Institutional Multiplicity and the Fight Against Corruption: A Research Agenda for the Brazilian Accountability Network

MULTIPLICIDADE INSTITUCIONAL E A LUTA CONTRA A CORRUPÇÃO: UMA AGENDA DE PESQUISA PARA A REDE DE ACCOUNTABILITY BRASILEIRA

Abstract

Since democratization, Brazil has established a robust network of accountability institutions that perform a myriad of functions in combating corruption. While there is empirical research on the inner workings of the Brazilian accountability network, many questions remain unanswered and many dimensions of the interactions between institutions in this system have yet to be analyzed and uncovered. This paper argues that the accountability literature can benefit from further descriptive empirical studies, especially detailed analyses that account for empirical differences in norms, procedures and sanctions. To promote this type of granular empirical analysis, we formulate a research agenda for the Brazilian accountability system, arguing that the concept of institutional multiplicity has much to offer in this endeavour. Considering the difficulty in capturing the complexity of Brazil's vast network of accountability institutions in a single paper, we focus on the accountability systems for civil servants working for the federal executive branch. Despite being focused on a particular dimension of the accountability system, we hope that the topics proposed here can inform research questions about other areas of the system as well.

Keywords
Corruption; accountability; institutional multiplicity; Brazil; civil service

Resumo

Desde a redemocratização, o Brasil estabeleceu uma robusta rede de instituições de accountability que desempenham uma ampla gama de funções no combate à corrupção. Apesar de haver pesquisas empíricas que analisam o funcionamento interno dessas instituições, há diversas questões ainda não abordadas, e muitas dimensões das interações entre as diversas instituições que integram a rede ainda precisam ser analisadas. Diante desse cenário, este artigo argumenta que a literatura sobre accountability pode beneficiar-se de estudos que abordem empiricamente como diferentes instituições da rede de accountability lidam com múltiplas normas, procedimentos e sanções. Com o objetivo de promover esse tipo de análise detalhada, o artigo propõe uma agenda de pesquisa para o sistema de accountability brasileiro a partir do conceito de multiplicidade institucional. Devido à dificuldade de capturar o complexo funcionamento da vasta rede de accountability brasileira, o artigo centra-se no sistema de accountability de servidores públicos civis do poder executivo federal. Apesar de tal recorte, espera-se que as questões aqui propostas possam inspirar perguntas de pesquisa sobre outras áreas do sistema de accountability.

Palavras-chave
Corrupção; accountability; multiplicidade institucional; Brasil; serviço público

Introduction

Since democratization, Brazil has established a robust network of accountability institutions that perform a myriad of functions in combating corruption. Since then, legal scholars have analyzed a series of law-related problems in the relationship between the different norms and institutions that form Brazil's accountability network, such as: is it fair for a public servant or private actor to be punished by different institutions at the same time? What are each institution's tasks? How should each institution's decision affect others? Should evidence produced in one procedure be used in other institution's procedures? The diversity and depth of these questions show that the perfection of the Brazilian accountability network depends on a sophisticated legal debate.

This paper argues that legal research on the Brazilian accountability network can benefit from empirical analyses based on a social sciences perspective. Recent empirical studies have raised important questions about the operation of this network: how does this system work in practice? Are the actors in different institutions influencing each other? Does the variety of different institutions with functional overlaps help or hamper the system's performance? ( ALENCAR and GICO JR., 2011 ; ARANHA, 2018 ; DA ROS, 2019 ; MACHADO and PASCHOAL, 2016 ; PRADO and CARSON, 2016 ; ROCHA and ALENCAR, 2009 ). While these studies provide a great start, much remains to be analyzed and uncovered.

In this regard, this paper argues that the concept of institutional multiplicity has much to offer. A multitude of institutions in Brazil perform the three core functions of an accountability network (monitoring, investigation and punishment), thus aptly described as an institutional multiplicity system. This term refers to “multiple rule systems that confront economic and political actors providing distinct and different normative frameworks and incentive structures in which they act” ( GOODFELLOW and LINDEMANN, 2013 ). Institutional multiplicity is characterized by the coexistence of independent normative systems, such as the Brazilian administrative, civil, and criminal spheres of responsibility. In other words, while preserving the idea that these normative systems are a form of accountability, the concept of institutional multiplicity also highlights the differences among them, the complex ways that these normative systems may interact with and influence each other, and how these differences may affect the behaviour of individual actors.

Using the concept of institutional multiplicity, this paper articulates a research agenda for empirical analyses of Brazil's network of accountability institutions. Considering the difficulty in capturing the complexity of Brazil's vast network of accountability institutions in a single paper, we focus on the accountability systems for civil servants working for the federal executive branch. More specifically, we focus on the investigation and sanctioning of these officials in the three spheres: administrative, civil, and criminal. The scope of this paper is purely descriptive: while there are important questions of whether this system should be better integrated/coordinated or further decentralized, a rich empirical analysis can produce data to enhance the quality of the debate on these normative questions.

To develop this research agenda, the article is structured as follows. Section 1 provides basic definitions of the concepts with which this paper works: corruption, accountability network, and institutional multiplicity. Section 2 reviews the Brazilian scholarship on the accountability of corruption, indicating the need for more systematic, detailed empirical analyses. Section 3 uses the concept of institutional multiplicity to propose a research agenda on accountability networks, focusing on civil servants in the federal executive branch. This section identifies three dimensions of institutional multiplicity, formulating a set of specific research questions for each.

1 Defining Corruption, Accountability Networks and Institutional Multiplicity

Before reviewing the literature on corruption, accountability networks and institutional multiplicity in Brazil, it is important to define these terms.

1.1 Corruption ¹

In the academic and policy literature, corruption is often defined as “the use of public office for private gain” – that is, it necessarily involves public officials (political or bureaucratic) ( BARDHAN, 2005 , p. 138; GRAY; KAUFMANN, 1998 , p. 7). ² More often than not, private gains refer to economic benefits, but the term can also encompass the use of public office to obtain non-economic gains (political or personal). ³ Although corruption comes in many forms, the common element is that an agent (usually a government official) pursues personal gain by bargaining away an item or benefit over which he or she has discretion, but which belongs to a principal (usually the public) ( KLITGAARD, 1991 , p. 69-74).

Two of the most important types of corruption for economic gain are embezzlement and bribes. While embezzlement entails the misappropriation of public funds, bribes involve the appropriation of private funds. Officials might ask for bribes to perform their functions (e.g. a city official asking for money to issue a permit that a citizen would otherwise be entitled to), not to perform their functions (e.g. a police officer asking for money not to issue a speed ticket), or not to perform acts that are not part of their functions but which they can threaten to perform through abuse of power (e.g. a police officer asking for money not to torture a prisoner, or a tax official demanding a bribe to forgo an audit).

Unless the word “corruption” is used in a particular legal provision (e.g. the Brazilian criminal code defines bribe paying and receiving as the crimes of active and passive corruption), lawyers prefer to talk about specific kinds of behaviour and their legal statuses. Thus, while a lawyer would agree that prohibiting bribery and embezzlement are ways of preventing the “use of public office for private gain,” they could also point to the fact that these two kinds of behaviour are significantly distinct ( DAVIS, 2019 ). For instance, one involves a private party, while the other can be fully executed without third party involvement. Therefore, they are, in a way, addressing quite different problems and also can lead to very different sets of sanctions.

1.2 Accountability Networks

Since the ground-breaking work of Susan Rose-Ackerman in the late 1970s, corruption has been conceived as a political economy problem, i.e. corruption happens when the costs are lower than benefits (1978). Building on this assumption, the literature has investigated what kind of mechanisms modify corruption incentives, which include salaries, levels of discretion, transparency and accountability ( FISMAN and MIGUEL, 2009 ). These incentives are generally divided into two types: ex-ante mechanisms that eliminate the temptation or the opportunities to engage in corruption (e.g. higher salaries and lower levels of discretion) and ex-post mechanisms that make those who engage in wrongdoing liable for their actions (e.g. dismissal, fines and prison sentences). While ex-ante and ex-post mechanisms are complementary and can jointly enhance an anti-corruption system's effectiveness, we focus on ex-post mechanisms, which are mostly sanctions.

Any form of punishment depends on mechanisms to uncover and investigate corruption, which in turn requires an intricate network of institutions that interact with each other in complex ways. This network can be described as a “web of accountability institutions” ( MAINWARING, 2003 ) or as “an integrity system” ( POPE, 2000 ; SPECK and NAGEL 2002 ). Despite comprising different institutions and arrangements in different countries, these networks of institutions usually perform three primary functions: (i) oversight, which entails monitoring those occupying positions of power and/or engaged in activities where there is a high risk of corruption in order to quickly identify anything suspicious or atypical; (ii) investigation, i.e. the process of obtaining more detailed information about acts or activities once suspicion has been raised; and (iii) punishment, i.e. the sanctioning of actors against whom there is enough evidence of wrongdoing ( POWER and TAYLOR, 2011a ; TAYLOR and BURANELLI, 2007 ). A comprehensive analysis of the complex interactions of institutions in these multifaceted accountability systems is necessary to understand the mechanisms of anti-corruption accountability. ⁴

1.3 Institutional Multiplicity

Institutional multiplicity describes any diversification of institutions performing a specific function. In the context of accountability networks, these institutions will be performing one of the three functions described earlier: monitoring, investigation or punishment. For instance, there will be multiplicity if there are multiple investigative bodies, that can operate simultaneously and independently of each other. While multiplicity can be found in any accountability system with at least more than one institution performing one of the three accountability functions, it can be structured in different ways. In performing the same function, these institutions may adopt distinct norms, procedures, and sanctions (or not). Graph 1 below used the Brazilian accountability network to provide an illustrative example of a system characterized by multiplicity.

