Social welfare policy was introduced in the 1988 Federal Constitution (FC-88). It operates by means of direct cash transfers to citizens and social welfare services, being offered mainly by the municipal level of government. We believe that studying this policy would be propitious in understanding the ability of the Brazilian federal government to induce the construction of a bureaucracy at the municipal level. On the one hand, the existing data show significant growth both in service and benefits offerings and in this bureaucracy, with the hiring of more than 100,000 direct officials in less than five years’ time. On the other hand, the literature points to the importance of regulatory authority centralization at the federal level in Brazil, which is a federal state marked by inequalities in subnational governments, particularly local governments (Arretche, 2012:174). We ask whether the results achieved by the policy regarding the construction of a bureaucratic structure at the local level, in addition to public facilities for service provision, are associated with the federal government’s capacity to induce the construction of this bureaucracy. Does the government have an effective transformative capacity that influences municipalities and therefore the capacity of the sub-national level to offer this policy? Bearing this question in mind, our goal is to analyze the role of the Union, specifically the Ministry of Social Development and Fight against Hunger (Ministério do Desenvolvimento Social e Combate à Fome - MDS), in inducing municipalities to create a local bureaucracy that would facilitate the institutionalization of the National Social Welfare Policy (Política Nacional de Assistência Social - Pnas), considering its legislative authority, the administrative resources available to it, and the agreements made between the federal entities.

We hypothesize that the MDS’s transformative capacity occurs via an incremental process that we call organizational learning, pertaining to both its own bureaucratic apparatus and administrative strategies that induce municipalities to act by producing specific regulations associated with financial incentives. This transformative capacity is made possible by agreements signed between the federal entities in federal policy coordination areas (the National Social Welfare Council (Conselho Nacional de Assistência Social - Cnas) and the Tripartite Interagency Commission (Comissão Intergestores Tripartite - CIT)), in which goals are established along with the means to achieve them, in this case, in an institutional learning process.

This hypothesis is supported by a theoretical combination that, on the one hand, considers the classic means of public administration control by imposing regulations and the use of hierarchical structures and standardized procedures and, on the other hand, introduces the use of deliberative spaces as a means of formulating federal agreements geared toward implementing policy in a coordinated manner (Evans, 2002). Considering the non-existent legacy (at that time) of state structures at the local level for the provision of social welfare services and money transfer, in addition to the low bureaucratic capacity of most of the 5,570 Brazilian municipalities, the structuring of this bureaucracy is only possible when based on a synergy between the classical model and the deliberative space for the agreement of federal entities (Evans, 2002).

This case study is structured using process tracing (Beach and Pedersen, 2013; Bennett and Checkel, 2014), which allows the identification and analysis of causal mechanisms, enabling inferences to be made regarding the links between actions and reactions, that is, of the existing causal links between the production of regulations and agreements and the construction of the municipal bureaucratic structure.

First, we briefly present some theoretical reflections that provide us with the analytical framework and that guide the data analysis. We then analyze the MDS’s organizational learning and its ability to induce the action of another federal entity. Subsequently, we proceed to analyze the results of the agreements and the actions that demonstrate the institutional learning that occurs in the exchange between the MDS and deliberative spaces, and finally, we present our conclusions about the analysis.

State capacities vary in time and space (Gomide, 2016), and different state functions require the mobilization of different capacities - the provision of social welfare demands different skills from those required to generate a surplus (see, e.g., Hanson and Sigman, 2011 apud Cingolani, 2013). The multi-dimensionality of the concept and its relationship with the various state functions greatly expand its reach, which is not very useful for analysis (Weiss, 1999). A broad concept such as is a conceptual background and it demands a narrower, a systematized concept for studies in order to be operationalizable (Adcock and Collier, 2001). Some authors show how the concept’s operationalization has been more difficult than strengthening the conceptual framework (Enriquez and Centeno, 2012). Thus, one cannot speak of state capacity in general but of specific arenas or dimensions of this capacity (Weiss, 1999:4; Cárdenas et al., 2015). Here, we choose to use the concept of state capacity in two dimensions: one related to the dependent variable and the other to the processes and mechanisms (which are our focus) required to achieve the former.

In Brazil, the implementation of social policy has been decentralized to municipalities with different levels of administrative capacity but that should also organize and offer public social services according to nationally established standards. In addition, a certain concentration of regulatory authority in the Union was established (Arretche, 2012). One must consider that the municipalities’ political autonomy does not necessarily correspond to their real capacity to formulate and implement public policies, insofar as there may not exist the administrative capacity, budget, or taxation for that purpose at the municipal level (Arretche, 2012; Pierson, 1996; Evans et al., 1985).

Under these conditions, we argue that the study of municipal state capacity, or even of a narrower concept, such as municipal bureaucratic capacity, must consider an analysis of not only the federal government’s transformative capacity, understood as its potential to mobilize consent and institutionalize cooperation, but also the relationship between the federal entities. The term “transformative capacity” was initially coined by Evans (1995) to explain the role of newly industrialized countries in shaping structural changes, occurring in the economy and in industrialization, that facilitate industrial development. This concept was later further developed not only by the author but also by others, such as Weiss (1999). We agree with Evans (1995) and Weiss (1999), who argue that the level of state intervention alone is not informative of its power but that its transformative capacity is (Cingolani, Thomsson, and Crombrugge, 2013).

