Abstract
The problems of violence in Latin America are often reiterated, yet understanding how and why violence declines is far less common. While urban violence takes different forms and has a range of motivations, we suggest that strengthening political and social institutions are important in violence reduction processes. We examine this using a comparative analysis of two cities which have recently seen unusual and marked reductions in lethal violence: Bogotá in Colombia and Recife in Brazil. Drawing on primary data collection, the case studies suggest that novel leaders who take advantage of critical junctures can deliver unexpected improvements to public security; and improvements are linked with institutionalising progressive security policies, increasing accountability of political institutions, and social reforms encouraging civic values and commitments to non-violence. While findings are specific to these two cases, they may plausibly apply to a broader range of cities, such that commitments to improve public policy and political institutions can overcome structural risk factors that foster violence.
Keywords
Introduction
Latin America is one of the world’s most violent regions (UNODC, 2014), with violence primarily occurring in cities. Given the developmental challenges urban violence poses (World Bank, 2011), urban violence reduction is imperative when the region’s cities are considered ‘fragile’ (Muggah and Savage, 2011) and experiencing a ‘humanitarian crisis’ (Bernal and Navas, 2013). The diversity and complexity of what is ‘urban’ in Latin America makes this a challenging process. Cities face a range of ‘violences’ with a complex array of motivations, expressions and policy responses, and discourses of urban settlements as universally ‘fractured’ provide an unhelpful ‘impasse’ (Rodgers et al., 2012: 5). Moreover, while conceptually distinct, political, civic, economic, social and criminal violence may in reality frequently overlap (Beall et al., 2013; Moser, 2004).
What then can we say about violence reduction in the Latin American city? While acknowledging the region’s diversity of violence, we suggest there is utility in understanding why in general, some cities are more or less violent than others; and why trajectories of violence change. Here we examine declines in lethal forms of urban social violence – namely violence between individuals or small groups that lacks an inherent political motivation (Fox and Hoelscher, 2012). Several explanations are used to account for urban social violence in Latin America, often related to political processes such as democratic transitions and social insecurity (Arias and Goldstein, 2010; Fox and Hoelscher, 2012), economic conditions related to liberalisation and socio-economic inequality (Rudqvist and Hettne, 2003), and socio-demographic structure associated with informal urbanisation and spatial segregation (Beall et al., 2010). Moreover, violence may also be a product of the state itself, whose history of abusive practices include involvement in right-wing paramilitary activities (Ramírez, 2010), extreme police violence (Ahnen, 2007; Caldeira and Holston, 1999) and large-scale corruption (Davis, 2010). Indeed, some consider violence – and particularly violence in the city – as a fundamental, constitutive aspect of Latin American democracies (Arias and Goldstein, 2010). Further, a range of other drivers cut across these categories, including small-arms availability, gang activity, the narcotics trade, alcohol and substance abuse, poor public education, large youth cohorts, and a poor criminal justice system (Briceño-León et al., 2008; Soares and Naritomi, 2010). Consequently, a complex political economy including top-down and bottom-up processes affects both causes of, and responses to, urban violence in Latin America; and we recognise distinct drivers may underlie violence escalation and violence reduction. 1
Our approach is based on two recent success cases which have markedly different trajectories of violence relative to most other Latin American cities: Bogotá, Colombia, and Recife, Brazil. We argue that despite certain differences related to city size and importance, governance structure and dynamics of violence, both cases provide evidence that institutionalising accountable public security policy as part of broader social and political institutional reforms coincided with the reduction of lethal violence. 2 In both cities, critical junctures related to an environment of political institutional reform, and a sense of emergency around the topic of violence created openings for public security policy innovation by novel political actors. These innovations received broad-based buy-in from key political stakeholders and civil society, and became part of the society’s institutional configuration that was difficult to revert; and consequently, have contributed to a sustained reduction of violence beyond the presence of initial promoters. Moreover, these processes differed from broader national dynamics and were inherently urban in nature, both in the mobilisation that supported early reforms, and where policy impacts occurred. 3
While related to findings about the role of the security sector regarding community policing, handgun restrictions and similar urban policy interventions (Felbab-Brown, 2011; Goertzel and Kahn, 2009), our analysis locates this in a broader context of what we term institutionalised policy change. Thus we situate our analytical position at a middle ground between macro-level explanations encompassing broad social processes such as civilisation, shifts in political regimes, economic development or changes in demographic structure (Eisner, 2003; Pinker, 2011); and micro-level accounts of specific policy measures. This framing thus situates institutionalised policy change at the centre of violence reduction processes in our cases, while also allowing for additional explanatory factors.
