Introduction

To guarantee the fairness of 2012’s municipal elections, the Regional Electoral Court of Rio de Janeiro petitioned the Federal Government to police eight municipalities in Rio de Janeiro’s metropolitan area.Footnote 1 In one of these municipalities, Japeri, the incumbent mayor was accused of involvement in the assassination of a political rival.Footnote 2 In Itaboraí, drug gangs did not allow some politicians to campaign in areas under their control.Footnote 3 Residents of Pedra da Guaratiba in Rio de Janeiro were told who to vote for by a local milícia group.Footnote 4 Again in 2016, the regional electoral prosecutor requested federal security forces remain in Rio de Janeiro after the Olympics to safeguard the integrity of municipal elections.

Each of these cases has something in common: politicians and criminal groups jointly act to violently influence elections. Yet, the cases also display certain differences. Sometimes criminal groups target candidates, killing them or barring them from campaigning in areas under their control. In other cases they coerce voters into selecting a specific candidate. Sometimes politicians are intimately linked with criminal organizations. At other times criminals and politicians are conjectural allies. Within one metropolitan area, strategies of political and criminal control both clash and are interconnected. How to make sense of these overlaps between criminal violence and electoral politics?

This article conceptualizes how criminal practices and electoral politics intersect at local levels (neighborhoods, small cities). I explore how criminal and political actors use criminal violence—understood as the intentional use of force or infliction of harm [29] by an organized criminal group as a means to reap profits from illicit activities—and clientelism to influence electoral outcomes. I call this criminalized electoral politics. Based on two dimensions–the nature of the relationship between political and criminal actors and the type of activity that provides a criminal group’s source of income–I also present four types of criminalized electoral politics: the neo-patrimonial, fusion, delegation and alliance types. In each type, different forms of clientelism and targets of criminal-political violence can be observed.

Currently, across many cities in the global South, criminal actors and politicians use criminal violence and material inducements to influence elections. However, the intersection of criminal violence and democratic processes is not a new historical phenomenon nor is it circumscribed to a particular world region. In New York City “it was estimated that by 1855 thirty thousand men [...] were in league with gang leaders who in turn were closely allied with influential politicians of Tammany Hall or the Know-Nothing party” ([55], 37). They used criminal violence and clientelistic schemes to influence vote choice. Politics in Southern Italy was (is) no less violent. Partnerships with the mafia resulted in “benefits” for politicians, for example, the assassination of political rivals and electoral intimidation [49, 52].

Nonetheless, research agendas in political science have for the most part ignored the conjunction of violence and clientelism. The literature on clientelism focuses almost exclusively on the material benefits proffered by politicians to garner political support, the target of these mobilization efforts, and the mechanisms of monitoring and compliance ([28, 31, 66], among others). They rarely explore how these exchanges are affected by criminal violence. Furthermore, studies of violence and elections usually focus on political violence (e.g. civil wars, ethnic violence, and state repression) and their effects on electoral outcomes [19, 27, 64]. While research on electoral violence provides valuable insights to understand when and how violence can be used strategically to achieve electoral success (e.g. [70]), we need to further understand how and why criminal actors get involved in elections. Finally, recent work on criminal violence in political science focuses mostly on its causes ([17, 35, 51, 68], among others), but seldom assesses how criminal violence can be used strategically to shape the linkages between politicians and voters.

By exploring the connections between criminal and political actors, this article offers a conceptual foundation to understand how criminal violence affects the linkages between citizens and politicians in competitive electoral regimes and, therefore, impact the quality of democracy and forms of local order in developing democracies. Criminal involvement in political processes has been seen as a consequence of state failure or diminished state capacity ([32, 72], among others). In this view, criminal violence indicates state absence and the existence of parallel authorities. However, in consonance with the major themes in this special issue, I argue that it can be misleading to treat criminal violence as an indicator of state absence or political disorder. This is particularly the case when state actors (like politicians) strategically use criminal violence to stabilize local—albeit authoritarian—political orders [3, 5, 10, 58]. A view of criminal violence as factor that shapes local electoral politics helps us better understand how existing local political orders function.

The linkage between criminal violence, clientelism, and elections is strong in Brazil’s urban peripheries. The growing importance of criminal organizations in the social and political life in these spaces have inspired exciting and sophisticated analyses of criminal governance in shantytowns, emphasizing the features of criminal organizations, their degree of territorial control, as well as their relationship with the community and state actors ([3, 5, 7, 42], among others). I build on this path-breaking research, but depart from it in two significant ways. Studies on urban marginalization and crime usually focused on shantytowns (favelas) in one city (for an exception see [6]). This limits our ability to observe a broader range of relationships between criminal groups and politicians, as well as different forms of criminal activity. My concept and descriptive typology of criminalized electoral politics are based on the empirical analysis of several peripheries (shantytowns, housing projects, and informal housing developments) and municipalities (large and small cities; urban and semi-rural areas) in Rio de Janeiro’s metropolitan area. This allowed me to include in my analysis different criminal groups and criminal-political linkages.Footnote 5 In the course of 17 months of fieldwork I conducted interviews with community leaders, activists, religious leaders, politicians, members of security forces, academics, and NGO staffers. I also collected information in archives, government reports, secondary literature, and media outlets, as well as created an original dataset of criminal-political violent events. With this information, I determined the characteristics of criminal groups and their relationships with politicians. I then defined categories to describe the nature of criminal activities and relationships between political and criminal actors. Using these categories, I developed my typology of criminalized electoral politics.

