation can be a function of all
Both policy studies and institutionalist approaches provide a foundation for
examining the politics of policy making. Policy studies highlight the ways that a policy’s
distinct characteristics affect its likelihood of being enacted, while institutionalist
approaches provide important insights on intergovernmental competition and the costs of
reform for local governments. Yet, both these approaches have analytic shortcomings
that are addressed in this study. First, this project aims to reach beyond particularistic
analysis of specific policy domains to uncover underlying similarities despite differences.
Second, by examining the replication of similar policies across Brazilian municipalities,
this study highlights the autonomous role of local governments and the potential for
uncoordinated horizontal learning in policy making. Policy diffusion accounts for both
these dimensions, as the following section explains.
Diffusion research dates back to the late 1960s and 1970s, when scholars such as
Walker (1969) and Collier and Messick (1975) noted that the study of policy trends and
their adoption needed to address the methodological problem of interdependence.
Specifically, researchers faced Galton’s problem, “that the findings based on the analysis
of causal relation within nations (or other units of analysis) may be distorted by the effect
of diffusion” (Collier and Messick 1975: 1300).
These early scholars posited that
research on the spread of policies could not simply examine “prerequisite” explanations
for policy decisions, e.g. levels of economic modernization or human capital, as tended to
be the case with more policy studies research. Rather, diffusion required an analysis of
external variables that could affect internal processes, such as spatial proximity to
innovative jurisdictions. In other words, diffusion research needed to account for policy
By acknowledging that jurisdictions are not isolated in their
decision-making processes, these early scholars highlighted the need to incorporate
analyses of both internal and external pressures for diffusion.
Although early research by Walker (1969), Grey (1973), and Collier and Messick
(1975) created significant interest among scholars for diffusion research, this area of
inquiry lay largely dormant until the 1990s. Part of the renewed interest in diffusion
research relates to worldwide political events, including the waves of democratization,
13 For greater elaboration on this issue, see Collier and Messick (1975) and Box-Steffensmeier and Jones
economic liberalization, and pension privatization (Brinks & Coppedge 2006; Madrid
2003; Simmons & Elkins 2004; Weyland 2004). Cross-national replication of state
agencies, such as science ministries and women’s bureaucracies, has been notable
because their implementation has occurred across a diverse set of countries (Finnemore
1993; 1996; True and Mintrom 2001). European integration has led to increasing interest
in policy transfer and learning across the continent (Bennett 1991; Rose 1993; 2004).
Recent trends toward devolution in the United States has also led scholars of state politics
to examine such diverse issues as the diffusion of charter schools, welfare reform, and
health maintenance organization regulation (Balla 2001; Karch 2007; Mintrom 1997;
Mintrom & Vergari 1998; Mooney 2001).
An important contributor to the resurgence of scholarship on diffusion includes
methodological innovations that allow for better analysis of the internal and external
pressures for policy diffusion in a large number of cases. Berry and Berry were among
the first diffusion scholars to apply event history analysis,14 a discrete time logistic
regression, which tests both internally and externally driven explanations (1990; 1992).
Many recent studies of policy diffusion have drawn on this statistical technique to test
various explanations for diffusion related to internal prerequisites and external influences
(Brinks & Coppedge 2006; Mooney 2001; True & Mintrom 2001). Event history models
can also account for different theoretical explanations for policy diffusion, including the
relative contributions of structural factors (e.g. intergovernmental relations) as well as
14 Event history analysis is also referred to as “survival analysis” because it allows for assessment of “risk”
or “failure” that an event will take place over a given period of time.
actor-orientated explanations (learning and membership in networks). Scholars have
examined such diverse factors as social norms, learning, competition, partisan ideology,
and geography. Most importantly, when diffusion research also relies on qualitative
methods, it allows for a more nuanced analysis of micro-processes that lead to emulation
Even though scholarship on policy diffusion is well established among scholars of
transnational politics and U.S. state politics, diffusion research has not been readily
applied to federal developing countries and is relatively new in the study of Brazilian
politics.15 Institutional similarities between Brazil and the United States make it possible
to draw lessons from the extensive literature on diffusion in the fifty states,16 where
scholars have identified a potpourri of factors that drive policy diffusion (see Rogers
2003 for a full review).
The existing literature offers a valuable starting point for examining potential
factors that explain diffusion in Brazil. Yet at the same time, there is a need to address
two shortcomings of standard diffusion approaches, one conceptual and the other
methodological. First, diffusion research has not tackled the question of what motivates
actors’ adoption decision. As Rogers notes, the “why” question about adoption is seldom
addressed by researchers, in part because of the difficulties in gathering information;
thus, researchers tend to assume motivations are economically driven and over rely on
15 To date, there are no studies of diffusion within Brazil or a developing country.
16 Both Brazil and the United States share a similar federal structure. Both countries also share
presidentialism and the executive control over the bureaucracy. The most substantial distinction between
the two countries is that Brazil, unlike the United States, explicitly recognizes the role of municipalities for
governance and social policy provision in its constitution (1988).
models that are rationalistic. In practice, this often means researchers assume adoption of
innovation is rational17 whereas those actors who reject innovations are misguided. Not
only does this type of classification lead to a pro-innovations bias, but it also fails to
account for people’s perceptions and their decision-making processes, which can be
idiosyncratic (Rogers 2003: 115-116).
While the efficiency bias is certainly pervasive throughout the diffusion literature,
the question of actors’ perceptions and motivations is still largely unanswered.
Researchers often incorporate their discipline’s approaches or their field’s dominant
paradigms and draw on their implicit assumptions about behavior. Political scientists
who study diffusion often ground their explanations in notions of rationality,18 electoral
competition, and economic competition (for example see Berry & Berry 1992; Walker
1969). In contrast, sociologists tend to examine learning in terms of the strength of
relational ties and organizational networks (for example see Berry & Berry 1992;
DiMaggio & Powell 1983; Granovetter 1973; Walker 1969). Underlying assumptions
about political behavior and the mechanisms that facilitate diffusion can be obscured by
scholars’ embedded assumptions.
Second, the most recent diffusion research has tended to draw on large-N
statistical analyses to test for the determinants of policy replication. The introduction of
statistical techniques such as event history analysis does have the advantage of capturing
17 Rogers uses rational in relation of efficiency; e.g. “the most effective means to reach a given end”
18 Walker uses rationality in terms of competition but also as bounded rationality where decision-makers
use cues and short-cuts to make decisions. Rather than seek all the information possible, individuals who
operate on limited time seek to satisfice. Ultimately, Walker argues that researchers should aim to predict
this behavior (1969:889).
both internal prerequisites and external pressures for policy diffusion (Berry & Berry
1992, 1999; Box-Steffensmeier & Jones 2004). Yet, even when statistical analyses offer
insights into diffusion, Meseguer and Gilardi note, the reliance on this statistical method
can contribute to an under-accounting of ‘causal complexity’ and difficulties interpreting
the relationship between indicators and concepts (2005). One solution is to elaborate on
event history models by addressing the causal mechanisms for diffusion through
qualitative interpretation and process tracing.
Resolution of these methodological
dilemmas is crucial because the problems associated with causal complexity have
important implications for theory building and generalization. For this reason, diffusion
research should draw on multiple methods, including qualitative analysis of the policy
process.19 Doing so will not only uncover potential problems of causal heterogeneity but
also lead to better causal explanation (for a useful discussion on the limits of statistical
analysis, see Brady 2004).
This dissertation seeks to uncover the motivating factors that lead actors to adopt
programs designed for other cities. To do this, it assesses the contributions that three
broad approaches, which highlight different motivations for political behavior, can make
to explaining emulation decisions. Do individuals make decisions based on rational selfinterest calculations? Or, do they make choices based on their ideological values and
beliefs, even when faced with the prospect of political costs?
Alternatively, do
19 For a particularly notable example of research that incorporates process tracing and includes discussion
of the policy process, see Kingdon (1995).
policymakers act because they are socialized into a community that defines and transmits
shared norms?
These three questions relate to fundamental issues of whether
policymakers act in a purely self-regarding way; in a principled way, regardless of self or
others; or in an other-regarding, community-oriented way. By framing the motivations
that drive political action into three distinct categories – individual political self-interest,
abstract ideology, and legitimation before social networks – I intend to clarify how
individual behavior drives diffusion.20
Conventional rational choice approaches suggest that in an electorally competitive
environment, policymakers are driven by self-interest or political incentives as they seek
to maintain and increase their political power. In this view, purely self-regarding21
instrumental rationality plays a fundamental role in regulating behavior as individuals
seek to realize their goals (Downs 1957; Riker & Ordeshook 1973: 11).22 Applications of
these principles have led to models where politicians make choices to maximize political
support, typically because they desire to win re-election, win a more competitive office in
the next election, or simply retain their partisan appointments.
20 As Elkins and Simmons note, research on diffusion has taken on numerous definitions each of which
has distinct underlying assumptions about decision-makers’ behavior. This study employs their definition
of diffusion, which considers governments as independent in terms of decision making but allows for
uncoordinated interdependence. This definition allows for the testing of numerous causal mechanisms that
drive diffusion (2005:25).
21 Anthony Downs’ (1957) seminal work, which applies economic theory to explain political competition
and government action, established many of the rationality assumptions which are later adopted by his
successors. His model follows the axiom that “every individual, though rational, is also selfish… [t]hus
whenever we speak of rational behavior, we always mean rational behavior directed primarily toward
selfish end (1957:27).
22 Riker and Ordeshook make clear that their notions about rationality play a fundamental role in social
science, and provide the regularity for generalization (1973:11).
Scholars draw clear linkages between the effects of rational calculus and public
The expectation is that politicians behave strategically and choose
policies after having assessed the political costs and benefits of various alternatives;
particular policies are simply means for attaining political power (Carmines & Stimson
1993; Downs 1957). In the context of diffusion, Walker (1969) argues that when stiff
party competition exists, there is an increased propensity for parties to initiate change and
propose new programs in order to distinguish themselves. Lowi also argues that new
programs are more likely to be instituted in the beginning of a new administration (1963).
Thus, electoral competition and frequent executive turnover contribute to the adoption of
new policy, including emulation of external policy models.
Berry and Berry have
extended this understanding of electoral cycles and competition in their diffusion studies,
embedding assumptions about politicians’ rational calculus in the context of elections and
politicians’ desire to win reelection despite enacting politically unpopular tax policy
(1992). They suggest for instance, that politicians time the enactment of policies based
on the electoral cycle; politically unpopular policies might be enacted early in order to
give the electorate time to forget the policy come election time (1992: 719). Similiarly,
politicians might enact popular policies just prior to an election in order to pander to the
All in all, these scholars assume that politicians make strategic decisions
regarding policy enactment and replications, with the ultimate goal being winning
Another, altogether different explanation for diffusion is that policymakers are
driven by their ideology and emulate policies irrespective of political incentives.
Douglass North argues that it is important to consider the role of ideology in accounting
for the allocation of resources because not all individual behavior can be explained
through neoclassical behavioral assumptions alone (1981:46-47; 1990: Ch. 5).23 In other
words, actors may make seemingly nonsensical and other-regarding choices that deviate
from rational choice explanations but that are driven by principled commitments. In the
context of electoral politics this could include a politician’s supporting a public policy
even when confronted with the significant political costs of doing so.
Ideology, while one of the most centrally important explanations used in the
social sciences, remains one of the most contested and vilified for its conceptual
murkiness24. The term has been used inconsistently among social science scholars,
broadly falling into two categories: ideology of knowledge (e.g. ideological doctrine) and
ideology of politics (e.g. ideological mentality) (Sartori 1969: 398). While operating
within this second category, North adds to the confusion by locating “ideology” as any
seemingly “irrational” behavior that conflicts with neoclassical economic behavioral
expectations. But this conceptualization is also unclear; ideology is not the same thing as
irrationality nor is irrationality a “motivation” for action. As this study seeks to contrast
distinct research traditions with differing views of what drives individuals’ motivations in
policymaking, I locate ideology in terms of decisions driven by abstract maxims
regardless of self or others.
23 North’s assertion is particularly remarkable as he is widely known for his application of neo-classical
economic principles for a theory of state and institution building; he shares a Nobel Prize with Robert W.
Fogel for their work in the field, New Economic History.
24 See especially, Sartori (1969).
In this work, ideology is understood as “a pattern of thoughts and beliefs
explaining each person’s attitude toward life and their existence in society, and
advocating a conduct and action pattern that is responsive to such thoughts and beliefs”
(Lowenstein 1953:52 as cited in Gerring 1997:958).25
The key for a study of
policymaking is that ideology can compel individuals into action by providing both
exigency and grounds for political activity.
As Mullins argues, one of the key
components of ideology is its action-orientation in policymaking; political ideology
provides actors with a “relatively structured and consistent conception of the causal
forces operating in the social world, it also incorporates evaluation of what is conceived”
(1972:508). Thus, not only does ideology structure people’s worldviews, but it also
shapes one’s interpretation of the consequence of action and non-action.
Social policies are often value-laden in politics; they require that politicians
prioritize certain groups or make difficult decisions about the distribution of their costs.
To uncover the potential effect of ideology for policy emulation, it is necessary to
understand to what extent political actors are driven by their ideological commitments.
Ideology likely works in two complementary ways: First, actors must display their own
values and then assess the extent to which policies like Bolsa Escola and PSF fit their
worldviews. There are observable implications for the role that ideology can play in
emulation decisions. Since these policies target the needy, enhance equity, and alleviate
25 This definition of ideology is a substantively different from Down’s view, which argues the
development of ideologies are a means to political power by social classes or groups, rather than a
representation of actual goals (1957: 96). Down’s argues that in the American context, ideologies represent
cues or short-cuts for voters who face uncertainty and information costs. Ultimately, ideologies are
functional in that they are used by parties to obtain votes.
poverty, we might expect that those politicians with strong left-of-center beliefs would be
more likely to emulate these programs. Political actors’ own narratives about their
ideological beliefs and the meaning26 they ascribe to these social policies can confirm the
extent to which a left-of-center self-identification drove policy emulation.
Sociological approaches, alternatively, suggest that change occurs as a function of
social context and relations to others. The premise is that human behavior is embedded
in a matrix of organizational and informal relationships that provide fundamental filters
through which preferences are formed (Kaufman 1999:367-368). Networks in particular
can play a crucial role in linking individuals with others, structuring meaning, and
defining individual perceptions and preferences (Friedkin 1993; Kilduff & Tsai 2003;
Passy 2003). Professional associational networks in particular, can define the scope of
legitimate action and structure values for “modern” administrative practices. Thus,
networks offer cognitive short-cuts and social cues, guiding policymaking (Walker 1969).
Both formal and informal social networks can play influential roles in
policymaking. Formal organizations, such as professional associations, link individuals
with structurally equivalent roles who reside in different organizations but nevertheless
pressure each other to behave in similar ways (Friedkin 1988:69-70 as cited in Kilduff &
Tsai 2003:58). Informal networks can also exist among individuals or across
geographical space as “neighborhood effects,” where social learning and information
exchange travel spatially (Collier & Messick 1975; Granovetter 1973, 1983; Mooney &
26 Michael Taylor (2006) offers an interesting analysis that contrasts incentives-based rational choice
explanations for human behavior with a more holistic approach to understanding people’s decisions, that
emphasizes meaning, feelings of connectivity, and narratives; see especially chapters 2 and 3.
Lee 1995; Walker 1969).27 For example, newspapers can have regional circulation or
neighboring city administrators can periodically meet to discuss common problems. The
more actors are connected through informal and professional associations, the more likely
they share similar values, norms, and discourse. Thus, we could expect that professional
norms conveyed by social networks would spur policy diffusion.
A social network approach offers important contrasts to both political incentives
and ideological frameworks. Social networks and the ways they structure preferences
need not contribute to “rational” decision-making (DiMaggio & Powell 1983).
Policymakers may simply desire to “keep up with the Joneses”, even when doing what
the Joneses do may not be functionally beneficial for them (Weyland 2004).
instance, Finnemore argues that the worldwide diffusion of science bureaucracies
occurred even though many countries lacked a domestic demand for such institutions and
had few resources to invest for scientific advances (1993). In a similar vein, there was
limited evidence to indicate that Bolsa Escola and PSF were appropriate policies for all
Brazilian municipalities.
Additionally, social networks need not be comprised of
individuals who share the same ideological beliefs.
This is particularly true for
membership in professional associations, where individuals share professional norms but
may diverge when it comes to their personal ideology.
As this study brings motivations to the forefront to explain diffusion processes,
this framework sets aside questions related to federal financing. In a federal system like
27 Although Keck and Sikkink do not specifically address diffusion, their work on transnational advocacy
networks offers valuable insights on how cross-border networks that link actors with shared norms, are key
to explaining policy advocacy on issues such as human rights, women’s rights and the environment (1998).
Brazil, it is likely that municipal governments and local actors would find it useful to tap
into federal fiscal resources to increase their budgets28. Yet financing alone does not
constitute a motivation for policy emulation, particularly when the matching grants are
limited29. For these reasons, the potential impact of fiscal transfers is treated as a control
Each of the three approaches to understanding the motivations that drive political
behavior, in this case, the decisions to emulate programs such as Bolsa Escola and PSF,
can be tested empirically. If a political incentives approach explains diffusion, we would
expect decision-makers to use these policies to gain political power by including them in
their campaigns for office.
In contrast, if ideology drives decision making, then
politicians would frame adoption of these programs in terms of their ideological
commitments and beliefs. They would stand by their choices even if political selfinterest pointed in a different direction. Alternatively, actors who are drawn to these
programs because of learning through professional networks would express their
decisions in terms of the professional norms and trends in their field. In doing so, they
would relate their decision-making to others, participate in the same networks, and seek
to demonstrate how their policies reflect new conventions.
28 Research from the U.S. context suggests that fiscal transfers or matching funds can be influential in
spurring local governments to participate in new social programs (Derthick 1970; Mossberger 1999; Rose
1973; Welch & Thompson 1980).
29 The ministries of education and health have played an inconsistent role in supporting the spread of
municipal Bolsa Escola and Programa Saúde da Família. None of the municipalities in this study benefited
from federal financing to spur the adoption of municipal Bolsa Escola programs. Funding for PSF adoption
has changed over the time period in this study; at first very few municipalities qualified for any incentive
Local level politics and regional interests have and continue to be an important
focus for Brazilian politicians.
Historically, differences in geography, settlement,
immigration, industry, and levels of economic development, have resulted in markedly
different regional interests. Since the founding of the First Republic (1889-1930), the
country’s politics has reflected tensions between divergent local interests and the national
government, as elites from states such as Rio Grande do Sul, São Paulo, Rio de Janeiro,
and Minas Gerais all vied for political and economic influence. Today, states and
municipalities continue to exert political influence through federalism and as a result state
and municipal politics continues to remain relevant, both locally and nationally. More
importantly, politicians in Brazil do not view local representation as a lesser post or
stepping stone for more influential office with the federal government.
politicians readily move from municipal, state, and national office and back, as
opportunities arise. Mayors of state capitals are particularly influential, thus it is not
uncommon to see a former senator or governor compete in mayoral elections (Samuels
2003). For instance, prior to winning the mayoral election in São Paulo in 2004, José
Serra served as a Congressional deputy, senator, planning minister, health minister, and
ran a competitive bid for the Presidency; today he is a leading national figure in politics.
Unlike many Latin Americans countries where the capital city tends to dominate the
national political arena (Myers & Dietz 2002)30, municipal-level politics in Brazil offers
30 There are multiple reasons for the dominance of capital cities in the political landscapes of Latin
American countries. Among them are simple demographics. In many Latin American countries, nearly 50
percent of the population resides in the capital city. Brazil on the other hand, has many large cities
throughout the country; at least 31 cities have populations over a half-million in the census year 2000.
multiple venues for high-stakes electioneering where local contests matter for politicians’
Municipal races for executive office tend to be competitive and both candidates
and incumbents campaign vigorously to win voters’ attention.
Debates, electoral
advertisements, and direct appeals to voters by candidates via neighborhood visits are
commonplace features of the campaign season. When incumbents are ineligible for
reelections, they nevertheless campaign with vigor beside their hand-picked successor.
After all, the election of an anointed successor serves as an affirmation for the incumbent
that she has done a well and provides her with further political capital.
While many
contests are driven by personalism and candidates’ charisma, candidates do make
reference to their policy positions. Because of the socio-demographic realities of the
country, where 46 percent of the population was classified as poor in 2000,31 candidates
cannot afford to ignore the lower classes in majoritarian elections with mandatory voting.
For this reason, politicians from all political parties and ideological persuasion court the
lower classes. While some politicians engage in unscrupulous clientelistic methods, such
as vote-buying or patronage politics,32 most others seek to win votes by promising social
policies that would benefit the poor.
Politicians of all parties and ideological stripes make direct appeals to the lower
classes. Given that Bolsa Escola and PSF are targeted programs that benefit indigent and
31 See Martins 2005.
32 The global anti-corruption watchdog organization Transparency International reports that in the March
2001 elections, 7 percent of survey respondents in Brazil reported they were offered money for votes.
Interestingly, their studies indicate that vote-buying is not isolated to the poor. See:
poor groups, it is conceivable that all politicians would find the social policy particularly
useful. As chapters 3 and 4 explore in greater detail, politicians who were rightists,
centrists, and leftists all endorsed Bolsa Escola and PSF at some point in time. For
instance in the 1998 campaign in Brasília, Joaquim Roriz made a campaign promise to
keep Cristovam Buarque’s well-known program Bolsa Escola if elected (Interview
Buarque 2004). Similarly, in the 2004 campaign season in Salvador, both rightist and
leftist candidates sought to win voters by mentioning PSF and appealing to program
beneficiaries. Since Bolsa Escola and PSF are well-known award winning programs that
can further the electoral goals of calculating politicians, it is reasonable to suspect that the
degree of electoral competition can influence the likelihood these programs will be
adopted. In highly contested electoral environments, incumbents will feel the need to
demonstrate to voters that they are doing something to secure their reelection or further
their subsequent electoral ambitions. Whereas in cities with low levels of electoral
competition, incumbents can rest on their laurels and are more likely to retain their
electoral advantage, regardless of their accomplishments.
While the political incentives hypothesis is intuitively appealing, there is reason to
believe that ideology might matter when it comes to social policymaking in Brazil.
Actors with firm ideological commitments and beliefs could respond very strongly to
programs such as Bolsa Escola and Programa Saúde da Família. In general, leftists have
favored pro-poor policies that seek to overcome the historic marginalization of large
segments of the population and guarantee full citizenship. At the same time however,
leftists have been wary of many features of neoliberal policy reforms, which tend to
emphasize features such as limited state intervention and targeting, rather than broad and
universal programs with full social protection. Actors to the right-of-center tend to favor
neoliberal reforms, the retraction of the state, and pro-business policies.
In many
respects, Bolsa Escola and PSF represent both a pro-poor and pro-citizenship approach to
social policy while at the same time encompassing features of targeted neoliberal social
policy. The degree to which actors are driven by their ideological commitments and how
they interpret these programs could very much influence emulation decisions.
Lastly, the social network approach also holds plausibility for understanding the
Brazilian political scenario.
Since the abertura33 (political opening) various social
movements, – e.g. women’s, student, public health, and labor – have all pushed for
democratization and a constitution that would enshrine social rights in addition to basic
political rights (Alves 1985; Alvarez 1990; Diamond 1999). As Alvarez et al. note, the
movements from the 1970s and 1980s have undergone a significant transformation,
responding not only to the new political scenario under democracy but also the realities
of globalization and the shrinking state under neoliberalism; the result has been a
growing predominance of non-governmental organizations (NGOs) and the ‘NGOization’
of movements (Alvarez et al. 1998). Today Brazil has an abundance of associations and
NGOs that engage in policy debates and “watch-dog” activities.34 They include now well
established professional organizations that promote advances in education, health, and
33 The abertura period consisted of a slow, gradual, and insecure political liberatization begun by the
Geisel administration in 1974. The slow transition from military rule to democracy took over fifteen years
to complete.
34 The Ministry of Justice reports there are 455 organizations classified as public interest organizations in
Brazil (Ministry of Justice Website, Downloaded February 18, 2007).
anti-poverty efforts. Given the robust civil society activity throughout Brazil, we might
expect that formal and informal organizations could be instrumental to the promotion of
programs such as Bolsa Escola and Programa Saúde da Família.
The following chapters will empirically test the three competing motivational
approaches for understanding emulation decisions. Chapter 3 does so by taking a bird’seye view of policy diffusion for Brazil’s largest cities and statistically tests the three
motivational approaches in this study. Chapter 4 contextualizes Bolsa Escola and Renda
Minima policies in terms of stalled education reforms in Brazil and provides an in-depth
analysis of how these programs were debated and interpreted by local policymakers.
Process tracing of policymaking in four cities over the course of three different
administrations allows for the uncovering of the mechanisms that led to emulation
decisions. Chapter 5 follows by framing Programa Saúde da Família within the context
of health policy reform and decentralization in the 1990s. It uncovers how various local
actors involved with health policy interpreted the family health program and the factors
that drove their replication decisions.
In the 1990s both Bolsa Escola and PSF captured the attention of policymakers
and spread throughout the country, despite differences in scope, purpose, and unique
challenges related to their policy venue. One of the surprising features of Bolsa Escola
diffusion is that cities were so quick to emulate the program. Both Brasília and Campinas
implemented the policy in 1995 and within two years approximately 88 cities had already
adopted the program (Araújo & de Souza 1998). Other cities enacted the program after
the federal government introduced a short-lived matching grant to promote the policy’s
expansion.1 By 2001, over 200 municipalities and seven states had adopted Bolsa Escola
(Villatoro 2004).
On the one hand, the adoption of Bolsa Escola by municipal
governments is notable because most cities financed the program directly out of their
budgets, reflecting a policy commitment to prioritize the program. On the other hand,
given that the program had received so many accolades, both nationally and
internationally, it is curious why so few governments among Brazil’s 5,500
municipalities had in fact adopted it.
The diffusion pattern for Programa Saúde da Família differed considerably from
that of conditional-cash transfer programs. At first, PSF operated on a small scale, with
only a few small municipalities adopting the program. Most of the early adopters were
poor, rural, and concentrated in the northeast of the country. By the late 1990s PSF
1 However, only cities with below their states’ average income per capita were eligible (L. Lavinas & M.
L. d. O. Barbosa 2000).
gained broader credibility and visibility, both within and outside of the Health Ministry,
spreading from 55 municipalities in the first year to 4,944 by 2003. (See Appendix C for
state-based information on the scope of PSF coverage, by state and population as well as
aggregate date on PSF adoption across the country’s municipalities).
This chapter first provides an overview of municipal Bolsa Escola and Programa
Saúde da Família diffusion.
After tracing the diffusion patterns, this chapter also
explains some broad tendencies among those cities that chose to adopt each of the two
programs. The heart of the chapter integrates that framework and tests the theoretical
motivations for policy emulation – political incentives, ideology, and social networks –
for Brazil’s largest municipalities, using an event history analysis. After elaborating on
the hypotheses and their measures, the third section presents the model results and
interpretation of the findings.
Surprisingly, even though Bolsa Escola and PSF are well-known programs that
have garnered national and international attention, there is very little systematic data on
when and where these programs first spread. The lack of data is due in part to the
organic nature of their diffusion; early-on these programs traveled without coordinated
stimulus or tracking by the national government’s ministries. Since municipal Bolsa
Escola programs were enacted independently from the Ministry of Education, there was
never any systematic data collection effort to track their spread. Although the federal
research agency IPEA conducted a series of evaluations of these policies, these rarely
examined diffusion and conducted only selected case studies (Lavinas & Barbosa 2000;
Lobato & Urani 1998). In the case of PSF, during its earliest phase (1994-1997), the
program was a small project tracked by a few federal civil servants and the Ministry only
collected aggregated data by state.2 It would take four years for the Ministry of Health to
systematically collect information on municipal adoption of PSF.
Given the limited information on diffusion that is available on Bolsa Escola and
Renda Mínima, this study draws on original data collection. To map the pattern of
diffusion, I conducted a phone survey of all the 224 cities with populations over 100,000
in the census year 2000. Researchers telephoned municipal civil servants in departments
of education and public assistance to inquire whether their cities had a municipal
education stipend program.3 If interviewees answered in the affirmative, they were asked
what year the program was enacted, how it was administered (i.e. which agency was
responsible for the program), and to describe the program more generally (see Appendix
D for interview questionnaire). Ascertaining a policy’s start date is one of the most
difficult methodological challenges for diffusion research, because researchers must often
ask respondents to reflect back in time to identify the date of adoption; recall data is a
characteristic weakness of diffusion research (Rogers 2003: 126). To minimize this
problem, interviewers prompted respondents to consider if there was any legislation or
2 State-based record keeping on municipal adoption of PSF from 1994-1997 is uneven; despite concerted
efforts to retrieve information on the earliest adopters it has been impossible to obtain this information. In
cooperation with this study, staff members from CONASEMS initiated a state-based inquiry to retrieve this
historic data. Unfortunately, many states had not maintained records from that time period or had never
systematically tracked the municipal adoption of PSF.
3 Research Assistants, Francisco Marques, Ana Paula Karruz and Evelyn Chaves carried out phone
interviews between November 2003 and October 2004. In some instances, these researchers made repeated
calls to reach those civil servants responsible for educational social programs.
administrative decree that would provide a clear timeline for the program’s start. Even
though many cities had programs with different names, if respondents reported their
cities administered programs that shared a similar design and programmatic goals (e.g.
cash transfers with educational conditionality) then those cities were classified as having
Bolsa Escola.
Respondents’ answers were also cross-checked with other published
records and municipal documents, when available.
The response rate for the phone survey was high, with a total of 93.3 percent of
cities participating in the phone interviews. Most respondents were willing to discuss
their municipality’s programs and services. However, of the total 244 cities in the
sample, 15 had officials who either refused to answer questions or were otherwise noncooperative. While it is difficult to explain non-responses, it is also the case that despite
nearly twenty years of democracy in Brazil, many local governments are still wary of
providing public information.4 Those non-participating cities were dropped as “missing
cases.”5 In total, 48 cities (22.9 percent) reported having a program similar to Bolsa
Escola by 2003; Figure 3.1 shows the overall diffusion pattern of Bolsa Escola programs
for the sample in this study.
4 Non-cooperation included instances where respondents explained they needed a supervisor’s permission
to provide information. Some administrators asserted they needed formal requests for information. When
interviewers sent formal letters of inquiry with follow-up phone calls, staff still declined to provide
5 Cities dropped as missing cases in Model 1 included: Fortaleza (CE), Abaetetuba (PA), Olinda (PE),
Vitória de Santo Antão (PE), Parnaíba (PI), Magé (RJ), Nilópolis (RJ), Rio de Janeiro (RJ), Viamão (RS),
Diadema (SP), Embu (SP), Franca (SP), Itaquaquecetuba (SP), Itu (SP), São Caetano (SP).
Figure 3.1: Cumulative Adoption of Bolsa Escola in the Sample
Number of Cities
A significant portion of the PSF data used in this study draws on information
made available by the Ministry of Health, Department of Basic Health Services
(Departamento de Atenção Básica). Where possible, data on the early years of the
program for the 224 municipalities in this study were cross-checked with secondary
sources or deduced based on other state-wide information.6 Otherwise, the decision-rule
was to code all cities with missing data between 1994 and 1997 as non-adopters. This
rule produces data that corresponds with general information about the program’s
beginnings. Since the program first spread to small rural municipalities in the northeast
(Sousa 2002; Viana and dal Poz 1998), it is reasonable that very few of the cities in the
6 For instance, aggregate state data provided from the Ministry of Health for 1994-1998 confirmed that
several municipalities in the study had not adopted PSF during that time period. A few other State Health
Departments, including Goiás and Ceará made their municipal-level data available for this study. In
addition, a few municipalities had websites that specified the date PSF began in their cities.
sample would have adopted PSF from 1994-1997. The imputed values for PSF also yield
an overall trend that is similar to the aggregate national data provided by the Ministry of
Health (see Appendix C, Figure C.1). By 2003, 89 percent of all the cities in this sample
had adopted PSF and most of those cities which had not adopted the program were in the
Southern region of the country. Figure 3.2 shows the resulting diffusion trend for PSF in
this sample.
Figure 3.2: Cumulative Adoption of Programa Saúde da Família in the Sample
Number of Cities
1994 1995 1996 1997 1998 1999 2000 2001 2002 2003
Scholars of diffusion note that adoption over time typically follows an S-shaped
curve, while a frequency distribution of the number of mean adopters per year approaches
a bell-shaped normal distribution (Rogers 2003: 273). The S-shaped distribution rises
slowly at first, as the first few “innovative” or risk-taking cities adopt the program; the
curve’s slope increases sharply until approximately half the population has adopted the
program, after which the slope decreases (Rogers 2003: 272). Figures 3.1 and 3.2 visually
demonstrate the cumulative adoption rates of Bolsa Escola and PSF, and show some
important differences between these programs’ trajectories. Programa Saúde da Família
represents the classic S-shaped distribution. While the Bolsa Escola pattern mirrors some
of the overall S-shaped distribution, the program adoption still constituted a relatively
rare event by 2003 and had yet to “take-off” and spread to half of the sample. In
comparing two programs that spread at different rates, we can determine whether these
programs nevertheless share similar underlying explanations for their diffusion.
The selection of Brazil’s 224 largest cities for the large-N analysis in this
dissertation is not intended to be representative of all municipalities, but is instead based
on several factors. First, Brazilian municipalities vary significantly in size, from the
smallest town with 795 residents to the largest mega-city São Paulo with over 10.4
million people (IBGE). In order to ensure some comparability between cities, this study
includes all cities with similar population parameters (i.e. medium to large) in the census
year 2000. Second, although these cities represent a small fraction of all cities, they
encompass over 51 percent of the total population and are distributed across all of the 26
states and the Federal District. Thus, although the number of cities is small in comparison
to all municipalities, it does allow for an analysis that affects more than half of the total
population. Third, given the significant barriers in relation to data access and the dataintensive requirements of event history modeling, feasibility precluded the inclusion of
all cities. Since data collection included phone surveys of municipal administrators, it
was infeasible to collect data for a larger number of cities.
Having uncovered the patterns of Bolsa Escola and PSF diffusion, we can now
ask the question: what are the main characteristics of those cities that adopted these
policies? And more importantly, how might electoral incentives, mayors’ ideological
convictions, social networks, and other control factors influence policy emulation? This
section explains how each of the theoretical variables of interest is measured and
provides a snapshot of each factor’s relationship to the dependent variables.
accomplish this, I draw on annual observations for the year 1998 to examine simple
cross-sectional relationships.7 The dataset for each policy issue has a different sample
size; Bolsa Escola (n=208) and PSF (n=2238). Thus, the following analysis differentiates
between tendencies for Bolsa Escola and PSF.
Political Incentives
To uncover the potential relationship between political interests and policy
emulation, I measure each city’s degree of electoral competitiveness, drawn from election
results for the 224 cities in this study from 1992,9 1996, and 2000. Data for this variable
came from the federal election bureau (Tribunal Superior Eleitoral), the 27 state election
7 The entire database includes annual observations for each city, from each policy’s start-year to 2003. For
the purposes of this analysis all cities are included in the sample (i.e. cities that previously adopted the
program are not dropped from the dataset). I select 1998 as it is the first year of the second administrative
period in this study.
8 The sample size is 223 because the municipality Timon (MA) was not incorporated in 1998. Timon is
included in the analysis for later years.
9 Data for the 1992 municipal elections are very difficult to obtain. The federal government had not
imposed guidelines for how states should collect and distribute election data. Some states, such as Minas
Gerais, collected systematic and detailed information on election results, while other states did not. In
subsequent elections voting was electronic, facilitating the distribution of election data.
bureaus (Tribunal Regional Eleitoral), the database Voto a Voto from the Fundação
Perseu Abramo, and archival records from newspaper articles. The mayoral contests are
classified into three categories: highly competitive, competitive, or non-competitive.
Highly competitive races are those where the winning candidate wins with less than 45
percent of the vote, in competitive elections the winning candidate garners between 45 to
55 percent of the vote, and in a non-competitive race the winner garners more than 56
percent of all votes.10 For the event history models, which require annual observations,
the competition values for these cities were repeated until the next election cycle.
Table 3.1 Electoral Competition & Bolsa Escola Adoption (1998)
BE Non-Adoption
BE Adoption
Total in Sample
High Competition
Low Competition
Within the Bolsa Escola dataset, we see that nearly half the cities had highly
competitive elections. In fact, about 80 percent of cities had medium to high levels of
competition, suggesting that among Brazil’s largest cities, meaningful electoral
competition is a characteristic of local politics. Interestingly, there are differences in
10 Brazilian election law requires that mayors win 50%+1 of the valid votes when cities have more than
200,000 electors. A few cities in this sample held second round elections for mayor. However, for even
comparison across the cases, this measure only examined first round results.
Bolsa Escola adoption decisions based on cities’ competitiveness, although the trends are
non-linear. Cities with highly competitive elections were most likely to adopt the policy
(17 percent), followed by cities with low levels of competition (15 percent).
Table 3.2 Electoral Competition & PSF Adoption (1998)
High Competition
Low Competition
PSF Non-Adoption
PSF Adoption
Total in Sample
Like Bolsa Escola, the cities in the PSF dataset also demonstrate a variation in
electoral competition. Nearly half of the cities have highly competitive elections and a
solid third fall in the medium competition category. However, unlike Bolsa Escola,
adoption of PSF happens most often among cities that have moderate levels of
competition; nearly half in this category adopt the health program. Although fewer cities
held low competition elections, this subset emulated PSF the least (19 percent).
Most Brazilian political parties fall along a left-right ideological spectrum
(Mainwaring 1999; Mainwaring et. al 2000). Following Mainwaring’s typology of
partisan ideology (1999: xvii-xix), I group each city’s mayor into one of three mayor
categories: Left, Center and Right.11 Classifications are based on the candidates’ partisan
affiliation at the time she or he filed with their local elections bureau.12 A very small
fourth category, “Non-Aligned” captures new parties or small regional parties for which
there is little information.
Table 3.3 Ideology & Bolsa Escola Adoption (1998)
BE Non-Adoption
BE Adoption
Total in Sample
Leftist Mayor
Centrist Mayor
Rightist Mayor
Mayoral Ideology
The Bolsa Escola dataset for 1998 shows that on aggregate, these cities are
governed by executives that represent various ideological perspectives.
No single
ideological perspective dominates the sample. Most cities’ mayors are centrists, followed
closely by rightists. Nearly a quarter of these cities have mayors from parties that fall on
the left of the political spectrum.
One of the interesting features of Bolsa Escola
adoption is that a greater percentage of centrist mayors adopted the program, compared to
leftists. Rightist rarely adopted the education cash transfer program.
11 Left Parties: PT, PC do B, PSB, PPS, PDT, PMN and PV; Center Parties: PSDB, PMDB, PTB; Right
Parties: PFL, PL, PDS/PPR/PPB, PRONA, PSC, PSL and PSD. The relatively new party, PV is included in
the left category as its members are often former party members of other left parties.
12 Party switching is relatively common in Brazil; although, mayors from well established and
institutionalized parties (e.g. PT, PSDB, PFL, etc.) typically retain their partisan affiliations through the
course of their term.