Violence reduction in Latin America
While violence in general (Buvinic et al., 1999; Frühling et al., 2003) and, more recently, urban violence have been of particular interest to scholars and policy-makers in Latin America (Auyero and Burbano de Lara, 2012; Koonings and Kruijt, 2007; United Nations Development Programme (UNDP), 2013), research on violence reduction is still incipient. Micro-level policy interventions are frequently touted as reasons for violence reduction. Successful examples include alcohol restriction (de Mello et al., 2013), handgun ownership controls (Restrepo and Aguirre, 2010) and focused police presence in key areas (Di Tella and Schargrodsky, 2004). In Brazil, studies have traced positive impacts of specific crime-reduction policies using qualitative historical approaches (Leeds, 2007; Ratton et al., 2014; Willis and Prado, 2014). In Mexico, community organisation has been identified as promising tool (Vilalta, 2013), while zero-tolerance policing policies have been critically appraised (Davis, 2013). Other recent studies focus on efforts of political coalitions and progressive public policy in achieving violence reduction in Colombian cities (Gutiérrez Sanín et al., 2013), and in explaining recent homicide declines in São Paolo (De Mello and Schneider, 2010; Goertzel and Kahn, 2009). The interrelated literatures on gang violence, illicit urban social orders and local drug trafficking have contributed alternative strategies to deal with violence. Increased contact between gang members to reduce intergroup enmity (Strocka, 2009), negotiations with criminal organisations such as gangs (Wennmann, 2014), humanitarian intervention in situations other than war (Bernal and Navas, 2013) and preventive second-generation strategies for gang members instead of repression (Jütersonke et al., 2009) have been cited as possible mechanisms to deal with ‘non-state armed groups’ (Rodgers and Muggah, 2009) and the ‘humanitarian crisis’ of violence in the Latin American city.
However, while public security policies and engaging with gangs and violence entrepreneurs may in some cases encourage urban violence reduction, these are largely exceptions to the norm. More often, policy initiatives in the medium to long term prove ineffective. They are captured by elite or criminal interests, lack local ownership or prove unsustainable because of funding or capacity deficits. One critical reason for these failures is that policy is initiated in institutional contexts inappropriate for sustained success (Basombrío and Dammert, 2013; Cano, 2006), and the development and consolidation of formal democratic institutions is hindered by informal, illegal arrangements involving both criminal organisations and state officials (Arias, 2013). However, while micro-level policy interventions may be suited to explain short-term reduction in urban criminality, sustained decreases in lethal violence may be more likely in situations where policy changes complement, or are integrated into, wider reforms to political and social institutions. Indeed, statements such as ‘for such processes to be sustainable, they must be embedded in broader social and political transformation processes’ (Wennmann, 2014: 269), or ‘lasting solutions to violence are likely to depend on broader efforts to establish systems of order and social institutions’ (Arias, 2013: 282) are recurrent in the literature on violence reduction in Latin America. Therefore, focusing on a higher level of analysis looking at broader institutional change rather than simply exploring various isolated micro-interventions may be promising.
Institutions and violence reduction
Inherent here is a position that institutions – formal and informal norms and guidelines that order societies (North et al., 2009) – can have a decisive impact upon violence. 4 Since O’Donnell suggested that states represent ‘manifold social relationships engaged on the basis of stable (if not necessarily approved) norms and expectations’ (1993: 1356), scholars have acknowledged the role political institutions play in shaping violence, particularly in urban areas (Auyero et al., 2014; Brinks, 2003). Perverse institutions of law and order can threaten human security (Caldeira and Holston, 1999), and limited political institutional legitimacy can incentivise violence (Nivette, 2014). Consequently, a ‘major factor shaping urban violence is the inability of state institutions to regulate and ultimately manage the legitimate use of force’ (Muggah, 2012: 49). As such, if weak democratic institutions encourage violence, improving political institutions may be important in violence reduction processes by regulating the (un)acceptance of illegitimate force.
Historical institutionalist perspectives emphasise the potential importance of critical junctures, where key actors take advantage of favourable conditions that enable them to follow particular trajectories of change which are self-reinforcing. Here, ‘structural influences on political action (can be) significantly relaxed for a short time’, such that ‘the range of plausible choices open to powerful political actors expands substantially, and the consequences of their decisions for the outcome of interest are potentially much more momentous’ (Capoccia and Kelemen, 2007: 343). While emphasising the importance of critical junctures, our analysis also considers ‘critical antecedents’ which ‘shape the choices and changes that emerge during critical junctures in causally significant ways’ (Slater and Simmons, 2010: 887). Here we pay attention to the political economy of how critical junctures shaping unique (and often unlikely) process outcomes emerge, especially where this involves reconfiguring institutional relationships between state and citizen. 5 Importantly, these processes are highly contingent and mutually constitutive involving multiple stakeholders (e.g. see Goldfrank, 2007; Rodgers, 2010).