I begin this article by introducing the characteristics of criminalized electoral politics, the use of criminal violence and clientelistic mechanisms. I also distinguish criminalized electoral politics from other forms of criminal politics. This is followed by a presentation of four types of criminalized politics with cases illustrating each type.

Conceptualizing criminalized electoral politics

Political transitions since the Third Wave of Democratization (1974) diversified the set of countries that satisfy the minimal criteria to be considered an electoral democracy. More low-income countries with high levels of socio-economic inequality and uneven state presence became democratic polities [48]. Many of these democracies suffer from a high incidence of organized violence [8, 63]. In Latin America, high levels of criminal violence have been a feature of many of the region’s electoral democracies [69]. Clientelism also survived the transition to democracy [31]. Despite their joint occurrence in the developing world, these phenomena have been mostly studied independently from one another.

In this article, I focus on the use of criminal violence and material inducements by politicians and criminal organizations to influence electoral outcomes. I call this criminalized electoral politics. It is a form of electoral politics that is based on a non-programmatic linkage between citizens and politicians, but in which clientelistic relationships are also structured by the use of criminal violence: the reduction of electoral competition through the use of criminal violence limits citizens’ ability to choose between political options. Politicians and their criminal allies combine the stick of criminal violence with the carrot of clientelistic practices to maintain political control over peripheral areas.

In what follows, I distinguish criminalized electoral politics from other forms of clientelistic politics. I begin by describing the nature of non-programmatic politics and, in particular, clientelism. I then briefly discuss criminal violence. It is important to highlight that the existence of high levels of violent crime (e.g. homicides, assault) in an area does not necessarily indicate the presence of criminalized electoral politics. While violent crimes can be perpetrated by individuals or groups, violence observed in criminalized electoral politics is carried out by organized criminal groups. Criminal violence—the intentional use of force by a criminal group—is generally used for non-political, profit-seeking purposes.Footnote 6 In cases of criminalized electoral politics, however, the use of force also has political implications: criminal groups are profit-seeking, but they profit from violently shaping the link between citizens and politicians.

Using criminal violence and clientelistic tactics to influence electoral behavior does not mean that electoral success is guaranteed. For one, in areas where criminalized electoral politics occurs, many politicians simultaneously use these strategies and, naturally, not all of them can win. Moreover, while criminalized electoral politics can occur at a regional and even national level, it is a predominantly local phenomenon. The uneven rule of law and very limited societal accountability at local levels are necessary conditions for its occurrence. Yet, some state and societal institutions (judiciary, electoral authorities, parliaments, media) at regional and national levels often still operate and may occasionally thwart or limit the extent to which criminals and politicians can successfully exercise electoral control through criminal means, including the use of violence.

Clientelism

Criminalized electoral politics is characterized by the presence of clientelistic linkages between citizens and politicians. Traditional models of democratic representation posit a programmatic relationship between citizens and politicians. Office-seekers propose policy positions seeking to win electoral support. Citizens vote for candidates whose policy proposals most resemble their preferences. However, politicians and voters can also exchange political support for a specific benefit [31]. Non-programmatic politics is a form of distribution of goods and services that does not follow public criteria of distribution or in which public criteria are subverted (informally) by particularistic criteria. A more specific form of non-programmatic politics is clientelism, or a politician’s (patron) provision of a private benefit to a citizen (client) in exchange for their political support [66].

The prevalence and diversity of clientelistic practices spurred a rich and growing literature. Clientelism can vary depending on the nature of the relationship between clients and patrons. While some relationships are more enduring and transcend electoral cycles (relational clientelism), others are limited to specific exchanges during elections (electoral clientelism) [46]. These practices can have different targets. Benefits can be directed towards voters, either to change their political preference, to “bribe” supporters to go the polls or to induce supporters from the opposition to abstain from voting [23]. This literature has also studied the different organizational forms that sustain clientelistic politics [44, 65].

Students of political clientelism do not usually analyze how clientelism can go hand in hand with physical violence and other forms of voter intimidation [40, 41]. Indeed, clientelism is not usually associated with the use of violence. However, it is not uncommon to see clientelistic practices mix with violent tactics. In Latin America and Africa, armed groups and their political allies use both violence and the distribution of material benefits to cement political control [1, 4, 20, 26, 33, 54, 34, 38].

In some cases, the line between criminal and more political specialists in violence is fluid, making it difficult to distinguish between criminalized and insurgent electoral politics. Despite this fluidity, it remains important to identify if an armed group is driven more by political or illicit profit-seeking motivations, as this has implications for the relationship between armed and political actors. Criminal groups, for example, may have a limited opportunity to become Olson’s stationary bandit [50]. They have shorter time horizons that tend to result in more predatory behaviors [67]. Even if they become local “rulers,” their overt criminal pasts can limit a group’s ability to run for highly visible regional or national offices.

The use of criminal violence can have severe effects on the relationship between patron and client. While greater political competition resulting from democratization can increase citizens’ leverage vis-à-vis patrons and induce changes in the organizational features of clientelistic exchanges [31], violence can dramatically inhibit any of these gains by de facto reducing political competition. Violent intimidation of voters and/or violence against politicians strengthens patron’s hands and limits voters’ choice. By violently reducing competition, a criminalized politician effectively reduces the “price” of a vote.

