The Paradox of Violence in Venezuela: Revolution, Crime, and Policing During Chavismo (Pitt Latin American Series) 0822947129, 9780822947127

Citation preview

THE PARADOX OF VIOLENCE IN VENEZUELA

Pitt Latin American Series Catherine M. Conaghan, Editor

THE PARADOX OF VIOLENCE IN VENEZUELA

REVOLUTION, CRIME, AND POLICING DURING CHAVISMO

-

EDITED BY DAVID SMILDE, VERÓNICA ZUBILLAGA, AND REBECCA HANSON

University of Pittsburgh Press

Published by the University of Pittsburgh Press, Pittsburgh, Pa., 15260 Copyright © 2022, University of Pittsburgh Press All rights reserved Manufactured in the United States of America Printed on acid-free paper 10 9 8 7 6 5 4 3 2 1 Cataloging-in-Publication data is available from the Library of Congress ISBN 13: 978-0-8229-4712-7 ISBN 10: 0-8229-4712-9 Cover photo: AFP / Leo Ramírez. The image, taken by photographer and sociologist Leo Ramírez shows a routine police roundup in the neighborhood of Petare. Cover design: Melissa Dias-Mandoly

Contents

Acknowledgments vii Foreword ix

Richard Snyder

Introduction: The Paradox of Violence in Venezuela  3 Rebecca Hanson, David Smilde, and Verónica Zubillaga

Part I. The Shape of Violence 1. The Problem with Venezuelan Homicide Data, and a Solution  29 Josbelk González Mejías and Dorothy Kronick

2. The Diversification of Venezuela’s Violence: A Quantitative Exploration 51

José Luis Fernández-Shaw Guerra

Part II. Causal Processes and Cycles of Violence 3. Violent Crime in Venezuela: Economics and Institutions  73 Josefina Bruni Celli and Javier Rodríguez

4. The Lights of Peonía: Violence and Prison Order in Venezuela  95 Andrés Antillano

5. “Everyone’s Armed Here”: Guns and Lethal Reactivity in a Caracas Barrio 114

Verónica Zubillaga

Part III. From Civilian Police Reform to Resurgent Militarized Policing 6. Criminal Violence and Government Responses under Chavismo 135

Luis Gerardo Gabaldón

7. Dilemmas of Reform: Crime, Policing, and Public Opinion in Venezuela 153

Rebecca Hanson and David Smilde

8. Police Raids in Venezuela: Necropolitics and the State of Exception 170

Keymer Ávila

9. The Pressure to Bring in a Body: How Systematic Killing Transformed Police Raids and Gangs in Post-Chávez Venezuela 187

Leonard Gómez and Rebecca Hanson

Part IV. Responses to Violence: Looking Back, Looking Forward 10. Civil Society Responses to Violence in Venezuela: Anesthesia, Revenge, and Empathy  211

Manuel Llorens

11. Criminal Justice Strategies and Violence in Venezuela  232 Enrique Desmond Arias

12. Final Reflections and Emerging Agendas for Research  250 Rebecca Hanson, David Smilde, and Verónica Zubillaga

Notes 263 References 279 List of Contributors  323 Index 327

vi

Acknowledgments We would like to thank Ludovico Feoli and the Center for InterAmerican Policy Research (CIPR) at Tulane University for supporting this project from beginning to end, including the first conference at Tulane University in October 2015. Sefira Fialkoff of CIPR was a patient administrator. Daniel Ortega and Ana María Sanjuán of the South American Development Bank; Ysrrael Camero, Ángel García, Sary Levy, Pily Méndez, Clemy Machado de Acedo, Pedro Planchart, Eduardo Valero from Asociación Fulbright de Venezuela supported the conference in Caracas in March 2017. The Latin American Program of the Open Society Foundations supported the research leading to several of the chapters in this book. We would like to warmly thank Pedro Abramovay, Mario Arriagada, Aram Barra, Felipe Cala, and Angélica Zamora. The Stone Center for Latin American Studies at Tulane University and the University of Florida Center for Latin American Studies supported the translation of five chapters. Verónica and later David worked on the manuscript while fellows at the Kellogg Center for International Studies at the University of Notre Dame. David worked with the Washington office on Latin America during this entire project and benefitted from innumerable discussions of citizen security with colleagues there, including Adriana Beltrán, Geoff Ramsey, Joy Olson, and John Walsh. Joshua Shanholtzer of the University of Pittsburgh Press has been a steady and supportive editor. Catherine Conaghan first suggested we submit the manuscript to the Pitt Latin America Series. A number of colleagues directly participated in the discussions leading to this book. Roberto Briceño-León, Fernando Carrión, Robert Gay, Benjamin Lessing, Dennis Rodgers, and Alejandro Velasco participated in one or both of the conferences and importantly impacted the direction of our analysis. This book also benefitted from multiple presentations at congresses of the Latin American Studies Association and elsewhere. Others in our network have significantly impacted our ideas: Guy Bajoit, Doris Barreto, Brenda Bellorín, Beatriz Borges, Yoletty Bracho, Colette Capriles, Abby Córdoba, Luis Cedeño, Andrea Chacón, Crisvii

viii

Acknowledgments

tián Correa, José Miguel Cruz, Carlos de la Torre, Luis Duno, Angélica Durán Martínez, Andreas Feldmann, Alberto Gruson, Vicente Lecuna, Magdalena López, Pancha Mayobre, Miguel Mónaco, Henry Moncrieff, Virginia Oliveros, Roberto Patiño, Aníbal Pérez-Liñán, Hugo Pérez Hernáiz, Michael Reed-Hurtado, Ana María Reyes, Magaly Sánchez-R, Francisco Sánchez, Chelina Sepulveda, Elisa Silva, Hugo José Suárez, Eduardo Silva, and Guillermo Trejo. Hugo Pérez expeditiously translated the Spanish-language chapters to English. Lucas Díaz worked through the translations to put them into native English. Francisco Sánchez has been a collaborator and expeditiously compiled the index. David would like to thank his wife María and daughters Yara and Annelies, who have withstood the inevitable grumbling and crises of a long-term writing project. Veronica would like to thank Zoe, her daughter, for being so sweet and patient when her mom was in the middle of this demanding, long, and above all, exciting project. She is also very grateful to her parents, Fran, Mariela, and Yleana; sisters Tilde, Nela, and Tere; and brothers-in-law, Agustín, Juancho, and Daniel, for their unconditional support. Rebecca would like to thank her husband and coauthor for keeping her (relatively) sane during this process and reminding her to eat even when deadlines were looming, and Billie and Robert O’Hara for all the adventures that took her mind off the project over the years. We’d also like to thank each other for a challenging, intense, and unforgettable intellectual experience. We agree it has been one of the most rewarding collaborations of our careers. And of course, we sincerely hope this project advances current understandings and debates on violence in Venezuela and the region, and helps create a safer world, with more solidarity and less violence.

Foreword The Paradox of Violence in Venezuela offers a rich, nuanced, and multifaceted case study of the massive upsurge in violent crime that Venezuela, like many Latin American countries, experienced over the past decade. The best case studies are multivocal: they speak effectively, as this book does, to more than one audience. A successful case study should satisfy readers who wish to learn “something about something”; in this case, those who seek a stronger substantive understanding of the causes, patterns, and human consequences of violent crime in contemporary Venezuela. In addition to nonspecialists like myself, this first audience includes other Venezuelanists who, like the contributors, have large stocks of accumulated knowledge about the country and who will scrutinize the work, looking to see that the authors “got the case right.” A second audience engaged by a successful case study is interested not in the case per se, but in where it fits into a broader set of cases and if it can advance knowledge that is applicable to this larger universe. Reaching this second audience of comparativists and generalists requires a compelling answer to the question, what is the upsurge of violent crime in Venezuela a case of? The kind of case-based research presented in The Paradox of Violence in Venezuela, which illuminates “something about something” in a way that speaks convincingly both to specialists and nonspecialists, depends on deep contextual knowledge, typically hard earned through years of intellectual and personal engagement with a place and its society and culture. Contextual understanding is enhanced by reliable access to networks of informants, the judgment to tell “informed informants” from uninformed ones, and the ability to recognize their biases. Together, these conditions for contextual insight are fostered by what economic sociologists call embeddedness—in rich networks of informants and also in the communities, organizations, and places that are the focus of research. The contributors to this book exemplify embedded scholars, with deep contextual knowledge about Venezuela and the ability to deploy their knowledge to provide a stronger understanding of the causes and consequences of violent crime. Verónica Zubillaga is a distinguished ix

x

Richard Snyder

Venezuelan sociologist with a cosmopolitan training and professional trajectory who is widely known for her long-standing ethnographic research program on gender and violence in the slums of Caracas. Her co-editors are gringos—that is, a gringo, David Smilde, and a gringa, Rebecca Hanson—which raises the thorny question of whether foreigners can acquire credible contextual knowledge. A routine response to this question points to Alexis de Tocqueville’s Democracy in America. Although Tocqueville was a Frenchman with imperfect English who made only a brief visit to the United States in 1831–1832, he managed to produce a study that is praised as “at once the best book ever written on democracy and the best book ever written on America.” Still, Democracy in America has been roundly and quite validly criticized for its shallow empirical foundation: not only was Tocqueville’s study replete with factual errors, it was stunningly blind to the Industrial Revolution, well under way during his visit to America, which was radically transforming the material basis of the country’s life. Fortunately, Smilde and Hanson are no de Tocquevilles. Both have deeply rooted professional and personal connections to Venezuela. Over the past quarter century, Smilde has produced a coherent and important body of research on Venezuelan society, culture, and politics. Hanson, while still at an early stage in her career, has already accumulated seven years of sustained ethnographic research on the Venezuelan police, and, as a result, she understands how this organization works better than any other social scientist of which I am aware. Moreover, the three co-editors are not only embedded but also engaged scholars who actively aim to communicate the implications of their research to wider audiences, including policymakers, by making media appearances and publishing in diverse outlets for nonexperts. The other contributors are a mix of similarly embedded and engaged Venezuelan and foreign scholars. Admirably, while the three coeditors are sociologists, the other contributors hail mostly from different disciplines, including political science, criminology, psychology, economics, and law. Moreover, while the contributors deploy a broad set of research methods—from deep ethnography to quantitative survey analysis—they converge in a shared commitment to producing contextually valid knowledge. The cross-disciplinary coalition not only shines light through many windows on an important human problem, violent crime, in a specific context, Venezuela. It also positions this book to command wide attention among comparativists and generalists because it is precisely this interdisciplinary, multi-methods combination of contextually grounded perspectives that holds the key to unraveling the paradox that frames the volume: why was the striking reduction in poverty and economic inequality achieved by Venezuela’s Chávez associated not with