In distinguishing transformative capacity and intervention, Weiss considers that the former represents the ability to coordinate change whereas the latter does not necessarily induce it. The agents of a transformative state seek to shape and coordinate resources across a broad spectrum and have the central role of “orchestrating cooperation of the main organized partners and, […] coordinating resources essential to their bargaining” (Weiss, 1999:14). Their transformative capacity includes providing infrastructure, socializing risk, and encouraging cooperation.

In turn, in democratic and federal states in which policies are implemented in a decentralized manner and in which partners are autonomous federal entities, with common and complementary skills, transformative capacity cannot be sustained only by the use of classic administration model control strategies, based on the imposition of regulations, or by the use of hierarchical structures and standardized procedures. The institutionalization of public democratic decision-making processes is necessary, with the participation of actors who can also control public resource allocation (Evans, 2002). Such deliberative processes provide the political support necessary for cooperation to occur and for the administrative measures that are required of the federal entities to be formulated and implemented.

Moreover, the dynamism of politics and the environment in which it occurs determines that the state’s transformative capacity depends on a set of instruments that must be constantly changing and adapting because new issues require new tools. In other words, the scope and targets should be redefined to be compatible with the dynamic context (Weiss, 1999:8). Therefore, transformative capacity is related to institutional and organizational learning, which occur incrementally, most of the time, and which allow the policy itself to be changed. Learning occurs through interaction and the development of knowledge in the ongoing context; that is, it is an interactive process between actors involved in certain institutional arrangements (North, 1990; Morgan, 1997). According to Van Brabant (1997), organizational learning differs from institutional learning, given that the former “refers to collective learning within an organization” (1997:2) and the latter refers to observable learning in a larger system, between different agencies. This learning is incremental in that new knowledge is generated from that acquired and tested in previously occurring interactions, both internally, within organizations, and between institutions.

The process tracing methodological approach allows the analysis of causal mechanisms, and that is our major contribution: the first one was about the causal links between the production of regulations and administrative acts and changes on the municipal bureaucratic structure, and the second one was about the federative relations and its importance to this process.

Basically, we traced the municipal bureaucracy construction by observing the installation of public facilities and Human Resource hiring in the municipalities and the factors that determined it. Hypothesis is that this process is an effect of the transformative capacity of the MDS as well as of the inter-federative cooperation. As seen, the literature suggests that the organizational and the institutional learning are central mechanisms to explain this process, insofar as they uncover the interactions that had shaped the decisions.

In order to reach the proposed objective, the analysis was carried out in two moments: one observing internal pieces of evidence from the MDS and its relation with the municipalities, and the other observing evidence of the interactions between the federative entities. For the first one, basically 3 factors were analyzed: the structuration of a federal bureaucracy, the organizational and institutional learning of this bureaucracy and its relation with other organs, and the inductive incentives by ordinances and decrees captained by the MDS. Analyzing each of these factors as a necessary condition to the installation of public facilities and human resource hiring in the municipalities. Another causal mechanism was about the necessary inter-federative relation on social assistance policy in Brazil, then, we observed several pieces of evidence of agreements and cooperation on the CIT, the political arena where the covenants are made. There, the agreements with municipalities by one hand, and the connection between the SNAS bureaucracy with the CIT by another hand, were linked to effects on installation of public facilities and human resource hiring in the municipalities.

The evidence raised for this analysis focuses on the period from 2004 to 2012 and is collected by means of documentary analysis of seven laws, four decrees, 236 ordinances, two regulations, 141 CIT resolutions, and five CNAS resolutions, in addition to MDS monitoring reports and in-depth interviews.¹ These documents allow us to observe the creation of direct and indirect incentives for the signing up and agreement of municipalities to national goals that were not necessarily theirs and the effect of these incentives on the construction of their own bureaucracies.

We classify the ordinances by nature: those relating to the PBF, to the CadÚnico, to funding, to integration between income transfer and services, and to other issues, showing their distribution over time and correlating with the human resource hiring in the municipalities. However, it is important to highlight that only the documental analysis would not permit the connection of all chain links. The interviews with who occupied strategic positions at the Senarc, and the Snas were essential to meaning the process and the mechanisms tracing because they fill informational gaps.

Finally, the Venn Diagram technic was used with the explicit objective to represent each of those factors act as necessary or sufficient condition, showing the equifinality of this process and that there has been no factor that has presented itself as a sufficient condition. The sufficient condition was due to the sequence and conjunction of factors.

The Brazilian federal organization and the decentralization process in this area indicated that the municipal government level should be responsible for the provision of social welfare services. However, the municipal structure was not ready to take on this task. What are we talking about? What is the phenomenon that the analyzed processes are revealing? Over a 10-year period, a significant change indicative of an effective construction of an administrative and bureaucratic structure at the local level in the social welfare area has occurred. This is the basis of what we call capacity and is represented here by the number of employees who are distributed in public facilities providing social welfare services (as well as by the construction of the facilities themselves). This hiring of staff and the installation and launching of public facilities are therefore expressions of what we call the municipality’s bureaucratic structure, used as a conceptual proxy for the bureaucratic capacity to which we are referring.

The data are impressive: until 2004, when a new left-wing government took over and the MDS was established, there were no specific and appropriate physical public units for the provision of social welfare; that is, there were no facilities for the provision of social welfare services, and therefore, there were no state human resources for the same.² In 2013, there were already 7,883 Social Welfare Reference Centers (Centros de Referência da Assistência Social - Cras) in 5,394 municipalities with 75,241 workers; 2,249 Specialized Social Welfare Reference Centers (Centros de Referência Especializados de Assistência Social - Creas) with 20,938 employees; and 131 Homeless Population Service Centers (Centros de Atendimento à População de Rua - POP Centers) with 1,914 workers, totaling 98,093 workers responsible for the delivery of social welfare services. Figure 1 shows the hiring over time.

