Defining Police and Violence

I treat the police as a unified actor. This includes street-level police officers who constitute the bulk of the organization, are the ones most likely to encounter (and engage in) violence, and “comprise the most visible face of the state for many civilians” (Lake Reference Lake2022, 858), as well as the high-ranking officers who most frequently engage with political superiors and make the first crucial choice to abide or defy their mandate. While I recognize that police commanders and rank-and-file have different incentives and their own principal-agent problems (Brehm and Gates Reference Brehm and Gates1999), I consider them as a singular actor in this simplified framework to center on their general organizational incentives to comply with or defy political incumbents. I will describe how this assumption may be relaxed in the final section.

This framework intentionally neglects officers whose main duties involve criminal investigations—except when they use torture when questioning a suspect (Ralph Reference Ralph2020)—intelligence operations, and administrative tasks, such as training, bookkeeping, technical support, and internal oversight, since they are not as frequently put in positions where they need to resort to violence. It also excludes other agencies that carry out policing functions—such as the military, customs, and border patrol—since these have different institutional mandates, socialization processes, and relations with politicians and the public. Finally, this framework also sidelines non-state actors who, whether due to state absence, forbearance, or complicity, maintain a given social order within their communities through surveillance and the threat of sanction, such as neighborhood patrols, self-defense groups, vigilantes, and organized criminal groups (Arias Reference Arias2017; Bateson Reference Bateson2021; LeBas Reference LeBas2013; Moncada Reference Moncada2021). While these actors can engage in policing, being outside the state releases them from certain requirements that are fundamental to the police as a differentiated, specialized institution (Bowling, Reiner, and Sheptycki Reference Bowling, Reiner and James2019, 5–8). Unlike the police, they are not “equipped, entitled and required to deal with every exigency in which force may have to be used” (Bittner Reference Bittner2001, 299).

Violence is quintessential to police work. As Bittner (Reference Bittner1970) has stated, the potential to use violence, even if its application is almost always unnecessary, “constitutes the quotidian power of policing.” Similarly, Seigel claims that the potential (use of) violence is the essence of [police] power (Reference Seigel2018, 9). The centrality of violence to policing explains why politicians, democratic or authoritarian, would be strongly interested in harnessing that potential, whether to mitigate or exacerbate it. Where my framework differs from these sociological analyses is that it problematizes the role of police within the state, given the conflict it might have with governing politicians. I also focus on a narrower definition of police violence as the realm in which these potential conflicts manifest.

I define police violence as the use of force by on- or off-duty police officers that results in the death or injury of an individual. While building on the World Health Organization’s definition of violence,Footnote ⁵ mine is more restricted in three ways. First, it does not include the threat to use force, which is less observable, frequently unreported, and more likely to be (mis)construed by officers as appropriate “warnings.” Second, I focus on the direct physical effects on victims and thus exclude potential manifestations of violence such as “psychological harm, maldevelopment and deprivation,” because these could arise from indirect as well as direct exposure to police brutality, making it difficult to ascertain how many individuals should be considered victims of police violence. Third, I focus on violence against individuals rather than groups or communities. While individuals are often victimized by police because of their real or perceived association with certain groups, this is not always apparent or easy to corroborate. Furthermore, individuals can possess multiple identities. In short, a police force is more violent than another—or when compared with itself over time—if its members resort to interventions involving bodily harm against individuals more frequently and systematically.

This definition is agnostic to the purpose for which police employ force and its legitimacy, which is usually defined ex post and is highly contested. While this paper’s empirical analysis mostly focuses on one form of police violence, that is, lethal interventions by police officers,Footnote ⁶ the framework can also apply to other non-lethal forms, such as non-fatal shootings, beatings, torture, and repression of social protests.

The definition encompasses off-duty as well as on-duty police officers. When off-duty officers apply violence, whether they identify themselves as law enforcement agents or act covertly, that use of violence is related to their occupation. For instance, off-duty officers can kill individuals when intervening to stop a robbery by using their service weapon, which, in many countries, they can carry home with them (Fyfe Reference Fyfe1980). When intervening, they will resort to the practices and techniques acquired during their training and socialization. While off-duty officers can perhaps more easily defy politically enacted restrictions on the use of force because not being in uniform weakens their supervision and increases their impunity, off-duty police can also apply violence in compliance with formal or informal political mandates. For instance, off-duty police can form or join militias or death squads that seek to wipe out presumed criminal elements, which are often sanctioned, if not lauded, by politicians (Cruz Reference Cruz2016). Consequently, in many cases a high share of police killings are carried out by off-duty police officers,Footnote ⁷ who are also killed at high rates.

