Public policy analysis presupposes that results depend on inputs and that variations in types or amount of input explain the differences in results. The concept of capacity encompasses a set of inputs that are necessary for policies to be implemented and produce results and impact ( Centeno, Kohli & Yashar 2017; Matthews 2012). In theory, the greater the capacities of the State, the greater the effectiveness of public policies. The problem is how to know which inputs explain the results. Capacities can be individual, organizational, or systemic but there is no consensus in the literature on the specific attributes or their weighting ( Cingolani 2013; Painter & Pierre 2005; Wu, Ramesh & Howlett 2015).

The analysis of these capacities presupposes structural institutional conditions. Capacity analysis is based on macro-structural factors as determinants of results. If administrative and political capacities were identifiable, it would be sufficient to replicate them for effective implementation. The structural analysis does not consider that the individuals composing public organizations may or may not adhere to their objectives and perform functions they are assigned ( Williams 2017).

This structural analysis of capacities fails to consider individual factors of the work performed by public managers. Organizations are set of individuals who act on standards and rules, design policies, and implement them. In this process, interests, opinions and perspectives of these agents converge in organizational inputs for the implementation of policies ( Cyert & March 1963).

The purpose of this article is to analyze policy capacities by observing how these capabilities are distributed in policy functions and how they are taken into consideration in the Brazilian federal civil service. In other words, the research question to be addressed in this investigation is “What do policy capacities and policy functions reveal about the bureaucratic roles in the implementation of the Brazilian public policies?”.

This article is based on some assumptions. First, capacity analysis must observe the performance of bureaucrats in fulfilling policy functions, which differ across policy sectors. Second, these functions are performed in organizational contexts, in which individuals act based on choices and interests that delimit organizational objectives. Individuals matter in capacity analysis and the way they organize their tasks implies different patterns of organizational capacity building.

After this Introduction (I), the second section of the article analyzes the concept of capacity and policy work, in order to build up the framework for performance of policy functions. The third section debates the institutional context of the federal civil service in Brazil. In the fourth section, the methodology of the research and the results (section V) will be presented. In the sixth section the results will be discussed looking at the literature debate and, final considerations will be presented in the last section (VII).

The concept of state capacity means different things to different scholars ( Cingolani 2013; Jessop 2001). In one perspective, the concept involves the creation of a political order and degree of institutionalization of the state within a territory. This perspective defines state capacity as a set of norms and rules that make governments sovereign, backed by coercive apparatus, which is the basis for effective public administration ( Skocpol 1985; Tilly 1975). Thus, state capacity is related to the actions of the state to assert its autonomy, collect taxes, and manage conflicts in order to transform the economy and society ( Besley & Persson 2009; Knutsen 2013; Levi 1988).

In another line of analysis, the concept of state capacity is used to understand how the state apparatus produces results for society and for the economy through public policies ( Matthews 2012). This second perspective seeks to analyze how public policies and public services are effective to positively impact the intended outcomes of the state action. In contrast to the first perspective, the second perspective analyzes state capacities as delivery services and policies to promote development ( Evans 1995; Geddes 1996; Mann 1993).

Both lines of analysis of the concept of state capacities involve an understanding of the infrastructure of the state apparatus associated with a macro-historical construction. The concepts of state capacities are multiple and involve, in particular, a conception of the potential for. This means observing how the state apparatus result in better or worse public policies. They see analysis of capacity as a stock of skills and abilities that affect implementation of policies ( Centeno, Kohli & Yashar 2017). The concept of state capacities is related to the formation and stockpiling of skills and abilities of state bureaucracy to implement policies and build services. It involves the construction of an institutional trajectory of bureaucracies that may present more or less effective results.

That historical-structural conceptions of state capacity disregard the dimension of action and behavior within bureaucracies. The assumption is that state capacity is a predictor of implementation performance ( Williams, 2017). But we cannot disregard the central role of individuals in policy implementation and the complexity of bureaucratic organizations ( Cyert & March 1963; Lipsky 2010; Lotta 2010; Pires 2009). The conception of state capacity as organizational and institutional stock needs to consider the role of individuals in the context of bureaucracies. In this way, state bureaucracies are collective actors whose behavior is reflected in state capacity vis-a-vis the organizations of which they are part ( Cyert & March 1963). State capacity, therefore, depend on individuals’ actions in the context of complex organizational structures. It involves, therefore, thinking about the institutional framework bringing together individuals in the context of organizations ( March & Olsen 1984).