Table 3.4 Ideology & PSF Adoption (1998)
PSF Non-Adoption
PSF Adoption
Total in Sample
Leftist Mayor
Centrist Mayor
Rightist Mayor
Mayoral Ideology
The dataset for PSF, which includes a few more cities, provides a similar
distribution along mayoral ideology: there is an equal number of centrists and rightists,
while leftists constitute a quarter of the sample. There are notable differences between
leftists, centrists, and their right-of-center counterparts when it comes to PSF adoption.
Only a third of rightists adopt the family health program, whereas over fifty percent of
centrists and leftists adopt it. In other words, centrists and leftists adopt PSF at similar
Social Networks
Social networks for policymaking generally reflect the unique specialities of the
professionals who work in that policy arena.
Thus, to understand the relationship
between professional networks and policy diffusion, this study draws on data from two
different organizations.
In the case of Bolsa Escola, the Programa Gestão Pública e Cidadania (Public
Management and Citizenship Program) was cited by practitioners as an influential source
of information and convener for professionals. The program conducts multiple activities,
including an annual competition on innovative public policy. Applicants for the award
receive information from the Gestão Pública e Cidadania program, and a select number
later participate in the organization’s meetings and conferences.
The program also
generates annual data on the cities that apply for an innovation award, dating back to
1996. Interestingly, the number of applications from cities does vary by administration,
so the annual data allows for a nuanced analysis of network connectivity.
When it comes to primary healthcare, health professionals often cited in
interviews that the Centro Brasileiro de Estudos de Saúde (CEBES) was an important
professional association. This organization, which publishes the well-known journal
Saúde em Debate, was founded by militants in the movimento sanitário (public health
movement). Most members are professionals engaged in public health and work for
local, state, or federal governments. CEBES staff members provided membership data by
municipality. Since membership and subscriptions to the journal have remained constant
over the years, data for 2003 was used for all annual observations. Both the network
variables were coded dichotomously, “1” for cities where at least one person was linked
to the network/a city applied for an award, and “0” for cities where no one had any
formal participation/had not applied for an award.13
Given Brazil’s continental size, geographic effects can also influence diffusion
processes as actors learn through informal social networks. To examine the potential
relationship between region and adoption, I classify all cities according to the Brazilian
13 In tracking Gestão Pública Award applications, I did not include those cities that had only applied for an
award for a Bolsa Escola/Renda Mínima program, so as to avoid the problem of autocorrelation.
government’s geographic categories: North, Northeast, Southeast, South and Central
Table 3.5 The Gestão Pública Network & Bolsa Escola Adoption (1998)
BE Non-Adoption
BE Adoption
Total in Sample
Formal Networks
Gestão Pública
No Gestão Pública
The Gestão Pública network permeated over a third (32.6%) of all municipalities
in the study. Cities that were connected to this network adopted the program much more
often than their non-networked counterparts; only 5 percent of those cities chose to adopt
Bolsa Escola.
Table 3.6 Region & Bolsa Escola Adoption (1998)
BE Non-Adoption
Informal Networks
Central West
BE Adoption
Total in Sample
14 Central West: Goiás, Mato Grosso do Sul, Mato Grosso, and Distrito Federal (Brasília); North: Acre,
Amapá, Amazonas, Pará, Rondônia, Roraima, and Tocantins; Northeast: Alagoas, Bahia, Ceará, Maranhão,
Paraíba, Piaui, Rio Grande do Norte, Sergipe; Southeast: Espirito Santo, Minas Gerais, Rio de Janeiro and
São Paulo; South: Paraná, Rio Grande do Sul, and Santa Catarina.
Given the selection criteria for cities in this study, it is understandable that the
distribution of cases is uneven across the five regions in the country. It is the differential
rate of adoption within each region that is most striking. Bolsa Escola adoption is rarest
among large cities in the north. In contrast, neighboring regions – the southeast and
central west – display higher rates of adoption. Not surprisingly, the cities to first adopt
conditional cash-transfer programs were located in these regions.
Table 3.7 Summary of Dichotomous Variables in Bolsa Escola Dataset (1998)
PSF Non-Adoption
Formal Networks
CEBES Network
PSF Adoption
Total in Sample
Unlike the professional network associated with Bolsa Escola, which represents a
small share of the overall sample, the presence of CEBES is much more pronounced
across Brazilian cities; nearly seventy percent of cities have some connection to the
health network. While cities with a linkage to CEBES evenly split between PSF adopters
and non-adopters, it is notable that the rates of adoption differ between cities with and
without a CEBES link. In other words, municipalities without a CEBES connection
adopted PSF less often; only 37.7 percent of those cities emulate the family health
Table 3.8 Summary of Dichotomous Variables in PSF Dataset (1998)
PSF Non-Adoption
Informal Networks
Central West
PSF Adoption
Total in Sample
Given that the family health program originated in the northeast, we would expect
that municipalities within this region would adopt the policy at greater rates than their
counterparts. Almost two thirds of the municipalities in the northeast do emulate PSF;
however it is the northern region which actually takes the lead with 78.6 percent adopting
the policy. A striking characteristic of PSF adoption by region is the clear geographic
pattern that emerges; municipalities in regions furthest from the north have the lowest
adoption rates. Thus, the south lags farthest behind the north, with only a third adopting
Overall, these descriptive statistics provide a snapshot of Bolsa Escola and PSF
adoption among Brazil’s largest cities.
Cross-sectional analysis of the theoretical
variables of interest – political incentives, ideology, and social networks – reveals some
broad tendencies. However, logistic regression analysis is necessary to test the relative
impact of each variable on policy adoption. The next section details the model which
tests three competing explanations for diffusion, while also accounting for the timing of
adoption decisions.
This study of policy diffusion tackles several interrelated questions. First, what
motivates policymakers to emulate programs like Bolsa Escola and Programa Saúde da
Família? Second, why did some cities quickly emulate the program while others lagged
farther behind? One of the important features to policy diffusion is timing. This chapter
examines both sets of questions by drawing on event history analysis to test the three
competing approaches that uncover actors’ motivations for policy adoption.15
Event history analysis is a standard statistical method used by diffusion scholars
to parse out both internal and external determinants of policy adoption decisions.16 The
model, which is similar to a discrete time logistic model that includes time controls,
allows for probabilistic interpretation of “risk” or “hazards” of an event occurring. In this
study, each logistic model will capture the likelihood that a city will adopt Bolsa
Escola/Renda Mínima and PSF, in any given year; Model 1 captures Bolsa Escola and
Renda Mínima, while Model 2 analyzes the adoption of PSF. The resulting analysis will
reveal the extent to which the independent variables – political incentives, ideology and
social networks – increase the probability for diffusion. To portray the time dimension of
Bolsa Escola diffusion, this model includes annual observations for the period 19952003, which corresponds to the year Bolsa Escola/Renda Minima programs were first
introduced and the last year for which data is available. Similarly, the family health
15 I use a logistic model with time controls.
16 For further details on event history methodology for the social sciences, see Allison 1984; BoxSteffensmeier & Jones 1997, 2004.
program started in 1994, thus the dataset for Model 2 includes the period from 19942003.
Model Specification & Hypotheses
The logit equation for each model produces a log-odds of an event occurring. The
model specifications follow below:
3.1a Model 1: Zb = ln (Pi/1-Pi) = α1 + ß1(highly competitive)I + ß 2(competitive)i +
ß 3(leftist mayors)i + ß 4(centrist mayors)i + ß 5(Gestão Pública network)i +
ß 6(Lagged neighborhood effect)I + ß 7(south)i + ß 8(southeast)i +
ß 9(central west)i+ ß 10(northeast)i + ß 11(medium pop. city)i + ß 12(large
pop. city)i + ß 13(t1)i+…. ß 20(t8).
3.2a Model 2: Zp = ln (Pi/1-Pi) = α 1 + ß1(highly competitive)i + ß 2(competitive)i +
ß 3(leftist mayors)i + ß 4(centrist mayors)i+ ß 5(CEBES network)i +
ß 6(Lagged neighborhood effect)i + ß 7(south)i + ß 8(southeast)i +
ß 9(central west)i+ ß 10(northeast)i + ß 11(medium pop. city)i + ß 12(large
pop. city)i + ß 13(t1)i+…. ß 21(t9).
Dependent Variables
To measure the effects of political incentives, ideology, and socialized norms on
social policy diffusion, I collected data on the adoption of Bolsa Escola/Renda Mínima
and Programa Saúde da Família for Brazil’s largest cities. In both cases, the dataset
requirements included determining which cities had these programs and in what year they
were replicated.
The dependent variables are coded dichotomously (“0” for cities
without the program in a given year and “1” for cities that had adopted the program).
Once the event occurs (i.e. the program emulation takes place), observations for that city
are dropped from the dataset.
Independent Variables
The two models test proxies for the three competing explanations of the diffusion
of social programs in Brazil’s largest municipalities.
Political Incentives
The first school of thought to explain the underlying motivations for political
behavior posits that individuals regularly respond to political self-interest. Walker argues,
for instance, that the more electorally competitive the jurisdiction, the more likely policy
replication will occur as actors compete for votes (Walker 1969). With this logic, we
would expect that cities with competitive mayoral races would be more likely to adopt
Bolsa Escola and PSF than those facing less contested races.
Hypothesis 1: The stiffer the jurisdiction’s electoral competition, the more likely
policy emulation will occur.
The second theoretical approach these models will test is whether policymakers’
ideological convictions influence the likelihood of program emulation.
As Mullins
argues, ideology can influence political actors in actionable ways, as it filters their
information, shapes their worldview, and guides their evaluation of particular policies
(1972). In this vein, politicians with dissimilar ideological convictions would respond
differently to particular policies. Both Bolsa Escola and PSF are programs that have
equity-enhancing goals and aim to extend social services to marginalized and poor
population groups. Thus, we would expect that left-leaning progressive actors would be
more willing than others, either on the right or center, to adopt these programs.
Hypothesis 2: When cities have left-of-center mayors, they will be more likely to
replicate Bolsa Escola and PSF.
Social Networks
A third alternative explanation is that political actors are motivated to emulate
programs for diffusion when they are socialized to do so through their professional
associations (Balla 2001). Connections between individual policymakers and social
networks could spur program diffusion for a number of reasons, including professional
socialization, peer pressures, legitimacy considerations, or information exchange. The
exact motivational relationship between individuals and their professional associations
can be difficult to discern without qualitative evidence, however, statistically we would
expect to see that those individuals with connections to networks would be more likely to
emulate Bolsa Escola or PSF. Since these policies relate to specific arenas (e.g. health,
education/poverty alleviation) we would expect that professionals in each sector would
turn to distinct associations.
Hypothesis 3a: When actors in these cities have linkages to the Gestão Pública e
Cidadania network, the likelihood that Bolsa Escola will be adopted increases.
Hypothesis 3b: When actors in these cities have linkages to the CEBES network,
the likelihood that PSF will be adopted increases.
In addition to the formal membership in professional associations, several
diffusion scholars note that informal socialization through “neighborhood effects” can
also drive diffusion (Mooney 2001; Walker 1969). For instance, several researchers posit
that learning can travel spatially across geographic territories, either because there are
opportunities for professionals to meet regionally, or because the circulation of
information tends to be regionally based (Mooney 2001; Walker 1969). In this logic,
governments that are geographically proximate will be more likely to replicate a
neighbor’s innovative program.
Hypothesis 4: The greater the proportion of neighboring cities with Bolsa
Escola/PSF, the greater the likelihood that a city will emulate its neighbors.
The measure used for “neighborhood effect” in this study was the proportion of
municipalities (in the sample) that had adopted either Bolsa Escola or PSF in their
respective state.
Since we would expect some time-delay between a jurisdiction’s
decision and its influence on their neighbors, this variable is lagged by a year.
Another measure of regional influence on diffusion is region; thus, municipalities
are classified into one of five areas: north, northeast, southeast, south and central-west.
Regional characteristics might matter because the policies in this study were also born in
different parts of the country; PSF is most associated with the northeast while Bolsa
Escola and Renda Mínima were developed in the Southeast and Central West.
Control Variables
Aside from the immediate goal of winning elections, scholars also argue that
political actors will respond to fiscal incentives when they are made available, typically
through the federal government (Derthick 1970; Mossberger 1999; Rose 1973; Welch &
Thompson 1980). Federal financing, through matching grants or incentive grants,
provides politicians with resources to demonstrate their accomplishment to their
constituents. In the case of Bolsa Escola, the only financing made available to
municipalities was limited to those cities with below their state’s average income per
capita. Few, if any, of the cities in this dataset fell into this category and the program
lasted no more than a year. For this reason, this variable is not included in Model 1.
Although the Ministry of Health has promoted PSF through financing, including
line-item transfers, it is very difficult to operationalize the magnitude of the funding
because of irregular data collection and availability.17 For this reason, Model 2 will test
the impact of federal transfers with the indirect proxy “year,” which allows us to isolate
whether the years in which the federal government initiated changes in financing
correspond to increased likelihoods that PSF would be replicated.
City size is another characteristic that might matter for municipal administration
and policy advocacy. Although Brazilian cities face similar levels of financing from the
federal government, other factors typically associated with city-size might matter.
Residents from smaller cities might welcome a health program that relies on home-visits
The Ministry of Health does have a sophisticated online database, Datasus, with information on health
financing, however this data is only available for the most recent periods, from 1998-present.
from neighbors, whereas residents in large urban cities might shun these intrusions on
their privacy. Larger cities also tend to have more universities and non-profit
organizations, enabling civil society engagement in policy debates. For the purpose of
this analysis, cities are grouped into three categories: Small, Medium and Large Cities.18
Other potential factors that could influence policy diffusion are: “internal needs”
such as levels of poverty, “internal capacity” such as resource availability, and the
mayor’s gendered priorities. However, none of these variables bore out in the analyses,
and were thus dropped from the models presented below.
Model Results
One of the most remarkable features of this study is that despite differences
between Bolsa Escola and PSF, such as policy area, extent and rate of their replication,
both policies share similar determinants for diffusion. Models 1 and 2 offer remarkably
consistent results for understanding the relationship between political incentives,
ideology, and social networks on social policy diffusion, see Table 3.13.
One of the most surprising results of the event history models is the null finding
that electoral competition does not spur policy emulation, for either social policy issue.
The theoretical literature on the relationship between political competition and the
incentives they create for policy renewal and experimentation, based primarily on the
United States (for example, Walker 1969; Lowi 1963), does not apply for the largest
Brazilian cities included in this sample. In other words, when we control for all other
Small Cities: population less than 150,000; Medium Cities: population between 150,000 and 300,000;
Large Cities: population greater than 300,000 (IBGE/Ipeadata).
Table 3.13 The Determinants of Social Policy Diffusion: Bolsa Escola and PSF
Coefficients (with standard error in parenthesis)
Political Competition
Highly Competitive
Leftist Mayors
Centrist Mayors
Social Networks
Gestão Pública Network
CEBES Network
Lagged Neighborhood Effect
Central West
Model 1
Bolsa Escola
Model 2
-.053 (.412)
.036 (.414)
-.112 (.248)
.150 (.254)
1.024 (.460)*
.732 (.115)
.516 (.257)*
.179 (.237)
1.382 (.347)**
-.213 (2.581)
.659 (1.115)
1.569 (1.069)
.684 (1.272)
1.180 (1.083)
.502 (.219)*
1.287 (.813)
-2.407 (.582) **
-2.009 (.552) **
-1.077 (.688)
-.980 (.583)
.376 (.458)
1.394 (.436)**
-.149 (.223)
.036 (.254)
-.427 (.691)
.395 (.644)
-.110 (.709)
-1.29 (.867)
-1.244 (.886)
-2.290 (.741)
.112 (1.441)
.812 (1.225)
5.594 (1.047)**
4.176 (1.054)**
5.131 (1.060)**
-6.174 (1.214)
-4.648 (1.126)
City-Size Controls
Medium City
Large City
Log Likelihood
Prob > chi2
Pseudo R2
*p<.05. **p<.01.
In Model 1, T1 is 1995; for Model 2 T1 is 1994.
The statistical program Stata dropped T9 because the model perfectly predicted non-adoption.
factors, cities with more electoral competition did not adopt either Bolsa Escola or PSF at
higher rates than those with less competitive elections.
How does ideology fare in explaining policy diffusion? When compared with
cities governed by the right, leftist mayors are significantly different and in the expected
direction; they are more likely to adopt both Bolsa Escola and PSF, even when
controlling for other factors. Interestingly, the impact of ideology is most pronounced for
partisans on the left, as centrists are not statistically different from their rightist
The last set of theoretically driven variables tested in these models corresponds to
a sociological approach for explaining diffusion. Although neither model reveals that a
lagged neighborhood effect matters for policy diffusion, the other social networks
variables clearly demonstrate that when cities are linked to influential professional
networks, those cities are more likely to adopt Bolsa Escola and PSF. In other words,
cities with a member of CEBES are more like to adopt PSF and cities with a connection
to the Gestão Pública e Cidadania network are more likely to emulate Bolsa Escola.
As expected, time, the indirect measure that captures the influence of federal
spending confirms that PSF adoption would increase with the introduction of greater
funding in 1998. The likelihood of PSF adoption increases significantly from 1998 to
2003 (T5-T10). Obviously, the availability of resources that help defray the costs of
program adoption facilitate diffusion. But it is noteworthy that despite this control factor,
policy emulation still depends in large part on the ideological perspective of municipal
mayors and ties professionals have to social networks.
The logistic models of Bolsa Escola and PSF can tell us more than just the
relationship between the outcome (policy adoption) and its contributing factors. Recall
that logit equations 3.1a and 3.2a produce a “log odds” of an event occurring. Since we
normally think in terms of probabilities of events occurring, these equations can be
transformed mathematically to yield bounded probabilities, which are more interpretable
(see 3.1b and 3.2b below).
3.1b Probability of a city adopting Bolsa Escola/Renda Mínima = 1/(1+(exp(-1* Zb))),
where Zb is the log-odds of Bolsa Escola/Renda Mínima.
3.2b Probability of a city adopting PSF = 1/(1+(exp(-1*Zp))), where Zp is the log-odds of
The above equations allow for an analysis of the relative impact that the variables
of interest have on the probability of Bolsa Escola diffusion. For instance, the modal city
in the dataset had the following characteristics: it had competitive elections, a centrist
mayor, was not a part of the Gestão Pública network, had a lagged neighborhood effect
score of 0, had a medium-sized population, was located in the southeast, and adopted the
program in 1997. Given this scenario, Model 1 predicts that in 1997 the probability the
modal city will adopt Bolsa Escola is 3 percent.19 Although this probability is low and
indicates the program is rarely emulated, it is consistent with what we know about its
overall adoption rate. When varying certain key characteristics of cities, however, the
19 In the sample, 2.9 percent adopt Bolsa Escola.
overall probability of Bolsa Escola adoption increases significantly. For example, when
the modal city is governed by a leftist rather than centrist mayor, the probability of
adoption increases by one third to 4 percent. An even greater effect occurs when the
modal city has both a leftist mayor and participates in the Gestão Pública network,
increasing the predicted probability of Bolsa Escola adoption to 13 percent. Overall the
model predicts the lowest probability of Bolsa Escola adoption, 1 percent, for a city that
has highly competitive elections, is governed by a mayor from the right, has a small
population, and is located in the south. By contrast the city with the greatest likelihood of
emulating the program, with a predicted probability of 41 percent, would be: a large city
located in the southeast, with competitive elections, a leftist mayor, and participation in
the Gestão Pública network.
Drawing on the equations above, we can uncover the relative probability of Bolsa
Escola adoption. Interestingly, “time” also has important effects on the likelihood of
policy adoption. In other words, the probability of diffusion of the modal city differs
according to the year under consideration (see Figure 3.3). Bolsa Escola programs were
more likely to be enacted in 1997 and 2001, which corresponds to executives’ first year
in office.
This seemingly cyclical pattern corroborates Lowi’s argument that new
policies are more likely to be enacted at the beginning of a new administration (1963).
Also noteworthy is that mayors choose to adopt municipal Bolsa Escola programs in
2001, even after the federal government institutes its own national program.
commitment to the program demonstrates that some local officials still felt deeply that
the program was an important priority for their administration. Had mayors decided to
enact Bolsa Ecola at the end of their administrations’ and prior to the next elections, this
would indicate electoral incentives drove the timing of emulation decisions. But since
mayors typically implemented the education program just after an election suggests that
elected officials took advantage of the freedom for ideological decision making granted
by their electoral mandate.
Figure 3.3 Annual Predicted Probability of Bolsa Escola Adoption for Modal City
In the case of Programa Saúde da Família, Model 2 predicts much higher
probabilities that cities will adopt the program, when compared with Bolsa Escola
diffusion. The modal city in this sample has low competition, a centrist mayor, a lagged
“neighborhood effect” score of 0, adopted the program in 1998, has a medium-sized
population, and is located in the southeast. Given these characteristics, the likelihood that
this average city will adopt PSF (in 1997) is 37 percent.20 However, if this same city
were to lose its linkage to the CEBES network, the overall probability would drop to 26
percent. Interestingly, the effect of ideology is especially strong. Holding all other
factors constant, if the average city elects a leftist mayor, then the predicted probability
PSF will be implemented increases to 45 percent. Overall the model shows that when
different characteristics are in play, the probability of PSF adoption can vary
considerably depending on the theoretical variables of interest.
In 1998 alone, the
probability of adoption is as high as 73 percent for a mid-sized city in the northeast with
medium competition, a leftist mayor, and CEBES network presence. When that same
city has a rightist mayor and loses its CEBES network connectivity, the probability of
PSF adoption falls precipitously to 43 percent. While variables such as region or citysize can significantly affect the probability of PSF adoption, it is clear that the
motivational variables – ideology and social networks – dramatically influence the
likelihood that the health program will diffuse.
There are several reasons to suspect that “time” would impact the probability of
PSF adoption. As Figure 3.2, on the cumulative adoption of PSF in the sample
demonstrates, there is a dramatic increase in PSF adoption in 1998. The results in Model
2 also indicate that the years 1998 to 2003 are statistically significant and positively
associated with PSF adoption. This finding reflects the history of policy development
and mirrors other diffusion research that shows federal financing has strong effects in
spurring diffusion decisions (Derthick 1970; Mossberger 1999; Rose 1973; Welch &
20 In the sample, 13.8 percent adopt PSF.
Thompson 1980).
For the modal city in this sample, the yearly effects are remarkable
(see Figure 3.4). The probability of adoption jumps dramatically after 1997, from .5 to
37 percent. Also noteworthy is that the annual effects are non-linear and non-cyclical.
Unlike the timing of Bolsa Escola adoption, the timing decisions behind PSF appear
unrelated to election cycles, thus reinforcing the findings in Model 2 that electoral
competition bears no influence over emulation decisions.
Figure 3.4 Annual Predicted Probability of PSF Adoption for Modal City
The event history analyses allow us to test three theoretically driven explanations
for the diffusion of social policies for a large number of municipalities in Brazil.
Surprisingly, the proxies for political competition provide little support for the argument
that electoral competition and vote-seeking behavior on the part of politicians spur policy
replication. This bucks conventional assumptions in political science that highly
contested elections will spur actors to emulate policies.
A particularly noteworthy
finding in this regards is that the degree of electoral competition did not matter;
politicians in low, medium, and highly competitive environment behaved the same. By
contrast, the proxies for ideology and formal social network connectivity do explain the
likelihood that Bolsa Escola and Programa Saúde da Família will diffuse. That both leftist
ideology and formal social networks connectivity positively influence the emulation of
two programs that spread at different rates is quite remarkable particularly since scholars
of education and health reform often comment that these sectors are more different than
similar (Corrales 1999; Grindle 2004). Chapter 6 will further explore the puzzle behind
these similar findings and will address the opportunities and limitations for generalization
about diffusion across policy arenas.
As King, Keohane and Verba argue, one of the benefits of quantitative analysis
based on a large number of observations is the greater leverage garnered to test theories
and draw causal inferences (1994). Certainly the event history models in this chapter
reinforce the basic argument that ideology and social networks spur diffusion in Brazil’s
largest cities. However, it is also important to acknowledge the limitations of this method.
At the most basic level, the models used in this chapter involve basic assumptions that
can influence our analysis. First, the models require a dichotomous categorization of
policy adoption – cities either do or do not have a given policy. Yet, in practice,
policymakers may modify or adapt policies for local environments. The “adaptability” of
a program may or may not relate to its overall diffusion pattern. Second, these models
require dropping observations for cities once they have adopted a program. This analysis
certainly allows for understanding what leads up to the decision to emulate a policy, but
assumes that policy reversal either does not happen or is irrelevant. Yet, in Brazil, social
policies are often changed, adopted, and reversed when new elected officials take office.
Given some of these methodological limitations, it is important to draw on qualitative
data to elaborate on the statistical findings and clarify the causal mechanisms that drive
This chapter offers valuable insights into the relationship between the variables of
interest and policy diffusion, yet the analysis also raises a number of questions that will
be addressed in subsequent chapters. First, to what extent do the proxies used to measure
political incentives, ideology, and social networks accurately reflect the real-world
political dynamics of policy emulation? For instance, political incentives could play out
in a number of non-electoral ways, including the use of clientelism or patronage. Are the
findings here simply a reflection of measurement error or limited data? Second, why is it
that leftist mayors would be so much more likely to adopt Bolsa Escola and Programa
Saúde da Família? While it is true both programs have equity-enhancing goals and seek
to improve access to education and healthcare, they also appeal to a broader range of
policymakers who like the targeting and “contract” features of these programs. In
addition, both programs can represent socially conservative objectives that reinforce
traditional gender roles. Thus, there is a need to clarify how these programs are
interpreted and how ideology affects decision-making and program emulation. Lastly,
the statistical analysis alone cannot reveal the way in which a formal “social network
presence” matters. What is it about network connectivity that compels emulation? How
do networks shape professional norms? The following chapters, which rely on case study
evidence from four Brazilian cities, will answer these questions.
Why is it that a relatively simple idea, an education grant for poor families to
increase children’s educational performance, appealed to municipal executives? Bolsa
Escola is not a cure-all for Brazil’s educational woes, yet many policymakers across the
country were enamored with the program. In what ways did the program appeal to these
policymakers? And what motivated municipal authorities to emulate the program for
their own cities? To answer these questions, this chapter first contextualizes Brazil’s
national educational deficits and delayed reforms. The second section addresses how
development of conditional cash-transfer programs like Bolsa Escola. The third section
draws on case study evidence to uncover the mechanisms that drove emulation decisions.
Interviews with policymakers from four research sites reveal why they chose to emulate
Bolsa Escola. Although Bolsa Escola lends itself to electioneering behavior, I show that
actors chose to emulate the policy because of their deeply held ideological convictions
and desire to keep up with their profession’s norms.
The abertura period that began in the late 1970s opened the door for discussions
on how best to improve Brazilian society and democracy. There was little doubt that
reform of the education sector was necessary as educational studies indicated Brazil
among the worst in the region. First, the system favored spending in secondary and
higher education, rather than prioritization of primary education. Since most children did
not progress to secondary or higher education, the vast majority of children were
underserved by the system. The socioeconomic segregation of education services also
meant the public system favored the needs of middle- and upper-class families, which
tended to send their children to prestigious public universities. The structurally unequal
spending in education contributed to inefficiencies, exacerbated by poor quality
education, contributed to Brazil’s growing inequality, and was an important factor in
persistent intergenerational poverty. Second, the country’s educational deficits raised
serious doubts as to whether Brazil could construct meaningful citizenship for all. Could
Brazil consolidate democracy with a population that had such low levels of educational
Education policy has continued to reflect the legacies of policies that favored
middle and upper income groups. Families with high incomes can bypass low quality
public primary and secondary schools, in favor of private schools which better prepare
their children for meritocratic university exams. While upper income families incur the
cost of private education at the primary and secondary levels, they disproportionately
benefit from education spending that includes free federal public university enrollment.
Low income families must rely on inadequate public education throughout, thus their
children encounter serious barriers to high education. Nearly sixty percent of university
students in Brazil belong to the top income quintile, while higher education accounts for
over 20 percent of total education spending. To put this figure into perspective, Brazil’s
expenditure for university students is nearly four times the average of countries in the
Organization for Economic Co-operation and Development (OECD 2004: 7).
Recent figures of Brazil’s contemporary education deficits have also troubled
policymakers. In ten years, the country had not made significant advances in literacy and
school completion. In the early 1990s, the combined repetition and dropout rates for
primary and secondary education were about 50 percent (Draibe 2004: 383). Data from
1994 also revealed that only 83 percent of children between ages 7 and 14 had access to
primary education; figures for secondary school access were even more dismal with less
than 20 percent of students with access (Draibe 2004: 383). Brazil’s low-quality
education continues to rank it among the worst in the region. It also places Brazil in a
worsening position vis-a-vis its international economic competitors. As Birdsall et.al.
note, in the 1960s the quality of Brazil’s education system of basic education matched
that of other countries with similar incomes. Yet, by 1990 Brazil had fallen behind with
lower average quality, particularly compared to high-performing East Asian economies
(Birdsall et. al. 1996:7-8). By the mid-1990s, it was clear that Brazil’s development
strategies and subsequent educational deficits would have profound social and economic
consequences for the nation and represented one of the most important arenas for reform.
Even though the need for education reform was clearly understood by both
Brazilian and international development experts, the period from the mid-1980s to mid1990s was marked by stalled reforms. Although the Constituent Assembly promoted the
right to education as a fundamental social right, the constitution only laid out general
principles for reform. Among them was the goal of decentralizing education, specifying
that municipalities should focus on pre-school and primary education, and states should
focus primarily on secondary education.
Particularly notable is the Constitutional
requirement that states and municipalities must spend 25 percent of tax revenue on
education; 60 percent of that sum must go to primary education (representing a minimum
percentage of 18 percent of tax revenue) (Ministry of Education: 2004:6). Despite the
fiscal and administrative outlines for decentralization, much of the specifics regarding
curricular reform, teacher training, student access, and funding equity would be left to
future congresses to address. Brazil’s delays in reforming the education sector at the
national level are far from surprising. As Corrales notes, education reform is difficult to
undertake; there are numerous political hurdles including: the concentration of costs on a
few actors, low incidences of policy entrepreneurship, political disengagement of
potential beneficiaries, and that cost-bearing groups often enjoy political advantages
(Corrales 1999: vii).
Institutional and historical tendencies have made education reform politically
unattractive for many politicians. Brazilian presidents often need to dole out cabinet
level appointments in a quasi-parliamentary manner in order to sustain coalitions with
other political parties. They typically negotiate ministerial positions with allied party
leaders who, in turn, agree to legislative unity in return for control over parts of the
This practice has contributed to de-prioritization of certain ministries,
among them the Ministry of Education and Culture (MEC). From 1985 to 1995 the MEC
was led by rightist Ministers (PFL and PL affiliates) who used education for political
patronage (Draibe 2004:380).
Thus, the strong tradition of clientelistic practices
remained intact throughout the period. Even if education ministers had been committed
to substantive reform, they would have required political insulation from the President to
implement change. As Corrales (1999) notes, one indicator of how political vulnerable
ministers are is the turnover rate. During Jose Sarney’s presidency (1985-1990), five
ministers held posts lasting on average a year, and during Fernando Collor’s Presidency
(1990-1992), the average duration of an education minister was 10 months.
Another factor that likely contributed to stalled reforms was the “low demand”
from the system’s beneficiaries. Unlike the health sector, the education sector has not
traditionally benefited from well organized civil society advocacy. As Draibe (2004:385)
notes, parents, who are normally advocates for their children’s educational interests, have
not formed or actively participated in fora such as: parent-teacher associations, city-wide
education councils, or school councils, to demand greater educational access and
improved quality.
Thus the beneficiaries of reform were largely absent from
policymaking venues. The absence of civil society organizing is in large part a product
of traditionally low participation in this sector, as well as the fact the large segments of
the middle-class exited from the primary education system in the 1970s, opting instead
for private schools that offered better quality education.
A significant turning point for reform in education came when President Fernando
Henrique Cardoso (PSDB) took office in January 1995.
Unlike his predecessors,
Cardoso retained tight control over the “social ministries,” including the education
ministry, and appointed a fellow PSDB partisan and member of his inner circle, Paulo
Renato Souza, to head it. Minister Souza would have an unprecedented tenure at the
MEC, lasting throughout Cardoso’s two-term presidency, and benefiting from some of
the political protection observers note is necessary to enact unpopular reforms. The
President’s willingness to insulate the MEC is notable because “demand” for reform on
the part of civil society had been and continues to be low.
The mid-1990s brought important reforms that would greatly change education in
Brazil. The General Law of Education (1996) and other specific laws and acts finally
detailed the responsibilities for education in the federal system (see Table in Appendix
E). Thus, decentralization, which had only been outlined in vague terms in the
Constitution, gained greater specificity. Changes included the decentralization of the
Ministry’s own programs in primary and secondary education, specification of criteria for
transferring resources to states and municipalities (based on a value per student), and the
passage of FUNDEF (Draibe 2004:390). Overall, the most significant changes would
take place in primary education, including municipalization,1 curricular modernization,
investment in teaching training, and decentralization of resources.
The FUNDEF2 (Fundo de Manutenção e Desenvolvimento do Ensino
Fundamental e de Valorização do Magistério – Fund for the Development and
Maintenance of Primary Education and Valuing of Teachers), represented the most
significant equalizing measure undertaken by the MEC during this period. First, it
altered the distribution of spending in education to prioritize primary education. Each
state would have its own FUNDEF-fund, comprised of 15 percent of all state and
1 In the case of education, municipalization involved the transfer of schools and their related administrative
apparatus from state to municipal control.
2 The Constitutional Amendment passed in September 1996 and went into effect on January 1st 1998.
municipal tax collection and constitutionally required transfers, which would need to be
used exclusively for primary education.3 The funds would then be allocated to each
school equally on a per capita basis. Second, although the per capita spending could
differ across states, the amendment ensured that greater equity across the country by
establishing a minimum threshold for spending per pupil.4 Poor states would receive
federal funds that were unable to meet the minimum spending requirement.
The combination of a federal ministry of education that was intent on devolving
programs to sub-national governments, together with increased resources for primary and
secondary schools, went a long way toward fulfilling the administration’s goal to
decentralize education. The new funding formulas by the MEC encouraged more
municipalities to take responsibility for primary education from state governments, as
they would have greater fiscal autonomy. Even low-income cities could do so, because of
FUNDEF’s equalizing effect. Moreover, MEC policymakers sought to create a program
that would encourage municipalities to go after and bring-in students because the
transfers were based on pupil enrollment rates, thus ensuring that all children were
enrolled in school. FUNDEF thus represented the fulfillment of a campaign and political
commitment to primary education, on the part of the administration (Interview, Sousa
Although the eight-year period under which Minister Souza oversaw the MEC did
bring about significant changes to the structure of education provision in Brazil,
3 Up to 60 percent of the fund can be used for teacher salaries; no minimum salary is stipulated.
4 In 2004, the minimum was R$564 per pupil in first through fourth grades and R$595 and fifth through
eighth grades (Presidential Decree 5.299, December 7, 2004).
instituting a Bolsa Escola or similar grant program was not among his top priorities. In
Sousa’s view, the FUNDEF was the mechanism that created a “supply” for education;
whereas policies like Programa de Garantia de Renda Mínima (PGRM) or Bolsa Escola
created a “demand” for education, which can only take effect after a solid supply is
available (Interview, Sousa 2004).
During the Cardoso administration, the MEC
administered two “demand-side” programs: the PGRM and Bolsa Escola Federal, which
replaced it, both of which drew from examples of innovations from local governments.
These federal initiatives had “starts-and-stops” and reflected bottom-up learning based on
municipal experimentation, rather than reflecting a “top-down” initiative developed by
the national government.
Although national efforts to reform the education sector largely stalled from the
mid-1980s to the mid-1990s, states and municipalities have been able to make
considerable advances by experimenting with new policies.
Certainly, once
decentralization and later FUNDEF were in place, municipalities had even greater
resources at their disposal to experiment with new innovative education models. The
Gestão Pública e Cidadania Program, which sponsors an annual innovations competition
for “good governance” practices, offers a useful vantage point from which to examine
larger trends in education. City administrators have submitted information on projects,
which included: developing programs for children with special needs; reform of
education administration and information systems; innovation of various classroom and
teaching methods; out of school activities to reinforce learning; programs focusing on the
environment and local development; inclusion of local culture in the curriculum; and
various outreach reading programs and libraries, including adult literacy efforts (Spink
2006:17). In general, many of these programs stayed within the standard domain of
education by either addressing learning or curriculum development, while others
expanded notions of teaching and education by emphasizing social inclusion or parental
The local governmental arena presented policy entrepreneurs with fertile ground
to plant their seeds for new ideas and programs. Interestingly, the policy entrepreneurs
who would lead the way to enactment of innovative “education” programs were not
strictly affiliated with the field of education. Rather, the idea for bundling education with
poverty alleviation goals would be developed by intellectuals, technocrats, and politicians
who wanted to address the pernicious effects of low education outcomes on poverty and
inequality, and the cyclical effect that poverty would have on educational attainment for
children. Two leading figures who served as “policy entrepreneurs” and later became
spokesmen for their replication are: Cristovam Buarque and Eduardo Suplicy. Making
sense of the origins of “Bolsa Escola” or “Renda Mínima” can be a tricky task. As in
many cases, ideas are seldom born from a single individual and policies often undergo
considerable adaptation before their implementation.
In addition, the perception of
policy “success” leads many individuals to seek credit for the innovation5. The task here
is neither to give political credit to one individual or another, but rather to present a sense
5 For more on the problem of assessing the origin of ideas, see especially Kingdon 1995: chap. 4.
of the murky landscape in which early experimentation occurred and how these programs
Since the 1970s, Brazilian economists have debated the merits and feasibility for
a guaranteed minimum income program (Programa de Renda Mínima).6 Although these
early intellectuals made important theoretical contributions to the debate on poverty,
inequality, and democracy, it was not until the 1990s that such a program would make
considerable advances in the political arena.
Eduardo Suplicy, a career politician and
economist7 by training, has been the most vocal advocate for a national program as a
mechanism for social redistribution of wealth and poverty alleviation. Once elected to
the senate in 1990, representing the state of São Paulo and affiliated with the Workers’
Party, he would propose legislation that would guarantee all individuals8 over 25 years of
age, who earned less than two monthly minimum wage salaries, a cash supplement.9
Suplicy’s proposal was heavily debated, with many technocrats and economists
questioning the potential economic effects such a program would have on inflation and
the economy, in general. Others argued the program was utopian and that Brazil was not
ready or even administratively capable of implementing it. Others still wondered if cash6 For instance, Antonio Maria da Silveira in 1975, and Edmar Bacha e Roberto Mangabeira Unger in 1978,
published on this topic (For discussion on the contribution of these early scholars, see Fonseca 2001: 99109; Aguilar and Araujo 1998:31).