Politicians Incite or Restrain Violence

This framework’s other protagonists are national and subnational democratic politicians, that is, elected or appointed government officials in the executive branch charged with supervising the police, who can either incite or restrain police violence. Examples of these politicians are presidents or prime ministers, governors, mayors, and their respective secretaries or ministers of security, justice, the interior, or defense, depending on the police’s position in the government structure. Even in contexts where politicians have limited operational control over the police, they can still designate authorities, approve their budget, enact criminal and procedural codes, and outline strategic security goals, all of which can shape police behavior.

Politicians typically respond to (and subsequently shape) citizens’ insecurity concerns by expanding or reducing police discretion to use force, inciting or cracking down on police extralegal violence, and tolerating or combating police participation in non-state armed criminal groups (Cruz Reference Cruz2016). Certainly, citizen interactions with police can affect how the latter carry out their jobs. However, in this two-actor model, citizens only play an indirect role in shaping security policies, as they can influence politicians’ decisions through electoral choices, public opinion polls, lobbying, or protests.

Two clarifications are in order regarding this framework’s scope conditions. First, I center on politicians in democratic regimes, who, in principle, face stronger constraints in their resort to police violence, whether from political opponents, the press, the judiciary, civil society organizations, or the broader public. On average, democratization improves respect for human rights and decreases state repression (Davenport and Armstrong Reference Davenport and Armstrong2004), although this does not negate the presence of authoritarian policing in formally democratic contexts (González Reference González2020). I also focus on democratic settings because clashes between politicians and police over the application of violence are likely more frequent than in authoritarian regimes, where police might have stronger incentives to comply with incumbents, as defiance can result in imprisonment, torture, or even death.

Second, this framework applies to democracies with varying levels of insecurity and violence. While factors such as homicide levels and the power of organized criminal groups surely influence the design and enforcement of policing strategies, similar policies to incite or restrain police violence have been implemented in myriad contexts irrespective of these indicators. The reproduction of broken-windows policing, on the one hand, and community policing, on the other, both originated in the United States and exported to generally more violent Latin American countries, constitute but two examples. More generally, security policies are not intrinsically related to levels of violent crime.

Policy shifts: Changes in police power. This paper focuses on policy changes that have the potential to alter the police’s core functions, internal procedures, and accountability, subsequently modifying police discretion and its relative power over citizens and vis-à-vis politicians. These shifts occur as politicians become more involved in determining the criteria by which police will recruit, train, equip, promote, and punish officers, as well as in deciding what the police will do and how they will do it. I focus on politicians’ attempts to modify regulations that affect the police’s propensity to use force, which requires an intention to alter the status quo. When there is no attempt to change policy, I consider politicians as tolerating existing levels of police violence. Governing politicians may issue symbolic gestures to condemn an act of police violence, or to demand greater discretion for police to control crime. However, these actions will not be considered unless accompanied by concrete attempts to modify police functions, accountability mechanisms, or procedural guidelines.

Politicians can change police functions by altering the range of situations in which police should intervene and the guidelines by which those interventions should occur. For instance, politicians can determine that multi-disciplinary crisis intervention teams instead of police officers will resolve incidents involving homeless individuals, persons under the effects of psychotropic substances, or persons with mental illnesses (Taheri Reference Taheri2016). Contrarily, they might decide that the police should be the first (and perhaps the only) agency responding to these situations (Stuart Reference Stuart2016). Politicians may also militarize their police, a shift that includes acquiring military equipment, such as automatic weapons, Kevlar vests, or armored vehicles, as well as changing police doctrine, tactics, training and the like. Militarization usually incites police violence, as it projects a “signal from the state that violence is an acceptable means of public control” (Stavro and Welch Reference Stavro and Welch2023, 2) and instills officers with a warrior mentality (Mummolo Reference Mummolo2018a). Politicians can also informally tolerate or mandate that police abstain from repressing several crimes, from squatting to drug dealing, whether to use their resources more efficiently or to capture rents from these activities (Holland Reference Holland2017; Flom Reference Flom2022). Finally, politicians can also alter the broader criminal justice system in which police operate by changing penal or penal procedural codes. They may designate new activities as crimes or decriminalize existing offenses and either expand or limit police prerogatives to arrest, interrogate, and hold suspects in custody (Hausman and Kronick Reference Hausman and Kronick2021; Magaloni and Rodriguez Reference Magaloni and Rodriguez2020).