Relative to violence reduction in our cases, these ‘contingency’ factors relate to changes in institutional arrangements, civil society demands, and – ultimately – trajectories of violence. We suggest such factors strengthen stakeholder agency; and actors exist not merely as observers, but are ‘capable of acting on “openings” provided by such shifting contextual conditions’ (Thelen and Steinmo, 1992: 17). Our contention is that following initial openings, ‘tipping point’ conditions can arise where political actors are able to introduce policy or institutional changes which cross certain thresholds of stakeholder buy-in, encouraging further commitment, thus reproducing earlier choices. As such, our study complements existing literatures with a comparative urban perspective; and theorises how micro-level interventions – organised around the objective of increased institutional accountability – contribute to violence reduction.
Empirical strategy and case description
We employ a comparative process tracing methodology (Bennett and Checkel, 2014) examining processes leading to outcomes of interest in our two cases, Bogotá and Recife. We use a deviant case approach, 6 choosing cases unique among Latin American cities which show extraordinary reductions in lethal violence; and appear distinctive compared with broader national or regional dynamics regarding policy reform and violence trajectories. Bogotá was a pioneer in implementing novel institutional approaches for reducing violence, and Recife adopted some of Bogotá’s lessons learned. 7 Unlike many previous studies of Latin American violence reduction focusing on single cases, we add inferential leverage using a comparative approach. Should our theory of institutionalised policy change apply to both cities, other cases with similar scope conditions could pursue violence reduction through similar means.
Primary data were collected during in-depth interviews 8 with key stakeholders including government officials, police officers, judicial figures, NGOs and academics in Bogotá (October 2013–March 2014) and Recife (February–April 2013). Key informants were either directly involved in administration of local-level policies, or experts with long-standing experience observing the two cities. Interviewees were asked about the evolution of public security, relevant actors in security governance, policies and reforms elaborated in recent years, and other factors contributing to violence and its reduction. To triangulate, contrast and historically situate interview material, secondary data from newspapers, NGO reports, public opinion surveys and official documents were collected.
Bogotá
Bogotá is the capital of Colombia with approximately eight million inhabitants. During the 1980s and early 1990s, it was one of the most dangerous cities in Colombia, with homicides peaking at over 80 per 100,000 inhabitants. However, contrary to national homicide rates that peaked in 2002, Figure 1 shows that with the exception of a few minor setbacks Bogotá has seen an unprecedented drop in lethal violence since the mid-1990s, reaching a 30-year low of 16 homicides per 100,000 in 2013 – lower than any other major Colombian city. 9
Homicide rates in Bogotá and Recife, 1983–2013.
Recife
Recife, with a metropolitan population of 3.8 million, is the capital of Pernambuco in the northeast of Brazil. At its peak during the late 1990s and early 2000s, Recife had one of the world’s highest homicide rates at over 90 per 100,000. Yet lethal violence has markedly declined over the past decade (Figure 1), sharply contrasting with the other eight states in Brazil’s northeast (Waiselfisz, 2011). While once triple the national average, homicides in Recife have declined 60% since May 2007 to under 30 per 100,000.
Comparative process tracing
We suggest that institutionalising public security policy and the related reform of key political institutions can account for reductions in violence in Bogotá and Recife. While refraining from claiming this is the only causal explanation given other factors likely played some role, evidence suggests violence reduction is strongly linked with successful changes in the policy and institutional context in each city. Importantly, key drivers of violence reduction appear related to local dynamics, distinct from the broader contemporaneous national dynamics in Brazil and Colombia. In Colombia, violence increased in most regions and cities throughout the 1990s and early 2000s because of heightened conflict activity which dominated the debate about public security during this period and left little room for urban violence reduction policies. In Brazil national-level public security policy was generally ineffective, and despite cerain selected exceptions, few state-level changes occurred.
The institutionalisation of policy change in both cases was different, but can be categorised according to three phases (Figure 2). Phase 1, commencement, represents some form of critical juncture, breaking from past practices of political indifference in the face of increasing violence, allowing path-dependent processes to take root. Phase 2, traction, highlights how policy shifts and political institutional reforms overcame moderate initial opposition within the state and ‘criminal resistance’, and how the alignment of interests among political coalitions and other major actors in government, police and civil society established practices which increased institutional accountability. Phase 3, sustained impact, sees institutionalised policy change, continued institutional transformations, and sustained violence reduction, resulting from effective law enforcement, institutional accountability and increased modalities of participation.
A stylised model of institutionalised policy change and violence reduction.
Phase 1: Commencement
Prior to declines in violence in Bogotá and Recife, the political institutional context was fragmented, dysfunctional and unconducive to societies with strong rule of law. However, in Bogotá during the mid-1990s and Recife during the mid-2000s interests began to shift in ways which encouraged a break with past practices.