Building a reputation as a ruthless politician who is willing to use violence to restrict electoral competition also contributes to reduce the compliance problem. If a politician is capable of assassinating his rivals and allying with criminal groups to intimidate the population, a voter will think twice before betraying a patron. Violence is thus a powerful signal to voters; it instills fear in them and, in their eyes, elevates the cost of non-compliance.

The signaling effect of criminal violence is even stronger when criminal groups provide politicians with the organizational structure and networks to mobilize voters. Often, criminal groups co-opt local civic organizations and use them to mobilize and monitor voters. In other contexts, formal and informal partisan structures and networks provide politicians with the information, organizational resources, and networks necessary to mobilize and monitor voters [65, 36]. Criminal groups can fulfill this role in criminalized electoral politics. If the political broker in a community is also a specialist in violence, the likelihood that a client will defect is much lower.

In short, in criminalized electoral politics, the use of criminal violence and clientelistic practices are not separate strategies of voter manipulation; they are complementary. The use of violence reduces the cost of a vote by limiting electoral competition. It also reduces problems of non-compliance and voter mobilization inherent in clientelistic politics. In areas—like urban peripheries—where the cost of inflicting physical harm are negligible, combining both strategies offers considerable benefits to politicians.

Criminal violence

Urban peripheries in Brazil and across the developing world can be violent places. Indeed, when one looks at the distribution of homicides in cities like Rio de Janeiro or São Paulo, the homicide rate in peripheral areas is considerably higher than in other neighborhoods. A well-established literature across the social sciences sees the causes of the varying levels of criminal violence in socio-economic inequality ([21], among others), as well as state incapacity to repress [60] or provide justice [56]. Other scholars have pointed out the enduring structural features of neighborhoods as factors that explain the spatial variation of criminal violence [57]. The geography of criminal groups, as well as their network structures, are also important predictors of violence [53]. Research in political science has focused on the political causes of criminal violence like the effects of political transitions, public security policies, the breakdown of state sponsored protection rackets, or the legacies of authoritarian security apparatuses ([15, 17, 35, 45, 51, 59]; among others). In these analyses, criminal groups are mostly profit-seekers that do not engage in acts of violence for ideological reasons. They do not seek to overthrow and replace the existing social and political order [30]. Criminal violence is a response to a political actions or institutions that affects illegal markets.

Recent studies have started to unveil the political consequences of criminal violence. Political participation tends to be higher in victims of violent crime [13]. However, when criminal groups directly target political campaigns and candidates, electoral participation decreases [37]. It is this latter, more strategic and political use of force by criminal groups that we observe in criminalized electoral politics. While still driven by a profit-seeking motivation, the use of violence by criminal groups in these cases also follows a political logic: criminal actors use or threaten the use of physical harm to advance the political ambitions of an allied office-seeker by obtaining or maintain electoral control. When criminal-political alliances use violence to intimidate voters or political rivals, elections are no longer mechanisms of accountability or leadership selection. Election results reflect more a capacity to inflict physical harm than the preferences of the electorate. In these cases, violence by criminal groups resembles forms of political violence [12]. Incorporating this strategic use of criminal violence in our analysis is vital to understand political life–for example the nature of clientelism–in many peripheral communities.

To affect electoral outcomes, violence can be employed against two targets: politicians or voters. When the target of violence is a politician, violence affects the “supply” side of the election, i.e. there are fewer candidates offered to voters. In turn, when voters are the focus of violence, violence impacts the “demand” side of the electoral equation, i.e. who can vote and how freely they can express their preferences. By targeting voters, politicians, or both, criminals and politicians re-configure electoral competition and distort electoral outcomes. In this way, criminal political violence can have similar effects on electoral politics as conventional forms of political violence [64].

The supply side: violence against/between politicians

Politicians can be the targets of criminal violence. Criminal organizations use violence against politicians and government officials–as it has been the case in Colombia or Mexico–because they decline to cooperate with criminal organizations and combat illegal businesses [35]. They also target politicians as part of extortion schemes, requiring them to pay a portion of municipal tax revenue to these organizations [67].

Politicians also suffer violent attacks in attempts to alter the supply of candidates in an election. Violence can be used against rival candidates in order to protect electoral turfs. The lethality of violence in these cases varies. The most common and less lethal use of violence is coercively banning campaign publicity (fliers, posters) in a particular space. “Unauthorized” candidates are forbidden from holding campaign rallies or other events. The assassination of political rivals to obtain or maintain control of political turfs is the most extreme manifestation of this form of violence. As shown in Fig. 1, between 1998 and 2016 107 politicians were victims of violent attacks (i.e. assassination or assassination attempts) in Rio de Janeiro (state). The occurrence of these violent events usually coincides with election years. During this time period, there was an average of 7.3 violent events against politicians in election years in Rio de Janeiro. In contrast, 3.8 violent events occurred in non-election years. In areas of Rio de Janeiro’s metropolitan area, like the Baixada Fluminense, political assassination is commonplace. Disputes between political groups are not only resolved at the ballot box, but rather in violent conflict between political competitors with the support of their criminal allies [61]. Similar patterns have been observed in cities across Brazil. In total, 97 Brazilian politicians were assassinated or suffered assassination attempts in 2016.Footnote 7

Fig. 1
figure 1

Assassination and assassination attempts of politicians in Rio de Janeiro (state), 1998–2016. Source: Original Dataset of Criminal-Political Violence in Brazil

Full size image

Violence against politicians can have important—indirect—effects on voters. It can increase a person’s perceived cost of voting, inducing lower levels of turnout [37]. Furthermore, it can also signal to voters’ the high costs of supporting candidates that oppose the political ambitions of criminal groups.