Foreward

less violence, as predicted by leading theories and explanations, but with more? A crucial first step toward resolving this paradox of violence is recognizing that the positive correlations found in prior research among poverty, inequality, and violent crime rest on economistic moorings anchored in a narrow, income-based understanding of inequality. Getting beyond the limits of this economistic perspective requires stepping outside it, a move this book achieves splendidly with its interdisciplinary team of contributors armed with contextual knowledge. These authors produce a far more nuanced understanding of the complex drivers of violent crime in Venezuela today, including intersectional, bundled, and “thick” inequalities of race, gender, and status, in addition to “thin,” income-based inequalities that, while they may be easier to quantify and measure, can lead to overly simplistic correlations and inferences—for instance, that the level of violent crime is directly related to the level of poverty and income inequality. This book’s framing around the paradox of violence thus produces an engaging answer to the question that comparativists and generalists ask of case studies: what is the case a case of? Venezuela is a puzzling case of increasing violence despite decreasing poverty and income inequality. Deploying Venezuela as a “deviant” case to destabilize and poke holes in existing explanations, while valuable, begs the question, if poverty and economic inequality cannot adequately account for patterns of criminal violence in Venezuela, then what does? Here, The Paradox of Violence in Venezuela breaks new ground by showing that state fragmentation serves as a key driver of criminal violence in Venezuela. This new focus on tensions and cleavages inside the state brings Venezuela into a lively comparative research program on criminal violence and policing in Latin America sparked by an exciting generation of young scholars, including Angelica Duran-Martinez, Benjamin Lessing, Sandra Ley, Eduardo Moncada, and Guillermo Trejo, which has focused mainly on Brazil, Colombia, and Mexico. While the effects of state fragmentation on criminal violence in Venezuela resemble the effects observed in these other countries, the origins of fragmentation are quite distinct. In Venezuela, in contrast to Mexico, for example, fragmentation and contentious factionalism across state agencies specializing in violence stems not from partisan conflicts across levels of government in a federal system, but from fissures in a neo-socialist revolutionary project exacerbated by personalistic leadership. A quick review of comparative social science research over the past forty years shows that Venezuela has been a fertile source of enduring conceptual and theoretical contributions with broad cross-national rele-

xi

xii

Richard Snyder

vance. In the study of political regimes, Venezuela served as a counterpoint example of stable, governable democracy during the 1960s, 1970s, and into the 1980s, a period when most Latin American countries were ruled by authoritarian regimes of varied stripes, from bureaucraticauthoritarian military regimes in the Southern Cone, to personalistic patrimonial dictatorships in Central America and the Caribbean, to a single-party corporatist regime in Mexico. Moreover, studies by Terry Karl, Daniel Levine, and Michael Coppedge highlighted the role of elite “pacts” and “partyarchy” as devices that, while helpful for understanding Venezuela’s exceptional status as a stable democracy, also had broad applicability for understanding transitions from authoritarianism, as well as the possibilities for stabilizing new democratic regimes, across the globe during the third wave of democratization that occurred in the last quarter of the twentieth century. With the rise to power of Hugo Chávez at the end of the 1990s, Venezuela became a harbinger of a set of “new left” and “neopopulist” leaders, from Bolivia to Ecuador to Nicaragua and Argentina. In turn, Venezuela spurred a flurry of research in comparative politics on the “left turn” in Latin America, with the collapse of the country’s two-party system and the associated crisis of representation serving as a modal causal pathway that could lead to the rise of an “outsider,” neopopulist leader. Likewise, Venezuela also played a key role in building and refining theory about the “resource curse”; that is, the potentially harmful consequences of natural resource wealth for state capacity, economic performance, and democracy. Fernando Coronil, Terry Karl, and Thad Dunning proposed novel theories about the pernicious effects of oil dependence on state institutions. With its important contributions to the vanguard of research on violent crime in Latin America, The Paradox of Violence in Venezuela joins the illustrious lineage that draws powerful insights with broad comparative relevance from a close and sustained engagement with Venezuela.

Richard Snyder

THE PARADOX OF VIOLENCE IN VENEZUELA

Introduction

The Paradox of Violence in Venezuela Rebecca Hanson, David Smilde, and Verónica Zubillaga

Inequality generates injustice and injustice generates violence. The Bible says so. —Hugo Chávez on his weekly Hello President television show (Chávez 2003)

The history of violence in Venezuela during the presidencies of Hugo Chávez and Nicolás Maduro challenges a common assumption that runs through much academic and everyday discourse on violence—that poverty and inequality are its underlying causes. Between 2004 and 2011 the Venezuelan government used windfall oil profits to strengthen social policies that economically benefited the poorest social sectors and the results were clear: the number of poor households, unemployment levels, and the rate of child malnourishment all declined.1 Life expectancy increased from 73 to 74.3 (PROVEA 2012, xxxvi), and the Gini coefficient used for measuring inequality dropped from .46 to .39 (viii). Yet during this same period, violence measured by homicide rates soared. Even according to official statistics—which tend to be more conservative—the homicide rate in Venezuela almost tripled. In 1998, the year before Chávez became president, the homicide rate was 20 per 100,000 inhabitants. Fifteen years later, it had risen to 56 per 100,000 inhabitants. Violence was already a concern in the 1990s, of course. Indeed, it 3

4

Rebecca Hanson, David Smilde, and Verónica Zubillaga

Graph I.1. Paradox of violence in Venezuela. The relationship of poverty, inequality, and homicide during the windfall of oil revenues in Venezuela. Source: Poverty, data.worldbank.org; Homicide, Ministerio del Poder Popular para Interior y Justicia reported in http://prodavinci.com/blogs/las-muertes-por-vi olencia-en-venezuela-comparadas-con-el-mundo-por-anabella-abadi-m-nu meralia/; Gini, National Institute for Statistics (Venezuela). Note: For the Gini coefficient, the vertical axis should be read as 0, 0.1, 0.2, 0.3, 0.4, 0.5, and 0.6.

routinely topped polls as citizens’ number one concern and was one reason for the anti–status quo sentiment that brought Chávez to power. But rates skyrocketed during the Chávez period. This is what we are calling the paradox of violence in Venezuela. Over the past several years, we have facilitated a discussion among an interdisciplinary group of scholars examining crime and violence in Venezuela—a discussion which led to this book. In it we seek to understand why violence steadily increased during the governments of Hugo Chávez and Nicolás Maduro. In doing so, we also seek to contribute to long-standing debates in the social sciences. These chapters reveal a need to reorient how we think about violence and its relationship to poverty, inequality, and the state.2 They do not argue there is no relationship between poverty, inequality, and crime, but that particular models of governance, citizen security policies, and persisting structural social deficits affect how this relationship plays out and have their own independent effects. We argue that during Chavismo, violence increased in Venezuela as a result of the following interrelated factors: extraordinary oil revenues and the hypertrophic growth of the state, a particular type of revolutionary governance, failed citizen security reform and the resurgence of militarized policing, and continued concentrated disadvantage. Extraordinary oil revenues contributed to intrastate struggles and state fragmentation, limiting state institutions’ interest in and capacity to regulate so-

The Paradox of Violence in Venezuela

cial life. The Chavista governments’ efforts to revolutionize Venezuelan society, branded “twenty-first-century socialism” after 2006, produced institutions that were even less capable of coordinating sustainable public policy and ordering society, and less able to provide security to its territory and populations. These institutions generated competition between actors and between institutions as well as the atomization of the means and exercise of force; in other words, they spurred a pluralization of violent, armed actors. This occurred in the midst of abundant resources, yet these processes and persisting pockets of exclusion and the growth of illicit markets in the region led to a surge in violence. The data analyzed by Josefina Bruni Celli and Javier Rodriguez in chapter 3 show that there is a long-term trend in Venezuela in which increases in oil revenue are tied not only to state spending but to violence. These findings coincide with those of Marcelo Bergman (2018), who shows that economic growth has led to more violence in a number of countries in the region, but point in a different direction. Bruni Celli and Rodriguez suggest that an influx of revenue can actually undermine state institutions, leading to institutional weakening and incentivizing impunity. In the period under study here, while the Chávez government used extraordinary oil revenues to reduce poverty and inequality, its revolutionary project effectively converted the state into a battlefield. And while there were more resources circulating in the economy—one of the factors Bergman identifies as driving the increase in violence—most of the Chávez government’s antipoverty efforts did not uniformly reduce poverty. Persisting spaces of “concentrated disadvantage” unaddressed by state benefits created cleavages and sources of conflict within lower and working-class communities, contributing to a continued expansion of crime and violence (Antillano 2016). Guiding our analysis is a sociological approach with a couple of key aspects. First, in contrast to common wisdom and even considerable criminological research, in this book we analyze violence not as a lack of norms or values, an inability to control emotions, or the result of biographical trauma, but rather as a practice that rational, contextually embedded actors use to assert or maintain control in social relations (Auyero and Berti 2015). Contexts in which there is competition over resources, social and institutional structures that do not distribute resources in a stable way, and institutions that do not provide for the resolution of conflict will see violence emerge as individuals and groups seek to capture resources and assert dominance over each other (Bergman 2018; Arias 2017; Lessing 2018). Interaction, competition, and violence are shot through with emotion, but even expressive violence is generally part of an effort to develop reputation—itself a resource or form of capi-

5

6

Rebecca Hanson, David Smilde, and Verónica Zubillaga

tal—for individuals and groups (Bergman 2018; Smilde 2007; Zubillaga 2007). Second, we see the state as a processual set of relationships; that is to say, not as a unitary or coherent entity, but as a heterogeneous field riddled by contradictions and struggles, sometimes more than others (Migdal 2001; Mann 2012). While terms such as state capacity or state fragility have their usefulness, we think it is necessary to move beyond their implicit assumption of a unitary interest in providing security. We see the state not just as a set of institutions but as a space in which conflicts and tensions between political actors with diverse interests play out; violence can result from these tensions but also be used as a resource to gain the upper hand in conflict, or permitted in processes of forbearance in which political actors do not enforce laws to maintain support or alliances (Holland 2017; Willis 2015). In other words, violence is not only a problem of the state’s inability to act but also the result of state actors’ actions and decisions. This is essential to explaining how and why the Venezuelan state not only failed to provide security to its citizens but also became a systematic violator of this most basic of rights, as will be seen in the chapters by Keymer Ávila (chapter 8) and Leonard Gómez and Rebecca Hanson (chapter 9). To be clear, we do not think that state capacity is irrelevant, but that concentrating on it alone misses the problem. Indeed, informed by the work of Javier Auyero, Jose Miguel Cruz, Angélica Durán-Martínez, Desmond Arias, and Benjamin Lessing we understand state policies and practices as central to explaining when and why violence is deployed by state and criminal actors.

Leading Explanations of Violence in Venezuela Explaining homicide rates has been central to Venezuela’s political conflict from the time Hugo Chávez campaigned for the presidency in the late 1990s. He campaigned and governed with a traditional leftist perspective that suggested reducing poverty and inequality would reduce crime. Less than a month after taking office in 1999 Chávez famously declared that if his daughter were dying of hunger, he would commit a crime to feed her. The assumption of a direct causal relationship, with increases in poverty and inequality leading to increases in crime and violence, is not just characteristic of Chavismo or traditional leftism. Social scientists in the United States have long posited such a connection (see, for example, Blau and Blau 1982; Ehrlich 1973; Hsieh and Pugh 1993; Messner 1982; Sachsida et al. 2009). Some scholars of Latin America have argued that poverty, unemployment, volatile and mediocre growth in the economy—often thought to be driven by neoliberal policies—have incentivized violence as a survival mechanism in the face of precarity,

The Paradox of Violence in Venezuela

insecurity, and exclusion (O’Neill and Kedron 2011; Zubillaga 2007; Humphrey 2013, Fajnzylber et al. 2002; Chinchilla 2003; PNUD 2009; Moser 2004; Kilanski and Auyero 2015). It is also a commonsense assumption accepted by broad sectors of many, perhaps most, societies. In fact, this assumption was one of the few points that Chávez had in common with his opponents. In Venezuela’s polarized political context, critical commentators frequently accepted the terms of this causal portrait, but inverted the equation, using the increase in violence as evidence of an increase in poverty. But as can be seen in graph I.1, violence did not increase during the Chávez years because of an increase in poverty, but despite a decrease. Without a doubt, the rise in violence has rightfully had a central place in opposition critique of the Chávez and then Maduro governments. But how to explain this increase and how to understand its relationship to Chavista governments is not straightforward. Opposition commentators and scholars have also frequently argued that Chávez’s combative form of governance and polarizing rhetoric produced social psychological effects leading to violence. Comparing homicide rates with elections and political events, Steven Tremaria (2016, 70, 74) claims that in Venezuela, polarization produced “tension and hostility among the population” with “interpersonal violence peaking at times of heightened political tension” (see also Humphrey and Valverde 2013). While it is true that increasing political and class conflict has led to a series of violent confrontations between state institutions and civil society, the vast majority of violence in Venezuela has not been an expression of Venezuela’s national-level political conflict. Rather, it is primarily intraclass violence occurring among men living in zones of exclusion, engaged in struggles for dominance over social spaces or illicit economies. Many of them are targeted by and in conflict with state security forces, or caught up as indirect victims in these conflicts. This violence is not linked to institutional politics, political parties, or political action in the sense of seeking state power. Another common explanation is that leftist and or socialist governments are simply ill equipped to address crime and violence since they misunderstand the problem. Indeed, the Left has historically had an uneasy relationship with the institutions of “law and order,” often intentionally marginalizing these institutions in its discussions of crime and insecurity. Jock Young (1986) has referred to this as “the characteristic syndrome of left idealism . . . Crime itself is played down, marginalized, and is not the focus of attention.” While it is true that during the first years of the Chávez government little attention was paid to the state security apparatus, after 2008 the government spent considerable resources implementing what began as a textbook citizen security reform