Politicians can also enact policies that affect the police’s internal procedures, such as recruitment, training, promotion, and discipline. These are crucial processes that mold police culture and behavior. Politicians can distance themselves from these procedural decisions, including those in relation to the use of force, which police have often used to expand their discretion and shield abusive officers. By contrast, political interventions can, in principle, align police procedures with democratic criteria, for instance, by ensuring diversity in recruitment and promotion. Of course, political proactiveness can also condone more aggressive law enforcement, such as when politicians promote officers with controversial human rights records.

Finally, politicians can also establish (or eliminate) oversight bodies—such as internal affairs units, auditing offices, ombudspersons, or judicial courts—to monitor police behavior. When adequately equipped, these oversight institutions can have a significant impact in reducing the frequency of human rights violations by the police (Hu and Conrad Reference Hu and Conrad2020). However, not all monitoring mechanisms are equally effective. Body-worn cameras that record police interactions with citizens have not reduced the frequency of police use of force (Lum et al. Reference Lum, Koper, Wilson, Stoltz, Goodier, Eggins, Higginson and Mazerolle2020), despite this being one of their primary objectives. Citizen security forums, while theoretically aligned with democratic values, can reinforce exclusionary stereotypes, unequal social relations, and violent policing practices (González and Mayka Reference González and Mayka2023). In short, the aforementioned policy shifts can serve to either restrain or incite police violence, depending on how they are implemented, and the extent to which police comply with them.

Politicians’ incentives. Politicians have multiple incentives to enact these various policy shifts. Adopting tough-on-crime positions can deploy a valence issue that attracts various constituencies or helps politicians avoid blame for high-profile crimes (Flom and Post Reference Flom and Post2016; Holland Reference Holland2013; Roberts et al. Reference Roberts, Stalans, Indermaur and Hough2002). Easing regulations on the use of force could also gain politicians the trust and support of police officers who cherish greater discretion to deal with criminals or suspects. Officers (and their families) might then rally in support of such politicians come election time (Papanicolaou and Papageorgiou Reference Papanicolaou and Papageorgiou2016). In some countries, politicians might also enact these changes in search of, or retribution for, contributions from police unions to their campaigns (Zoorob Reference Zoorob2019).

Nonetheless, inciting police violence can also be costly for political incumbents. Given police officers’ often subpar preparation, the stressful situations they routinely encounter, and the inevitable agency costs involved in monitoring them, incidents involving “unjustifiable” police violence are highly probable. Civil society organizations, rival politicians, and the media can blame incumbents for these incidents, especially when these are perceived as “disproportionate” relative to the risk posed by the victim or harm “innocent bystanders.” Subsequently, politicians may face street protests, media probes, legislative inquiries, and even judicial investigations impugning their expansion of police discretion. González has shown how police scandals can unify typically fragmented societal views on policing and ignite reform (González Reference González2020). Furthermore, police brutality can spark violent responses from criminal actors, undermining citizen safety and withering the incumbent’s approval.

Restraining police violence is not without costs, either. It requires implementing various policies that increase political supervision of police activities, which demand significant fiscal costs, require trained professionals, and sour police since they reduce its autonomy. Implementation is thus costly and imperfect. Political mandates may be acquiesced by the leadership but then ignored by street-level officers. Reformist politicians have often also faced more aggressive opposition by police, including work stoppages, authorizations to underworld acolytes to commit crimes, and personal threats. Police have even murdered prosecutors and security officials investigating their complicity with organized crime (Cruz Reference Cruz2016, 375). Moreover, public opinion may still criticize incumbents for lingering incidents of police violence or demand more aggressive policing if violent crimes are perceived to increase.