The study of capacities poses deep theoretical challenges. Firstly, the problem of information and incentives for individuals to collaborate with the organizations of which they are part. The second problem is how to best allocate individuals according to their competences within the organization. Third, organizations are made of multiple individual and management mechanisms according to contract terms and organizational culture that affect performance ( March & Olsen 1985; Simon 1951).

With regard to the problem of the implementation of public policies, the concept of state capacities, with a view to a more structural and macro-historical conception, is generalized, abstract and without a clear analysis of the effects of bureaucratic behavior ( Williams 2017).

In this sense, it is worth mentioning the additional criticism raised by the literature towards the formulation of the concept of state capacity. On the one hand, the concept and malleable conformation allows different aspects of the policy implementation complexity to be incorporated. On the other hand, the concept of capacity holds an intrinsic limitation of empirical observation given that its analytical power is restrained to the level of stock of resources thus of potentialities of the state action. In other words, state capacity has analytical key problems of direct observation given that it does not look at action per se but only at the potential of the state action ( Cingolani 2013; Mazzucca 2012; Williams 2017). The problem remains to understand what in fact explains capacities activation dynamics.

Looking at bureaucratic performance and policy results, Williams (2017) argues that there is an additional challenge regarding the transition from the individual level towards organizational and systemic levels of capacity. The scholar claims that organizational capacity is not the sum of individual capacities, as usually assumed in the concept operationalization. This assumption relies on the misleading conception of bureaucracy homogeneity and uniformity and does not acknowledge the product of the collective action ( Williams 2017).

The analysis of the implementation of public policies and the role of bureaucracies has advanced to the concept of policy capacity. The concept of policy capacity seeks to understand the role of bureaucracies in implementation by bringing together a perspective based on the interaction between individuals and organizations. Policy capacity also has multiple definitions. The assumption is that policy capacity is an implementation predictor. Policy capacity is “... the ability to marshal the necessary resources to make intelligent collective choices, in particular to set strategic directions, for the allocation of scarce resources to public ends” ( Painter & Pierre 2005, p. 2).

This assumption involves a conception of skills such as the construction of competences relevant to the implementation of policies based on knowledge that involves the use of research for policy problems, effective use of communications and strategic management of stakeholders ( Howlett 2009). The development of policy capacity involves the application of resources to review, formulate and implement public policies ( Fellegi 1996). It can also be defined as “... set of skills and resources - or competences and capabilities - necessary to perform policy functions” ( Wu, Ramesh & Howlett 2015, p. 166).

From this perspective of skills and abilities, the concept of policy capacity brings together the necessary skills and resources for policy implementation, considering individual, organizational and systemic elements. These three factors bring together the resources and skills that would be predictors to produce results in policies, broken down by analytical, operational, and political competencies ( Wu, Ramesh & Howlett 2015).

The concept of policy capacity examines the resources needed to implement policies, in the dimension of individuals ( Hsu 2015) and organizations ( Pattyn & Brans 2015). The concept of policy capacity becomes attractive in the eyes of the public policy analyst. It incorporates both dimensions: the collective actors of state bureaucracies and the political elements of legitimacy building. Political elements become essential in building the legitimacy of policies ( Moore 1995). However, like the concept of state capacity, the concept of policy capacity presupposes a set of resources - financial and human - that are predictors of good public policy.

The understanding of policy capacity continues to address bureaucratic dynamics as a stock of capacities that are analytically predictors of policy outcomes. The understanding of the stock of capacities does not consider some theoretical hypotheses. First, it can be said that a given organization may have a high stock of capacities. But bureaucrats in this organization do not want to collaborate with policies, or there may be problems with the policy process that negatively impact outcomes. On the other hand, there may be a low stock of capacities, but even with this low inventory, policies can yield good results ( Williams 2017).

Policy capacities must be understood as a dynamic flow and process than as a stock that works as a predictor of good public policy. More than the objectives of organizations and the stock of bureaucratic capacities, it is necessary to understand the dynamics of agents in institutional contexts. Therefore, this investigation focus on the policy capacity debate by means of an analytical approach that looks precisely at the work of the bureaucratic agents to implement a public policy.

Relevant contributions from the policy work debate can be identified to discuss the role and performance of bureaucracy in policy production. Firstly, policy work approach starts from the observation of the field and not from prescriptive models, allowing thus a broader comprehension of flows and dynamics embedded in specific existing contexts ( Colebatch, Hoppe & Noodgegraaf 2010). As a consequence, that approach provides elements to recognize a wider range of roles and functions played by bureaucrats in policy production, which are not restricted to the policy cycle stage model neither to the policy analysts conception which is not recognized in the Brazilian context ¹ . Instead, it acknowledges as functions performed by policy makers not only the production of formal analysis, but also managerial tasks, mediation, translation and building shared meaning amongst all policy interested parts ( Colebatch, Hoppe & Noodgegraaf 2010).