7

8

Rebecca Hanson, David Smilde, and Verónica Zubillaga

(although other aspects of the justice system remained largely ignored). Furthermore, after the launch of militarized policies, massive incarceration took place and the prison’s population increased by almost 67 percent between 2009 and 2011, representing “the highest confined population in history,” according to the minister of justice (see Antillano et al. 2016).3 A quick scan around the region shows there is no necessary link between left-wing governing projects and increasing crime and violence. Central America is one of the most violent regions in the world. Yet most of the violence is concentrated in the northern triangle of Honduras, El Salvador, and Guatemala, countries not led for leftist governments. In contrast, the increase in crime and violence in leftist Nicaragua has been comparatively small.4 Homicide rates in countries like Ecuador under President Rafael Correa and Bolivia under President Evo Morales increased only slightly, but remained some of the lowest in the region. Finally, another term frequently used to explain violence is one borrowed from twentieth-century sociology: anomie. The Venezuelan sociologist Roberto Briceño-Leon (2017) suggests that the looting in the 1989 Caracazo undermined respect for private property, and Hugo Chávez’s 1992 coup d’etat undercut norms against taking power by force. The result was anomie, a break in the social pact and institutions of social control. We agree that a fracturing of the rule of law and fragmentation of institutions of social control is a central causal factor in Venezuela’s surge in violence—what Natalie Gan (2020) has called an “anomic state” that no longer follows its own official rules. But we disagree with uses of anomie to denote normlessness and a vacuum of social relations that presumably leads to a chaotic war of all against all. As the chapters in this book show, the mechanisms driving violence are not social psychological, nor the result of a lack of shared values and norms. Rather, institutions that first were challenged under the neoliberalism of the 1990s eroded further under the revolutionary governance of Chavismo, with sources and forms of social control multiplying and competing. As we discuss further, by 2002 state actors believed it was in their best interest to permit the existence of or provide outright support for non-state armed actors. The number of armed actors has increased dramatically since then, such that state institutions no longer have the capacity to regulate violence. Nevertheless, the violence that occurred was not anomic, but rather formed part of projects seeking to establish social control within complex microsocial configurations. In Andrés Antillano’s chapter in this volume (chapter 4) we see that violent contexts too are structured by norms and values that regulate it as a form of interaction (see also our discussion of Waverly Duck’s work below). As in Javier Auyero et al.’s (2014) description of Buenos Aires, the Venezuelan state is neither

The Paradox of Violence in Venezuela

completely absent, nor does its iron fist lead to absolute control; instead, the intermittent, selective, and contradictory presence of the state contributes to widespread depacification. Furthermore, the reach of the state has actually expanded under Chavismo, and this includes security institutions and alliances with parastate armed actors. There are more police, soldiers, and parastate armed actors than ever before, and they kill and incarcerate more people than ever before. This contributes to disruption and disorder in communities and facilitates the growth of illicit markets (see chapter 9). Indeed, the increased rates of incarceration combined with inattention to the penal system generated spaces where criminal networks developed, exacerbating violence on the outside. Later, under the government of Nicolás Maduro, lethal tactical units invaded sectors for varying lengths of time, usually leaving behind a power vacuum that would quickly be filled by more organized non-state armed actors. Thus, the problem is not so much an absence of state as it is a violent presence. The concept of anomie leads analysis away from these processes, obfuscating more than it reveals. Reducing violence is not about filling a normless void, but challenging and altering existing norms and social structures inside and outside of the state. So is Chavismo to blame for the surge in violence? After twenty years in power, how could it be otherwise? Nevertheless, the leading explanations forwarded in a politically polarized context are wide of the mark. The revolutionary project championed by Hugo Chávez is responsible for the surge in violence, but not in the way most people think. And, while Venezuela is often discussed as an exceptional case, violence in the country is also driven by broader regional trends, three of which are relevant to our explanation. Embedding our four causal factors in these regional trends can help us better understand violence in Venezuela.

Venezuela and Violence in the Latin American Context Since the early 2000s, Latin America has been considered the most violent region in the world. According to a 2013 report by the United Nation’s Development Program (PNUD 2013), it is the only region where lethal violence increased between 2000 and 2010.5 While homicide rates in most regions have fallen by as much as 50 percent, in Latin America they increased by 12 percent. Latin America has, of course, witnessed mass violence in the past due to colonialism, slavery, civil wars, dictatorships, and revolutions, but there have been unprecedented changes in the quantity and quality of violence in the twenty-first century. Many countries in the region suffer from rates of armed violence as high or higher

9

10

Rebecca Hanson, David Smilde, and Verónica Zubillaga

than those in countries affected by war (Geneva Declaration Secretariat 2011, 51–65), and that violence has taken on new forms and expressions (see, for example, Arias and Goldstein 2010; Civico 2012; Koonings and Krujit 2007). While political, state-driven violence was dominant in the twentieth century, contemporary violence has taken on a lateral form—that is, civilian-on-civilian violence, carried out by non-state actors across relatively horizontal networks. Guns, illicit economies, and the pluralization of violent actors are fundamental to understanding how violence has evolved across the region (see Koonings and Krujit 2004; Arias 2006; Cruz 2010; Arias and Barnes 2017). These three trends run through the four factors we develop in the following sections to explain the paradox of violence in Venezuela. This book locates these regional trends in a particular time and place, specifying how they matter within the Venezuelan context. The proliferation of guns in Latin America is essential to explaining increasing violence throughout Latin America (Briceño-León and Zubillaga 2002).6 According to the Small Arms Survey (2012), Latin America has the greatest proportion of firearm-related fatalities in the world. But even among countries with high homicide levels, Venezuela stands out—among the population between ten and nineteen years of age, Venezuela has the highest percentage of homicides using guns, 94 percent (Otamendi 2019, 6).7 Firearms are an important part of our explanation precisely because the circulation of legal and illegal guns is intimately connected (Cano 2001). Legal firearms are one of the primary channels through which guns are obtained for the commission of crimes in the region. Firearms and other weapons are often stolen from, sold, or “rented” by state security forces, another way in which loose connections between state and non-state groups contribute to violence. In fact, Verónica Zubillaga et al.’s research with armed youths has demonstrated that police officers are their principal source of weapons (Zubillaga, Llorens and Souto 2015). At the microsocial level, the uncontrolled proliferation of guns has meant that everyday interpersonal conflicts become lethal conflicts as they are no longer solved by insults or fisticuffs but by bullets (see chapter 5). In many Latin American countries armed actors have pluralized, challenging the state as the main coercive actor. The proliferation of civilian militias, criminal organizations, and paramilitary groups (Rodgers 2006; Arjona 2016; Civico 2012) have led Enrique Desmond Arias and Daniel M. Goldstein (2010) to conceptualize much of Latin American as “violently plural.” For example, in Colombia, military and paramilitary groups have killed with impunity in both urban and rural areas. These same paramilitary groups, aided by government initiatives

The Paradox of Violence in Venezuela

and even funding, eventually transformed into major players in the drug trade. In Venezuela, gangs, armed community groups often referred to as colectivos, and even security forces have multiplied over the past twenty years. In some cases, state policies intended to fight crime have contributed to this proliferation. As Andrés Antillano shows in chapter 4, militarized interventions and the massive imprisonment of young men with knowledge and experience in illicit economies favored the advancement of illicit networks and the development of more sophisticated criminal groups. In other cases, state actors have actively supported non-state armed groups. This was particularly true after the short-lived coup d’etat in April 2002, in which many officers from the Metropolitan Police of Caracas participated. This marked the beginning of their marginalization and probably their increased involvement in crime, and contributed to the government’s support for armed colectivos. An increase in illicit markets is also an important part of the regional surge in violence (Yashar 2018). Dynamic transnational actors look for spaces of weak or complicit state control and develop routes accordingly. Changes in drug trafficking routes after crackdowns are an important part of the story of surges in violence across the region. As Plan Colombia successfully diminished drug trafficking from the Caribbean coast of Colombia, it forced some routes to move through neighboring states, including Ecuador, Brazil, and Venezuela. The map of violence in Venezuela does indeed show that armed violence has increased along drug transportation routes as well as border regions in which drugs are one among several illicit markets. Nevertheless, research shows that illicit markets in themselves do not cause violence—contested illicit markets do (see Thoumi 2012; Andreas and Wallman 2009). Thus, while increased drug trafficking is part of the equation, it is not a sufficient cause of violence. For example, Juan Camilo Castillo and Dorothy Kronick (2020) argue that state seizure of illegal goods, by disrupting supply, can fuel violence. Using case studies in Mexico and Burma, Richard Snyder and Angélica Durán-Martínez (2009) propose that the exercise of violence might be a function of the capacity of actors—namely, illicit traffickers and public officials—to establish pacts in which public officials establish protection and traffickers demonstrate that they are trustworthy (also see Lessing 2018 on Mexico). When these pacts rupture, the disruption catalyzes violence (Snyder and Durán-Martínez 2009). Similarly, in his reflection on drug markets and the selective use of violence, Richard Friman (2009, 287) affirms that “once markets are consolidated in the hands of organized networks . . . the levels of large-scale violence tend to decrease” (see also Rodgers 2015 for Nicaragua). Disruption gets us closer to understanding the rise

11

12

Rebecca Hanson, David Smilde, and Verónica Zubillaga

of violence in Venezuela, pushing us beyond drug trafficking in and of itself. At a more micro level, research by Antillano and Zubillaga (2014) has also demonstrated that various social, institutional, and cultural factors must be taken into account in order to explain why microtrafficking produces violence in some barrios and not others. Changes in delivery practices (delivering drugs to the buyer’s home, for example) and informal social control mechanisms can modify the relationship between violence and drug trafficking. Guns, pluralized violence, and illicit markets are an essential part of our story, then, but their impact in Venezuela cannot be understood in isolation from the particular factors at play inside the country: the hypertrophic growth of the state, revolutionary governance, failed citizen security reform and the resurgence of militarized policing, and continued concentrated disadvantage. The proliferation of guns and armed groups in the country, for example, is intimately connected to the context of revolutionary governance, which facilitated access to weapons by non-state actors. Failed police reform and the persistence of militarized policing—which gave the police access to military-grade weapons—help us to understand the circulation of weapons and the pluralization of violent actors. Illicit activities are not inherent or exclusive to marginalized areas, but in neighborhoods where disadvantage is concentrated, illicit markets emerge as residents seek alternative survival strategies; thus, while poverty and inequality declined during much of the Chávez period, the persistence of concentrated disadvantage provided space in which illicit markets could flourish. While conflicts occur among illicit organizations without the presence of state security forces, the militarization of security and incredibly violent police raids have caused consistent disruption in illicit markets; this has produced a defensive response by criminal groups and, as a result, their strengthening. Thus, they are essential to explaining conflicts in these spaces.