GRAPH 1
Institutional multiplicity in the Brazilian accountability network

While many countries' anti-corruption systems are characterized by institutional multiplicity ( DAVIS, MACHADO and JORGE, 2014 ), it is unclear if this arrangement is more effective than other alternatives, such as the creation of specialized anti-corruption agencies. The response to these policy questions depends on a careful assessment of how institutional multiplicity works. From an empirical perspective, there is no certainty about how multiple institutions will interact with each other ( WEIJER, 2013 ). In some cases, such interaction may lead to conflict and inefficiency, while in others it may foster institutional competition, collaboration, complementarity, and compensation. Therefore, only a detailed empirical analysis can determine if multiplicity generates problems and, most importantly, under which conditions (if any) the benefits of multiplicity surpass its costs.

2 The Brazilian Accountability System to Combat Corruption: a Literature Review

There is a growing empirical literature on the Brazilian system to combat corruption. While there has been increasing attention to the topic, much remains to be done. This section shows how the concept of institutional multiplicity could be useful in systematizing research questions and setting up a research agenda.

To develop this claim, we map the existing literature, including high-level and granular empirical analyses of the Brazilian accountability network. Then, we point to the need for more granular empirical analyses, indicating how the concept of institutional multiplicity may be useful in this endeavour.

2.1 Three Types of Empirical Analyses of Brazil's Anti-Corruption System

As scholars turned their attention to the Brazilian anti-corruption system, several studies of particular institutions in the Brazilian accountability network were conducted. For instance, the National Court of Accounts ( Tribunal de Contas da União , TCU) functions as Brazil's leading government audit institution, but multiple scandals in the 1990s (e.g. Collorgate, São Paulo Regional Labour Court (TRT)) highlighted numerous procedural and institutional deficiencies in the TCU's functioning, which have been mapped in the literature ( LOUREIRO et al ., 2012 ; SPECK, 2011 ; SPECK and NAGEL, 2002 ; FONSECA, 2019 ). Along with TCU, the Office of the Comptroller General of the Union ( Controladoria-Geral da União , CGU) has also been the object of studies ( OLIVIERI, 2011 ; RICO, 2013 ). Responsible for internal accountability within the executive branch, one of CGU's initiatives is the Random Audits Program ( Programa de Fiscalização a partir de Sorteios Públicos ), which uses a lottery to randomly select municipalities whose books are audited to oversee the use of federal transfers ( FERRAZ and FINAN, 2008 , 2011 ).

While studies of isolated institutions are useful, they are the most voluminous and need to be complemented by more analyses of the anti-corruption network, i.e. empirical studies of how the different parts in the web of accountability institutions interact with each other ( ARANHA, 2018 , p. 4). Timothy Power and Matthew Taylor (2011a) are pioneers in this effort, not only calling attention to the importance of empirical analysis about the complex interactions between institutions in the Brazilian accountability system, but inviting contributors to an edited volume to map some of the points in the Brazilian accountability network that merit further attention. Questions raised in the volume include: how electoral courts interact with the federal judiciary ( TAYLOR, 2011 ), what is the relationship between the federal public prosecutor's office and the federal police ( ARANTES, 2011 ), and how do federal criminal justice institutions interact with the state criminal justice systems ( MACAULAY, 2011 ).

Along the same lines of Power and Taylor's edited volume, a series of scholars have developed studies mapping the Brazilian anti-corruption accountability network by identifying which institutions are part of it, and their different functions within the system. For instance, Da Ros centers his analysis on legal accountability, i.e. the sanctions created by statutes and imposed by officials authorized to enforce the law. After classifying legal sanctions into three ideal types: administrative, civil and criminal, he shows how they can vary in the type of punishment imposed and its severity. He then shows that the accountability process involves different institutional frameworks, depending on the severity of the sanction ( DA ROS, 2019 ). Based on that, one could hypothesize that the probability of the sanction being imposed is likely to be inversely correlated with the number of institutions involved. This, in turn, has relevant policy implications:

to invest in criminal sanctions as the main strategy of legal accountability for corruption is possibly an inefficient strategy: while [criminal] sanctions are harsher, they only apply to few and very specific situations, and only to individuals [in Brazil]. Moreover, [criminal] procedures involve the highest number of veto players and, as a result, [criminal] sanctions take longer and are less likely to be imposed in contrast with other types of legal accountability. ( DA ROS, 2019 , p. 1262, translation from the original by the authors)

This mapping exercise can be developed further to account for important details in the actual operation of these legal accountability systems. For instance, some scholars claim that the lack of coordination among different accountability institutions may explain the lack of robust results in the system ( TAYLOR and BURANELLI, 2007 ; MACHADO and FERREIRA, 2014 ).

It is worth noting that some of these studies have used the concept of institutional multiplicity to describe the Brazilian anti-corruption network ( MACHADO and PASCHOAL, 2016 ; MACHADO 2019 ; PRADO and CARSON, 2015 , 2016 ), exploring hypotheses about whether multiplicity has strengthened or weakened the system. For instance, Prado and Carson (2015 , 2016 ) hypothesize that Brazilian institutions' recent success in investigating corruption may be attributed, in part, to institutional multiplicity. They argue that functional overlap, at the very least, improves the likelihood of an investigation of suspicious cases. However, this hypothesis is hard to prove. The increase in Brazilian institutions' overall investigative capacity has occurred in parallel with initiatives that have significantly strengthened these investigative institutions, such as the federal police and the federal public prosecutor's office ( Ministério Público Federal , MPF). The former has received greater financial and personnel resources, a newly defined focus on corruption, and the publicity generated by high-profile operations; these have all contributed to the federal police's emergence as an increasingly potent force in fighting corruption in Brazil ( ARANTES, 2011 ). The MPF has also had an increase in financial resources (its budget doubled between 2004 and 2017 (LOPEZ and GUEDES, 2018)), increased independence since 2001 with a new process to choose the Prosecutor-General ( Procurador Geral da República ) and increased institutional capacity to conduct criminal and civil investigations (RAGAZZO and FREITAS, 2020). In this context, lacking a counterfactual scenario, it is hard to provide clear evidence of a causal relationship between institutional multiplicity and enhanced investigative capacity.

Another type of empirical studies about the Brazilian accountability network is what we call granular analyses. While a high-level analysis focuses on similarities (e.g. which institutions perform the same accountability function) or mapping of one function (e.g. what institutions are involved in punishment), a granular analysis focuses on details of the system operation and the complex interactions among different institutions. These granular analyses seem even rarer than high-level studies, but there are some concrete examples of this kind of analysis.

Alencar and Gico Jr. (2011) measured the overall effectiveness of the accountability network by comparing the number of administrative sanctions with that of civil and criminal sanctions between 1993-2005. They found that judicial (civil and criminal) accountability is much less effective than administrative accountability. ⁵ While this conclusion offers a starting point to consider strategies to fight corruption, it raises important questions about the mechanisms that could explain these findings. Could the administrative sanctions be the cause of the lack of judicial punishment (i.e. judges would be more resistant to impose a sanction if the actor has been punished already)? Or maybe there was a broken communication system, and not all administrative cases were being sent to the MPF? Uncovering the mechanisms that explain these results can provide valuable information on institutional alternatives to maximize the system effectiveness.

Another important study was conducted by Machado and Paschoal (2016) . Using qualitative analysis, the authors try to open the black box of internal operations within the accountability system. Using interviews with government actors, they uncover how the system operates. For example, they show that, in criminal cases, courts refuse to accept evidence collected in investigations conducted in other cases (administrative or civil), arguing that they are unlikely to comply with the procedural requirements of a criminal trial. ⁶ Along similar lines, administrative disciplinary procedures are not always sent to the MPF. ⁷ This piece of information raises a series of empirical questions that could be addressed in future studies. For instance, what is the percentage of cases that are not sent? Of the ones sent, how many are false positives (i.e. cases that do not involve violation of other laws)? Among the ones that do involve violations of civil or criminal legislation, how many are eventually prosecuted and punished? What informs the decision of a public prosecutor to further investigate and bring one of these cases to a court of law? Answers to these questions would provide valuable information to understand how the administrative accountability system interacts with the civil and criminal systems.

Aranha is perhaps today the leading scholar in granular empirical analyses of Brazil's accountability network. Through a series of studies that combine quantitative and qualitative data ( ARANHA 2015 , 2017 , 2018 ), she has engaged in an unprecedented and ingenious effort to map the flux of corruption cases within the Brazilian accountability network. Her research uncovers the existence of institutional coordination in some points of the Brazilian system and the dearth of coordination in others. Cases of coordination include monitoring institutions sharing information with institutions with investigative powers, joint investigative efforts (with each institution bringing their expertise) and formal cooperation agreements. However, existing coordination efforts are not homogeneous, do not necessarily result in higher rates of punishment, and can raise important questions about the problems derived from too much coordination, as illustrated by the Lava Jato case ( ARANHA 2020 ). While this is a wonderful example of the type of empirical analysis advocated here, we believe that there is still much more to be explored about the topic, and an even higher level of granularity could be added to this kind of analysis, as we illustrate in the next section.