The demand side: repressing voters

The direct threat or use of physical harm against voters is another way to influence electoral results. In contexts of civil war violence can be used against civilians aligned with certain political preferences. Forced displacement, for example, completely reconfigures the geography of electoral support [64]. Violent groups can also use coercion to depress voter turnout or influence voter choice [1, 24].

Intimidation of voters also happens in contexts of mass criminal violence. A case in point are areas under the control of milícias in Rio de Janeiro and adjacent municipalities. These paramilitary-like groups establish protection rackets and use mechanisms of social control—including the ruthless use of violence against the population—to advance the political ambitions of some politicians. They force all residents to register, often record voter ID numbers, and use this information to contrast expected with actual voting results by voting booth.Footnote 8 Although milícias cannot perfectly monitor voters and thus target “non-compliant” voters, the fact that they collect private information is a powerful signal. For residents, it is reasonable to believe that they can be targeted if they do not comply with a milícia’s political directives.

In many cases when criminal violence is employed to influence voters’ electoral behavior, it is difficult to distinguish it from conventional criminal violence used to control criminal markets. In contexts when criminal groups exercise population—and not only territorial—control, a group’s use of violence for criminal activities creates a reputation of ruthlessness and instills fear in residents. Criminal groups can capitalize on this reputation to pressure people into voting for their candidates without having to actually physically harm voters.

Distinctions with other forms of criminal politics

Criminalized electoral politics is not the only form through which criminal activities intersect with political processes [62, 12]. Criminal actors can try to influence policy-outcomes through bribes or threats [16]. During the 1980s and 1990s in Colombia, the Medellín drug cartel sought to influence the Colombian government’s policy positions, particularly regarding extradition agreements with the United States, using terror tactics and political assassination [35]. The Cali Cartel in Colombia was notorious for financing political campaigns as a way to gain leverage over politicians. Moreover, it has been widely discussed that the Primeiro Comando da Capital’sFootnote 9 wave of attacks against police forces and the generalized uprising that led to the shutdown of the city of São Paulo in 2006 was an attempt to force concessions from the government regarding policies in its prison system [11].

In all these cases, criminal activities and politics intersect. Using bribes or violence, criminal actors seek to influence policy-making or policy implementation. However, these forms of criminal politics have substantive differences with criminalized electoral politics. First, criminalized electoral politics refers to attempts by criminal and political actors to directly influence electoral outcomes by regulating voters’ behavior. Unlike the previously discussed examples of criminal politics, criminalized electoral politics directly infringes on citizens’ right to freely choose their representatives.

Second, criminalized electoral politics usually occurs in very limited geographic areas like neighborhoods or small municipalities. Although there have been cases at a regional and even national scale, the visibility of criminal-political partnerships and the closeness between voters and criminal brokers characteristic of this phenomenon limits its viability at a regional or national level.Footnote 10 This, however, does not mean that local criminalized politics do not reverberate in other levels of government. Politicians at regional and national levels frequently ally with local criminalized politicians as part of their electoral coalitions.

Types of criminalized electoral politics

Criminalized electoral politics is always characterized by the use of criminal violence and clientelism to influence electoral outcomes. However, there are differences in the way it is structured across peripheries. In this section, I introduce four types of criminalized electoral politics based on the combination two dimensions: the relationship between political and criminal actors (integrated or distinct) and the type of (illicit) activity that generate a criminal group’s primary income (extractive and non-extractive). For each type, I will discuss how these two variables incentivize certain forms of clientelism and criminal-political violence.

Relationship between criminal and political actors

Types of criminalized electoral politics can be distinguished by the existing relationship between criminal and political actors at the local level [5, 12, 43]. Often, these actors are distinct, each having their own set of goals, strategies and resources. When this occurs, politicians are essentially office-seekers with organizations meant to mobilize voters. Criminals reap benefits from illicit activities and are primarily profit-seekers. The relationship that emerges between these actors is usually an agreement of mutual benefit. Criminal groups provide exclusive access to a population and can mobilize voters in exchange for benefits. Politicians can also provide local public goods and private benefits to a population through these groups; criminal groups become intermediaries between citizens and politicians [3]. However, alliances between crime and politics do not involve the integration of criminal and political organizations. Political actors are not involved in the criminal market. Moreover, the transaction between a politician and the criminal intermediary can be limited to one electoral cycle. In each election, the criminal group can change its allegiance to another political partner who is willing to offer a larger payoff.

In other cases, political and criminal groups cannot be easily distinguished from one another. In fact, criminal enterprises and political activities are two faces of one organization: crime and politics are integrated into one criminal-political organization [12]. This implies that the relationship between criminal and political actors is longer lasting and does not vary from one election to the next. The politician is involved in criminal enterprises. When criminal and political groups are more integrated, criminal actors cannot easily change their political partner.

The nature of the relationship between the criminal and political actors has important implications for the form of clientelistic practices that develop in areas experiencing criminalized electoral politics. When criminal and political groups are distinct, politicians do not have an incentive to invest in durable clientelistic relationships. The broker—the criminal group—is external to the political organization's operations and can easily change allegiances in future electoral cycles. In this sense, we should observe clientelistic exchanges occurring primarily during political campaigns (electoral clientelism). In contrast, when the political and criminal groups are integrated, politicians do have an incentive to create in long-lasting clientelistic structures that provide services to voters. Since the brokerage structure (criminal actors) is part of the criminal-political group, the politician does not expect that the broker will defect to another politician. Therefore, we should observe clientelistic exchanges occurring on an ongoing basis (i.e. relational clientelism) when there is criminal-political integration.