Rethinking Violence in Venezuela The conceptual framework presented here has emerged out of a crossnational, cross-cultural, interdisciplinary discussion and debate over the past seven years among historians, political scientists, sociologists, and psychologists from across the region. Contributors brought their disciplines’ approaches and questions to bear on the paradox presented in this introduction; over time each contribution was strengthened through interdisciplinary engagement. Taken together, the chapters in this volume suggest that violence has increased due to the interrelated factors mentioned earlier: extraordinary oil revenues and the hypertrophic growth

The Paradox of Violence in Venezuela

of the state, a particular type of revolutionary governance, failed citizen security reform and resurgent militarized policing, and continued concentrated disadvantage. These four factors are obviously not discrete but overlap, mutually reinforcing each other. Hypertrophic Growth of the State and Economy Marcelo Bergman has argued that one of the reasons for soaring crime rates throughout Latin America is economic growth leading to rising consumerism. He suggests that “prosperity usually provides greater returns on criminal activity because there are more opportunities and targets” (Bergman 2018, 140; see also Rosenfeld and Levin 2016). Also, rising consumption also creates secondary markets for illicit consumer goods that fuel property crime and the violence associated with it. He suggests that the rapid increase in opportunities for crime and violence can overwhelm police forces and the justice system, further reducing the “costs” of crime, leading a country’s crime rate to spiral out of control. Bergman’s explanation of Latin America’s surge in violence would certainly seem to apply to Venezuela, given the extraordinary oil windfall Venezuela received during the presidency of Hugo Chávez. Of course, the case of Venezuela may be a little different from others in the region since the economy is dominated by the state and its spending and distribution of oil rents. Nevertheless, given that Chavismo’s economic policies led to soaring levels of consumption (PROVEA 2012), the processes that Bergman describes may have been at work. However, there is one pesky fact that is often overlooked with respect to Venezuela: in the 1990s and through 2010 there is no clear evidence of an increase in criminality in Venezuela apart from homicide. Crime figures are notoriously unreliable, of course, but both in the 1990s (Zubillaga 2003) and 2000s (PROVEA 2010; Sanjuan 2011), under different governments, measured by both official statistics and victimization surveys, violence surged at the same time that rates of burglary, robbery, and auto theft were largely stagnant (although high). Thus, while the “more money, more crime” thesis points us in the right direction, we cannot assume that crime is a direct function of consumption, nor that violence is a direct function of other forms of crime. What seems to have happened in Venezuela is that while crime itself did not increase, it became increasingly violent (Zubillaga 2003; Sanjuan 2011). In chapter 3, Josefina Bruni Celli and Javier Rodriguez show a strong correlation between oil income and violence between 1970 and 1988 and then again from 1999 to 2016. The relationship was presumably different from 1989 to 1998, during the neoliberal period when violence rose without oil revenue rising. But examining more closely, they find no re-

13

14

Rebecca Hanson, David Smilde, and Verónica Zubillaga

lationship between economic growth and violence. Instead they focus on the way that an influx of resources can lead to hypertrophic expansion; rather than solidifying the state, this influx reduces its institutional capacity, which in turn increases impunity. An influx of extraordinary resources can undermine incentives for political leaders to build effective state institutions, reducing the state’s ability to exercise social control. Though this phenomenon precedes Chávez, from 2004 to 2013 his government received the largest influx of oil wealth in Venezuelan history, which led to hypertrophic growth of the state, reducing capacity and willingness to exercise social control at multiple levels. It is important to remember that oil in and of itself is not a curse (Dunning 2008), and that the trade-off between the state’s ability to provide social inclusion and order have been pointed out more broadly (Centeno, Kohli and Yashar 2017). But petro-economies do have a number of consistent structural problems that challenge economic policy makers. In the case of Venezuela, “the combination of an incoherent socialist program made classic petro-state problems all the more severe, drawing from the worst features of each” (Velasco 2016). This in part explains the surprising inverse relationship between inequality and violence in Venezuela. When oil revenue goes up, inequality goes down, as the government spends more money on its people. Yet hypertrophic growth of the state reduces the strength of the same institutions that regulate society. What is more, the same oil revenue that creates hypertrophic state growth, produces struggles within state sectors as actors fight to secure control over resources. Revolutionary Governance We know from elsewhere that political conflict and change can exacerbate violence. Jose Miguel Cruz (2016, 388) writes that the democratic transition in El Salvador produced disruption and violence, in part because it “increased state fragmentation, pitching state operators at different levels and institutions against each other. This struggle has translated into the constant renegotiation of power and authority between state representatives and other social actors outside the electoral field” (see also Yashar 2018). Angélica Durán-Martínez (2015) points out that violence connected to drug trafficking in Colombia and Mexico becomes more visible and frequent when trafficking organizations compete, and the state security apparatus is fragmented. Guillermo Trejo and Sandra Ley Gutierrez (2020) show how electoral contestation can lead to violence when the state is penetrated by criminal rackets. Venezuela has its own version of political conflict facilitating violence. Hugo Chávez’s revolutionary project generated intense fragmenta-

The Paradox of Violence in Venezuela

tion and power struggles. On the one hand, previously dominant social, economic, and political groups being displaced by Hugo Chávez’s efforts to revolutionize Venezuela fought intensely to maintain their positions. Of course, a number of governmental attempts to transform socioeconomic relationships in the region have generated political polarization and potent opposition mobilization in recent decades (see Roberts 1998). But in Venezuela, the process was even more conflictive. Chavismo’s desire to overturn Venezuelan politics was expressed by their referring to the 1958–1999 period as the Fourth Republic and the Chavista period as the Fifth Republic. Some Chavista leaders even spoke of themselves as working for a “new hegemony.” In response, many of those who belonged to the previously dominant political class saw themselves defending pluralist democracy against the Chavista political project. Opposition responses included street mobilization, a coup d’etat attempt in 2002, a general strike in 2002–2003, and multiple waves of mobilization over the next decade. This is not the place to provide a comprehensive review of these political battles, but we can suggest that each side’s interpretation of the situation as a fight to the death hamstrung efforts to address violence. What is important for our purposes is that these political clashes diminshed not only the state’s capacity but state actors’ interest in intervening in key areas of citizen security—such as gun control and judicial and penal reform. Public officials across the board were simply not focused on the challenges presented by changes in criminality. Furthermore, the Chávez government’s citizen security failures were a primary campaign issue and some opposition governors and mayors refused to work with Chavista state institutions that were attempting to implement security reform. On the other hand, the parts of the state apparatus controlled by Chavismo were continually mobilized and fighting for the survival of their project. Distrustful of the police after 2002, and fearful of foreign military intervention after the US invasion of Iraq and tensions with Colombia, Chávez began to promote what he called a “peaceful but armed revolution.” He made explicit that “the revolution is armed to defend its achievements, to defend its advancement, to defend itself against threats and conspiracies” (see chapter 5). This resulted in an “outsourcing” of some security functions to non-state armed actors. Although some are supported by the communities from which they emerged and are more trusted than the police (see Velasco 2015), the proliferation of armed parastate groups with political goals and economic interests implies the establishment of alternative regimes of domination in many geographic sectors (Arias and Barnes 2017). This pluralization of violent actors destabilized police-state relations, fragmented the state security apparatus,

15

16

Rebecca Hanson, David Smilde, and Verónica Zubillaga

and led to more frequent and more visible violence (Durán-Martínez 2015; Davis 2017). More broadly, the coalitions Chávez brought together were incredibly diverse and, in some cases, had opposing, even contradictory, goals as well as visions for how to achieve them. In a governance system marked by personalism—Chávez was the lynchpin holding together a patchwork of civilian, military, intellectual, and activist groups from early on (Lopez Maya 2011)—and the erosion of institutionalization—decisions regarding resources generally depended on a small group—internal struggles for access to spaces of power became fierce. These struggles likewise impeded an effective response to violence. For example, during the Chávez and Maduro administrations there has been a consistent rotation of leaders in and out of the Ministry of Justice—seventeen in the span of twenty-three years.8 As a rule, new ministers suspend, obstruct, or reverse the decisions made by their predecessors (see chapter 6). Clashes between civilian and military sectors within Chavismo also weakened the government’s ability to follow through with a number of citizen security reforms. Those aligned with the military staunchly rejected the process of civilian police reform starting in 2006, actually stopping the legislative project for a year in 2007. They also fought and obstructed the three-year process during which the Presidential Disarmament Commission—led by human rights activists, and the National Assembly’s Mixed Commission—heavily influenced by former military officers, fought over restrictions on the selling and carrying of guns and control over ammunition production. Some government officials, many of whom were either active or retired military officers with financial interests in arms imports or ammunition production, opposed plans to prohibit retail sale of guns, provide personal defense licenses, and mark ammunition. By the time the bill became a law, it had been watered down to assuage the armed forces. As Durán-Martínez (2015, 6) notes, “Enforcement efficacy depends on the ability to coordinate enforcement actions.” Throughout this period struggles between Chavismo and the opposition and within Chavismo made coordination next to impossible, impeding effective action by the state to intervene in matters of citizen security. This polarization and conflict moved far beyond struggle for control of the central institutions of government and affected multiple sectors of society. When social hierarchies are challenged, individuals and groups frequently attempt to (re)assert social dominance by deploying violence (Kruger and Fitzgerald 2012; Karakurt and Cumbie 2012). A clear example of this can be seen with conflict between labor unions. Since 2001 the Chávez government tried to open up organized labor by, on the

The Paradox of Violence in Venezuela

one hand, obliging unions to hold elections supervised by the National Electoral Council, and on the other hand, sponsoring parallel Chavista unions.9 This disruption resulted in a continuing, shifting collage of new and old unions vying for dominance in any given work or organizational site. As the economy boomed during the Chávez years in both the public and private sector, union-related armed violence surged as unions fought to dominate work sites, especially in the construction sector. While sindicariato, a term combining sindicato (union) and sicariato (assassination), is not a main cause of Venezuela’s murder rate, it illustrates larger dynamics of disruption and competition in violence (Lucena 2011). In sum, Venezuela’s tragic levels of violence were not the direct result of its long-term political conflict, and the great majority of Venezuela’s violence is not motivated by the quest for state power (with an important exception described at the end of the next section). However, it is also important to understand the indirect effects of this conflict. The Chávez and then Maduro government’s efforts to revolutionize the state apparatus and construct a new hegemony destabilized institutions of governance from the central government to states, to municipalities to neighborhoods. The reaction against these changes and competition within government institutions in some cases led to violence, but more often reduced the state’s interest in and ability to exercise monopoly over violence as well as social control more broadly.

Failed Citizen Security Reform and Resurgent Militarized Policing Guided by the idea that the best way to reduce crime and violence is to increase investment in social and economic policies addressing poverty and inequality, the Chávez government, during its first seven years, paid little attention to the effectiveness of police forces. Worse yet, police forces became a site for political polarization and conflict. In the period of heightened confrontation starting in 2001, different police forces were commanded by political actors struggling against one another (Antillano 2006). The police’s continued deterioration and involvement in crime came to the fore in 2006 with a series of high-profile cases of violence in which police were involved (see chapters 7 and 8). In the wake of public outrage, the Chávez government created an initiative to study Venezuela’s police forces and recommend reforms: the National Commission on  Police Reform (CONAREPOL). It succeeded in proposing a new police law that was eventually passed in 2008 and put into effect in 2009, leading to an extensive restructuring of Venezuela’s police forces, police training, and oversight mechanisms. The 2009 reform regulated the types of weapons that officers could carry, banning officers from carrying

17

18

Rebecca Hanson, David Smilde, and Verónica Zubillaga

military-grade arms and limiting them to nine-millimeter pistols. New offices were created where citizens could report officer misconduct, and human rights became part of the core curriculum for police training and retraining. Police officers were trained in the progressive and differential use of force. The reform, however, became another example of how competing interests within Chavismo hamstrung the state’s ability to effectively intervene in citizen security. For the most part, it was human rights activists, under the skeptical eye of some leftist factions and the military components of the Chávez government, who carried out the reforms with backing from some state actors.10 From the beginning, certain factions within the government lambasted reform and viewed policing as a right-wing impediment to the construction of socialism. In 2007 the CONAREPOL’s recommendations were shelved by new minister of interior and justice and retired military officer Pedro Carreño, who referred to the reform as “right wing” and “bourgeois.” The reform was revived a year later when a new minister was appointed. These divergent perspectives resulted in sporadic and erratic implementation. And the Chávez government was never exclusively committed to civilian policing. Within months of hitting the streets at the end of 2009, the Bolivarian National Police coexisted with the Bicentennial Security Program (DIBISE), which had heavily armed National Guard officers carrying out operations in which they roared into the barrios on motorcycles in the middle of the night, dragged suspects out of their houses without warrants, and then declared success to the media. Similar operativos (raids), such as Madrugonazo al hampa (Dawn Raid on Crime) in 2011 and Plan Patria Segura (Secure Nation Plan) in 2013, conducted by other security forces continued, delegitimizing the nonrepressive approach to citizen security advocated by reformers. Over time, the need to gain and maintain the support of Chavista factions resistant to the plan led police reformers to emphasize the “humanist” elements of their reform over the enforcement elements. They pushed forward not only with human rights training but also placed more emphasis on community activities and youth programs that emphasized sports and music. Nevertheless, many Chavista leaders remained dubious. The military resented a loss of control over police forces as well as the ample opportunities for corruption and kickbacks provided by equipping them. Police officers also resented the change, believing that being handed over to civilians discredited and delegitimized them (Hanson 2017). The background, of course, is the long-term control of policing by the military in Venezuela, a phenomenon typical of the region (Bailey and Dammert 2006; Ungar 2011; Withers, Santos, and Isacson 2010).