Finally, politicians can also benefit from restraining police violence. Police abuse of force most often targets the poor as well as racial, ethnic, and religious minorities, who may be a key constituency of most politicians—not just center-left progressives. As Wilkinson has shown, Indian politicians are more likely to mandate that police protect Muslim minorities from violence by Hindu majorities when they face more competitive elections or when these minorities are part of their constituency or that of their coalition partners (Wilkinson Reference Wilkinson2006). Finally, restraining police violence could increase police legitimacy by expanding citizen cooperation with police and increasing its effectiveness (Mazerolle et al. Reference Mazerolle, Bennett, Davis, Sargeant and Manning2013), which can benefit the incumbent by enhancing citizen safety.

Police Comply with or Defy the Incumbent

Police are not unproblematic agents of elected politicians. This judgment does not just apply to rank-and-file officers, who as “street-level bureaucrats” have great discretion to deal with the situations they encounter (Lipsky Reference Lipsky2010). Higher-ranking officers can also deviate significantly from politicians’ mandates.

The general incentive for police, especially high-ranking officers, to comply with politicians’ preferences is straightforward: it can make or break their careers and affect their organization’s budget, consequently influencing their subordinates’ loyalty. In contexts where politicians can decide or influence police promotions, transfers, or removals, especially where such decisions are typically arbitrary and obscure, it certainly helps officers’ careers to have political support—or at least not to have stepped on any major politician’s toes. Moreover, politicians can sign off on police budgets, and while the margin to cut police salaries or pensions is likely low, they can impose hiring freezes, cut overtime pay, or slash funds earmarked for upgraded technology, equipment, or infrastructure. Finally, in weakly institutionalized contexts, where police—and sometimes politicians—partake in illicit enterprises, politicians can offer judicial immunity to police in return for a cut of the rents they extract from crime or the operations they carry out against political opponents (Faull Reference Faull2007; Flom Reference Flom2022; Verma Reference Verma1999).

Given this incentive structure, why and how would police defy politicians and deviate from their mandates? Police may defy incumbents when they perceive that political subservience would harm their individual career trajectory or organizational well-being, whether in terms of material interests, social prestige, or corporate cohesion (Pion-Berlin and Trinkunas Reference Pion-Berlin and Trinkunas2010). For instance, police can perceive increased political supervision as interfering with their capacity to collect rents from criminal activities, which can be a substantial share of their informal budget. Alternatively, police may interpret the standard goals of punitive policies, such as more arrests, seizures, or clearance rates, as impinging on the flexibility they require to regulate crime (Muir Reference Muir1979, 10). They may also perceive that punitive policies undermine their social legitimacy and reduce citizens’ willingness to cooperate with them, which is fundamental to solve crimes (Denny et al. Reference Denny, Dow, Pitts and Wibbels2022). Finally, police commanders may worry that political mandates, either inciting or restraining violence, will spark dissatisfaction or dissent amongst their subordinates, thus reducing the organization’s cohesion.

When police do not view political mandates as legitimate, they can resist in multiple ways. When politicians incite violence, police can drag their feet, pretending to follow instructions while effectively ignoring them, thus undermining the administration and waiting for a politician more favorable to their interests to come to power (shirking). Alternatively, when politicians seek to restrain violence, police may control crime more aggressively than mandated, whether by engaging in more discretionary stops, searches, and seizures—which are more likely to result in human rights abuses—or by confronting with non-state armed actors (sabotage). In short, while both shirking and sabotage challenge politicians’ authority, they involve different actions by the police.

Shirking and sabotaging might be more convenient for police than protesting, which requires greater collective action and produces identifiable ringleaders. Additionally, the informational advantage that police agents have over their political principals makes shirking and sabotaging more appealing. Negative outputs of police work such as homicides or police brutality can be masked as unavoidable amid confrontations with criminals, or concealed altogether, such as when victims are disappeared. In this sense, my typology sheds light on two understudied scenarios of police defiance against politicians: when they disobey mandates to restrain violence (sabotage) and the less likely but still plausible scenario, where police defy incumbent mandates of punitive policing (shirking). Table 1 shows how the interaction between political decisions and police responses yields four different types of police strategies to administer violence, in other words, to use violence for political purposes.

Table 1 Police administration of violence: types and empirical predictions

Source: Author’s elaboration.