One may also argue that policy work relational perspective in contrast to top-down comprehension of policy governance allows one to examine the products of the interaction between bureaucracy and policy stakeholders as part of policy production ( Repetto 2004; Wu, Ramesh & Howlett 2015).

The concept of policy work seeks more than a dynamic of stock of capabilities. Policy work is active and involve attitudes and beliefs of bureaucratic agents. It addresses policy styles and the set of attitudes and tasks performed by bureaucratic agents, so that the production of results in public policies depends on a more complex bureaucracy ( Brodkin 2011). Policy work is the set of activities disbursed by bureaucrats who bring together attitudes, tasks and world conceptions that specify the dynamics of policy formulation, implementation and evaluation ( Veselý, Wellstead & Evans 2014). The attitudes and beliefs of bureaucrats matter to understand the outcome of public policies ( Ball et al., 2011 ). Bureaucratic agents mobilize their individual capacities in a complex flow of public policies, which involves different sources of knowledge, dimensions and expected results.

These dynamics involve a complex framework of activities that need to be performed by bureaucrats for policy outcomes to be produced. Understanding these activities and how they are mobilized by agent’s matters directly in the construction of a public policy ( Brodkin 2011). These dynamics need to be analyzed in a framework that allows the understanding of internal and external activity of bureaucratic agents, as well as the interaction between the individual and the organization. The framework below presents this policy work and dynamics of capacities which we will test empirically.

Table 1 presents an analytical framework to understand the policy work performed by bureaucrats. The construction of this framework started from the literature review, covering a set of activities that are considered essential for policy implementation ( Wu, Ramesh & Howlett 2015).

The framework presented above brings together a complex set of activities aimed at understanding the activation of bureaucracy capacity in the form of policy work. Capacities do not represent a rational organization of governments but a collection of individuals who perform functions and activities in a context of bounded rationality, of undefined goals and in a game of discretion and obligations ( March 1994, March & Simon 1958; Simon 1991). This means that these capacities can be mobilized or not in the dynamics of policy implementation. The degree to which these skills and competences are mobilized or not can impact policies in a varied way, demonstrating that implementation is complex. Policy workers may be oriented toward formal strategies to implementation, such as a managerial strategy. However, they adapt and adjust the implementation of policies in the context in which they act, adjusting the processes in the field of practice ( Brodkin 2011).

First, capacities can be understood within bureaucracy. The capacities required for a public policy to be implemented involve skills and competences within organizations, regarding analytical and managerial dimension. The analytical dimension of policy work involves the dynamics from which bureaucratic agents generate capacity to produce data to inform the public policies being implemented. These capacities require the agents of the bureaucracy to gather and analyze data, as well as specific knowledge to produce evidence in public policies, to improve information management and to generate institutional learning during implementation ( Carney 2016, Wu, Ramesh & Howlett 2015). This capacity of policy workers involves building of knowledge on data production and technical expertise involving the specific policy, to allow the development of this analytical capacity. That dimension entails the original conception of the role of policy analysts ² .

The second dimension of these capacities of policy workers involves management knowledge and skills, norms and regulations of a given policy, administrative skills of the policy workers, personnel management, allocation and management of financial resources and technological resources of organizations. This second dimension concerns a set of public administration back-office activities, which aim to guarantee technical conditions for the performance of policy workers. They involve punctual managerial attitudes to support implementation activities, which may impact or not policy outcomes.

The first two dimensions involve the activities of the administrative back-office. But the activities of policy workers are not restricted to this back-office, nor are they restricted to the conditions most intricately linked to the administration’s infrastructure. Policy work also involves the mobilization of relational capacities, which aim to ensure internal operational conditions for public management ( Peters and Pierre 2001), aiming at the coherence of public policy, as well as political support that provides the conditions of legitimacy and public value production ( Moore 1995).

The relational capacities that can be mobilized by policy workers involves two dimensions. The dimension of internal coordination relates to a set of activities aimed at improving the state agency. They aim to ensure institutional conditions for the operation of the policy and the maintenance of internal coherence ( Peters 2004). This involves a decision-making process that needs to be coordinated, given the fact that bureaucracies have multiple principals ( Wilson 1989; Dixit 1996). Internal coordination relates to the constitution of elements of monitoring and control of the policy, aiming to ensure its legality and probity in the public management ( Doig & McIvor 2003). It also draws in the creation of inter and intraorganizational informal or formal structures that allow this process of coordination of the activities of policy workers, involving both intraorganizational coordination and the coordination of organizations involved in the implementation - interorganizational coordination ( Alexander 1993; Peters 2004).