Bogotá
Increasing levels of homicide in Bogotá in the 1980s and early 1990s corresponded to national trends driven by drug trafficking, police corruption and impunity. Reflecting an understanding of security as national issue, the Colombian president César Gaviria published his ‘National Strategy against Violence’ in 1991 (and 1993) warning against multiple forms of violence related to guerrilla groups, paramilitaries, narco-traffickers and corrupt agents of the state. While using the term ‘citizen security’, his focus remained clearly on national security issues (Nussio, 2013). Consequently, at this time the mayors of Bogotá – who are responsible for public security and oversee the Metropolitan Police – did not focus on security policy, but saw security rather as a naturally occurring result of economic and social policies, and essentially ceded responsibility to the police for its day-to-day management.
Yet from 1995, the policies proposed and implemented under the leadership of mayor Antanas Mockus marked an important turning point for Bogotá. A series of serendipitous dynamics in Colombian politics and intellectual life relaxed earlier structural constraints and opened up space for new agency. Altering processes of centralised appointment, the popular election of mayors was introduced in 1988 as a means to decentralise state power (Gilbert and Dávila, 2002). In Bogotá, the first three mayors elected by popular vote belonged to the traditional conservative and liberal parties. However, after the 1994 presidential elections and related Proceso 8000 corruption scandal that linked financing from the Cali drug cartel with the presidential campaign of elected president Ernesto Samper, general disenchantment with political elites reached new heights. Importantly, this increased the chances of independent candidates in Bogotá mayoral election later that same year, and Mockus, an otherwise unlikely candidate to win an election in Colombia, emerged as a realistic option. While president of the Universidad Nacional, Mockus came to sudden nationwide popularity through his unconventional and self-deprecating manner of engaging with student protest. As a widely recognised ‘anti-politician’, he won the election with a landslide victory receiving 65% of the votes.
Mockus’ interest in violence and citizenship culture relates to a series of critical antecedents. A large number of scholars started studying violence during the 1980s when it increased dramatically because of heightened competition between drug-traffickers, conflict activity and increased petty delinquency. When these technocrats accompanied Mockus into city government, this interest in violence suddenly became a policy priority. The city of Bogotá (in addition to Colombia as a whole) faced a security crisis with extreme levels of violent criminality, with President Virgilio Barco calling Bogotá in the late 1980s a ‘tremendous chaos, an immense disorder, a collosal mess’ (cf. Martin and Ceballos, 2004: 65). Moreover, a series of choices targeting violence reduction were beginning to emerge in Colombia, with the epidemiologist mayor of Cali Rodrigo Guerrero pioneering several policies in his city, though with limited success. Importantly, Guerrero, and later also Mockus, benefited from the 1991 constitution which explicitly stated that the mayors were responsible for public order in their jurisdictions, a responsibility most other mayors did not assume. Mockus also benefited from a favourable context for policy implementation. His predecessor Jaime Castro saved the city from bankruptcy and increased its revenues with a new tax scheme; and developed the ‘Organic Statute’ that gave the city administration greater independence in policy-making from the clientelist city council (Pasotti, 2010). 10
In this context, Mockus proposed a new strategy to deal with urban violence. His flagship policy called cultura ciudadana (citizenship culture) focused on combining institutional accountability with integrated policies to address violence. Reflecting on the novel agent who guided Bogotá’s public security policies into a new direction, the then director of the Metropolitan Police Luis Ernesto Gilibert said at his retirement that, ‘it was necessary that an unprejudiced and sensitive person would become mayor, so he would understand that it was his own responsibility to look after the social order and the security in the city’ (El Tiempo, 1996b).
As depicted in Figure 1, homicides started declining from 1993, with some suggesting this was due to independent criminal dynamics rather than Mockus’ policy innovations (FIP, 2013). However, while criminal dynamics might explain short-term homicide trends – such as peak levels in 1993 – they cannot account for a sustained reduction as experienced in Bogotá. Our argument therefore relates less to the exact turning point in 1993, but rather to longer-term declines in violence, for which Bogotá stands alone in a country afflicted by conflict and large-scale organised crime.
Recife
In Recife, violence during the 1990s and 2000s coincided with poorly conceived state security policies, the expansion of the northeastern drug trade, and police brutality absent of civilian oversight (Hoelscher, 2013). Mayors of Recife and governors of Pernambuco considered crime a national-level concern, reflecting the ‘hands-off’ approach to public security of most Brazilian subnational politicians. Under the new governorship of Eduardo Campos, however, the introduction of the Pacto Pela Vida (PPV) public security programme in 2007 marked a turning point, 11 and coincided with sustained violence reduction. Facing a decade of endemic violence, the critical antecedents of this juncture are associated with opportune civil society and political mobilisations which matched in purpose and goal.