I label these four types of police administration of violence as 1) peacekeeping, 2) punishing, 3) sabotaging, and 4) shirking. If these outcomes were understood as equilibria, compliance scenarios should be more stable than defiance scenarios. When police comply with politicians, external factors such as societal mobilization against police violence or electoral turnover are necessary for government officials to modify security policies and, subsequently, for police to reevaluate their compliance. Meanwhile, under the police defiance scenarios, conflict should not persist for long. Either politicians will prevail and force police compliance by removing the leadership, changing statutes, and modifying strategies or tactics, or the police will prevail, and politicians will have to modify their initial position, such as by changing cabinet officials, or, more rarely, resigning from office. It is beyond the scope of this paper to analyze the factors that cause political mandates or police compliance (or lack thereof). However, I will address some potential explanations, as well as permutations of this strategic interaction, in the final section.

Peacekeeping (Preventive Policing)

The first type, peacekeeping, covers contexts in which police comply with politicians’ mandates to restrain the use of force. In this type, politicians instruct police to avoid continuous confrontation with criminal actors and forgo, or at least limit, aggressive crime control strategies. To this end, politicians may reduce the scope of situations addressed through forceful tactics, establish de-escalation procedures as the norm for police interactions with citizens, deploy police strategically to minimize confrontations with criminal actors, and promote organizational changes to strengthen relations between police and the community. If the police comply with these policies, one would expect a decrease in the deaths and injuries caused by police as well as in the number of police raids, arrests, and deployment of tactical units.

One of the foremost examples of peacekeeping is the Pacification Police Units (UPP in Portuguese) program carried out in the state of Rio de Janeiro, Brazil, which features one of the most violent police forces in the world. This program was most successful between 2008 and 2013, before succumbing to the state’s political and fiscal crises. With newly incorporated police trained in human rights and de-escalation settling permanently in various favelas, the state government managed to decrease homicide rates but accomplished an even greater reduction in police lethal interventions. Police, while initially resistant, complied with the government’s policy shifts, unlike in other moments in Rio’s history. The partisan alignment between the state, local, and federal governmentFootnote ⁸ ensured political and financial support for this program, thus decreasing motivation for police resistance. As one police commander told me, “If the governor decides he wants to put all the police in the UPP, what can the commander do? He has to obey, otherwise he’s out.”Footnote ⁹ At the same time, the fact that new recruits were ushered in directly from the academy to the UPPs meant that they would not—in theory—be tarred by the shady practices of veteran officers. The program had a massive impact. Magaloni and her coauthors (Reference Magaloni, Franco-Vivanco and Melo2020) estimate than police killings fell by 40% in neighborhoods where UPPs were installed. Throughout the state, police killings decreased from 1330 in 2007 to 400 in 2013. During this period, largely due to the perceived effectiveness of the UPP, the governing party (the PMDB) won three consecutive gubernatorial elections.

Rio offers another effective, albeit more limited, instance where governments successfully restrained police violence. In 2020, the state government enforced a mandate by the Brazilian Supreme Court that suspended police raids into favelas after a fourteen-year-old was killed by the police. Trudeau (Reference Trudeau2022) reveals that, in the five months after the order was enacted, police lethal interventions decreased by more than 60%. Similarly, in the United States, Mummolo (Reference Mummolo2018b) reports that police largely complied with changes in protocols for stopping, questioning, and frisking criminal suspects, aimed at decreasing police arbitrary stops which predominantly targeted individuals from racial and ethnic minorities. Several other transitional democracies, including Georgia (Light Reference Light2014), Nicaragua (Cruz Reference Cruz2011), and South Africa (Altbeker Reference Altbeker, Hinton and Newburn2009), successfully reformed their police’s organizational structure and operating procedures, and managed to—at least temporarily—reduce police lethality and human rights abuses. Whether through criminal justice or police reforms that restrict police’s use of torture during interrogation (Magaloni and Rodriguez Reference Magaloni and Rodriguez2020), changes in policing strategies—such as in Rio’s Pacification program—or temporary suspensions of aggressive tactics, governments can effectively extract police compliance and restrain its violence, even where it was previously rampant.

Punishing (Punitive Policing)

Punitive policing, often referred to as tough-on-crime, “mano dura,” or law-and-order policies, involves politicians inciting police violence by granting the police more discretion to use force, and police complying with such directives. In this scenario, politicians may change the criminal justice code or departmental regulations to ease restrictions on police brutality, while also using the bully pulpit to rouse aggressive police behavior. Punitive policing usually results in higher arrest and incarceration rates, as well as increased reports of torture, beatings, and lethal interventions, without decreasing criminal violence or increasing police legitimacy (Braga and Kennedy Reference Braga and Kennedy2020).