In summary, the literature on the accountability network in Brazil needs to be further expanded. Understanding the complex interactions between institutions and sanctions within the accountability network can provide valuable information for policy makers, and may help combating corruption more effectively. A granular analysis that accounts for the details in each legal accountability system can make this understanding even more accurate and perhaps more useful for policy makers. In the next section, we argue that the concept of institutional multiplicity may be particularly useful to guide those interested in granular empirical analyses.

2.2 Using Institutional Multiplicity to Develop Granular Empirical Analyses

As mentioned earlier, the concept of institutional multiplicity points to multiple institutions performing the same function within the same system, often referring also to a plurality of normative orders. Methodologically, the concept simultaneously embraces a functionalist approach to identify the commonality in the functions performed by these different institutions (high-level analysis) and a detailed analysis of the different normative orders in which these institutions operate (granular analysis). While the political science literature on institutional multiplicity often uses the concept to describe the interaction between formal orders (sanctioned by the state) and informal orders (based on traditions, costumes and habits of a community), this article will use the term to describe formal normative orders within a legal system.

There are at least three ways in which the concept of institutional multiplicity can be useful in granular analyses of accountability networks.

First, there are different ways in which the interaction between different normative orders can be structured: i) different institutions applying the same rules, procedures and sanctions; ii) the same institution applying different rules, procedures and sanctions; and iii) different institutions applying different rules, procedures and sanctions. The concept of institutional multiplicity allows us to map the complex forms in which these institutions and normative orders interact and how individual actors navigate their differences and similarities.

This map raises interesting questions about Brazil's accountability network. Could the relationship between the disciplining bodies within the ministries ( corregedorias ) and CGU be described as a case of different institutions operating within the same normative order? Considering that the MPF can seek civil and criminal accountability, do these different normative orders interact within the institution? If so, how? Is there any interaction between normative orders that are distinct and managed by different institutions, such as administrative sanctions and criminal ones?

Second, the literature on institutional multiplicity also explores the different forms in which these institutions and their normative orders interact. For instance, in exploring how traditional authorities are integrated in the structure of the modern state in Africa, Goodfellow and Lindemann (2013) show that when formal and informal institutions have overlapping functions – called discordant institutional multiplicity – there is a higher probability of conflict. However, conflict is not inevitable. It depends on how the interactions between these orders are structured. Across the British Empire, colonizers adopted divergent approaches to the integration/separation of indigenous and transplanted legal institutions. In so-called integrated jurisdictions, the legal systems, while still functioning semi-autonomously, regularly interacted, drawing on one another and promoting the creation of a relatively consistent corpus of law that could be retained by the colony post-independence. In contrast, in parallel jurisdictional models, the segregation of the British and native legal regimes impeded mutually advantageous institutional adaptation and thwarted the realization of robust rule of law outcomes ( DANIELS, TREBILCOCK and CARSON, 2011 ).

These analyses raise interesting hypotheses about the interaction of different normative orders within an accountability system. For instance, when the administrative, civil and criminal accountability systems overlap, is there a risk of conflict between these orders (similarly to the phenomenon of discordant institutional multiplicity in colonized nations)? Do these systems interact regularly? If so, do they influence each other, as the traditional and transplanted systems in the British colonies? Or do they operate in isolation, as paralell orders? If so, how do these systems evolve?

Third, the literature on institutional multiplicity focuses on individual incentives for action, within a diverse institutional framework. As Tempel and Walgenbach indicate, “several scholars have pointed to institutional multiplicity as a fruitful way to integrate agency into new institutionalism” (2012, p. 232). Specifically, sociology has been using the concept to explore if the existence of more than one institutional option and heterogeneity of models of action can change the choices made by actors and individual behaviour ( CLEMENS and COOK, 1999 ). This can promote change, or simply reinforce the status quo ( PACHE and SANTOS, 2010 ; TEMPEL and WALGENBACH, 2012 ).

Does institutional multiplicity change individual behaviour? How? The creation or existence of alternative institutional paths can generate contradictions that destabilize the existing regularities of action. More specifically, institutional multiplicity has the potential to generate an external contradiction, i.e. the behavioural regularities observed in one institution are challenged by contradictory behavioural patterns followed by another institution. There are two potential mechanisms that can effect change in behavioural patterns: socialization and institutional incentives. Both have significant power in explaining observed changes, as illustrated by the following analogy: “if a mouse repeatedly takes the same path across a table, this regular path may be due to either the presence of a maze that obstructs many possible changes in direction or to effective socialization through behaviour modification. However, since the mouse may be well socialized and in a maze, these ‘institutionalisms’ are properly understood as complements, rather than mutually exclusive explanations” ( CLEMENS and COOK, 1999 ). Thus, while the existence of multiple institutional avenues may offer alternative paths (the new “maze”) that may modify behaviour, it does not exclude the possibility that these changes may also be caused at least in part by informal mechanisms, such as the socialization of actors in a different institutional culture.

This raises a number of interesting questions for analyses of the accountability network. How does the existence of multiple institutions performing monitoring, investigative and sanctioning functions in different ways influence the behaviour of individual actors inside and outside the accountability system? Are the mechanisms formal or informal rules and norms? The concept of institutional multiplicity enriches these analyses by offering hypotheses that can be tested to determine how multiplicity influences individual agency.

By adding these three dimensions to the analysis of the multitude of systems in the accountability network, the literature on institutional multiplicity can illuminate how the system works. The following section presents a research agenda for accountability networks based on the concept of institutional multiplicity, using the example of the accountability of public employees.

3. Accountability Networks and Institutional Multiplicity in Brazil: Concepts and Research Questions

The previous section showed how scholars have been devoting more attention to the Brazilian anti-corruption system, and how this literature could benefit from the concept of institutional multiplicity to develop empirical studies that map how the network of accountability works. Building on this, this section will focus on the accountability of public servants in Brazil, proposing a research agenda for empirical analyses. To develop this agenda, we: i) provide a description of the accountability system of federal workers as a system characterized by institutional multiplicity; ii) identify methodological challenges to design research about accountability networks (using some existing studies as illustrative examples); and iii) propose concepts that may help to formulate questions for empirical research about the network's inner workings.

3.1 Accountability of Federal Employees in Brazil

In general, civil servants are subject to three types of legal accountability: administrative, civil, and criminal ( DA ROS, 2019 ). In Brazil, the accountability system for civil servants can be described as one in which there is institutional multiplicity because different institutions may share the same accountability function. For instance, both the police and the MPF can investigate criminal cases. Similarly, both ministries' internal institutions and the CGU can perform investigative and sanctioning functions in administrative cases.

In addition to the multitude of institutions involved in each type of accountability, the accountability of federal employees also involves different normative orders. More specifically, there are statutes that govern each type of legal accountability, covering different conducts, following different procedures, and imposing different sanctions.

Each of these three types of accountability sanctions a set of acts. There is great overlap in what each accountability type prohibits, but different statutes often use distinct legal classifications for the same conduct. Administrative accountability applies sanctions to misconduct listed in the Public Servants Act (Law 8.112), which range from absenteeism to bribery ( BRAZIL, 1990a ). Civil accountability can be divided in two. First, the Administrative Improbity Act (Law 8.429) punishes illicit personal enrichment, damage to the public budget, and violation of administrative principles, presenting a non-exhaustive list of acts to illustrate each of these violations ( BRAZIL, 1992 ). Second, civil accountability can involve general tort law (civil wrong) to ensure reparation for any damage caused. Criminal accountability, in turn, is regulated by the Criminal Code ( BRAZIL, 1940 ), which lists crimes related to corruption, such as to accept or demand a bribe (art. 317), offer or pay a bribe (art. 333), embezzlement (art. 312), graft (art. 316), influence peddling (art. 332), malfeasance, misfeasance, and nonfeasance in office (art. 319).

For each accountability type, different investigative and decision-making procedures are set in motion when there is a suspicion or evidence of wrongdoing. Administrative accountability involves solely institutions within the executive branch. For investigations taking place inside an office of the executive branch, e.g. a ministry, two or more federal civil servants are appointed to an ad hoc investigative committee ( sindicância ). If there is probable cause, another committee is formed to try the accused. The procedures are rather informal and not heavily regulated, ⁸ but the accused may be represented by a lawyer. The trial committee then writes a final report, indicating if they find the accused guilty or innocent, and if guilty, recommending punishment. Harsher punishments require a second report, from the internal legal department of the office the accused is affiliated with. The report(s) is(are) then forwarded to a decision-maker, who is a high-level official within the executive branch. The decision-maker varies depending on the punishment suggested. For example, in case of dismissal (one of the harsher sanctions in administrative processes), the office of the presidency makes the final decision. The final decision may be subject to judicial review, but the judiciary tends to limit its analysis solely to procedural matters. ⁹

In turn, civil accountability for administrative improbity has a different set of procedures. It requires an indictment by the MPF or the office of the executive branch affected by the wrongdoing. Before proceeding to trial, the MPF may conduct its own investigation ( inquérito civil ). In contrast, civil accountability based on tort law requires an indictment by the federal government against the employee. These lawsuits can include damages caused by the employee to the government or to a third party (e.g. if a third party sues the government for damages caused by an employee, the government can sue the employee to obtain reparation). Finally, criminal accountability depends on an indictment by the MPF and the trial is preceded by a criminal investigation conducted by the police, sometimes in collaboration with the MPF ( inquérito penal ).