During the early 2000s, violence primarily occurred in the urban margins. Yet by the middle of the decade, civil society engagement increased when violence encroached on middle-class interests – typified by several murders of wealthy Recifenses. 12 One official recalled this time, saying: ‘If you don’t have a city that is secure for everybody, you don’t have security for anybody. If life is worth nothing in the favelas, then (those) that live there can do anything … People came to understand this because there were so many crimes (with wealthy citizens killed during robberies)’. 13 This was bolstered by greater media demands for urban violence reduction (with Bogotá held up as a success); 14 and the installation of an electronic billboard by journalists – provocatively titled Bodycount Pernambuco – showing the number of murders in the city. Increasing civil society engagement was also supported by a traditionally strong activist culture in Pernambuco, with many NGOs interested in the issue of violence reduction. 15
This collective sentiment marked a break from past indifference, and Campos capitalised by running and winning 2006 state elections on a platform heavily focused on reducing violent crime. Yet beyond simply being an electoral platform responding to social mobilisation, Campos’ ability to politically mobilise following his election victory was both enabled by the initiatives, and encouraged by the failures of previous administrations’ public security policies. Specifically, the Jarbas Vasconcelos governorship (1999–2006) attempted to reform the public security apparatus by creating the Secretaria de Defesa Social in 2001, but did so ineffectively, reactively, non-transparently, and without a long-term view to establish deeper meritocratic institutions governing public security. 16 Therefore, co-occurring ‘mobilisations’ created an environment whereby civil society increasingly demanded a safer city; while politically there was precedent that shallow reforms could not address underlying institutional illegitimacy, nor achieve violence reduction.
These events, including favourable media and public opinion, Campos’ vision for deeper reforms in response to past failures, and an electoral mandate to enact change created a critical juncture where violence reduction processes emerged. This saw the introduction of a technocratic programme – which included stakeholders from politics, civil society and academia – that emphasised new forms of accountability in measuring when and where violent criminality occurred; and monitoring ways the police and judiciary responded. This was predicated upon a new set of institutional ideals designed to modify how both the state and citizens should view the issue of violence, and policies that introduced credible incentives and punishments to reduce violence both in the short and long term (Hoelscher, 2013; Macêdo, 2012). This coalescence of an invigorated civil society and a dynamic political coalition led by a strong central figure created a platform for unique policy approaches to emerge.
Phase 2: Traction
Once initial openings were established, new approaches gained traction in different ways. In Bogotá, changes focused on the creation of a broad ‘civilising culture’ and political reforms addressing the police. In Recife, PPV’s management-led changes increased effectiveness of policing, while also enhancing centralised forms of oversight and institutional accountability.
Bogotá
Under cultura ciudadana, several policies were enacted to modify dysfunctional citizen behaviour and reduce violence. Several emblematic policies included voluntary disarmament, the ley zanahoria restricting nightclub opening hours, more effective judiciary services accessible to citizens, street actors used to enforce traffic rules, and a series of strategies related to policing and reform (Mockus, 2001; Pasotti, 2010). Yet both Mockus and Gilibert were convinced that broader police reforms were needed, addressing not only the availability of technical resources but also greater training. 17 Consequently, over 8000 police officers received tutoring in private universities on topics including human rights and community policing (Acero, 2002). The Frentes de Seguridad Local, launched by Gilibert, were the main community policing initiative during this time. Involving a range of citizens from all neighbourhoods of Bogotá, by 2004 almost 900,000 people participated in this initiative (Bello, 2004).
Citizens were also invited to participate in Escuelas de Seguridad Ciudadana (Citizen Security Schools) to receive training in security issues, with more than 37,000 community leaders participating between 1996 and 2004. This large-scale collaboration between communities and the police developed increased trust, making police officers directly accountable to communities (Martin and Ceballos, 2004: 395). Increased Metropolitan Police spending also covered a series of rewards, housing benefits, higher salaries and civic education programmes for police officers; and countered lingering opposition to new oversight the changes mandated. While the Metropolitan Police report both to the national government and the mayor of Bogotá, civic–police collaboration in Bogotá was unique in the Colombian context, and Bogotá was the only locality where community policing was effectively enacted (Ruiz Vásquez, 2012).
An important element in the institutionalisation of violence reduction policies was the creation of a Unified Information System on Violence and Crime (Spanish acronym SUIVD) in 1995 (Martin and Ceballos, 2004: 219–262). The SUIVD was the result of a technocratic turn in the administration of security, directed first by Álvaro Camacho and later by Hugo Acero, the security councillors of the Mockus administration. Based on SUIVDs transparently tracked violence indicators, an inter-agency committee (Comité de Vigilancia Epidemiológica) met weekly to discuss relevant strategies, and included both traditional and non-traditional actors in security governance. This standing committee allowed for quick responses to the dynamics of violence affecting Bogotá, and while mayoral leadership was key, the holistic approach in accommodating a range of actors strengthened the committee’s institutional foundations and increased accountability. 18
While public opinion and major media outlets were initially sceptical about cultura ciudadana (El Tiempo, 1996a), policies eventually created a virtuous circle which confronted collective action problems related to fear and distrust. This coincided with Riaño’s (2011) account of street mimes (an emblematic part of cultura ciudana), who encouraged pedestrians and drivers to respect zebra crossings and promote the respect of social norms and rules. Bogotá’s former mayor Paul Bromberg, who replaced Mockus when he became a presidential candidate in 1997, remarked: ‘Before, Bogotá was only news when it was bad news. The journalists of that time only wrote about the disasters of the city. That is why Bogotá was seen as one of the worst cities in the world. With cultura ciudadana, Bogotá turned into good news for several years’. 19 Similar sentiments are reflected in scholarly research that praised Bogotá as ‘model city’ (Gilbert, 2006) or ‘pedagogical city’ (Berney, 2011).