Illicit activities

Another relevant dimension to distinguish type of criminalized electoral politics is the activity that provides a criminal group’s primary source of income. I distinguish between extractive and non-extractive illicit activities. Extractive illicit activities produce income through the direct extortion (or “taxation”) of a population. A common example of are protection rackets. Residents of an area are forced to pay a “protection fee” to avoid being targeted by a criminal group. A criminal organization can also force businesses to pay a percentage of their revenue. These groups might also hold a monopoly in the provision of a particular good and, through monopolistic pricing, extract rents from a population. Extractive activities require an organization capable of monitoring and punishing a population under its control. These organizational characteristics can be useful for electoral purposes as well. Technologies for extraction can be employed to force populations to vote in a specific way.

In contrast, non-extractive illicit activities do not involve the “taxation” of a population by a criminal group or “expropriating” resources from people in areas under its control. The retail sale of illicit substances can be considered a non-extractive activity, as the profit from the sale of this product is obtained from consumers who are (generally) not forced to buy from a specific vendor.Footnote 11 The illegal appropriation of state funds can also be considered a non-extractive activity. Legal taxation by the state is indeed a (licit) extractive activity. However, when a criminal group “plunders” a city’s finances it is not directly taking personal income away from residents. Non-extractive activities do not require building organizational structures to monitor and regulate populations. However, they do require control over an illicit activity or market, for example by controlling access to a territory. While gaining a population’s support, for instance through the provision of some goods, might facilitate control over criminal markets, these minor governance tasks can be fulfilled by organizations that do not primarily regulate populations.

The source of criminal income affects who the target of criminal-political violence is. Criminal groups that undertake extractive activities develop organizations that control populations and territories, akin to some forms of rebel governance [9]. They can use these organizations to coercively mobilize voters to support their political partners. Conversely, groups that engage in non-extractive activities tend to exercise mostly territorial control. These organizations have the capability to regulate access to a criminal and political territory. Lacking the capacity to effectively monitor residents (populations), non-extractive criminal groups cannot effectively use violence against voters, but can regulate who can enter their territory, making politicians the preferred target of non-extractive groups.

The following typology of criminalized electoral politics resulted from empirical research in the metropolitan area of Rio de Janeiro. By analyzing the similarities and differences between criminal groups and electoral politics in eight cities in this area, I identified categories to describe criminal activities, as well as the relationships between political and criminal actors. I then used these categories to construct types. These should be understood as ideal types in a Weberian sense. They are analytical constructs that help us understand and classify cases, yet they do not exist in reality. Table 1 displays the four types of criminalized electoral politics based on the two dimensions described: neo-patrimonial, delegation, fusion, and alliance.

Table 1 Types of criminalized electoral politics
Full size table

A possible refinement of this typology would be to conceptualize and measure the two dimensions in a continuous rather than categorical way. Because criminal groups seldom engage only in exclusively extractive or non-extractive criminal activities nor criminal and political actors are completely integrated or distinct, a continuous measure may more accurately reflect reality on the ground.

Neo-patrimonial type

Characteristics

The neo-patrimonial type of criminalized electoral politics is characterized by a high degree of integration between criminal and political groups. In this type, these actors are often one and the same. Political-criminal organizations are close-knit groups whose members usually belong to a same family or are very close associates. One member of the group may be the public (political) face of the group, while another close associate handles its criminal component. The integration between criminal and political organizations in this type is frequently related to the usual main source of income: the misappropriation of state resources, a non-extractive illicit activity.

This form of criminalized electoral politics resembles past forms of rural, local patrimonial orders in Brazil [47]. However, while the neo-patrimonial type can be observed in small cities, it does not exhibit the same mechanisms of social and political control of populations present in (past) rural areas. This contemporary type of criminalized politics is characteristic of urban, albeit peripheral, locations. Control of land and labor is not a decisive factor. These groups are patrimonial because they take over institutions and their resources and use them as if they were their own private property [39]. State funds end up in their private accounts; infrastructure projects are assigned to their construction companies or those of their allies. City workers are employed in their political campaigns. In short, these groups completely take over local institutions to plunder them. While they are often involved in other forms of criminal activity or regulate other criminal markets in their areas of influence, these groups tend not to directly extort populations.

Given the close association between criminal and political actors, we tend to observe long-term clientelistic relationships (relational clientelism) in this form of criminalized electoral politics. Through criminal brokerage structures, these groups provide particularistic services to voters (for example, healthcare), not only during elections. Moreover, these organizations’ source of criminal income incentivizes them to develop violent capabilities suited for the control of a territory. As the plundering of local state institutions is a major source of income, these groups particularly specialize in violently attacking political rivals or social leaders opposed to their political interests. This means that they seek to influence electoral outcomes by artificially reducing the number of candidates, not directly employing electoral violence against voters. In this way, the neo-patrimonial type of criminalized electoral politics is different from previous forms of patrimonial rule in Brazil, in which both residents and political rivals were the target violence.