Figure 1 Source: Elaborated by the authors, from Suas Census data.

In addition, in 2013, 99.7% of the municipalities were operating the Single Registry (Cadastro Único - CadÚnico), a means of identifying families in vulnerable income situations that require the state’s social protection. Although it is managed by the MDS, the registry is operated by local governments. There were 25.3 million registered families in that year and 18 million in 2010. Additionally, 13,872,243 families received benefits from the Family Grant Program (Programa Bolsa Família - PBF), 50.4% of which were in the Northeast region (Camargo et al., 2013).

The proposed tracing verifies the extent to which the production of regulations at the central level (both administrative, which induce municipalities to action, and those associated with financial incentives) influenced the hiring of human resources, i.e., the municipal capacity to offer services, and shows whether this process denoted an organizational and institutional learning and whether the deliberative spaces promoting coordinated decisions boosted this construction process.

Because municipalities have the political autonomy to adopt a national policy or not, the question is, why did the municipalities accept the transfer of responsibility for the management and provision of social welfare services that basically rely on human resources? After all, the installation of administrative capacity involves high costs for most of the 5,570 Brazilian municipalities, which have less than 20,000 inhabitants and that, in addition to their low existing administrative capacity, also have low fiscal capacity.

We work with two hypotheses: first, that the MDS’s transformative capacity to induce behavior is relevant in understanding the changes occurring at the municipal government level starting in 2004; and, second, that this alone would not be sufficient for the cooperation of municipalities and that, therefore, the agreements occurring in the deliberative bodies were also a necessary condition for such a process to occur.

State capacity must therefore be considered not only in its administrative and transformative dimensions but also in relational terms; that is, one must consider the relationships established between the federal entities. In this sense, the transformative capacity of the federal government, the process of organizational and institutional learning, and cooperation negotiated in deliberative environments can directly affect the construction of the municipality’s bureaucratic-administrative structure. Therefore, we argue that transformative capacity, in a federal state with autonomous entities, is not without relational processes, which associate administrative actions (such as regulations and incentives) with cooperation negotiated in deliberative environments.

Initially, one aspect to note is the growth of the MDS’s bureaucratic structure in relation to its responsibilities. Created in 2004,³ the MDS was established without sufficient technical staff to perform its duties, but by mid-2014, it had enjoyed a substantial increase from 383 to 981 (256.13%) direct administration employees. It still has approximately 20% fewer employees than ministries such as Social Security and Education, which at that time had 1,231 and 1,209 employees, respectively.

A second important aspect is that the MDS, in establishing its organizational structure,⁴ distinguished four areas of activity: national social development policy; national policy on food and nutrition security; national social welfare policy; and national citizen income policy. These were organized in their own manner, with public organs, funding, and specific management models,⁵ the latter two of which are analyzed in this work. Initially, this structure launched income transfer actions via the National Bureau of Citizenship Income (Secretaria Nacional de Renda de Cidadania - Senarc), the structure responsible for social welfare services and benefits, located at the National Social Welfare Bureau (Secretaria Nacional de Assistência Social - Snas). This reflected how social welfare was institutionalized and implemented.

Due to this structure, two main bureaucracies were created in the Ministry, each with different characteristics and abilities. Initially, actions were developed by third-party technicians and consultants with specific functions (interview 1). Partnerships with international organizations such as the United Nations Educational, Scientific and Cultural Organization (Unesco), the Inter-American Development Bank, and the World Bank made it possible to build frameworks, from 2005 onward, for performing the Senarc’s institutional functions (Dutra, 2014), which now has a professional bureaucracy, recruited specifically for the functions that it has to perform. The manner in which policy was conducted in that Bureau attracted a considerable amount of public policy managers at the federal level, who were highly skilled and known as Public Policy and Governmental Management Specialists. At the Snas, the bureaucracy that assumed these functions came from a long history in the social welfare area, with political recognition of its involvement. In practice, the Snas had human resource weaknesses that significantly hindered the performance of its duties (interview 2).⁶ This organizational distinction, in principle, would not cause problems for the institutionalization of social welfare if there was concern for creating organic unity between the bureaus, but this did not occur (interview 1), which had consequences for policy in general and for the integration of income transfer and social welfare services in particular, as discussed below.

A first aspect to be highlighted is the effect of regulation on income transfer and on social welfare services and its consequences for the institutionalization of policy. Until 2004, there was no specific regulation on income transfer, which was standardized in Law 10,835/2004, which established the Citizen’s Basic Income (Renda Básica de Cidadania) but which did not establish a formal structure for its organization or its link with social welfare. Therefore, the Senarc had relative freedom to create rules for its area of operation, including the maintenance of a direct relationship with municipalities and, subsequently, with states, establishing a centralized form of coordination (interview 1). The Snas was guided by pre-existing regulations (such as the Organic Social Welfare Law of 1993 and Basic Operational Regulations) and was organized with a coordinated decentralization, in a relationship with states and municipalities mediated through the CIT and the Cnas. The rules needed to be agreed upon within these bodies. The characteristics of the Snas demanded strong federal connections, requiring of its bureaucracy a significant political and relational capacity, whereas the income transfer area was based on top-down government decisions, requiring the bureaucracy to have a predominantly technical capacity (interview 2). Thus, although belonging to the same ministerial structure, income transfer and social welfare benefits and services were treated as separate policies, and initially, little was done to create organic unity between them. In practice, they competed with each other for presidential attention and for funding, which were clearly at that time tipped toward transfer (interview 2).