7 Suplicy’s interest in a cash-based minimum income program dates back to his graduate training the
United States, where he learned about a Milton Friedman’s ideas for a negative income tax, and policies
such as Aid to Families with Dependent Children (AFDC) and later the Alaskan Permanent Fund Divided
8 Suplicy was a firm supporter of a universal program geared towards individuals, not families.
9 At the time, a minimum income was approximately C$ 45 thousand. The Bill specifically called for a
supplement for the difference between 2 minimum income threshold and actual income. In the event an
individual had no income, the supplement would not exceed 50 percent of the threshold to encourage
people to work.
grants were the proper vehicle for poverty alleviation, preferring other programs and
approaches. Despite these criticisms the bill gained enough traction to pass in the Senate
unanimously on December 16, 1991, only to then stall in the house as Presidents Collor,
Franco and Cardoso declined to support it. Suplicy would continue to champion the
legislation10 nevertheless and advocated for governors and majors to adopt variations of
his legislation at the local level.11
In the public sphere, economist Jose Márcio Camargo joined the debate on the
merits of Suplicy’s proposal when he published an influential opinion piece in the
respected newspaper, the Folha de São Paulo (1993).
He argued on behalf of a
framework for a social policy that would alleviate poverty in the short-term and resolve
multigenerational of poverty in the long term, largely supporting Suplicy’s efforts.
However, Camargo suggested several notable modifications to the existing Bill, including
the prioritizing families with school age children and making transfers contingent on
families regularly sending their children to school. These modifications would later
become the backbone for future Renda Mínima programs.
Given the lack of enthusiasm for a guaranteed minimum income policy at the
national level, local governments were a natural venue for advancement of the policy. In
1995 two cities simultaneously adopted minimum income programs: Campinas and
Ribeirão Preto, both in the state of São Paulo. The Campinas program, implemented by
10 An amended version of the Renda Mínima legislation was finally passed and signed by President Lula
on January 8, 2004. However, even at the bill’s signing, Lula and others noted they were uncertain whether
the legislation could ever be implemented.
11 Although local governments would take up programs with the name Renda Mínima, they often differed
from the original proposed by Suplicy. He nevertheless “consulted” with cities and “welcomed”
modifications, suggesting they represent a single idea (Suplicy, Interview 2003).
José Roberto Magalhães Teixeira (PSDB), is the more widely known of the two, having
received awards and undergone several evaluations12 (Fonseca 2001). Although the
program shares the name most closely affiliated with Suplicy, the program differed
considerably from the national legislation. First, Campinas’ Programa de Garantia de
Renda Familiar Mínima (PGRFM) policy was not geared at individuals but rather
prioritized families as a whole.13 Further, the program only served families in extreme
poverty and with children between the ages of 0 to 14.14 Second, unlike the national bill,
the local initiative required beneficiaries to ensure their children’s regular school
attendance among other requirements.15 Notably, although the Campinas program had a
strong educational component to the program, it was viewed more broadly to include
social assistance.
Though contemporaneous to the Renda Mínima program, Bolsa Escola originated
among participants of the Center for Contemporary Brazilian Studies at the Universidade
de Brasília (UnB). The interdisciplinary center, which was spearheaded by then Rector
Cristovam Buarque, provided a venue for faculty, students, and intellectuals to meet and
develop policy that could respond to Brazil’s most pressing problems.
participants quickly identified the problem of basic education to be an urgent issue, and
noted the strong correlation between poverty and high dropout rates (Aguiar & Araújo
12 The city is also home to nationally renown Unicamp (The University of Campinas), which has a well
known center in social policy evaluation, lending to both interest and ease for studies of the city’s
Programa de Garantia de Renda Mínima.
13 Lei No. 8.261, dated June 6, 1995 and regulamentado pelo Decree No. 11.471 of March 3, 1995.
14 Families had to have a per capita income of R$35 or below to meet eligibility requirements.
15 Heads of households signed a “Termo de Responsabilidade e de Compromisso”, which required the
following: the children’s regular school attendance, regularized attention to healthcare, and children could
not reside on the streets. In addition, families were required to participate in monthly meetings.
2002: 38). Based on this observation, participants concluded that that poverty and loweducational attainment were inter-related and positively reinforcing, and Bolsa Escola
could tackle both. Even though education is compulsory and free in Brazil, parents must
still provide a minimum level of resources for their children to attend school (e.g. clothes,
shoes, and school supplies). Furthermore, parents face an opportunity cost when sending
their children to school (i.e. they must forgo potential labor and income their children can
generate through informal work). Second, Buarque argued that just as government has
supported students through scholarship to attend institutions of high education (masters
and doctoral degrees), it would be appropriate for children to receive scholarships for
primary education. Thus, the group proposed a school grant (scholarship) for families in
poverty, but on the condition their children regularly attend school; families whose
children failed to attend school, would lose a monthly payment.
When Cristovam Buarque took office in the federal district of Brasília in 1995, he
quickly implemented a Bolsa Escola program.16
The program started small, first
prioritizing families in the neighborhood Paranoá, which had the lowest socio-economic
indicators in the city. City administrators later expanded it to include more families; by
1998, 25,680 families and 50,673 children were enrolled in the program (Aguiar &
Araújo 2002: 43). Eligible families included those whose income fell bellow half a
minimum salary per capita, the monthly benefit was fixed at one minimum salary
16 Decree 16.270 on January 22, 1995 and regimented by Portaria 16 on February 9, 1995.
(R$130)17; the program raised roughly 10,000 families above the poverty line (Lavinas
and Barbosa 2000: 449).
The city also implemented a complementary program,
Poupança-Escola (School Savings) as an additional incentive to encourage students to
stay in school until completion of the lower secondary school, and for those families with
older children who stay in school. In all, the Bolsa Escola program cost the district
approximately 1 percent of its annual revenue.
Some policy specialists argue that municipal Renda Mínima and Bolsa Escola
policies are substantially different policies because of their distinct origins, nuances in
policy design, and names (Paulics 2004; Interview Lavinas 2004; Interview Rocha 2003).
Table 4.1 provides an overview of both the similarities and differences between two
exemplary municipal programs: Campinas (SP) and Brasília (DF). While it is certainly
the case that each program displayed unique features, I argue that these programs are
essentially similar.
First, both programs incorporate two objectives within a single program: to
improve educational performance and alleviate poverty. Interestingly, although both
programs have important education goals, neither was conceived by education specialists,
but rather framed within the larger context of economic and social development. Second,
both programs have strict eligibility requirements based on family income and make
benefits conditional based on parents’ behavior (e.g. regular school attendance).
17 At the time, a minimum salary was approximately $76 US dollars. The benefit was fixed, regardless of
the number of children in the household.
Table 4.1 Characterics of the Municipal Bolsa Escola & Renda Mínima Programs
Income Eligibility
Benefit Amount
Administered by:
Renda Mínima
(Campinas, SP)
Families in extreme poverty,
incomes per capita below R$35,
and with children from 0 to 14
years of age.
Two years from date of the
Regular School Attendance
Regular Health Check-ups
Monthly Meetings
Head of household
The difference between actual
family income (per capita) and
the minimum income (R$35 per
Department of family, child,
adolescent, and Social Services.
Bolsa Escola
(Brasília, Federal District)
Income per capita below ½
Minimum Salary (R$130), and
with children ages 7 to 14 and
matriculated in school.
Five year residency in Brasília.
Regular School Attendance
(90% attendance rate)
Mothers of Children
1 Minimum Salary (Flat)
Department of Education
Third, both policies provide cash grants and parents can determine how best to spend
those resources. Fourth, both Renda Mínima and Bolsa Escola represent a flexible policy
that can incorporate additional components. For instance, Campinas’ Renda Mínima had
healthcare requirements and Brasília added parent literacy programs; one of the important
elements of the program was that it appealed to technocrats who wanted to have a more
integrated approach to social services. Lastly, and perhaps most importantly, the policy
entrepreneurs who have been most vocal in advocating for replication of their respective
program, have acknowledged that in practice, Bolsa Escola and Renda Mínima are
essentially similar (Interview Buarque 2004; Interview Suplicy 2003). For all these
reasons, this project examines the diffusion of Renda Mínima and Bolsa Escola as part of
the same phenomenon.
Both Bolsa Escola and Renda Mínima policies received early recognition for their
“innovativeness” and won numerous awards. For instance, in 1996, the first year of the
Innovations Award Program administered by the Public Management and Citizenship
Program of the Fundação Getúlio Vargas, city administrators from Brasília and Campinas
submitted applications for their respective programs; both received awards that year.18
Bolsa Escola in particular, became the “darling policy” of both Brazilian and
international development policy specialists who supported assessments of the program.
For instance, the international organization UNESCO office in Brazil produced an
evaluation and was an early enthusiast of the program (Waiselfisz et al. 1998). In funding
the evaluation, UNESCO introduced the program to policy professionals in the national
and international arena. A few other domestic and international organizations followed
suit and produced policy papers on Bolsa Escola (Bava et. al 1999; Lavinas et al. 2001;
Lobato & Urani 1998; Vawda n.d.). Although UNESCO19 and the World Bank never
directly supported the policies through financial contributions in Brazil, the program was
consistent with their organization’s values and priorities. This is particularly true in the
case of the World Bank, which had been advocating for more efficient use of social
spending and for targeting to prioritize the poorest groups rather than creation of policies
that entailed universal coverage. In addition, the news media took considerable interest
in Bolsa Escola and Renda Mínima programs, as cities began receiving awards. Major
18 In 1996, the Campinas program received a semi-finalist award and the Brasília program received the top
finalist award. Bolsa Escola also won the award, Criança e Paz (Children and Peace), from UNICEF.
19 Interestingly, of all the organizations to evaluate and support the earliest Bolsa Escola efforts, only
UNESCO was a truly education-oriented institution; all others are generalist organizations that focus on
“good governance,” poverty alleviation, and development.
news outlets, the Correio Braziliense, Jornal do Brasil, Estado de São Paulo, and weekly
news magazine Istoé, ran articles and opinion pieces on Bolsa Escola.20
Aside from the wide recognition that Bolsa Escola’s received, part of the
program’s broad appeal was due to its very policy design. Although some of the staff in
Brasília who designed the particularities of the program had a feminist perspective on
social policy21 – for instance having known about microcredit programs for women in
Bangledesh such as the Grameen Bank – the program still appealed to mainstream
Brazilians who hold more conservative views on women’s roles. In targeting women,
policymakers designed a program that was largely consistent with traditional gender
norms and notions of maternity. As Marisa Pacheco, the coordinator of Bolsa Escola
explained, the decision to target the payments to mothers was well received and noncontroversial.
We thought it was important to recognize the culture here in Brazil.
Women take a more active role in the family and with their kids. We
believe mothers are more likely to keep track of their children’s
attendance at school, make sure they dress well, eat well, etc. And I had a
strong belief that women would manage the resources of the Bolsa Escola
well (Interview Pacheco, 2004).
20 See for instance: Assunes 1997; Ibañez 1996, 1997; Dimenstein 1997; Geraldes 1999; Fernandes 2001;
Lago 2001; Silva 1996; Steck 1997; Jornal do Brasil 1997; Villaméa 2001; Mello 2000; Correio Braziliense
1997ª, 1997b; Rossi 1997; Husseini 1996.
21 In practice, Pacheco and Conseição Zotta Lopes had strong theoretical and administrative reasons for
directing the cash benefit to mothers. In their previous work in public housing, they noticed how
problematic it was when women were not included on deeds with their husbands. They noticed that without
explicit attention to women’s status, they could become even more vulnerable. In addition, there were
practical considerations. Women were more likely to have custody of children and were often heads of
households (Interview Pacheco 2004).
In this way, Bolsa Escola reinforced notions that women would be more responsible
because of their maternal roles. Men were perceived to be less trustworthy whereas
women were thought to be self-sacrificing and would put their children first. Discussions
about “women’s roles” and “women’s work” were an explicit part of the policy design,
which Governor Buarque was quick to use.
One justification I used for focusing on women was that we needed to
value the work women do as mothers and that the state had an obligation
to support this work. Buarque liked this idea and even pushed it a little
further, saying that we were generating work for women. Basically,
spinning it as a payment for women for work they already do (Interview,
Pacheco 2004).
The implications of the program’s construction of gender norms generated very little
discussion, both in Brasília and later when it would be replicated elsewhere. As Aguilar
noted, the policy is attractive in part, because it does not attempt to restructure social
relations in a radical way (Interview 2003). Conservative segments in Brazilian society
appreciated that the program reflected views about women’s self-sacrificing nature while
progressive feminists remained largely silent.
Policy entrepreneurs such as Cristovam Buarque and Eduardo Suplicy were vocal
advocates for the expansion of Bolsa Escola and Renda Mínima programs. One of their
principal targets was the federal government and both men approached President Cardoso
and his senior policy staff to get a national program off the ground (Interview Buarque
2004; Interview Suplicy 2003; Interview Souza 2004). Despite their efforts, the Cardoso
administration was primarily committed to other social policy approaches, such as
changing education financing through FUNDEF. In a concession and despite different
priorities, in 1997 the federal government introduced a national program – Programa de
Renda Mínima Vinculada à Educação (Lei 9.553, December 10, 1997). The policy
supported municipal efforts for Bolsa Escola and Renda Mínima by providing a matching
grant to those communities that instituted their own program; the federal program would
cover 50 percent of beneficiaries’ payments. However, the program was limited in scope
and focused on the poorest cities. Only those cities with per capita incomes below their
states’ average were eligible to participate.
The federal Programa de Renda Mínima Vinculada à Educação was short-lived.
Renda Mínima and Bolsa Escola enthusiasts believed the program was doomed from the
start and represented a half-hearted attempt to promote the program’s expansion. Federal
technocrats assumed that offering funding would be sufficient incentive for local
governments to participate in the program. But eligible cities were often smaller and
poorer and had limited capacity to establish their own program. In practice, a few state
governments, such as the state of Bahia, urged their municipalities to adopt Bolsa Escola.
But the vast majority of eligible cities never participated. Overall, it represented an oldfashioned design that required formalized cooperative agreements between municipalities
and the federal government (Interview Pesaro 2004).
In 2001 the federal government
would make a second attempt to support conditional-cash transfers through the program,
Bolsa Escola Federal. In this iteration, federal authorities would by-pass municipal
governments altogether by directly paying poor mothers a school grant. (For more
information on federal conditional-grant programs, see Appendix F).
Predictably the enactment of the federal program in cities that had their own
municipal Bolsa Escola and Renda Mínima programs posed especially challenging.
Federal administrators and politicians criticized these local governments for failing to
enact their program in a timely manner and accused politicians from opposition parties of
political posturing and purposeful delays.22
Local officials complained of being
railroaded by the federal government (Interview Leitão 2004). Since they had their own
experiences, these local officials argued they should be granted greater flexibility and
given the option of integrating the municipal and federal programs. Technocrats were
especially concerned about the registration process and wanted to insure that the
programs would extend coverage to new families rather than provide overlapping
benefits. In the end, cities that had already established municipal programs continued to
operate their own programs and simply added Bolsa Escola Federal, essentially operating
two separate programs. In other words, municipalities across Brazil continued to design,
administer, and finance their own Bolsa Escola programs despite the complications
caused by the entry of a federal program bearing the same name.
Given the early enthusiasm for Bolsa Escola and Renda Mínima programs, with
awards and enthusiastic reviews from international development agencies, it is not
22 The politics related to São Paulo’s enactment were particularly intense as it became a battleground
between two prominent politicians, Mayor Marta Suplicy (PT) and Education Minister Paulo Renato Souza
(PSDB). Each politician defended the merits of its program. Mayor Suplicy offered a “compromise” that
included an integrated program and the municipal governments’ logo on the Federal Bolsa Escola debit
card. MEC officials took that as a sign of clear partisan politics and dismissed the idea decisively.
Billboards later went up across the city detailing how much money low-income citizens were losing
because of the Suplicy administration delayed enactment of the Federal Bolsa Escola.
surprising that other cities would choose to replicate these programs. Some cities across
the country were so quick to adopt similar programs they did so within a year of Brasília
and Campinas’ enactment. For instance, Salvador emulated the program before major
research organizations had widely distributed policy evaluations.23 That cities replicated
the programs so quickly and before assessments of these policies established their
effectiveness is remarkable; particularly since most of the earliest publications on these
programs were based on case studies (i.e. usually based on one or two cities), and it was
unclear whether cities with different socio-demographic characteristics, educational
difficulties, or financial resources could benefit from adopting a similar program. Thus,
what explains why certain cities were so eager to replicate these programs? Why were
some cities quick in doing so? Why did others take a slower pace to adoption? And,
why did other cities choose not to replicate the programs at all?
Case Studies
This chapter details case study evidence on the motivations behind decisions to
adopt (or not adopt) Bolsa Escola and Renda Mínima policies. Table 4.2 presents a
synopsis of the replication decisions across the 10 cases in this study.
23 Rocha noted that cities such as Belo Horizonte and Belém replicated the Brasília program wholesale;
there were few adjustments made to target the particularities of their cities (Interview 2003).
Table 4.2 Ten Case Studies: Adoption & Non-Adoption
Brasília (DF)
Belo Horizonte (MG)
Executive in Office & Party IDa
Joaquim Roriz (PTR)
Cristovam Buarque (PT)
Joaquim Roriz (PMDB)
Patrus Ananias (PT)
Célio de Castro (PSB)
Célio de Castro (PSB)
Fernando Damata Pimentel (PT)e
Bolsa Escola/Renda Mínima
Salvador (BA)
Lídice da Mata (PSDB)
Antônio José Imbassahy (PFL)
Antônio José Imbassahy (PFL)
São Paulo (SP)
Paulo Maluf (PDS)
Celso Pitta (PPB)
Marta Suplicy (PT)
Mayor’s partisan affiliation at the time he or she ran for office.
Brasília, the Federal District , operates under the gubernatorial electoral calendar.
As the originating city for Bolsa Escola, its adoption in 1995 does not constitute a case of diffusion.
The program was suspended or discontinued and then reintroduced under new names.
Fernando Damata Pimentel (PT) assumed office in November 2001, after Célio de Castro suffered a
There are several broad patterns across the 12 cases that are worth noting at the
onset. First, adoption of Bolsa Escola occurs at different points in time; each city adopts
the policy during different administrative cycles. After Brasília’s innovation Salvador
was the first city to replicate it, followed by Belo Horizonte and then São Paulo. Second,
all of the administrations to adopt Bolsa Escola were left-of center, yet not all cities
governed by a leftist adopted the program. For instance, Belo Horizonte (1992-1996) had
a Workers’ Party mayor who did not adopt the policy. This suggests perhaps that leftleaning majors are necessary but not sufficient for emulation of this program. Third, one
of the most striking features of Bolsa Escola replication is that it did not necessarily
“stick”. There are several instances of policy reversal – Brasília and Salvador – where
programs were suspended following the start of a new administration. Thus, the case
studies provide an opportunity to examine both the determinants of policy diffusion as
well as the reasons for policy reversal.
Political Incentives
A political incentives approach offers an intuitively appealing explanation for the
spread of Bolsa Escola in Brazil. Policy advocates for Bolsa Escola and Renda Mínima
suggest these programs spread because they are politically attractive (Interviews Buarque
2004; Suplicy 2003). In what ways does the nature of the policy itself allow for a
political incentives explanation of its adoption? Does political competition drive local
politicians’ day-to-day decision-making? Do political leaders decide to adopt these
programs to win elections? Do political incentives explain the timing of diffusion?
In many ways, Bolsa Escola is the type of policy that calculating politicians are
eager to adopt in order to sustain their electoral popularity.
Given that Brazil has
compulsory voting, Bolsa Escola, which targets poor constituencies, can be especially
useful in electoral politics. To contextualize the potential impact of the poor’s vote, in
Salvador in 1991 approximately 35 percent of the population had monthly per capita
incomes of 75 reais24 (Martins & Libâno 2005). While figures for the other case study
cities is less dramatic – Belo Horizonte 19 percent; Brasília 17 percent, and São Paulo 8
percent (Martins & Libâno 2005) – this population is still substantial enough to sway
24 This figure represents half a minimum monthly salary (in 2000).
elections in competitive races. It is only natural to imagine that families that receive the
cash grant and rise out of extreme poverty would choose to reward the politicians that
backed the policy.
Even though Bolsa Escola targeted a specific group of beneficiaries and
distinguished between the “haves” and “have nots,” the policy did not exacerbate class
cleavages and was widely appealing across income groups. Several factors contributed to
its wide acceptance. First, the school grant program was similar to other social assistance
programs local governments have long offered to the poorest and most vulnerable.
Municipalities have long provided poor families with school uniforms, school supplies,
and food baskets, albeit on an ad-hoc and irregular basis. Others liked that the program
could empower the poor to manage household resources and take responsibility for their
children’s education. Among others still, there was speculation that increasing cashresources to local economies could have positive economic effects. While Bolsa Escola’s
policy design differed from previous public assistance efforts, it fit in line with historic
municipal efforts to alleviate poverty and assist the poor.
Electoral competition is an important feature of local politics for Salvador, São
Paulo, Belo Horizonte and Brasília and we would expect that candidates for executive
office and current mayors would use their policy position to attract and retain voters.
Certainly, the pressure to win elections and distinguish oneself from one’s competitor is
crucially important in the Brazilian municipal arena. Though personalism is still a
characteristic of Brazilian politics, candidates do refer to their policy preferences or
specify programs they would enact once in office. Campaigning on the provision of
these social programs can offer a clear opportunity to gain votes and clarify the field
when voters are faced with numerous candidates. So how did Bolsa Escola come to play
in the electoral arena?
Despite the name recognition of the policy, few candidates actively campaigned
on their intentions to implement program.
Those who did were leftist candidates,
including Cristovam Buarque, Marta Suplicy, and Célio Castro. (Lidice da Mata, who
also implemented the program, only became familiar with the program after she was in
office.) Their centerist and rightist counterparts however, did not, despite the potential to
garner a similar benefit for publicly supporting the program.
Interestingly, when the time came to implement the policy, Bolsa Escola
administrators perceived the program to be risky. As a coordinator of the program in
Belo Horizonte explained:
When executives carry out effective programs and the population views it
favorably, it can result in votes. But, I think it was difficult at the time to
determine what the electoral payoffs would be (Interview Leitao, 2004).
There were several administrative uncertainties that gave technocrats pause. First, Bolsa
Escola could increase school attendance, but by incorporating previously marginalized
and failing students into the system, other performance indicators would likely decline.
Second, Bolsa Escola was actually disliked by teachers who were skeptical about the
program and its benefits.25 As Marisa Pacheco noted, the BE is not a singular solution to
education. “Once you bring children into the classroom, other problems arise, including:
25 Merilee Grindle offers a useful analysis on the ways in which teachers’ unions have perceived
themselves as “losers” when it comes to education reform in Latin America (2004).
limited classroom space, the need to develop strategies to help students catch up, and the
need for improved teacher instruction” (Interview Pacheco, 2004). But the program was
also unpopular in terms of traditional corporatist politics; teachers’ unions preferred that
education policies enhance teacher pay and classroom supplies (Interview Aguilar 2003).
Despite these challenges, the Buarque administration proceeded in instituting the
pioneering Bolsa Escola program. Mayors Célio Castro in Belo Horizonte and Lídice da
Mata in Salvador also signed on to Bolsa Escola very early on, in spite of the risks of
policy failure. Only Marta Suplicy, who emulated the program in 2000 in São Paulo,
would benefit from having a cadre of technocrats already familiar with similar municipal
For these case study cities, the replication of Bolsa Escola offers some surprising
findings. In theory, the education stipend could have been attractive to a broad set of
politicians. The program could have been a natural extension of “politics as usual”
practices, including patronage politics, clientalism, and pre-election payoffs. But only
left-of-center mayors choose to emulate Bolsa Escola, suggesting that politicians made
their policy choices based on other factors. Given that these programs cost municipalities
their own resources and that federal incentives27 were non-existent, there must have been
26 Ana Fonsceca, the director of Renda Mínima in São Paulo, had conducted evaluations of similar
programs in Belo Horizonte, Brasília, Campinas, and Salavdor. At the time, she held an academic post at
27 None of the cities that adopted municipal Bolsa Escola or Renda Mínima programs benefited from
federal funds. Vertical diffusion – such as that caused by federal inducements through financing – was not
a feature of municipal Bolsa Escola replication. Belo Horizonte, São Paulo, and Salvador all created their
own municipal programs without federal subsidies demonstrates that horizontal diffusion across cities does
take place. Surprisingly, even when the federal government later initiated its own program, many cities
including Belo Horizonte and São Paulo balked at the chance to integrate their municipal program with the
something else that drove mayors da Mata, Castro, and Suplicy to emulate the education
Traditional ideological divides between the left and the right had a particularly
strong impact on the adoption of Bolsa Escola and Renda Mínima programs. Politicians
to the left-of-center, from the Workers’ Party (PT), Brazilian Socialist Party (PSB), and
the Party of Brazilian Social Democracy (PSBD), tended to emphasize social programs in
their campaigns and policymaking. Elected officials revealed a dramatically consistent
framing of ideological objectives and values when prioritizing issues and selecting public
policies. Nearly every politician and technocrat from the Workers’ Party, for instance,
justified his or her policy choices with notions of “social rights,” governmental
responsibilities, and the need to invert spending to prioritize the poorest and most
vulnerable sectors of the population.
For many actors ideologically to the left-of-center, Bolsa Escola and Renda
Mínima represented a profound transformation in the relationship between the state and
citizens. In their analysis, public assistance programs had historically reflected traditional
clientalistic approaches to social assistance. These programs were often administered by
the wives of mayors who took them on as part of their charitable first-lady obligations,
regardless of whether she had professional credentials in the field. Thus, critics on the
left argued that municipal-run programs that offered hand-outs, such as electronic
federal one, which would have allowed them to reduce municipal expenditures. They also maintained their
own municipal programs even after their cities began administering the federal program.
appliances or baby clothes, were more often than not vehicles for vote-buying.
Moreover, they also failed to address the causes of poverty. In contrast, advocates of
Bolsa Escola and Renda Mínima argued their program would give children a chance to
get out of poverty, while also empowering mothers to decide how to spend the grant.
Bolsa Escola program coordinators displayed remarkable convergence of ideological
discourse around these general themes. They also expressed a desire to address social
exclusion and a belief that education was an important component of citizenship. When
politicians discussed why they had chosen to adopt a school grant program, they all cited
problems like social inequality and the need to address the “social deficit.”
Cities governed by executives from the right-of-center parties took a very
different approach and mostly ignored Bolsa Escola and Renda Mínima proposals. In
general, right-of-center mayors emphasized policies that encouraged business interests or
market competition and enacted policies that were framed along these conservative
rationales. Unlike their left-of-center opponents, conservatives’ political campaigns often
highlighted and prioritized their progress in non-social policy arenas. In campaigns for
re-election for instance, Mayor Antônio Imbassahy in Salvador emphasized his
administration’s accomplishments in infrastructure projects, while Governor Joaquim
Roriz in Brasília highlighted the construction of an award-winning bridge.
The different discourse of actors on the left and right could be easily dismissed as
a rhetorical device were it not for the fact that left-of-center politicians were consistent in
their follow-through and implementation of Bolsa Escola. In the case studies, emulation
of Bolsa and Renda Mínima occurred under left and left-of-center politicians: Belo
Horizonte under Célio Castro (PSB), São Paulo under Marta Suplicy (PT), and Salvador
under Lídice da Mata28 (PSDB). Some politicians were so committed to the ideals
behind Bolsa Escola and Renda Mínima that they implemented and defended the
programs in ways that perplexed even their own allies.
Longtime advisors to Governor Cristovam Buarque and Mayor Lídice da Mata
admitted they could not logically explain the actions taken by their candidates. For
instance, Mayor Lídice da Mata implemented the Programa Renda Mínima Familiar in
her last year of office even though it was clear she would lose her bid for re-election and
understood that her successor would most likely dismantle the program once his term
began. She also faced criticism from her supporters and inner circle of confidants, who
argued that a Renda Mínima program was only feasible for cities flush with resources;
they argued Salvador faced too many deficits for this type of specialized effort. Even so,
she went ahead out of principle because she was personally committed to the goals of the
program (Interview, Mata 2004). Cristovam Buarque also deviated from instrumental
political rationality in a way that could only be understood as grounded in his ideological
commitments. During his campaign for re-election in 1998, he did not reach out to the
mothers of Bolsa Escola and consistently told his audiences that the social programs
enacted during his administration were part of the state’s obligations and constituted their
social rights. Accordingly, he told beneficiaries of programs like Bolsa Escola that they
did not owe him their votes and should feel free to vote for whomever they wished
28 According to Coppedge (1997) the PSDB is a centrist party. Lídice da Mata’s own ideological
dispositions place her on the left-of-center. After her term as mayor of Salvador, she joins the leftist party,
(Interviews Buarque 2004; Ibañez 2003). The Buarque campaign staff admitted his
ideological speeches confused voters and contributed to his electoral defeat (Interview
Aguiar 2003). These examples of seemingly irrational decision-making by Lídice da
Mata and Cristovam Buarque confirm the hypotheses that some politicians are indeed
driven by their own deeply held values and will make decisions that go against their own
Right-of-center politicians and their senior staff also displayed their own
ideological tendencies when it came to the social policy development. In general, right of
center politicians and their politically appointed technocrats did not mention a “social
deficit” when discussing their policy priorities.
Rather, administrators and political
appointees associated with the Imbassahy (PFL), Roriz29 (PMDB), Maluf (PDS) and Pitta
(PPB) administrations emphasized market-oriented priorities for economic development,
including tourism, business development, and major public works.
Given their
ideological predispositions, it is not surprisingly that when left-leaning mayors lost their
bids for re-election, their successors dismantled their predecessor’s education stipend
programs. For instance, in Salvador, Mayor Imbassahy (PFL) simply dissolved the
Programa de Renda Mínima Familiar. The city’s secretary of Social Development,
Raimundo Caires Araujo, noted that social policies were not among the mayor’s top
priorities and the secretariat for social assistance had a limited budget (Interview 2004).
For this reason, their projects were small and often included sponsorship from private
29 According to Coppedge (1997) the PMDB is a centrist party. Joaquim Roriz’s own ideological
dispositions place him on the right.
firms. In general, the city of Salvador administered federal programs (such as PETI,
Bolsa Escola Federal, Agente Jovem), a few municipal social assistance programs30, and
targeted projects such as a youth orchestra and small-scale cooperatives where
participates would develop arts and crafts goods for sale (Interview Araújo, 2004).
Mayor da Mata’s Renda Mínima program was so short lived and absent from the public
memory that Secretary Araujo acknowledged he was unfamiliar with it. He also admitted
that it had not occurred to him to institute a municipal Renda Mínima program. While
staff in Salvador who worked on social policy were committed to their public assistance
work, what was most striking was their rhetoric. Or rather, what was notable was the
absence of a “left” rhetoric. City officials never mentioned legacies of exploitation,
social exclusion, racism, or lack of citizenship. Nor did they frame their work in the
context of democratic practice or empowering the poor.
The administration of Governor Roriz (PMDB) from 1998-2002 in Brasília offers
a parallel account. As in Salvador, when the right-of-center governor entered office in
1998 after defeating left-of-center Buarque, he quickly moved to terminate the Bolsa
Escola program. His staff declared the program unnecessary because the district did not
have problems with irregular school attendance but rather low academic performance.
The district suspended enrollment of new families in the Bolsa Escola program and
designed an alternative program, Successo no Aprender (Success in Learning), that
provided students with school uniforms, school supplies, eye exams, and extra classes on
30 For instance in 2000, the city provided needy citizens with a “cesta baxica” (food basket), containing
basic goods such as rice, beans, noodles, etc. Shortly thereafter, the municipality began working with
grocery stores to enable beneficiaries to use supermarket cash cards for food purchases.
Saturdays, eliminating the cash-grant altogether.
In other words, the administration
returned to a more traditional mode of public assistance. At the time, the administration’s
decision to suspend registration into the Bolsa Escola program drew considerable
criticism from the news media but officials pressed on with their intention to evaluate
Buarque’s Bolsa Escola and create their own education programs31 (Interview Lima
In 2001, the Roriz administration provided the only instance in this study of the
Bolsa Escola reinstatement. The city reintroduced a school grant program under the new
name, Renda Minha (My Income), combining elements of both Bolsa Escola and
Sussesso no Aprender. The director of the program, Lílian Carneiro Lima, downplayed
the notion that politics or media pressures led to the decision to reintroduce the program;
rather, she emphasized that their decision reflected policy evaluations and assessments of
various programs (Interview 2003). Unfortunately, it is difficult to disentangle these
actors’ motivations for re-instituting a education stipend program. Tracing the internal
decision-making process of the Roriz administration is particularly difficult due to a lack
of transparency.
Public officials associated with the Renda Minha program were
reluctant to discuss internal processes, provide documents, or discuss the number of
Several technocrats explained that unless their supervisors granted
approval, they were barred from providing “private internal documents” as they were not
“public.” Higher-level officials also refused requests for interviews. Nevertheless, it is
31 Officials disregarded the evaluations by the Fundação Getúlio Vargas-São Paulo, UNESCO, and the
World Bank and commissioned their own study. The study the Roriz administration commissioned was not
made available.
possible to conclude that the decision-making process for reintroducing Renda Minha in
2001 was considerably different from that of Bolsa Escola in 1995.
Unlike his
predecessor, Renda Minha was not a major symbol of Governor Roriz’s administration.
The policy had less visibility both in terms of his politics and personal discourse (e.g. in
campaigns and in the media). One implication of this case may be that left-of center
ideology matters more for instances of first-time policy emulation than it does in the rare
circumstances when policy reenactment occurs.
Ideology offers an important lens for understanding why politicians and their
senior staff implemented Bolsa Escola. Politicians and their politically appointed senior
staff shared similar dispositions and worldviews about the relative importance of social
policy. These individuals shared likeminded commitments to prioritize policies that
would address the long-standing social inequalities and persistent poverty. While left-ofcenter political actors generally shared similar partisan affiliations, such as the PT,
PSDB, and PSB, they noted that their decision-making was independent of partisan
directives. In fact, early adopters of Bolsa Escola noted that their emulation decision
preceded their party’s decision to endorse the program (Interview Buarque 2004;
Interview da Mata 2004). As Buarque explained, the policy itself was in line with a
subset of members of the Workers Party who favored a progressive vision of social
policy, but in the early and mid-1990s, the party itself was reluctant to officially endorse
the policy (2004).32
32 In Buarque’s assessment, in the early and mid-1990s the Workers Party was comprised of three distinct
streams: 1) those connected via unions; 2) those who interested in economic issues; and 3) those who were
For most cities, the timing of Bolsa Escola and Renda Mínima emulation
coincided with changes in administration as left-of-center mayors took office. The only
instance where a city with a leftist administration did not adopt the program was in Belo
Horizonte under the Patrus Ananias administration (1992-1996). This suggests that while
a left-of-center ideological commitment is necessary for emulation to occur, it might not
be sufficient. Early adoption in particular, requires that actors learn about innovations
quickly. In this way, decision-making might entail more than a self-regarding decision
process by include a socializing process as well. The next section on social networks
explores the extent to which emulation decisions reflected a process of social networking.
Social Networks
Civil society organization can serve a crucial function by creating opportunities
for formal networking and learning. Education policy has been a central thematic interest
among Brazilian associations; according to the Ministry of Justice, there are nearly two
hundred public interest non-governmental organizations whose primary focus is
Yet, one of the important features of this sector is the way in which
traditional corporatist interests, represented by teachers’ unions, have retained their
influence in larger policy debates. In practice, teachers have taken a more narrow view
of education policy, focusing on debates regarding curriculum development, pedagogy,
interested in social priorities. PT affiliates who had commitments to social issues were more interested in
Bolsa Escola. Since the PT has a São Paulo bias, it took a while for Bolsa Escola to garner the attention of
the mainstream in the party (Interview 2004).
33 The Ministry of Justice monitors public interest civil society organizations in Brazil and provides a
directory of these organizations on this website.
textbooks, classroom conditions, teacher training, and teacher pay. Education stipend
programs challenged the notion of what constituted an “education policy,” as they
integrated components of poverty alleviation with education goals. In other words,
features of the policy itself created some cognitive dissonance for educators. But in
addition to these conceptual policy differences, Bolsa Escola represented a political
conflict over resources and funding priorities. Unions generally worried that the funds
for the program would come from allocated set-asides for primary education, which
federal law mandates, rather than municipalities’ general operating budget. One of the
ironies about Bolsa Escola is that Brazil’s most internationally recognized education
policy was hardly a central issue among education professionals.34
Despite the low levels of interest in Bolsa Escola among education professionals,
there were a few formal associations and avenues for learning that were important.
Quasi-governmental associations such as Conselho Nacional de Secretários de Educação
(CONSED) and União dos Dirigentes Municipais de Educação (UNDIME) were key
institutions for UNESCO officials who wanted to engage in education policy
development (Interview Cunha 2004).
An award from UNICEF in 1996 also lent
international credibility for the policy. Yet, one of the most important organizations to
contribute to the spread of education stipend programs in Brazil was not specifically an
“education” association, but a generalist policy entity: Programa Gestão Pública e
Cidadania (Public Management and Citizenship Program), housed in the prestigious
34 When Cristovam Buarque lost his bid for re-election, rather than work with an existing education
organization, he established his non-governmental organization. Missão Criança was created to promote the
spread of Bolsa Escola in Brazil and worldwide.
public management school of the Getúlio Vargas Foundation in São Paulo.
As a
UNESCO official explained, it should not be surprising that Bolsa Escola would appeal
to a generalist policy audience:
Bolsa Escola appeals to a lot of different people… I think that as an idea,
BE was most attractive to people in non-education sectors; people who
work on social policy, poverty, social assistance, etc. You see, I always
consider education to be a very conservative field. There are three
conservative institutions in society: the church, the military, and the
schools… This phenomenon of being conservative, i.e. slow to change, is
very much present in education (Interview Cunha, 2004).
The Public Management Citizenship Program emerged as a particularly important
organization among generalist policy professionals by socializing them to follow the
latest trends in their field. Part of its influence relates to its very institutional design,
which includes a dissemination strategy to publicize award winning good governance
programs. They also hold public awards ceremonies so that national and local press can
provide media coverage on the finalists.
The program also has a general outreach
component that includes working with the press, producing of videos, and developing
materials for municipal, states, and federal use. Since it is housed in a school of public
management, the staff also write books and case studies based on the award winning
entries and hold thematic conferences for practitioners and scholars alike. From the
perspective of public administrators, participation in the program’s activities offers
several benefits. First and foremost, when administrators submit entries for the annual
innovations competition they gain recognition and visibility for their work.35 While some
elected officials and their political appointees encourage their staff to submit entries for
the competition, more often policy professional themselves seek out the legitimacy the
competition offers.
Winning the award also creates the potential that applicants’
programs will survive the turmoil that comes with elections and new executives.36 It also
provides civil servants with the affirmation they seek from their peers, that their work
meets the profession’s standards of excellence.
Both Brasília and Campinas received Public Management and Citizenship awards
in 1996 for their respective education stipend policies. In partnership with the NGO
Instituto Pólis, the Fundação Getúlio Vargas- São Paulo (FGV-SP) produced publications
describing Bolsa Escola and hosted conferences and meetings that featured officials from
Brasília. As Marisa Pacheco, the coordinator of Bolsa Escola in Brasília noted, the
Department of Education received many invitations to participate in conferences to talk
about Bolsa Escola. Those invitations were normally divided between Governor Buarque,
Secretary Ibañez and herself. Nearly ten years after she started directing the program,
Pacheco recalled that some of the most important venues for disseminating information
about the Bolsa Escola in Brasília were at the seminars held by the FGV-SP (Interview
2004). Civil servants in São Paulo, Belo Horizonte, and Salvador confirmed that they
35 All applicants receive a certificate from the program and all entries are included in a public database,
available on the Internet.