Magé

The neo-patrimonial type of criminalized electoral politics can be observed in the Baixada Fluminense of Rio de Janeiro and in smaller cities of Northern and Northeastern Brazil. In the Baixada Fluminense, there has been a historical strong link between criminal groups and politicians, surviving well past Brazil’s democratic transition in 1985 [61]. A contemporary case of the neo-patrimonial type of criminalized electoral politics can be found in the city of Magé. Since 1982, when Renato Cozzolino was elected mayor, the Cozzolino family has governed Magé almost uninterruptedly. Besides electing four mayors, they have also elected councilmen, four state deputies, and a federal deputy.Footnote 12

Political hegemony in the municipality has been very profitable for the family. In fact, most of the family’s politicians, like former mayor Nubia Cozzolino, were declared ineligible for public office because of massive fraudulent procurement schemes and misappropriation of public funds.Footnote 13 The family and their allies also continually use public employees and infrastructure in their campaigns. Public spaces are often used to perform regular clientelistic exchanges. While the use of state institutions to establish durable clientelistic schemes is not unique to the Cozzolinos, they took this practice to an entirely new level. Public buildings in the city were rebranded to bear the names of family members, many of whom were not even notorious public figures. Despite their numerous criminal indictments and removals from office, the clan maintains a firm grip over the city.

As we should expect in neo-patrimonial forms of criminalized electoral politics, violence against politicians is a characteristic of political life in Magé. Since 1998, 12 politicians, including seven councilmen and a deputy mayor, have been killed. Ten (or 83%) of these assassinations occurred in election years. While the Cozzolino family disputes any relationship with criminal-political violence, political opponents claim to be the victims of their threats and abuse. Moreover, a close political ally of the Cozzolinos, councilman Genivaldo Ferreira Nogueira, was a suspect in the assassination of four local politicians.

Fusion type

Characteristics

Like the neo-patrimonial type of criminalized electoral politics, there is a high level of integration between political and criminal actors in the fusion type. Although public attention may force politicians and criminals to re-calibrate strategies,Footnote 14 they always maintain very close political ties. Politics is fundamental to understand criminal violence and criminal markets in this type. Political connections protect sources of criminal revenue by securing impunity. In this latter sense, they resemble neo-patrimonial types of criminalized electoral politics.

In contrast to the neo-patrimonial type, criminal groups in the fusion type engage primarily in extractive enterprises. The primary source of their income is the extortion of local populations: from protection fees charged to residents, to operating fees demanded from local businesspeople. They can also control certain legal markets and use abusive (monopolistic) pricing strategies to extract rents from the population. These extractive activities require population control and organizations, among them armed groups, to secure it. Criminal groups invest in organizations that exercise direct control over community organizations (like neighborhood councils) to monitor and collect protection fees from residents. They also engage in conflict adjudication and the provision of some public goods.

Like in the neo-patrimonial type, the close linkage between criminal and political actors has implications for the form of clientelism we are likely to observe in fusion forms of criminalized electoral politics. Criminal-political groups tend to set up stable clientelistic schemes, regularly providing residents certain public and private goods in exchange for political support.Footnote 15 Many of these groups establish long-term clientelistic machines, many even build physical infrastructure like community centers, to provide material benefits in exchange for political support beyond electoral cycles.

Unlike the neo-patrimonial type, voters tend to be the target of criminal-political violence in the fusion type. Criminal groups use their already existing violent control over a population to coercively advance the candidacies of politicians closely linked to them. Residents are frequently told who to vote for and community organizations are also employed to monitor voters’ behavior. The same physical violence (or threat thereof) that is used to extract payments is used to advanced political ambitions. In fact, distinguishing between regular criminal violence and more political criminal violence in these communities is difficult.

Often, criminal groups do not need to employ physical violence in electoral season because the continuous use of violence to sustain their criminal local order already predisposes residents to comply with the criminal group’s electoral ambitions. While voters are the central target of criminal-political violence in this type of criminalized electoral politics, rivaling politicians can also be targeted, for example, by not allowing them to campaign in areas under their control or, in more extreme cases, killing them.Footnote 16

Duque de Caxias

The case of the “Família é nós” (We are family) milícia in Duque de Caxias provides a good example of the fusion type of criminalized electoral politics. This criminal group’s central source of revenue is extractive: it established protection rackets by charging “security fees” from residents, provided illegal cable and internet services, as well as transportation. They sold cooking gas illegally and forced residents to buy groceries at higher than market prices. While mostly engaging in extractive activities, Família é nós was also involved in some forms of non-extractive criminal activities like arms trafficking, illegally sold public lands, as well as regulated other illegal businesses like the jogo do bicho and retail drug trafficking in their territory.Footnote 17

Besides running a successful criminal business, Familia é nós also had a successful political operation. Sebastião Ferreira da Silva (Chiquinho Grandão) and Jonas Goncalves da Silva, two Caxias councilmen, ran this integrated criminal-political group. Chiquinho Grandão, the alleged leader of the group, was elected councilman in Caxias in 2004 and 2008, each time for a different party. His votes, not surprisingly, were highly concentrated in neighborhoods controlled by Familia é nós. In 2008, 64% of his votes came from four neighborhoods in which his criminal group operated.