As to sanctions, the administrative procedures can culminate in penalties that range from warnings to dismissal. Civil accountability based on tort law may lead only to compensation for damages. Civil accountability based on the Administrative Improbity Act is much broader, consequences ranging from forfeiture of assets to fines, and includes dismissal and the suspension of political rights (i.e. prohibition to run for office). Finally, criminal accountability may lead to penalties such as forfeiture of assets, fines, community service, temporary suspension of rights (e.g. suspension of professional licence) and prison sentences. Depending on the crime committed, the punished individual may also be fired and required to pay compensation for damages.

In summary, by unpacking the details of each normative order that exists within Brazil's accountability system for federal employees one can identify significant differences and also important overlaps. Using the concept of institutional multiplicity, the next subsections will discuss how these differences and overlaps present challenges and raise interesting questions for researchers.

3.2 Institutional Multiplicity and its Methodological Challenges: Defining Corruption

The existence of several laws punishing civil servants for corruption in Brazil indicates that this is a system of institutional multiplicity characterized by a variety of normative orders. This multiplicity of laws poses a methodological challenge to empirical investigation of the functioning of the network: by comparing and contrasting different types of accountability, how can researchers be sure that different legal categories are commensurable? This will be particularly pressing for projects that need to define a priori which unlawful behaviour will comprise their sample. A list of challenges follows.

First, as in any research on corruption, the roll of legal provisions that can be labelled as combating corruption will depend on the definition of corruption that one adopts. For instance, campaign finance laws may not prevent the “use of public office for private gains” but it may prevent what is known as political corruption, i.e. the “use of public office for partisan gains”. Therefore, campaign finance laws can be considered instruments to combat corruption if a broader definition is adopted. Another example is lobbying, which is legal in the United States but could be classified as what Lessig names “dependence corruption”, which occurs “when individuals within that institution become dependent upon an influence that distracts them from the intended purpose of the institution” ( LESSIG, 2011 , p. 15).

The second challenge is that not all situations usually classified as the “use of public office for private gain” are necessarily prohibited by law. Indeed, most countries prohibit conducts that fit into the traditional definition of corruption, such as bribes and embezzlement ( FISMAN and GOLDEN, 2017 , p. 26-29). However, some countries have laws that explicitly allow behaviour classified as corrupt under the traditional definition, such as sending gifts to public officials. As a result, an act may not be prohibited by law, despite being perceived as corruption by citizens, or despite not advancing the public interest ( GARDINER, 2017 ; SCOTT, 1972 ). On the other hand, a broad definition of corruption that goes beyond what is legally prohibited may dilute its value and may not be enough to spark the complex institutional reforms required to tackle them ( ROSE-ACKERMAN, 2018 ).

The third challenge is that, as Davis (2019) points out, legislative provisions may prohibit conduct that could be classified as “the use of public office for private gain,” while not labelling it “corruption”. This means that researchers may need to construct samples for empirical research by navigating a complex legal array of distinct legal provisions. We will focus the remaining of this section on this third challenge.

In Brazil, scholars have to choose, among the many conducts classified as administrative infraction, improbity and crime, which ones amount to the researcher's definition of corruption. The first challenge researchers must deal with is the fact that conducts described as administrative infraction, improbity and crime may or may not fall into the category of “use of public office for private gain”. For example, absenteeism is an administrative infraction but not a corrupt act. Because government search engines and databases are based on legal classifications, the researcher needs to define which of these classifications can be considered corruption. In other words, a sampling process guided by a theoretical concept of corruption has to deal with cases arranged by a folk concept of corruption.

The following paragraphs deal with this challenge by looking at three empirical studies that constructed their samples using data from the three types of accountability: administrative infraction, improbity and crime. Our goal is not to criticize these works, but to show how they illustrate the inherent challenges of dealing with legally-based data on corruption cases.

We begin with a research that used data based on administrative definitions. Alencar and Gico Jr. (2011) evaluated the Brazilian judicial system's efficacy in punishing corrupt public servants. To do so, they first identified cases in the federal gazette ( Diário Oficial da União ) in which employees were dismissed from a tenured or appointed position, or lost pensions. To choose which of these cases are related to corruption, they selected six categories listed as corruption in the Public Servants Act: abuse of position, administrative improbity, ¹⁰ damage to the public budget, receiving advantages, corruption, and administrative advocacy.

As any other selection criterion, Alencar and Gico Jr.'s had to decide which categories classified as non-corrupt practices and exclude these. However, this exercise still involves risks. For instance, the Public Servants Act does not define most administrative infractions clearly. “Corruption”, for example, is not defined at all. As a result, law enforcement officers may be classifying a particular behaviour as “corruption” under the Public Servants Act, but that behaviour may not fit the definition of corruption adopted by scholars. Also, Alencar and Gico Jr.'s choice may have excluded other acts that could be defined as corruption. For example, their sample excludes cases in which public employees allow someone else to perform their function. Yet, this could be an instance of “use of private office for public gain” if the employee had contracted out the service provision, paying a fraction of their salary to the person actually delivering the service. In summary, the lack of clarity on legal classification of conducts provided by statutes makes it hard to select corruption-related cases.

Research based on data on the first type of civil accountability (administrative improbity) faces a similar challenge. For example, Santos (2014) studied the impact of the creation of the National Justice Council (CNJ) on the speed and severity of improbity procedures, with a separate analysis of those related to corruption. Using a CNJ search engine, Santos collected cases of administrative improbity. To select those related to corruption, he chose the ones legally classified as improbity by illicit personal enrichment, defined by law as “to gain any kind of illicit patrimonial advantage due to one's position, mandate, function or job or activity” ( BRAZIL, 1992 , art. 9). According to Santos (2014 , p. 27), this type of improbity fits the concept of “abuse of public office for private gains”.

However, similarly to Alencar and Gico Jr.'s, this methodological choice brings some challenges. First, it excludes cases in which corrupt agents gained non-economic advantages. Second, the legal definition is so vague that acts that involve illicit personal enrichment could also be investigated and punished under other legal provisions. For example, improbity by damage to the public treasury is defined as the “loss, misuse, misappropriation, or giving discount, dilapidation of the property or assets.” Thus, embezzlement may be classified either as illicit personal enrichment or damage to the public treasury, but the latter was excluded from Santos's sample. As in the case of administrative accountability, the lack of precision in the Administrative Improbity Act challenges researchers interested in civil accountability.

In contrast, defining what crimes amount to corruption is an easier task. Criminal laws tend to be clearer and more precise. For example, the difference between the crime of bribery (labelled “corruption” by the Criminal Code) and embezzlement is much clearer than the difference between improbity by damaging public treasury and improbity by illicit personal enrichment. In this vein, Lecovitz suggests a list of 18 crimes related to public corruption ( LECOVITZ, 2014 , p. 22). While he also had to face choices, producing such a list of crimes is easier than dealing with the vagueness of the administrative and civil accountability laws. For example, Lecovitz argued that he excluded tax-related crimes committed by public servants – such as receiving a bribe to charge taxes (art. 3, II, Law 8.137/90) – from his sample, because they did not fit his definition of corruption (BRAZIL, 2019a).

While more precise, criminal accountability also presents challenges to researchers. In recent work, Madeira and Geliski (2019) investigate criminal corruption cases brought to federal courts. Unlike the studies previously cited, the authors did not resort to a definition of corruption to define their sample. Rather, they searched for the word “corruption” using a federal court research engine. While this strategy is useful to show what crimes judges associate to corruption, selecting cases based on a folk classification may include behaviour that does not fit into a theoretical concept of corruption. For example, the most frequent crime in their sample is smuggling, which usually does fit into the social sciences' definition of corruption. Also, it is possible that their sample missed other corruption-related crimes. For example, in ruling on a case of embezzlement, it is possible that judges would not use the word corruption because this term, in criminal law, is used for bribery.

These challenges regarding sampling are augmented in projects focused on institutional multiplicity because often these scholars are comparing and contrasting multiple accountability systems. In order to study cases that have been subjected to different types of accountability, a researcher not only needs to ask if a legal provision fits the concept of corruption adopted in the project, but also how the classification used in one type of accountability process (and provided by one statute) translates into a classification provided in another process (and often another statute). As Alencar and Gico Jr. (2011) show, administrative cases hardly ever reach the criminal and civil accountability spheres, but should they? The answer depends on determining empirically how many of these administrative cases were also considered crimes. In sum, the concept of institutional multiplicity points to the difficulties of studying corruption cases subjected to different types of accountability.

3.3 Multiplicity as a Question for Empirical Institutional Research

Besides the conceptual questions presented in the previous subsection, the concept of institutional multiplicity also helps formulating research questions about the anti-corruption accountability network. To formulate these questions, we propose a series of questions based on three different aspects of institutional multiplicity.