Added to that was the distinct nature of the areas. A clear difference between the bureaucracies in the income transfer and social welfare service and benefits areas can be observed - the former had a more administrative and managerial profile, due to higher insulation and the quality of its specialized bureaucracy, taking regulatory initiatives of a more technical and operational nature; the latter had a more political profile and was involved in spaces that require intense institutional relationships, having to negotiate regulations with federal entities and civil society. The process of interaction, which was not always friendly, produced an instance of organizational learning that was important to the construction of the transformative capacity of the MDS: the need for technical and operational strengthening of the SNAS and the importance of political debate for the Senarc (interview 1).

From this process, it can be observed that the federal government acted as a behavior inducer at other government levels. Although it was structured with bureaucracies with distinct characteristics, soon after it had been created, the MDS began to move toward its intended transformations. One of these was to use its regulatory authority to establish flows and procedures and to standardize service offerings and income transfer, introducing the financial incentives associated with these changes. The Ministry systematically regulated the transfer of federal resources, encouraging the signing up of local governments. The transformative capacity of the Ministry, therefore, can be verified by regulatory and administrative acts, which, in addition to regulating its own activities, also placed demands upon other government levels, particularly the municipalities. This strategy was embodied in ordinances issued by the MDS, following the same approach as used by Health (Arretche, 2002, 2004), which is configured as institutional learning on the part of the federal government.

Figure 2 shows the MDS’s regulatory activity in the studied period, distinguished according to the income transfer and social welfare services and benefits areas. A total of 236 ordinances were issued in nine years, specifically addressing social welfare policy (inter-ministerial ordinances are not included here). The peak in 2006 and the stronger decrease after 2009 are noteworthy.

Figure 2 Source: Elaborated by the authors

However, the transformative capacity of the MDS can only be analyzed based on what constitutes the studied policy, in this case, the operationalization of income transfers and services.

The year 2004 is the starting point for our analysis because it was in this year that two legal instruments that were important for the area were formulated: the PBF⁷ and the National Policy for Social Welfare (Política Nacional de Assistência Social (Cnas Resolution 145/2004)), which details and regulates the regulatory and operational aspects of the Organic Law on Social Assistance (Lei Orgânica da Assistência Social - Loas), which had not previously been done. These regulations allow us to verify the MDS’s actions in regard to effecting them, considering essential intergovernmental management, encompassing the three government levels. It is important to note Decree No. 5209/2004, which regulated the PBF and defined its decentralized and intersectoral form, with community participation and social control. It stipulated that other government levels that wished to join the program should formally sign up to it and that local managers had until the end of 2005 to make the necessary provisions to develop it in the municipality. To effect sign-up, Ordinance No. 246/2005, which defined the instruments of municipality sign-up to the PBF and the Single Registry,⁸ was established. This extended the validity of the terms previously signed for the School Grant and Food Grant and established the appointment of a local manager for the PBF and a formal social control body to monitor its implementation. This period was undoubtedly a strong organizational learning period in a Ministry that was at the beginning of its structuring.

Of all of the above ordinances, 49 are related to income transfer, published by the Senarc. We can classify these ordinances by nature: those relating to the PBF, to the CadÚnico, to funding, to integration between income transfer and services, and to other issues. Figure 3 shows their distribution over time and allows us to observe the intensity of PBF regulation.

Figure 3 Source: Elaborated by the authors.

Additionally, in 2005, three more ordinances relating to the PBF were issued: Ordinance No. 360, which established the criteria and procedures for the transfer of financial resources allocated to federal entities for the implementation and development of the PBF and for the maintenance and improvement of CadÚnico; Ordinance No. 551, which regulated PBF conditionalities management; and Ordinance No. 555, which established the rules and procedures for PBF benefits management. In addition to these, there was Ordinance No. 666, which regulated integration between the PBF and the Child Labor Eradication Program (Programa de Erradicação do Trabalho Infantil - Peti),⁹ this last program, which existed previously and had specific legislation, also was led to the CadÚnico.

CadÚnico was an important inducer of changes in municipal management¹⁰ and was the second most addressed theme in the analysis period (Mostafa and Sátyro, 2014). The first audit of the database, performed in 2005 after allegations of irregularities in the PBF, enabled the SENARC to identify the municipal weaknesses that generated the inconsistencies identified. It was thereby able to establish a series of actions directed toward the municipalities to guide them and enable them to use the Registry properly (interview 2). This was a time of intense learning, not only the organizational learning of the MDS, in the sense of understanding how it worked, but also institutional learning, in understanding the strategies that actually achieved results at the municipal level. According to the interviewees, the dialogue with partner municipalities was systematic in this process. Some municipalities served almost as laboratories and, because they shared their experiences with the Ministry, helped federal technicians understand certain processes at the local level and, in particular, at the street level (interview 3).

This process led to the registry updating performed by municipalities, established by Decree No. 360, and introduced intergovernmental transfers direct from the Union to the municipalities that had joined the Suas and the PBF, through the National Social Welfare Fund, which aimed to support local management of the program. Municipalities that did not join the SUAS received federal funding by agreement, and management was by means of the states. The re-registration process gained centrality in the actions of municipalities and demonstrated their institutional weaknesses, leading most municipalities to review their organization and infrastructure (Suárez and Libardoni, 2007 apud Dutra, 2014; Mostafa and Sátyro, 2014). This moment is central to this chain because it is when the federal government effectively learned the most effective form of dialogue with the municipalities (interview 3).