Given the high level of integration between criminal and political actors in this case, it is not surprising that the group established more durable clientelistic schemes. This criminal-political group operates “centros sociais”, permanent community centers that offer medical and other services in exchange for political support. This criminal-political organization is also accused of engaging in vote-buying operations.Footnote 18

Moreover, the extractive nature of its main income had incentivized the group to create an organization that can monitor the population under its control. Quite importantly, its armed apparatus is populated with members and former members of state security forces. The arrest of five former military policemen, two marines, and one policeman belonging to the group illustrates the level of involvement by state specialists in violence in this organization. Members of this armed group were accused of harassment, killing witnesses, torture, threats and hiding human bodies.Footnote 19 The violence used to sustain their criminal control was easily harnessed to advance political ambitions by coercing the population.

Delegation type

Characteristics

In contrast to the neo-patrimonial and fusion types, criminal and political actors are loosely integrated in areas with delegation type criminalized electoral politics. In cases characteristic of this type, state actors informally delegate to local criminal strongmen the maintenance of (informal) social and political order without including them in their political organizations or being involved in their illicit sources of revenue. Criminal groups in this model establish primitive forms of neighborhood governance and provide some public goods to the population; however, access to these goods is contingent on the acceptance of the rules underlining the local criminal order.

Commonly, these groups create or take over civic associations and neighborhood groups to use these to monitor the population. It is therefore not surprising that that criminal actors in this type of criminalized electoral politics are engaged in extractive illicit activities: they derive income mostly through the extortion of a population. Dwellers of these communities are required to pay protection fees. In some cases the criminal group provides public services (water, cooking gas) at exorbitant, above market values. In this way, the delegation model resembles the fusion type.

The lack of deep integration between local criminal groups and political machines has implications for the nature of the clientelistic relationship that is likely to emerge in delegative forms of criminalized electoral politics. Criminal groups act as intermediaries between politicians and citizens. However, unlike the fusion and neo-patrimonial type, the criminal group can act as a “free agent” and change political patron after an election. They are criminal brokers that are not permanently aligned with one specific political machine.

The focus on extractive criminal activity makes criminal groups focus more intensely in the repression of the local population. These criminal groups have to develop networks and organizations that enable them to monitor and punish the population. Given these organizational characteristics, they can offer coerced voter mobilization to politicians in exchange for protection (impunity) or material benefits. Thus, violent capabilities are mostly directed towards restricting dissent and mobilizing local residents to vote for the group’s candidates. These local criminals generally do not target politicians.

Primitive Milicias in the Baixada Fluminense

The informal delegation of political power and the structures that develop in the delegation type of criminalized electoral politics can be observed in two settings in the metropolitan area of Rio de Janeiro. Some of the patterns of the delegation model are consistent with some of the early milícias that emerged in the western area of the city of Rio de Janeiro. These protection rackets primarily focused on the extortion of protection fees from the population. In time, some of these groups began controlling other (legal) markets–like the provision of cooking gas, cable TV–through illegal means and abusive pricing [2]. While they had very close links with some state agents, particularly members of the police, many were not—at least initially—integrated into political machines. These criminal groups acted as intermediaries between a population and a politician, mediating clientelistic exchanges between them.

Delegation currently occurs in newly established informal settlements in some areas of the Baixada Fluminense. Many communities under this delegation model are, in fact, the result of criminal enterprises. For example, a criminal group illegally appropriated publicly owned lands in a city of the Baixada. It sold plots of the land (with a property title!) to low-income settlers. The criminal group established a small protection racket in this newly created neighborhood, charging fees for “security” and other services (like distributing mail), i.e. they engaged in extractive criminal practices.

Moreover, the group did not have enough political strength to field a candidate belonging to the criminal group. In contrast to cases of fusion or neo-patrimonial criminalized electoral politics, the local strongman did not belong to a political machine. Politicians, in turn, did not establish durable clientelistic infrastructure, like community centers, in this community. Clientelistic practices were mostly circumscribed to campaign season. During regional and local elections, the group did serve as a broker for politicians, mobilizing residents in their support. In this, its control of the local residents association and local coercive structure was instrumental. The threat of criminal-political violence was directed primarily towards the local population.Footnote 20

Alliance type

Characteristics

In areas where the alliance type is prevalent, criminal groups and politicians are not integrated organizations. They are distinct actors who agree to cooperate for mutual benefit over a limited time period. Their relationship is analogous to a business exchange: the criminal group provides access and exclusivity to a territory for benefits. In contrast to fusion and neo-patrimonial types, however, the relationship between criminal and political groups can end on election day and may not repeat itself in future campaigns. In their activities as a legislator, for example, a politician does not provide enduring protection to the criminal group. Neither does the criminal group “work” for the politician in-between elections.

Moreover, in this type, a criminal organization’s primary source of income are not fundamentally extractive activities seeking to take resources from a local population. Revenue can be derived from selling illicit products—for instance drugs—or providing illegal entertainment—like gambling—to consumers. Thus, criminal organizations do not develop very complex governance structures or monitor residents too closely. However, some minimal level of control over the population—for example through the provision of public goods and the co-optation of local institutions—can generate legitimacy and local support, as well as maintain levels of social order that are beneficial to non-extractive illicit activities [5].

The lack of integration between criminal and political actors has implications for the way clientelism is structured in this type of criminalized electoral politics. Similarly to the delegation type of criminalized politics, during elections, criminal groups act as intermediaries (brokers) between citizens and politicians, mediating the clientelistic exchanges [4]. However, given the lack of integration between criminal and political actors, politicians do not invest in long-term clientelistic infrastructure. In this type of criminalized electoral politics we should observe more electoral than relational clientelism.