Punitive policing was a recurring security policy in Rio de Janeiro before the advent of the previously described Pacification program. In the mid-1990s, Rio’s governor, Marcello Alencar, instituted a system known as “Faroeste” (Far West), which rewarded officers who engaged in more “courageous” actions—including the use of lethal force—with bonuses and promotions. As a former Military Police captain told me, “[Faroeste] rewarded the police force’s repressive actions: seizures, arrests, and confrontations, even if they resulted in the death of the criminal; it increased police salaries. That really mobilized the military police apparatus, where the value of promotions is huge.”Footnote ¹⁰ This policy shift led to an immediate surge in state violence: police killings in Rio increased by 63% the first year Faroeste was in place, as officers complied with the government’s incitement to use (lethal) force (Cano Reference Cano1997).

Politicians from other countries have enacted similar policy shifts, adopting the rhetoric of zero tolerance policing. When running for the governorship in the province of Buenos Aires, Peronist politician Carlos Ruckauf campaigned (and won) on a security platform that promised to “put bullets into criminals” (López Echagüe Reference López Echagüe2000). Once in office, he reversed the criminal justice reforms introduced by his predecessor and granted police more discretion to detain, interrogate, and use force against suspects. Reports of police beatings, torture, and killings increased dramatically during his tenure. Many other developing and established democracies have also resorted to punitive policies, expanding criminalization and police discretion, leading to mass incarceration and more casualties of police violence (Bonner Reference Bonner2019; Newburn Reference Newburn2007; Roberts et al. Reference Roberts, Stalans, Indermaur and Hough2002).

Punitive policing is still pervasive across the world. The most extreme contemporary examples are El Salvador under President Nayib Bukele (2019–present) and the Philippines under former president Rodrigo Duterte (2016–2022). In El Salvador, Bukele has tested the boundaries of democracy by declaring a state of exception and suspending various constitutional rights. He has authorized police to lash out against presumed gang members, instituting daily arrest quotas (Renteria Reference Renteria2022), and filling up prisons with detainees, many of whom are still awaiting trial (Phillips Reference Phillips2022).Footnote ¹¹ Police compliance has, in this case, proven highly convenient to Bukele, whose popularity sits above a whopping 80% (Galdamez Reference Galdamez2022). This paradigm is not new in El Salvador as in the 1990s, Nationalist Republican Alliance (ARENA in Spanish) politicians expanded the definition of “gang membership” and deployed mass detentions and incarcerations to harness electoral support (Holland Reference Holland2013; Wolf Reference Wolf2017).

Meanwhile, in the Philippines, former president Rodrigo Duterte unleashed a no-holds-barred war on drugs since taking office in June 2016. On the eve of his electoral victory, he announced: “You drug pushers, holdup men, and do-nothings, you better get out because I’ll kill you” (Baldwin and Marshall Reference Baldwin and Andrew2017). He appointed an officer from his hometown of Davao as the head of the national police and “gave him free rein to roll out Davao’s crime fighting model across the Philippines” (Bouckaert Reference Bouckaert2017). Duterte’s war on drugs has led to more than 12,000 killings to date, of which at least 2,555 are attributed to the national police (Jensen and Hapal Reference Jensen and Hapal2018). Like Bukele, Duterte was not the first to adopt this approach in his country but took it to an unprecedented scale.

Sabotaging (Active Police Defiance)

Most of the literature on police violence has focused on the previous two types—peacekeeping and punishing—assuming police compliance with governing officials. By contrast, the typology introduced in this paper expands our understanding of police violence as a political response by also exploring how police may administer it in defiance of political incumbents, whether through sabotaging or shirking.