Multiplicity in function highlights the fact that institutions perform different accountability functions: monitoring, investigating and sanctioning, but investigating these can be a complex task if one considers the multitude of normative orders embedded in these systems. Multiplicity from the perspective of law enforcers emphasizes that there are rules governing institutional multiplicity, which can be imposed by statutes (exogenous) or created by law enforcers (endogenous). Institutions and individuals in charge of enforcing the law design different strategies to deal with institutional multiplicity, but these often involve a combination of interpreting and enforcing exogenous norms and making their own, endogenous norms, to orient and regulate their work in a multiple system. Multiplicity from the perspective of the actors held accountable underscores the fact that not only accountability institutions but also those investigated and punished have to deal with a multiple system (both private and public actors).

It is worth noting that these three dimensions of multiplicity do not represent discrete and independent realities. Instead, they are analytical cuts that may facilitate the analysis of the phenomenon in some cases, but in the cases in which these dimensions are intertwined it may be difficult to make analytical choices based on the distinctions proposed here. Acknowledging this limitation, the next subsections present research questions for each of these dimensions of institutional multiplicity.

3.3.1. Multiplicity in Function: Investigation and Punishment

In order to understand the functioning of the Brazilian accountability network in the fight against public servants' corruption, we argue that it is important to examine: i) in which institutions investigations often start (and how that influences case outcomes); ii) how information flows among institutions; iii) how the actions of one institution influence the functioning of other institutions.

How do investigations begin? In theory, investigations can be initiated by the executive branch, the police, or the MPF. We can hypothesize that most investigations start in the executive branch (inside the public administration), since that is the closest institution to employees and thus the one best able to monitor their daily activities. The mechanisms that may allow for the executive branch to be an effective monitor may include the employees' direct superiors or an internal whistleblower system (e.g. the institution's ombudsman, ouvidoria in Portuguese). However, this is a hypothesis that needs to be empirically verified.

Even if we can determine that most investigations start within the executive branch, it is also important to examine in which cases they start somewhere else. This knowledge can reveal the strengths and weaknesses of different institutions performing investigative functions. For example, Prado and Pimenta (unpublished) suggest that when there is systemic corruption within the executive branch (i.e. when high-ranked employees are involved in corruption schemes), independent bodies such as the police and the MP are better equipped to deal with the case. Another hypothesis is that whistleblowers outside of the executive branch, such as private actors whose activities depend on the corrupt institution, could prefer denouncing corruption to the MPF or the police rather than to the executive branch itself, due to fear of retaliation from the institution. In sum, it is essential to assess where and how investigations start.

The second point worthy of inquiry is how different institutions interact after an investigation starts. Machado and Paschoal (2016) propose five forms of interaction between institutions: indifference; competition; independence with indirect recognition; independent coordination; mutual work. This mapping provides theoretical categories to investigate which interaction prevails in each situation. To understand the functioning of the accountability network, it is important to map empirically the nature and timing of these cross-institutional interactions. For instance, how often each of these forms of interaction happen? Which institutions are involved? What are the mechanisms that allow for these interactions, i.e. once an investigation starts in one institution, how do other institutions get involved?

While the Brazilian legislation establishes general norms to regulate the information flow among different institutions, it also grants a great deal of discretion to those in charge of investigations ( MACHADO, 2019 ). Given that discretion, empirical research is required to understand how the system actually works. We can analyze what happens in practice through the mapping of cross-institutional interactions. An example is the study of CGU's lottery program described in section 2 ( ARANHA, 2018 ). To conduct this kind of analysis, we must learn what circumstances or mechanisms lead institutions to interact in different ways and how that affects case outcomes.

This leads to a series of research questions about institutional interactions in the accountability system for civil servants.

First, how do cases move from the administrative to the criminal sphere of accountability? The law establishes that if an investigation starts inside the executive branch through an administrative disciplinary procedure ¹¹ and the report concludes the actions committed were criminal, the officials involved in the disciplinary procedure must inform the MPF. At the same time, CGU allows officials to inform the MPF about the case before the administrative trial has ended ( BRAZIL, 2019 , p. 23). How do public servants, most of them with no legal background, decide if an administrative infraction constitutes a crime? How often does the internal legal department participate in this decision?

Moreover, it is important to investigate under which circumstances the executive branch informs the MPF about the case before the administrative trial has a final report. Would they consider, for example, the statute of limitations, so that the notification is timely to allow for a criminal trial to proceed? Would the decision be based, instead, on the fact that the MPF (or the police) may have legal tools and investigative powers that are not available to the executive branch? These tools include the temporary layoff of the accused to prevent them from meddling with evidence, seizure of assets and preventive detention. In an investigation that starts at the administrative level, the accused might be able to destroy evidence before more restrictive actions are used, which can affect the case outcome. ¹² It is interesting to note that while the first set of circumstances could reveal a concern with the proper functioning of criminal accountability, the second seems more related to an instrumental use of the criminal system to allow for the administrative accountability to function properly. Uncovering which motivation guides the actions of accountability officials can provide information about multiplicity that one cannot obtain with an analysis of legislation or cases.

Another important question is what happens once the MPF and the police are informed about the case. When and how does this occur? What do these institutions usually do in such cases? Do they wait for the administrative procedure to end? Do they start their own investigation? How are these decisions made? In sum, it is important to understand how information flows from the executive branch to other accountability institutions, and how it is processed.

Likewise, we must consider information flows in civil accountability. ¹³ In cases of improbity, the law establishes that the MPF and the TCU must be informed when an administrative disciplinary procedure starts. In these cases, the questions above also apply. How does the public servant in charge of the investigation decide whether the case constitutes improbity? Given the lack of clarity in the legal provisions (see subsection 3.2) about what can be considered improbity, potentially any administrative violation could meet the requirements. Yet, arguably not every administrative disciplinary case is forwarded to the MPF. It is possible that some instances of actual improbity never make it to the MPF, which could lead to impunity; or that too many cases that are not improbity reach the MP, overburdening the institution.

Once the case reaches the MPF, it is important to assess how the MPF reacts. Firstly, the law establishes that the MPF might request to observe the disciplinary administrative procedure (Law 8.112). Does this occur? In which cases? How do public servants in charge of the procedure perceive the MPF's participation in the case? Secondly, when the MPF is informed of an administrative procedure, does it wait for the end of the procedure to act? Or does it start a civil inquiry immediately? How is this decision made? For example, it is possible that the MPF considers that the administrative report provides insufficient information to determine whether the case must be brought to a court of law. If the MPF decides to conduct its own investigation to determine if there was wrongdoing, there might be an unnecessary duplication of investigative work. That is, had the executive branch done a more reliable investigation, the civil inquiry might have been unnecessary. Thirdly, the law allows for the executive branch itself to file a lawsuit. When does this occur? Is there any variation in the outcome of cases brought by the MPF or by the executive?

The literature on institutional multiplicity emphasizes not only the interactions between institutions, but also how multiple normative orders may influence the behaviour of actors, effectively allowing for one institution and its norms to influence other institutions ( CLEMENS and COOK, 1999 ). In the Brazilian example, one could consider how the investigation and punishment of actors in one institution influence the behaviour of actors in another accountability institution. Accountability spheres are legally independent, and it is possible for a civil servant to be convicted in one sphere and acquitted in another through independent processes and decisions. However, it is possible that the conviction in one sphere affects decisions in other spheres. For example: if a civil servant is harshly punished in the administrative sphere, does that influence investigation or punishment in other spheres? Are those in charge of civil/criminal investigation or punishment considering the harsh administrative sanction to decide how to handle the case? If so, how does this happen: are actors in other spheres more likely to condemn someone who was harshly punished, or do they consider the administrative sanction enough and decide not to investigate or punish any further? While legally actors in all spheres – administrative, civil, and criminal – are required to investigate wrongdoings, in practice some actors might choose to save time and resources when a case is being handled by another institution. A comprehensive empirical inquiry could reveal how often, and in which circumstances, this logic applies. For example, it is possible that in widely publicized corruption scandals, all institutions feel obliged to act. Understanding these dynamics is essential to understand the functioning of accountability networks.

Another matter that requires attention is the expectations of individuals within each institution. For example, when the MPF is informed of a conduct that possibly constitutes improbity, it can request an investigation by the executive branch. Does this request influence in any way how the investigation is conducted or its outcome? For instance, is the request by the public prosecutor's office interpreted by executive branch officials as a signal that there is wrongdoing? In that case, are the actors tasked with the investigation at the administrative level more inclined to convict the accused, due to a belief that there is an expectation from the MPF? On the other hand, it is possible that officials in the executive branch may see the MPF's request as an intrusion in their sphere. If so, does this make them more inclined to acquit the public servant? The same can be asked about the MPF when a court requests an investigation. Again, these questions illustrate the importance of granular empirical research to understand how the law works in practice.

Finally, there is a series of interesting questions about direct influences. Judicial review provides an interesting example. Knowing that the administrative disciplinary process can be annulled in a court of law for failure to comply with procedural requirements, do the officials within the executive branch modify their behaviour? Are they considering the possibility of a judicial appeal when deciding which procedure to follow, and which defence opportunities and guarantees to offer to the accused? In sum, it is important to learn how each institution's behaviour is shaped by the norms and expectations of other institutions.

