Public Policy Networks and the Implementation of the Bolsa-Família Program: An Analysis Based on the Monitoring of School Attendance

The paper analyzes how street-level bureaucrats construct and activate the intersectoral network induced by the implementation of the Bolsa Família Program (BFP) in a region of extreme poverty in Brazil. BFP is a federal cash transfer program with conditionalities, benefiting 13.8 million families. Based on the educational conditionality, BFP is articulated with other social policies, using beneficiaries, schools, health units, and social assistance centers as vectors that mobilize public organs, processes and agents, triggering complex and dynamic multi-sectoral and intersectoral social Education Policy Analysis Archives Vol. 28 No. 120 SPECIAL ISSUE 2 networks. Thirty-one multilevel public agents were interviewed, including 14 education professionals in two schools and eight social work professionals in two local centers. The analysis reveals that street-level bureaucracy emerges as hubs at the local level, indicating the social protection network in the territory depends on personal links and mediations among agents, institutions, and public equipment materialized by processes derived from the monitoring of school attendance. The articulation and interaction between school and social agents reflect degrees of intersectoral cooperation, marked by different patterns of comprehension/translation of routines and strategies. The analysis explicitly highlights connections among (im)material elements of the networks induced by the BFP, emphasizing their potential for understanding the implementation of social and educational policies.

networks. Thirty-one multilevel public agents were interviewed, including 14 education professionals in two schools and eight social work professionals in two local centers. The analysis reveals that street-level bureaucracy emerges as hubs at the local level, indicating the social protection network in the territory depends on personal links and mediations among agents, institutions, and public equipment materialized by processes derived from the monitoring of school attendance. The articulation and interaction between school and social agents reflect degrees of intersectoral cooperation, marked by different patterns of comprehension/translation of routines and strategies. The analysis explicitly highlights connections among (im)material elements of the networks induced by the BFP, emphasizing their potential for understanding the implementation of social and educational policies. Keywords: public policy networks; policy implementation; educational policy Redes de políticas públicas e implementación del Programa Bolsa Familia: Un análisis basado en procesos derivados del monitoreo de la asistencia escolar Resumen: El artículo analiza cómo los burócratas de la calle construyen y activan la red intersectorial inducida por la implementación del Programa Bolsa Família (PBF) en una región de extrema pobreza en Brasil. PBF es un programa federal de transferencia de efectivo con condiciones, que beneficia a 13.8 millones de familias. Basado en la condicionalidad educativa, PBF se articula con otras políticas sociales, utilizando beneficiarios, escuelas, unidades de salud y centros de asistencia social como vectores que movilizan órganos, procesos y agentes públicos, desencadenando redes sociales multisectoriales e intersectoriales complejas y dinámicas. Se entrevistaron 31 agentes públicos multinivel, incluidos catorce profesionales de la educación en dos escuelas y ocho trabajadores sociales en dos centros locales. El análisis revela que la burocracia a nivel de la calle emerge como hubs a nivel local, lo que indica que la red de protección social en el territorio depende de enlaces personales y mediaciones entre agentes, instituciones y equipos públicos materializados por procesos derivados del monitoreo de la asistencia escolar. La articulación y la interacción entre la escuela y los agentes sociales reflejan grados de cooperación intersectorial, marcados por diferentes patrones de comprensión/ traducción de rutinas y estrategias. El análisis resalta explícitamente las conexiones entre elementos (im) materiales de las redes inducidas por el PBF, enfatizando su potencial para comprender la implementación de políticas sociales y educativas. Palabras-clave: redes de políticas públicas; implementación de políticas; política educativa Redes de políticas pública e implementação do Programa Bolsa Família: Uma análise baseada nos processos derivados do monitoramento da frequência escolar Resumo: O artigo analisa como os agentes implementadores de base constroem e ativam a rede intersetorial induzida pela operacionalizaçãio do Programa Bolsa Família (PBF) em uma região de extrema pobreza no Brasil. O PBF é um programa federal de transferência de renda com condicionalidades, beneficiando 13,8 milhões de famílias. Com base na condicionalidade educacional, o PBF articula-se com outras políticas sociais, utilizando beneficiários, escolas, unidades de saúde e centros de assistência social como vetores que mobilizam órgãos, processos e agentes públicos, desencadeando redes sociais multissetoriais e intersetoriais complexas e dinâmicas. Foram entrevistados 31 agentes públicos multiníveis, incluindo quatorze profissionais de educação em duas escolas e oito profissionais da assistência social em dois centros locais. A análise revela que a burocracia no nível local emerge como articuladora na ponta, indicando que a rede de proteção social