Therefore, community participation, transparent information, accountable police services and a holistic policy design were the foundation of sustained reductions in violence (see also Moncada, 2009). Econometric evidence also supports an association between several implemented policies – particularly normative shifts provoked by the cultura ciudadana programmes, and public space interventions – and homicide reduction (Sánchez et al., 2003). As such, it was not only the police and political allies of the mayor’s office, but also citizens, civil society organisations and economic elites who started buying into a technocratically inspired policy approach that had reform of key institutions as a foundation, and which promised increasing collective returns (Gutiérrez Sanín et al., 2013).
Recife
By establishing agencies to collect and manage data on criminality and policing operations, the Pernambucan government both increased information to address criminality, and introduced important accountability mechanisms. 20 This was a paradigm shift in how the state addressed violence and brought police forces and the judiciary – agencies that have closely guarded their independence – under formalised oversight mechanisms. From its initiation in May 2007, regular weekly meetings reinforced new programme values, opening formal channels of communication between the Policia Civil (PC), Policia Militar (PM) and judiciary; and institutionalising new forms of accountability among senior police related to achieving targets in their areas of responsibility (Macêdo, 2012; Ratton et al., 2014).
There was initial resistance from some senior colonels in the PM accustomed to autonomy. Many had established networks of graft and clientelism that supported earnings and future career prospects; and ingrained attitudes supported the ‘legitimacy’ of killing ‘marginals’ among the urban poor. Many police supplemented incomes by protecting narco-trafficking groups, or were involved in illegal paramilitary organisations, grupos de exterminio, responsible for extra-judicial killings. While reforms created ‘protection vacuums’ leading to short-term criminal contestation, these were addressed through new strategies that moved police to high-crime areas. While improving the effectiveness of criminal deterrence, new approaches also shifted norms and behaviours within the police towards reducing the use of state violence, making senior police accountable for homicide reduction in their areas of responsibility, and for the conduct of subordinates. A senior PM colonel remarked: The culture before was: ‘He’s a criminal? Let him die’. Not today. Today we observe ‘The Pact’ is to save lives … [Recently] a colonel in [another northeastern state] told me: ‘The bandidos, they’re to be killed. It’s social control’. I said ‘it’s still going to be counted as a homicide! You could have caught him, interrogated him, arrested his gang. Instead, you just killed the guy!’
21
The working lives of police and emergency services were also improved through the restructuring of the police and bureaucracy. Police previously worked in a system that provided inadequate resources, equipment and salaries; was morally opaque in citizen engagement; and lacked formal oversight mechanisms. Instead, PPV articulated cooperation between police and judiciary, rewarded conduct conducive to violence reduction and human rights, and incentivised accountability throughout the chain of command. 22 Senior officers were promoted based on how successful their units were in achieving targets including weapons and drug apprehensions, arrests and reducing homicides; while lower-ranking officers received target-based bonuses. Further, investigative capacity was strengthened by increased training and investments in evidence cataloguing and protection. 23 Moreover, PPV introduced community policing initiatives which increased information gathered in communities, enabled police to target operations more effectively, 24 and instilled trust between police and citizens. 25
This overview of operational changes in PPVs initial period between 2007 and 2009 highlights that while technocratic, the approach was underpinned by policy changes encouraging widespread buy-in to new norms that no longer condoned violence as acceptable. Minor initial resistance occurred where illegitimate practices among police and criminal actors were threatened, yet were overcome by commitments to credible oversight mechanisms and institutionalising reforms to structures and practices in the criminal justice system. It was upon this new foundation further investments were made and sustained declines in violence occurred.
Phase 3: Sustained impact
Despite critical junctures enabling new policy approaches, and political institutional changes that expanded early successes, the temporal nature of transformations in Bogotá and Recife differ. Figure 1 shows Bogotá’s sustained decline in violence lasting 20 years, while Recife’s remains incipient.