3.3.2. Multiplicity from the Perspective of Law Enforcers

The previous subsection presented empirical questions about how accountability institutions exercise their functions within a system characterized by multiplicity, i.e. a system with a multitude of institutions and distinct and independent normative orders. This subsection turns to another dimension of institutional multiplicity: the rules that govern the interactions of these multiple orders. In general, institutional multiplicity systems are governed by exogenous norms established by statutes. Depending on how precise or vague these may be, institutions or individuals may create their own endogenous norms to navigate the institutional maze. We propose that, apart from investigating the practices embedded in everyday activities surrounding investigation and punishment of corruption cases, a research agenda must also look at norm-producing activities (formal and informal). In other words, it is important to investigate how institutions create endogenous norms to guide their actors' behaviour vis-à-vis a multiple system of exogenous norms (the law). Determining which institutions and actors are involved in norm-making, their interests and concerns, and the resistance they face (if any), might provide an interesting map of the inner workings of the accountability system.

This matter is particularly relevant to the Brazilian case, since federal laws that govern the system's institutional multiplicity do so in vague and broad terms, granting significant levels of discretion to actors involved in the investigation of corruption cases in each institution. This in turn leads to the creation of internal, institutional-specific norms to deal with multiplicity. Understanding these norms, which can be informal in some cases, is fundamental to understand the workings of the accountability network.

Research questions can be divided in those focusing on the process of norm creation – or their absence – and on the impact of these norms on the system. The first topic brings questions such as: How often are norms to deal with multiplicity created and used? Do they tend to be formal or informal? What type of institution creates the norms? Do norms emerge from actors' daily activities and experiences or are they established by high-ranked subdivisions within institutions? Are other institutions consulted in the process of norm-making? Is there an effort to coordinate new norms with other institutions' formal norms and informal practices?

A second set of questions relates to the impact of these norms on accountability institutions' activities. The presence or absence of internal rules and norms, despite appearing obscure to the outside observer, strongly inform how accountability networks operate. If there is an absence of norms, do players in the system perceive this to be dysfunctional? How does the system work, in the absence of norms? Are there spontaneous forms of coordination between actors, do they act in complete isolation with no regard for what is happening in other institutions, or does the absence of norms create clashes and tensions that could be, in theory, prevented by some governance structure?

In Brazil, the CGU, AGU, and specific accountability departments within the executive are examples of institutions involved in norm-making activities that should be studied. For example, an investigation of wrongdoing by a civil servant may start at different institutions. It can start with an administrative inquiry at the executive branch, a criminal inquiry through the Federal Police, or a civil inquiry by the MPF. As previously mentioned, this multiplicity leads to questions about institutional coordination and collaboration. But it also leads us to wonder how institutions create internal norms to deal with multiplicity during the investigation. For example, do administrative institutions have norms on how to share the investigation with the police and the MPF?

One example of how institutions have dealt with this question is the creation of Corruption Fighting Divisions ( Núcleos de Combate à Corrupção ) within MPF units ( Procuradorias da República ). While normally the investigation of improbity and crime is decentralized and uncoordinated, some units have unified both activities under specialized divisions ( QUEIROZ, 2018 ). This allows for the same prosecutor to investigate the same case in different spheres (civil and criminal). According to Queiroz (2018) , there are two types of MPF internal norms that regulate this institutional design: i) the rules that establish the mixed Corruption Fighting Divisions, tasked with investigating cases involving improbity, crime, or both; and ii) the rules that establish the regular Corruption Fighting Divisions, tasked with investigating cases in which there is both improbity and crime. This example of variation within the governance structure raises broader questions on the circumstances in which different institutions create internal formal norms to regulate multiplicity.

Let us consider another example. The Public Servants Act (Law 8.112, art. 154) provides that “in the occasion that the investigative report concludes that the infraction is a crime, the responsible authorities shall forward the casefile to the MPF” ( BRAZIL, 1990a ). Given that not every disciplinary administrative infraction constitutes a crime, and that public servants are not trained in criminal law, it is possible that institutions establish guidelines to guide officials in charge of administrative accountability. Determining if these guidelines exist, who set them up and what is their content is important to understand the functioning of the system. Alternatively, institutions might designate specific actors to investigate these cases, such as their internal legal departments. Does the involvement of personnel with legal training increase the precision in the analysis and selection of cases, making it more likely that the cases that reach the MPF are actually those involving criminal activity? It is also possible that institutions create bilateral partnerships to better coordinate their efforts. For example, are there specific norms within the MPF and the executive branch that establish a communication channel?

In addition to investigating what kind of rules exist, it is also possible to analyze if they are effective. In some cases, internal rules and guidelines may strongly influence the behaviour of individual actors, but in others they may be simply ignored. For instance, if there is an internal system to facilitate the communication between the investigator at the administrative level and the MPF when the act is deemed criminal, it is worth asking if the system is often used, by whom and what are the results. It is also interesting to look at the other end of this equation, i.e. the institution receiving the information, to evaluate how it is handled. Are there effective norms in that end too? This can help us understand how the accountability system works in Brazil.

3.3.3. Multiplicity from the Perspective of the Actors Held Accountable

So far, we have discussed multiplicity from the perspective of the accountability institutions. Yet, a matter still incipient in the literature concerns the impact of institutional multiplicity on those investigated and punished. Discussions about institutional design and the system's efficiency would benefit from a research about the impact of multiplicity on the experiences of the accused public servants. How do those who are investigated and punished react when being held accountable by multiple institutions?

This matter is essential to evaluate the fairness and efficiency of the accountability system. It is possible that multiplicity increases the likelihood of punishment ( PRADO and CARSON, 2016 ), having a greater deterrent effect. But it also increases the probability that the accused will be punished in multiple spheres. Moreover, given that going through an investigative procedure generates financial and emotional strain for the accused (even if a sanction is not applied), it is likely that going through multiple procedures multiplies this effect. Moreover, repeated interactions between client and lawyers can have negative effects ( SARAT and FELSTINER, 1995 ). The procedural justice literature shows how trust in institutions is highly affected by the perception of the accused on how they are treated ( TYLER, 1988 ). This perception can affect individual behaviour toward norms. Understanding how accused civil servants experience multiple investigations and processes, and which impact this has on their attitudes toward norms and their trust in the system can foster debates around the fairness and legitimacy of systems characterized by institutional multiplicity.

Let us consider an example. A corrupt civil servant might be required to go through two or more of the following: a disciplinary administrative procedure at the institution where they work; an improbity lawsuit in the civil sphere; and a criminal procedure. Each one of these procedures involves the participation of the accused, and some require legal representation. For example, an improbity lawsuit might require them to testify at the MPF in a civil inquiry, their participation in several hearings throughout the investigation, and the hiring of a lawyer. A criminal procedure might require that the accused testify to the police, participate in hearings, and hire a different lawyer. An administrative procedure, although more informal than judicial lawsuits, also requires the participation of the accused in an inquiry at their workplace, where they are investigated and judged by their co-workers. Although this procedure does not require a lawyer, the accused might choose to hire one. Moreover, administrative procedures might be judicialized later and they frequently are, in cases where the outcome is a layoff ( ALENCAR and GICO JR., 2011 , p. 87).

When a civil servant goes through multiple procedures, they are required to talk about the same facts in different institutions, hire and interact with different lawyers, and deal with potential punishment at different stages of the investigation. This situation raises empirical questions: What is the financial cost, for the accused, of multiple procedures in more than one sphere? How does the accused perceive the consequences of dealing with multiple procedures? What is their relationship to lawyers? What are their biggest concerns – forfeiture of assets, loss of employment, stigma, penal punishment, etc.? Is there variation in these matters among civil servants from different institutions?

These questions affect not only civil servants accused or punished. The civil servants who staff the investigative committees and judge their co-workers are also working under institutional multiplicity. As such, civil servants are in a peculiar position as potential investigators and accused in the anti-corruption accountability system. This leads us to wonder how the questions posed in the paragraph above apply for civil servants in the position of accusers. For example: How does knowing that a co-worker will go through multiple procedures affect the way actors conduct investigations at the administrative level? Does the investigator believe they must be less rigorous, knowing the accused will go through multiple procedures? Or does the knowledge that the event will be investigated in other spheres lead the investigator to be more rigorous? Does this knowledge impact the investigator's decision to classify an act as crime or as improbity?

Conclusion

Fighting corruption has been a central theme in policy circles in the last decade, and Brazil is not an exception. The rise and fall of the Lava Jato case has sparked even more intense debates and analysis about what is required to design an effective system to combat corruption. In an attempt to contribute to the specialized literature on the topic, this paper argued that legal scholarship can strongly benefit from detailed and descriptive empirical analysis of the existing system. Understanding the inner workings of the system allows for a more informed and fruitful discussion about where and how it can be improved.

To develop this claim, the paper delved into the Brazilian accountability system to combat corruption. The Brazilian system has been receiving increasing attention by scholars recently, and a number of important empirical studies have analyzed it. In addition to offering a map of these efforts (classifying these studies into three types), we indicated that analysis of the network (as opposed to analyses of a particular accountability institution) are still rare, and the granular detailed analyses that seem particularly helpful for lawyers are rarer still. This article hopes to have convinced the readers that legal scholarship can strongly benefit from more granular and descriptive empirical analyses of this system.

The paper's main argument was the concept of institutional multiplicity, and the literature associated with it can help guide a research agenda for Brazil's accountability network. While preserving the idea that these normative systems are a form of accountability, the concept of institutional multiplicity also draws attention to the different institutions and normative orders involved in different forms of accountability, the complex ways these normative systems may interact with and influence each other, and how these differences may affect the behaviour of individual actors.