Introduction
In Brazil, the origin of the Bolsa Família Program (BFP) is in the several programs financed and implemented by the municipal governments along with the nineties, ultimately resulting in the cash transfer federal program adopted by the government of Fernando Henrique Cardoso, named Bolsa Escola Federal Program. Inexpressive in terms of budget and coverage, it has been enlarged, redesigned and reinforced by the government of Luis Inácio Lula da Silva, since 2003. Kept by Dilma Rousseff and by Jair Bolsonaro, elected the President of the Republic in 2018, it is the social program with the most extensive coverage in the country. The BFP covers 4734 municipalities of the federation, with approximately 14.2 million of families assisted (Brasil, 2020).
The BFP is a cash transfer program to target families under the poverty line. It incorporates, in its architecture, a set of conditionalities that must be accomplished by the beneficiary families. As outlined in Ordinance No. 551/2005(MDS, 2009, conditionalities are the counterparts in the areas of education, health and social assistance that must be met by the families so that they can continue to receive the allowance. There are three conditionalities in the BFP: (1) keep school-age children and adolescents attending school; (2) comply with the primary healthcare for children from registered families through the vaccination calendar for minors between zero and six years old and (3) prenatal and postnatal medical appointments for pregnant and breastfeeding women 1 . The amounts paid vary from R$32.00 (32 reais -about US$8.00) to R$306.00 (306 reais -US$77.00), according to the monthly income (individual and familiar) with the number of children and adolescents up to 15 years-old and youth aged 16 and 17 years-old.
At the central level in the federal government, the coordination falls on the Ministry of Citizenship (former Ministry of Social Development/Ministério do Desenvolvimento Social -MDS 2 ). At the local level, the BFP is managed by the Secretariat of the area in which it is located (Social Assistance, Education, Health and Planning, among others), according to the determination of the local executive branch.
1 When referring to conditionalities, our intention is basically to inform the reader about one of the pillars of the BFP. Although it is not the focus of the article, it is worth mentioning that this is a controversial issue. Institutionalist perspectives reinforce the thesis that educational conditionalities contribute to the increase in attendance rates and, consequently, to the performance of the students attended. On the other hand, empirical studies on the implementation of Cash Transfer programs have revealed that such causality is not necessarily verified and that the implementation can produce perverse social effects, increasing inequality and reinforcing social exclusion. 2 The research was carried out before the ministerial reform of 2019. Therefore, we will use the acronym MDS to describe the actions within the scope of the federal government.
In Belo Horizonte, the BFP is housed in the Municipal Secretariat of Social Policy (Secretaria Municipal de Política Social -SMPS) and the management of the school attendance binds Education and Social Assistance Secretariats. The management of the program is made in three hierarchical levels, with the participation of the sectors involved in each of them, horizontally. The Inter-sector Chamber of Social Policy (Câmara Intersetorial de Política Social -CIPS), comprising the secretary of each of the areas involved and the sub-secretary of each regional administrative unit, is responsible for the inter-sectorial cooperation. Some policies, as the BFP, had thematic groups at the Chamber. The Thematic Group of the BFP had representatives, at the strategic level, from the Municipal Secretariat of Education, Municipal Secretariat of Health and the Municipal Secretariat of Social Assistance as members. On the second level, there was the Inter-sector Regional Center 3 (Núcleo Intersetorial Regional -NIR), which gathered the regional managers of each of the Secretariats. Each one of the nine regional administrative units had its NIR. At the local level, the articulation was made by the Local Thematic Groups, formed by the managers or directors of the public bureaus based in the territory (healthcare centers, social assistance centers, schools, and community centers).
Although the BFP is a social policy, the school attendance demand links the social to the educational policy, producing a fluid movement between these fields. When the student is at school, he/she becomes the subject of the educational policy, induced by the social policy. Hence, the BFP articulates with other social and educational policies, using the beneficiaries, schools, health units and social assistance centers (CRAS 4 ) operating at the local level as vectors that deploy, from a bottom-up perspective, several entities, processes and public agents, driving complex and dynamic multilevel and inter-sector social networks.
Instead of understanding social networks with an exclusive focus on the relations among single actors -individuals or corporations -with effects on the policies, we understand its connections from a multifocal perspective, anchored in the formulations developed by Latour (1997): human and non-human actors (regulatory frameworks, management technologies and public equipment) build network policies 5 .
In the BFP, a technology that promotes the articulation of the actors -and the policies -is the process of monitoring of school attendance, linked to the education conditionality. The attendance (and the processes derived from it) is a technology that links the school, the CRAS, the beneficiaries and the public actors in the territory to the ministries and the bureaucracy at the federal level, responsible for the design, follow-up and financial transfers at the Union level to the beneficiary families, characterizing the inter-sector and multilevel action.
This article analyzes the implementation of the BFP in a peripheral area of extreme poverty located in the city of Belo Horizonte 6 . Based on the dynamics generated by the control of school attendance, it highlights how the program creates and drives a set of local practices that give materiality to the policies, and are mediated by human actors and technologies (non-human actors). We assume that these sets of practices involve aspects of management, but also visions, values and meanings assigned by the managers that operate at the local level -the street-level bureaucratsrelated to the BFP and that do the mediation with the beneficiary families. That dynamics is not free of conflicts, tensions or personal sharing and relationships that end up by setting disputes that shape the results of the policy in the territory. The local practices are, therefore, spaces for disputes that contribute to the implementation and transformation of the policy, as well as for its results. As the practices depend on the connections and actions of the actors in the territory and those connections are measured by technologies, the network approach used must be capable of placing at the same level of importance human (their decisions, actions, visions and relations) and non-human actors (technologies, routines and the territory).
Considering the local practices, the management processes, and the routines linked to the monitoring and control of school attendance, this article also intends to analyze how a network is configured, taking into account the territorial dynamics of the implementation of the BFP through the practices of the actors involved. Assuming that these routines allow tracking flows and the articulations that the BFP causes, the article describes this dynamic from the perspective of a network.
Usually, the role of technologies as articulators of the networks is neglected in the analysis of social networks. We start from an extended definition of technology that approaches the overlapping among technical, social, political and economic issues. From the contributions by Law (2002) and Labatut, Aggeri and Girard (2012), we assume that technologies present both material aspects linked to their functions, and immaterial aspects, associated with the meanings given by the actors that create and use them. In this regard, they help to establish connections among the actors that comprise the network.
In order to organize this discussion, this article is structured in five sections, besides the Introduction and final considerations. The first section theoretically discusses the network approach from the perspective of the Actor-Network Theory (ANT). The section analyses four elements that are considered fundamental to the construction of the network and its implications to public policy implementation: technologies, routines, street-level bureaucrats and the territory. The following section presents the methodological strategies. In the next three sections, the results are presented and analyzed. It begins with the presentation of the BFP management processes at the federal level and in the city of Belo Horizonte, indicating the different instances linked to the program at the two levels of government. Then the focus of the analysis falls on the network (and its flows) that materializes from the controls and routines derived from school attendance. Finally, the fifth section deals with the challenges and difficulties that street-level bureaucrats linked to school attendance monitoring routines face in the policy implementation process.