The challenge of institutional coordination requires a multi-level governance system, with shared strategy of governing as well as decision-making and coordination bodies that define a coordinated and negotiated pattern of public action, taking into consideration the involvement of various institutions. This perspective on coordination of institutional action involves non-hierarchical structure to exchange between institutions in different levels ( Hix 1998; Peters & Pierre 2001; Smith 1997). Moreover, as stated by Peters (2004), it recognizes that ideas play a crucial role in creating a more integrated pattern of governance. These ideas need to be relevant to range of policy areas, and they also need to be sufficiently powerful to pressure organizations and actors which might as soon persist in their established patterns of action.

Finally, relational capacities also involve political dimension. It acknowledges the diversity of sources of power and the multiple interests and ideologies distributed in society ( Repetto 2004). State action does not develop in isolation and is not neutral in relation to the issues of power which, in contemporary states, originate not only from the state apparatus, but also from civil society and the international environment ( Repetto 2004). This political dimension of policy work means mobilizing capacities to promote the necessary interactions with the external environment of management. First, policy work demands activities that aim to promote social participation and accountability. Social participation is a fundamental element for promoting the legitimacy and publicity of public policy ( Avritzer 2012), increasing the conditions of accountability ( Filgueiras 2016).

This demands that policy-makers be able to act in participatory spaces, representing governments in contexts of civil society participation in the decision-making process and implementation of public policy ( Avritzer 2012). In addition, the democratization of the state requires that policy makers act within the administration to build bridges with civil society through participatory institutions ( Warren 2009). Thirdly, relational capacities in the political dimension include negotiation skills to dialogue with the legislative and judicial branches.

The legitimacy of public policy demands that there be permanent interlocution with the powers, to secure spaces of negotiation ( March & Olsen 1995). Finally, relational capacities must account for international communities, to disseminate social policies and technologies in order to promote international improvement and interlocution ( Olsen 2010; Underdal 1995). Relational capacities require managers to engage in networks with transnational policy actors so that they can gain operational and analytical support for the adoption of implementation practices ( Dolowitz & Marsh 2000; Stone 2004).

This analytical framework assumes a set of activities that policy workers need to play in the implementation process. It is important to emphasize that the study of policy work demands the mobilization of individual capacities inserted in organizational contexts. Policy work involves understanding this relationship between the individual capacities of bureaucrats and organizational contexts to provide a robust framework for analyzing bureaucratic predictors for public policy. The next section looks at the Brazilian case, addressing the decision-making and the implementation processes of public policies.

Brazilian federal civil service is this investigation’s main unit of analysis. As debated in international comparative works, there is no common definition of civil service and its scope and size strongly vary according to each country’s public employment trajectory and structure ( Evans 2008; Rao 2013).

A Weberian delimitation of civil service, identified in the Organization for Economic Co-operation Development (OECD)’s countries and some developing countries, involves the category of public civil servants that are under a merit-based system which provides some level of standard regulations and procedures regarding servants `selection, appointment and discretion’ ( Evans 2008; Ramos & Milanesi 2018).

This investigation acknowledges that delimitation but includes two additional functional aspects also recognized in the comparative studies in the field, which are civil servants’ permanence and relevance for core activities of the state ( Rao 2013). For the purpose of this research, civil service is defined as the civil servants that work, under a merit-based system, in the Federal Executive Branch and are involved more directly and permanently in the public policy production. In that delimitation, teachers, health workers and policemen were not included given to the fact that, for the Brazilian case, each of these groups have specific employment regimes and precise roles in the policy production.

In this investigation, the universe of analysis was reduced to the Brazilian civil service which works in the federal level and is involved in the direct administration. That decisions were made given to the research limitations of time and resources to operationalize the investigation with the whole universe. Thus, the investigation looks at federal civil servants working in all ministries, vice-presidency and presidency.

In the Brazilian public administration, the legal regime of organizations differentiates between direct administration and indirect administration. The direct administration represents the core of the government, bringing together organizations that perform functions of government and state. Indirect administration, on the other hand, is the set of organizations that, linked to organizations of the direct administration, perform decentralized functions of State and functions of interest of the government. Indirect administration comprises foundations and autarchies, such as universities and research institutes.