Bogotá
Bogotá’s homicide rates peaked in 1993, yet the sustained reduction was chiefly due to policies enacted by Mockus (1995–1997) continued by his successor Enrique Peñalosa (1998–2000) and during Mockus’ own second mayorship (2001–2003). Principally, policies focused on gradually strengthening political institutions supporting violence reduction. Within the mayor’s office, a secretariat for citizen security was created in 1997 – and directed by Hugo Acero until 2003 – allowing for centralised organisation of security policies. Further, security budgets, specifically those dedicated to the Metropolitan Police, increased continuously throughout this entire period (Acero, 2002: 366).
During the Peñalosa administration, cultura ciudadana policies were complemented with public space interventions, including the construction of the Transmilenio bus system and the destruction of the criminally governed Cartucho neighbourhood – much in line with broader neoliberal prescriptions to transform Bogotá into a global hub (Berney, 2010). According to Peñalosa’s secretary of government, the policies in New York at the time which drew on broken windows theory were used as an example for a new conception of public space in Bogotá. 26 Moreover, during these years of uninterrupted year-on-year violence reduction, civil society organisations, including the Chamber of Commerce of Bogotá and the Bogotá Como Vamos organisation, began civic oversight of security policies by means of novel perception surveys and policy evaluation. The initial policy changes were thus strengthened by efforts of important allies in civil society interested in their continuity (see also Gutiérrez Sanín et al., 2013).
This broad-based support implied significant costs in abandoning the institutionalised path of violence reduction, as exemplified by the mayorship of Luis Eduardo Garzón who took office in 2004. A leftist mayor, Garzón identified the concept of security with a rightist ideology, and adopted alternative strategies to bring about public security, including his flagship food security policy Bogotá sin hambre (Nussio and Pernet, 2013). 27 Coinciding with his ‘lack of leadership’ 28 on public security, and a reduction of formalised cooperation within the local administration, El Tiempo (2006) heralded 2005 as the first year in over a decade where homicide rates increased. This represented an important public declaration of support for more than 10 years of the city’s engagement with the issue of homicide reduction, and a strong example of civic oversight. Such statements obliged Garzón to adopt a firm position on the security issue (Semana, 2006), showing that homicide numbers were a critical point of leverage to compel the mayor to embrace previous approaches to addressing violence. Despite this, homicides increased slightly between 2007 and 2010 because of public policies that contradicted the previous decades’ reforms. and new criminal dynamics related to the demobilisation of paramilitary groups and the re-organisation of urban criminality (FIP, 2013). A lack of interest in accountable security policy, a disastrous term under currently imprisoned former-mayor Samuel Moreno, and a confrontational relationship between the leftist city administration and Álvaro Uribe’s national government all served to temporarily undercut institutional gains achieved since the mid-1990s. 29 Following these setbacks under the Moreno administration, the broader dynamic of violence reduction has since resumed and continued. This owes to simplified accountability measures, increased returns for actors involved, and a set of ‘rules’ about civic behaviour that citizens of Bogotá adopted in response to sustained interest in, and institutionalised response to, the topic of public security.
Recife
Following modest initial homicide rate declines of around 4% in Recife during 2007 and 2008, annual reductions during 2009 exceeded PPV’s stated goal of 12%. Increasing commitment within the bureaucracy, police and criminal justice system created a greater institutional awareness of the issue of addressing violent crime. Particularly important was the improved effectiveness of both the investigative and repressive state apparatus (Hoelscher, 2013); and the suppression of grupos de exterminio, with almost 200 arrests and over 600 police expulsions during PPV’s first year (Alston, 2008). Judicial actors and police increasingly worked together to prosecute cases that were delayed or left unprocessed; 30 and greater efforts were made to improve state–citizen relations following commencement of the Policia Amiga community policing initiative in 2008.
By 2009, the effectiveness of PPV was evident, with declines in violence stemming from changes in – and greater institutionalisation of – policy rather than major increases in investment. 31 Building on initial successes through renewed policy commitments and resource mobilisation, 3000 new police were hired and significant improvements made to homicide investigation capacity. Importantly, these have transcended ‘policies of government’ and have become ‘policies of the State’. 32 Reflecting further commitments, PPV was supported by the introduction of a multi-sectoral municipal-level programme, Pacto Pela Vida do Recife, in 2013. Designed to support PPV in areas of municipal responsibility, Pacto Pela Vida do Recife encompassed public health and drug treatment; at-risk youth outreach initiatives; urban planning and upgrading; and improving public facilities. 33 Discussing this consolidation, a former president of the Chambers of Justice remarked: ‘an expectation about crime control so great [has been created] that no governor who follows will have the courage to change the structures [of PPV] that have been put in place by this government that are so accepted by the population’. 34 While transformations have happened rapidly, evidence suggests policy changes have been founded upon deep shifts in the organisation, structure and accountability of local political institutions; and this institutionalised policy change is the principal driver of declining violence.