Networks and the Implementation of Public Policies
Policies networks have been approached as a metaphor, an analytical tool, or as a method in order to analyze the social structure. Despite their different approaches, the studies of public policies through its networks have in common the understanding that various actors (besides the state and expertise in building local social welfare network. Thus, the city has been considered a reference in the management of social policies, sharing its expertise with other states and municipalities, besides the federal government itself. the government) are connected for the formulation and implementation of the policies. They also seek to comprehend how the linkages among them are established (Börzel, 2002;Fleury, 2005).
The network is a multiple concept that can mean both a space of connections (relational aspect) and a space of coordinates (structural aspect), in addition to a set of mathematical techniques from graph theory. It has been used for purposes as varied as descriptions about scientific cooperation, environmental issues, digital networks, political behavior, among others. The network approach we use in this article derives from Actor-Network Theory 7 (ANT), which focuses on the movements, flows, changes and actions of the actors who build the network (Venturini, Munk & Jacomy, 2018). These create a continuous movement of (re)construction within the network, opening up the possibility of social transformation as an action is carried out collectively.
One of ANT's main distinctive features is that the network and, hence, collective action, is not built only by individuals, but also by non-human actors (such as organizations, natural elements, technologies and even scientific concepts or laws). Next to individuals, the non-human actors are equally important in the description and formation of a network. The reason is that everyone can influence the course of action as long as they promote links among actors or are enrolled in these links. According to Fenwick and Edwards (2011), the core of ANT lies in describing what the actors do, not what they are or mean.
In this perspective, networks are also characterized by the fact that it is the network itself that defines its action frameworks. Then, the parameters used by analysts are defined within the network. There is no pre-established social order with social categories whose functioning can be defined as a priori or by the assumptions of analysts. ANT, according to Latour (2006), is a sociology of associations. The way associations among actors are established in the network and defines its scope and effects. In other words, the social is shaped within the construction of the network by the actors' agency. Thus, to understand a network, it is necessary to follow its actors and their ties (Latour, 2005).
Therefore, ANT describes practices, relations, and qualitative characteristics, besides clarifying the disputes that occur among actors in a territory and their consequences on the implementation and the results of a policy (Law, 2002). According to Latour (2006), the actor is always a network-actor because he is formed from the association with other actors, sharing meanings, negotiating interests and changing his subjectivity and positions.
Following the approach proposed by ANT, we consider that the constitution of public policy networks occurs through the interrelation of four elements: technologies, routines, human actors and territory. These elements promote the crucial links to the implementation and the achievement of results by public policies. In the following paragraphs, we will discuss these elements and how each of them is related to network formation.
We understand technologies as "a nexus of assumptions, rational myths, belief systems, hypotheses and material constraints which stem from broader institutional forces, intervene in the building of patterns of actions, and open new performance possibilities and inventions" (Labatut, Aggeri, & Girard, 2012). Technologies can be understood through their functions, but also through the meanings given by those who create and use them. Also, they can be understood from the different links of actors that they mobilize. Thus, technologies emerge as nexus among human actors, organizations and institutions for the realization of the public action and contribute to the configuration of the network in the territory, where public policies effectively materialize. Regarding public policies, technologies take a fundamental role, as the policy itself standardizes the instruments that will be used for its implementation and monitoring, defining the routines that makeup policy management.
When implementing policies in the territory, actors establish processes, mobilize resources, thereby institutionalizing, obeying and transforming instruments and norms. According to Feldman and Pentland (2003), these interactional processes usually occur in routines, patterns of repetitive and recognizable actions performed by multiple actors.
According to Feldman and Pentland (2003), routines have two aspects: ostensive and performative. The ostensive aspect corresponds to the structure of the routine, its script. The performative aspect corresponds to the specific actions of actors in a specific space and time; it is how the actors perform the routines constituting their practices. When actors use technologies and perform routines, they promote other actors' enrolment 8 and create the network's social fabric that will be responsible for public policy implementation in the territory.
In addition to technologies and routines, human actors also promote network linkages. A category of human actors already identified as fundamental for the implementation of public policies is street-level bureaucrats, public agents who work in the implementation process at the local level. For Lipsky (1980), street-level bureaucracy refers to the public servant that directly interacts with the citizens in the course of his/her job, and who also has some leeway to choose how to perform his/her work. For the author, these actors are public servants who grant access to government programs and provide services in these programs. Although there are considerable theoretical developments of street-level bureaucracy, 9 the focus of this study lies on its human actors, the streetlevel bureaucrats and their role in the network construction. Teachers, police officers, healthcare professionals, social assistants are some classical examples described in his research. For Musheno and Moody (2006), street-level bureaucrats are policymakers, since they are those who deal with the reality of the policy. Thanks to them the population manages to connect with the public administration. According to May and Winter (2009), all this control exercised by the implementing agents on the beneficiaries is related to the discretion embedded in the decisionmaking process. It is through this choice and decision-making capacity that agents can exercise control on the beneficiaries of the policy, as evidenced by Eiró (2019), to the case of the relationship between social assistants and beneficiaries of BFP. In many cases, what happens with public policyits success or failure -is influenced and, many times, determined by the discretionary acting of these agents and how they deal with it and its external influences.
This argument is also addressed by Dubois (2019). For the author, the relations of the petits fonctionnaires and their interactions with the public in the guichet reveal something at the macro level about the way of managing, carrying out public policies and governing society. The reforms and new content of public policies oblige the street-level bureaucrats to decide on uncertain bases and to expand their influence in the definition of public policies. Despite the uncertainty, this dynamic holds employees accountable and also the beneficiaries who are required to contribute to the efficient and effective implementation of public policies.
According to Hill and Hupe (2014), the implementation of public policies is a complex and dynamic process with different degrees of unpredictability. The decisions are made under the events that take place at the local level. For Oliveira (2019), discussing the implementation of education policies at the micro-level, the decision-making capacity and scope of action of the actors involved in the process of implementation at schools become essential to understand how the policy is put into action. The decisions made in these contexts may or may not be aligned with the educational objectives of the governments. Oliveira (2014) and Oliveira and Pena (2018) makes an interesting remark on the understanding of the discretionary power of these agents during the implementation process. According to them, the discretion exercised by street-level bureaucrats results from interactional strategies and patterns among them. In this perspective, Lessard and Carpentier (2016) affirm that the interpretation and translation of their values, the values of other actors involved, the procedures, restrictions, structures, incentives, encouragements, and prohibitions are crucial to comprehend the policy dynamics. Thus, it is necessary to examine interaction standards in order to understand how and why actions and decisions were made that way. The multidimensionality of street-level bureaucrats' discretionary decision and action can be discussed from Ball's (2010) study of performance and performativity, which can promote new social identities and professional subjectivities. For the author, performativity is a technology, a culture and a mode of regulation that occurs in a disputed "field of judgment." The question of who controls the "field of judgment" is crucial. In educational institutions, control takes both the form of rituals that naturalize discourses and routines that address identities. The issue, therefore, is to look at the relational and institutional environment in which those actors operate and how they build routines to meet public policy objectives. When street-level bureaucrats build routines, they promote the association of human and non-human actors, constituting the public policy network.
What can be understood from this discussion is that street-level bureaucrats are fundamental mediators for the construction of policy networks. When they execute the practices, they deploy other human and non-human actors, whose actions define, transform and configure the results of public policies at the local level and, simultaneously, at the macro sphere, at federal and state levels. The territory, then, is coextensive to the practices of the actors in the network. This also indicates that the microanalysis sphere, within the scope of the individuals and organizations in which they operate locally, is indispensable for the understanding of the macro sphere, within the scope of the policy and strategies of formulation and the design of the policies in the federal government.
In addition to technologies, routines and street-level bureaucrats, a fourth fundamental element in our discussion is the notion of network-territory. For Santos (2017), the interactions among actors, technologies and routines mobilization, (re)construction and results take place in the territory. The territory has its dynamics that are (re) configured by human and non-human actors and their relationships. Public action takes place in this territory that includes not only geographical limits but also social, symbolic and material relations. These relations outline a local reference experienced in its diversity, in its internal connections and connection with other territories (Raffestin, 1993).
In this sense, it can be said that the subjects, in their territories, given multiple connection possibilities, experience and build network-territories (Haesbaert, 2019). This network-territory coincides with the connections established by the actors. According to Latour (1997), it is possible to understand that the limits of a network depend on the way humans organize their experience, especially in their relationships with non-humans. When actors organize individual and collective experience (that is also historical, cultural, technical, geographical, environmental, and political), the routines defined by public policies are shaped and transformed in the territory, generating ways of implementation and results that are specific to that place.
From the previous discussion, we understand that, for the BFP, the monitoring of school attendance is the technology that engages education and social assistance professionals (street-level bureaucrats who work in the territory, according to Lipsky (1980) and Musheno and Moody (2006)). By following school attendance and the administrative routines that arise from it, it is possible to describe and analyze the links among human and non-human actors located at different levels of state management, such as municipal secretariats, state governments, ministries, computerized systems, indicators, among others. Therefore, the network promoted by these actors is intersectoral and multilevel.
The constitution of a network depends on the way actors use technologies, build routines and experiences, which are permeated both by formal objectives explained in the BFP official texts and by their values and meanings. This double characteristic of the BFP network formally explained in the text of the program itself and in the daily practice of the actors who work in schools and in the CRAS constitutes the dynamism of the network and its transformations. This dynamic can change not only the results but also the objectives of the program.