In the Brazilian federal system, which resulted from a decentralization process, municipalities and state governments are essential parts of the policy implementation. In many policies, such as in health and educational fields, local governments hold authority and autonomy for making decisions and for managing different sort of resources for service delivery ( Abrucio & Franzese, 2007). Moreover, federal indirect agencies also play fundamental roles in the policy production, particularly in the territory ( Paula et al., 2017 ).

Notwithstanding the relevance of both groups of public servants in local governments or in indirect federal agencies, this investigation looks at the specific context of the civil servants that work in the direct agencies. Two main reasons support that decision. The first one relates to the still reduced literature that analyzes in depth this particular group functions’ specificities and conditions for policy production ( Cavalcante & Lotta 2015; Howlett 2011; Pires 2012). The second one is a practical reason related to the size and heterogeneity of the three groups which would make the data collection unfeasible for the project time framing.

Looking at the developments of the bureaucracy’s literature in Brazil, one can argue that this investigation’s universe of analysis entail, mainly, the context of high level and middle level bureaucrats in the federal administration.

Most of the research carried out in Brazil on public bureaucracy focuses on high-level bureaucracy, especially the federal bureaucracy and how it acts in the formulation of public policies ( Loureiro, Abrucio & Rosa 1998). However, there are other works that seek to understand the role played by public policy-making bureaucrats at the “front line”, that is, those actors who operate the policies formulated by the high level of government, called “street level bureaucracy” ( Lipsky 2010; Lotta 2010, 2012; Pires 2009, 2017). Little understanding exists about those actors who are between the policy-making elite and those who implement it, that is, about the federal bureaucracy, which plays an important role in the process of coordination, articulation and translation of public policy production ( Abers 2015; Cavalcante & Lotta 2015; Lotta, Pires & Oliveira 2014; Oliveira & Abrucio 2018; Pires 2015). Thus, the findings of this research contribute to the theoretical-analytical consolidation of the performance of this bureaucracy.

Another aspect that must be highlighted regarding the justification of this investigation’s design is the adoption of Colebatch, Hoppe and Noodgegraaf (2010) assumption that policy work is apprehended and developed in practice. As mentioned in the first section, contingencies matter and distinct national contexts produce different forms of action and combinations of policy functions. It is worth noting that policy analysis in Brazil is not understood as a professional field with specific training, techniques and methodologies, which comprises a community with shared identity and agenda, such as in the Anglo-Saxon countries where the field of knowledge firstly evolved. However, this does not mean that this role is not being played in the Brazilian public service ( Farah 2016; Vaitsman, Ribeiro & Lobato 2014). However, one must argue that there is no formal distinction, for instance, between the ones who can work with policy analysis and the ones who can formulate or implement policies ( Farah 2016).

Recent investigations contest that a vast and more complex range of policy works can be identified in the examination of the field, challenging thus the traditional dichotomy between technical and political roles in the policy production ( Howlett 2011). To provide mediation between the state and policy stakeholders, to clarify meaning, to democratize information related to the policy production are some of other policy works found in empirical investigations in different countries ( Colebatch, Hoppe & Noodgegraaf 2010; Lotta 2010; Mayer, van Daalen & Bots, 2013; Pires 2012; Veselý, Wellstad & Evans 2014).

To conclude, this investigation intends to explore how the policy work and policy functions are played by the Brazilian federal civil service and their existing resources and performing conditions.

This section presents the methodology used in the research to explore and analyze the different functions performed by the Brazilian federal civil service in the direct administration.

Based on data from the Integrated System for the Administration of Human Resources (Siape) of March 2017, this investigation looked at a universe of 96,534 individuals, which represented the group of federal civil servants of the direct administration ³ .

The election of that sample of that universe was carried by means of the technique of the simple stratification ⁴ , according to each one of the 24 existing agencies ⁵ of the direct administration in the base of the Siape in March of 2017. The use of the sampling is justified by the great variability of public servers between the organizations. After the calculation ⁶ of the size of each relative sample to the respective agency, respondents of organization were randomly selected to compose each group. Information regarding the universe of the research, the definition of the sample and the collection of valid data are reported in Appendix A ( Table 1A).

The elected sample totalized 6,055 individuals who received a survey tool designed based on the analytical framework shown in Table 1 in order to collect data on activities performed and capacities accumulated by the Brazilian federal civil service. The survey tool was chosen as a research technique for two reasons, namely: i) potential to capture as many perceptions as possible; and ii) the possibility of generating new research hypotheses from the data provided by the field ( Babbie, 1990). The development of instrument survey was conducted from March to July of 2017 and was pre-tested with professionals of different agencies of the Brazilian federal government from August to September of 2017.