Sabotaging implies that politicians seek to restrain police violence, but police disobey this directive. In this case, politicians enact restrictive measures, while police officers, whether at the instigation of their commanders or due to internal revolt, adopt unsanctioned aggressive behaviors. This police strategy sabotages politicians’ objectives not just because it deviates from the mandate of those who—unlike police—have electoral legitimacy, but also because it increases levels of state—and potentially criminal—violence in the incumbent’s district, sparking discontent from civil society, the media, and rival politicians, and undermining the incumbent’s support. In other words, police sabotage produces a “negative output” regarding the government’s purported goal of decreasing violence and ensuring security (Brehm and Gates Reference Brehm and Gates1999). Police sabotage might also compel politicians to remove appointed security officials, reverse their policy course, or even resign their position. In short, this type of administration of violence involves the police actively defying political orders and ramping up violence against civilians (or even politicians themselves) to undermine or destabilize the incumbent.

Sabotage was a frequent reaction by police to the myriad reform attempts promoted by politicians in transitional democracies during the 1980s and 1990s, as they sought to cleanse departments from the violent and corrupt reputation they had (appropriately) gained during the preceding dictatorships. Reformist politicians often encountered active resistance from the police when they attempted to restrain its use of lethal force. In Rio de Janeiro, populist governor Leonel Brizola (1982–1986) attempted to democratize the state’s Military Police by prioritizing the protection of human rights, changing training modules, and restricting police operations in marginalized neighborhoods. Subsequently, he endured not just police protests but also a flurry of police killings and acts of criminal violence authorized by police to undercut his effort (Hollanda Reference Hollanda2005; McCann Reference McCann2014). The ensuing violence sapped Brizola’s popularity and aided the election of his rival, Wellington Moreira Franco, who reversed his predecessor’s security policies and expanded police discretion to raid favelas and apply lethal force (Arias Reference Arias2017, 67). The governor of neighboring São Paulo, Franco Montoro (1982–1986), faced similar sabotage ploys from the Civil and Military Police to derail his reform efforts in the early 1980s (Caldeira Reference Caldeira2000; González Reference González2020; Mingardi Reference Mingardi1992).

Meanwhile, in Argentina, the police of the province of Buenos Aires carried out similarly destabilizing maneuvers against reformist politicians in the late 1990s and early 2000s. In response to government interventions, which included widespread police purges, crackdowns on police rackets, and greater political supervision, police threatened politicians, enabled their underworld associates to commit violent crimes, and increased use of deadly force against alleged criminals (Saín Reference Saín2004; Reference Saín2008). Their strategy seemed to work as, between 1996 and 2004, Buenos Aires’ three governors shuffled fourteen security ministers, some of whom remained in office for only a few months and most of whom left due to police resistance to their authority. A former security minister described the police’s intimidatory practices following the onset of reform at length:

Similarly, in the neighboring Argentine province of Santa Fe, various security officials of two Socialist administrations (2008–2016) recounted the pressures and intimidation they faced during their attempts to reform the police. One former security minister told me that he found notes in his desk that said, “Get out, usurper,” was constantly followed, and stopped going to the cafeteria for fear that police had poisoned his food.Footnote ¹³ He resigned after suffering a near heart attack. Another former minister described the police’s intimidatory capacity quite graphically: “Hitting these guys in the head requires a big dick. The police know everything about you.”Footnote ¹⁴ The failure of their intended reforms, in large part due to police resistance, did not directly increase police violence, but enabled the proliferation of drug trafficking rackets with police complicity, which contributed to a doubling of homicide rates in the province between 2008 and 2013 that undermined the government’s popularity.

Many other cases of police reforms succumbing to violent police resistance have been documented. Tankebe has argued that some African police forces have used violence in defiance of the incumbent to enhance their own legitimacy among the population (Tankebe Reference Tankebe and Cheliotis2010). Cruz has shown that police in Honduras have even murdered prosecutors and political officials who were investigating their corrupt practices (Cruz Reference Cruz2016).

The police’s unsanctioned use of violence, whether directed against citizens or politicians themselves, sends a clear political signal of the extent to which the force will tolerate changes that threaten its interests and power, including its rent extraction from criminal enterprises. While perhaps less frequent than non-violent resistance, it is a recurring strategy used by police to oppose, undermine, and even destabilize the political principals they supposedly serve.