Method
In order to describe the practices of the actors in the network driven by the process of monitoring school attendance of students assisted by the BFP, the methodological procedures adopted were: (i) selection of the actor to be followed (Latour, 2005); (ii) network description/cartography of controversies (Latour, 2006;Venturini, 2010); (iii) documental analysis and collection of data on the architecture of the program and its implementation; (iv) informal interviews with agents in charge of the program at federal and local levels to define the analytical pre-categories (Minayo, 2012) and (v) preparation and test of questionnaires; selection of two municipal public schools located in an extreme poverty area; selection and interview with the implementation agents involved; analysis of the interviews and definition of categories (Abdal et al., 2016).
The analysis of the data on the BFP supported the preparation of the scripts for the interviews that were carried out. Three levels of governance in the program were defined: national and local strategic (MDS and City Hall of Belo Horizonte); managerial (sector managers at the municipal, regional administrative unit) and operational in the frontline (education professionals, social assistants and healthcare agents, in two neighborhoods of an extreme-poverty area of the city).
Semi-structured scripts for the interviews with the agents at strategic levels in the MDS and the City Hall were prepared to include the following categories: (1) perception of the program; (2) values and representations on the target audience; (3) perception on the actions of the governments; (3) institutional and operative design of the program; (4) impressions on the conditionalities of the program and the demands for agents at the local level and (5) the routines of the agents in the daily routine of the policy.
The scripts were adapted for different subjects, depending on their involvement and position in the hierarchy of the program. Thirty-one interviews were carried out, distributed as follows: (four) strategic body of the program -MDS; (two) strategic body of the program -City Hall of Belo Horizonte; (three) managerial body of the program -Municipal Regional Administrative Unit; (14) street-level bureaucrats that operate in the schools (directors, teachers and pedagogical coordinators in two schools); (eight) street-level bureaucrats linked to the social assistance in two CRAS. The following sections present the data analysis.

The BFP Network and its Processes: Design and Program Management at the Federal Level and in the City of Belo Horizonte
All actions linked to the BFP are financed and coordinated by MDS, created in 2003. This is an important point, especially as regards the process of implementation of public policies from the state bureaucracy viewpoint: one Ministry took over, institutionally, the function of financing, coordinating and articulating the social policies of the country. It is up to the municipalities, in turn, to implement, develop complementary socio-educational actions and monitor compliance with the conditions. They face, then, the decentralization dilemma: on the one hand, they are induced to implement local public policies; on the other, budgetary resources are not sufficient. Figure 1 presents the levels and areas involved in the federal, state and municipal governments, with an emphasis on the moments of transformation that the policy suffers every time a change of level takes place, up to the local level. Besides being the holder of the budget fund of the BFP, the federal Executive is responsible for many key-processes: defining the characteristics of the benefitted families, strategically managing the information contained in the Single Registry 10 (Cadastro Único -CADU) and managing the structure of incentives to local governments through the Decentralized Management Index 11 (Índice de Gestão Descentralizada -IGD). That architecture permits some considerations: when transferring the funds directly to the families through the bank system, the federal government considerably reduces the costs of operation. As the money intended to these families does not go through state and municipal budgets, waste, deviations and, consequently, corruption is avoided. On the other hand, focus, transparency and accountability, which are strategic requirements for the implementation of public policies, are increased. Figure 2 presents the institutional arrangement of the BFP in Belo Horizonte. As previously described in the Introduction, the BFP is managed by the Municipal Secretariat of Social Assistance, according to the determination of the local executive branch. Besides being responsible for the political interaction with the Federal and State government, the Secretariat, through the Inter-sector Committee, articulates the central, regional and local spheres in the city. At the regional and local level, the Regional Center coordinates the areas, public bureaus and street-level bureaucrats involved in the operation -social assistance, education and health.