Building on the concept of institutional multiplicity, the paper focused on the accountability systems for federal employees to design a research agenda for empirical research. The system is characterized by multiplicity (as there is administrative, civil and criminal accountability), which includes not only different institutions, but also different norms, statutes and sanctions for each type of accountability. We start with the methodological challenge of identifying what classifies as corruption, when legal categories across each of these types of accountability are not identical in law and in practice. Then, we move to questions about how institutions in charge of investigation and punishment interact with each other, and may influence each others' actions. We end this section with a discussion of the impact of multiplicity on the accused, a theme that has not yet received much attention in the literature.

Despite focusing on civil servants, the research agenda proposed here is not intended to be limited to them. Rather, it is designed as an illustration of specific research questions and concepts that may inform and/or inspire the study of the accountability of other actors (e.g. private actors, civil servants at the state level, etc.). Also, despite being descriptive, the research agenda proposed here is designed to uncover information that can be useful for normative analyses and proposals for reform. Sometimes mapping what is working or not is an important first step to fix the system. And in this regard, institutional multiplicity constitutes a unique vantage point, since it offers different instances of comparison. Indeed, the presence of multiple institutional referents “enlarges the toolbox from which reformers can draw in crafting new solutions, facilitating deeper change” ( ANDREWS, 2013 , p. 182). Thus, while the research questions proposed in this paper are descriptive, the answer to these questions can inform important normative concerns and inspire proposals for reforms.

In summary, while a modest contribution, with the research agenda proposed here we hope to have chartered a fruitful way to enrich legal scholarship on combating corruption.

• 1
  This section is based on Trebilcock and Prado (2014) .
• 2
  Transparency International, on the other hand, defines corruption as “the use of entrusted power for private gain”, which also includes what is known as private corruption (TRANSPARENCY INTERNATIONAL, [s.d.]).
• 3
  Forms of corruption that involve non-economic benefits include clientelism or patronage. These take place where public resources and political processes are manipulated to increase personal power. This is a form of “institutional corruption”, which generates gains in a way that benefits officials in their capacity as public officials rather than in their personal life (individual corruption) ( HEIDENHEIMER and JOHNSTON, 2002 , p. 26-42).
• 4
  As Power and Taylor (2011 , p. 9) argue, “the study of accountability and political corruption poses a great challenge to political scientists. For academics accustomed to inhabiting narrow institutional bailiwicks (…) the question of how to fight corruption requires a major readjustment of analytical perspective. Rather than focusing on institutions in isolation, we are forced to switch to a wide-angle lens and a more broadly systemic approach”.
• 5
  After Operation Car Wash (Lava Jato) in 2014, these numbers may have changed. For more details about this case, cf. Lagunes and Svejnar (2020) .
• 6
  Machado and Paschoal (2016 , p. 34): “Hoje, grande parte dos juízes rejeita o intercâmbio de provas entre procedimentos de diferentes áreas, argumentando que as regras para produção de provas nas esferas civil e administrativa são diferentes das da esfera criminal, estas muito mais protetivas dos direitos do réu”.
• 7
  Machado and Paschoal found that “often, administrative accountability procedures, when finished, are sent to the MP when there is evidence of possible violation of civil and criminal law (interview 3)” (2016, p. 25).
• 8
  The Public Servants Act (Law 8.112) does not regulate this procedure in detail. For further reference see Brazil (2019, p. 58-65).
• 9
  Alencar and Gico Jr. (2011) found that more than half of all public servants fired due to administrative procedures file for judicial review, but are rarely successful in reverting the case outcome.
• 10
  Although administrative improbity is also part of civil accountability, the Civil Servants Act (Law 8.112) lists it as an administrative infraction. However, unlike the Administrative Improbity Act (Law 8.429), the Civil Servants Act does not define administrative improbity.
• 11
  There are other types of investigation at the administrative level regulated by decrees and ministerial norms, but we focus only on one regulated by statutes (disciplinary administrative procedure) since it is a well-known form of administrative inquiry.
• 12
  In theory, in order to avoid such problems, the CGU can claim responsibility for the procedure in cases with more notoriety.
• 13
  Here we use an example related to improbity because it raises more complex questions than lawsuits seeking damages, but the latter surely deserve to be empirically studied.

ACKNOWLEDGEMENTS

The authors would like to thank the financial support of the Social Sciences and Humanities Research Council (SSHRC), and Bruno Monteiro for the research assistance.