The survey was hosted by an online platform and was sent by email directly to the public servants. Data collection took place from October to December 2017. The research obtained, after data processing, a database composed of 2,000 responses, representing a total response rate of 32% ⁷ .

Demographic data show similarities between the universe’s profile and the respondents’ sample composition. In the universe of federal civil service, there is a predominance of the male gender (56.81%). The same pattern can be found amongst the respondents’ sample, in which the male gender represents 55.2% of the sample. Age average in the federal civil service corresponds to 45 years old, while in the respondents’ sample it reaches 47 years old. Regarding race data, though, a relative difference can be found among the two groups regarding the percentage of the black and “parda” groups, as it is shown in Table 2. That difference can be related to the source of data. The former set comes from an administrative record (Siape), while the latter was self-reported by respondents in the survey. This work argues that the respondents’ profile generally corresponds to the universe one.

In relation to the analysis of the data, a multivariate exploratory analysis was carried out to find possible associations between the variables studied. In this sense, it was used factor analysis to reduce the number of initial variables, with the lowest possible loss of information in factors and test the analysis framework for policy work. The application of this technique made it possible to evaluate interrelationships between variables, so that these could be adequately described by new groups of basic categories, namely “factors” ( Pestana & Gageiro 2005).

The main issue is to understand how the activities carried out by bureaucrats are organized in collective action and how they can be understood theoretically in the process of implementing public policies. This factorial analysis aims to demonstrate how the policy work can be gathered in abstract categories of analysis from the framework presented previously and what are the specifics of policy work in Brazil. From the application of these techniques, it was possible to observe different profiles and functions of public policies performed by federal bureaucrats of the Brazilian Civil Service, as it will be discussed in the next section.

The Brazilian policy workers who responded the survey have a high degree of education. 59.4% have some graduate qualification. They are relatively older, with an average age of 45. They are mostly male (56.30%), slightly lower than the government average of 59%.

In relation to the government sectors in which they operate, the respondents are relatively distributed, with greater participation in social and security area (33.15%). In relation to the time of performance in public policy, they are experienced: 34.5% have been working for more than 10 years in the same policy, with higher rate of those that are located in the area of territorial sovereignty and management (38%) and lower rate of those located in the area of infrastructure (24%) ⁸ .

With regard to the trajectory of being in positions of trust within the bureaucracy, the respondents showed a high occupancy rate: 57.15% said they had already held some strategic position in the public sector.

In Table 3, factor analysis which cumulatively explains 59.3%, reveals four main components of works performance. The policy work within the Brazilian bureaucracy makes it possible to constitute categories of analysis that show a differentiation of the activities performed. Bureaucracy is not a single, coherent aggregate of capabilities, but a complex set of dynamic activities that can be aggregated into general categories of analysis. These categories suggest a theoretical framework that specifies dynamics of policy work.

The result of the factorial analysis returned 4 principal components that aggregate different activities performed by bureaucrats. The Eigenvalues are adequate and the KMO test of 0.902 suggests the suitability of the sample.

The first component shows a relational type of work, involving representation, negotiation and coordination tasks with internal and external partners of policy. The relational component reveals capacity of policy workers to build interactions with other partner organizations in the implementation of policies, to provide networks and synergy that facilitate the process. The fulfillment of this policy function builds policy capacity outlined as political skills of public managers.

The second component revealed an analytical-oversee work that grouped data production tasks with activities related to the compliance with accountability agencies recommendations. This analytical component is the fulfillment of a policy function based on the production of information and analysis. It is interesting to observe how in this component the production of information is aligned with recommendations originating from the performance of accountability institutions. This component suggests that the production of information and analysis is aggregated with audit processes. This important policy function is performed by policy workers as a result of institutional controls more than to enhance evidence based public policy.

The third component relates managerial role which involves technical tasks of resource mobilization regarding dealing with contracts, agreements and prospect funding. This policy function involves a set of tasks performed by policy workers in a back office required for public policy. This back-office involves elements of management in organizational processes.

Finally, a fourth component corresponds to administrative work that deals with operational tasks. These administrative tasks involve human resources activities, prosecuting and day-to-day operational tasks for maintaining organizations.

Looking this data in face of the proposed analytical framework in Table 1, one must highlight that the portrait above suggests that the analytical capacity of the Brazilian civil service is mostly aimed at attending accountability demands, not prioritizing policy production or policy development, as stated by the literature.