The Network Materialized from the School Attendance Monitoring in Belo Horizonte
Besides articulation difficulties, which are understandable because the culture of policy networks is still fragile in Brazil, the interviews analyzed point out that the issues related to the monitoring of the educational conditionality seem to make the implementation more complex, tensioning the relation among the areas. Based on the dynamics that originated from the following of the routines of school attendance monitoring, Figure 3 represents the BFP network described in this section.

Routines and flows from the central level: The relationship between the Secretariat of Education and Social Assistance
Belo Horizonte had its own local cash transfer program before the BFP, which was linked to the Municipal Secretariat of Education (Secretaria Municipal de Educação -SMED), the city had already implemented some specific routines regarding school attendance. These processes were already under the responsibility of a management unit, which, accordingly, was responsible for following up the beneficiary families. So, there were several instruments developed for that purpose, which also served as a reference for the BFP at the beginning of its implementation.
With the creation of the BFP, because of the know-how acquired with the municipal program, the management started to accumulate these procedures for the federal program, including the schools of the state and the private philanthropic ones. According to the manager interviewed, it had been decided, since 2009, that such sector would not only monitor the attendance of the beneficiaries but of all the students of the municipal network. In other words: the methodology used for the beneficiaries was applied for those that did not receive the cash transfer program. Initially, a survey was made by the School Management System (Sistema de Gestão Escolar -SGE) to detect the non-attending students. Once this was done, efforts were made to understand the profile of these students. Depending on the situation, there was a routine of home visits.
The first tensioning point was in the monitoring of the school attendance in state schools. According to that manager, the pressure was being made so that the Municipal Secretariat of Education would concentrate on the students of its network, transferring the assignment to the areas of assistance in the state and the municipality.
The attendance control was a process full of bureaucratic procedures, which generated essential results. In general lines, it was up to the school to assess the attendance and feed the School Management System, linked to the Ministry of Education (Ministério da Educação -MEC). Based on that system, MDS would assess the number of families that were in an irregular regarding the education criteria. Once that was done, each municipality received a list of the beneficiaries that should justify their absence. In parallel to that, each family received a notice from the MDS and, then, they should present an appeal, a process in which they indicated the reasons why their children had, in that period, absences above the permitted limit. The appeals were analyzed by the municipal sectors assigned and the result was entered in CADU, regularizing the pending issue (administrative routine). According to the situation, the family was suspended, withdrawn, kept under observation, or released. Data show that the attendance monitoring is essential to the BFP when joining local to global, promoting the mediation that permits the connection among actors that are located at several levels of the state bureaucracy. Part of the network may be so represented from a bottom-up perspective by the vertical flow of school attendance and its return to the local level for the cash transfer payment.
In Belo Horizonte, the Family School Program team was responsible for the management of the process. It was up to the sector to coordinate the actions in each of the regional administrative units of the administration, besides articulating the process of the resources related to the lack of school attendance in the CRAS. Despite the process being centralized on that management, the responsibility for judging the resources fell on CRAS (or to the Administrative Regional Units, in case the territory did not have any of such units). In other words: the school was not responsible for accepting or not the justifications.
As the CRAS was not part of the organizational structure of the Municipal Secretariat of Education and was at the municipal level, linked to the Municipal Secretariat of Social Policy, the process of the resources caused some tension between these Secretariats and the sectors directly involved in each one of them. As that was one of the most time-consuming administrative routines for the teams, which were generally small and with an insufficient number of public agents, the areas tried to escape their responsibilities. Concerning the education, the responsible management defended that its focus should be on the qualitative follow-up of the family, whereas the administrative routines should be assigned to the areas directly linked to the BFP in Social Assistance.
Manager: We have discussed a lot. For many years, [the Management] did a role that belongs to Social Assistance. We did, for example, that stuff of appeals. The families are entitled to file an appeal when the MDS notifies them, it was education that used to do all these appeals with the family. This is not the role of the school. In 2009, we started to share these appeals: "ok, we are going to do it, but Bolsa Família has to do it, and the assistance also has to do it." At least we shared it because there are many appeals. We have discussed that we do not have to deal with any appeal; the appeal is an administrative issue of Bolsa Família. We have to follow up on that family, which has a non-attending student, we have to see that issue of non-attendance and do the pedagogical procedures related to that attendance. There are cases in which we have to change the group, and there are cases in which we have to change shift, there are cases in which we have even to change the school. These are our issues of education. There are many cases in which the children go to school, but they do not remain within the classroom. This is a problem of ours. So, this is our part. Now, doing the appeal, which is an administrative instrument of the Bolsa Família program, this must be there, at Bolsa Família [referring to Municipal Secretariat of Social Policy]. Then, this is the conversation we are having.
The interviewee evidences the difficulty in making inter-sector connections in the horizontal dimension of the network. Even though school attendance demands the establishment of links among the municipal secretariats, the public equipment and their human actors, they understand their functions in a specific way. The appeal for the absence of attendance involves not only administrative routines but also touches the meanings assigned by the human actors to their work, as well as their role in the network. The meanings, the roles and the available structure (number of public agents) are negotiated among the actors, while they perform and (re)build their practices, even when these practices are already oriented by the model established in the public policy.