Therefore, what we observe are the initial regulations governing the PBF directly affecting the municipalities to the extent that they signed up to the program, particularly in the aspects of management and demanding the appointment of a manager and the establishment of a local social control body, regulating funding destined for its implementation and defining procedures for the management of the benefit and its conditionalities. Dutra (2014) notes that the Sign-Up Contract (required) signed by the municipalities reveals the size and complexity of their new roles. This contract generates an agreement of responsibility for the registration of potential beneficiaries, the management of benefits, the monitoring of conditionalities connected to municipal and state schools, and linking with complementary programs that can be offered by the municipality or the state, including conditioning the expansion of the program to the performance of municipal management. Contrary to what the literature highlights regarding the municipalities’ resistance to signing up to such a policy, in which implementation and services all occur at the local level, sign-up was quick: between July and December 2005, 4,991 municipalities (89.7%) had already signed the contract, and during 2006, another 547 (9.8%) did so, leaving only 25 in 2007, which shows the strong inducive ability of the federal government.

Thus, management standards, flows and procedures established by the national coordination, were imposed upon the municipalities that formed the PBF provision network, which forced them to build a previously nonexistent capacity to organize management and take on new functions. It was therefore necessary to build an administrative area focusing on municipal management in the social welfare area, which had a direct effect on existing structures. Interestingly, these determinations, particularly the registration and monitoring of conditionalities, had the potential to integrate income transfer and social welfare services at the local level, but in practice, the organization of the two bodies occurred in a sectarian manner in which the two competed for resources.

After the successful signing up of the municipalities, efforts to establish more regular and less discretionary standards for PBF funding at the municipal level were made, forming the content of Ordinance No. 148/2006, which established the Municipal Decentralized Management Index (Índice de Gestão Descentralizada Municipal - IGD-M). This index was undoubtedly one of the greatest induction instruments used by the MDS. It determined the quality parameters for the municipalities’ management of the PBF (indicators relating to the single registry, the certification of registration, the monitoring of education, and health conditionalities), through which federal funds to be transferred directly by the Union would be calculated, with monthly calculations, based on updates made at the local level. The index was the inducive instrument of the quality of the municipalities’ response because the determination of the funds transfer amount would depend on the quantity and quality of the submitted registrations. Although the implementation of this process initially encountered difficulties arising from the administrative incompetence of many municipalities (Lindert et al., 2007; Dutra, 2014), the IGD-M became the parameter for the funding of municipal PBF management.

The institution of the IGD-M led to the transfer of 2.2 billion reais to the municipalities between 2006 and 2013, and from 2010 to 2012, there was a nominal increase in resources of 60% (Paiva, Falcão, and Bartholo, 2013). This resource contribution model, aimed at strengthening municipal management, was sustained by the regularity of transfers and encouraged the existence of the municipality’s own bureaucracy, including strengthening the area of social welfare at the local level, one more link in the chain described here. The IGD-M led municipalities to form teams for monitoring beneficiary families and for the operationalization of the CadÚnico to effectively use federal transfer balances and to form control agencies. In this regard, Decree No. 7322/2010 and Decree No. 754/2010 imposed major changes on the IGD-M, among them, the municipality’s signing up to SUAS and the effective role of the municipal social welfare councils as PBF control bodies. This integration process culminated in Law No. 12,435/2011, which amends the Loas and defines the organizational form of the Suas, clearly a step forward in the construction of the municipal structure based on the learning process described here.

In 2010, there were important changes in the calculation of the IGD-M (which began to be called the IGD-PBF) due to Decree No. 7,322/2010 and Decree No. 754/2010. The indicators were transformed into one factor (the PBF Operation Factor), which includes proof of expenditure of funds allocated to municipalities through direct transfer. This change occurred because it was diagnosed that many municipalities did not implement the budget, which is relatively free of constraints within the area. This innovation demonstrates the low funding implementation capacity of municipalities and points to federal government inducement, in the sense that local governments effectively spent the funds transferred fund-to-fund¹¹ (Sátyro and Cunha, 2016).

Regarding services, it is first important to stress the legacy: their history was marked by the fragmentation of supply according to government levels,¹² types of actions and suppliers, and the lack of regulation effectively determining the role of each federal entity in the provision of services and in the execution of policy. This legacy speaks to the challenge faced when assembling a system and to the very conception of political management that results in unity in the scope of protection (Dutra, 2014).

If municipalities need resources to implement the PBF, which is an income transfer made directly to the recipient and therefore, one would assume, less complex to manage at the local level, then it is possible to infer the difficulties of deploying a service network. The inducing role of the federal level has been essential in this regard, though not sufficient by itself, as discussed below. As shown in figure 1, most ministerial orders have fallen within the domain of the Snas. Figure 3 shows 85 ordinances that are classified according to their nature: funding, the Suas, the BPC, and a heterogeneous set classified as “other”. Interestingly, 52.94% (45) of them relate to funding.

Two factors are relevant to understand the process of inducing the construction of municipal bureaucratic capacity. The first is the formulation of the Basic Operational Norm of Suas Human Resources (Norma Operacional Básica de Recursos Humanos do SUAS - NOB-RH), agreed upon in the CIT and expressed in Cnas Resolution No. 269/2006, which specified the infrastructural bases relating to human resources required for the general realization of the Pnas and for the municipalities. This NOB establishes the basic composition of the reference teams according to the management level and size of the municipality and relates them to the actions to be developed. This demonstrates an awareness that the construction of the Suas needed human resources at the local level, a strong link in the chain. Below, we discuss how it has directly impacted recruitment at the municipal level.