This component, when aggregating this set of activities, relates the production of evidence and information on public policies with audit reports and activities carried out by accountability institutions. The analytical dynamics of federal government bureaucracy is guided by compliance and auditing actions.

Power (1994; 1997) has pointed to the development of the “audit society” as one of many dimensions of the New Public Management, as well as a more general movement toward quantification in society. Measure performance and use auditing to control government produced many unintended, and undesirable, consequences. For example, strict auditing tended to decrease innovation in the public sector and created distrust among the participants in governance processes ( Power 1997, p. 127).

The production of information and evidence is related and moves in the same direction of compliance monitoring and responses to audit reports. This dynamic of the analytical capacity of the bureaucracy circumscribes a set of activities in which information is related to audit processes. The role played by accountability institutions ( Filgueiras 2018) and control activities move the public policies implemented by the bureaucracy, imposing red tape effects and directing the practices of public servants ( Boyne 2002).

Relational capacity, on the other hand, is typical of the center of government, with the prospect of coordinating actions. The result presented suggests that this set of policy work constitutes complex coordination activities of the government, in order to promote policy coherence. Bureaucrats engage in policy coordination activities, both within the size of the center of government and in organizations implementing the policies ( Cavalcante 2018).

The solution to the constant problems identified by different stakeholders in relation to public policies is related not only to the availability of information but also to the coordination during formulation and implementation of public policies ( Peters 2010; Souza 2018). Many government programs are contradictory, and others may have gaps that fail to provide public services. These failures result from coordination difficulties ( Bouckaert, Peters & Verhoest 2016; Peters 2004). Going further, reforming public policies has a mantra: governments must improve coordination.

The federal government bureaucracy engages in coordination activities through relational skills that are activated in the dynamics of public policies. Coordination of public policies is also required between political and bureaucratic actors or organizations that comprise the same level of government - horizontal coordination - or that comprise different levels of government - vertical coordination ( Souza 2018). In the sense of coordination, actors and agencies play different roles in different phases of the same public policy. The challenge for coordination is related to collective action involving different public organizations and policy stakeholders for the provision of services to citizens ( Bouckaert, Peters & Verhoest 2016).

The bureaucracy performs policy work with a relational capacity approach. The main functions are negotiating, representing, participating, meeting interest groups and joint projects, and preparing normative bills. These policy works are essential of work for governance and increase a pluralistic state.

There is, in fact, a group dedicated to managerial tasks which is distinct from the group responsible for more operational activities. The capabilities are organized according to the framework presented previously, considering the flow of policy functions that are performed by the Brazilian Civil Service. These groupings mark back-office work, considering activities essential for the implementation of policies, such as public procurement, human resources management and information technology management.

Analytically, the definition of these categories of capacities demonstrates the differentiation of policy work, in which the activities performed will depend on different capacities. This finding is important in understanding policy work and bureaucratic activities, which may have a different impact on policy outcomes. Thus, understanding state capacities does not involve understanding the macro-structures of public organizations, but rather the collective action of bureaucrats in the policy implementation process. Taking the concept of state capacity to explain policy outcomes depends on understanding the work of bureaucrats, their differentiation, and dynamics in the context of public organizations.

Public policy areas present different dynamics of policy work. The policy functions performed by the Brazilian Civil Service vary according to each of the components and each of the policy areas. It is interesting to observe in the charts below how the factorial indices highlighted in the model above are distributed in the percentiles, if observed in the dimension of the policy areas.

Looking at Graph 1 the relational type function is less common in the area of urban mobility and energy and in the area of energy and communications. The areas linked to social policy and citizenship and economic policy demand more relational activities, be it in the structure of Brazilian federalism, or with other powers and civil society. These areas have multiple stakeholders and policies are carried out intersectorally, demanding broader relational capacities that favor coordination processes. For example, social policies require greater coordination skills with areas such as health and education.

With regards to the type of analytical / oversee function and policy areas, according to Graph 2, the areas of urban planning, housing, sanitation, and land use respond strongly and very frequently to the question of the performance of accountability institutions. Within the area of infrastructure, this result demonstrates how this area of public policy demands that public managers work with analytical resources to respond to audits and accountability systems ( Gomide & Pereira 2018). Also, the social area also mobilizes its analytical capacities to perform tasks-related to responding to accountability institutions.

With respect to managerial skills, according to Graph 3, the mobilization of work in relation to process, partnership and contract management, as well as exploration of resources is more common in the area of economic policy, urban planning, housing, sanitation and land use. This result is expected, considering the amount of economic resources with which these areas operate, demanding more activities for the management of contracts and negotiation.