Shirking (Passive Police Defiance)

Finally, police can deviate from politicians’ mandates by not abiding with incumbents’ orders to carry out more aggressive enforcement. I refer to this strategy as shirking or passive police defiance. In this scenario, police commanders might prefer to maintain their institutional legitimacy and protect their officers’ physical integrity rather than follow controversial political mandates. While politicians may incite police violence through policies and discourse, police can exploit informational asymmetries or policy ambiguities to avoid or minimize confrontation with organized criminal actors, aggressive crime control, or brutal repression of social protests (Brehm and Gates Reference Brehm and Gates1999; Pion-Berlin and Trinkunas Reference Pion-Berlin and Trinkunas2010, 396). While politicians might turn a blind eye toward police defiance if police are able to maintain acceptable levels of social order, this equilibrium might not endure if police passivity leads to dramatic increases in violent crime. In such cases, politicians are likely to change police authorities and modify institutional incentives to seek greater police compliance.

This response is the rarest for multiple reasons. Police officers often self-select into an occupation that involves the exercise of violence (Dharmapala, Garoupa, and McAdams Reference Dharmapala, Garoupa and McAdams2016). Therefore, they are more likely to align with (and obey) politicians who espouse punitive policies. Furthermore, politicians might try to conceal situations where police officers question their authority. They might, therefore, reframe their directives to police commanders or shift their public stance.

A highly controversial episode involving police violence in Argentina illustrates the dynamics of police shirking. In 2018, center-right president Mauricio Macri (2016-2019) and his security minister Patricia Bullrich publicly congratulated a police officer who had shot and killed a fleeing criminal while off-duty. Bullrich subsequently expanded the federal security forces’ authorization to use firearms, further legitimating this officer’s actions—and signaling greater tolerance of aggressive enforcement to other officers.Footnote ¹⁵ This was not the first time that the administration had sided with security forces amid allegations of human rights violations or expressed support for a tougher stance on crime.Footnote ¹⁶ Nonetheless, according to the human rights organization Center for Legal and Social Studies (CELS), police killings in Greater Buenos Aires—which is under jurisdiction of the Federal security forces as well as the police of the province and the city of Buenos Aires, all of which were governed by members of Macri’s center-right party—decreased by 21% from 116 in 2018 to 91 in 2019, whereas they had been stable the previous two years (117 and 107, respectively) (Centro de Estudios Legales y Sociales 2023). There was also a 25% drop in the number of police killed that same year (from 31 to 23), which could indicate that officers were evading riskier encounters. The fact that Macri’s administration was in its final year and declining in popular support due to the growing economic crisis could be an explanation for this police response.

Police shirking in law enforcement can also be observed in how some forces respond to social protests. While these demonstrations frequently break the law, police will often take an arm’s length approach or use “velvet gloves” rather than “iron fists” to handle these events, thus avoiding confrontations with activists and potential repercussions for themselves (Earl Reference Earl2011). In this vein, Bolkvadze and Lindvall show that police in London and Paris have frequently chosen to manage or stand back from protests rather than repress them. For instance, during the 2011 London riots following the murder of a British man of African-Caribbean descent, police refrained from repression partly because they were “worried about accountability should they act improperly in the lens of the media” (Briggs Reference Briggs2012, 388–89, quoted in Bolkvadze and Lindvall Reference Bolkvadze and Lindvall2022, 27). While the government may initially be supportive of tougher enforcement, police might not trust that the government will uphold that stance in the face of media scrutiny. They might also fear that they will be investigated and disciplined by the political opposition or the courts, regardless of the incumbents’ support.Footnote ¹⁷

In some cases, police shirking can also prove destabilizing toward administrations. When former Bolivian president Evo Morales ordered police forces to repress protests against his controversial victory in the 2019 elections, the police not only refused to follow this mandate but ultimately joined the protesters against the government (Collyns Reference Collyns2019). This institutional refusal paved the way for Morales’ undemocratic ousting from office. Former Peruvian President Pedro Castillo experienced a similar fate when he tried to impose a self-coup in late 2022 (Buschschluter Reference Buschschluter2022). Although there is no counterfactual, one can assume that the costs of police repression, not just in terms of lives lost but also of political legitimacy, would have been substantial, perhaps also leading to the president’s early exit from office.

Despite the punitive mandates of executive politicians, officers’ self-selection toward violent occupations, and the level of conflict and risk they encounter, police forces have repeatedly deviated from carrying out politicians’ calls for greater violence. Like the military, which can opt to not repress civilian uprisings, defying politicians’ orders—and thus seriously undermining their power—police can also challenge mandates of aggressive enforcement when they perceive them to conflict with their individual or collective interests.