The Paradox of Violence in Venezuela

Political polarization, as chapter 7 shows, also placed police reformers in a difficult position. When people were asked about the Chávez government’s police reform in general, government supporters praised it while government opponents criticized it. However, when people were asked about the actual content of the reforms—for example, the progressive and differential use of force—government opponents were more likely to respond positively than government supporters. Though police reformers framed their efforts as part of the transition to twentyfirst-century socialism, many Chavistas did not support them in practice. Worse yet, those who identified as opposition supporters and actually supported the content of the reforms rejected the police reform project because it was associated with a political project they opposed. Despite these challenges, reformers achieved a lot. The Chávez administration created the Bolivarian National Police, founded the National Experimental Security University (Universidad Experimental de la Seguridad, UNES), created the General Police Council (Consejo General de Policia) and the Presidential Commission on Gun Control (Comisión Presidencial para el Control de Armas, Municiones y el Desarme). By 2012 it seemed as if civilian reformers were gaining the upper hand. That year Chávez consolidated the various citizen security reform initiatives into the Grand Mission Full Life Venezuela (Gran Misión a Toda Vida Venezuela), developed and implemented mostly by progovernment civilians and human rights activists. However, the tables quickly turned after Hugo Chávez’s reelection in 2012. In October of that year, he asked the minister of interior and justice, Tarek El Aissami, who had been a main proponent of civilian police reform within the government, to resign from his post and run for governor of Aragua State. With El Aissami out of the game and Chávez clearly ill, civilian police reform lost its momentum. One important casualty was the effort at control of guns and ammunition. Initially an ambitious effort, it was progressively picked apart by stakeholders. Once Maduro was president a largely ineffectual, watered-down law was put into effect. It is during the Maduro presidency that we see a more direct link between Venezuela’s political conflict and violence (see Cruz 2016; Antillano and Ávila 2017; Zubillaga and Hanson 2018). When Chávez passed away in March 2013, Nicolás Maduro won the April snap election by a percentage so small it shocked everyone—roughly 2 percentage points, despite Chávez having left him with a twenty-point advantage at the time of his death. Starting his presidency in a weak position, Maduro identified the Armed Forces as a sort of security blanket. If we trace the heads of ministries in the country, the growing influence of the military across state institutions becomes evident. In 1999, at the beginning of

19

20

Rebecca Hanson, David Smilde, and Verónica Zubillaga

Hugh Chávez’s first administration, 10 percent of heads of government ministries were military officers. By 2009 this number had increased to 22 percent, and between 2014 (Maduro’s second year in officer) and 2017 it rose from 29 percent to 50 percent (Zubillaga and Hanson 2018). One of Maduro’s first decisions as president was the creation of the Secure Nation Plan (Plan Patria Segura), which redeployed the Armed Forces to perform citizen security functions. This plan had the usual nefarious results of increasing civilian death at the hands of armed security forces while making no headway against crime. In January 2014, after the public uproar over the horrendous murder of former Miss Venezuela Monica Spear and her husband while on a visit to Venezuela, Maduro put the symbolic nail in the coffin of citizen police reform, removing human rights activist Soraya El Achkar from her position as rector of the UNES and replacing her with a military officer. In chapter 8 Keymer Ávila reminds us that police militarization is not a new phenomenon in Venezuela. The police in Venezuela remained under military control even after democratization, and, like military governments throughout the region, Venezuela continued to conceptualize policing in terms of the national security doctrine (see Hernández 1986; Skurski and Coronil 2005). While militarized policing never disappeared under Chávez, it reached new heights and took new directions under the Maduro government. In 2015 the Maduro government initiated the most nefarious citizen security program of all. The Operation Liberation of the People (Operación de Liberación del Pueblo, OLP) brought together various civilian and military security forces to carry out raids, shooting up neighborhoods in broad daylight in a media-savvy display of force with hooded police officers. Human rights group PROVEA reported that over six hundred people were killed in OLP raids in 2016 alone. Though the police in Venezuela have long had a license to kill, in chapter 9 Leonard Gómez and Rebecca Hanson suggest that new initiatives like the OLP have in effect an order to kill.11 As they discuss in their chapter, this change in policy altered criminal organizations, in some cases motivating increased cooperation between them and in others generating conflict as gangs moved into new sectors or migrated out of the city into rural areas. In sum, the early years of the Chávez government saw police forces that were largely abandoned and in many cases shunned in favor of non-state armed actors as well as the proliferation of firearms. Efforts at police reform after 2006 never fully convinced Hugo Chávez and always competed with militarized policing initiatives.12 But when Nicolás Maduro assumed power as a weak president in 2013 he progressively gave more space in the state apparatus to the military, including control over the police forces. This resurgence of militarized policing expanded the

The Paradox of Violence in Venezuela

role of the state as a key violent actor (Zubillaga and Hanson 2018). In addition to being a significant cause of violent deaths, these militarized operations have had the effect of disrupting settled criminal networks and thereby creating violent battles for dominance, incentivizing criminal networks to band together and become further organized to confront violent police raids. Failure to Reduce Concentrated Disadvantage The fact that violence in Venezuela correlates with economic growth does not mean that poverty and inequality are irrelevant. The mere fact that most lethal violence takes place in popular 13 barrios and has poor, young men as its victims is indication enough that poverty and exclusion are part of the causal portrait. Understanding the idea of concentrated disadvantage can help us understand how, and point us to some of the shortcomings of the Chavista project. Debates on crime in the United States have tended to focus on the dramatic discrepancies between Black and white populations, which have long led to racist and other essentialist interpretations of a “culture of poverty.” In recent decades, however, sociologists have developed a perspective that looks at “concentrated disadvantage,” which refers to a convergence of various structural conditions that result from residential segregation and social and economic isolation.14 Concentrated disadvantage includes poverty but also unemployment and underemployment, female-headed families, lack of professional workers, and absence of college graduates; these better predict patterns of violence (see Sampson and Wilson 1995, 2018; Krivo and Peterson 1996). By undermining formal and informal crime control mechanisms at the level of the household and the community, as well as preventing opportunities for economic and social inclusion, concentrated disadvantage creates opportunities for crime and violence. Highlighting the role of political institutions in the production of disadvantage, Robert Vargas (2016) shows how conflicts over political power within local governments undermine violence reduction efforts by obstructing local level organizing and fracturing collective efficacy. For our purposes here it is important to see how these factors work. Concentrated disadvantage should not be understood as a focalized mechanisms of frustration-aggression or relative deprivation. It is not that people are poor and therefore they act out. Rather, over generations public policies and other processes produce social and economic isolation, erecting barriers to the consolidation and reproduction of formal and informal mechanisms of social control. Where families and communities are persistently barred from accessing social and economic re-

21

22

Rebecca Hanson, David Smilde, and Verónica Zubillaga

sources, it may be more difficult for collective efficacy—or the trust and cohesion among neighbors that facilitates informal social control—to develop (Sampson and Wilson 1995; Sampson, Raudenbush, and Earls 1997). Robert Sampson et al. emphasize that this is a structural theory that is not based in motivations, but rather the microdynamics of neighborhoods and households produced by social structural features, while Lauren Krivo and Ruth Peterson (2012) highlight the historical patterns of social policies that reproduce structural disadvantage in these places. There have been useful corrections of this perspective, and the social disorganization school of thought from which it emerges. Waverly Duck (2012, 2009) taps into a tradition that challenges the notion of “disorder” in poor neighborhoods. He shows that poor neighborhoods, even those with high rates of violence, are organized by complex interactional orders, and reveals the identifiable logic of seemingly “senseless” killings (2009). Duck argues that given the high stakes residents of violent neighborhoods confront, they are even more likely to learn and observe the local order than in other contexts. However, he makes an important distinction that gets at the source of much of the confusion around the terms anomie and disorder. He points out that the highly structured interactional orders that he describes are often quite at odds with the values, beliefs, and goals of the people who inhabit them. Thus, the issue is not a lack of values or norms, but that residents’ values and beliefs may be largely irrelevant to navigating the contexts they live in and therefore do not actually impact their behavior. Rather, their survival strategies in violent contexts are governed by locally grounded codes. The Chávez government’s achievements in reducing poverty and inequality when measured at the national level cover over significant gaps of structural disadvantage. While social policies undoubtedly empowered some people and communities and generated cohesion and social capital in some places, in others they resulted in new lines of social conflict within popular sectors (Antillano 2016). Despite considerable efforts, housing construction did not keep up with demand, leaving overcrowded neighborhoods with high rates of residential mobility. While overall employment went down, it remained high among poor young men. A gender analysis is instructive here, as social policy and participatory initiatives had very different impacts on the lives of men and women. While education coverage increased, over half of all young men were not enrolled in school or education initiatives. Women participated in much higher numbers in the government’s many participatory initiatives (Fernandes 2008). The continued decay of the judicial system (Smilde 2015) disproportionately affected the poorer social classes, leaving them to their own

The Paradox of Violence in Venezuela

resources for conflict resolution (Brinks 2008). One state institution did become much more influential in the lives of young men under Chavismo: prison. Increasing rates of incarceration carried with it all of the typical deleterious biographical and social effects. While our chapters do not directly focus on concentrated disadvantage, they point to it its importance in explaining violence in Venezuela (see chapters 5, 6, and 11); future research should include concentrated disadvantage in its analysis of crime and violence in the country (see chapter 12).

Plan of the Book This book represents an interdisciplinary effort at explaining violence in Venezuela. Chapters cover a range of methods, including ethnography, survey research, statistical analysis, and interviews, thereby providing a multifaceted and multilayered approach to the problem. The framework presented in this introduction does not represent a series of hypotheses “proven” by the empirical chapters of this book; rather, it presents a series of constructs to orient the reader and weave together overlapping but distinct insights. We hope this book will focus discussion of violence in Venezuela and the region not just by providing explanations, but through exhibiting tension points that can point to agendas for further research. Part I, “The Shape of Violence,” seeks to provide a nuanced and empirically grounded portrait of the problem of violence in the country. In chapter 1 Josbelk González Mejías and Dorothy Kronick look at the controversy over the homicide rate in Venezuela, showing how different institutions’ methods for counting homicides complicate estimation. They use a novel dataset to estimate the violent death rate, suggesting it provides a more accurate indicator than homicides. They arrive at an astronomical rate of 70 deaths per 100,000 inhabitants for 2015. In chapter 2 José Luis Fernández-Shaw Guerra looks at the increasing heterogeneity of violence in Venezuela. While violence in Latin America has generally been considered an urban problem since the third wave of democratization, he shows the growth of rural violence, evidencing the need for contextually based analysis. Part II, “Causals Processes and Cycles of Violence,” looks at the causes of violence. In chapter 3 Josefina Bruni Celli and Javier Rodríguez show a striking direct relationship between state oil income and violence, except during the neoliberal period of the 1990s when state fragility probably had other causes. They make clear the need to step back from the standard assumption that violence is the result of poverty and inequality and consider how certain influxes of resources might under-

23

24

Rebecca Hanson, David Smilde, and Verónica Zubillaga

mine institution building. In the absence of effective state institutions, armed conflict and coercive social orders emerge. In chapter 4 Andrés Antillano uses ethnographic research in a Venezuelan prison to reveal the way violence operates as a means of domination and sovereignty in social relations at the same time that it can revalorize socially excluded subjects. In chapter 5 Verónica Zubillaga looks at the importance of handguns in Venezuela’s violence. Part III, “From Civilian Police Reform to Resurgent Militarized Policing,” looks at the various ways authorities have sought to address violence as well as the challenges they have faced. In chapter 6 Luis Gerardo Gabaldón details the government’s efforts at police reform and how these efforts came up short, showing how politicization and ideological discrepancies within the government weakened citizen security initiatives. In chapter 7 Rebecca Hanson and David Smilde use polling data to reveal the Catch-22 police reformers found themselves in, with politicization and polarization destabilizing reform not only from within the state but within society. We show that government backers supported the reform in the abstract, but did not sympathize with its main tenets, while those who sympathized with its progressive elements tended not to support the government. Chapters 8 and 9 look at the at the growth of militarized policing and how it has contributed to violence, with Keymer Ávila focusing on weak state institutions and Leonard Gómez and Rebecca Hanson looking at performances of state power. These elaborate spectacles not only produce state violence; they also alter criminal networks in such a way that incentivizes the use of violence between gangs or between increasingly organized gangs and the state. Part IV, “Responses to Violence: Looking Back, Looking Forward,” looks at responses to violence. In chapter 10 Manuel Llorens asks how communities have responded to chronic violence in a context of mounting insecurity, pluralized violence, and state institutions too fragmented to provide effective solutions. Although communities have been transformed by trauma, Llorens also finds that local organizations in some spaces have engaged in violence prevention, with anti-violence pacts among communities being notably effective in some contexts. In chapter 11 Enrique Desmond Arias provides some regional points of comparison. He looks at the variety of policy hits and misses in Brazil and Colombia and suggests that lessons can be drawn out for Venezuela. While acknowledging that “high levels of crime establish a pernicious resilience,” Arias details how some countries in the region have successfully decreased record breaking rates of violence. In the final reflections we look at the current context in which crime has decreased and illicit markets reorganized in the midst of an economic collapse and a wave

The Paradox of Violence in Venezuela

of out-migration, and discuss what this might mean for the conceptual framework we put forward in the introduction.