Finally, according to Graph 4, more administrative activities are carried out, especially in the areas of social and rural development, citizenship, diversity and security. These administrative activities require public managers to work with human resources, finance and organizations. It is interesting to note that the area of economic policy is the one that least mobilizes these administrative capacities.

Looking at the four graphs, one can deduce that capacities are fluid and different, depending on the policy areas and the organizational contexts in which they operate. These capacities can be mobilized differently by the areas and according to the activities that are demanded. Capacities are dynamic and depend on the performance of policy functions and not necessarily on an organizational macro-structure that operates above individuals. Organizational contexts matter, but within a set of behaviors, actions, beliefs and attitudes of the individuals that making work in a policy.

Theoretically, observing the policy work at the individual level may respond to a more complex set of variables that makes it possible to identify the dynamics of policy processes, the specificities of different bureaucratic organizations, and how this dynamic of capacities impacts policy outcomes. The findings of this research demonstrate that the policy capacities in the Brazilian federal government are circumscribed in a governance perspective. This perspective of governance comprises a more plural state, whose analytical focus must be on organizations and their institutional environment, where policy implementation depends on policy work performed by bureaucrats. This policy work involves permanent negotiation of values, senses and relationships, creation of networks and projects with multiple stakeholders, managerial activities and administrative functions ( Osborne 2010; Peters 2010).

The findings of this article point to the fact that capacities are not inventories of skills, not even macrostructures that move policies. Capacities depend on being activated and put in flux, mobilizing the different activities needed for policy implementation. Capacities depend on the action of agents contextualized in institutions ( Brodkin 2011; Colebatch, Hoppe, & Noodgegraaf 2010; Williams 2017). These actions occur in complex processes, making policy work depend on relational, analytical, managerial, and administrative capacities.

This study would benefit from comparative studies that show the differences in the public policy process between different countries. Theoretically, bureaucracies with different profiles and different institutional arrangements will imply in different dynamics of policy work and, therefore, in the activation of different capacities according to the organizational context. This is an important future agenda for the analysis of policy capacities to understand how different institutional arrangements activate agents and perform policy work.

Capacity building in the public service cannot be understood only as a set of organizations that fulfill delegated goals for the realization of public policies. These policies depend on a set of actions performed by individuals who carry with them different individual capacities, which are brought together in a complex set of organizations, which demand commitment and motivation to achieve their ends ( Cyert and March, 1963).

The concept of state capacities does not necessarily represent what bureaucracies do. What bureaucracies do depend on a set of contingent actions for the performance of various public policy functions. These public policy functions fulfill the role of complex activities that, in the aggregate of all organizations, comprise a complex and highly differentiated map that cannot be interpreted only within the scope of organizational objectives. The functions that are performed by public managers compose a complex framework that can explain the performance of policies. This demonstrates that the concept of capacities should not only focus on a macro conception of organizations, but on a more nuanced concept of capacity, which is associated with the concept of policy work ( Colebatch, Hoppe, & Noodgegraaf 2010).

Understanding what public managers do and what gaps are related to relational, analytical and managerial capacities can serve to outline a more effective public policy perspective. This nuance can make reformers better understand the performance of functions and to cover the institutional and managerial incentives for the realization of policy work. Also, it can serve to understand eventual management failures in a more realistic way, including the specifics policy domains.

This research sought to understand this nuance of the policy works performed by public servants in Brazil and how this policy work aggregates in different capacities. This research needs to establish a dependent variable that can associate the capacities listed above with policy performance. Having done that, it will be possible to map which capacities are relevant to producing results. And, similarly, what capacities, when they fail, explain the failure of certain policies. This task is not simple, so a robust discussion about the validity of the results in terms of evidence is required and collected in order to provide analytical results.

Capacity building and change promoted in bureaucracies depend on institutional incentives and on an organization based on motivation and leadership. In the case of public bureaucracies’ political leadership must play a central role, not in the sense of creating organizational goals, but in producing incentives that are capable of mobilizing a complex set of individuals who perform actions based on attitudes, beliefs and interests within bureaucracies ( Cyert & March 1963; March 1994; March & Olsen, 1995) .

Understanding of state capacities should occur not as organizational stocks but as continuous and repeated flows of work performed by individuals within organizations ( Wu, Ramesh & Howlett, 2015). Our findings indicate that the capacities are dynamic aggregates of policy work, with the goal of delivering public policies and services to society. Capacities can be predictors of results. But they need to be dealt with in a dynamic and contingent way, taking into consideration organizational and political demands.