36 Policy continuity across different mayoral and gubernatorial administrations is relatively rare in the
Brazilian context.
were familiar with the awards program, had attended a meeting, or had received
publications from Public Management and Citizenship program.
In addition to the opportunities for formal socialization afforded by professional
networks, elected officials and policy professionals cited “informal” contacts as crucial
for convincing them to initiate change, oftentimes these developed in highly idiosyncratic
ways. Mayor Lídice da Mata decided to implement a Renda Mínima program after
hearing the Mayor of Campinas, a friend of hers, describe his city’s program at a
conference (Interview Mata 2004). The mayors knew each other well, as they were both
in PSDB and occasionally attended the same events. In Belo Horizonte, the first efforts
to institute Bolsa Escola originated from the city council, when Rogério Correia (PT)
proposed replicating it in 1996. Correia reported that as a fellow partisan, he had been
following Buarque’s campaign for Governor in 1995; as a fellow “educator” he took
special interest in his education proposals (Interview 2004). Rogério drafted legislation
to initiate a Bolsa Escola in Belo Horizonte, emulating every feature of the Brasília
program.37 He also invited Buarque to testify before the city council to explain the
program (Interview Correia 2004). As a well known politician in the Workers’ Party,
Marta Suplicy met Buarque on numerous occasions and learned about the well publicized
program, Bolsa Escola. But in Marta Suplicy’s case, her informal socialization process
37 The legislation for Belo Horionte is a replica of the program in Brasília, including defining eligible
families, the stipend, and targeting of women as beneficiaries. When it came time to implement the
program, municipal administrators retained these program features (Interview Rocha 2003; Interview
Leitão 2004; Interview Céres 2004).
occurred closer to home as her husband Senator Eduardo Suplicy, had been a major
supporter of local Renda Mínima initiatives.38
These informal network processes were not only important for elected officials,
but also for their political appointees and senior staff. Horizontal learning was a major
feature of Bolsa Escola emulation, as cities seeking to implement the policy visited
innovating cities. The Secretariat of Education in Brasília frequently hosted visitors from
other cities and states (Interview Pacecho 2004; Interview Ibañez 2003; Interview
Aguilar 2003). Officials in Salvador for instance, traveled to Campinas to see how the
Renda Mínima Familiar worked there. When Belo Horizonte took up the policy, Mayor
Celio Castro’s wife, who served as Secretary of Public Assistance, visited Campinas.
The city’s technical staff however, visited Brasília to learn how they had determined
eligibility for families and designed their registry (Interview Leitão 2004). Although
technocrats in Belo Horizonte conducted their own poverty assessment in anticipation of
the program, they closely followed Brasília’s plan. That decision was largely due to the
fact that Belo Horizonte’s legislation essentially copied Brasília’s plan. When Marta
Suplicy enacted a Renda Mínima program of her own, she tapped Ana Fonseca to direct
the program. As a scholar at UNICAMP, Fonseca had evaluated the Renda Mínima
program in Campinas and was aware of similar programs in other cities (Fonseca 2001).
In the case of Bolsa Escola, social networks served two important, albeit separate,
First, formal and informal networks socialized actors on the latest
developments and norms in their respective fields. Both technocrats and politicians
38 The couple separated in 2001 and divorced in 2003.
wanted to demonstrate they were aware of the latest trends and sought to gain legitimacy
among their peers. Being a “follower” or “emulator” for cities such as Belo Horizonte,
Salvador, or São Paulo was not perceived negatively. Technocrats and politicians would
simply emphasize their city’s unique features (e.g. high poverty rates, lower Human
Development indicators) and accomplishments (e.g. program size and speed of
implementation). Second, connectivity to social networks facilitated the learning process
by providing policymakers with cognitive shortcuts that enabled them to emulate policies
fairly quickly and with few adjustments. Even though the cities that adopted these
programs had highly skilled technocrats who could have tailored these programs for local
conditions, administrators largely engaged in wholesale replication of Bolsa Escola and
Renda Mínima programs.
These case studies of Bolsa Escola adoption reveal the way left-of-center
ideology and linkages to social networks both contributed to emulation decisions.
Similarly, instances of administrations that failed to replicate the education grant program
show that neither a mayor’s leftist ideology nor the presence of professional networks,
was by itself sufficient to bring about diffusion.
Both social norms and ideology
mattered by shaping actors’ motivations and reinforcing decision-making.
Implementation of the education program required that executives seek policies
consistent with their deeply held values. These politicians all held a desire to remedy
long-standing inequality and prioritize programs that would enhance citizenship by
alleviating social exclusion. If a leftist worldview was a prerequisite for emulation
decisions, it certainly was not sufficient to ensure adoption. For example, Mayor Patrus
Ananias (PT) in Belo Horizonte chose not to implement the program in his last year in
office, despite legislative efforts by leftist city council members. What also mattered in
all these cases was not only the presence of a committed politician, but also his or her
connection to a professional network.
Municipal executives often met one another
through formal events and informal contacts, both of which offered opportunities to share
information about the latest development and highlight their administrations’
Similar network relationships matter for technical staff and high-level technocrats.
Those individuals who worked on poverty and social development were often familiar
with educational stipend programs and could name those cities that were ahead of the
curve. Yet, technocrats’ desires to demonstrate their knowledge of professional norms
were insufficient to lead to Bolsa Escola emulation.
For instance, civil servants in the
office of Work and Social Development in the Antônio Imbassahy administration had
publications from the Public Management and Citizenship Office on their bookshelves.
They also identified Porto Alegre and Campinas as cities in the vanguard for designing
innovative social policies.39 That his administration had abandoned an education stipend
program was not for lack of technocratic socialization or knowledge, but rather, reflected
the absence of an ideological commitment on the part of the city’s leadership to do so.
39 Porto Alegre in Rio Grande do Sul, is well known for the Orçamento Participativo (Participatory
The central role that ideology and social norms hold for Bolsa Escola/Renda
Mínima emulation is surprising because it conflicts with so much of what we have come
to expect about the politics of redistribution. Local governments in Brazil are well
known for their history of local caciques who dominate the electoral arena through
political patronage. Education stipend programs like Bolsa Escola and Renda Mínima
could have had powerful electoral effects for politicians of all stripes. It would have been
logical for calculating policymakers to emulate these programs as a vehicle for selfinterested political behavior. Yet, these ten case studies demonstrate that, despite the
logic of a political incentives explanation for policy emulation, it is actors’ social justice
commitments and connections to their peers that matter.
Today, the Programa Saúde da Família (PSF) is widely accepted as a new model
for basic health service provision in Brazil. But to explain how and why it grew from a
small project adopted by a few local governments to widespread health care reform
across the country, this chapter explores the process by which it diffused and why
policymakers came to emulate it. In the early 1990s it was not at all evident that PSF
would come to represent a useful framework for all cities. After all, the program is based
on innovative experiments in preventive health that were designed for rather unique
Brazilian cities; the early precursors of PSF were community-based programs in the arid
poor state of Ceará in the Northeast, and the large city of Niterói, which displayed high
levels of human development. Yet, despite the unique features of the program’s earliest
adopters, the policy would eventually spark a tidal wave of emulations.
To explain why policymakers came to embrace PSF and were motivated to
emulate it in their own municipalities, this chapter first contextualizes national health
reform efforts since the democratic opening. Central issues during this time period
include real advances in social rights for health care access as well as stalled efforts to
fulfill state obligations. Decentralization nevertheless opened the door for municipal
experimentation in health policy in the late 1980s and 1990s. The second section of the
chapter provides an overview of state and local innovation, which laid the foundation for
the family health program. The last section draws on twelve case studies to uncover the
mechanisms that led to PSF emulation decisions. Qualitative evidence from interviews
with policymakers reveals how their decisions reflected deeply held ideological beliefs
and desires to seek professional legitimacy by following social norms. These findings are
remarkable given that the provision of basic healthcare can easily contribute to rentseeking behavior and yield electoral payoffs for mayors.
Historically, health policy in Brazil was tied to a larger public social security
system that included old age assistance. The origins of Brazil’s social security system
date back to the 1920s, when the state preempted an emerging working class movement
by granting social protections to selected sectors (Malloy 1979, chap 2 as cited by
Weyland 1996: 89). It was later extended by the authoritarian government of Getúlio
Vargas (1930-1945) to include more sectors of the economy into the state-corporatist
system (Collier and Collier 1979, as cited by Weyland 1996: 89). Like the Bismarckian
social insurance system, the Brazilian state tied pensions and health care entitlements to
worker and employer contributions. This corporatist policy favored formal sectors of the
economy, organized labor, and excluded informal and rural sectors of the economy
(Huber 1996). In so doing, Brazilian social policy not only left out those groups that
were most in need of social protection, but also exacerbated social inequality.
During the military regime (1964-1985) the public social insurance system largely
retained its Biskmarckian characteristics.1
The highly centralized federal agency,
Instituto Nacional de Assistência Médica da Previdência Social (INAMPS) administered
health care benefits. INAMPS managed public health care facilities and contracts for
services with private hospitals, clinics, laboratories, and doctors to provide state-funded
medical services for federal and private sector workers (Arretche 2002). The system
favored specialized medical services as doctors, public hospitals, and clinics were paid
according to the type of services rendered. This created incentives for physicians to
maximize expensive, complex services to increase revenue (Arretche 2002:160). Basic
medicine was left to two entities: the federal Ministry of Health, which drew on the
general revenue budget, and state-level health agencies, which operated independently
with their own budgetary resources. By the late 1980s however, the federal government
had largely abandoned basic health care altogether. Nearly all primary care was managed
by state and local governments (Arretche 2002: 160-161).2
The transition to democracy ushered in the first stage of major health care reform.
Some advocates for change noted that the existing social insurance system was expensive
due to its focus on curative medicine, which only benefited a select few, and called for
1 A notable change was the expansion of benefits to agricultural workers. To curb the swell of rural-urban
migration and preempt rural opposition and mobilization, the repressive government of General Emílio
Médici (1969-1974) created a social security scheme for agricultural workers (FUNRURAL) and delegated
its administration to the National Confederation of Rural Workers (CONTAG). This strategy sought to
create a positive constituency for ARENA (Stepan 1978, chap. 2-3 would call this “inclusionary
corporatism”). Their benefits would be subsidized by the urban sectors since these workers often lived in
conditions of extreme poverty and could not afford a contribution scheme (Weyland 1996:90). This
arrangement would later lay the foundation for incorporating other marginalized workers.
2 Sub-national governments were responsible for 93 percent of all primary care.
greater emphasis on prevention. Others noted that the social insurance system had left
out informal sectors.3
Sanitaristas, who represented a new generation of health
professionals, local health authorities, and left-wing health experts, called for universal
health coverage.4 The sanitary movement (movimento sanitário) successfully allied with
other social movement mobilizations for democratization to advance progressive health
reform. These reformers found an audience for their advocacy at the VIII Conferência
Nacional de Saúde (March 1986), which was convened by the presidency and the
Ministry of Health. Their participation at the event allowed them to substantially shape
the conference’s final resolutions, which called for reorganization of national health care.
It also declared it the state’s obligation to fulfill objectives such as universalization,
participation, and decentralization. The debate and outcomes of the conference laid the
foundation for subsequent debates over health policy, which would take place during the
Constitutional Assembly.
The sanitarista movement made impressive strides in the late 1980s to enshrine
progressive social rights in the constitutional text.5 They successfully articulated their
vision, including universal rights to health care, prioritization of basic health, and
3 Not only did the INAMPS system reflect a Bismarckian contribution scheme that favord elite interests,
but the uneven distribution of medical services across the country further exacerbated social exclusion. For
instance, in the mid-1960s, of the 3,972 municipalities in the country, 2,089 (53 percent) lacked a physician
(Mello 1977:107). Aggregate numbers however, mask regional disparities: in the southeast and south only
36.7 percent and 40.1 percent of cities lacked physicians, respectively; figures for the central west (66.3
percent), North (75.7 percent) and Northeast (70.8 percent) were dramatically higher (Mello 1977:108)
4 From 1960 to 1970, there was a dramatic increase in the number of schools of medicine; from 29 to 73,
representing an average increase of 5 university programs per annum (Mello: 1977:179). The 1970s was
also an important turning point for the burgeoning field of public health. Previously, medical schools
emphasized social medicine (medicina social) which is conceptually different from a public health
framework (Cohn 1989:126).
5 According to Cohn, in formulating their political strategies, the Sanitary Movement would collaborate
with various sectors, including other social movements (1989: 129).
While the constituent assembly would water down some of their
proposals, the venue proved particularly amenable to the movement’s political strategies.
In the end, Brazil’s democratic constitution proclaimed a universal right to health and
reinforced the state’s obligation to carry out those responsibilities through a free unified
health system.
Once the constitution was promulgated, however, changes in health policy would
largely stall. Although the constitution had articulated broad principles for progressive
health care, many of the details on the unified health system were left unspecified and
reforms would falter through much of the early 1990s. Several factors contributed to
delays in transforming the system. First, once Congress met to institute new legislation,
opponents of progressive health reform (medical businesses, INAMPS bureaucrats, and
conservative politicians) successfully resisted equalizing proposals.
As a result,
decentralization efforts that would have prioritized primary care came to a standstill.
Efforts to shift the ministry’s resources for basic health would falter as profit-seeking
hospitals would continue to benefit contracts for mid- and high-level complex services.
Second, Brazil faced a fiscal crisis in the mid-1990s that made it difficult to
increase spending for health care without cutting expenditures elsewhere.
6 Although the SUS includes principles of universal coverage and access, the sanitaristas were unable to
achieve the goal of a single national health policy. Public sector expansion would take place, but Brazil
would retain a dual system of public and private health insurance. In practice, middle and upper class
families have access to private health insurance. The universality of the SUS however, means that private
insurance carriers have few incentives to cover the most expensive medical cases, such as organ
transplants, and well-off Brazilians return to the public for the most sophisticated high-cost care. The
current system amounts to an indirect public subsidy to well-off Brazilians, further contributing to inequity;
at least 15 percent of SUS spending goes to the well-off (top three income deciles), mostly for expensive
treatments reducing resources for less well-off (World Bank 2004:164)
Minister Adib Jatene lobbied extensively to increase revenue for health and won
legislative approval for a constitutional amendment creating earmarked revenue for
health care. Yet, the increase in revenue for the Health Ministry would represent a short
lived victory.
Minister Jatene lacked political clout in the cabinet; the Minister of
Finance Pedro Malan opposed new taxes, especially those targeted for particular
spending areas. Thus, the finance ministry simply cut the health ministry’s resources to
offset the gains from the new earmarked taxes (Arretche 2004:175). This episode over
budgetary allocations demonstrated the weakness of the Ministry of Health, which would
suffer under politically vulnerable ministers until President Cardoso would select José
Serra, a close political ally, in 1998.
Third, while the sanitaristas had been effective in pushing for reforms during the
democratic transition, their political influence diminished thereafter.
decentralization had been a major goal of the sanitary movement, to combat the influence
of the medical industry at the national level, in practice it also had the effect of diverting
its attention from the national policy arena. Additionally, the sanitary movement was
unable to sustain its broad advocacy coalition. Not only did it lose the strength in
partnership with other social movements, but internal coalitions began to splinter
reflecting the divisions among the movement’s members (Cohn 1989:132). As a result,
many sanitaristas turned their attention to local initiatives for preventive medicine.
Despite all the political, administrative, and fiscal challenges for health reform in
the 1990s, Brazil did manage to enact incremental changes. The second wave of reforms
(1990-1995) led to the consolidation of the unified system including the
“municipalization” of service delivery and implementation of financial mechanisms for
the allocation of federal funds(World Bank 2004: 157).7
The third wave of reforms
(1996-2001) focused on the prioritization of basic care, specification of institutional
roles, legal and regulatory changes, and the introduction of alternative payment
mechanisms (World Bank 2004:157).8
One of the most important changes for
municipalities was the specification of federal transfers. Under new regulations, transfers
would vary by program and the level of service delivered. Cities could opt out of the
unified health system but doing so would require that local governments finance health
care from their own budgets.
Overall, the movement for progressive health reform at the national level yielded
mixed results. Without a doubt, sanitaristas won a major victory when the Constituent
Assembly enshrined the right to health care and made it the state’s obligation to provide
universal access. But other equity-enhancing efforts, such as prioritization of basic
health, stalled under political and fiscal pressure.
The Ministry’s efforts to foster
decentralization of health services resulted in a shift in responsibility, as sub-national
governments would take on the substantial efforts to prioritize preventive and basic
medicine. In the next section, we shift our gaze away from the national context to
7 New administrative rules were codified in the 1993 Norma Operacional Básica (NOB), which specified
the rules for decentralization and allowed municipalities to choose the degree of health care complexity
they could offer. Municipalities were required to demonstrate their capacity to deliver the level of service
they wanted to provide, and the majority of local governments were quick to participate in health
decentralization. By 1997, 3127 of 4,973 municipalities would participate in decentralization (Arretche
8 These changes were codified in the 1996 Norma Operacional Básica (NOB). Although 1996 NOB was
published in 1996, it did not go into effect until 1998.
examine how sub-national governments advanced healthcare access within their
While the Ministry of Health sought to advance national health reform and define
intergovernmental responsibilities, many states and municipalities forged ahead by
designing and implementing their own health care policies. In some instances, states
shifted spending priorities and introduced new partnerships with local governments. In
others, municipal governments took advantage of their newfound authority in the public
health arena to experiment with new modes of health care delivery. This section provides
an overview of some innovative local experiments in health care delivery during the late
1980s and early 1990s. As the cases reveal, in most instances policymakers sought to
introduce progressive health reforms that would reach historically underserved
populations and emphasize preventive and basic health care. The experience of São Paulo
served as a notable counterpoint. During this period, local experimentation led to policy
diversity and programs typically addressed the unique challenges each jurisdiction faced.
All in all, local governments would serve as “laboratories” that experimented with
different models of health care.
Their successes would inform and inspire the
development of Programa Saúde da Família.
In the early 1990s, several Brazilian municipalities began experimenting with
public health care models that would emphasize preventive and basic health care and
reverse the course of curative and doctor-centric approaches to medicine.9 One of the
most well recognized efforts to emphasize community health occurred in Niterói, in the
state of Rio de Janeiro, which instituted the Programa Médico de Família (PMF, Family
Doctor Program). At the time, this large city had strong social development indicators,
including high literacy rates, high median household incomes, and an average life
expectancy of 70 years. Despite high levels of human development, aggregate figures
masked social inequalities and pockets of deep poverty; city officials considered a quarter
of the population to be at “social risk” and in need of specialized attention.
In 1991, Mayor Jorge Roberto Silveira (PDT) of Niterói visited Cuba, learned
about its world-renowned health care model, and resolved to implement a similar system
back home.
The Cuban Health Ministry provided the municipality with technical
assistance to implement its own Médico de Família program. In Niterói, doctors and
nurses aides worked collaboratively in clinics embedded in the communities they served.
Each clinic included three or four teams (each team had a general practitioner and a
nurse’s aide) that was responsible for a designated jurisdiction that included 200 to 250
families. This approach allowed PMF teams to resolve 70 percent of medical issues
through clinical and home-based care. Unlike the Cuban model however, officials in
Niterói did not require that physicians, only their nurse’s aides, reside in the
communities. The decision to deviate from the Cuban model was due to necessity; very
few Brazilian doctors would have been willing to live in these impoverished
9 These experiments inlcude Porto Alegre, Curitiba, Londrina, Marília, São Paulo, Botucatu, Fortaleza
(Terra and Malik 1998).
neighborhoods (Interview D’Angelo 2004). Today, the family doctor program is one of
the most highly regarded municipal public health efforts in Brazil.10
Officials from Niterói were not alone in their desire to prioritize basic health
problems and enhance prevention. However, Niterói was uniquely positioned to initiate a
Cuban-inspired program; the city had the political, fiscal, and human resources it needed
to develop a family doctor model. Health experts in small cities in the Northeast also
wanted to deviate from curative medicine to prioritize prevention, but lacked the human
and physical resources to replicate the Cuban model. The Northeast faced high levels of
poverty and had limited infrastructure; in general the region lacked sufficient clinics and
hospitals, and physicians were in short supply.
Given these structural challenges,
officials sought an alternative to doctor-centric care that could emphasize communitybased health.
In 1987, the state of Ceará in the Northeast of Brazil designed and implemented
the pioneering preventive health program, Programa de Agentes de Saúde (PAS). The
program relied on two sets of actors: community health agents and nurses who would
supervise them. Like the family doctor program in Niterói, state officials in Ceará
wanted to promote basic health, emphasize prevention, and build ties with local
communities. Health agents were selected from within communities to work directly
with families. After receiving training, health agents would work with nurse-supervisors
to register families’ health care needs and encourage basic sanitary practices such as
10 In 1997, the Programa Médico de Família in Niterói won a national innovations award from the Gestão
Pública e Cidadania Program housed in the prestigious Getúlio Vargas Foundation in São Paulo.
water filtration, proper nutrition, and the promotion of vaccinations. Since the state also
had particularly high levels of infant mortality, health agents also monitored the height
and weight of children.11
Although the state of Ceará conceived and promoted the program, implementation
of the program still required municipal participation. Mayors who adopted the program
would have to find the funds to cover 15 percent of operating costs while also sharing
administrative responsibilities with state officials.12 The adoption of PAS across the state
took several years and depended on mayors’ willingness to opt-into the program. But by
1992, basic health indicators across the state had improved dramatically, infant deaths
had declined by a third, and vaccination coverage for measles and polio had tripled
(Tendler 1997:22). For these and other accomplishments, the state won the prestigious
UNICEF Maurice Pate award for child programs in 1993. In addition to Ceará, the
southern states of Paraná and Mato Grosso do Sul also instituted their own community
health agent programs.
The early experience of the PAS program stimulated the federal government to
support its expansion in other states. In 1991 the Ministry of Health instituted the
Programa de Agentes Comunitários de Saúde (PACS), which largely mirrored the PAS
program from Ceará. The objective of the PACS was to reduce infant and maternal
mortality, primarily in the North and Northeast, by extending basic health services to the
11 In 1987, the rate of infant death, in Ceará was 102 per every 1000; this was double the national figure
(Tendler 1997:21).
12 State officials sought to minimize rent-seeking behavior and traditional clientelism on the part of mayors
by retaining control over the hiring of community health agents (Tendler 1997: 24)
poorest and most destitute areas (Viana and Dal Poz 1998:18). The Ministry of Health
started offering federal funds to stimulate the program’s adoption across the Northeast
and provided guidelines detailing minimum requirements for health agents and their
An altogether different approach to basic health care during this time period was
adopted by in the city of São Paulo. Under Mayor Paulo Maluf (1992-1996), a leader of
the right wing party PDS, city administrators sought to address several problems in public
health services, including high costs, tremendous inefficiencies, and low quality of public
services. Secretary of Health Getúlio Hanashiro, who also shared Maluf’s disposition for
market-oriented solutions to public management problems, sought to dramatically
reconfigure basic health services. The policy, the Plano de Atendimento à Saúde (PAS),
integrated business sector principles into health services and represented a dramatic
departure from mainstream public health strategies.14 Although the constitution ensured
health care rights for all, city officials made clear their plan would prioritize service to the
most indigent.
Under the PAS, municipal authorities planned to create market-based incentives
for doctors and clinics and prioritized services for the neediest groups. Maluf proposed
the creation of regional clinics that would serve a given area’s designated population. To
13 For instance, health agents should be at least 18 years old, proficient in reading and writing, and have a
disposition for community health. These workers are responsible for: registering families, assessing
families’ health and living conditions, collecting updated information for a national database, conducting
home visits, identifying children for schooling, mapping community needs and identifying at-risk areas,
just to name a few.
14 The World Bank has cited the PAS as an experiment with innovative forms of organization and
management (World Bank 2004: 178).
encourage greater efficiency, each clinic would operate as a doctor-owned cooperative.15
Similar to private health care providers, each cooperative would, in theory, have an
incentive to rein in costs and maximize efficiency of operations. The city would in turn
provide per capita transfers for the cooperative, based on the number of registered
beneficiaries assigned to each clinic.16 Residents of São Paulo would have their own
health care card, similar to insurance cards provided by private insurers, and could visit
their designated regional cooperatives for medical care. The PAS proposal endured stiff
political opposition but was implemented in 1996 during Maluf’s last year in office.
Although the PAS is credited for making achievements in public management, such as
improvements in work absenteeism, the plan has been widely discredited as financially
unsustainable and was dismantled in 2000.
The Programa Saúde da Família (PSF, Family Health Program) was born out of
various community health experiences, such as the Programa de Médico de Família,
Programa de Agentes de Saúde from Ceará, and the nationally supported Programa de
Agentes Comunitários de Saúde. On December 27-28, 1993, the Ministry of Health held
a meeting of leading public health officials to discuss municipal health services and
financing. As Viana and dal Poz (1998) describe, the gathering was a response to
demands from municipal secretaries of health, who sought greater financial support for
15 This model required that civil servants, who had been employed in municipal health care, depart from
the public system and opt into the semi-privatized cooperative clinics. Approximately 35,000 civil servants
(88.3 percent of all workers) were removed from their original positions when they declined to integrate
into PAS cooperatives. Of those, 17,705 found positions in other municipal agencies, accepted demotions,
or accepted positions in the municipal health secretariat (Gouveia and Palma 1999: 143).
16 Under the plan, the municipality would pay cooperatives approximately US $15 (or R$15) per capita per
month. This figure was based on estimates of costs for similar services offered by the private sector (Cohn
et. al. 1999: 19).
basic health care.
The meeting was sponsored by the minister’s cabinet and included
officials representing the ministry, as well as bureaucrats from municipal and state
secretaries of health; also present were two officials representing international
development organizations, namely UNICEF and PAHO.17 An important feature of this
technical meeting was that it included a broad spectrum of participants from throughout
the country; technocrats who were involved in innovative experiences in the south and
southeast engaged with officials from the Northeast, who worked with PACS. The
models from Niterói and Ceará were very influential in shaping the eventual design of
PSF (Viana and dal Poz 1998; Interview Andrade 2004; Interview Machado 2003).
Health professionals liked the territorial organization of health services and the potential
to focus on prevention rather than demand-side service delivery. They also embraced the
role of the community health agent, but wanted to integrate other health professionals.
To accomplish these goals, PSF would draw on a larger team of workers including: a
doctor, nurses, nurse’s aides, and community health agents; nurses would retain their
central supervisory role over their aides and health agents.18 In this way, the PSF
program represented an upgrading of the PACS program; nurses were still central as
administrators and the community health agents’ roles remained the same.
Over the years, the staff in the Ministry of Health would nurture the Programa
Saúde da Família, protect it from administrative upheavals, and eventually champion its
17 According to Viana and dal Poz (1998) meeting participants included: Eugenio Villaça Mendes
(PAHO), Oscar Castillo (UNICEF); Halim Antônio Girade (UNICEF); Luis Odorico de Andrade
(Municipal Secretary of Health of the city of Quixada in the state of Ceará) (1998: 19).
18 Programmatic details were conceived at the December meeting, but were elaborated by staff from the
Ministry of Health in 1994 (Viana and dal Poz 1998:20).
central role in the unified health system. While meeting participants such as Heloísa
Machado and Luis Odorico de Andrade had always envisioned that the program would
become a central organizing model for basic health (Interview Machado 2003; Interview
Andrade 2004), PSF started out with modest institutional support and limited fanfare.
The program was one of many efforts in community health, a small project embedded
among other ministerial programs. In its first year of operation (1994) the ministry
signed limited convênios (funding agreements) with states and municipalities, which
required that sub-national governments contribute to the program’s cost. Selection of
eligible cities was restricted to high-priority cities based on a needs assessment conducted
by IPEA.19 In total, 55 municipalities signed agreements and instituted PSF in the first
year of operation.
Over the course of several years, PSF program coordinators Heloisa Machado and
Fátima de Sousa would defend the program internally within the Ministry. They not only
weathered the restructuring that would come with a constant stream of new health
ministers, many of whom lacked the political support to initiate significant health
reforms, but they eventually garnered the support they would need to institute and expand
the program within the ministry.20
A turning point for PSF came in 1995, when
prominent heart surgeon, Dr. Abib Jatene, became the health minister for the second time
19 The federal research agency, IPEA, produced a report Mapa de Fome, which identified the poorst cities.
This report served as the basis for determining eligibility for PSF participation (Vasconcellos 1999: 156).
20 Since December 1993, the Ministry of Health has been under the leadership of: Henrique Antônio
Santillo (August 1993-January 1995), Adib Domingos Jatene (January 1995 to November 1996), José
Carlos Seixas (November-December 1996), José Carlos de Albuquerque (December 1996 – March 1998),
José Serra (March 1998-February 2002), Barjas Negri (February 2002-December 2002), Humberto Sérgio
Costa Lima (January 2003 to July 2005).
in his career. As Jatene explained, he first learned about community health efforts and the
role of the agentes comunitários de saúde when he was health minister in 1992. At the
time, they were doing a wonderful job to combat the spread of cholera in the North and
Northeast (Interview Jatene 2003). Upon his return to that position in 1995, he met with
Machado and Sousa, who persuaded him that the PSF would work more broadly. After
visiting cities that had instituted the program, such as Camaragibe in the state of
Pernambuco and Sobral in the state of Ceará, Minister Jatene agreed to support the
program (2003).21 In January 1996, PSF was transferred to the Secretaria de Assistência
da Saúde (SAS) finding a more central home within the ministry and allowing for its
institutionalization (Viana and dal Poz 1998: 22).
The administrative changes had several important consequences. First, the PSF
gained broader visibility and the staff started articulating the idea that it should move
from an isolated project to represent an organizing principle for basic health care.
Second, the ministry also moved towards integrating PSF with PACS and connecting it
with the broader efforts to decentralize health and institute the SUS.
Programa Saúde da Família sprung up at a moment when national health care
reform was encountering some of its greatest political and fiscal difficulties. Local
governments on the other hand, were taking advantage of newfound authority to develop
and implement new policies for better preventive health; these early experiments
reflected a potpourri of approaches. As Judith Tendler observed in her case study of
21 Minister Adib Jatene was so enthusiastic about the program that he took it to the Presidents Cabinet, and
introduced it to the President and first lady, Ruth Cardoso. They traveled together to the Northeast to see it
on the ground (Interview 2003).
Ceará, decentralization of health policy has entailed a mix of both central and local
efforts (1997:23), as municipalities, states, and the federal government work
collaboratively to deliver and finance health services. This observation certainly holds
true for the PSF, which benefited from early municipal health policy experimentation,
was conceived by a broad group of experts involved in all tiers of health care provision,
and drew on federal financing offered through the Ministry of Health. Yet, to understand
the evolution of the program’s spread, from 55 municipalities in primarily small rural
towns in the North and Northeast, to extensive national adoption by 4,944 cities in 2003,
requires more than a simple tale of vertical pressures for diffusion. While PSF did
provide some political opportunities for mayors, such as the potential to dole out coveted
health agent jobs to political cronies, the program was still highly complex and required
restructuring health care services.
Furthermore, some politicians and health policy
technocrats would challenge the notion that PSF was a desirable policy for their cities. In
practice, the adoption of PSF was not always guaranteed or automatic. Each city would
undergo its own policymaking process related to PSF. Thus to uncover the adoption
decisions for PSF, we must turn to the local dimensions of policy making and open the
black box of policymaking to uncover actors’ motivations for policy emulation.
Why did some cities adopt the Programa Saúde da Família quickly, while others
lagged behind? Why would health policy makers emulate the program, particularly when
many local governments had already experimented with alternative health care models?
Also puzzling was that actors from large urban cities with sophisticated health
infrastructure would emulate a program that largely drew its inspiration from cities that
are far from typical, the poor rural Northeast (with the PACS) and Niterói (with PMF)
with high levels of human development and incomes per capita.
The case studies in this chapter examine the motivations behind PSF adoption in
four research sites over three municipal administrations. These cities had great flexibility
in determining their basic health models. Some administrations would tailor health
policy to suit the needs of their municipalities while others would adopt PSF. In other
words, emulation of PSF was far from automatic or a forgone conclusion. Electoral
competition varied across these cities and the voters selected mayors representing various
ideological predispositions, from rightists to leftists.
In addition, health policy
technocrats engaged in professional networking activities, but not necessarily the same
ones. Table 5.1 provides an overview of the case studies, when PSF adoption took place,
and the partisan affiliations of the mayors who adopted the family health program.
Table 5.1 Case Studies: Adoption & Non-Adoption
Brasília (DF)
Belo Horizonte (MG)
Salvador (BA)
São Paulo (SP)
Executive in Office & Party IDa
Joaquim Roriz (PTR)
Cristovam Buarque (PT)
Joaquim Roriz (PMDB)
Programa Saúde da Família
Patrus Ananias (PT)
Célio de Castro (PSB)
Célio de Castro (PSB)
Fernando Damata Pimentel (PT)d
Lídice da Mata (PSDB)
Antônio José Imbassahy (PFL)
Antônio José Imbassahy (PFL)
Paulo Maluf (PDS)
Celso Pitta (PPB)
Marta Suplicy (PT)
Mayor’s partisan affiliation at the time he or she ran for office.
Brasília, the Federal District, operates under the gubernatorial electoral calendar.
The program was suspended and then reinstated.
Fernando Damata Pimentel (PT) assumed office in November 2001, after Célio de Castro suffered a
Political Incentives
Health care is one of those areas of public policy that is particularly visible to the
electorate because irrespective of age and income status, health policy affects the entire
population. For instance, the outbreak of infectious diseases, such as cholera, dengue,
and HIV-AIDS can affect an entire city’s population, regardless of gender, age, and
income. The Programa Saúde da Família can generate considerable public attention as its
aims include prevention and basic health care. For beneficiaries of the program, PSF is
also highly visible because it brings the state into the private sphere of domestic life
through home visits and neighborhood outreach. In addition, communities are able to
discern if a neighboring area has PSF while their district remains underserved. Given the
high visibility of PSF, it would certainly make sense for self-interested politicians
focused on electoral politics to embrace the program. In cities that lack PSF, candidates
might campaign on implementing the program. Politicians in cities that already have a
PSF policy, might campaign on extending the program to new communities.
In addition to the general appeal that PSF holds among the poor, the program also
has the potential to generate rent-seeking behavior as politicians have the ability to dole
out particularistic benefits. Implementation of PSF involves job-creation as municipal
health teams need local residents to serve as agentes comunitários de saúde (ACS). In a
context where unemployment is high and the working poor encounter tremendous
difficulties in making ends meet, a job as a community health agent is very attractive.
The position does not require specialized skills in primary health;22 rather the minimum
requirements to qualify for the position include: literacy, basic schooling, and
“leadership” skills. In many ways, the ACS position is an extension of “women’s work”
in the domestic (private) sphere, which helps explain why the vast majority of the
positions go to women.23 The PSF also has the potential to generate a second type of
political patronage.
The geographical demarcation of neighborhoods served by the
program offers a clear benefit for politicians who want to reach out to communities for
22 ACS training on sanitation and preventive medicine occurs after they are hired.
23 Women in Latin America often face a triple burden of working, family care, and community care. The
job of an ACS blends traditional gender roles that delegate women as responsible for family life and
domesticity as well as that of their community.
electoral support. In other words, mayors who adopt the program can influence voters by
deciding which neighborhoods will be served. This is particularly important as PSF
typically targets economically and epidemiologically vulnerable areas, rather than
extending the policy throughout the entire city. In practice, politicians can deviate from
serving the neediest areas to fulfill their electoral agenda. For these reasons, savvy
politicians who want to engage in traditional patronage politics can benefit tremendously
from the family health program’s design.
Mayors in each of the case study sites faced electoral competition and
campaigned on issues of health care delivery. Given the electoral potential of enacting
PSF, we might expect all mayors, regardless of their ideology, would emulate the
program. In cities with low levels of health infrastructure, PSF represented an important
extension of new services to communities with limited access to health care. Thus, PSF
represented the creation of new services. For cities with existing health services already
in place (e.g. clinics and hospitals), the PSF program offered the potential to restructure
healthcare to work with families and communities in a more integrated fashion.24 In
these instances, voters would benefit from better quality services, greater interaction with
healthcare providers, and easier access.
Despite the potential that PSF could offer given the competitive electoral
environment in the case study cities, the qualitative evidence reveals that mayoral
candidate did not systematically endorse the program and instead sought a diverse set of
24 Adib Jatene argued that even in cities like São Paulo, with sophisticated health infrastructure, PSF
represented a new ‘add-on’services because there were terrorities of the city that were underserved by
clinics and hospitals. In his view, the conditions of urban poverty in the city’s periphery are similar to that
of underserved rural communities in the Northeast (Interview Jatene 2003).
health policies. Conservative politicians often advocated for market-oriented proposals
whereas leftist politicians tended to embrace PSF. Candidates’ approaches to public
health issues were certainly important to their campaigns and entered into the electoral
debates in these cities. Of the four research sites the city of São Paulo, where the PAS
quasi-privatized system served as a counterpoint to PSF, offers the best example that
debates over health received widespread media attention (Cohn et. al 1999:67-94). All
three mayoral campaigns dedicated considerable attention to health care issues, but
candidates differed in their vision for the city. Some candidates like Maluf and Pitta
argued São Paulo should institute private market incentives into its system. It was the
leftist candidate, Marta Suplicy, who would campaign on health reform and announced
she would implement the PSF if elected
Even though PSF had the potential to garner electoral votes for Marta Suplicy, her
advisors dismissed the notion that her favorable position toward the policy during the
campaign represented a vote-buying strategy (Interview Manfredini 2003). The local
media widely covered health issues in São Paulo, yet most residents were unaware of the
technical dimensions of the various proposals under consideration. The groups that
might benefit most from PSF, the poor and most vulnerable populations, were unfamiliar
with the family health program. Once her administration moved to implement PSF,
many citizen representatives who served on local health councils were skeptical of the
program as they viewed clinics and hospitals, not PSF health teams, as appropriate places
to go to for their health care needs. Moreover, several unions expressed deep concern
about contract negotiations under PSF (Interview Costa 2003; Interview D'Agostini 2003;
Interview Oliveira 2003). In this light, Suplicy’s emulation decision appears to reflect a
calculated administrative risk rather than a clear-cut electoral strategy to win the election.
Brasília was the first of the case study cities to emulate PSF in the second half of
Cristovam Buarque’s administration. While Buarque was a staunch leftist and often
spoke about social rights and citizenship on the campaign trail, he was not a strong
advocate for PSF. Rather, his focus was on Bolsa Escola and other educational social
policies. The decision to implement the family health program was largely delegated to
his senior political appointees, notably his Secretary of Health, Maria José da Conceição
(Maninha). Although both Bolsa Escola mothers and PSF community health agents
could have provided significant electoral support for Buarque’s reelection campaign, his
closest advisor asserted that the governor refused to exploit beneficiaries of his social
programs (Interview Aguiar 2003).