Routines and flows from the local level: The forums and the relationship of beneficiary families with the school and CRAS
The school is where the educational policies are implemented, regardless of its orientation (Federal, State, or Municipal). In Belo Horizonte, the Family-School Forums are important actors for the access to and debate of these policies within the scope of the family, seeking to promote a higher involvement of it with the educational policies. That understanding associates to street-level bureaucrats that are in the school, who are responsible for promoting the interaction between the BFP and other educational policies.
In the BFP, school attendance establishes the link of the beneficiaries directly to the school, at the local level. In case a student is non-attendant, the payment may be suspended, generating the administrative routine of appeal for the absence of attendance. This appeal is analyzed by the CRAS, at the local level. As the CRAS is responsible for registering the family in the system linked to MDS, and for home visits, the street-level bureaucrats connected to the CRAS make decisions on the justification presented in the appeal and, for that purpose, they also consider their knowledge about the families.
In order to analyze the appeals on school absence and other subjects of the BFP, both CRAS and school participate in the Regional Administrative Unit, which links municipal secretariats of Education and Social Policy and operates as a decision-making forum on the BFP. Similarly, as it occurs with the school, the CRAS is the public equipment in which the street-level bureaucrats' guide and follow-up the families on the other social assistance and health policies.
As a result of its high institutional capacity and the expertise accumulated, the Municipal Secretariat of Education, since 2009, was striving so that its actions focused on the family, understanding that that was the fundamental link so that the educational policies in the municipality were more effective and more visible. Thus, Municipal Secretariat of Education held, on an annual basis, three Family School Forums, arenas of participation in which the families could gather to debate, in an extended manner, the issues directly and undirectedly connected to the education.
For the forums to take place, a certain number of participants were deployed per school, families' transportation was paid, and workshops were organized for small children and children over six years so that there did not impede the participation. At these opportunities, the discussions concentrated on educational issues: child education, integrated school, evaluations in the municipal network, among others. According to one of the interviewed managers, these spaces were important arenas because they stimulated the debate, put the citizen closer to the public authorities and potentiated changes in the design of the policies.
Manager: (…) it is there that they claim, there they complain, there they praise, put their issues. Moreover, our ombudsman has always been present in this Forum, manager, to respond to any issue as regards the regional unit that may appear. The family that does not go to the microphone can write, the ombudsman has ten days to answer to that family. It is an exciting process of dialogue with that family. Some issues of the education policy of Belo Horizonte started from the issues placed by the family, such as, for instance, the school bulletin, that did not exist, it started to exist after the demand in the Forum.
Even though that was a local initiative, not only made for the families benefitted by the BFP, the Forum brought with it two critical questions: (i) making the beneficiary families more committed to their children's schooling, and (ii) creating mechanisms to get the family closer to the state and, consequently, to make them know other social policies and have access to them. In other words, such Forum emerges as one of the strategies adopted by the local government so that, through the Municipal Secretariat of Education, it can drive the social protection network and link the public schools that operate in the municipality to the citizen and vice-versa. When taking part in the Forum, the families served by the BFP started to know the actions and policies implemented by the municipality, among which those focused on the promotion of the education and its various dimensions. Thus, the forums stimulated the inter-sector nature of the program, approximating the social dimension of the educational sphere, induced by the conditionality of the school attendance.
As these families did not have how to move towards the state, they waited, sometimes passively, for any initiative by local administrations. In this regard, the design of the BFP depended on the involvement of several sectors, equipment and actors at the local level. Once schools and CRAS were spaces institutionally linked to the families by the school attendance control, such bureaus were expected to develop strategies of action or approximation with the targeted citizen, which, according to one of the social assistants interviewed, was one of the main challenges to provide inter-sector actions to the families.
The interviews held with the managers indicate that Belo Horizonte was trying, with difficulty, to fill that gap. According to one of the interviewees, there was an effort so that the beneficiary families knew what social policies the municipality and the federal government offered, supported by the CRAS. At the same time, the school professionals held meetings to present the Pedagogical Project and the School Regulation to create ways of dialogue and stimulate parents' responsibility and interest in the performance of their children and youngsters. They also wished the parents to know the educational projects and programs developed by the Municipal Secretariat of Education. As regards education, the schools were trying to get closer to the families served by the BFP, inserting them into discussions and routines and making them, together with the enrolled students, subjects of the educational policies. This is significant evidence in the sense of reinforcing the thesis that the program transits between the fields of education and social assistance, articulating, in a network, the actors and the spheres that are driven by the school attendance.