Discussion and conclusion
Despite unique public security dynamics in Bogotá and Recife, striking similarities suggest that institutionalising changes in progressive public security policy can contribute to urban violence reduction. Following Soifer (2012), our analysis of critical junctures highlights particular socio-political circumstances in each city as ‘permissive conditions’ which altered underlying contexts; and the process of institutionalising policy changes as ‘productive conditions’ that fostered violence reduction. Responding to civil society mobilisation that was highly urban in nature, serendipitous circumstances created critical junctures allowing leaders with non-traditional policy agendas to move the issue of violence into the public consciousness. Moreover, violence reduction was driven by a technocratic elite relying on a data and management-led policy approach, which itself became institutionalised. In both processes, regular meetings among a large group of stakeholders in public security debates opened previously ineffective channels of communication. While the broad approach taken by the Bogotá administration contrasts with Recife’s focus on policing and judicial operations, both processes gained support from a diverse range of key actors, including economic elites, police forces, civil society organisations and citizens.
The institutionalisation of new public security policies has also been critical, particularly the positive police presence in marginal urban areas and increased engagement of local communities. This has reinforced the accountability of the police and civic authorities responsible for security policies, and was matched with multi-sectoral approaches to create civil cultures that eschewed violence. As a consequence, politicians who resist pursuing approaches consistent with new expectations face high political costs. Therefore, beyond the operational restructuring and reform of the security apparatus through policy changes, this has been underpinned by a more profound shift away from the normative acceptance of violence.
Although we do not reject potential alternative explanations, we conclude that similar processes of institutionalised policy change strongly contributed to reductions in lethal violence in both cities. Bogotá’s sustainable violence reduction sees homicide rates comparatively low for a major Andean city; and while Recife’s reforms need to prove to be sustainable, initial homicide rate declines are impressive (though remaining above the Brazilian average; and almost twice Bogotá’s). However, reflecting continuously high levels of petty delinquency, inhabitants in both Bogotá (CCB, 2013) and Recife 35 still hold unfavourable perceptions of security in their cities despite successfully reducing lethal violence. Given the challenges in reporting and tracking petty crime, and the seriousness of lethal violence, addressing homicides may be a more pragmatic target for policy innovation.
Our comparative analysis allows for some generalisation when considering scope conditions of the analysed cities. Specifically we identify several factors that may support violence reduction that could generalise across cases, namely: commitment of political leaders interested in prioritising public security; sufficient resources for designing and implementing novel policies; an evidence-based approach to policy evaluation; police forces acquiescent to civilian administrations; and a civil society able to support and internalise changes that reforms entail. Consistent with our findings, the UNDP notes that: improving citizen security does not stem from a single isolated policy or action, but from a multi-sector approach and a series of policies including preventive measures, institutional reforms, sufficient public investment, changes in the relationship between the State and communities, broad and sustained political will, and the adoption of more modern and effective systems of information and intervention. (UNDP, 2013: 12)
Overall, our analysis presents several key points. First, it highlights that the purported causes of violence ‘escalation’ may differ from drivers of violence ‘reduction’; and escalation and reduction are not symmetrical processes. Importantly, institutionalised policy change can contribute to a reduction of violence even where many structural factors favouring violence escalation persist; though unfavourable structural conditions may limit the potential of these changes. 36 As such, institutional accounts should not be viewed as solely explanatory, but should be regarded as integral in understanding sustained violence reduction processes. Second, it emphasises the role of institutionalised policy change as a politically led process of policy innovation and institutional reform; and one that likely benefits from urban-based civil society support. Finally, it presents a simple three-phase model identifying stages of commencement, traction and sustained impact that can be used to examine violence reduction processes in other contexts, calling particular attention to the dynamics of critical junctures that can enable meaningful policy shifts.
Future research could address whether similar mechanisms operate in other cities by identifying and testing cases where institutional or policy change either improved or impaired public security (negative cases may be particularly instructive). Moreover, research should consider how to promote civilian oversight of police forces and the legal apparatus, and how reforms can support public policy. Finally, while institutionalising public security policy appears important, this should not preclude supporting multi-sectoral approaches addressing broader political, economic and social dynamics.
Footnotes
Acknowledgements
Kristian Hoelscher is a Senior Researcher at the Peace Research Institute Oslo (PRIO). Enzo Nussio a Post-Doctoral Researcher at the Institute of Latin American Studies, Stockholm University. Authors contributed equally. We are very grateful to all interviewees for this study, both in Bogotá and in Recife. We also thank Laura Huerfano, Miguel González, Laura Tovar, Giselle Hoover Silveira and Patrícia Correia de Oliveira for excellent research assistance. Enzo Nussio was supported by a Swiss National Science Foundation post-doctoral fellowship and would like to thank his colleagues at the Political Science Department of Universidad de los Andes. Kristian Hoelsher would like to thank José Luiz Ratton and Michael Jerome Wolff for useful discussions and feedback.
Funding
This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.