In Belo Horizonte, where the political competition was concentrated to the left of
the political spectrum, the decision to implement PSF was hotly contested among
technocrats. Mayor Célio de Castro was said to have embraced PSF early-on in his first
mayoral term because of his own familiarity with the program; he was a physician by
training. Although he was an early enthusiast of the family health program it would
nevertheless, take him until 2000 to implement the program in the city. While the delay
in emulating PSF gives the appearance of an electoral incentive, health experts in Belo
Horizonte told a different story. Senior health policy technocrats in the municipal health
department opposed the mayor’s plan to emulate PSF. In their view, PSF was not an
appropriate strategy for Belo Horizonte (Interview F. Santos 2004; Interview Franco
2004). Thus, for several years the mayor encountered opposition from senior technocrats
who favored other types of integrated health services. Ultimately, in order to implement
PSF, the municipal Department of Health would need to undergo personnel changes.
Opponents of PSF left the agency and new technocrats, who had adopted the program
elsewhere, were brought-in to administer the program in Belo Horizonte.
Like Belo Horizonte, Salvador was a late adopter of PSF. It would take several
years for Mayor Antônio Imbassahy to announce he would institute the health program.
Health policy specialists, both inside and outside of government, explained that the
mayor was relatively uninvolved in the decision to implement PSF in Salvador and
delegated these issues to his Secretary of Health, Aldely Rocha (Interview Queiroz 2004;
Interview Nossa 2004). Technocrats in the city’s Department of Health noted that the
mayor was skeptical of the program, preferred that it expand slowly, and expressed
concern over the expense associated with it (Interview Queiroz 2004; Interview Nossa
2004). One observer of municipal health noted that it was ironic that the mayor would
initially fail to recognize the electoral potential behind PSF and attributed the city’s
delays to a lack of political imagination on the part of the political elite (Interview Boa
Sorte 2004).
In all these cases, mayors were rarely motivated to adopted PSF for electoral
gains. Mayors’ initial emulation decisions were largely delegated to senior political
appointees, such as municipal secretaries of health, and technocrats, who served as civil
servants. Once PSF was in place however, the program was difficult to dismantle and a
few politicians found ways to use it for political gain. Two rightist politicians, Joaquim
Roriz (PMDB) from (1998-2002; 2002-2006) and Antônio Imbassahy (2000-2004), are
exemplary in this regard.
Both are well known for their general use of political
patronage, and those tendencies continued once PSF was enacted.
Joaquim Roriz of Brasília inherited Saúde em Casa25 from Cristovam Buarque’s
administration. During his campaign, Roriz promised loyal partisans jobs as community
health agents, a much coveted position among the lower classes, and handed out slips of
paper that indicated they would be in line for the jobs. Once the administration was in
office and the health secretariat opened applications for the ACS positions, individuals
would arrive with their letters indicating they were promised positions.
bureaucrats were reluctant to admit their own participation in this hiring scheme, they had
all heard or seen evidence to this effect. Since the district’s health secretariat experienced
several changes in leadership, technocrats felt free to say that their predecessors engaged
in problematic hiring practices but that they followed proper rules.
The allegations of unethical recruitment practices for PSF were not limited to the
community health agent position. One high-ranking health administrator in Brasília
informed me that patronage was such a pervasive and engrained part of the local political
culture that upon announcing the resumption of the PSF program he received over a
thousand personal requests from the politically connected for jobs associated with the
program, including positions for doctors and nurses.
Overall, the administration of
25 The Buarque administration gave the PSF program its own name, Saúde em Casa.
26 The Federal District briefly suspended the PSF program under allegations of fraud. The suspension
coincided with the Federal Ministry of Health suspension of PSF funds due to inquiries of improper usage
of health financing. When the program resumed, it did so under new leadership and with new personnel.
Joaquim Roriz in Brasília was especially notorious for irregularities related to the health
sector (including PSF). Accusations were so pervasive that Brasília’s Ministério Público
(Public Prosecutor’s Office), in collaboration with federal auditors, undertook
investigations into allegations of widespread corruption and the misuse of funds. The
irregularities were so extensive that Jairo Bisol, a public defender with the Ministério
Público, asserted that PSF in Brasília was synonymous with corruption (Interview
2004).27 Both federal and district audits of PSF concluded that the program was unoperational; hiring practices had been based on political favoritism and many personnel
were operating with incomplete teams (i.e. they lacked nurses or doctors).
In Salvador, technocrats reported they faced very little interference when it came
to hiring PSF personnel. But even so, “political interests” had impeded their ability to
implement the family health program as they saw fit. Prior to establishing PSF in
Salvador, staff members in the Health Department conducted a city-wide epidemiological
study and assessment of health services. Results from the study had defined which
districts should receive priority for PSF (i.e. which sanitary districts should be served and
in which order).28 While technocrats had sought to follow their plan, they acknowledged
27 As Bisol (2004) explained, the entire health care system was a source of a lot of money (for services,
medication, operational expenses, etc.) and represented an opportunity for graft. He estimates that R$ 40
million (approximately $20 million U.S. dollars) in health funds went to the construction of the district’s
bridge. In 2004, the Ministério Público was planning to investigate new allegations into clientelistic
practices with health care delivery, including a scheme where doctors on the government payroll offered
preferential services to patients with political connections. The new allegations involved doctors paid by
the federal district but contracted through the Instituto Candango de Solidariedade, a non-profit association,
which had ties to Wesliam Roriz, Governor Roriz’s wife.
28 Initially, technocrats in the city government wanted to build off of the existing ACS program. They also
favored complete PSF coverage for the health district – Subúrbio Ferroviário – which has some of the worst
health indicators. But nearly four years after the program’s start, only about 50 percent of that district’s
that Secretary of Health Aldely Rocha, Mayor Antônio Imbassahy, and the “political
leadership” had directed them to territorially expand PSF to include other parts of the
city. As such, it was not surprising that at an inaugural event for a PSF clinic in the
neighborhood Altos dos Coutos in June 2004, the ribbon cutting event included dozens of
the mayor’s political allies. The mayor’s political cronies (city and state officials, city
council members, and candidates for elective office) gave speeches to the crowds
amassed near the health center.
That traditional conservative politicians such as Roriz and Imbassahy would
continue to engaged in “politics as usual” with PSF is not surprising; allegations of
patronage and clientalism were certainly features of all their departments. Yet, what is
truly remarkable is that PSF emulation decisions rarely came down to these electoral
incentives. Most mayors who adopted the program delegated these policy decisions to
their politically appointed senior staff. Cristovam Buarque gave his Secretary of Health,
Maria José da Conceição (Maninha), wide latitude to emulate PSF (Interview Conceição
2003). Mayor Imbahassy of Salvador was reluctant to embrace the program and only did
so under pressure from health experts.29 Marta Suplicy of São Paulo let her technical
advisors decide how the city should recover from the debilitating experience with PAS.30
population was covered by PSF (35 PSF teams). Rather than extend full coverage within that district, the
city started extending PSF with teams in districts 2 (Itapagipe), 3 (São Caetano), and 12 (Cajazeiras).
29 Health policy specialists both inside and outside of government asserted that Mayor Imbassahy was
uninvolved in the decision to implement PSF in Salvador. Technocrats in the city’s Department of Health
noted that the mayor was skeptical about the program and preferred that the program’s implementation
proceed slowly.
30 Marta Suplicy’s supporters were themselves divided on whether to embrace PSF. In the early stages of
the campaign, the PT-aligned think tank Instituto Florestan Fernandes sponsored meetings for health
Only Mayor Célio de Castro of Belo Horizonte was in the odd position of having
embraced the program early on, but lacked the administrative support to implement it
(Interview F. Santos 2004).31 Given that electoral incentives for PSF emulation offer a
relatively weak explanation for the motivations for policy enactment, we now turn to the
two alternative explanations: ideology and social networks.
Do actors’ ideological commitments drive their emulation decisions? Do actors
perceive this program to be “leftist” or “rightist”, and if so, does the ideological meaning
behind the program influence emulation decisions?
While PSF does have equity
enhancing goals and reflects a reprioritization of basic and preventive services, actor’s
perceptions about the program and its ideological meaning changed over time. As the
case studies reveal, mayors’ ideological commitments mattered, as did those of their
politically appointed technocrats and civil servants.
Mayors in each of the administrations had clear ideological tendencies, ranging
from leftists to rightists (see Table 5.1). In practice, mayors’ ideological predispositions
shaped the character of their administration. Senior appointed officials, such as municipal
secretaries of health, were usually close political allies who shared the mayor’s
ideological viewpoints. Mayors also set the overall tone for their administration by
through budgetary allocations and championing their trademark programs. For instance,
experts to debate the merits of PSF. Ultimately, key technocrats made the decision to endorse PSF for São
Paulo (Interview Manfredini 2003).
31 Mayor Célio de Castro suffered a massive stroke in November 2001; his vice-Mayor Fernando Pimentel
served the rest of his term. Mayor Castro died prior to the start of field research.
left and center-left mayors all championed education-related programs and emphasized
the need to invert social spending to prioritize the needy and enhance “citizenship”;
Cristovam Buarque’s trademark policy was Bolsa Escola, Lídice da Mata created the
program Cidade Mãe, and Marta Suplicy created integrated “community schools” called
Centros Educacionais Unificados (Unified Education Centers, CEU). Politicians on the
right, on the other hand, emphasized business-oriented initiatives; Antônio Imbassahy
highlighted tourism and business development, Joaquim Roriz championed construction
of a bridge (Ponte Juscelino Kubitschek), and both Paulo Maluf and Celso Pitta defended
the market-oriented health program PAS.
Table 5.2 Case Studies: Mayor’s Ideology
Brasília (DF)
Belo Horizonte (MG)
Salvador (BA)
São Paulo (SP)
Executive in Office & Party IDa
Joaquim Roriz (PTR)
Cristovam Buarque (PT)
Joaquim Roriz (PMDB)
Patrus Ananias (PT)
Célio de Castro (PSB)
Célio de Castro (PSB)
Fernando Damata Pimentel (PT)d
Lídice da Mata (PSDB)
Antônio José Imbassahy (PFL)
Antônio José Imbassahy (PFL)
Paulo Maluf (PDS)
Celso Pitta (PPB)
Marta Suplicy (PT)
Mayor’s partisan affiliation at the time he or she ran for office.
Brasília, the Federal District, operates under the gubernatorial electoral calendar.
Election data unavailable for 1992.
Fernando Damata Pimentel (PT) assumed office in November 2001, after Célio de Castro suffered a
While elected officials’ ideological predispositions framed the range of options
for health care, the most rampant ideological debates surrounding PSF occurred among
technocrats and experts in public health. When PSF was first introduced in 1994, health
policy specialists and activists associated with the movimento sanitário had mixed
reactions to the policy. Their assessment of the program reflected broader debates in the
international public health community over prevention, access, and prioritization of
services. Many staunch leftists affiliated with the movimento sanitário in Brazil strongly
embraced the goals asserted at the Alma Ata World Health Organization conference, held
September 6-12, 1978.32 The conference declaration included the maxim, “Health for All
by the Year 2000” and reflected an ambitious effort to transform the entire health system.
The holistic approach at Alma Ata linked public health issues with broader questions of
development. It also challenged major assumptions about health care by emphasizing
appropriate technology and calling for the training lay health professionals. Implicitly,
the declaration criticized advanced industrialized countries’ approach to medicine with its
emphasis on disease-oriented technology, overly specialized care, and elitist bias. The
viewpoint articulated in the Alma Ata declaration strongly resonated with many of the
sanitary movement’s goals to prioritize primary health care and engage local
32 As Cueto (2004) explains, the international situation surrounding the Alma Ata conference was highly
politicized given the context of the Cold War. Despite early difficulties in identifying an appropriate
location for the event, the conference included 3000 delegates from around the world, with 70 participants
from Latin America, 97 percent of whom represented public health ministries (Cueto 2004:1867). The
conference and its proceedings had a lasting effect on Brazilian public health officials, who often reference
Alma Ata goals during interviews (for example: Interview Machado 2003; Interview Andrade 2004).
While the Alma Ata declaration provided a framework for many public health
professionals worldwide, the international development community diverged in its
approach to public health. The Rockefeller Foundation, along with UNICEF, articulated
an alternative or “minimalist” approach to the Alma Ata declaration.
Instead of
transforming the system, these actors favored small add-on programs. These institutions
supported limited, short-term, and selective strategies best known as GOBI: growth
monitoring, oral rehydration techniques, breast feeding, and immunization (Cueto 2004:
1869). From these international donors’ point of view, this strategy offered the possiblity
of tracking measurable results through program monitoring and evaluation. Limiting
efforts in selective primary care also helped these institutions avoid the political and the
costs associated with the Alma Ata declaration. In the Brazilian context, this approach
certainly caught on; the PACS program in the Northeast reflected modest GOBI
The international public health debate surrounding Alma Ata created two camps:
one that favored major transformation for universal primary care and another that
embraced complementary activities leading to selective primary care (Cueto 2004: 1869).
As Brazilian public health experts debated the merits of PSF, their assessments often
hinged on whether they saw PSF as reflecting the more radical universal primary care or
neoliberal selective primary care. Many staunch leftists tended to embrace the holistic
vision of Alma Ata and contested whether PSF was compatible with the declaration.
Early critics of the program declared PSF a “programa pobre para pobre” (a poor
program for the poor) and said it represented neoliberal policy (Interview Junkeira 2003;
Interview La Forgia 2004).
There were several reasons staunch leftists initially opposed the program. First,
when the Ministry of Health first introduced PSF it did so in a limited fashion, as one of
multiple projects. Second, the program’s design had been largely influenced by the
PACS program – demonstrating a type of GOBI agenda - and had essentially served as an
extension of it (Viana and dal Poz 1998:21). Since PACS was a predecessor of PSF, the
earliest cities to adopt the family health program were in the Northeast. This convinced
some that PSF was intended as a targeted, not universal approach to basic health. All this
contributed to their belief that PSF was a “poor program for the poor.”
It would take
years for the Ministry of Health to articulate its ambitious vision for PSF, namely that
PSF could serve as a model to reorient primary health care for all.33 Lastly, when José
Serra led the health ministry (1998-2002), he made efforts to obtain supplemental
funding for PSF from the World Bank. The implicit approval of PSF from the World
Bank further convinced those actors that the program was a “World Bank program”
representing Washington-consensus style social policy.34
33 Adib Jatene, who became health minister in 1995, would be supportive of the PSF and argue that it was
not a “poor program for the poor”. In 1997, the ministry would publish “Saúde da Família: uma estratégia
para reorientação do modelo assistencial” (Family Health: a strategy for reorienting the assistential
model). In 2000, the ministry would frame the PSF program as the principal model for basic health (World
Bank 2004) and encourage municipalities to hire enough PSF health teams to cover 100 percent of their
jurisdiction’s population.
34 This viewpoint was expressed by the researcher and state health policy analyst Virginia Junkeira
(Interview 2003). However, World Bank health policy program officer Jerry La Forgia asserted that the
idea that the Bank imposed PSF was misguided; bank staff follow Brazilian government ministers’ lead
(Interview 2004).
Nevertheless, other long-term public health experts associated with the movimento
sanitário were quick to embrace the policy and argued it was compatible with the Alma
Ata declaration. The policy coincided with their priorities to focus resources on primary
care, centered on basic health and prevention. It also reflected previous sanitary policy
and public health initiatives that sought to build close community ties, dating back to the
1970s and 1980s.35 Also important was that decentralization had created opportunities
for many sanitaristas to put principles into practice.
Those in the Northeast were
especially encouraged by the outcomes of both PACS and PSF programs. Research
suggested that these strategies could prevent the spread of cholera and lead to dramatic
declines in infant and maternal mortality (Lima et. al. 2001; Mendonça et. al. 2004;
Mishima et. al. 1999; Solla et. al. 1996). Leading participants who helped design the PSF
also argued that the policy should be much more than an isolated program; actors such as
Machado, F. Sousa, and Andrade wanted PSF to be the basis of a restructured primary
health care system for all (Interview Andrade 2004; Interview Machado 2003).
Given these “mixed” signals about the program’s broader aims, it is
understandable that actors committed to universal primary care might balk at the PSF.
But key actors, such as David Capistrano, Adib Jatene, and Luis Odorico de Andrade
were crucial in convincing health experts that the program was compatible with full
primary health coverage. Capistrano was a highly visible public health expert with ties to
the Workers’ Party. When he agreed to partner with Minister Jatene to implement a
35 For example, the Programa de Interiorização de Ações de Saúde e Saneamento (PIASS) was first
introduced in the Northeast in 1976 and lasted until 1979. From 1980-1985, the PIASS was extended to
other areas, including São Paulo. For a history of sanitary policy and community health, see Silva and
Dalmaso (2002).
state-sponsored PSF program, he signaled his support for the model. In doing so,
Capistrano would recruit and convert skeptical leftists that the program was compatible
with their values (Interview Silveira 2003; Presentation Mattos 2003). Although Jatene
never affiliated with a political movement or party, he enjoyed a solid reputation among
technocrats for his deep commitment to Brazilian healthcare. His endorsement of the
program carried special weight. Luis Odorico Andrade, a self-identified sanitarista who
had formed part of the team to design PSF and later implemented it in Quixadá and
Sobral in Ceará, also advocated for the program. Not only did he enthusiastically defend
it within networks such as CONASEMS but he also created a training school for PSF and
published the journal “SANARE: Revista Sobralense de Politicas Públicas” to showcase
the program.36 These individuals’ activities caught many leftists’ attention and convinced
them that PSF could represent a progressive transformation of Brazilian public health.
Surprisingly, rightist politicians largely stayed out of the ideological debate
surrounding PSF.
Some advocates of the program also sought to appeal to fiscal
conservatives by suggesting that PSF could lead to greater economic efficiency given the
rising costs of curative medicine in Brazil. The logic is that effective prevention can be
more cost-effective in the long term, than expensive hospital-based treatment.37 While
publications by the World Bank (2004) support this general viewpoint, arguments for
36 The first volume of SANARE focuses on the PSF as strategy to restructure public health (Vol. I, No. 1,
Oct-Dec. 1999).
37 This line of reasoning is highly debated and longtime advocates from the sanitary movement are
reluctant to make this appeal. As Dr. Luis Odorico de Andrade noted, universal coverage of PSF is costly
(Interview 2004).
PSF based on potential cost-effectiveness gained little traction and rightist politicians and
technocrats were largely absent in their support of PSF.
The development of health policy over three administrations in the city of São
Paulo offers valuable insight into the ways ideology influenced emulation decisions.
While most mayors delegated health policy issues to their secretaries of health, Mayor
Maluf in São Paulo was unusually hands-on in selecting, advocating, and defending the
PAS for the municipality. As discussed earlier in this chapter, Mayors Maluf and Pitta
sought to implement their own public health policy for the city of São Paulo. The semiprivatized cooperative system was consistent with both mayors’ conservative ideology
and Maluf held steadfast in defending the model from its earliest opponents.
From the onset, the PAS proposal fueled considerable debate and led to both
legislative and judicial machinations as various groups tried to block or support the
implementation of the program. Various sectors, including some physicians, embraced
the logic of doctor-controlled cooperatives and saw potential for greater financial and
professional rewards. Others, including staunch leftists and activists in the sanitary
movement, were immediately critical of the plan. Not only did they prefer that municipal
authorities seek universal coverage and access, as mandated by the constitution, but they
argued that the nature of public health problems, such as communicable diseases, made
private sector solutions financially unsustainable.
Another concern among PAS
opponents related to Maluf’s electoral appeal with the public as someone who “rouba
mas faz” (steals but gets things done). In other words, the mayor’s widespread reputation
for corruption led political opponents to anticipate that his health reform plan would
serve as a vehicle for graft as well.38
Mayor Maluf announced the PAS on January 17, 1995 and moved to implement
the program through decree in April of that year. However, the Mayor encountered
significant resistance to the plan which delayed its start. City council members from the
Workers’ Party were among the most vocal opponents and sought to block the PAS
through both judicial appeals and interventions in the city council.
implementation of the PAS would require City Council approval, which passed along
partisan lines on September 12, 1995.39 Even though Paulo Maluf was nearing the end of
his mayoral term and was ineligible for reelection, the city went ahead in instituting the
PAS. When Celso Pitta, a political ally of Maluf’s, won the 1996 election, he kept his
campaign promise to maintain the PAS program during his administration.
Paulo Maluf’s decision to adopt the PAS not only encountered resistance from
actors within the city of São Paulo, but also a host of external actors who appealed for
him to reverse course. Minister Adib Jatene was personally interested in seeing the PSF
program implemented in the city (Interview Jatene 2003). Aside from the fact that the
minister had been a resident of São Paulo and took a special interest in it, Jatene was also
38 Paulo Maluf has consistently asserted his innocence against accusations of corruption. Nevertheless, in
2005 he spent 40 days in jail on accusations of racketeering, tax evasion and money laundering. In March
2007, the New York City District Attorney’s office indicted him in a corrupt kickback scheme at the
expense of Brazilian taxpayers. According to District Attorney Morgenthau, Maluf and four coconspirators stole at least $11.6 million, although they are believed to have stolen more than $140 million
as that sum passed through the New York bank account linked to Maluf (Rohter & Hartocollis 2007: A9).
39 According to Cohn et al., votes on the PAS corresponded strictly to partisan lines: all center-left and
left city counselors from the PSDB, PT, PV and PC do B voted against the proposal; all partisans of the
right-of-center parties including PPR, PTB, PL, and PFL voted in favor; of the twelve members from the
PMDB, 1 abstained and 1 broke with their party to oppose the measure (1999:45-47).
convinced that the program could have positive effect; he had seen model PSF programs
in Northeastern cities and believed parts of São Paulo faced similar difficulties. He also
wanted to see the program extended to the biggest metropolitan areas, such as Belo
Horizonte, São Paulo, and Rio de Janeiro (Capistrano 1999). Therefore Minister Jatene
reached out to Paulo Maluf on at least two occasions to personally appeal to him to enact
PSF in São Paulo; but Maluf simply turned him down, citing other plans (Interview
Jatene 2003).
Minister Jatene refused to give up his goal of implanting PSF in São Paulo.
Instead, he asked the Governor of São Paulo state to install a state-run program within the
city’s borders. The center-left governor Mário Covas (PSDB) agreed and the State
Secretary of Health, José da Silva Guedes, initiated a state-run PSF program within the
city’s borders called QUALIS (Qualidade Integral à Saúde). The program started in 1996
in the district of Itaquera, under a partnership agreement between the federal Ministry of
Health, State Department of Health, and the Santa Marcelina Hospital. Over time, the
QUALIS program expanded to include additional districts in the municipality.40 To
shepherd QUALIS, Adib Jatene selected a trusted advocate for public health with a solid
reputation for his commitment to primary health care – David Capistrano Filho. The
selection of Capistrano was especially notable because of his high profile background and
links to the leftist Workers’ Party; he had been active in the movimento sanitário, was the
former health secretary of the city of Bauru (1984-1986), and former mayor of the city of
40 For a full description of QUALIS, see Capistrano (1999).
Santos41 (1992-1996). As this episode suggests, Dr. Adib Jatene was firmly committed to
PSF and was willing to work with politicians of all ideological stripes to get the program
off the ground. But at least in the early years, the Minister would find much greater
receptivity for PSF among left and left-of-center politicians. Ultimately, it would take
the election of a leftist, Marta Suplicy, for the municipality of São Paulo to adopt the
family health program.
In sum, the decision to implement PSF reflected the ideological beliefs of two sets
of actors: mayors and their senior technical staff. Mayors who chose to implement the
family health program were overwhelmingly leftist or center-left and had decided that the
programmatic goals of working directly with patients in the community and expanding
access to primary health care were consistent with their core beliefs. In three of the four
cities, PSF was introduced by left-of-center administrations, including Cristovam
Buarque (PT) in Brasília, Célio de Castro (PSB) in Belo Horizonte and Marta Suplicy
(PT) in São Paulo. The only non-leftist to adopt the program was Mayor Imbassahy in
Salvador, who according to technical staff did so unenthusiastically and with uncertainty
about the program (Interview Queiroz 2004; Interview Nossa 2004).
A second set of key actors involved in emulation decisions were politically
appointed Secretaries of Health and their senior technical staff.
In most instances,
mayors selected Secretaries who shared similar ideological beliefs and commitments.
Leftist mayors were more likely to hire health secretaries and senior staff who self41 As Mayor of Santos in the state of São Paulo, Capistrano initiated several public health policies to
address the city’s unique challenges as one of the country’s most important port cities. Santos developed
programs for girls and young women to reduce the incidence of prostitution and programs to curb the
spread of HIV-AIDS.
identified as being part of the movimento sanitário. Sanitaristas adhered to similar leftist
ideologies and used consistent rhetorical framing to discuss how they approached their
work in health policy; these actors emphasized their identities, life experiences, and
desire to “make a difference” through their work. They also shared how the family health
program reflected their values and beliefs to improve equity and access to health care and
reverse the tide of government spending that had left out the poor. In their view, full
democracy would require a reprioritization of social policies to address historic and
longstanding inequality and create meaningful citizenship.
Social Networks
The health arena in Brazil has seen particularly robust civil society activity, which
dates back to the mobilization efforts of the Movimento Sanitário during the 1970s and
80s. The sanitaristas promoted the development of university public health programs
across the country, sent doctors far into the country’s interior, published public health
journals, promoted universal health care rights in the constitution, and advocated for
wider civic participation in policymaking through decentralization. Though the
movement has since dissipated as a single entity, its legacy remains in numerous health
care organizations and professional associations it helped establish. Given the vibrancy
of associational life in the health arena and the central role that PSF took in primary
healthcare debates, it is not surprising that the program would spark the attention of
numerous associations.
Two associations created by the Movimento Sanitário,
ABRASCO and CEBES, have been at the forefront of the debate regarding the program’s
Since the inception of PSF, health professionals in Brazilian municipalities had
frequent opportunities to assess the policy through the sector’s numerous and
professional associations. Technocrats had access to official publications from the
Ministry of Health, but more often turned to other sources for the latest information in
their field. PSF administrators for instance, consistently identified ABRASCO and
CONASEMS as important associations and cited the journal Saúde em Debate, published
by CEBES, as a key reference.43 The CEBES publication served as a forum for debate on
the programs, including critiques as well as case-studies from the Northeast, where
authors detailed the merits of PSF. CONASEMS’s annual meetings also became an
important meeting ground for local health officials. When PSF was first introduced in
the mid-1990s, most members were skeptical of the program. Yet, within ten years most
of the participants reported that they had adopted PSF. This turn-around also extended to
academic circles, and by 2003 ABRASCO’s annual meeting included 59 paper
presentations on PSF alone.
What brought about the turn-around of opinion on PSF? First, these organizations
effectively brought together individuals from across the country and provided a forum for
42 Associação Brasileira de Pós-Graduação em Saúde Coletiva (ABRASCO) is a membership organization
for post-graduates in public health, which includes academics as well as government officials. Centro
Brasileiro de Estudos de Saúde (CEBES) is a public health organization with roots in the movimento
sanitário, and its membership comprises health care professionals both within and outside of government.
43 Conselho Nacional de Secretários Municipais de Saúde (CONASEMS) is a national association of
municipal health secretaries; the association publishes research and organizes national and regional
PSF enthusiasts to advocate for the program. The regional diversity of professionals was
crucial because so many proponents were from the Northeast and they might have
otherwise been shut-out of policy debates typically dominated by Southeastern
professionals.44 This Northeastern perspective was crucial because PSF had created
obvious concrete benefits which offered convinving evidence of the policy’s merits.
Second, the health sector and its professional associations comprised of professionals
from different partisan affiliations, many of whom were willing to work across partisan
lines. A notable example of such collaboration applies to two leading figures in the
medical profession, former Minister Adib Jatene and David Capistrano Filho; both men
worked for administrations with very different ideological profiles, yet partnered together
to promote a pilot PSF program in São Paulo.45 This partnership was highly influential as
Jatene drew on his extensive network in the specialized medical field and Capistrano
tapped his network of leftist public health officials.
Lastly, several high-profile
administrators began to show that PSF need not represent “a poor program for the poor.”
Several municipalities demonstrated that PSF could be an all-encompassing strategy for
basic health services and could provide coverage for the entire population. In these ways,
44 This point was made by Luis Odorico de Andrade in reference to CONASEMS (Interview 2004).
45 Adib Jatene, a prominent cardiologist, has worked in various administrative public health positions, first
serving as State Secretary of Health under Governor Paulo Maluf during the military regime. Later he
accepted the position of Minister of Health under Presidents Fernando Collor do Mello (PRN) and
Fernando Henrique Cardoso (PSDB). Dr. David Capistrano Jr. was firmly entrenched in the leftist
Movimento Sanitário and had served as the Mayor of Santos (PT), a port-city in the state of São Paulo.
David Capistrano not only worked across partisan lines by partnering with Adib Jatane, but he also built a
solid relationship with José Serra of the PSDBç see for example, Serra’s warmly obituary for Capistrano
(2000: 195-198).
professional associations connected individuals, filtered information to their members,
and shaped their views and professional norms.
Given the rich and overlapping networks associated with health policy, it is not
surprising that administrators offered sophisticated and consistent analyses for why they
wanted to adopt PSF.
Technocrats frequently invoked similar explanations for the
benefits of the program, including a belief that Brazil should focus on preventive
medicine, a determination to engage directly with communities, and the conviction that
basic medicine should move away from doctor-centric models. In some cases, albeit not
all, technocrats specifically mentioned the Alma Ata conference as having articulated an
important set of goals. Others mentioned the ground-breaking VIII Conferência Nacional
de Saúde, held in March 1986, which demanded that the state provide universal health
care a basic social right. Although a few policymakers expressed skepticism about the
program’s applicability for their cities and even discussed the ways they tried to block the
program, they acknowledged that in ten years the PSF model had become the
professional norm in their field (Interview Franco 2004; Interview F. Santos 2004). This
helps explain why eventually cities like Salvador and Belo Horizonte adopted PSF. In
Salvador, city health administrators acknowledged with some discomfort that they were
relatively “late” in adopting PSF and that several nearby cities were ahead of them in
implementing “model” programs (Interview Queiroz 2004; Interview Nossa 2004). The
sheer density of health care associations and their ability to shape professional norms thus
helps explain the phenomenal spread of PSF across the country.
Although formal overlapping networks tell us a great deal about professional
socialization, it is also important to acknowledge the ways in which informal networks
provided individual policymakers with wide-ranging connections and convinced them to
initiate change.
Oftentimes the informal networks revolved around personal
relationships. For instance, Adib Jatene first met David Capistrano Filho in 1979 when
he served as State Secretary of Health in São Paulo and Capistrano was a sanitarista
working for the state’s public health agency (Jatene 2002).46 Over the years, Jatene had
taken note of Capistrano’s accomplishments in Bauru and Santos, and quickly identified
him as a natural choice for director of QUALIS. In selecting Capistrano, Jatene chose
someone who held tremendous influence within the political left and the movement.
Capistrano would become an important spokesperson for the PSF program and would
persuade many skeptical colleagues on the merits of the policy and that PSF was not a
“poor program for the poor.”47 On an individual level, Capistrano connected informally
with countless actors in the health policy arena and convinced them that PSF was an
appropriate model for primary health delivery, even in the largest metropolitan areas.48
46 As Adib Jatene (2003) explained David Capistrano’s father was a communist and activist who had been
persecuted by the government. Although the military government had later issued an amnesty to political
opponents, then Secretary Adib Jatene was under pressure to fire him. Since there was no evidence that
Capistrano had misused his post for political reasons, Jatene refused to do so. Although their paths did not
cross for many years, Jatene would take an interest in Capistrano’s later accomplishments in Bauru and
Santos. Dr. Jatene held David Capistrano Jr. in the highest regard for his work in public health. He was
especially impressed by the fact that Capistrano left both Bauru and Santos with empty pockets; in other
words, unlike other politicians, he had not enriched himself by stealing public resources (Jatene Interview
47 For instance, members of his health policy team from Santos expressed deep skepticism about the PSF
program; they thought it was a “poor program for the poor”. But he managed to turn around Silveira and
Mattos, who joined him in administering QUALIS (Interview Silveira 2003; Presentation Mattos 2003).
48 Capistrano died in October 2000. Although I was unable to interview him for this research project,
many interview subjects in São Paulo mentioned him with fondness and his endorsement of PSF carried
Aside from the personal contacts that drew actors to learn about PSF, other
informal opportunities arose for learning and socialization. For instance, when technical
staff at the Ministry of Health wanted to promote PSF, they strategically identified
influential staff members in different cities and invited them to visit a “model” city. It
was their belief that once visitors saw how effective the program was, they would be
motivated to adopt it in their own hometowns and would advocate for it with their
supervisors (Interview F. Sousa 2003). For instance, around 1995 and 1996 the ministry
sent key actors in the Federal District to the Northeast to visit model PSF programs
(Interview F. Sousa 2003). One of these individuals went to Recife in the state of
Pernambuco to see how city administrators had adopted PSF; she would later advocate
for it to the Secretary of Heath, Maria José da Conceição (Maninha) (Interview Peixinho
2004). This type of first-hand experience was also crucial for São Paulo’s Secretary of
Health Eduardo Jorge Martins Alves, who credits his enthusiasm for the program on
having seen PSF first-hand in the Northeast as a member of Congress (Interview Martins
Alves 2004).
In these ways social and professional networks, both formal and informal, played
an important role in transmitting ideas and shaping new norms. When individuals were
socialized to believe that a particular policy represented “the model” in their field, they
were especially eager to adopt a similar approach, lest they fall behind their peers. This
dynamic was particularly true for the family health program, which was embraced by the
heavy weight (for example, Interview Silveira 2003; Interview Gouvea 2003; Interview Manfredini 2003;
Interview R. Santos 2003).
densely organized health networks. Interviews with Brazilian policymakers in all four
cities revealed that social and professional networks prompted individuals to influence
the policy agenda by proposing new programs. Social network connectivity was thus a
necessary component for policy diffusion to occur.
Controlling for Federal Transfers
Though the Ministry of Health had an important role in promoting the Programa
Saúde da Família, the impact of its fiscal transfers on diffusion is mixed. Thus, it cannot
rightfully be seen as a motivating cause for actors’ desires to adopt PSF. Starting in the
mid-1990s, a few administrators in the ministry began advocating for a preventive health
approach, represented in PSF. Programa Saúde da Família would eventually evolve from
a small isolated project embedded precariously in the bureaucracy to an established
division in the Ministry that would eventually reorganize basic health care policy around
the program’s design. By the late 90s, the Ministry began providing line-item fiscal
transfers for cities that chose to adopt the program.49 Yet despite the opportunity to gain
access to resources through PSF, many mayors and city administrators refused to institute
the program. In São Paulo, Mayor Paulo Maluf rejected federal funds and a personal
appeal from Minister Adib Jatene to introduce PSF (Interview Jatene 2003). Throughout
the 1990s, health administrators in Belo Horizonte were skeptical of the PSF model and
refused to implement it, despite losing federal resources. Ironically, technocrats affiliated
49 Fiscal transfers for PSF were calculated on a sliding scale and depended on the overall percentage of the
population covered under the program and the number of health teams in place. The formula favored cities
that adopted the program with expansive coverage of the population.
with both the Cristovam Buarque and Joaquim Roriz administrations in Brasília reported
that they implemented PSF despite not having received regular earmarked funding from
the federal government; both sets of officials claimed that political rivalries between
officials in the federal and district governments led to irregular funding.50 As for the
other city administrations, once they chose to adopt PSF, officials from Belo Horizonte,
Brasília, and São Paulo all explained that they appreciated the “extra” PSF financing.
However, they also emphasized that the program was nevertheless very expensive to
operate; the fiscal transfers alone were insufficient to motivate them to adopt the program
because the costs far exceeded the transfers.51 Officials from Salvador were the only
program administrators to cite their policy choices as a reaction to “directives from the
federal Ministry of Health”. Though part of their response was due to the transfers, they
also acknowledged they were “behind” other neighboring cities and needed to respond to
the new paradigm.
Throughout much of the 1990s, when the federal government struggled to define
health care reform, municipal governments became laboratories for experimentation and
innovation in basic health. Cities like Niterói and Londrina were inspired by leftist
50 Cristovam Buarque was affiliated with the PT, which was in opposition to then-President Fernando
Henrique Cardoso of the PSDB. During the Joaquim Roriz (PMDB) administrations (1998-2002; 20022006) Brasília officials encountered resistance from the federal government, which only intensified once
President Luis Inácio Lula da Silva (PT) took office in 2002.
51 For instance, officials in Salvador estimated that each PSF team has an operational costs of R$30,000, of
which the Ministry of Health provides R$2,800 (i.e. less than 10 percent). The Ministry also provides
R$40,000 to for each new facility for PSF. Thus by their calculations, the financial incentives were too
minimal for a large city like Salvador (Interview Queiroz 2004; Interview Nossa 2004).
primary care health models and drew on Cuba’s family doctor model. In São Paulo, the
mayor favored a more conservative approach that integrated public-private partnerships.
Others still embraced a nascent model that integrated community health with integrated
teams of health care professionals, called Programa Saúde da Família. Most importantly,
this period of experimentation demonstrates that for municipalities across Brazil, issues
of basic health care delivery were far from clear or automatic. Rather, politicians and
technical health policy staff debated the merits of different programs and made calculated
decisions on which policies to adopt.
Contrary to conventional wisdom in political science, which emphasizes neoclassical behavioral assumptions that individuals pursue their own self-interest, the actors
in this study did not emulate PSF for electoral gain. In other words, there is no evidence
that mayors adopted the family health program for the purpose of winning an election.
Rather, most emulation decisions were delegated to politically appointed staff and highly
specialized health experts who assessed the ideological and technical merits of the
program. In the few instances where rightist politicians came to embrace PSF for its
political benefits, they did so belatedly and after staff members had initiated the policy’s
emulation. This finding is remarkable given that PSF easily lends itself to clientalism
and patronage politics on the part of mayors.
As this chapter argues, actors’ motivations to adopt PSF were largely driven by
two factors: ideological commitments to social justice and desires to keep up with
evolving professional norms. While these factors were consistently important for all the
case studies, actors’ ideological interpretation of PSF changed over time. Staunch leftists
had to analyze and interpret the PSF to determine if it was consistent with their
ideological commitments, and in many cases their assessments evolved. Professional and
informal social networks were often important features in these processes because they
reinforced actors’ ideological commitments. In other words, professional networks such
as CEBES, ABRASCO, and CONASEMS created opportunities for professionals in
different jurisdictions to learn about PSF and debate the program’s merits and meaning.
Informal networks and friendship ties were also important among leftist politicians and
technocrats. When key opinion leaders such as David Capistrano Filho and Eduardo
Jorge Martins Alves embraced the family health program, it served as an important signal
for others that the policy was consistent with their leftist ideological commitments. The
following chapter will further explore how ideology and social norms operate together to
spur emulation decisions, drawing on evidence from PSF and Bolsa Escola.
Decentralization in Brazil creates the opportunity for thousands of local municipal
governments to tailor social policies for their constituents. This flexibility is particularly
attractive because the country, continental in size, has tremendous regional inequalities.