REFERENCES

• ALENCAR, C. H. R. DE; GICO JR., I. Corrupção e judiciário: a (in)eficácia do sistema judicial no combate à corrupção. Revista Direito GV , v. 7, n. 1, p. 75-98, Jun. 2011.
• ANDREWS, M. The Limits of Institutional Reform in Development: Changing Rules for Realistic Solutions . Cambridge: Cambridge University Press, 2013.
• ARANHA, A. L. Corruption and the Web of Accountability Institutions in Brazil , PhD dissertation, Universidade Federal de Minas Gerais, 2015. Available at https://bit.ly/2A64SfR
  » https://bit.ly/2A64SfR
• ARANHA, A. L. Accountability, Corruption and Local Government: Mapping the Control Steps. Brazilian Political Science Review , v. 11, n. 2, Aug. 2017.
• ARANHA, A. L. The Web of Accountability Institutions and Corruption Control in Brazil. In: 2018 OECD GLOBAL ANTI-CORRUPTION & INTEGRITY FORUM. 2018 https://www.oecd.org/corruption/integrity-forum/academic-papers/Aranha.pdf
  » https://www.oecd.org/corruption/integrity-forum/academic-papers/Aranha.pdf
• ARANHA, A. L. Lava Jato and Brazil's Web of Accountability Institutions. In: LAGUNES, P.; SVEJNAR, J. (ed.). Corruption and the Lava Jato scandal in Latin America . New York: Routledge, 2020. p. 94-112.
• ARANTES, R. B. The federal police and the Ministério Público. In: POWER, T. J.; TAYLOR, M. M. (ed.). Corruption and Democracy in Brazil: The Struggle for Accountability . From the Helen Kellogg Institute for International Studies. Notre Dame, IN: University of Notre Dame Press, 2011. p. 184-217.
• BARDHAN, P. Scarcity, Conflicts, and Cooperation: Essays in the Political and Institutional Economics of Development . Cambridge, MA: MIT Press, 2005.
• BRAZIL. Decreto-Lei n° 2.848, de 7 de dezembro de 1940. Rio de Janeiro, RJ: Presidência da República, 1940. Available at: http://www.planalto.gov.br/ccivil_03/decreto-lei/del2848compilado.htm Accessed on: 14 Aug. 2020.
  » http://www.planalto.gov.br/ccivil_03/decreto-lei/del2848compilado.htm
• BRAZIL. Lei n° 8.112, de 11 de dezembro de 1990. Brasília, DF: Presidência da República, 1990a. Available at: http://www.planalto.gov.br/ccivil_03/leis/l8112cons.htm Accessed on: 14 Aug. 2020.
  » http://www.planalto.gov.br/ccivil_03/leis/l8112cons.htm
• BRAZIL. Lei n° 8.137, de 27 de dezembro de 1990. Brasília, DF: Presidência da República, 1990b. Available at: http://www.planalto.gov.br/ccivil_03/leis/l8137.htm Accessed on: 14 Aug. 2020.
  » http://www.planalto.gov.br/ccivil_03/leis/l8137.htm
• BRAZIL. Lei n° 8.429, de 02 de junho de 1992. Brasília, DF: Presidência da República, 1992. Available at: http://www.planalto.gov.br/ccivil_03/leis/l8429.htm Accessed on: 14 Aug. 2020.
  » http://www.planalto.gov.br/ccivil_03/leis/l8429.htm
• BRAZIL. Controladoria-Geral da União. Manual de Processo Administrativo Disciplinar. Brasília, DF, 2019. Available at: https://repositorio.cgu.gov.br/handle/1/42052 Accessed on: 14 Aug. 2020.
  » https://repositorio.cgu.gov.br/handle/1/42052
• CLEMENS, E. S.; COOK, J. M. Politics and Institutionalism: Explaining Durability and Change. Annual Review of Sociology , v. 25, p. 441-466, 1999.
• DA ROS, L. Accountability legal e corrupção. Revista da CGU , Oct. 2019.
• DANIELS, R. J.; TREBILCOCK, M. J.; CARSON, L. D. The Legacy of Empire: The Common Law Inheritance and Commitments to Legality in Former British Colonies. The American Journal of Comparative Law , v. 59, n. 1, p. 111-178, 1 Jan. 2011.
• DAVIS, K. E. Between Impunity and Imperialism: The Regulation of Transnational Bribery . New York, NY: Oxford University Press, 2019.
• FERRAZ, C.; FINAN, F. Electoral Accountability and Corruption: Evidence from the Audits of Local Governments. American Economic Review , v. 101, n. 4, p. 1274-1311, Jun. 2011.
• FERRAZ, C.; FINAN, F. Exposing Corrupt Politicians: The Effects of Brazil's Publicly Released Audits on Electoral Outcomes. The Quarterly Journal of Economics , v. 123, n. 2, p. 703-745, 2008.
• FILGUEIRAS, F.; ARANHA, A. L. Institutional Change and Accountability: Procedural Ecology and Defiance of the Brazilian Case. Revista Brasileira de Ciência Política , n. 28, p. 33-64, Apr. 2019.
• FISMAN, R.; GOLDEN, M. A. Corruption: What Everyone Needs to Know. [ s.l. ]: Oxford University Press, 2017.
• FISMAN, R.; MIGUEL, E. Economic Gangsters: Corruption, Violence, and the Poverty of Nations. Boulder: Princeton University Press, 2009.
• FONSECA, T. DO N. Nem indicação política, nem cooperação: a importância da autonomia do TCU no controle de recursos públicos federais transferidos aos municípios. Dados , v. 62, n. 3, e20170150, 2019.
• GARDINER, J. Defining Corruption. In: HEIDENHEIMER, A. J.; JOHNSTON, M. (ed.). Political Corruption: Concepts & Contexts . 3rd ed. New Brunswick, NJ: Transaction Publishers, 2017. p. 25-40.
• GOODFELLOW, T.; LINDEMANN, S. The Clash of Institutions: Traditional Authority, Conflict and the Failure of ‘Hybridity’ in Buganda. Commonwealth & Comparative Politics , v. 51, n. 1, p. 3-26, 1 Feb. 2013.
• GRAY, C. W.; KAUFMANN, D. Corruption and Development. Finance & Development , v. 35, n. 1, p. 7-10, 1998.
• HEIDENHEIMER, A. J.; JOHNSTON, M. (ed.). Political Corruption: Concepts & Contexts. 3rd ed. New Brunswick, NJ: Transaction Publishers, 2002.
• KLITGAARD, R. Controlling Corruption . Berkeley, CA: University of California Press, 1991.
• LAGUNES, P.; SVEJNAR, J. (ed.). Corruption and the Lava Jato scandal in Latin America . New York: Routledge, 2020.
• LECOVITZ, S. A corrupção e a atuação do judiciário federal 1991-2010 . PhD dissertation. São Carlos: Universidade Federal de São Carlos, 2014.
• LESSIG, L. Republic, Lost: How Money Corrupts Congress – And a Plan to Stop It . New York: Twelve, 2011.
• LOUREIRO, M. R. et al . Do controle interno ao controle social: a múltipla atuação da CGU na democracia brasileira. Cadernos Gestão Pública e Cidadania , v. 17, n. 60, 9 Aug. 2012.
• MACAULAY, F. Federalism and State Criminal Justice Systems. In: POWER, T. J.; TAYLOR, M. M. (ed.). Corruption and Democracy in Brazil: The Struggle for Accountability. From the Helen Kellogg Institute for International Studies. Notre Dame, IN: University of Notre Dame Press, 2011. p. 218-249.
• MACHADO, M. R. Independência como indiferença: a face jurídica da multiplicidade institucional em casos de corrupção. Revista Direito, Estado e Sociedade , v. 0, n. 55, 2019.
• MACHADO, M. R.; FERREIRA, L. M. A. (ed.). Estudos sobre o Caso TRT . São Paulo: FGV Direito SP, 2014.
• MACHADO, M. R.; PASCHOAL, B. Monitorar, investigar, responsabilizar, sancionar. A multiplicidade institucional em casos de corrupção. Novos Estudos , n. 104, p. 11-36, 2016.
• MAINWARING, S. Introduction: Democratic Accountability in Latin America. In: MAINWARING, S.; WELNA, C. (ed.). Democratic Accountability in Latin America . Oxford, UK: Oxford University Press, 2003. p. 3-33.
• OLIVIERI, C. Combate à corrupção e controle interno. Cadernos Adenauer , v. 3, p. 99-109, 2011.
• PACHE, A.-C.; SANTOS, F. When Worlds Collide: The Internal Dynamics of Organizational Responses to Conflicting Institutional Demands. Academy of Management Review , v. 35, n. 3, p. 455-476, 1 Jul. 2010.
• POPE, J. Confronting Corruption: The Elements of a National Integrity System. London, UK: Transparency International, 2000.
• POWER, T. J.; TAYLOR, M. M. (ed.). Corruption and Democracy in Brazil: The Struggle for Accountability. Notre Dame, IN: University of Notre Dame Press, 2011a.
• POWER, T. J.; TAYLOR, M. M. Introduction: Accountability Institutions and Political Corruption in Brazil. In: POWER, T. J.; TAYLOR, M. M. (ed.). Corruption and Democracy in Brazil: The Struggle for Accountability. Notre Dame: University of Notre Dame Press, 2011b, p. 1-28.
• PRADO, M. M.; CARSON, L. Brazil: Tackling Corruption Through Institutional Multiplicity. Policy in Focus , v. 12, n. 3, p. 30-33, 2015.
• PRADO, M. M.; CARSON, L. Brazilian Anti-Corruption Legislation and Its Enforcement: Potential Lessons for Institutional Design. Journal of Self-Governance and Management Economics , v. 4, n. 1, p. 34, 2016.
• QUEIROZ, R. P. DE. Núcleos de Combate à Corrupção: um novo arranjo institucional para uma atuação ministerial eficiente. In: Coletânea de artigos: avanços e desafios no combate à corrupção após 25 anos de vigência da Lei de Improbidade Administrativa / 5ª Câmara de Coordenação e Revisão, Criminal. Brasília: MPF, 2018. p. 123-140.
• RAGAZZO, C.; FREITAS, F. Brazilian Anti-Corruption Structure: Normative Systematization and Institutional Coordination Issues. VRÜ Verfassung und Recht in Übersee , v. 52, n. 4, p. 463-482, 2019.
• RICO, B. O papel da GCU no combate à corrupção: dez anos de atividade. ponto-e-vírgula , v. 12, p. 165-196, 2013.
• ROCHA, M. N. N. DA; ALENCAR, C. H. R. DE. Análise das penalidades graves aplicadas a servidores federais no Brasil: principais causas, modalidades punitivas, avaliação quantitativa e qualitativa. XIV Congreso Internacional del Clad sobre la Reforma del Estado y de la Administración Pública. Anais … In: XIV CONGRESO INTERNACIONAL DEL CLAD SOBRE LA REFORMA DEL ESTADO Y DE LA ADMINISTRACIÓN PÚBLICA. Salvador: 2009.
• ROSE-ACKERMAN, S. Corruption & Purity. Daedalus , v. 147, n. 3, p. 98-110, 1 Jul. 2018.
• ROSE-ACKERMAN, S. Corruption: A Study in Political Economy. New York, NY: Academic Press, 1978.
• SANTOS, D. DA S. Improbidade administrativa no Brasil: uma análise de output. PhD dissertation. Maceió: Universidade Federal de Alagoas, 2014.
• SARAT, A.; FELSTINER, W. L. F. Divorce Lawyers and Their Clients: Power and Meaning in the Legal Process. New York: Oxford University Press, 1995.
• SCOTT, J. C. Comparative Political Corruption . Englewood Cliffs, NJ: Prentice-Hall, 1972.
• SPECK, B. W. Auditing Institutions. In: POWER, T. J.; TAYLOR, M. M. (ed.). Corruption and Democracy in Brazil: The Struggle for Accountability. From the Helen Kellogg Institute for International Studies. Notre Dame, IN: University of Notre Dame Press, 2011. p. 127-161.
• SPECK, B. W.; NAGEL, J. A fiscalização dos recursos públicos pelos Tribunais de Contas. In: SPECK, B. W. (ed.). Caminhos da transparência: análise dos componentes de um sistema nacional de integridade. Campinas: Editora da Unicamp, 2002. p. 102-115.
• TAYLOR, M. M. The Federal Judiciary and Electoral Courts. In: POWER, T. J.; TAYLOR, M. M. (ed.). Corruption and Democracy in Brazil: The Struggle for Accountability. From the Helen Kellogg Institute for International Studies. Notre Dame, IN: University of Notre Dame Press, 2011. p. 162-183.
• TAYLOR, M. M.; BURANELLI, V. C. Ending Up in Pizza: Accountability as a Problem of Institutional Arrangement in Brazil. Latin American Politics and Society , v. 49, n. 1, p. 59-87, 2007.
• TEMPEL, A.; WALGENBACH, P. Subsidiary Managers and the Transfer of Human Resource Practices in Multinational Companies – Institutional Work at the Intersection of Multiple Institutional Frameworks. Schmalenbach Business Review , v. 64, n. 3, p. 230-247, 1 Jul. 2012.
• TRANSPARENCY INTERNATIONAL. What Is Corruption? Available at: https://www.transparency.org/en/what-is-corruption Accessed on: 8 Aug. 2020.
  » https://www.transparency.org/en/what-is-corruption
• TREBILCOCK, M. J.; PRADO, M. M. Corruption and Development. In: Advanced Introduction to Law and Development . Cheltenham, UK: Edward Elgar Publishing, 2014. p. 147-162.
• TYLER, T. R. What Is Procedural Justice? Criteria Used by Citizens to Assess the Fairness of Legal Procedures. Law & Society Review , v. 22, n. 1, p. 103-135, 1988.
• WEIJER, F. DE. A Context of Multiple Institutions: Engaging in Fragile Settings. The Broker Online , 4 Nov. 2013.

Publication Dates

• Publication in this collection
  11 Dec 2020
• Date of issue
  2020

History

• Received
  13 Mar 2020
• Accepted
  23 Oct 2020