Conclusion During his presidency, Hugo Chávez succeeded in reducing poverty and incorporating sectors that had been long marginalized and ignored into Venezuela’s political life. Nevertheless, conflicts and struggles within Chavismo resulted in an approach that focused on state-spending that failed to address structural inequalities in a universal way, constant institutional change, ambivalent and eclectic attempts to reform criminal justice institutions, and continued militarized policing. This approach not only failed to reduce violence, it facilitated a dramatic increase in it. Taken together, the material presented in this book underlines the fact that reducing poverty and inequality—while important goals in themselves—are not the same as reducing crime and violence. The latter is driven by multiple causal dynamics that are related to poverty and inequality, but largely in indirect, nonlinear ways. As has occurred under previous governments, a boom in oil prices was followed by massive public spending. Under Chávez, oil revenue was invested in health, education, and food initiatives. However, this approach did not alter pockets of concentrated disadvantage in poor neighborhoods and coexisted with a considerable de-structuration of the policing, justice, and corrections systems. As a result, the most important causal factors of violence were not improved—they worsened. Indeed, the Chávez period shows that success in reducing poverty and inequality is undercut when state institutions become increasingly fragmented, there is unequal distribution of government institutional coverage—including in crime prevention (Fajnzylber, Lederman and Loayza 2000, 247)—and the provision of security falls to communities or other non-state actors.

25

Part I The Shape of Violence

1

The Problem with Venezuelan Homicide Data, and a Solution Josbelk González Mejías and Dorothy Kronick

The cliché that it is easy to lie with statistics needs a corollary: it is even easier to lie without statistics. —Monkkonen 2001b, 7

The academic debate over homicides in Latin America generally turns on competing explanations, not competing facts. Researchers studying Mexico, for example, might disagree over whether the government’s military offensive against drug cartels (Dell 2015; Calderón et al. 2015; Lessing 2018) or (say) fluctuating drug profits (Castillo and Kronick 2020) explain more of the violence—but no one questions the fact that the Mexican homicide rate nearly tripled between 2007 and 2010. Likewise, students of Colombia might compare the effects of counternarcotics policies (Mejía et al. 2017), pro-state para-state armed actors (Acemoglu et al. 2013), or price shocks (Dube and Vargas 2013) in driving conflict—but everyone knows that the homicide rate plummeted during the presidency of Alvaro Uribe. Of course, body counts are fraught everywhere (Monkkonen 2001b; Ball and Reed 2016; Willis 2017). But the governments of Mexico, Colombia, Brazil, and Argentina, among others, publish well-documented homicide data in timely fashion, making it easy for journalists and scholars to describe homicide trends. Venezuela does not afford this luxury. Researchers for Brazil’s Igarapé Institute have written of “a near blackout of reliable [Venezuelan] government crime statistics” (Garzón and Muggah 2017). The veteran crime journalist Ioan Grillo, reporting for 29

30

Josbelk González Mejías and Dorothy Kronick

Time magazine, described Venezuela’s murder rate as a matter of fierce debate (Grillo and Benezra 2016). And the Miami Herald published an article under the headline “Dueling Data Blur Venezuelan Murder Rate” (Wyss 2016). Researchers studying Venezuela dispute even basic facts like whether the homicide rate rose or fell in certain years (see Kronick 2016; BriceñoLeón 2016; Toro 2016). This hampers research on Venezuela, a case important in and of itself (as readers of this volume no doubt agree). More than that, were it not for the dearth of data, Venezuela could help scholars evaluate competing theories about the determinants of lethal violence in Latin America. Subnational variation within Venezuela could speak to the consequences of local political changes (Dell 2015; Fergusson et al. 2018), demographic trends (De Mello and Schneider 2010), drug trafficking (Angrist and Kugler 2008), and gun supply (Dube et al. 2013), among others. So: what is wrong with Venezuelan homicide statistics? Is it simply that the government conceals official data? Or is there something else? We find that there is something else. Based on original interviews with government officials and analysis of privately obtained data, we find that a problem with the underlying datasets makes it especially difficult to count homicides in Venezuela—even relative to other countries in the region. The problem is that the Venezuelan government underuses the label homicide, and that the extent of underuse varies dramatically over time and across Venezuelan states and municipalities. Imagine that a person dies from a gunshot wound. This death might be a homicide—an “unlawful death inflicted upon a person with the intent to cause death or serious injury” (UNODC 2015)—or it might be an accident, a suicide, or the result of legal police action. Both of the Venezuelan government agencies that produce homicide statistics classify many homicides as something else. The Venezuelan health ministry classifies many homicides as violent deaths of unknown intent, a residual category; while the national investigative police classify many homicides as cases of resistance to authority, which implies police action. Because violent deaths of unknown intent include many deaths that are not homicides, and because cases of resistance to authority include many incidents that do not result in death at all, researchers cannot obtain a count of homicides by simply adding these categories to the counts of deaths labeled homicide. The upshot is that analyzing deaths coded as homicides—whether in vital statistics, produced by the health ministry, or in police data—both understates the number of homicides and misrepresents the trend (how homicides rise or fall over time).

The Problem with Venezuelan Homicide Data, and a Solution

To some extent, these problems plague homicide statistics everywhere. In other major Latin American countries, though, the problem is circumscribed: relatively few deaths fall into health ministries’ unknown intent category, allowing researchers to use vital statistics to estimate the homicide rate. Moreover, in Mexico or Colombia, for example, coding rules appear consistent over time. But in Venezuela, a large proportion of homicides are classified as violent deaths of unknown intent or as cases of resistance to authority. Worse, this proportion varies dramatically over time and across the country. We propose a partial solution to this problem. Rather than analyze intentional homicide, we propose that researchers studying Venezuela consider instead all violent deaths, which include intentional homicides, legal intervention, and violent deaths of unknown intent. While a count of violent deaths does not facilitate simple comparisons with other countries, it does allow us to make comparisons within Venezuela over time, and across Venezuelan states or municipalities. These comparisons are valuable for assessing hypotheses about the causes of Venezuela’s homicide wave (or rather, lethal violence wave). To execute our proposed solution, we first use original interviews with officials in the police and in the health ministry—the two principal sources of relevant data—to document their respective methods for counting homicides and other violent deaths. Second, we use data provided by these sources—together with data we digitize from historical records—to construct and validate a measure of the national violent death rate. Finally, we use these data to characterize descriptive facts about the Venezuelan violence wave. Some of these facts challenge popular and academic explanations. For one thing, the violent death rate doubled in the first half of the 1990s, well before the election of Hugo Chávez in 1998. This bears on theories that relate violence to rhetoric or policies of the Chávez administration. For another, the violent death rate remained low from the 1930s through the mid-1980s, throughout regime changes, an oil boom, many attempted coups d’etat, and guerrilla activity. While this is not necessarily inconsistent with analysis that points to the role of political conflict as a driver of violence in the post-1989 period, it does invite an explanation for why the mechanisms linking political conflict to lethal violence were apparently inoperative for most of the twentieth century. The same applies to work focused on oil rents as a driver of conflict. If the oil boom of the 2000s drove violence in the 2000s, did the oil boom of the 1970s similarly stoke lethal violence? If not, why not? Our goal is to set out clearly what it is that the remaining chapters of this volume seek to explain. While we would not expect any one theory to account for every feature of the lethal violence wave—to quote the

31

32

Josbelk González Mejías and Dorothy Kronick

great historian Erik Monkkonen, “Violence is complex and multicausal and no one has all of the answers” (2001a, 181)—no valid explanation will be ruled out by the facts.

Sources of Homicide Data in Venezuela In September 2015, researchers and government officials from twelve Latin American countries gathered in Bogotá, Colombia, to develop tools for measuring and improving the quality of homicide data in the region (Protocolo de Bogotá 2015). The resultant Bogotá Protocol defined a set of best practices for counting homicides in the two principal sources of homicide data: police records and vital statistics. The Bogotá Protocol also proposes standards against which to gauge the quality of the data. The protocol suggests, for example, that best practice requires recording the age of each victim, and, as a standard, states that no more than 5 percent of homicide records should fail to register victim age. The protocol also emphasizes the importance of “a high degree of convergence” between the counts based on vital statistics and those based on police records, suggesting a “20% discrepancy” as the maximum acceptable disparity between the two sources (7).1 Police Records The process of measuring lethal violence using police records2 comprises four steps: (1) opening a case file, (2) classifying the case by type, (3) counting the cases by type and by state, and (4) the use of these counts by government agencies and researchers. We describe both the de jure and the de facto procedures at each of these steps. The Investigative Police Open a Case File When a person dies from external causes in Venezuela, someone— most often a state or local police officer, other times a family member or doctor—notifies the national investigative police. Unlike in the United States, where local police forces also have investigative functions, in Venezuela, these functions are the responsibility of a single organization: the national investigative police. In theory, three agents of the investigative police then attend the scene: one agent from the national Homicide Division, another from one of four regional homicide units, and a third from the local branch office.3 The agents are meant to jointly determine which of their respective divisions should handle the investigation. The Homicide Division generally handles high-visibility cases, including deaths of women and children and police-involved shootings; the four regional homicide units

The Problem with Venezuelan Homicide Data, and a Solution

generally handle cases known internally as “routine homicides,” such as those involving gang conflict; and the local branch offices generally handle cases that might or might not be intentional homicides, such as deaths in hit-and-runs. In practice, however, many deaths are attended by agents from only one or two of the divisions, rather than all three. Officers of a Caracas-area municipal police force said in interviews that, when they notify the investigative police of a violent death, they sometimes describe it as the result of a gang shootout, so as not to have to wait for an agent of the Homicide Division. Since the Homicide Division covers the entire country with only approximately one hundred agents—many of whom are tied up in administrative work—agents from one of the four regional homicide units generally arrive much more quickly. In the words of one municipal police officer, “Sometimes you are tired and about to end your shift when you come across a body. It’s your responsibility to guard the scene and wait for the investigative police, but, uff, they can take a long time. So what you can do is call and say, ‘Hey, this was a gang shootout,’ which makes everything go much faster.”4 This behavior obviously affects the proportion of cases labeled as gang shootouts. Whichever of the three investigative police divisions takes the case then oversees the recording of evidence from the scene (fingerprints and photographs, initial interviews with witnesses, etc.). In theory, a coroner from the National Forensic Medicine Service and a prosecutor from the Ministerio Público (Public Prosecutions Service) are also present; in practice, both agencies are short-staffed, and cadavers are often removed to the morgue without the presence of either.5 The detective assigned to the case then opens an expediente policial (case file), which, at a minimum, includes a copy of the autopsy report and of the death certificate. The case file also generally contains officers’ notes from the scene, but there is no standardized incident report form. Therefore, while certain details generally appear in every case file, the included (and excluded) information varies, perhaps reflecting officers’ views (expressed in interviews) that each case is unique. Officers Classify Cases by Type Of the many types of cases handled by the national investigative police, three types clearly relate to violent death: cases of homicide, cases classified as resistance to authority, and cases in which a person dies in what might (or might not) have been a homicide. This last category is called averiguación de muerte, which we translate as “death inquest.” Because the investigative police do not use a manual, guide, memorandum, or other written document that specifies definitions of these