Despite the autonomy that local governments enjoy in policymaking, the 1990s has
ushered in an era of tremendous policy replication as cities across the country have
emulated program that were designed elsewhere. The question that drives this study is:
what motivates policy makers to emulate innovative social policy? Are actors motivated
by self-interested pursuits to win elections? Or are actors motivated by principled
commitments regardless of self or others? Alternatively, are they motivated by
community-oriented action in an other-regarding way? In answering these questions
about actors’ motivations, this dissertation explores how and why policy diffusion occurs,
and explains how diffusion has been an important feature of Brazilian social policy since
the early 1990s. Chapter 3 provides a statistical analysis of the determinants of social
policy diffusion for Brazil’s 224 largest cities. Chapters 4 and 5 draw on in-depth case
studies to uncover actors’ motivations to adopt Bolsa Escola and Programa Saúde da
The conclusion unifies this work by bringing together the previous chapters. To
accomplish this aim, the first section designs an overarching theory for social policy
diffusion in Brazil. As the second section explains, these findings are consistent despite
differences between qualitative and quantitative methodologies and the two issue areas of
this study.
The third section addresses the implications of this study for diffusion
research and scholarship on Brazilian politics. It concludes by setting an agenda for
future research.
This dissertation tests three distinct approaches to understanding individuals’
motivations for emulating innovative social policies in Brazil. Each corresponds to
distinct analytic research traditions, found in political science and sociology, and their
paradigmatic response to questions of what drives political behavior. Do actors follow
conventional rational choice assumptions about political self-interest and pursue social
policies to win elections? Or, do politicians make policy choices regardless of self and
others and emulate policies because of their deep seated ideological commitments?
Alternatively, do actors respond to a community of shared norms and emulate policies to
demonstrate to their peers that they are in line with their profession’s latest trends? To
answer these questions, this study draws on both statistical and qualitative evidence to
uncover the process by which actors make emulation decisions.
Both Bolsa Escola and Programa Saúde da Família share similar causal
mechanisms that explain why actors are motivated to emulate them. The statistical and
case study evidence reveals similar findings, despite differences between qualitative and
quantitative research designs.
Figure 6.1 provides a visual representation of the
theoretical findings of this study. A more in-depth analysis of each factor, as well
evidence from Bolsa Escola and PSF emulation, follows.
Figure 6.1 Explaining Social Policy Diffusion: A Motivational Approach
Overlapping and Mutually
Reinforcing Effects
A political incentives approach offers an intuitively appealing explanation for the
spread of social policies in Brazil. This rationale speaks directly to rational choice
assumptions about actors’ determination to pursue their political self-interest and to gain
electoral power; it thus represents both the theoretical and instinctive conventional
wisdom. Within political science, the assumption of rationality is one of the principal
ways researchers have sought the regularity necessary for generalization (Riker and
Ordeshook 1973:11). Given this, one of the most surprising findings of this dissertation
is that political incentives fail to explain actors’ motivations for social policy emulation.
As Chapters 3, 4, and 5 show politicians rarely make emulation decisions as
vehicles to advance their electoral self-interest. Certainly, the pressure to win elections
and distinguish oneself from one’s competitor is crucially important in the Brazilian
municipal arena. Campaigning on the provision of these social programs can offer a
clear opportunity to gain votes. Mayors Cristovam Buarque, Marta Suplicy and Célio
Castro all campaigned on their intentions to implement either Bolsa Escola or PSF. Once
in place, these programs offer other electoral benefits, including the potential to target
benefits to voters for future municipal elections.
Mayors who adopt PSF can also
influence voters by deciding which neighborhoods will be served by the program. Some
executives do indeed use these electioneering strategies.
For instance, under the
administration of Mayor Antônio Imbassahy in Salvador, the political leadership
determined which neighborhoods should be included in the health program. Often, the
favored neighborhoods were not those with the greatest epidemiological needs.
For some mayors, these programs are politically appealing not because they offer
opportunities for policy-based electoral competition, but rather because they can
perpetuate “politics as usual” through patronage and corruption.
Since the health
program includes the hiring of new personnel, the power behind job creation offers the
opportunity for significant political payoffs. The community health agent job under PSF
is an especially desirable position among the lower classes because it requires relatively
little education or background in healthcare; often, the only requirements are residency in
the neighborhood served and demonstration of leadership skills. The administration of
Joaquim Roriz in Brasília was especially notorious for irregularities related to the PSF, as
investigations into widespread corruption and the misuse of funds unveiled (Interview
Bisol 2004).
Despite some of the electoral benefits that Bolsa Escola or PSF entail, there is a
wealth of evidence to suggest that mayors and their staffs mostly did not seek to replicate
these programs because of self-interested calculi for electoral or economic gains. As the
event history models reveal, greater electoral competition does not explain the likelihood
that Bolsa Escola or PSF will diffuse. Moreover, the degree of electoral competition does
not explain differences in adoption for the 224 larges cities. The case studies reinforce
these findings in several ways. Politicians in this study never considered these programs
as responses to citizen demand.
Though Bolsa Escola received widespread media
coverage, none of the cities had citizen groups who demanded that candidates or
incumbent mayors adopt Bolsa Escola. In addition, citizen delegates who served on local
health councils rarely advocated introducing PSF; if anything they were resistant to the
program and wary it would not result in improved services.
The poor and most
vulnerable populations served by PSF were unfamiliar with the program and they still
viewed clinics and hospitals – not PSF health teams – as appropriate places to go to for
their healthcare needs. City officials who would adopt PSF would often have to persuade
citizens and local health councils that the program would be an improvement. Thus, the
adoption of Bolsa Escola and PSF often took place in the absence of electoral “demand.”
In fact, mayors’ decisions to adopt these social policies often entailed political
risk that made electoral payoffs far from assured, even when technocrats were confident
in the benefits flowing from these programs. The uncertainties were especially high for
administrations that decided to replicate the programs soon after the innovations garnered
For example, Bolsa Escola could increase school attendance, but by
incorporating previously marginalized and failing students into the system, other
performance indicators could have declined. The complications were even greater for
those administrations that adopted PSF because doing so required reorganizing health
services, updating facilities, conducting new training, and formulating new relationships
with patients. For many cities, adopting PSF also involved assuming responsibility for
services that were previously in the hands of state governments. Even after several years,
technocrats would often report that they had not yet reached their operational goals
because of difficulties related to training, recruiting personnel, and retaining staff.
Despite the risks of policy failure, both Célio Castro in Belo Horizonte and Lídice da
Mata in Salvador signed on to Bolsa Escola very early on; in São Paulo Marta Suplicy
also committed to PSF even though very few other cities with a comparable health
network had chosen to do so.
A further disincentive for politicians to adopt Bolsa Escola and PSF is that they
understood these programs would garner opposition from well-organized segments of the
population. Traditional corporatist sectors such as teachers, medical practitioners, and
civil servants disliked many elements of these social policies.
Bolsa Escola was
unpopular with many teachers and teachers unions, who traditionally favored other
education policies that enhanced curriculum development, school materials, and pay.
Similarly, in nearly every case study, plans for the adoption of PSF caused conflicts with
existing city personnel and resulted in walk-outs or work-stoppages. Unions representing
healthcare workers, doctors, nurses, and other city employees were often skeptical of the
policy because it meant a substantive change in their workday and a reformulation of
their contracts. Given that the programmatic outcomes would be uncertain by the next
election cycle and that politicians would face known resistance from an organized
segment of the electorate, it was hard for elected officials to see clear political gains.
A political incentives approach thus provides a deficient explanation for
understanding social policy diffusion. These cities had competitive municipal elections
with hotly contested campaigns, but political competition did not drive the selection of
the public policies and the speed at which policy replication occurred. Mayors adopted
these policies when there was little demand even knowing there was a chance the policies
could fail to provide positive results by the next election.
Because the political incentives approach cannot convincingly explain diffusion,
we need to turn to other motivational explanations. If conventional rational choice
explanations can not account for the diffusion of social programs, what role, if any, does
ideology play in motivating individuals into action? Did policy innovations spread across
cities governed by mayors on both the left and right? Or do only individuals with certain
ideological commitments feel compelled to adopt programs such as Bolsa Escola and
Do political actors make decisions to implement programs because they are
strongly motivated by their ideological beliefs, even when these choices appear
politically inexpedient or “irrational”?
As the case studies and event history analyses reveal, leftist politicians were much
more likely to adopt Bolsa Escola and PSF; whereas both centrist and right-wing mayors
were unlikely to do so.
Leftist ideology not only mattered to mayors, but also to
technocrats and political appointees, such as secretaries of education and health. Staff
members regularly displayed their ideologically driven preferences and discussed their
policy choices in relation to their values and beliefs. In nearly every case of program
adoption, technocrats, political appointees, and elected officials, who self-identified as
leftist discussed these programs with similar rhetoric.
These individuals invoked
comparable themes and concepts to explain program emulation; one common description
was their desire to address a “social deficit” and prioritize funding to alleviate social
Another theme was the construction of “citizenship” and the view that
education and health care were important social rights. In other words, for these actors,
full citizenship rights in democratic Brazil included the state’s obligation to fulfill basic
social services.
While ideology played a central role in emulation decisions, ideological
interpretations of these programs were not static. Bolsa Escola was quickly embraced by
left-of-center politicians who found the policy consistent with their own ideological
convictions and priorities. This was very different with respect to PSF; many leftist
health policy specialists were initially skeptical about the program and wondered whether
the PSF was really aligned with their ideological commitments. In this case, many health
policy experts consistently expressed their leftist beliefs but it would take time for them
to assess whether PSF could be consistent with their commitment for universal primary
As Chapter 5 explores, health policy specialists on the left found PSF’s mission
laudable but they had to reconcile its potential with other information that caused
cognitive dissonance. For instance, while the program sought to advance health care
access for the poorest, most vulnerable, and socially excluded groups, other features of
the policy seemed to be inconsistent with leftist values. Some actors identified signals
that the policy might reflect more conservative neoliberal values. Since the family health
program had followed in the footsteps of an earlier program targeted to the poor rural
Northeast, some leftists coined PSF a “poor program for the poor.” During Fernando
Henrique Cardoso’s administration, Health Minister José Serra had also approached the
World Bank about financing for the program.
That the World Bank was open to
supporting the program’s expansion served as evidence to some that the policy was a
“Washington consensus” project.
On the other hand, well-known leftist David
Capistrano Filho supported PSF by prominently advocating for it. Many progressive
health policy advocates would have to sort through these mixed messages and interpret
the policy for themselves. In practice, this meant the PSF program was subject to a
longer review process by politicians and their technical advisors to determine whether it
was consistent with their ideological beliefs. Left-of-center ideological commitments
mattered for both policy domains and were consistently important. Yet actors’ leftist
ideology alone fails to account for an all-inclusive explanation of social policy emulation.
Rahter, networks were crucial in making leftists see PSF as compatible with their
ideology. To uncover how education and health policy experts came to embrace the
program, we now turn to the role that social networks played in shaping professional
norms, socializing actors, and motivating them to seek legitimacy in relation to their
Social networks played an important role in motivating actors across Brazil to
emulate Bolsa Escola and PSF. At the most basic level, they provide actors with
knowledge about their profession’s latest trends. In this sense, networks operated as
conduits of information.
But more importantly for a study that focuses on actors’
motivations, networks also socialized actors.
When professional networks provided
space for debate, evaluated policies, and declared them successful or “innovative”, they
went beyond “information exchange” to place a value on that information. Interviews
with politicians and technocrats in Brazil revealed that most actors were embedded in
various forms of associational life. These actors often discussed their work with their
peers; i.e. they referenced whether they were proudly ahead of the curve, quick to notice
the latest developments, or were embarrassed to be behind their peers and thus a late
Municipal policy makers in both Bolsa Escola and PSF policy domains
consistently referred to their peers, suggesting that they sought to keep up “with the
Joneses”, especially when the Joneses were highly regarded innovators. In framing their
work in this manner, actors sought the legitimacy to be gained from their colleagues;
when individuals were linked to social networks their emulation decisions were driven by
social norms and other-regarding motivations.
Social networks mattered in shaping actors’ quest to gain legitimacy.
professional associations were crucial for most technocrats and a few politicians who
held close ties to their profession. These associations were fundamental conduits for
socialization and information exchange across geographical space. As Friedkin notes (as
cited in Kilduff and Tsai 2003:58), professional networks are important because they link
people with similar functional roles but who might otherwise be isolated and lack peers
within their jurisdiction.
Second, informal associations such as neighborhood and
friendship ties also mattered, but these sprouted up in more idiosyncratic and less
predictable ways. Differences between education and health policy arenas explain how
the structure, scope, and density of professional networks can influence emulation
decisions. Thus we now turn to uncovering the differences in social network activity,
both formal and informal, related to Bolsa Escola and PSF emulation.
As Chapter 4 explains, Bolsa Escola represented a hybrid program that embraced
both educational achievement and poverty alleviation.
As a result of these
complementary aims, the program never found a home in a single policy domain. Some
cities housed their programs in departments of education, such as Brasília and Belo
Horizonte; others in departments of public assistance, for example Campinas; yet others
created new agencies, as in the case of São Paulo; or entered into private-public
partnerships, as was the case with Salvador. While the program had clear links to
education and set goals related to increasing school attendance and decreasing grade
repetition, the policy never captured the center of the education sector’s professional
agenda. Rather, Bolsa Escola’s main enthusiasts would be general policy practitioners
interested in innovative public policies, not education specialists.
In general, the
professional associations in education were lukewarm towards the education grant
program, and traditional teachers unions were uneasy about losing resources. For these
reasons it would take a generalist network of policy practitioners associated with the
Gestão Pública e Cidadania network to stimulate actors’ desires for emulation.
Even if professional organizations in the education sector had formally embraced
Bolsa Escola, there would still be important differences between the education and health
sectors. Unlike the field of public health, advocates for progressive socially inclusive
education had never formed a large-scale social movement during the democratic
transition. While university students did join pro-democracy movements, their activities
would not serve as the backbone for primary and secondary educational reforms. Nor
would student groups plant the seeds for long-standing education associations, as was the
case with the sanitary movement. Much of the depth and breadth found in today’s health
sector associations, including organizations such as ABRASCO, CEBES, and
CONASEMS, date back to early organizing by sanitaristas. Today, these public health
associations serve an important role in bringing together influential health policy
technocrats, creating opportunities for learning, socialization, and informal exchanges.
The evidence from the event history analysis and case studies also reveals that
informal networks promoted policy diffusion, albeit in less consistent and more
idiosyncratic ways. In the case of PSF, there were strong regional effects that influenced
adoption decisions; states furthest from the early adopting region of the northeast were
least likely to emulate PSF. In interviews, technocrats also identified which neighboring
cities were ahead of the curve, and many technocrats who promoted the program said
they did so after personally visiting innovative programs (for instance Interview Martins
Alves 2004; Peixinho 2004).
While regional effects did not promote Bolsa Escola
diffusion, informal relationships did. Policymakers involved with Bolsa Escola often
cited learning about the program through colleagues and meetings. One notable instance
offered by Mayor Lídice da Mata is illustrative; she first learned about Renda Mínima at
a meeting where the Mayor of Campinas made a presentation (Interview 2003).
In these ways, politicians and especially technocrats responsible for social policy
development were strongly influenced by their informal and professional associations.
PSF spread quickly because of the health sector’s dense social and professional networks
which shaped experts’ desire to keep up with new professional norms. The education
sector, by contrast, has fewer formal organizations, explaining the relatively slow pace of
Bolsa Escola and Renda Mínima diffusion. Though informal networks mattered, their
effects were more haphazard and weaker than those produced through formal
professional channels. As the analysis shows, social policy diffusion at the municipal
level was strongly influenced by two motivational factors: first, actors’ ideological
commitments and deep-seated desires to enact policies that were consistent with their
leftist worldviews and second, actors’ quest for professional legitimacy and desire to
demonstrate to peers that they were keeping up with professional norms. Each of these
motivational approaches is treated separately in this study for conceptual clarity. Yet
these motivational factors need not operate in isolation; in fact the case studies reveal that
at times, ideology and social networks work together and display mutually reinforcing
effects. For a select number of politicians and politically appointed technocrats, both
these motivations operated together and in complementary ways. The analysis of Bolsa
Escola and PSF emulation offers clarification in this regard.
On occasion, like-minded leftist politicians participated in meetings that not only
introduced them to new social policy ideas but also helped them interpret these policies
as progressive solutions to social exclusion.
This type of serendipitous event was
important for Bolsa Escola diffusion; Mayor Lídice da Mata in Salvador met with
Campinas’ Mayor Magalhães Teixeira at an event and learned about the Renda Mínima
program (Interview Mata 2004). Similarly, Belo Horizonte city Councilman Rogério
Correia said that as a former educator and Workers’ Party representative, he took an
interest in Cristovam Buarque’s education proposals in Brasília (Interview 2004).
Partisan meetings created opportunities for social networking among officials who shared
ideological beliefs. It also gave mayors a sense of trends in local governance as their
colleagues presented information on their most innovative practices. These networking
events served a similar function as professional network meetings for technocrats, except
that it was mayors who sought to gain legitimacy and recognition from their peers.
In the case of PSF, social networks and ideology worked together in slightly
different ways; rather than operating among elected officials, health policy technocrats
and politically appointed health secretaries would find that professional networks could
assist them in evaluating the merits of the policy in ideological terms. Many leftist health
experts rejected the family health program in the early years of its enactment in Brazil. It
was not that they were unfamiliar with it, but rather they experienced cognitive
dissonance between their ideological commitments to expand universal primary care and
PSF’s more limited design. Many leftists wanted universal access for primary health care
and worried that PSF was a poor program for the poor; i.e. a band-aid approach to
offering services in areas that lacked basic health infrastructure rather than a full
coverage program with similar strategies across class lines. Professional networks, such
as CEBES, ABRASCO, and CONASEMS, became venues for debating the merits and
limits of these programs. And when well-traveled and nationally recognized public
health experts such as Adib Jatene, David Capistrano and Luis Odorico Andrade
endorsed the program, the larger community took notice. Moreover, leftists committed to
PSF advocated for the program to offer coverage to the entire population, not just in
selected shanty towns. Expanded coverage helped convince many committed leftists that
the program could offer universal primary care.
In these ways, professional social
networks played a significant role in convincing them that PSF was in line with their
ideological commitments to progressive health care.
To summarize, one of the most surprising findings is that the conventional
political incentives approach, which assumes rationality and emphasizes individuals’
pursuit of their political self-interest, failed to explain social policy diffusion in Brazil.
Many early adopters of these models embraced these programs even though their
electoral repercussions were far from clear. The central theoretical finding of this study
is that ideology and socialized norms drove individuals’ decision-making and their
desires to replicate new policy models in Brazil. Local elected executives often made
choices based on their own ideological commitments and deeply held beliefs.
Technocrats and politicians with strong professional ties, on the other hand, consistently
cited their profession’s norms and their commitment to following the latest trends and
models. The speed and extent of policy diffusion were tied to the density of professional
This study draws on two diffusion traditions to examine the spread of social
policies across Brazilian municipalities. While research on diffusion extends to multiple
social science disciplines, there is a notable divide between those scholars who employ
qualitative methods and those who use advanced statistical analysis. Researchers who
draw largely on qualitative methods and process tracing tend to embrace more
constructivist, or interpretative, approaches to diffusion that emphasize emulation,
learning, “policy transfer,” and norms (see for instance Bennett 1991, 2001; Finnemore
1996; Weyland 2007; Mossberger 2000). By contrast, scholars that employ statistical
analyses with a large number of cases tend to embrace more rationalist assumptions about
actors’ strategic behavior in making adoption decisions (see for instance Berry and Berry
1992; Mooney 2001; Simmons and Elkins 2004). Event history models can also analyze
broader trends, including the ability to understand the risks or likelihood that a given
jurisdiction will adopt a policy.
While Mahoney and Goertz (2006) suggest that qualitative and quantitative
research should be undertaken separately because each is marked by its own values,
assumptions, and presuppositions, this work embraces the view that both methods can be
complementary to explain policy diffusion processes. The large-N component of this
study not only captures diffusion trends in social policy adoption, but also allows for
greater generalization of causality (King, et. al. 1994).
The event history findings
provide a probabilistic interpretation of the likelihood that a jurisdiction adopts either
Bolsa Escola or PSF.
The qualitative case studies offer in-depth analyses of the
mechanisms that lead to diffusion. Process tracing also allows for the possibility of
causal heterogeneity and importantly clarifies policy makers’ motivations for emulating
Bolsa Escola and PSF.
While both methods produce similar explanations for the
motivations that drive policymakers’ adoption of these social policies, it is worthwhile to
explore the differences and similarities between the qualitative and quantitative findings.
One of the most important distinctions between the two methods employed in this
dissertation relates to how we understand diffusion processes. The event history findings
focus on the likelihood that diffusion will occur; once a jurisdiction has adopted a
program the city is dropped from the analysis. The assumption underlying this method is
that once a policy is implemented, it will remain in place. For many governments, this
notion about the nature of public policy is reasonable. Once a new policy is enacted, it
can be difficult to reverse course because actors and institutions have vested interests in
maintaining the status quo.1 However, as the case studies reveal, the Brazilian public
policy environment is more unstable; some municipalities make symbolic name changes
whereas others undergo more serious alterations as policies are suspended or reversed
altogether when new executives take office. For instance, the renaming of programs is
fairly common as politicians seek to place their own stamp on public policy. In Brasília
the family health program has been called Saúde em Casa and Programa Saúde da
Família, and Bolsa Escola has also been Renda Minha and Renda Mínima. For the most
part, the renaming of programs by city mayors remains a symbolic gesture that does not
interfere with the programs’ administration or continuity.
But, a more serious
phenomenon for diffusion research is policy suspension or reversal. The case study
evidence from Brasília reveals that turnover does occur and corresponds with electoral
shifts from a leftist to a rightist mayor.
By incorporating two research methods that examine diffusion processes that can
capture emulation, non-emulation, and reversal, this study is able to uncover the
importance of ideology in decision-making processes. Both the large and small-N studies
offer remarkably consistent findings on the role that ideology plays in social policy
emulation: Left-of-center politicians are more likely, to adopt Bolsa Escola and PSF;
while reversal is relatively rare those instances also inform the general findings of the
importance of mayors’ ideological commitments as only rightlists chose to reverse
course. In addition, the case studies of the Buarque and Roriz administrations show that
1 In the study of American politics, incrementalism is a main feature of policymaking, rather than
wholesale reinvention or reversal (Lindblom 1959; Pierson 1994).
ideology also matters when it comes to policy reversal. In other words, when a rightist
comes into office and lacks the same ideological predisposition to address social
inequality, this mayor is more willing to suspend a program he identifies as “leftist.” In
this way, the qualitative research, which includes an in-depth analysis of “outlier” cases,
can contribute to a more nuanced understanding of both diffusion processes as well as the
role of ideology in motivating politicians’ decisions.
Given that large- and small-N research draws on different types of data, it is only
natural that differences in measurement for the theoretical variables of interest could also
contribute to meaningful insights. Both qualitative and quantitative approaches test the
extent to which electoral competition, ideology, and social networks matter in driving
emulation decisions, but these causal factors are conceptualized slightly differently. The
data intensive requirement for the event history analysis necessitate in some instances
imperfect measures for the theoretical variables of interest. Assumptions about ideology
for instance are based on an actor’s partisan affiliation, but their partisanship might not
truly capture their worldviews.
By contrast, process tracing allows for “thick
description” that can reveal interpretive accounts such as the meaning an actor ascribes to
her ideological beliefs.
The measures of social network connectivity and their effect on stimulating social
policy emulation serve as a clear example of how both methods draw on slightly different
measures to test the theoretically driven hypothesis that policymakers who are part of
networks will seek to ‘keep up with the Joneses.’ Recall, for instance, that the measure
of formal network connectivity is based on membership and participation data provided
by CEBES and the Public Management and Citizenship Program. The data from the
Public Management program is of better quality because it includes more information on
who participated and accounts for annual fluctuations. In the case of CEBES however,
administrators had not kept annual databases on membership, but rather continuously
updated a single list.2 Since staff members emphasized that the geographical distribution
of members was fairly constant over time, values were simply repeated for every year in
this study. Nevertheless this solution to a shortcoming in the data likely leads to a less
powerful network effect for the PSF model.
The case study analyses enhance
understanding of socialized norms by allowing subjects to identify which networks
matter to them. The interviews also uncover how actors view different associations and
engage in various professional and informal networks. In this way the qualitative cases
show how participation in the Public Management and Citizenship Program and CEBES
networks mattered for policy makers. In addition, the case studies also shed light on the
unique features of civil society activity in the health sector, to explain PSF emulation.
Debates on the merits of PSF extended beyond CEBES to fora organized by ABRASCO
and CONASEMS among others, and emphasized the nature of the field’s dense and
overlapping health associations.
On the face of it, both the qualitative and quantitative methods examine ideology
similarly – elected officials and their senior political appointees are classified as leftists,
centrists, or rightists. The statistical models reveal that the presence of a leftist mayor has
a statistically significant impact on the likelihood of Bolsa Escola and PSF emulation.
2 The ABRASCO membership database was managed in a similar manner.
Also important is that centrists and rightists behave alike in the large-N study; in other
words centrists and rightists adopt these policies at similar rates. But what is it about
leftists that makes them so eager to emulate these social policies? As the case studies
reveal, leftists who adopted these programs shared similar commitments, beliefs, and
desires to govern in a way that fit their worldview. They were driven by a dedication to
address historic social exclusion and reverse decades-old spending priorities that
benefited the elite and middle-class. Interestingly, most of the leftists interviewed for this
study framed their policy choices as driven by their belief that Brazil needed to construct
“citizenship.” Pro-poor and equity enhancing policies like Bolsa Escola and PSF were a
means for expanding those citizenship rights.
Since questions regarding actors’
motivations are fundamentally about how they interpret social policy and attribute
meaning to their actions, research must go beyond the statistical correlates to include
qualitative approaches that explore individuals’ motivations.
Lastly, the methodologies employed in this study examine the potential impact
that electoral competition has on policy diffusion. Capturing electoral competition is one
of the most challenging issues for a longitudinal study of Brazilian municipal elections
because of irregularities of electoral data collection during the early 1990s.3 Thus, the
case studies provide an opportunity to fill-in information where data is lacking. At the
same time, the case studies provide “texture” on the nature of electoral competition for
the research sites. For instance, when left-of-center Mayor Lídice da Mata won executive
3 Data for the 1992 mayoral elections are unavailable for most municipalities. Even data on the city of São
Paulo is unavailable from the state’s election tribunal.
office of Salvador in 1992, her rise reflected a brief political opening due to the declining
influence of the conservative politician Antônio Carlos Magalhães who had dominated
Bahian politics for decades.4 By 1996, ACM would once again reign supreme in both
state and municipal politics, and his candidate for office, Antônio Imbassahy, would win
handily. In this way knowledge of specific local elections not only elucidates the nature
of the campaign debates and election, but it also reveals how political contexts sometimes
reflect broader political dynamics tied to state and national level alliances.
Scholarship often focuses on a single policy or political phenomenon to explore
the determinants of policy diffusion (for recent examples see, Brinks and Coppedge
2006; Mossberger 2000; Orenstein 2003; Simmons and Elkins 2004). By contrast, one of
the important features of this project is that it has examined two distinct policies, one
related to health and the other to education. There are important theoretical rationales for
selecting two social policies. First, this study seeks to avoid the problem of a proinnovations bias by examining only those programs that diffuse broadly and rapidly.
Both policies analyzed in this dissertation diffuse, but do so at dramatically different
rates; PSF spreads more extensively than Bolsa Escola. Second, research on public
policy tends to emphasize differences across issue area, rather than similarity.
Conventional policy studies highlight the distinct features of each policy domain,
examining entrenched interests, class conflicts, and the unique actors and institutions
involved in each sector. However, it is also possible that distinct policies are driven by
4 Senator Antônio Carlos Magalhães had supported President Collor. When Collor was impeached on
corruption charges, ACM who had long been implicated in shady deals himself, lost some political capital.
similar causal processes when it comes to emulation. To explore some of the unique and
shared features of Bolsa Escola and PSF, this section draws on the analysis presented in
Chapters 4 and 5, specifically highlighting their inherent qualities including: the degree
of policy design complexity, policy “flexibility”, and the gendered political appeals of
these programs.
One of the most striking differences between the two social policy issues in this
study is their degree of complexity. Bolsa Escola represents a straight-forward and
simple idea with modest goals to improve educational access.5 Key policy entrepreneurs
such as Cristovam Buarque made the analogy that Bolsa Escola was just like collegelevel grants for needy students, the only difference being that the school grant targets
needy children. The simple comparison made it easier for many Brazilians to relate to
the program, even though the policy’s design seeks to address more complex issues of
intergenerational poverty, child labor and the opportunity costs of education, and the need
to induce attitudinal and behavioral changes among parents.
implementation of Bolsa Escola was also uncomplicated; it merely required the city to
create a database of beneficiaries with corresponding school attendance information.
Teachers would need to submit information to the city, but Bolsa Escola did not
otherwise change the core of their job duties.
The clear-cut features of Bolsa Escola contrasted dramatically with the complex
nature of the family health program. Unlike Bolsa Escola, PSF represented a highly
5 Scholars in various fields have examined how simple or “flexible” ideas make them more likely to
diffuse. See for instance: Mossberger (2000) on the diffusion of empire zones in the American context and
Heath and Heath (2007) on why some ideas stick and others do not.
sophisticated re-visioning of health care delivery. The health program encapsulated
decades of domestic and international debates on how to structure primary health care
(see Chapter 5). Not only did it require that municipal health agencies shift resources in
public health, it also called for many health professionals to embrace the new model.
Unlike Bolsa Escola, PSF implied a dramatic change in the nature of health workers’ day.
For instance, many doctors and nurses who had existing civil service employment with
city governments had to renegotiate contracts.6 To hire community health agents, whose
job description conflicted with traditional civil servant job contracts, city officials entered
into partnerships with local non-profit associations.7
The sheer complexity of the
program’s administration offered a striking contrast to Bolsa Escola. Many policymakers
viewed PSF as a “big idea” which involved substantial commitment to reorganize health
care delivery in Brazil. For these reasons, it would seem that these two policies reflect
dramatically different types of social policy.
These differences in complexity (both ideational and administrative) did have an
impact on policymakers’ acceptance of these social policies.8 Key policy entrepreneurs
associated with Bolsa Escola and PSF would experiment with how to frame their policies
6 Most doctors with municipal contacts who worked in the four research sites had contracts with part-time
shifts. Thus, in practice many doctors in Brazil work hold two jobs to fill a full-time schedule. PSF
however, requires that doctors not only work within a designated territory but that they do so full-time.
Contract negotiations with municipal health care workers were often a source of conflict between labor
unions and municipal administrators (Interview M. Souza 2004: Interview Costa 2003; Interview
D'Agostini 2003; Interview Camara 2004).
7 One of the job requirements for the ACS is residency in the district in which they work. If a community
health agent moves, she is subject to job loss. Thus, administrators often need flexibility in both hiring and
firing. One mechanism cities have used to ensure flexibility is to sub-contract work through non-profit
associations. For instance, São Paulo established partnerships with fourteen institutions to implement PSF
throughout the municipality (Sousa 2003: 93).
8 Several scholars have examined the impact of complex, simple, and flexible ideas on policymaking; see
especially (Mossberger 2000).
to maximize their acceptance.
For instance, Cristovam Buarque had at one point
described Bolsa Escola as a policy that paid mothers for the work they already do; this
characterization did not sit well with feminists or conservatives and thus he quickly
abandoned that description to describe the program as a school scholarship (Interview
Almira 2003). In the case of PSF, the framing issue did have substantive implications.
Did the program represent “primary health” care or “selective primary care?”9 While
framing issues are important to both social policies, the simplicity behind Bolsa Escola
was an especially important feature that facilitated its acceptance.10
Policy flexibility was another element of Bolsa Escola and PSF that a few
technocrats found to be attractive. While most cities adopted these programs wholesale
and made few modifications to their policy design, in a few cities technocrats found ways
to build upon these policies once they were in place. Bolsa Escola and PSF could serve
as the foundation for additional program elements.11 For instance, in Belo Horizonte,
administrators responsible for Bolsa Escola had previous professional experience in
public assistance and social work. While they could have simply created a registry and
made cash payments to beneficiaries, the staff decided to institute monthly meetings for
the mothers in the program. Mothers were invited to select topics for discussion; among
the issues Bolsa Escola beneficiaries were most eager to address was how to search for
9 I borrow from Cueto (2004) in making this distinction.
10 Cristovam Buarque himself credits the simplicity of the idea as one of its general appeals (Interview
11 I am grateful to Peter K. Spink who made this observation.
employment. In this way, Bolsa Escola in Belo Horizonte became more than a school
grant program, namely an outreach and development program for poor mothers as well.
While the Ministry of Health set clear rules for how the PSF program ought to be
implemented in order for cities to qualify for federal grants, many cities would enhance
their PSF program with complementary services or distinct approaches.12 For instance,
Sobral in the state of Ceará was an early adopter of the family health program. But one
of the modifications city officials embraced was the expansion of services to include oral
hygiene by dentists as part of their system for primary medicine.
In São Paulo,
administrators affiliated with PSF created a birthing center for women as an alternative to
hospital deliveries. Sometimes these local experiments with PSF were viewed with
skepticism by federal health administrators who preferred a unified approach, while at
other times, new ideas would “trickle-up” and the addition would be embraced by
national policymakers.
A central question in diffusion studies is whether adoption decisions truly reflect
emulation of an existing policy. Do policymakers adapt an innovative policy before
implementing it, thus rendering it different? Or do actors simply apply a “policy label”
without adopting the specifics of policy design?13 In this study, both Bolsa Escola and
PSF were largely emulated wholesale by policymakers; few actors tweaked these
programs before adopting them. Rather, it was after cities adopted these policies that
12 For instance, the Ministry’s preference for PSF meant that in practice, cities such as Niterói with the
Programa Médico da Família did not qualify for federal PSF funds. Later, the Ministry would support
“similar” programs albeit with reduced levels of funding.
13 Mossberger defines policy labels as general concepts, with or without some elements of policy design,
and symbolism (2000:116-117). For greater discussion of policy labels and diffusion process, see for
technocrats would experiment by adding different components. That Bolsa Escola and
PSF were perceived to be useful foundations for complementary activity speaks to their
general appeal. It can also help explain why technocrats were willing to buy into these
programs and why these policies may have “stuck” as city administrators established
vested interests in these programs.
To summarize, one of the important features of this study is that is brings together
two different research traditions to uncover actors’ motivations for social policy
emulation. Employing distinct approaches in a single study makes up for the inherent
weaknesses of each method, clarifying the mechanisms that drive diffusion while also
allowing for greater generalization. A second analytic characteristic of this dissertation is
its comparison between two different types of policy issues. Bolsa Escola is a relatively
small program with specific aims geared towards a particular constituency. Programa
Saúde da Família, on the other hand, has broad goals to redefine an entire class of health
care services to all families. These differences alone might suggest these programs are
not comparable. However, both social polices also share characteristics that make them
similarly attractive to administrators seeking to make their mark.
Ultimately, both
policies appeal to leftist politicians and technocrats who embrace the policies for their
social justice and citizenship goals.
The findings in this study contribute to several research endeavors, including
disciplinary debates on actors’ motivations in political decision making, theories of
policy diffusion, and Brazilian politics. Also important for those with a particular interest
in poverty alleviation and equity enhancing public policy, this project informs
practitioners of the mechanisms that enable innovative strategies to diffuse.
One of the most central debates in social science has been: what motivates
individuals’ political behavior? Different disciplinary traditions have emphasized such
divergent factors as self-interest, socialized norms, or the power of ideas. While fields
such as political science and sociology have pursued theory-building exercises based on
these underlying assumptions, diffusion research which crosses disciplinary divides has
tended to leave motivations unspecified and untested.
In other words, diffusion
scholarship in political science has emphasized the importance of rationality and electoral
competition to explain policymaking while similar research by sociologists has
emphasized the role of social networks (see for instance, Berry and Berry 1992;
DiMaggio and Powell 1983; Granovetter 1973; Walker 1969). One of this dissertation’s
main contributions is its conceptual framework, which focuses on individuals’
motivations for policy emulation, filling a theoretical lacuna by highlighting actors’
motivations in diffusion research.
In testing three competing explanations for emulation decisions, this research
reveals that electoral competition fails to explain the diffusion of Bolsa Escola and PSF.
This finding is particularly noteworthy given that electoral self-interest is one of the basic
behavioral assumptions in political science. Rather, actors can be driven to emulate
innovative policies for principled and other-regarding reasons. This surprising result
suggests that future research on diffusion should take care to avoid embedding behavioral
For this reason, research designs should make sure to assess the
mechanisms that drive emulation decisions. While statistical techniques such as event
history analysis can facilitate generalization for a larger number of cases, case study
approaches and process tracing are necessary to uncover the mechanisms that drive
emulation decisions.
This research project also offers important insights for scholars of Brazilian
politics. Most research on social policy in Brazil has emphasized the role of institutions
or sector-specific features of health and education (see for instance Arretche 2000; 2002;
2004). Yet, with greater municipal authority in decision making, diffusion research
provides a new analytic lens for understanding both the horizontal and vertical processes
that drive the spread of similar policies across the country. The replication of social
policy in Brazil offers a unique opportunity to explore diffusion; it has occurred in a
country with thousands of local governments that is well-known for its vast contrasts
rather than similarities. Policy emulation in Brazil is puzzling precisely because of
dramatic regional differences; that a mega-city such as São Paulo with elaborate health
networks, would adopt a health care model which owes its origin to poor small cities in
the rural northeast, is remarkable. Similarly, it is surprising that a city like Salvador with
high rates of poverty and poor educational infrastructure would implement a school-grant
program developed for wealthy cities with some of the highest rates of human
development (Martins and Libânio 2005). The spread of the same policy model across
such diverse settings is worthy of explanation.
Lastly, this study has important implications for those concerned with the practice
of social policy development in Brazil’s local governments. Having contrasted two
policy arenas of central concern to local governments, we see that the motivations that
drive policy emulation can result in different outcomes. Not only did policies diffuse at
different rates, but the likelihood that a policy experienced “longevity” differed as well.
When ideology was the predominant motivation for policymakers, as was the case with
Bolsa Escola and Renda Mínima, the programs were much more vulnerable to policy
reversal once there was a turnover in government. Municipal Bolsa Escola never became
a standard model for education reform that encompassed new norms across education and
public assistance associations. Rather, supporters of the educational grant program were
socialized through the generalist Public Management and Citizenship network. This
contrasts significantly with programs such as PSF which become defined as the new
professional standard. In this case, its diffusion was extensive and the program was much
more likely to remain in place once implemented. Within four years of PSF’s earliest
adoption, associations such as CONASEMS, CEBES, and ABRASCO served as fora for
programmatic debate. If we consider “good governance practices” to be those that
encompass some measure of policy regularity across administrations, then there is reason
to believe that the existence of a professional society really does matter. By investing in
social networks that can cross ideological divides and socialize individuals into shared
professional norms, advocates of social policy innovation can go far to promote diffusion.
As is often the case with research endeavors, this dissertation has generated
several questions for future research to address. Field research offers many benefits
including the ability to observe, “soak and poke.” Thus, some questions sprang up in the
process of conducting field research, but were set aside for the sake of coherence. Other
questions are generated from the findings themselves. I will address each in turn.