The NOB-Suas (2005, 2012) had a significant impact on the relationship between the federal government and local governments as they changed the Union’s funding system, introducing criteria for sharing federal resources. This meant that the direct fund-to-fund transfer started to follow non-discretionary criteria and acquired regularity. This was a fundamental step toward consolidation of the ongoing process because it gave security to the local manager to hire staff. An initial measure was the establishment of management levels, consistent with municipal capabilities, which introduced positive selective incentives for municipalities to fall into line with the regulations and to comply with the minimum requirements for management. Transition rules were also established so that municipalities could adapt to the new guidelines. These criteria were subsequently amended but played an important inducing role in the construction of municipal capacities.

A second factor was the issuing of Law No. 12,435/2011, amending the Loas and organizing the Suas. In article 12-A, the Suas’s Decentralized Management Index (Índice de Gestão Descentralizada do Suas - IGD-Suas) was created. It is structured based on the Cras Development Index (IDCRAS); that is, the law reinforces the role of municipal management in the execution of social welfare services and strengthens the CRAS as central government units for social welfare policy. Here, an instance of organizational learning should be emphasized: as the MDS managers learned the operation of the IGD-M, subsequently renamed the IGD-PBF, they used the same strategy for the IDCRAS and, now, for the IGD-Suas. Decree No. 7,636/2011, which regulates the IGD-Suas, in turn, is an important step toward the integration of transfers and services under the ambit of Social Welfare because it establishes the index as a reference for monitoring and evaluating the system as a whole - monetary transfers and services.

Figure 4 Source: Elaborated by the authors.

The process tracing method that we have used, however, can only be complete when we connect the above chain of events to the hiring of staff to perform such services. Social welfare services basically need qualified human resources and do not require sophisticated equipment, as in the case of health. Thus, what is observed is that NOB-RH has an immediate effect on the hiring of staff. Figure 5 shows the history of the ordinances and the evolution of staffing in these newly created facilities, demonstrating the strength of the links described here. The tracing process shows how these ordinances have shaped municipal action and, consequently, the number of staff hired. One can observe that the number of ordinances was particularly high in the early years of 2006 and 2007, particularly those relating to funding in 2006, with a clear purpose of creating direct incentives for municipalities; in addition, the NOB-RH was created in the same year. Figure 5 shows the evolution of staff numbers and allows us to infer the causal links discussed above. Starting in 2006, the ordinances relating to services begin to take prominence, revealing significant numbers in subsequent years, with a reduction in 2011. In turn, the number of staff at the local level, particularly at the Cras, grows by approximately 290% in this period.

Figure 5 Source: Suas Census data.

Undoubtedly, the federal government has the institutional resources to influence the choices of local government, thus affecting their agenda and making the transfer of federal funds to municipalities conditional upon signing up to national policy objectives, and to that end, it used its regulatory authority. As in health (Arretche, 2002, 2004), the policy has been heavily funded by the federal government; since its implementation, federal co-funding of the policy increased from 75.5% in 2004 to 79.9% in 2011 (Mesquita et al., 2012).

This process of organizational and institutional learning generated the National Typification of Social Welfare Services, via MDS/Cnas Resolution No. 109/2009, which standardizes services specific to social welfare. The next step was the reorganization of these services at the state and local levels; that is, managers began to bring into line actions that existed prior to typification. In the same year, the Services, Benefits, and Income Transfer Integrated Management Protocol was issued, expressed in MDS/CIT Resolution No. 07/2009, clarifying the integration of income transfer and social welfare services as a management strategy necessary to effectively ensure social welfare rights. Importantly, neither the typification nor the protocol was an exclusively MDS initiative; rather, they were negotiated and agreed upon in the appropriate bodies.

The political autonomy of the municipalities means that they may choose to join, or not to join, national projects. This makes the inducing ability of the federal government central to the achievement of national goals. In the case discussed here, it was first necessary to make the municipalities sign up to the program and then to think of methods of inducing them into standardizing actions to guarantee qualitative and quantitative levels and to guarantee the budget for this. The phenomenon that is observed here is necessarily procedural and shows institutional learning during its course, with each step generating results that lead to others; that is, incremental changes occur that allow new steps to be performed. Regulatory acts (Laws, Decrees, Ordinances, Resolutions) were the mechanism by which this occurred. They were issued successively and, after verification of the results achieved, demonstrated a real institutional learning in the intergovernmental relationship. Every action of the MDS generated reactions in the municipalities, and when each goal was achieved, another MDS action was necessary, promoting the creation and development of capacities at the local level. In this process, it was significant that the MDS was flexible in defining the “achieved target”, given that the 5,570 municipalities had very different response capabilities.

Analyzing the ordinances with the interviewees allowed us to trace the incremental dynamics of this induction process and the capacity to transform the municipalities in respect to signing up to the program. If, before, it was possible to observe the organizational learning of the MDS, then what is observed here is that, when each goal was achieved, the next trigger was fired in a relational learning process. Municipalities had two great reasons to sign up and provide more services: funding transfers, which no longer depended on the discretionary nature of the agreements and became fund-to-fund, and the credit received from the population at the local level for implementing more services.