The Management of School Attendance in the BFP: Challenges, Dilemmas and Difficulties Faced by Street-level Bureaucrats in the Territory
In general, it was noted that the design of the BFP contained some loopholes or vague instructions, which permitted or potentiated different interpretations. Although the street-level bureaucrats occupy lower hierarchic positions in the implementation flow, the proximity with the beneficiaries potentiated the spectrum of decisions. Besides, the lack or impossibility of administrative control and systematic observation or follow-up of the work of these agents contributed to their empowerment. In the analyzed network-territory, the street-level bureaucrats developed or accumulated expertise along the time, adjusting to the existing routines and developing operation practices in order to perform the work better or avoid embarrassments imposed by the institutions to which they were submitted. The connection of actors helped explain how discretion was exercised.
Besides containing vague instructions that were hard for agents to understand, they presented a partial lack of knowledge of the BFP rules, because they have not been properly and sufficiently trained or because the staff is changeable in the organization. Thus, the street-level bureaucrats interviewed combined several strategies for their tasks to be completed. In the interviews made, it is found that when making decisions, they created rules: CRAS Coordinator A: There are five sanctions. The warning does not generate anything for the family, it generally receives a letter from the federal government, stating that there has been a non-compliance for attendance and it is being warned. After that, the family has the allowance blocked, and then it starts to find it odd. Then it moves. Then there is the second suspension and the cancellation. When we receive the list that comes with 80 to 200 families, some months it happens to come with more than 300, we filtrate. First, we filter the families that CRAS knows, because we have been in this territory for a long time, and there are some families that CRAS knows and some families that CRAS follows. Then we separate the families that CRAS follows and check whether they are noncompliant. Then we check those that are registered in CRAS to warn them to come to CRAS. Then we send the letters. However, we send the letters prioritizing the families that are in the cancellation. Moreover, this is a challenge for us because, in that universe of 150 families, we cannot read, which are the most vulnerable families.
Reports are indicating that the street-level bureaucrats made decisions according to what they believed, their values and beliefs, exceeding the limits many times imposed by the lack of resources. Regardless of the complexity of what they decided, their decisions had direct implications on the lives of the people who were served by them. The beneficiaries were classified according to the decisions of the street-level bureaucrats. In the interview, the school director explained how her decision was influenced by the information she had on a certain mother or for her knowledge on her profile. So, she suggested that the chances of accepting the absence justification presented by mothers considered severe and committed to the education of their children were higher. Similarly, the CRAS social assistant selected the families to fill the vacancies made available by the Administration of Belo Horizonte, considering the remembrance she had of such people. Thus, she seemed to favor the families that were more present, who showed up more frequently at the CRAS.
The interviews also indicate that the administrative routines were the basis of the practices, even though the actors performed them in different ways, as in the case of the service reports for registration of the families. These practices were influenced by functional roles, by volume of documentation to fill, by the time dedicated to service and registration was shared with other activities performed both at school and CRAS.
In general, these agents perceived their profession from the front line activities viewpoints, such as teaching (teachers), making visits and home services (social assistants) and caring for a patient (healthcare professionals), because they consumed more time, were tiresome, repetitive, strenuous and deviated them from the attributions that gave visibility to their work and because they did not recognize their importance in the macro-level bureaucratic system. Thus, the meansactivities were, as a rule, less prestigious and even relegated as secondary or disregarded.
An additional element must also be considered. As in democracies, governments have limited political cycles, with alternance of power, changes in the partisan coalitions and constant redefinition of agendas might create, redesign, or extinguish public policies. This ends up exercising influences on the practices and routines. When school secretary affirmed "things change fast at the Municipal Secretariat of Education," it seems that she was referring to that issue: School secretary: Things change a lot; they change a lot! The municipal Secretariat of Education of Belo Horizonte constantly changes the strategies. Currently, we have people more specific at school that care for the students' attendance through "BH at School." So, I work with them, and as they do for BH, and I do for Bolsa, then I am behind them.
According to the above statement, it is observed that, again, school attendance enters the scene making new practices arise when establishing new connections. Difficulties also arose when the routines of CRAS and school professionals crossed because of the education conditionality. As the school was responsible for the information on absences and the CRAS was in charge of processing the resources generated, the lack of coordination between these processes compromised the qualitative services to the families, reinforcing its bureaucratic dimension: CRAS Social assistant A: In the BFP routine, I confess that it is a challenge for us because there is no built legitimated flow so that we have information from the school of the child that has non-attendance. Because the child's non-attendance at school comes to us two months after it happened. Then we are doing an appeal related to Bolsa Família in June, related to an absence in February and March. It is as if we were always running behind something that has already lapsed. Currently, if we are to think about the routine of the school today, I cannot know who are the non-attendant children today, I will know it two months later.
The street-level bureaucrats in the network-territory noticed the tensions described between both municipal secretariats. The questioning of the practices involved the distribution of functions, roles, and meanings: the social assistants claimed that the CRAS was responsible for analyzing the appeals to non-attendance to school, once, for them, the qualitative follow-up of school attendance of these children should be an assignment of the school. Besides, they affirmed that the follow-up of the attendance produced consequences linked to the daily routine of the school, such as issues related to the students' performance, the relationship with the school and the links with the families.
CRAS Coordinator A: If the conditionality refers to education and it is only an attendance, I think that the school is not very attentive to whether the child is attendant or non-attendant. Then I cannot say whether at school there is anyone attentive: John Doe receives Bolsa Família and this month he was absent twice, what is going on? I think there is no such thing. School attendance is something bureaucratic. It is evident that they are compiling a list of children with lower attendance and are taking actions. I believe that the school does this. However, attendance itself is not enough; it says nothing about the student's performance, the school value for him. It says nothing about the learning process. If he/she produces, if he/she has access, in short, the conditionality of education says only about the insertion of the child in the school or not. That is the point when there is a knot for assistance because it is an attribution of the assistance to follow up the school attendance. Because that family is not assuring the right to education, which is a duty of the family. If the appeal of Bolsa Família were made at school, maybe it would be different because it would be in the school that that family would appear instead of appearing at the assistance. Furthermore, it brings things that refer to the educational system. When the mother comes and says my child no longer wants to be at school, she does not like the school. It is not a specific work of the assistance, and I may try with the mother. However, I have to understand who that child is within the school.
On the other hand, the school did not relinquish the responsibility for analyzing the justifications that the families presented before the data being entered into the management system. If the student was absent, the guardian(s) should inform the reason to the school office, which decided whether to accept it or not. Once this was done, the sector entered the absences in the system and informed whether they were, or not, justified, using the corresponding codes. Justified absences did not generate suspension or withdrawal for the families. The interview with the school secretary showed that: [E1] School secretary: This student is ours. Moreover, he is here with us and you, whether you like it or not, know the life history of this child. Many times, you know, dad and mom, uncle, dog, and cat. So, no one knows better than us what is happening. Whether something can be justified or not. Then, this was made clear to us at the last meeting. Moreover, look, I've been in the office for four years and this was the first time I heard that. So, for me, it was more evident now.
[E2] School secretary: We monitor the students' attendance, and then, depending on the number of absences the student has, we enter it in the system. This is online, and we need to be monitoring the justifications for the absences and the medical leaves of these students. Moreover, all justifications and leaves I have to send to Bolsa [in the regional unit]. If the person does not have any justification, then we do not send anything, and then there are some functions they apply depending on the seriousness of the situation of the number of absences. It is entirely up to the school justifying or not. What they explain is that the school's view is going to determine whether that will be justifiable or not.
The considerations made suggest that the work of these agents was intricate, demanding, hard, tiresome, and, sometimes, insecure. There were many interviews exposing challenges and dilemmas of the daily routine. The CRAS coordinator, for example, shared the anguish she felt when dealing with complex situations in the territory in which she worked, the difficulty to motivate and coordinate the work team and the demands that the exercise of the profession imposed, against the tendency towards accommodation induced by routine. Besides, the reduced number of employees, the insecurity of the territory of work, and the denounce related to fraudulent receiving of the allowance was contextual factors to influence the practices related to the BFP and to cause concern in the team.