As this dissertation briefly discusses in Chapters 4 and 5, Bolsa Escola and
Programa Saúde da Família have important implications for women: as direct
beneficiaries of the program, as workers, and through the social construction of gender
relations. Mothers were the targets of Bolsa Escola grants, women make up the bulk of
PSF health agents and team members, and these programs have the potential to either
transform gender relations or reinforce traditional family norms. An interesting similarity
across both social policies examined in this study, includes the way in which motherhood
and “family” served as focal points for organizing social services.
These socially
“neutral” themes do not challenge mainstream constructions of gender and family values
in Brazil, which likely enabled their adopters to see them as generally acceptable.14
The gender politics surrounding Bolsa Escola was particularly explicit in that
mothers were identified as the most responsible adults in children’s lives; only mothers
could be trusted to spend their children’s grants to further their education and basic living
Innovations are more likely to spread when innovative ideas are compatible with pre-existing values and
beliefs (Rogers 2003: 240-243). Mensch et. al (1999) argue that gendered norms can be difficult to change,
for instance, despite laws banning female circumcision, clitoridectomy remained a common practice in
Ghana; a diffusion campaign to outlaw the practice was slowly adopted because the policy conflicted with
cultural norms (as cited in Rogers 2003: 242-243).
needs. Fathers were largely dismissed by policymakers as either impractical to work with
or less likely to fulfill their paternal roles. Some administrators pointed to pragmatic
concerns such as the large number of female-headed households and noted potential
problems that could arise if men fathered children with several women. A few others
embraced progressive discourse about empowering women, but these viewpoints were
rare and largely absent from the public discourse on the program. Non-specialists and
politicians were more likely to revert to stereotypes to explain the gendered appeal of
Bolsa Escola; that fathers might use the grants to drink alcohol, was the most common
dismissal. In practice, Bolsa Escola’s political appeal was tied to these traditional social
norms surrounding motherhood and the prioritization of family.15
In some ways, the gendered dimensions of the PSF program were more
ambiguous. On the face of it, the PSF program seemed at odds with the broader trends in
the field of public health to make sex and gender explicit in organizing health services.
Public health in Brazil has included a focus on women’s health issues; particularly as
they relate to women’s reproduction, infant, and maternal health.16 In the last decade,
public education campaigns on domestic violence also shed light on health impacts of
violence against women. These larger trends in public health to provide specialized care
for women and acknowledge gender relations in Brazil were less explicit under the
15 As Chapter 4 explains, Senator Eduardo Suplicy had initially proposed a minimum income for all
individuals, regardless of marital status and number of dependents. This individualistic approach to social
policy was rejected by Brazilian policymakers who embrace the idea that social policy ought to be centered
on vulnerable families with children.
16 Here I refer to traditional development concerns surrounding family planning and fertility as well as
international public health goals (GOBI) that include promotion of maternal breast feeding.
family health program. The family,17 rather than the individual, was the “unit” to receive
Special attention to girls and women would require that health care
professionals mainstream gender in their work.
In general, it is possible that an
integrated family approach can result in better care, as health workers are able to
contextualize health problems and see patients’ home environment.
But successful
integration of women’s health depends on the quality of training programs and the
sensitivity of nurse supervisors.
Yet these concerns were rarely expressed by
administrators of the program in research sites I visited. Politically, PSF has a genderneutral façade and appeals to traditional constructions of nuclear family life. The logo
that accompanies the program in posters, clinics, and brochures includes the image of two
parents with two children within a house. Unlike specialized health programs based on
women’s reproduction or disease-centric care, the PSF image has a broad appeal that
everyone can relate to while also reinforcing notions of family life embraced by social
The gender dimensions of Bolsa Escola and PSF rarely entered into policy
debates on their merits.18 Social policy experts hardly ever highlighted issues of gender
norms when discussing these programs and non-governmental feminist advocacy
organizations seldom commented on the value of these programs for women. The low
levels of gendered discourse on social policymaking is surprising because Brazil is home
17 In practice, health teams in the four research sites explain that they view “family” in terms of
“household”. The registry of families includes all persons in the same domicile, regardless of relational
18 Street level bureaucrats acknowledged gender and family relations much more so than technocrats or
to a women’s movement that made significant strides in pressing for democratization
(Alvarez 1990; Jacquette and Wolchik 1998) and that has since formed professional nongovernmental organizations to press the women’s rights agenda (Alvarez 1990; 1999). In
addition, the country has also made efforts to create spaces for women through the
institution of a women’s bureaucracy and gender quotas for political parties. What does
it take to mainstream gender into social policy debates in Brazil? This question certainly
speaks to central research questions in comparative feminist scholarship, including
studies on feminist policy in the European welfare state, as well as other developing
countries (see for example Jordan 2006; Mazur 2002). This question for the Brazilian
context is for now left unanswered and awaits a future research opportunity.
Another question outside the scope of this dissertation is whether Bolsa Escola
and PSF are effective or should diffuse throughout the country. Both policies have won
awards for policy “innovation,” yet accolades are generally based on a few case studies.
It is less clear for instance, if these programs will be effective once installed elsewhere.
Can an educational cash-grant program produce desired results if implemented in a city
that lacks other educational resources for school achievement?
Should cities with
complex health infrastructure, where residents prefer clinician-based services, adopt a
family health program? As this dissertation has addressed, many of the cities that have
adopted these policies did so without undertaking research to determine if these programs
are indeed appropriate solutions to their city’s social problems. These issues about policy
efficacy are important, especially since actors who adopt them assume they will address
problems of social exclusion, poverty, and inequality. To answer these questions, social
policy researchers would need to conduct policy evaluation research; it is my hope that
specialists in Brazilian public policy analysis will undertake this kind of investigation in
the near future.
Scholars of Brazilian politics may also take to heart that this dissertation’s
findings have emphasized the importance of ideology in guiding social policy decisionmaking at the municipal level. At the same time, recent politics in Brazil at the national
level suggests that the role of ideology might matter less than it does for local politicians
and technocrats. The re-election of President Luiz Inácio Lula da Silva after the broad
extension of conditional cash transfer programs suggests that electoral competition might
matter more at the national level than at the local level (see Appendix F; Hunter and
Power 2007). One possibility is that the electorate is more aware of trends in national
policy and has difficulty ascribing credit for local policy innovation. Another possibility
is that the effects of national social policy can be felt more broadly. Regardless, this
difference in national and local policy making has remained outside of the purview of this
dissertation, which has focused on local policy emulation. Future research on social
policy should address whether policymakers’ motivations to enact social policy reform
differ according to the arena where they work; i.e. do electoral incentives, ideology, and
socialized norms operate differently at the national level? This is yet another question for
future investigation.
This dissertation has set an agenda for future scholarship on policy diffusion. As
I have argued, the findings in this study underscore the need to test actors’ motivations in
political decision making.
When diffusion research embeds disciplinary and
paradigmatic behavioral assumptions about political behavior, scholars run the risk of
misinterpreting political activity and the mechanisms that drive diffusion. For these
reasons, future diffusion research should specify and test assumptions about political
motivations. In doing so, diffusion scholarship may shed light on broader political
science debates on rationality and self-interest. This project bucks conventional wisdom
that self-interest, as demonstrated through electoral competition, drives resource
allocation decision making. While these findings may be specific to local politics in
Brazil, it is also possible that similar trends exist elsewhere and for other diffusion
phenomena. In order to integrate diffusion studies with larger theory-building exercises
in political science, future research will need to explore the explanatory power of
electoral competition, ideology, and social networks for other country studies and
transnational cases of diffusion.
Rio Branco
Vitória da Conquista
Teixeira de Freitas
Lauro de Freitas
Feira de Santana
Juazeiro do Norte
Distrito Federal
Vila Velha
Cachoeiro de Itapemirim
Rio Verde
Aparecida de Goiânia
Águas Lindas de Goiás
São Luís
São José de Ribamar
Teófilo Otoni
Sete Lagoas
Santa Luzia
Ribeirão das Neves
Pouso Alegre
Poços de Caldas
Patos de Minas
Montes Claros
Juiz de Fora
Governador Valadares
Conselheiro Lafaiete
Belo Horizonte
Campo Grande
Várzea Grande
Santa Rita
João Pessoa
Campinha Grande
Vitória de Santo Antão*
Jaboatão dos Guararapes
Cabo de Santo Agostinho
São José dos Pinhais
Ponta Grossa
Foz de Iguaçu
Volta Redonda
São João de Meriti
São Gonçalo
Rio de Janeiro*
Nova Iguaçu
Nova Friburgo
Duque de Caxias
Campos dos Goytacazes
Cabo Frio
Belford Roxo
Barra Mansa
Angra dos Reis
Porto Velho
Boa Vista
Sapucaia do Sul
São Leopoldo
Santa Maria
Santa Cruz do Sul
Rio Grande
Porto Alegre
Passo Fundo
Novo Hamburgo
Caxias do Sul
São José
Jaraguá do Sul
Nossa Senhora do Socorro
Taboão da Serra
São Vicente
São Paulo
São José dos Campos
São José do Rio Preto
São Carlos
São Caetano do Sul*
São Bernanrdo do Campo
Santo André
Santa Bárbara d'Oeste
Rio Claro
Ribeirão Preto
Ribeirão Pires
Presidente Prudente
Praia Grande
Moji das Cruzes
Mogi Guaçu
Itapecerica da Serra
Franco da Rocha
Francisco Morato
Ferraz de Vasconcellos
Bragança Paulista
* Denotes cities where municipal authorities declined to participate in the phone survey, and thus
excluded from Bolsa Escola analysis.
Alves, Rita de Cassia. Administrator, Finance Department in Municipal Department of
Health, São Paulo, September 17, 2003.
Almeida, Aide. Administrator in Health Program, UNESCO-Brasília. December 16,
Almeida, Ivonette Santiago de. Administrator, Finance Department with Secretary of
Health (Federal District). Brasília, December 3, 2003.
Andrade, Luis Odorico de. President of CONASEMS and Municipal Secretary of Health
of Sobral (CE), Natal, March 20, 2004..
Aguiar, Marcelo. Chief of Staff at Ministry of Education, Brasília, November 24, 2003.
Araújo, Raimundo Caires. Municipal Secretary of Work and Social Development,
Salvador, July 2004.
Augusti, Maria Teresa. President, Instituto Florestan Fernandes. São Paulo, October 27,
Barbosa, Alfonso Celso Renan. Former Coordinator of Bolsa Escola, Municipal
Department of Education, Belo Horizonte. March 22, 2003.
Bandeira, Célia. Former Special Secretary for Monitoring (Secretária Extraordinaria de
Acompanhamento), under Lídice da Mata, Salvador, July 9, 2004.
Bisol, Jairo. Public Prosecutor, Public Prosecutor’s Office for the Federal District,
Brasília, May 4, 2004.
Boa Sorte, Alfredo. President of Sindicato dos Médicos. Salvador, May 31, 2004.
Borio, Jõao Carlos. Administrator, Municipal Department of Workforce and Social
Development. São Paulo, September 15, 2003.
Bruno, Naire. Staff member, Municipal Department of Public Assistance, São Paulo,
November 13, 2000.
Buarque, Cristovam. Former Governor of the Federal District, Brasília, April 26, 2004.
Camara, Gilherme Ribeiro. Representative, Sindicato dos Médicos-MG. Belo Horizonte,
April 6, 2004.
Campos, Claúdia Valentina de Arruda. Researcher with Fundação Getúlio Vargas-São
Paulo consulting group. São Paulo, September 17, 2003.
Castro, Maria Céres Pimenta Spínola. Former Secretary of Education, Municipal
Department of Education of Belo Horizonte. Belo Horizonte, February 11, 2004.
Castro, Laura Alfonso. Bolsa Escola Administrator, Municipal Department of Education,
Belo Horizonte, March 4, 2004.
Castro, Marcelo Lúcio Ottoni de. Senate Health Researcher, Brazilian Senate, Brasília,
May 3, 2004.
Coelho, Cristina. Administrator of PSF, Municipal Department of Health for Belo
Horizonte, Belo Horizonte, February 16, 2004.
Conceição, Maria José da (Maninha), Former Health Secretary of the Federal District,
Brasília,December 1, 2003.
Correia, Rogério. Former City Counselor for Belo Horizonte, Belo Horizonte, March 29,
Costa, Célia Regina. Union Representative, SINDESAÙDE, São Paulo, October 30,
Cunha, Célio da. Program Staff. UNESCO, Brasília, January 22, 2004.
D’Agostini, Angelo. Union Representative, SINDESAÙDE, São Paulo, October 30,
D’Angelo Francisco. Secretary of Health, Municipal Department of Health for Niterói,
Natal, March 17, 2004.
Dimitrov, Pedro. Senior Advisor to Secretary of Health, Eduardo Jorge Martins Alves.
São Paulo, November 6, 2003.
Escorel de Moraes, Sarah Maria. Professor at Fundação Oswaldo Cruz and Consultant to
Ministry of Health, Natal, March 2004.
Elias, Paulo. Professor and Researcher, CEDEC, São Paulo, September 19, 2003.
Eon, Fábio. Administrator, Communications Department, UNESCO-Brasília. Brasília,
December 16, 2003.
Fernandes, Silvio. Secretary of Health, Municipal Department of Health of Londrina.
Natal, March 19, 2004.
Fontes, Alexandre. Archivst with Fundação Perseu Abramo, São Paulo, September 18,
Franco, Túlio Batista. Adjunct Secretary, Municipal Department of Health, Belo
Horizonte, February 12, 2004.
Freitas, Estanislau de. Administrator, Municipal Department of Workforce and Social
Development, São Paulo, October 2, 2003.
Galeano, Paula. Chief of Staff to Coordinator of Bolsa Escola Federal, Ministry of
Education, São Paulo ,April 16, 2004.
Geddes, José. Former State Secretary, State Department of Health of São Paulo, São
Paulo November 4, 2003.
Goldbaum, Moisés. Professor and President of ABRASCO, São Paulo, October 27, 2003.
Gomes, Cid. Mayor of Sobral (Ceará). Sobral, July 8, 2004.
Gouvea, Isamara. Administrator of PSF, Municipal Department of Health, São Paulo,
October 1, 2003.
Guedes, Ana Cláudia. Executive Secretary of CEBES, Rio de Janeiro, October 20, 2003.
Ibañez, Antonio. Secretary, Department of Education (Federal District). Brasília,
December 15, 2003.
Jatene, Adib. Former Minister, Ministry of Health, São Paulo, October 16, 2003.
Junkeira, Virgínia. Researcher, State Institute of Health, São Paulo, October 15, 2003.
Kayano, Jorge. Staff member, PÒLIS Institute. São Paulo, October 18, 2003.
La Forgia, Gerald M. Program Staff, Health Division, World Bank-Brazil, Brasília, May
4, 2004.
Leitão, Elizabeth Former Director of Bolsa Escola, Municipal Secretary of Education of
Belo Horizonte, Brasília, January 19, 2004.
Lima, Lílian Carneiro. Coordinator, Renda Minha, Department of Education of the
Federal District, Brasília, December 10, 2003.
Lopes, Conceição Zotta. Bolsa Escola Administrator, Department of Education of the
Federal District, Brasília, November 20, 2003.
Lorenzo, Rosicler Aparecida Viegas Di. State Coordinator of Programa Saúde da
Família, São Paulo, November 18, 2003.
Machado, Heloisa. Former Director of PSF, Ministry of Health, Brasília, November 21,
Madeira, Wilma. Staff member, Instituto Florestan Fernandes, São Paulo, October 27,
Manfredini, Marco. Chief of Staff, Office of City Counselor Carlos Neder, São Paulo,
September 24, 2003.
Mariani, Mônica. Staff member, ABRASCO, Rio de Janeiro, October 20, 2003.
Mata, Lídice da, Former Mayor of Salvador, Salvador, July 16, 2004.
Magalhães, Ines. Former Official at Workers’ Party National Headquarters, Brasília,
January 28, 2004.
Martins Alves Sobrinho, Jorge Eduardo. Former Secretary of Health in São Paulo, Natal,
March 19, 2004.
Mendes, Vera. Professor School of Public Administration, at the Federal University of
Bahia, Salvador, May 23, 2004.
Meneses, Milton. Director of PSF in the Federal District, Brasília, January 2004.
Miura, Hiromi. Administrator, Secretary of Health of the Federal District, Brasília,
December 9, 2003.
Nossa, Sonia. Staff member of Programa Saúde da Família, Municipal Department of
Health, Salvador, June 28, 2004.
Paixão, Marcia. Coordinator, Municipal Renda Mínima Program, São Paulo, October 9,
Pesaro, Antonio Floriano. Former Coordinator of Bolsa Escola Federal, Ministry of
Education. São Paulo, April 16, 2004.
Ribeiro, Teresa. Staff member of the Liderança do PT, Municipal City Council, São
Paulo, September 12, 2003.
Rocha, Sonia. Former IPEA Researcher, Rio de Janeiro, October 21, 2003.
Rodrigues, Almira Correia de Caldas. Professor of Sociology at the National University
of Brasília and affiliate of Centro Feminista de Estudos e Assessoria (CFEMEA).
Brasília, December 4, 2003.
Romero, Luis Carlos Pelizari. Senate Health Researcher, Brazilian Senate, Brasília, May
3, 2004.
Santos, Rosa Maria Barros dos. Coordinator, QUALIS-São Paulo, São Paulo, October 14,
Santos, Fausto Perreira. Former Adjunct Secretary of Health in Belo Horizonte, Brasília,
January 26, 2004.
Suassuna, Afra. Director of Basic Healthcare, Ministry of Health, Brasília, December 17,
Schneider, Alessandra. Administrator in Health Program, UNESCO-Brasília. Brasília,
December 16, 2003.
Silva, Edimar Gomes da. Chief of Staff, President’s Cabinet of the Municipal City
Council, São Paulo, October 3, 2003.
Silva, Joanna. Coordinator, Programa Sáude da Família, Municipal Department of
Health, São Paulo, October 15, 2003.
Silva, Maria de Salete. Former Secretary of Administration & Secretary of Education in
the Lídice da Mata administration, Salvador, June 1, 2004.
Silveira, Lídia Tobias. Administrator, Municipal Department of Health, São Paulo,
October 6, 2003.
Souza, Paulo Renato. Former Minister of Education, São Paulo, April 16, 2004.
Sousa, Valdomiro. Administrator, Federal Bolsa Escola Program. Ministry of Education
and Culture, Brasília, November 24, 2003.
Oliveira, Leandro Valquer JL de. President, SINDSEP-SP, São Paulo, November 10,
Oswaldo, José. Coordinator of the Belo Horizonte Municipal Health Council. Belo
Horizonte, March 12, 2004.
Passoni, Armelindo. Administrator, Department of Workforce and Social Development,
São Paulo, October 2, 2003.
Pacheco, Marisa. Former Coordinator of Bolsa Escola. Department of Education of the
Federal District. Brasília, January 27, 2004.
Pena, Moacir Ricoy. Administrator of Bolsa Escola, Municipal Department of Education.
Belo Horizonte,March 9, 2004.
Pochman, Márcio. Secretary, Municipal Department of Workforce and Social
Development, São Paulo, November 4, 2003.
Paulics, Veronika. Staff member, PÓLIS Institute, São Paulo, October 8, 2003.
Queiróz Jorge da Silva, Iêda Zilmara de. Staff member of Programa Saúde da Família,
Municipal Department of Health, Salvador, June 28, 2004.
Santos, Fausto Perreira. Former Adjunct Secretary, Municipal Department of Health of
Belo Horizonte. Brasília, January 26, 2004.
Santos Filho, Serafim Barbosa. Administrator, PROESF, Ministry of Health, Brasília,
November 25, 2003.
Silva, Célio. President, Missão Criança. Brasília, November 25, 2003.
Soares, Ana Maria da Silva. Staff member of Rede Feminista & Member of the
Municipal Health Council, Belo Horizonte, March 10, 2004.
Sousa, Maria Fátima de. Chief of Staff at Ministry of Health, Brasília, December 1, 2003.
Souza, Maria Aladilce de. President of Sindisaúde-Salvador, Salvador, June 16, 2004.
Suplicy, Eduardo. Senator, Brazilian Senate. São Paulo, October 31, 2003.
Turci, Maria. Advisor to the Department of Basic Health, Municipal Department of
Health, Belo Horizonte, February 16, 2004.
Vasconcelos, Sonia. Administrator, Bolsa Escola program, Municipal Department of
Education, Belo Horizonte, March 4, 2004.
Vaz, José Carlos. Staff member, PÒLIS Institute. São Paulo, October 8, 2003.
Woo, William. City Counselor and leader of the PSDB, São Paulo, November 12, 2003.
Focus Groups
Citizen Representatives to the Belo Horizonte Health Council. Belo Horizonte, February
16, 2004.
Citizen Representatives to the Salvador Health Council & Affiliates with Pastroral da
Saúde. Salvador, June 29, 2004.
Table C.1 PSF coverage of State’s population over Time (%)
Rio Grande do Norte
Mato Grosso do Sul
Mato Grosso
Distrito Federal
Espírito Santo
Minas Gerais
Rio de Janeiro
São Paulo
Santa Catarina
Rio Grande do Sul
Provided by the Ministry of Health, Departamento de Atenção Básica.
Sources: IBGE – Population Estimates and CAPSI - Sistema de Captação de Dados para Pagamento.
Table C.2 PSF coverage of Region’s population over Time (%)
17.87 26.35 31.87 35.69
Provided by the Ministry of Health, Departamento de Atenção Básica.
Sources: IBGE – Population Estimates and CAPSI - Sistema de Captação de Dados para Pagamento.
Figure C.1: National Cumulative Adoption of Programa Saúde da Família
Number of Cities
Source: Ministry of Health, Departamento de Atenção Básica.
Phone Interview Protocol:
Research Assistants introduced themselves and the research study. After identifying the
municipal official responsible for socio-educative programs for the city, researchers
asked the following questions:
A. Does your city administer a municipal conditional cash-grant for educational
purposes? Some cities call these programs, Bolsa Escola or Renda Minima.
If yes:
1) What is the program called?
2) When was it instituted?
3) If respondent cannot recall the year it was enacted, inquire about any legislation
or decrees that preceded the program’s start.
4) While department is responsible for the program?
5) Does the program include conditionality, e.g. regular school attendance?
6) Please describe the main features of your city’s program.
If no:
Was there such a program in the past?
B. Is there anything else you would like to share about your city’s socio-educational
C. Thank you for your time in answering these questions.
Table E.1: Structure of the Brazilian Education System
Specific Title
Early Childhood
4 years
3 years
1st grade
2nd grade
3rd grade
4th grade
Primary Education
Education (compulsory)
5 grade
6th grade
8 grade
7 grade
Education Postgraduate
Source: (Brazil 2004)
& Federal
States, and
States &
Perhaps no other policy in the last ten years has captured the attention of more
Brazilian policymakers than Bolsa Escola. The origins of the program are with municipal
governments, who were the first to design and administer the policy.
But after
conditional cash-grants for education began receiving national and international attention,
the federal government took small steps to institute its own program. A simple idea
designed by and for local governments caught-on among policymakers in Brazil’s
national government, who decided to replicate the program at the national level. This
epilogue, which focuses on federal initiatives, discusses the evolution of socioeducational social policy policies. Since the federal government’s earliest forays in this
area, Bolsa Escola has undergone profound transformation. The policy has changed
substantially, in terms of its goals, size and political significance. In the process, Bolsa
Escola Federal also experienced a name-change: Bolsa Família.
The first federal attempt to support municipal Bolsa Escola efforts was both
limited and short-lived.
In late 1997, President Fernando Henrique Cardoso’s
administration introduced the Programa Renda Mínima Vinculada à Educaçao, to provide
matching grants to municipalities that instituted their own Bolsa Escola program. Cities
with per capita incomes below their states’ averages were eligible for funds. These
policy design issues alone would have complicated the program’s expansion, but there
were other reasons the program got off to a bad start. Most notably, the federal Renda
Mínima program had not instituted controls for oversight of funds and municipal officials
were accused of corrupt practices by local media. The opportunities for mismanagement
of funds were widespread; federal matching dollars were directed to municipal
governments and many had not established a proper registry of beneficiaries or regular
payment method through a bank. In evaluating the policy, economist Sonia Rocha
quickly identified the program as a policy failure. One city simply disregarded the
programmatic objectives and distributed funds as political patronage (Interview Rocha
2003). Another problem was that each city established its own criteria for eligibility
making it became impossible to track beneficiaries, conduct program evaluations, and
allow for a transparent program with community oversight (Interview Galeano 2004).
From the point of view of municipalities, the federal government failed to transfer funds
on a regular basis and families were left without benefits for months; when funds finally
arrived, one city had officials hand out cash to beneficiaries in the middle of town square
(Interview Rocha 2003). Given all these problems, the first national experiment to
support municipal adoption of Renda Mínima died a quick and quiet death19.
In 2001, just prior to the the presidential campaign season, the Cardoso
government announced a new federal school grant program: Bolsa Escola Federal.
Administratively, the program represented an extension of the law that created the
Programa de Renda Mínima Vinculada à Educação (1997) but it differed considerably
from the previous program in both design and political visibility. First, Bolsa Escola
19 Rocha noted that at the time, administrators in the federal government downplayed problems with the
program. But the administrative irregularities were so problematic that the program ended about a year
after its inception (Interview 2003).
Federal sought to bypass municipalities altogether. Participating municipalities were no
longer required to provide matching grants to families; family grants would come
exclusively from the federal government. Nor would local government have the
responsibility for disbursing funds. Rather, the federal government would make direct
deposits into beneficiaries’ bank accounts. In this iteration, the program targeted families
whose per capita incomes were below half a minimum wage; the program covered
children from 6 to 15 years of age and provided each child with 15 reais (up to 90 reais) a
month. As before, the program operated as conditional cash-transfer so municipalities
would undertake registering eligible families, based on federal criteria, and monitoring
school attendance.
Unlike the previous program, Bolsa Escola Federal sought to achieve rapid
adoption throughout the country. MEC hired 120 staff to ensure nearly all cities would
adopt the program within a year (Interview Pesaro 2004).
The staff exceeded
expectations and within the first six months, over 4000 municipalities had registered
beneficiaries (Interview V. Sousa 2003, Interview Souza 2004).
Although Minister
Souza attributed the decision to implement a federal Bolsa Escola program to President
Fernando Henrique Cardoso and Vilmar Faria, his senior social policy advisor (Interview
Souza 2004), the policy’s timing clearly coincided with his own presidential campaign
drive to win the PSDB nomination. The political implications of a federal cash-grant
program on the eve of Presidential elections were quite clear for political opponents.
Although the name of the program had been largely associated with the Workers’ Party,
PSDB officials made efforts to claim the program as their own by citing early
experiences in the city of Campinas and emphasizing that the federal program was the
largest with the greatest coverage.20
Bolsa Escola Federal was not the only federal cash-transfer program under the
Cardoso administration. Many federal agencies had developed cash-assistance programs,
subsidies or voucher programs to assist the poor. For instance, the Ministry of Health
had a program, Programa Bolsa Alimentação (Nutritional Grant Program) for pregnant
women, nursing mothers, and children under six years of age who at risk of malnutrition.
The Ministry of Mines and Energy also had a program, Auxilio Gás21 (Gas Stipend) for
poor families to purchase household gas cylinders. In the last year of the Cardoso
administration, senior technocrats in the office of the President pushed for better policy
coordination between social assistance programs. At that point, each ministry had its
own rolls of beneficiaries and criteria for determining eligibility. In Cardoso’s last year
of office, the federal government began steps to move toward a single registry to track
beneficiaries. A unified registry would also facilitate governmental administration with
better oversight of its social services and avoid duplication of benefits.
When Luis Inácio Lula da Silva (PT) won the presidential election in 2002, he
surprised many political observers by introducing eradication of hunger as a major
priority. Lula promised to focus his attention on the issues of hunger, malnutrition, and
extreme poverty and announced a new program that would mark the early years of his
20 On May 23, 2002 the PSDB hosted a meeting “Brasil 2010: Desenvolvimento e Inclusão Social: O
Brasil no Rumo Certo” in Brasilia. The agenda for that meeting included a session led by Floriano Pesaro
entitled, “Bolsa-Escola: Um Programa Tucano de Inclusão Social” (The Bolsa Escola, a Toucan program
for social inclusion). The Toucan is the party’s symbol.
21 Also known as Vale-Gás, this program provided poor families with R$7.50 subsidy per a month.
social policy agenda: Programa Fome Zero (Zero Hunger Program). That he had risen
out of childhood poverty and a humble upbringing, made his quest to alleviate hunger all
the more captivating for domestic and international audiences. The goals and scope of
the program were ambitious from the start.
The policy was also administratively
complex given that there were numerous local and federal programs already in place to
address hunger and malnutrition. Despite considerable fanfare the Fome Zero program,
which includes both food stamps and in-kind disbursements to needy families, got off to
a slow start and has been plagued by problems, ranging from administrative
inefficiencies, local corruption and bureaucratic inexperience (Hunter & Power
At the same time the Lula administration introduced its signature policy initiative,
Fome Zero, officials also moved forward with plans to integrate the preexisting federal
cash grant programs into a single registry. Under the Lula administration, a minimumincome program providing cash-grants for the poor would take the name Bolsa Família22
(Family Grant). Similar to earlier efforts at the end of the Cardoso administration, the
Bolsa Família sought to provide beneficiaries with a single stipend unifying disparate
programs. As was the case with Fome Zero, Bolsa Família also ran into administrative
trouble as different ministries vied for control over the program. Cristovam Buarque,
who was Minister of Education (2003-2004), firmly defended the need for the Bolsa
Escola Federal and argued that combining the program with other cash-grants would
The Bolsa Família legislation went into effect on January 9, 2004; Law nº 10.836.
dilute its educational effects.
Reports that a main feature of the program – the
requirement for regular school attendance – was being overlooked by the federal
government generated even more criticism. It would take several years for the Lula
government to address the administrative hurdles associated with Bolsa Família and
implement it on a grand scale.
Ironically, even though the Lula administration introduced Fome Zero with
considerable zeal and sold it as his signature policy, Bolsa Família rose in importance to
represent his social policy focus. The Bolsa Família unified four previously distinct
programmatic objectives (educational stipends to boost school attendance, maternal
nutrition, food supplements, and a household gas subsidy) into a single conditional cash
transfer policy. In 2006, the program benefited 44 million of Brazil’s poorest citizens (11
million families). This impressive coverage has represented a third of the federal
government’s spending on social assistance for the poor (Hall 2006:689).
That the
program would grow so quickly and represent such a large share of public assistance
spending is somewhat surprising given that the policy has yet to undergo large scale
While President Lula and his senior staff’s motivations for adopting Bolsa
Família fall outside the confines of this dissertation and are therefore unknown, it is clear
the social policy had significant electoral consequences in the 2006 presidential election.
President Lula entered the reelection campaign under serious fire. Key members of his
party and senior staff had been implicated in a corruption scandal in June 200523. Shortly
thereafter, other allegations of serious malfeasance became widespread news headlines.
The series of corruption scandals threatened Lula’s reelection prospects and challenged
the Workers’ Party’s historic claims of “good governance” practices.
What explains
Lula’s second-round victory in the face of these political challenges? Hunter and Power
(2007) argue that a key component of Lula’s electoral victory in 2006 is due to his ability
to draw on voters from the lowest income and education brackets in the North and
Northeast. Given Bolsa Família’s emphasis on targeting resources, it is not surprising
that these regions, with high levels of poverty and low human development indicators,
would benefit disproportionately from the conditional cash-grant program.24
political payoff for Lula was clear, as states with the greatest Bolsa Família coverage
voted overwhelmingly for Lula in the first round of elections.
23 The “Mensalao” scandal detailed a monthly kick-back scheme directed at Members of congress in
exchange for their votes in favor of President Lula’s legislative agenda. Those implicated in the scandal
included: Senior Workers Party officials, close Presidential advisors, and congressional representatives.
24 For instance, the states Marahnão, Piaui, Ceará, Rio Grande do Norte, Pernanbuco, Alagoas, Sergipe,
and Bahia all had widespread coverage, with 42 to 50 percent of the population receiving benefits.
Abers, Rebecca. 2000. Inventing Local Democracy. Boulder: Lynne Rienner.
Abrucio, Fernando Luiz. 2002. Os Barões da Federação: os governadores e a
redemocratização brasileira (2nd. ed.). São Paulo: Editora Hucitec.
Affonso, Rui de Britto Álvares, and Pedro Luiz Barros Silva (eds.). 1996.
Descentralização e Políticas Sociais. São Paulo: FUNDAP.
Aguiar, Marcelo, and Carlos Henrique Araújo. 2002. Bolsa-Escola: Education to
Confront Poverty. Brasília: UNESCO.
Allison, Paul D. 1984. Event History Analysis: Regression for Longitudinal Event Data
(Vol. 07-046). Beverly Hills: Sage Publications.
Almeida, Carlos, Tatiana Baptista, Claudia Travassos, and Silvia Porto. 2000. Health
Sector Reform in Brazil. Inter-American Development Bank. Retrieved April 2,
2007, from
Alvarez, Sonia E. 1990. Engendering democracy in Brazil: Women's Movements in
Transition Politics. Princeton, N.J.: Princeton University Press.
Alvarez, Sonia E. 1994. The (Trans)formation of Feminism(s) and Gender Politics in
Democratizing Brazil. In Jane S. Jaquette ed., The Women's Movement in Latin
America. Boulder: Westview Press.
Alvarez, Sonia E. 1999. Advocating Feminism: The Latin American Feminist NGO
'Boom'. International Feminist Journal of Politics. 1(2), 181-209.
Alvarez, Sonia E., Evelina Dagnino, and Arturo Escobar, Arturo eds. 1998. Cultures of
Politics/Politics of Cultures: Re-visioning Latin American Social Movements.
Boulder: Westview Press.
Alvarez, Sonia E., Evelina Dagnino, and Arturo Escobar, Arturo. 1998. Introduction: The
Cultural and the Political in Latin American Social Movements. In Alvarez, Sonia
E., Evelina Dagnino and Arturo Escobar, Arturo eds. Cultures of Politics/Politics
of Cultures: Re-Visitioning Latin American Social Movements. Boulder:
Westview. pp. 1-29.
Ames, Barry. 2001. The Deadlock of Democracy in Brazil. Ann Arbor: University of
Michigan Press.
Apter, David Ernest. 1964. Ideology and Discontent. London: Free Press of Glencoe.
Araújo, Carlos Henrique, and Nair Heloísa Bicalho de Souza eds. 1998. Programas de
Renda Mínima no Brasil: Impactos e Potencialidades (Vol. 30). São Paulo:
Araújo, Carlos Henrique, and Eliezer Lopes. 1996. Falso Dilema. Jornal de Brasília,
May 31, 2006. p. 2.
Arretche, Marta. 2000. Estado Federativo e Políticas Sociais: Determinantes da
Decentralização. Rio de Janeiro: Editora Reven.
Arretche, Marta. 2004. Federalismo e Políticas Sociais no Brasil: Problemas de
Coordenação e Autonomia. São Paulo em Perspectiva, 18(2), 17-26.
Arretche, Marta, and Eduardo.Marques. 2002. Municipalização da Saúde no Brasil:
Diferenças Regionais, Poder do Voto e Estratégias de Governo. Ciência & Saúde
Coletiva, 7(3), 455-479.
Assunes, Márcia. 1997. Corte na Bolsa-Escola irrita mães. Jornal de Brasília, April 17,
1997. p. 20.
Baiocchi, Gianpaolo. 2003. Emergent Public Spheres: Talking Politics in Participatory
Governance. .American Sociological Review, 68(1), 52-47.
Baiocchi, Gianpaolo ed. 2003. Radicals in Power: the Workers' Party (PT) and
Experiments in Urban Democracy. New York: Zed Books/ Palgrave.
Balla, Steven J. 2001. Interstate Professional Associations and the Diffusion of Policy
Innovations. American Politics Research, 29(3), 221-245.
Bava, Silvio Caccia. n.d. Bolsa-Escola (school bursary program): a Public Policy on
Minimum Income and Education. São Paulo: Polis/Instituto de Estudos,
Formação e Assessoria em Políticas Sociais.
Bava, Silvio Caccia , Vera da Silva Telles, Selva Ribas Bejarano, Carlos Henrique
Araújo, and Nair Heloísa Bicalho de Sousa. n.d. Programa de Renda Mínima no
Brasil: Impactos e Potencialidades. International Development Research Centre
(Canada). Retrieved August 16, 2005, 2005, from
Bava, Silvio Caccia, Vera da Silva Telles, Selva Ribas Bejarano, Carlos Henrique
Araújo, and Nair Heloísa Bicalho de Sousa. 1999. Bolsa-Escola: Uma política
pública de renda mínima e educacional. Paper presented at the Seminario
Internacional: Reformas a la Política Social en América Latina: Resultados y
Perspectivas. Washington, D.C., May 11-12, 1999.
Bava, Silvio Caccia, Vera da Silva Telles, Selva Ribas Bejarano, Carlos Henrique
Araújo, and Nair Heloísa Bicalho de Sousa. 1998. Programas de Renda Mínima
no Brasil. Pólis: Estudos, Formação e Assessoria em Políticas Sociais, 30, 1-130.
Beck, Nathaniel, Jonathan N Katz, and Richard Tucker. 1998. Taking Time Seriously:
Time-Series-Cross-Section Analysis with a Binary Dependent Variable. American
Journal of Political Science, 42(4), 1260-1288.
Bennett, Colin J. 1991. What Is Policy Convergence and What Causes It? British
Journal of Political Science, 21(2), 215-233.
Berman, Sheri. 2001. Ideas, Norms, and Culture in Political Analysis. Comparative
Politics, 33(2), 231-250.
Berry, Frances Stokes, and William D.Berry. 1999. Innovation and Diffusion Models in
Policy Research. In Paul. A. Sabatier Ed., Theories of the Policy Process.
Boulder: Westview.
Berry, Frances Stokes, and William D.Berry. 1990. State Lottery Adoptions as Policy
Innovations: An Event History Analysis. American Political Science Review,
84(2), 395-415.
Berry, Frances Stokes, and William D.Berry. 1992. Tax Innovation in the States:
Capitalizing on Political Opportunity. American Journal of Political Science,
36(3), 715-742.
Birdsall, Nancy, Barbara Bruns, and Richard H Sabot. 1996. Education in Brazil: Playing
a Band Hand Badly. In Nancy Birdsall & Richard H. Sabot eds. Opportunity
Foregone: Education in Brazil. Washington, D.C.: Inter-American Development
Bank. pp. 7-47.
Birdsall, Nancy, and Richard H Sabot eds. 1996. Opportunity Foregone: Education in
Brazil. Washington, D.C.: Inter-American Development Bank.
Bolsa da Sobrevivência. 1997, November 9. Jornal de Brasília, p. 6.
Bolsa-Escola Um Exemplo de Brasília. 1996. Ciência Hoje, 21 (November/December),
pp. 56-62.
Box-Steffensmeier, Janet M., and Bradford S. Jones. 1997. Time is of the Essence: Event
History Models in Political Science. American Journal of Political Science, 41(4),
Box-Steffensmeier, Janet M., and Bradford S. Jones. 2004. Event History Modeling: a
Guide for Social Scientists. New York: Cambridge University Press.