33

34

Josbelk González Mejías and Dorothy Kronick

three types of cases, it is difficult for outsiders (such as the authors) or even insiders (such as our interviewees) to describe the categories with any precision. Cases of homicide generally include violent deaths that meet the definition of homicide in the Venezuelan penal code: a person who commits homicide “is one who has voluntarily killed another person” (Código Penal 2000, Article 407). In practice, the police category homicide cases includes both intentional homicides and manslaughter. Deaths at the hands of police or the armed forces are an exception. The Venezuelan penal code defines the crime of resistance to authority as “the use of violence or threats” against government officials (Article 216); in general, victims of police-involved shootings are classified as perpetrators of the crime of resistance to authority (along with many other people who are not victims of shootings). Thus, resistance-toauthority cases include cases of death at the hands of the police or armed forces, injury at the hands of the police or armed forces, and also incidents of “violence or threats” against any government official, whether or not the perpetrator was harmed. The third category, death inquest, is the most nebulous. Nothing in the Venezuelan penal code or in internal investigative police documents formally defines this category, and informal descriptions differ across officers within the investigative police. Some describe death inquest cases as cases of violent deaths in which it is not immediately clear whether the death was an accident—for example, a body found with injuries that could be consistent either with assault or with an accident. Others describe death inquest cases as those in which the deceased shows no obvious signs of violence—no bullet wounds, for example—but might nevertheless have been murdered; for example, a person killed by poison. Most likely, death inquest includes cases both of these types, in unknown proportion. Within the universe of homicide cases, the police further disaggregate into six subcategories: (1) “brawl” homicides (por riña, “scoresettling” homicides [in practice, those associated with gang conflicts; ajuste de cuenta]); (2) contract homicides (por encargo); (3) lynchings; (4) robbery-homicides; (5) homicides “of passion” and; (6) homicide of unknown motive. In recent years, approximately 70 percent of homicide cases have been categorized as score-settling, which is generally understood as gang-related; however, this is almost certainly an overestimate. The patrol officers who notify the investigative police of violent deaths are responsible for guarding the scene until the investigative police arrive; for personnel reasons, the division that handles gang violence arrives faster than the division that handles higher-visibility violence. Patrol officers therefore sometimes call in a death as score-settling even

The Problem with Venezuelan Homicide Data, and a Solution

Table 1.1. Construction of police data on lethal violence.

when it is not. The artificially inflated gang violence numbers have found their way into public statements by government officials. Formally, the Fiscalía (Public Prosecution Service) is the agency responsible for categorizing cases as, for example, a homicide or an accident. In practice, however, the initial categorization of the investigative police officers who open the case rarely changes, at least within the investigative police records. Each Division Counts Its Cases To the best of our knowledge, the investigative police do not create a master list of cases. In other words, case numbers, victim details, and other case-level information reside only in the physical case files themselves, as opposed to in an index or spreadsheet. Only aggregate case counts—by type, by state, and, more recently, by municipality—are recorded in spreadsheets. Since 2012, the investigative police have recorded the aggregate number of lethal violence victims in addition to the aggregate number of lethal violence cases. The number of cases is different from the number of deaths both because a single homicide case could involve multiple deaths and because many resistance-to-authority cases involve no deaths at all. Since 2012, the investigative police have counted deaths in three categories: deaths from homicide, deaths at the hands of state security forces (resistance to authority), and violent deaths in prisons. Recording the total number of violent deaths—as opposed to creating a victim-level

35

36

Josbelk González Mejías and Dorothy Kronick

database—falls short of best practice as defined in the Bogotá Protocol, which suggests that the unit of registration should be the victim. Each of the three administrative divisions—the Homicide Division, regional homicide units, and branch offices—conducts an initial count of its own cases. (Case files initiated by the local branch offices are counted at the corresponding state-level branches). After recording the number of cases by type and by state, the Homicide Division, the four regional homicide units, and the twenty-three state branches send their respective case counts on to the national Statistics Division of the investigative police, located in Caracas. The Statistics Division then aggregates the case counts (and, more recently, victim counts) from all sources and creates a consolidated spreadsheet with national and state-level totals. Table 1.1 illustrates the flow of information. The mechanics of conveying the summary tables from the administrative divisions to the central Statistics Division in Caracas illustrate some of the challenges of data management within the investigative police. In early years, the summary tables were conveyed via a private courier service, but finding the budget for these services was a constant struggle, so much so that the work of the central Statistics Division was sometimes delayed as a result of inability to pay the couriers. Later, the investigative police began transmitting the summary tables via fax, which was cheaper but introduced new problems. The fax machine paper was not durable, which meant that agents in the central Statistics Division had to transcribe every table. Besides creating additional work, transcription likely produced errors: faxed tables were sometimes difficult to read, and long shifts increased the risk of typographical mistakes. Even the arrival of the internet and email did not resolve transmission difficulties. Because many of the regional investigative police offices lacked reliable internet connections, agents would sometimes save the files on pen drives and send them from nearby cybercafes. The mechanics of recording the case counts have also changed. Commissioner Hugo Gamarra, in his role as the head of the Statistics Division from the early 1980s through the early 2000s, oversaw the replacement of handwritten or typewritten tables with Microsoft Excel spreadsheets. Despite subsequent attempts to replace Excel spreadsheets with a dedicated internal information system,6 Excel spreadsheets persisted as the dominant method of recording and storing investigative police data. Just as there is no written guide to categorizing cases, there is likewise no written guide or manual for the procedures by which the Statistics Division aggregates the figures it receives from the Homicide Division, the homicide units, and the state branch offices. These procedures are therefore sensitive to personnel changes. In interviews, agents in the

The Problem with Venezuelan Homicide Data, and a Solution

Statistics Division said that the content and format of the Excel spreadsheets can change in response to how the minister of the interior “likes to have the information presented” (the investigative police is part of the Ministry of Justice). On the other hand, continuity in the directorship of the Statistics Division has provided considerable stability. When Commissioner Hugo Gamarra left the Statistics Division in the early 2000s, after nearly two decades as director, one of his deputies—José Antonio Rojas—took his place. Rojas generally ran the division in Gamarra’s image. When Rojas left, two inexperienced directors followed in quick succession; then, in 2010, Cristina Rojas—formerly of the Human Resources Department within the investigative police—took the helm. While she had little experience with data, her organizational and managerial capacities improved the functioning of the division, and she was able to learn from agents who served under Gamarra and Rojas. Access to and Use of the Police Data From early in the twentieth century until 2003, national and state-level counts of each type of case were published in a government document called the Statistical Yearbook of Venezuela.7 (The Statistical Yearbook also published demographic, economic, climatological, and epidemiological data, along with hundreds of pages of other information.) After 2003, the Chávez administration discontinued the Statistical Yearbook series entirely, thereby ending the regular publication of the data collected by the investigative police. While the annual reports of the Ministry of Justice sporadically mention certain case counts, they do not regularly or reliably publish the data previously available in the Statistical Yearbooks.8 In the absence of any regular or reliable publication of the case counts constructed by the investigative police, researchers and journalists have requested the data both formally (in writing, to the director of the investigative police) and informally (though personal connections). To the best of our knowledge, the latter is the only viable way to access the investigative police data. Health Ministry Data The Venezuelan health ministry provides a second source of data on violent deaths, and its procedures are more formal and better documented.9 The health ministry process comprises three steps: the collection of paper death certificates, the transcription of the death certificates into a central database, and then use of the microdata to produce a Mortality Yearbook.

37

38

Josbelk González Mejías and Dorothy Kronick

A Medical Professional Fills Out a Death Certificate When a person dies in Venezuela, a medical professional fills out a paper death certificate (Form EV-14, reproduced in the online appendix, figure A.5). The type of medical professional who fills out the certificate—whether an attending physician, an agent of the forensic medicine service, or an officer of the health ministry—depends on the location and nature of the death (MPPS 2012, 14).10 De jure, coroners from the National Forensic Medicine Service fill out the certificate in cases of violent death; in practice, especially when victims die in a hospital, physicians complete the certificates. The death certificate records, among other information, the identity of the deceased, their location of residence, the location of death, and the cause of death. Critically, the medical professional who fills out the certificate describes the causes of death in words (section 54 of the certificate; see figure A.5) but does not fill out the corresponding code in the International Classification of Diseases (section 56). Coverage In practice, the vast majority of deaths produce a death certificate. A Venezuelan government study estimated coverage at 96 percent in 1998; in 2008 the World Bank estimated coverage at 96.8 percent, and a more recent report by Venezuela’s National Statistics Institute indicates slightly higher coverage by including late registration (OCEI 1991; Danel and Bortman 2008; INE 2013). Another way to assess the coverage of the health ministry mortality data is to compare the number of death certificates to the United Nations’s estimate of the overall crude death rate, which is based not only on death registration itself but also on survey data and predictions based on population structure (United Nations 2017a, 4). For 2010–2015, the United Nations estimated Venezuela’s crude death rate at 5.5 per 1,000 (United Nations 2017b); that would imply 163,824 deaths in 2013, given the population reported by the Venezuelan National Statistics Institute. The mortality microdata for that year contain 149,903 records, suggesting coverage of 92 percent. While Venezuela’s Organic Law of Civil Registry formally requires family members, police officers, medical professionals, and “any civilian aware of the death of an unknown person” to declare deaths (Article 126), the high coverage is likely due not to this formal requirement, but rather to the fact that, in practice, a death certificate is needed to obtain the permissions required for cremation or burial. (Coverage for birth registration is much lower.)

The Problem with Venezuelan Homicide Data, and a Solution

Coding and Transcription of Paper Death Certificates The original paper death certificates are sent to the health ministry in Caracas, via the municipal and state branch offices (MPPS 2012, 12).11 Within the ministry, the group that receives and processes the certificates is called the Division of Information and Health Statistics (Dirección de Información y Estadísticas en Salud, or DIES).12 DIES is part of the General Directorate of Epidemiology, which in turn is part of the Viceministry of Public Health (one of five viceministries; see figure A.6 for an organizational chart). While there has been considerable turnover in the directorship of DIES, many of the technical staff have had long tenures. The physicians and other medical professionals who fill out the paper death certificates simply describe the cause of death in words, rather than choosing a code from the International Classification of Diseases (ICD) (MPPS 2012, 20, 37). Choosing and recording the appropriate ICD code is the responsibility of technical staff within DIES. For sixty-three years (1955–2018), the World Health Organization maintained a Collaborating Center for the Classification of Diseases within the Venezuelan ministry of health; this collaborating center was effectively a part of DIES, and its staff assisted in the correct coding of causes of death (Gabaldón 2018; Danel and Bortman 2008, 93). Officers of the health ministry transcribe the paper death certificates using dedicated software called the System of Information in Health (Sistema de Información en Salud, or SIS). This software was designed to minimize transcription errors by employing drop-down menus and radio buttons wherever possible (see screenshots in online appendix figures A.3 and A.4). Unlike the national investigative police, the health ministry thus complies with the Bogotá Protocol recommendation that the victim constitute the unit of registration. Publication of the Vital Statistics Data In defining best practices for the dissemination of mortality data, the Bogotá Protocol (2015, 8) states that: (1) “Official data on homicides, both national and local, will be publicly disseminated,” and that (2) “In addition to aggregate numbers, the micro-data of homicides, victim by victim, will be freely accessible, with the exception of information that may lead to individual identification of the persons involved . . . with the lowest level of geographic disaggregation that does not compromise the confidentiality of the victims or alleged perpetrators.”