Figure 6 is a table showing the relationship (of necessity and sufficiency) of each of the factors and processes analyzed with the structuring of a specialized bureaucracy at the local level. Although the factors and processes presented here are relevant to demonstrating the performance of the MDS, both its institutional and organizational learning and its regulatory role in national policy coordination, it is plausible that none of these alone would be sufficient to induce municipalities to hire staff and provide social welfare services. Funding may have played a strong role, but we still consider that all aspects listed are necessary conditions for this result, despite being insufficient on their own to cause it. However, it is possible to infer that the absence of any of these factors would constitute a sufficient condition for the negation of the result, that is, for the non-structuring of a service provision system at the local level with its own bureaucracy, which has as a real counterfactual case the welfare policy developed during the administration of President Fernando Henrique Cardoso (Sátyro and Cunha, 2014).

Figure 6 Source: Elaborated by the authors.

The structuring of social welfare policy constituted two important spaces of deliberation in regard to the Suas: Interagency Commissions (state and national), particularly national (CIT), and social welfare councils (the Cnas at the federal level). These two bodies are distinct from each other, particularly in regard to their composition and nature. The commissions have the role of promoting negotiations and federal agreement within the policy’s scope; they are necessary for the protection unit; they are composed of representative managers of the Union, states, and municipalities and have no regulatory authority. The councils are deliberation spaces that allow the participation of civil society and federal entities in the formulation and social control of the policy, and they have the authority to regulate the policy. Therefore, decisions about policy management are made in the CIT, and when regulation is required, they are passed through to the Cnas. These spaces of negotiation and deliberation between federal entities and civil society demonstrate the importance of the existence and development of these entities’ relational capacity.

The role of the CIT¹³ in the intergovernmental management process is very important, particularly if we consider that the policy should be offered via minimally standardized services across the country, which depends on administrative capacity, bureaucratic structure, and attention to local demands, among other aspects. Out of respect for the autonomy of federal entities, the CIT’s internal regulations determine that its decisions must be consensual; that is, they must obtain the consent of all representatives, even if doing so means that subjects take longer to debate. These decisions are translated into resolutions. From its creation in 1999 to mid-2014, the CIT issued 141 resolutions expressing such decisions.

The reach of the CIT’s decisions is evidenced by its recipients: the majority is addressed to the municipalities (47.5% - 67), which increases to 82% when those addressing more than one federal entity, including municipalities, are included. Of the rest, 19.15% (27) are resolutions directed toward the national level, 16.31% (23) to the states, Federal Districts, and municipalities together, 13.47% (19) to the states and Federal Districts, 2.84% (4) at the federal government level, and 0.71% (1) to the Federal Districts alone. The subjects covered by these resolutions include management and funding, which total (61.7%): 33% of all resolutions concerned management and 28.37% funding.

Resolutions relating to services and benefits totaled 9.22% and internal organization 9.93%, with planning, monitoring, evaluation, and specific issues being less significant. Interestingly, the specific issue of social welfare services is not significant. In addition, we sought to understand whether the decisions made by the CIT potentially generate conflicts or cooperation between federal entities. We found that 57% (81) potentially generate cooperation, 16.31% (23) address material that is clearly conflictive in regard to interfederal interests, and 26.24% (37) address other matters that do not involve potential conflicts or cooperation.

In asserting that the municipality is the recipient in over 80% of the resolutions and that 57% generate the idea of cooperation, it is clear that the resolutions made within the CIT’s ambit sharpen the focus on the municipalities as important actors in the implementation of Suas actions and on management and funding as the most frequent issues in debates and decisions. It also indicates a greater potential for cooperation than intergovernmental conflict. Many of the agreements occurring within the CIT’s ambit are related to regulations governing the management of the social welfare policy outlined above. These include the Basic Operational Norm of Suas, the Basic Operational Norm of Suas’s Human Resources, the Integrated Services and Benefits Management Protocol, the National Typification of Social Welfare Services, and, more recently, the Suas Management Improvement Pact. This regulation was the subject of intense debate in the CIT, and the agreements produced were enacted into law by the Cnas (which published them as resolutions). Other agreements that did not require regulation were published as ministerial ordinances so that they would have a regulatory effect on all federal entities.

These legal instruments address issues that directly affect the municipalities, particularly those relating to the management and funding of the policy, such as the organization of care levels and the structure of the governing body and of facilities offering services (including human resources), the type of services to be offered (including mandatory services), the format that the integration between these services and benefits must have, the goals to be achieved by the municipalities, and the corresponding deadlines. Therefore, the CIT is not only an important space for promoting the relationship between the entities, with strong potential to induce cooperation, but also a locus where decisions that directly affect the management of the Suas are agreed upon.

This negotiated and agreed-upon process of decision production, which, in turn, generates rules and regulations that effectively focus on all federal entities and affect policy management, points to the importance of the relational dimension among the necessary state capacities for conducting democratic politics in federal states. Cooperation is achieved not only through the existence of constitutional guidelines or administrative regulations but also principally through the construction of spaces and procedures that enable the expression and resolution of conflicts, in addition to agreeing upon the rules governing intergovernmental relationships. In the case of social welfare, the CIT is an important mechanism for the development of this capacity, including the making of decisions that were absolutely vital to the structuring of policy at a national level. These decision-making characteristics in the Suas make it possible to infer the relevance of the difference in profile of Senarc and Snas bureaucrats. In this case, the more political profile of Snas bureaucrats allowed this seam to be sewn.