Box-Steffensmeier, Janet M., and Bradford S. Jones. 2004. Timing and Political Change:
Event History Modeling in Political Science. Ann Arbor: University of Michigan
Brady, Henry E. 2004. Doing Good and Doing Better: How Far Does the Quantitative
Template Get Us? In Henry E. Brady and David Collier eds. Rethinking Social
Inquiry: Diverse Tools, Shared Standards. New York: Rowman & Littlefield. pp.
Brazil. Câmara dos Deputados Comissião de Educação, Cultura e Desporto. 2002. Renda
Mínima Vinculada à Educação: Seminiario Nacional. Brasília: Centro de
Documentação e Informação (CEDI)/Coordenação de Publicações (CODEP).
Brazil, Ministry of Education. 2004. The Development of Education - Quality Education
for all Young People: Challenges, Trends and Priorities. Paper presented at the
International Conference on Education, 47th Session, Geneva, Switzerland.
September 8-11, 2004.
Brazil, Ministry of Education and Sports. 1996. Development of Education in Brazil.
Paper presented at the International Conference on Education, Geneza,
Switzerland. September 30-October 5, 1996.
Brazil, Ministry of Education. Secretaria do Programa Nacional do Bolsa Escola,
availbable: http://www.mec.gov.br/home/bolsaesc/default.shtm
Brazil, Ministry of Education. 2002. Relatório de Atividades 2001. Available:
Brazil, Ministry of Health. 2002. Desafios e Conquistas do PSF. Revista Brasileira de
Saúde da Família 2:5 (May): 7-24.
Brazil, Ministry of Health. 1999. I Seminário de Experiências Internacionais em Saúde
da Família: Relatório Final. Brasília: MS.
Brazil, Ministry of Health. 1997. Saúde da Família: Uma Estratégia para a Reorientação
do Modelo Assistencial. Brasília: MS.
Brazil, Ministry of Health. (MS. DAB Departamento de Atenção Básica). 2002. Relatório
de Gestão, 1998-2002. Brasília: MS. DAB.
Brazil, FUNASA (Fundação Nacional de Saúde). 1994. Programa de Saúde da Família:
Saúde dentro da Casa. Brasília: MS. FUNASA.
Brazil, Ministry of Health (MS. SAS Secretaria de Assistência à Saúde). 2002.
Experiências Inovadoras no SUS. Brasília: MS. SAS.
Brazil, Ministry of Health. 2002. Avaliação da Implementação do Programa Saúde da
Família em Dez Grandes Centros Urbanos: Sintese dos Principais Resultados.
Brasília: Ministério de Saúde.
Brazil, Ministry of Health. 2003. A Produção Sobre Saúde da Família: Trabalhos
Apresentados no VII Congresso Brasilieiro de Saúde Coletiva ABRASCO 2003.
Brasília: Ministério de Saúde.
Brazil, Ministry of Justice. 2007. Organizações da Sociedade Civil de Interesse Público OSCIP Lista de Entidades. Available at: http://www.mj.gov.br/snj/oscip/
Brinks, Daniel, and Michael Coppedge. 2006. Diffusion is No Illusion: Neighbor
Emulation in the Third Wave of Democracy. Comparative Political Studies,
39(5), 463 - 489.
Brooks, Sarah M. 2002. Social Protection and Economic Integration. Comparative
Political Studies, 35(5), 491-523.
Buarque, Cristovam, Vanessa Castro, and Marcelo Aguiar. 2002. Um pouco da história
do Bolsa Escola. Serviço Social e Sociedade, 66, 126-144.
Burt, Ronald S. Burt. 1987. Social Contagion and Innovation: Cohesion versus Structural
Equivalence. American Journal of Sociology, 92(6), 1287-1335.
Caldas, Eduardo de Lima, and Estêvão Passos Eller. 2005. Programa Paidéia de Saúde da
Família: Campinas (SP). In F. M. d. Oliveira, H. B. Barboza & M. A. C. Teixeira
eds., 20 Experiências de Gestão Pública e Cidadania - Ciclo de Premiação 2003.
São Paulo: Programa Gestão Pública e Cidadania, Fundação Getúlio Vargas-São
Paulo, pp. 287-297.
Camargo, José Márcio. 1993. Os miseraveis. Folha de São Paulo. March 23, 1993.
Capistrano Filho, David. 1999. O Programa de Saúde da Família em São Paulo. Estudos
Avançados, 13(35), 89-100.
Carmines, Edward G., and James A. Stimson. 1993. On the Evolution of Political Issues.
In William H. Riker Ed., Agenda Formation. Ann Arbor: University of Michigan
Press, pp. 151-168.
Castro, Claudio de Moura, and Philip Musgrove. 1998. On the Nonexistence of 'The
Social Sector' or Why Education and Health are More Different than Alike.
Washington, D.C.: Inter-American Development Bank/Education Unit
Sustainable Development Department.
CENPEC. 2001. Fontes de Educação: Guia para Jornalistas. Brasilia, DF: Fórum Mídia
& Educação.
Cesar, Juraci A., Marcelo A.Cavaleti, Ricardo S. Holthausen, and Luis Gustavo S. de
Lima. 2002. Mudanças em Indicadores de Saúde Infantil em um Município com
Agentes Comunitários: o caso de Itapirapuã Paulista, Vale do Ribeira, São Paulo,
Brasil. Caderno de Saúde Pública, 18(6), 1647-1654.
Chiaravalloti Neto, Francisco, Angelita A. C. Barbosa, Marisa B. Cesarino, et al. 2006.
Controle do Dengue em uma Area Urbana do Brasil: avaliação do impacto do
Programa Saúde da Família com relação ao programa tradicional de controle.
Caderno de Saúde Pública, 22(5), 987-997.
Cohn, Amélia. 1989. Caminhos da Reforma Sanitária. Lua Nova, 19, 123-140.
Cohn, Amélia, and Paulo E.Elias eds. 1999. O Público e o Privado na Saúde: o PAS em
São Paulo. São Paulo: Editora Cortez.
Collier, David, and Richard E. Messick. 1975. Prerequisites versus Diffusion: Testing
Alternative Explanations of Social Security Adoption. American Political Science
Review, 69(4), 1299-1315.
Coppedge, Michael. 1996. A Classification of Latin American Political Parties.
Unpublished manuscript. University of Notre Dame.
Coppedge, Michael. 1999. Thickening Thin Concepts and Theories: Combining Large N
and Small in Comparative Politics. Comparative Politics, 31(4), 465-476.
Corrales, Javier. 1999. The Politics of Education Reform: Bolstering the Supply and
Demand; Overcoming Institutional Blocks. Washington, D.C.: World Bank.
Costa, Cláudia Soares Costa, and Gabriel Rangel Visconti. 2001. Terceiro Setor e
Desenvolvimento Social. Brazil: BNDES. Available at:
Crianças darão nota à Bolsa-Escola. 1997. Correio Braziliense, September 24, 2007 .p.
Derthick, Martha. 1970. The Influence of Federal Grants: Public Assistance in
Massachusetts. Cambridge, Mass: Harvard University Press.
DiMaggio, Paul J., and Powell, Walter W. 1983. The Iron Cage Revisited: Institutional
Isomorphism and Collective Rationality in Organizational Fields. American
Sociological Review, 48(2), 147-160.
Dimenstein, Gilberto. 1997. Bolsa Mágica. Folha de São Paulo, August 6, 1997. p. A19.
Downs, Anthony. 1957. An Economic Theory of Democracy. New York: Harper.
Draibe, Sonia. n.d. Programa de Agentes Comunitários de Saúde. Retrieved March 4,
2007, available at:
Draibe, Sonia, Marta Arretche, Ana Maria Fonseca, et. al. 1994. Estratégias para
combater a probreza no Brasil: programas, institutições e recursos. Working Paper
190. Washington, D.C.: Inter-American Development Bank.
Draibe, Sonia M. 1985. Rumos e metamorfoses: um estudo sobre a constituição do
Estado e as alternativas da industrializacao no Brasil, 1930-1960. Rio de Janeiro,
RJ: Paz e Terra.
Draibe, Sonia. 2004. Federal Leverage in a Decentralized System: Education Reform in
Brazil. In Robert R Kaufman and Joan M.Nelson, eds. Crucial Needs Weak
Incentives: Social Sector Reform, Democratization, and Globalization in Latin
America. Baltimore: Johns Hopkins University Press. pp. 375-406.
Dresang, Lee T., Laurie Brebrick, Danielle Murray, et. al. 2005. Family Medicine in
Cuba: Community-Oriented Primary Care and Complementary and Alternative
Medicine. Journal of the American Board of Family Practice, 18(4), 297-303.
Dugger, Celia W. 2004, To Help Poor Be Pupils, Not Wage Earners, Brazil Pays Parents.
New York Times, January 3, 2004. pp. A1and A6.
Elkins, Zachary, and Beth Simmons. 2005. On Waves, Clusters, and Diffusion: A
Conceptual Framework. Annals of the American Academy of Political and Social
Science, 598(1), 33-51.
Encarnación, Omar G. 2003. The Myth of Civil Society: Social Capital and Democratic
Consolidation in Spain and Brazil. New York: Palgrave MacMillan.
Escobar, Arturo, and Sonia E. Alvarez. 1992. The Making of social movements in Latin
America: identity, strategy, and democracy. Boulder: Westview Press.
Escorel, Sarah. 1999. Reviravolta na Saúde: Origem e Articulacao do Movimento
Sanitário. Rio de Janeiro: Editora Fiocruz.
Farah, Marta Ferreira Santos, and Hélio Batista Barboza eds. 2000. Novas Experiências
de Gestão Pública e Cidadania. Rio de Janeiro:: Editora FGV.
Fausto, Boris. 1999. A Concise History of Brazil. New York: Cambridge University
Fernandes, Antônio Sérgio Araújo. 2004. Gestão Municipal e Participação Social no
Brasil: A trajetória de Recífe e Salvador (1986-2000). São Paulo: AnnaBlume.
Fernandes, Diana. 2001. Dois em Um. O Globo, January 5, 2001. p. 2.
Finnemore, Martha. 1993. International Organizations as Teachers of Norms: The United
Nations Educational, Scientific, and Cutural Organization and Science Policy.
International Organization, 47(4), 565-597.
Finnemore, Martha. 1996. National Interests in International Society. Ithaca, N.Y.:
Cornell University Press.
Finnemore, Martha and Kathryn Sikkink. 1998. International Norm Dynamics and
Political Change. International Organization, 52(4), 887-917.
Fonseca, Ana Maria Medeiros da. 2001. Família e Política da Renda Mínima. São Paulo:
Cortez Editora.
Fortes, Paulo Antonio de Carvalho, and Simone Ribeiro Spinetti. 2004. O agente
comunitário de saúde e a privacidade das informações dos usuários. Caderno de
Saúde Pública, 20(5), 1328-1333.
Friedkin, Noah E. 1993. Structural Bases of Interpersonal Influence in Groups: A
Longitudinal Case Study. American Sociological Review, 58(6), 861-872.
Geddes, Barbara. 1990. How the Cases you Choose Affect the Answers You Get.
Political Analysis, 2, 131-152.
Geddes, Barbara. 2003. Paradigms and Sand Castles: Theory Building and Research
Design in Comparative Politics. Ann Arbor: University of Michigan Press.
Geraldes, Déborah. 1999. ONG Leva Bolsa-Escola a Paracatu. Correio Braziliense,
September 5, 1999, p. 8.
Gerring, John. 1997. Ideology: A Definitional Analysis. Political Research Quarterly,
50(4), 957-994.
Goldsteen, Raymond L., Julio Cesar Pereira, and Karen Goldsteen. 1990. The engine or
the caboose: health policy in developing countries. Revista de Saúde Pública,
24(6), 523-527.
Goulart, Flavio A. de Andrade. 1995. Municipalização: Veredas (Caminhos do
Movimento Municipalista de Saúde no Brasil). Unpublished manuscript, Brasília.
Gouveia, Roberto and José João Palma. 1999. SUS: na contamão do neoliberalism e da
exclusão social. Estudos Avançados, 13(35), 139-146.
Graham, Carol. 1995. The Politics of Safety Nets. In Larry Diamond and Mark F.
Plattner eds. Economic Reform and Democracy. Baltimore: Johns Hopkins
University Press.
Granovetter, Mark. 1973. The Strength of Weak Ties. American Journal of Sociology,
78(6), 1360-1380.
Granovetter, Mark. 1983. The Strength of Weak Ties: A Network Theory Revisited.
Sociological Theory, 1, 201-233.
Gray, Virginia. 1973. Innovation in the States: A Diffusion Study. American Political
Science Review, 67(4), 1174-1185.
Grindle, Merilee Serrill. 2000. Audacious Reforms: Institutional Invention and
Democracy in Latin America. Baltimore: Johns Hopkins University Press.
Grindle, Merilee Serrill. 2004. Despite the Odds: the Contentious Politics of Education
Reform. Princeton, NJ: Princeton University Press.
Grindle, Merilee Serrill, and John W. Thomas. 1991. Public Choices and Policy Change:
the political economy of reform in developing countries. Baltimore: Johns
Hopkins University Press.
Grossback, Lawrence J., Sean Nicholson-Crotty, and David A. M Peterson. 2004.
Ideology and Learning in Policy Diffusion. American Politics Research, 32(5),
Hall, Anthony. 2006. From Fome Zero to Bolsa Família: Social Policies and Poverty
Alleviation under Lula. Journal of Latin American Studies, 38(4), 689-709.
Hassim, Shireen. 2005. Turning Gender Rights into Entitlements: Women and Welfare
Provision in Postapartheid South Africa. Social Research, 72(3), 621-646.
Heath, Chip, and Dan Heath. 2007. Made to Stick: Why Some Ideas Survive and Others
Die. New York: Random House.
Hechter, Michael, and Karl-Dieter Opp eds. 2001. Social Norms. New York: Russell
Sage Foundation.
Holzmann, Robert, Mitchell Orenstein, and Michal Rutkowski eds. 2003. Pension
Reform in Europe: Process and Progress. Washington, D.C.: World Bank.
Horne, Christine. 2001. Sociological Perspectives on the Emergence of Social Norms. In
Michael Hechter and Karl-Dieter Opp eds., Social Norms. New York: Russell
Sage Foundation, pp. 3-34.
Hunter, Wendy. 2003. Brazil's New Direction. Journal of Democracy, 14(2), 151-162.
Hunter, Wendy, and Timothy J Power. 2005. Lula's Brazil at Midterm. Journal of
Democracy, 16(3), 127-139.
Hunter, Wendy, and Timothy J Power. 2007. Rewarding Lula: Executive Power, Social
Policy, and the Brazilian Elections of 2006. Latin American Politics & Society,
49(1), 1-30.
Huntington, Samuel P. 1957. Conservatism as an Ideology. American Political Science
Review, 51(2), 454-473.
Husseini, Samya. 1996. Bolsa Escola Recebe Prêmio do UNICEF. Correio Braziliense,
November 12, 1996, p. 14.
Ibañez Ruiz, Antonio. 1996. Querem Acabar com a Bolsa-Escola. Jornal do Brasil,
October 22, 1996, p. 6.
Ibañez Ruiz, Antonio. 1997. A Coerência da Bolsa Escola. Correio Braziliense, May 26,
1997, p. 2.
Jacobo Waiselfisz, Julio, Miriam Abramovay, and Carla Andrade. 1998. Bolsa Escola
Melhoria Educacional e Redução da Pobreza. Brasília: UNESCO.
Jaquette, Jane S. 1994. The Women's Movement in Latin America: Participation and
Democracy (2nd Ed.). Boulder: Westview Press.
Jaquette, Jane S., and Sharon L.Wolchik, 1998. Women and Democracy: Latin America
and Central and Eastern Europe. Baltimore, Md.: Johns Hopkins University
Jatene, Adib. 2000. Um Rebelde com Causa. Folha de São Paulo. October 17, 2000.
Jordan, Jason. 2006. Mothers, Wives, and Workers: Explaining Gendered Dimensions of
the Welfare State. Comparative Political Studies, 39(9), 1109-1132.
Jordana, Jacint, and David Levi-Faur. 2005. The Diffusion of Regulatory Capitalism in
Latin America: Sectoral and National Channels in the Making of a New Order.
Annals of the American Academy of Political and Social Science, 598, 102-124.
Karch, Andrew. 2007. Democratic Laboratories: Policy Diffusion among the American
States. Ann Arbor: University of Michigan Press.
Kaufman, Jason, and Orlando Patterson. 2005. Cross-National Cultural Diffusion: The
Global Spread of Cricket. American Sociological Review, 70(1), 82-110.
Kaufman, Robert R. 1999. Approaches to the Study of State Reform in Latin American
and Postsocialist Countries. Comparative Politics, 31(3), 357-375.
Kaufman, Robert R., and Joan M. Nelson eds. 2004. Crucial Needs, Weak Incentives:
Social Sector Reform, Democratization, and Globalization in Latin America.
Washington, D.C.: Woodrow Wilson Center Press.
Keck, Margaret E. 1992. The Workers' Party and Democratization in Brazil. New Haven:
Yale University Press.
Keck, Margaret E., and Kathryn Sikkink. 1998. Activists Beyond Borders: Advocacy
Networks in International Politics. Ithaca, N.Y.: Cornell University Press.
Kilduff, Martin, and Wenpin Tsai. 2003. Social Networks and Organizations. Thousand
Oaks, CA: Sage Publications.
King, Gary, Robert Keohane, and Sidney Verba. 1994. Designing Social Inquiry.
Princeton: Princeton University Press.
Kingdon, John W. 1995. Agendas, Alternatives, and Public Policies (2nd Ed.). New
York: Longman.
Knight, Kathleen. 2006. Transformations of the Concept of Ideology in the Twentieth
Century. American Political Science Review, 100(4), 619-626.
Lago, Rudolfo. 2001. Lento e Bem-Vindo Investimento. Correio Braziliense, January 4,
2006, p. 16.
Lavinas, Lena, and Maria Lígia Barbosa. (2000). Combating Poverty and Increasing
School Attendance: The Case Study of Bolsa-Escola in Recife. Paper presented at
the LCSHD Conference: Education and Poverty: Including the Excluded.
Lavinas, Lena, and Maria Ligia de Oliveira Barbosa. 2000. Combater a Pobreza
Estimulando a Freqüência Escolar: O Estudo de Case do Programa Bolsa-Escola
do Recife. DADOS - Revista de Ciências Sociais, 43(3), 447-477.
Lavinas, Lena, Maria Ligia de Oliveira Barbosa and Octávio Tourinho. 2001. Assessing
Local Minimum Income Programs in Brazil. Geneva: International Labour
Organisation/World Bank.
Levi-Faur, David. 2005. The Global Diffusion of Regulatory Capitalism. Annals of the
American Academy of Political and Social Science, 598, 12-32.
Lima, Roberto Teixeira, and Nilton Maciel Mangueira. 2001. Avaliação de Implantação
de Ações Básicas de Saúde. Revista Brasileira de Ciências de Saúde, 5(2), 149158.
Lindblom, Charles E. 1959. The 'Science' of Muddling Through. Public Administration
Review, 19(2), 79-88.
Lobato, Ana Lucia, and André Urani eds. 1998. Garantia de renda mínima: ensaios e
propostas. Brasília: Instituto de Pesquisa Econômica Aplicada.
Lowi, Theodore. 1963. Toward Functionalism in Political Science: The Case of
Innovation in Party Systems. American Political Science Review, 57(3), 570-583.
Lowi, Theodore. 1995. Distribution, Regulation, Redistribution. In Stella Theodoulou &
Mathew Cahn eds., Public Policy: The Essential Readings. Englewood Cliffs,
New Jersey: Prentice Hall.
Madrid, Raúl L. 2003. Retiring the State: the Politics of Pension Privatization in Latin
America and Beyond. Stanford: Stanford University Press.
Mahoney, James, and Goertz, Gary. 2006. A Tale of Two Cultures: Contrasting
Quantitative and Qualitative Research. Political Analysis, 14, 227-249.
Mainwaring, Scott. 1999. Rethinking Party Systems in the Third Wave of
Democratization: the Case of Brazil. Stanford: Stanford University Press.
Mainwaring, Scott, Rachel Meneguello, and Timothy J. Power. 2000. Conservative
Parties, Democracy, and Economic Reform in Contemporary Brazil. In Kevin
Middlebrook Ed., Conservative Parties, the Right, and Democracy in Latin
America. Baltimore: Johns Hopkins University Press. pp. 164-222.
Martins, Roberto Borges, and José Carlos Libânio eds. 2005. Atlas do Desenvolvimento
Humano no Brasil. Retrieved May 17, 2005. Available at:
Mazur, Amy G. 2002. Theorizing Feminist Policy. New York: Oxford University Press.
Medeiros, Marcelo. 2001. A Trajetória do Welfare State no Brasil: Paper Redistributivo
das Políticas Sociais dos Anos 1930 aos Anos 1990. Brasília: Instituto de Pequisa
Econômica Aplicada (IPEA) Texto Para Discussão, 852, 1-24.
Melamed, Clarice, and Nilson do Rosário Costa. 2003. Inovações no financiamento
federal à Atenção Básica’. Ciência & Saúde Coletiva, 8(2), 393-401.
Mello, Alessandra. 2000. Bolsa-Escola Reduz Pobreza no País. Jornal do Brasil,
December 3, 2000, p. 13.
Mello, Carlos Gentile de. 1977. Saúde e Assistência Médica no Brasil. São Paulo:
Mendonça, Ana Valéria Machado, Fernanda Morais, Keyla Antunes Kikushi Câmara, et.
al. 2004. Primeira Mostra de Saúde da Família - Camaragibe dez anos: uma
síntese das rodadas de diálogo. Divulgação Saúde em Debate, 31, 28-37.
Mesa-Lago, Carmelo. 1997. Social Welfare Reform in the Context of Economic-Political
Liberalization: Latin American cases. World Development, 25(4), 497-517.
Meseguer, Covadonga, and Fabrizio Gilardi. 2005. What is new in the study of policy
diffusion? A critical review. Unpublished Manuscript. Mexico City.
Mintrom, Michael. 1997. Policy Entrepreneurs and the Diffusion of Innovation.
American Journal of Political Science, 41(3), 738-770.
Mintrom, Michael, and Sandra Vergari. 1998. Policy Networks and Innovation Diffusion:
The Case of State Education Reforms. Journal of Politics, 60(1), 126-148.
Mishima, Silvana Martins, Eliete Maria Silva, Maria Luiza Anselmi, et. al. 1992. Agentes
comunitários de saúde: bom para o Ceará... bom para o Brasil?. Saúde em Debate,
37, 70-75.
Montero, Alfred P. 2001. Decentralizing Democracy: Spain and Brazil in Comparative
Perspective. Comparative Politics, 33(2), 146-169.
Montero, Alfred P., and David Samuels eds. 2004. Decentralization and Democracy in
Latin America. Notre Dame: University of Notre Dame Press.
Montero, Alfred P., and David Samuels. 2004. The Political Determinants of
Decentralization in Latin America: Causes and Consequences. In Alfred P.
Montero and David Samuels eds., Decentralization and Democracy in Latin
America. Notre Dame: University of Notre Dame Press. pp. 3-32.
Mooney, Christopher Z. 2001. Modeling Regional Effects on State Policy Diffusion.
Political Research Quarterly, 54(1), 103-124.
Mooney, Christopher Z., and Mei-Hsien Lee. 1995. Legislating Morality in the American
States: The Case of Pre-Roe Abortio Lee n Regulation Reform. American Journal
of Political Science, 39(3), 599-627.
Moraes, Reginaldo C. 2002. Reformas Neoliberais e Políticas Públicas: Hegemonia
Ideológia e Redefinição das Relações Estado-Sociedade. Educação e Sociedade,
23(80), 13-24.
Mossberger, Karen. 1999. State-Federal Diffusion and Policy Learning: From Enterprise
Zones to Empowerment Zones. Publius, 29(3), 31-50.
Mossberger, Karen. 2000. The Politics of Ideas and the Spread of Enterprise Zones.
Washington, D.C.: Georgetown University Press.
Mullins, Willard A. 1972. On the Concept of Ideology in Political Science. American
Political Science Review, 66(2), 498-510.
Munck, Gerardo L. 2004. Tools for Qualitative Research. In Henry E. Brady and David
Collier eds., Rethinking Social Inquiry: Diverse Tools, Shared Standards. New
York: Rowman & Littlefield Publishers. pp. 105-121.
Myers, David J., and Henry A. Dietz eds. 2002. Capital City Politics in Latin America:
Democratization and Empowerment. Boulder, Colo.: Lynne Rienner.
North, Douglass C. 1981. Structure and change in Economic History (1st Ed.). New
York: Norton.
North, Douglass C. 1990. Institutions, Institutional Change, and Economic Performance.
New York: Cambridge University Press.
Nylen, William R. 2003. Participatory Democracy vs. Elitist Democracy: Lessons from
Brazil. New York: Palgrave Macmillan.
OECD. 2004. OECD Economic Survey of Brazil: Better Targeting Government
Spending. OECD Economic Survey of Brazil. in Available at:
Oliveira, Marina. 2001. À Margem do Bolsa Escola: Pedir dá Mais Lucro que Renda
Mínima. Correio Braziliense, January 2, 2001. p. 6.
Olson, Mancur. 1965. The logic of Collective Action: Public Goods and the Theory of
Groups. Cambridge: Harvard University Press.
Orenstein, Mitchell. 2003. Mapping the Diffusion of Pension Innovation. In Robert
Holzmann, Mitchell Orenstein, and Michal Rutkowski eds. Pension Reform in
Europe: Process and Progress. Washington, D.C.: World Bank. pp. 171-194.
Passy, Florence. 2003. Social Networks Matter. But How? In Mario Diani and Doug
McAdam eds., Social Movements and Networks: Relational Approaches to
Collective Action. New York: Oxford University Press.
Paulics, Veronika. 2000. 125 Dicas: Ideais para a Ação Municipal. São Paulo: Pólis/ Rio
de Janeiro: Banco Nacional de Desenvolvimento Econômico e Social.
Paulics, Veronika. 2003. Disseminação do Programa de Garantia de Renda Mínima no
Brasil (1991-1997). Masters Thesis, Fundação Getúlio Vargas-São Paulo, São
Paulics, Veronika. 2004. Disseminação de Inovações em Gestão Local. Paper presented
at the Prepared for delivery at the 2004 Meeting of the Latin American Studies
Association, Las Vegas, Nevada, October 7-9, 2004.
Pierson, Paul. 1994. Dismantling the Welfare State. New York: Cambridge University
Peterson, George E. 1997. Decentralization in Latin America: Learning through
Experience. Washington, D.C.: World Bank.
Plank, David N., José Amaral Sobrinho and Antonio Carlos da Ressureição Xavier. 1996.
Why Brazil Lags Behind in Educational Development. In Nancy Birdsall and
Richard Sabot eds., Opportunity Forgone: Education in Brazil. Washington,
D.C.: Inter-American Development Bank. pp. 117-145.
Riker, William H., and Peter C. Ordeshook. 1973. An Introduction to Positive Political
Theory. Englewood Cliffs, New Jersey: Prentice Hall.
Rocha, Sonia, and Guilherme Sedlack. 1999. The Bolsa-Escola Program. Paper presented
at the Social Protection: Strategies for Improving Policies in LAC, Washington,
D.C. June 7, 1999.
Rodrigues, Leôncio Martins. 2002. Partidos, Ideologia e Composição Social. Revista
Brasileira de Ciências Sociais, 17(48), 31-47.
Rogers, Everett M. 2003. Diffusion of Innovations (5th Ed.). New York: Free Press.
Rohter, Larry, and Anemona Hartocollis. 2007. Brazilian Politician Indicted in New York
in Kickback Scheme. New York Times. March 9, 2007. p. A9.
Roma, Celso. 2002. A Institucionalização do PSDB entre 1988 e 1999. Revista Brasileira
de Ciências Sociais, 17(49), 71-92.
Roncalli, Angelo Giuseppe, and Kenio Costa de Lima. 2006. Impacto do Programa Saúde
da Família sobre indicadores de saúde da criança em municípios de grande porte
da região Nordeste do Brasil. Ciência & Saúde Coletiva, 11(3), 713-724.
Rose, Douglas D. 1973. National and Local Forces in State Politics: The Implications of
Multi-Level Policy Analysis. American Political Science Review, 67(4), 11621173.
Rose, Richard. 1989. Ordinary People in Public Policy: A Behavioural Analysis.
Newbury Park, CA: Sage Publications.
Rose, Richard. 1993. Lesson-drawing in Public Policy: A Guide to Learning across Time
and Space. Chatham, N.J.: Chatham House Publishers.
Rose, Richard. 2004. Learning from Comparative Public Policy: A Guide to Lessondrawing. New York, NY: Routledge.
Rossi, Amamaria. 1997, Cristovam faz fé na Bolsa Escola. Correio Braziliense,
September 30, 1997. p. 2.
Rueschemeyer, Dietrich, and Peter B. Evans. 1985. The State and Economic
Transformation: Toward an Analysis of the Conditions Underlying Effective
Intervention. In Peter B. Evans, Dietrich Rueschemeyer, and Theda Skocpol eds.,
Bringing the State Back In. New York: Cambridge University Press. pp. 44-77
Ryan, Bryce, and Neal C. Gross. 1943. The Diffusion of Hybrid Seed Corn in Two Iowa
Communities. Rural Sociology, 8, 15-24.
Sabatier, Paul A. 1999. Theories of the Policy Process. Boulder: Westview Press.
Sabatier, Paul A. and Hank C. Jenkins-Smith 1993. Policy Change and Learning: an
Advocacy Coalition Approach. Boulder: Westview Press.
Samuels, David. 2000. Reinventing Local Government: Municipalities and
Intergovernmental Relations in Democracy Brazil. In Peter R. Kingstone and
Timothy J. Power eds., Democratic Brazil: Actors, Institutions, and Processes.
Pittsburgh: University of Pittsburgh Press. (pp. 77-98)
Samuels, David. 2003. Ambition, Federalism, and Legislative Politics in Brazil. New
York: Cambridge University Press.
Samuels, David. 2004. The Political Logic of Decentralization in Brazil. In Alfred P.
Montero and David Samuels eds., Decentralization and Democracy in Latin
America. Notre Dame: University of Notre Dame Press. pp. 67-93.
Sartori, Giovanni. 1969. Politics, Ideology, and Belief Systems. American Political
Science Review, 63(2), 398-411.
Serra, José. 2000. Ampliando o Possível: A Política de Saúde do Brasil. São Paulo:
Silva, Joana Azevedo da, and Ana Sílvia Whitaker Dalmaso. 2002. Agente Comunitário
de Saúde. Rio de Janeiro: Editora Fiocruz.
Silva, Maria Josefina da, and Rui Martinho Rodrigues. 2000. O agente comunitário de
saúde no processo de municipalização da saúde. Revista Eletrônica de
Enfermagem, 2(1).
Silva, Sônia Cristina. 1996. Complemento a Renda Familiar Reduz Evasão Escolar no
Distrito Federal. O Estado de São Paulo, February 5, 1996. p. A11.
Simmons, Beth A., and Zachary Elkins. 2004. The Globalization of Liberalization: Policy
Diffusion in the International Political Economy. American Political Science
Review, 98(1), 171-189.
Skocpol, Theda. 1979. States and Social Revolutions: a comparative analysis of France,
Russia, and China. New York: Cambridge University Press.
Snyder, Richard, and Richard Owen. 2001. Devaluing the Vote in Latin America.
Journal of Democracy, 12(1), 146-159.
Solla, Jorge José Pereira, Maria Guadalupe Medina, and Maria Beatriz Pragana Dantas.
1996. O PACS na Bahia: avaliacäo do trabalho dos agentes comunitários de
saúde. Saúde em Debate, 51, 4-15.
Sousa, Maria de Fátima de. 2003. O PSF an Cidade de São Paulo Trabalhando em
Parceria: O Compromisso Possível. In Maria Fatima de Sousa and Áquilas
Mendes eds., Tempos Radicais da Saúde em São Paulo: A Construção do SUS na
Maior Cidade Brasiliera. São Paulo: Hucitec. pp. 91-94.
Sousa, Maria Fatima de, Ed. 2002. Os Sinais Vermelhos do PSF. São Paulo: Editora
Sousa, Maria Fátima de, and Áquilas Mendes. 2003. Tempos Radicais da Saúde em São
Paulo: A Construção do SUS na Maior Cidade Brasiliera. São Paulo: Editora
Spink, Peter K., Silvio Caccia Bava, and Veronika Paulics eds.. 2002. Novos Contornos
da Gestão Local: Conceitos em Construção. São Paulo: Pólis/Programa Gestão
Pública e Cidadania-FGV-EAESP.
Steck, Juliana. 1997. Unesco Adota Project da Bolsa-Escola: Programa do GDF vai ser
implantado em outros países com o financiamento do órgão do ONU. Jornal de
Brasília. July 5, 1997. p. 21.
Stepan, Alfred. 1978. The State and Society: Peru in Comparative Perspective.
Princeton: Princeton University Press.
Stone, Deborah A. 1989. Causal Stories and the Formation of Policy Agendas. Political
Science Quarterly, 104(2), 281-300.
Stone, Diane. 1999. Learning Lessons and Transferring Policy across Time, Space and
Disciplines. Politics, 19(1), 51-59.
Strang, David, and John W.Meyer. 1993. Institutional Conditions for Diffusion. Theory
and Society, 22(4), 487-511.
Strang, David, and Sarah A. Soule. 1998. Diffusion in Organizations and Social
Movements: From Hybrid Corn to Poison Pills. Annual Review of Sociology, 24,
Suplicy, Eduardo Matarazzo. 2002. Renda de Cidadania: A Saída é Pela Porta. São
Paulo: Cortez Editora/Editora Fundação Perseu Abramo.
Taylor, Michael. 2006. Rationality and the Ideology of Disconnection. New York:
Cambridge University Press.
Teixeira, Suely C.S., Valéria de O Monteiro, and Verônica A.Miranda, 1999. Programa
Médico deFamília no Município de Niterói. Estudos Avançados, 13(35), 147-155.
Tendler, Judith. 1997. Good Government in the Tropics. Baltimore: Johns Hopkins
University Press.
Terra, Valéria, and Ana Maria Malik. 1998. Programa Médico de Família de Niterói. In
Luis Fujiwara, Nelson Alessio, and Marta Farah eds., 20 Experiências de Gestão
Pública e Cidadania. São Paulo: Programa Gestão Pública e Cidadania, FGVEAESP.
Thelen, Kathleen, and Sven Steinmo. 1992. Historical Institutionalism in Comparative
Politics. In Sven Steinmo, Kathleen Thelen and Frank Longstenth eds.,
Structuring Politics: Historical Institutionalism in Comparative Analysis. New
York: Cambridge University Press. pp. 1-32.
Theodoulou, Stella, and Matthew Cahn. 1995. Public Policy: the Essential Readings.
Englewood Cliffs, N.J.: Prentice Hall.
True, Jacqui, and Michael Mintrom. 2001. Transnational Networks and Policy Diffusion.
International Studies Quarterly, 45(1), 27-57.
Tulchin, Joseph S., and Allison M. Garland. 2000. Social Development in Latin America:
The Politics of Reform. Boulder: Lynne Rienner.
UNESCO. 2007. Brazil. World Data on Education. UNESCO-International Bureau of
Education. Downloaded, January 9, 2007. Available at:
União espera beneficiar 5 milhões de crainças com Bolsa-Escola. 2000. Estado de São
Paulo, p. C1.
Vasconcellos, Maria da Penha Costa. 1999. Reflexões Sobre a Saúde da Família. In
Eugênio Villaça Mendes (Ed.), A Saúde no Nível Local São Paulo: HUCITEC.
pp. 155-172.
Vawda, Ayesha. n.d.. Brazil: Stipends to Increase School Enrollment and Decrease Child
Labor, A Case of Demand-Side Financing. Available at:
Viana, Ana Luiza D´Ávila, and Mario Roberto dal Poz. 1998. A Reforma do Sistema de
Saúde no Brasil e o Programa de Saúde da Família. PHYSIS: Revista Saúde
Coletiva, 8(2), 11-42.
Villaméa, Luiza. 2001. Made in Brasil. Istoé, 1634, 38-39.
Villatoro, Pablo. 2004. Programas de Reducción de la Pobreza en América Latina: Un
Análisis de Cinco Experiencias (No. 87). Santiago, Chile.
Walker, Jack L. 1969. The Diffusion of Innovations among the American States.
American Political Science Review, 63(3), 880-899.
Walt, Stephen. 2000. Fads, Fevers, and Firestorms. Foreign Policy, 121, 34-42.
Wampler, Brian. 2004. Expanding Accountability Through Participatory Institutions:
Mayors, Citizens, and Budgeting in Three Brazilian Municipalities. Latin
American Politics & Society, 46(2), 73-99.
Wampler, Brian and Leonardo Avritzer. 2004. Participatory Publics. Comparative
Politics, 36(3), 291-312.
Wejnert, Barbara. 2005. Diffusion, Development, and Democracy, 1800-1999. American
Sociological Review, 70(1), 53-81.
Welch, Susan and Kay Thompson. 1980. The Impact of Federal Incentives on State
Policy Innovation. American Journal of Political Science, 24(4), 715-729.
Werneck, Felipe. 2001. Bolsa Escola terã R$1.7 bilhão. O Estado de São Paulo, p. A9.
Weyland, Kurt. 1996. Democracy without Equity: Failures of Reform in Brazil.
Pittsburgh: University of Pittsburgh Press.
Weyland, Kurt, Ed. 2004. Learning from Foreign Models in Latin American Policy
Reform. Washington, D.C.: Woodrow Wilson Center Press.
Weyland, Kurt. 2005. Theories of Policy Diffusion: Lessons from Latin American
Pension Reform. World Politics, 57(2), 262-295.
Weyland, Kurt. 2007. Bounded Rationality and Policy Diffusion: Social Sector Reform in
Latin America. Princeton: Princeton University Press.
World Bank. 2001. Brazil: An Assessment of the Bolsa Escola Programs. Retrieved June
6, 2005, 2005, from
Natasha Borges Sugiyama was born in Gainesville, Florida on September 10,
Shortly thereafter her parents, Maria Lúcia Borges Sugiyama and Sugiyama
Iutaka, who are both originally from Rio de Janeiro, Brazil, settled in New York, New
York. Natasha is a graduate of The Spence School. She received a B.A. from Oberlin
College, Ohio in 1996, with a double major in Politics and Latin American Studies. As
the recipient of a Woodrow Wilson Public Policy and International Affairs Fellowship,
she completed an M.P.Aff. at the Lyndon B. Johnson School of Public Affairs at the
University of Texas at Austin in 1998. Afterwords, she joined the Governance and Civil
Society program of the Ford Foundation, New York, as a Program Associate. Since
entering the doctoral program in Government at the University of Texas at Austin,
Natasha has won a number of grants and fellowships including the: American Political
Science Association Minority Fellowship (2000), Fulbright-IIE Scholarship (2003),
David L. Boren Graduate Fellowship (2003), Spencer Dissertation Writing Fellowship
(2004), Esther Garrett Edgerton P.E.O. Scholar Award (2004), and MacDonald
Fellowship from the University of Texas at Austin (2005).
Permanent address:
4305 Duval Street, Apartment 312, Austin, Texas 78751
This dissertation was typed by the author.
Fly UP