Conclusion
By following school attendance and its routines in the city of Belo Horizonte, the analyzed data reveal that the implementation of the BFP induces intersectoral cooperation, articulating the fields of social assistance and education. The BFP, through these routines triggered by the school and the CRAS, links these two fields in the territory.
With this analysis, it is possible to describe the network performed by the actors, that is, the network that is established based on the technologies and practices performed in the territory. Thus, we demonstrate that the implemented routines modify the official network expressed in the policy formulated at the federal level. In this sense, although the BFP is a national program, its results are constrained by local characteristics given the role that municipalities play in the implementation process.
It is important to note that the policy network presented in this paper is one of the other possible networks. Since the analysis made was based on following school attendance control, this configures specific flows that emerge as responses to the challenges faced in its operationalization. Thus, the choice of which actor to follow is fundamental to the description and analysis of the network, as expressed by the Actor-Network-Theory. The contribution of this approach to the study of public policy networks is evidenced both by the understanding of how non-human actors (through mediation and linking) induce the network and by the possibility of understanding the flows from the enrolment and the performances of the actors in the territory.
With the approach presented in this article, we propose a theoretical-methodological path for the analysis of public policy networks taking into account the Actor-Network-Theory. The fundamental aspect is the choice of the human or non-human actor to follow. The actor (or actors) to be followed must be able to enroll others and allow the tracing of flows that can express the links among the formulation of the policy, its implementation and its results in order to explicit its multilevel and intersectoral relations.
A second aspect concerns the elements of analysis and their performance in the constitution of the network: technologies, routines, street-level bureaucrats, and the territory particularize the implementation of policies and their results. What (or who) makes up these elements cannot be established a priori, but based on the structured relationships in the territory and the flows that arise when following the actor (or actors) initially defined. The connections that emerge from school attendance reveal actors and routines that configure a network-territory 12 that extends from the local to the federal level. Thus, following the actor provides essential data that can contribute to elucidate and understand the requirements inherent to the unfolding of the policy in the territory.
Concerning the data derived from the implementation process of the BFP, the interviews analyzed show that, even though the BFP is not officially an educational policy, it operates in a grayish zone between social assistance and education. The requirement to meet the school attendance causes the program to have a strong connection with education, its role in the process of formulating public policies, and the effects that the program potentiates and induces in the schools where the students are inserted.
The BFP is, therefore, a social policy that guarantees rights and allows families to have a higher income, increasing schooling opportunities and, consequently, reducing the intergenerational cycle of poverty. Hence, the implementation of the BFP depends on the official capacities of states and municipalities, particularly the latter. For the case of Belo Horizonte, a city that counts on its teaching network and official education policy, the effects of its implementation evidence the intersector relations and the social protection network that has been built in order to fight poverty in vulnerable territories.
Similarly, the guarantee of social rights depends on the organization's commitment and financial availability of municipal governments in promoting a social protection network, as Belo Horizonte did. When analyzing the implementation of the BFP in that city considering the perspective of school attendance, the network that emerges from the routines derived from there permits that the beneficiary families build a new comprehension of the notion of citizenship. The BFP has, then, the outlines of an educational policy that promotes school access because it links the granting of a minimum income to school attendance, making student rights effective.
At the same time, it can also have the contours of the educational policy of permanence because, when guaranteeing that income to the families, it is an incentive to keep students enrolled instead of abandoning the school towards the informal economy. School permanence is not, however, fully assured because it will only be reached when students have full conditions to attend and remain at school, which, necessarily, demands social and pedagogical policies that effectively promote such purposes.

About the Guest Editors
Emily M. Hodge Montclair State University hodgee@montclair.edu Emily M. Hodge, Ph.D., is an assistant professor in the Department of Educational Leadership at Montclair State University. She received her PhD from the Department of Education Policy Studies at the Pennsylvania State University. Her work uses qualitative methods and social network analysis to understand the changing nature of strategies for educational equity. Recent projects have explored how educational systems, schools, and teachers negotiate the tension between standardization and differentiation in the context of the Common Core State Standards, and the varied strategies state education agencies are using to support standards implementation.

Joshua Childs
The University of Texas at Austin joshuachilds@austin.utexas.edu Joshua Childs is an assistant professor of Educational Policy and Planning (EPP) in the Department of Educational Leadership and Policy. Joshua received his PhD in Learning Sciences and Policy at the University of Pittsburgh. Joshua's research focuses on the role of interorganizational networks, cross-sector collaborations, and strategic alliances to address complex educational issues. Specifically, his work examines collaborative approaches involving community organizations and stakeholders that have the potential to improve academic achievement and reduce opportunity gaps for students in urban and rural schools.

Wayne Au
University of Washington, Bothell wayneau@uw.edu Wayne Au is an educator, activist, and scholar who focuses on issues of race, class, and power in schooling. He is a professor in the School of Educational Studies at the University of Washington-Bothell, where he currently serves as dean of diversity and equity. Au is an editor of the social justice teacher magazine Rethinking Schools and the author or editor of numerous other publications, including Teaching for Black Lives, Rethinking Ethnic Studies, and A Marxist Education: Learning to Change the World.

SPECIAL ISSUE Researching 21st Century Education Policy
Through Social Network Analysis education policy analysis archives Volume 28 Number 120 August 17, 2020ISSN 1068-2341 Readers are free to copy, display, distribute, and adapt this article, as long as the work is attributed to the author(s) and Education Policy Analysis Archives, the changes are identified, and the same license applies to the derivative work. More details of this Creative Commons license are available at https://creativecommons.