One can observe from the table below the relationship of each of these processes with the assembly of facilities and hiring of staff in regard to their condition of need or sufficiency. One may imagine that a completely top-down relationship with the municipalities worked at the beginning of the implementation of the Family Grant when it needed a smaller, though essential structure. However, both for the implementation of social welfare services and to sustain the size that the Family Grant reached, the real cooperation of the federal entities was necessary. The CIT provided the space for the agreement that made this cooperation possible. Without the connection between the CIT, the Cnas, and the MDS, the results would not have been possible in the given time.

We can infer from the processes involved that all conditions as a whole were sufficient for personnel hiring and the construction of facilities at the local level. Separately, each of these factors and processes would not produce the result, but all were necessary to produce it, making the whole a sufficient condition. Having an institution responsible for the coordination of policy at the national level, with a bureaucracy specializing in the MDS ambit is a precondition for inducing behavior at the local level. However, given that the policy in question is characterized by different “deliveries”, in this case, two great pillars - money transfer and social welfare services and benefits - a learning process over time was needed in terms of how to induce the municipalities to build bureaucratic capacity, that is, to establish public facilities and to hire staff for income transfer service provision and realization.

Figure 7 Source: Elaborated by the authors

Furthermore, we argue that this induction could not be sustained if it were simply a top-down process. In this case, the centrality of the agreements in the CIT were central in creating legitimacy and, therefore, in achieving the signing up of autonomous entities. These factors did not occur at random; there was a concerted action by actors involved in the bodies (the MDS, the Cnas, and the CIT). The relationship between them and its sufficient effects on the assembly of a bureaucratic-administrative structure at the local level can be observed in the figure below.

Figure 8 Source: Elaborated by the authors.

In summary, we reiterate what Evans (2002) proposes regarding the need for the institutionalization of democratic public decision-making processes involving the different actors concerned to achieve the desired result - in this case, to offer services that require the hiring of human resources. The transformative capacity of the MDS is dependent on an effective dialogue between its bureaucrats, here, more specifically, those of the Snas, the Cnas, and the CIT. These were deliberative processes within the CIT that politically sustained cooperation.

The choice of social welfare policy as a case study to verify the transformative capacity of the Brazilian state has proven very favorable, and the process tracing method has allowed us to trace the mechanisms that were used by the MDS in this process, building a causal chain. The institutional legacy of the MDS demonstrates the fragility of its structure at all levels of government, with a low administrative capacity in general. Added to this is the fact that it is structured within a national system, which links the three government levels and aims to offer action in a standardized manner, reaching users equally across the country. This indicates that, to some extent, there need to be common standards to guide the actions of municipalities.

The study of the MDS’s transformative capacity allowed us, first, to verify the importance of both organizational and institutional learning in the Ministry, the municipalities, and the CIT in this process of the constitution of state capacities. It is possible to observe, by means of the decisions expressed in the regulatory acts, that the Ministry itself has learned in the process, reviewing rules when necessary and adapting them to the specificities that were revealed in the process of implementing the regulations themselves, such as in cases in which sign-up deadlines were revised so that municipalities could adapt to the rules. This learning process can also be found in the manner itself in which this induction occurred: if, initially, income transfer and services were handled separately, throughout the process, these actions came together, creating regulations that integrated them, particularly in their provision, at the municipal level.

It is also relevat to emphasize the importance to the construction of the MDS’s transformative capacity to the end of the competition between the Snas and the Senarc, each with its specific learning. The need is clear for the technical and operational strengthening of the Snas, on the one hand, and the centrality of the political debate in the Senarc, on the other hand. This learning is reflected in the reduction in the number of ordinances issued by the MDS regarding social welfare and the simultaneous increase in CIT resolutions in the same period, showing that, to some degree, the arena of political debate grew in importance in the discussion of management and funding issues, which also indicates the increasing political power of the federal entities in this process.

The purpose of verifying the transformative capacity of the MDS, using its ability to induce the behavior of municipalities, particularly in regard to the construction of their own administrative capacity, as a parameter, allowed us to identify two important results. The MDS, through various normative acts, which have mainly (but not only) included selective incentives relating to federal public funding, was able to induce municipalities to sign up to the PBF and the Suas and to take on important aspects of the policy’s management at the local level. On the other hand, once they had signed up, municipalities built bureaucratic structures that allowed them to manage income transfer and offer social welfare services and to effectively perform the actions that fell to them following their signing up. The actual existence of public facilities at a municipal level is indicative of this construction, in addition to the increasing the number of professionals (although many with fragile labor relations) and the significant increase in the number of families enrolled in CadÚnico and beneficiaries of the PBF. Finally, it should be noted that the transformative capacity of the MDS, verified by means of the construction of the municipalities’ bureaucratic structure, has been exerted to mobilize consent and to institutionalize cooperation, demonstrating the MDS’s central role in this change. In conclusion, the centrality of the ordinances as a tool to induce this process and, in particular, the centrality of the policy’s federal government funding capacity is evident. However, it can be observed that funding alone would not be able to create this agreement, particularly in regard to management and services; in this respect, the role of the agreement bodies was crucial. Agreements were reached in the CIT that guided many of the analyzed ordinances.

By analyzing the Brazilian case of social welfare policy, it is possible to verify that the transformative capacity of the MDS was not the result of an intervention but of a concerted action, in which the federal government played an effective role in coordinating the change in this arena at a local level; however, this was only possible due to the agreements that were reached in the federal space for linking and interaction, the CIT. That is, this concerted action is the consolidation of what we call transformative capacity, derived from the organizational learning within the MDS and its institutional learning and that of other agencies involved in this process. This led to the construction of municipal bureaucratic structures, reflected in the creation of public facilities and the hiring of staff.