Moreover, given the generally non-extractive nature of their illicit activities, these groups do not have the organizational resources to use criminal violence against the population for electoral purposes. Violence is not used to coerce residents to vote for a particular candidate. In this type we should observe politicians being the main target of criminal violence during elections. Threatening the use of violence, some candidates are denied the right to campaign in an area unless they have an agreement with the criminal group. Campaign materials of unauthorized candidates might be removed and destroyed.

Drug gangs in Rio de Janeiro

The alliance form of criminalized electoral politics has been observed throughout many favelas controlled by drug gangs in Rio de Janeiro (e.g. [4, 5, 25, 34, 42]). With the rise of more developed retail drug trafficking organizations in the 1980s, the dynamics of local politics in the peripheries shifted. Although primarily focused on the profit from retail drug sales, drug gangs started to regulate some aspects of social life—particularly movement in and out of some favelas of the city. They became arbiters of neighborhood disputes and provided some public goods. Many drug gangs violently took over civic organizations, like neighborhood associations, and became political intermediaries [3, 4, 42]. Their provision of some public goods as intermediaries generated some support from the community–beneficial for their business–but also provided other minor sources of income. Even though they controlled some civic associations, they did not initially establish the complex taxation and monitoring structures created by other criminal groups in Rio de Janeiro.

These organizations were also traditionally distinct from political machines. They negotiated access to areas under their control without integrating into political organizations. Candidates could hold campaign rallies or distribute fliers in areas under their control only if there was an agreement with the local drug gang. If politicians do not have permission, their campaign posters would be removed.Footnote 21 Residents supporting unauthorized candidates were unable to openly campaign for their preferred candidate.Footnote 22 In return for access, criminal group received payments. These arrangements, however, were strictly limited to electoral periods. Politicians did not build their own permanent clientelistic infrastructure.

Moreover, criminal violence used for political purposes by drug gangs mostly targeted politicians and tended to be non-lethal, i.e. denying access to a community but not assassinating politicians. Drug gangs in Rio’s favelas did not usually coerce residents into voting for a particular candidate. While violence against the population is observed in these communities to enforce certain rules—like not cooperating with the police—it is not used to maintain political (electoral) control.

Table 2 provides a summary of the features of clientelism and criminal violence that should be observed in each type of criminalized electoral politics. As the table shows, each type can share some features with other types; nonetheless, each type is distinct from the other. Although all types of criminalized electoral politics are characterized by the use of criminal violence and clientelistic exchanges to influence electoral outcomes, the dynamics of clientelism and violence can vary ostensibly: The target and often the lethality of the criminal violence differ. In some types these clientelistic relationships are managed directly by the criminal-political group in long-term (relational) clientelistic relationships. In others, the criminal group is only the intermediary of short-term (electoral) clientelistic exchanges.

Table 2 Features of clientelism and criminal-political violence by type of criminalized electoral politics
Full size table

Conclusion

Criminal violence is not necessarily a sign of disorder and anomie [58]. Often, criminal violence can indicate the existence of local orders and is profoundly shaped by political calculations. This article explores one way through which criminal groups and politicians create local orders in urban peripheries to restrict electoral competition using criminal violence and clientelism. To account for the variation in the forms of criminalized electoral politics, I also develop a descriptive typology and highlight the expected forms of clientelism and criminal-political violence for each type. In this way, I seek to bridge literatures that have focused exclusively on clientelism, electoral violence, or criminal violence.

While many forms of criminal influence in local politics have been analyzed in previous scholarship, this article provides a framework to understand what are often treated as distinct criminal and political phenomena through one conceptual lens. Frequently, scholarship focuses on the actions of one criminal actor (drug gangs, vigilante groups, smugglers, etc.) and their criminal and/or political actions. This article assumes that all local criminal groups are capable of engaging in political activities. In some contexts, certain criminal groups might gravitate towards a type of criminalized electoral politics. Nevertheless, they will not always engage in the same pattern of violent electoral influence. For example, some milícias in Rio de Janeiro are involved in fusion forms of criminalized electoral politics. Others are more accurately described as delegation types. Drug gangs in Rio de Janeiro are more likely to build temporary alliances with politicians. Yet, they may make attempts at fusion forms of criminalized electoral politics.

This analytical strategy can provide a useful foundation for future research. Often, criminal groups with the same label behave in markedly different ways. Yet, by using the same label we often ignore the considerable variation that hides behind it. In other cases, we neglect to see similarities between criminal groups with different labels. This problem is even more acute when we attempt to compare criminal groups and criminal-political relationships across regions and countries. To further advance the comparative study of criminal violence and, in particular, how criminal groups interact with political actors to create order, we need concepts and typologies that are less rooted in everyday language. While my article is grounded in my study of crime and politics in Brazilian urban peripheries, it produces a typology that can be adapted for other contexts.

Finally, this article lays the groundwork for more empirical research on criminal violence and politics in Brazilian urban peripheries. It can guide the collection of more fine-grained data on criminal-political violence, for example, data on political assassination and coerced voter mobilization. It can also lead to more qualitative studies that unpack the clientelistic and violent mechanisms that sustain political orders in which criminalized electoral politics is prevalent. Finally, it serves as a starting point to develop a theory that explains how and why the relationship between criminal and political actors changes, as well as the impact of these transformations for local criminal markets and political life. This will not only enable us to better understand the shortcomings of many contemporary democracies, but also inform policies seeking to address these.