39

40

Josbelk González Mejías and Dorothy Kronick

Table 1.2. Does Venezuela meet the Bogotá Protocol standards?

While the national investigative police publish neither aggregate numbers nor microdata (indeed, the police do not even create microdata), the health ministry publishes aggregate numbers (“both national and local”) and, while the anonymized microdata are not available online, they have been made available to academic researchers upon request. More specifically, summary tables on mortality within Venezuela have been published continuously since the nineteenth century. From 1877 through 2003, these tables were published in the Statistical Yearbook of Venezuela, alongside the police records described above and hundreds of pages of data on unrelated topics.13 While the earliest volumes published only total deaths by location within the country (that is, not disaggregating by cause), already by the early twentieth century the Statistical Yearbooks published the number of deaths by location and by cause (see online appendix figure A.7 for an example). Beginning in 1938, another periodical—the Yearbook of Epidemiology and Vital Statistics, published by the health ministry—provided even more disaggregated figures than those available in the Statistical Yearbook of Venezuela. To the best of our knowledge, the first microdata files date from the early 1990s. Online appendix table A.2 summarizes the availability of the health ministry data.

The Problem with Venezuelan Homicide Data, and a Solution

Table 1.3. Cases and violent deaths by type, 2013.

Comparing the Quality of the Two Sources Table 1.2 summarizes how Venezuelan data on violent deaths perform against the standards set by the Bogotá Protocol. Overall, the health ministry data performs much better, in that it uses the victim as the unit of registration, and in that it publishes aggregate data and makes the microdata available to researchers. Despite the deficiencies of the police data, however, the trends it reveals are largely consistent with those visible in the health ministry data.

The Impossibility of Counting Intentional Homicides, and an Alternative Public laments about Venezuelan homicide statistics sometimes imply that the relevant data do not exist, or that the government never make them available to researchers. We documented that the relevant data do exist, and that while they are not as easy to access as one might like, researchers can obtain them. In practice, though, these data have a fatal flaw: inconsistent application of the label homicide. Here, we explain the problem and propose a partial solution. The Impossibility of Counting Intentional Homicides in the Health Ministry Data Despite the strengths of the health ministry data on violent deaths, these data undercount the number of homicides. Worse, the extent of the undercounting changes over time and varies across Venezuelan states and municipalities. Consider, for example, the data from the year 2013, presented in table 1.3. While the police recorded 14,781 deaths from intentional homicides, the health ministry recorded only 8,841 deaths from inten-

41

42

Josbelk González Mejías and Dorothy Kronick

Figure 1.1. The impossibility of counting intentional homicides in Venezuelan vital statistics data.

tional homicides—and 9,565 violent deaths of unknown intent. (The vast majority of these were gun deaths.) In other words, many—if not most—of the violent deaths that the health ministry codes as of unknown intent were in fact homicides. Few other Latin American countries classify such a high proportion of violent deaths as of unknown intent: in Colombia, for example, the proportion is approximately 10 percent; in Mexico it is approximately 15 percent. Second, the proportion of violent deaths that the health ministry codes as intentional homicides (as opposed to of unknown intent) varies dramatically over time and across states and municipalities, in ways that suggest changes in coding practices rather than changes in homicide incidence. The changes in coding practice stem in part from revisions in the International Classification of Diseases. From the 1950s through 1967, the Venezuelan health ministry coded causes of death according to the seventh revision (ICD-7), which did not have a code for “of unknown intent” (NCHS 1975, 37–38, 52). The unknown intent category appeared

The Problem with Venezuelan Homicide Data, and a Solution

for the first time in the eighth revision (ICD-8). When Venezuela adopted ICD-8 in 1968, many deaths that would previously have been classified as homicides—and a few that would have been classified as accidents—were subsequently coded as of unknown intent, as table 1.3 makes clear. When Venezuela adopted the ninth revision of the ICD, in 1979, use of the unknown intent category changed sharply for a second time (figure 1.1). But revisions to the ICD are not the only reason for shifts in the use of the unknown intent category in Venezuelan mortality statistics. Figure 1.1 reveals that after falling sharply with the introduction of ICD-9 in 1979, the proportion of violent deaths coded as of unknown intent increased from the mid-1980s through the mid-1990s—a period in which ICD-9 was used continuously (Venezuela adopted ICD-10 in 1996). In 1997–1999, the number of unknown intent violent deaths actually surpassed the number of known intentional homicides (figure 1.1). This problem is less pronounced in other countries. Figures 1.2–1.4 plot the proportion of violent deaths classified as homicides or as violent deaths of unknown intent in the vital statistics of the United States, Mexico, and Colombia (we omit accidents and legal intervention for visual clarity). The introduction of the unknown intent category with ICD-8 in the 1960s clearly affected use of homicide codes in all three countries, and most dramatically in Mexico. But by the 1980s, coding rules appear to have stabilized: the homicide proportion rises slightly when conflict increases (as in Mexico after 2006), or falls when crime falls (as in the United States in the early 1990s), and the unknown intent proportion generally does not surpass 20 percent. In Venezuela, in contrast, the proportion of violent deaths labeled unknown intent rises from the mid-1980s through the late 1990s, even exceeding 50 percent. Figure 1.1 makes clear why the Venezuelan vital statistics data both underestimates the homicide rate and misstates the trend: because many homicides appear to be miscoded as violent deaths of unknown intent. Moreover, use of the unknown intent category varies across Venezuelan municipalities and within municipality over time. In the Federal District (part of Caracas), for example, officials coding causes of death moved away from the intentional homicide category over the past half-century; the proportion of violent deaths coded as intentional homicides fell from 100 percent in 1960 nearly to 0 percent by 2012.14 In Maracaibo, Zulia, in contrast, officials code 80–90 percent of violent deaths as intentional homicides. Using these data to compare the homicide rate in the Federal District and in Maracaibo would thus be uninformative. Despite the strengths of the health ministry data, therefore, we cannot use it to measure the intentional homicide rate, both because inten-

43

44

Josbelk González Mejías and Dorothy Kronick

tional homicides are undercounted, and because the extent of undercounting varies over time and across states and municipalities within Venezuela. The Impossibility of Counting Intentional Homicides in the Police Data The police data suffer from a related, but distinct, classification problem. Deaths perpetrated by the police or other state security forces—whether legal or illegal—are generally classified as cases of resistance to authority. As noted earlier, cases of resistance to authority include: (1) Cases of death at the hands of the police or armed forces (2) Cases of injury at the hands of the police or armed forces (3) Incidents of “violence or threats” against any government official, whether or not the perpetrator was harmed.

This means that we cannot obtain a count of homicides from the police data, even if we were willing to assume that all police-involved killings were homicides. If we were to add the police count of resistance-to-authority cases to the police count of homicide cases, we would (in most years) overestimate the number of homicides because many resistance-to-authority cases do not involve any deaths at all. Moreover, the extent of the overestimate would vary dramatically over time. Since 2012, the police have maintained counts of victims in addition to counts of cases; these data provide a glimpse into variation in the relationship between resistance-to-authority cases and deaths. In 2013 deaths from resistance to authority made up just 22 percent of resistance-to-authority cases (see figure 1.4); by 2015 it was 57 percent; and by 2017 there were more deaths than cases (implying that a single case of resistance to authority involved multiple victims). This variation precludes simply deflating the number of resistance-to-authority cases by a constant in order to infer deaths from cases. To summarize, the police count of homicide cases both understates the number of homicides (because it excludes homicides committed by state security forces) and misstates the trend in homicides (because state violence has fluctuated dramatically over time).15 Nor can we add homicide cases and resistance-to-authority cases, because a large and variable fraction of resistance-to-authority cases entail no deaths. If the state did not perpetrate much lethal violence in Venezuela, this might not be cause for concern. No count of homicides is perfect; if 1 percent or even 5 percent of homicides were committed by police, perhaps we could simply exclude these deaths for the purposes of some studies. But in fact, a large proportion of all lethal violence victims are

The Problem with Venezuelan Homicide Data, and a Solution

Figure 1.2. Comparison of trends in vital statistics and police data.

killed by the state. The police data collected since 2012 allow us to estimate this proportion with some precision: the state killed 7 percent of all lethal violence victims in 2013, and 27 percent by 2017. While we lack reliable figures for earlier years, we know that the state was responsible for a large fraction of lethal violence in 1989, the year of the Caracazo. To study homicide in Venezuela without attention to police violence would be misleading at best. A second problem with the police data is that a large number of deaths fall into the averiguación de muerte (death inquest category). As explained earlier, death inquest includes both violent deaths in which the intent (homicide, suicide, or accident) was unknown and deaths which might or might not have been from external causes (usually found bodies). In 2013, 21 percent of deaths in the police data fell into the death inquest category—more than double the maximum of 10 percent recommended in the Bogotá Protocol. We cannot know what proportion of these deaths were in fact intentional homicides. Unlike the unknown intent category in the ministry of health data, however, rules for using the death inquest category do not obviously change over time. Counting Violent Deaths We cannot use the Venezuelan health ministry data to count homicides because a large and variable proportion of homicides are misclassified as violent deaths of unknown intent, a category that also includes many nonhomicides. And we cannot use police data to count homicides because a large and variable proportion of homicides are misclassified as cases of resistance to authority, a category that also includes not only legal police killings but also many nonlethal interactions. The problem is

45

46

Josbelk González Mejías and Dorothy Kronick

Figure 1.3. Comparison of lethal violence in vital statistics and police data.

minor compared to the impossibility of counting homicides committed by state security forces. Given the impossibility of counting intentional homicides in Venezuela in a way that makes for meaningful comparisons across time and space, we propose instead that researchers use a count of violent deaths, based on the health ministry data. Our measure includes homicides, lethal injuries of undetermined intent, “legal intervention,” and firearm accidents; we exclude motor vehicle deaths, known suicides, and accidents other than firearm accidents (such as accidental drownings). Appendix table A.3 lists the specific ICD codes included for each revision. To provide a sense of how this measure compares with other possible measures of homicide and of lethal violence in Venezuela, consider figure 1.2. Figure 1.2 plots the homicide rate and the violent death rate as measured in the health ministry data together with the number of homicide cases per 100,000 and the number of violent-death-related cases per 100,000 from the police data. By number of violent-death-related cases, we mean the sum of homicide cases, cases of resistance to authority, and cases of death inquest. The trends are not entirely dissimilar; indeed, the correlation between the differenced series—that is, the changes from one year to the next—is 0.88, which boosts our confidence in both sources.16 But considering the violent death rate rather than the homicide rate also pro-

The Problem with Venezuelan Homicide Data, and a Solution

Figure 1.4. National and regional trends in violent death, 1958–2013.

duces corrections: the health ministry’s homicide rate is artificially low especially in the 1960s and in the post-1989 period, due to changes in coding practices. Figure 1.3 clarifies these differences by using a log scale on the y-axis (which approximates percent change rather than absolute change). This emphasizes that while the homicide rate as measured in the police data increased sharply in the early 1970s, the violent death rate did not. The same is true of cross-sectional variation (figure 1.3). That stateand municipality- level violent death rates are highly correlated across the two sources increases our trust in our proposed measure of violent death rates. But using the violent death rate rather than the homicide rate allows us to make credible comparisons both over time and across Venezuelan states and municipalities.

Venezuela’s Lethal Violence Wave The goal of the previous section was to introduce and validate our proposed measure of lethal violence; here we use that measure to characterize Venezuela’s violence wave; that is, the phenomenon that this book seeks to explain. From the 1950s through 1988, the violent death rate hovered between 10 and 15 per 100,000 (figure 1.4). This fact bears on theories that point to political unrest and/or a surge in the oil price as the cause of violence in later years. For example, Roberto Briceño-León (2012, 3236) cites the coups d’etat of 1992 as the cause of a “institutional breakdown that provoked some substantial changes in the social contract,”

47

48

Josbelk González Mejías and Dorothy Kronick

and thereby led to more lethal violence in that year and the following year. Observing the (relatively) stable violent death rate through the 1960s, we might ask why the coups d’etat attempts of 1961 or 1962 did not have similar consequences (or, for that matter, the actual regime change in 1958). If the coups d’etat “made democratic rules dispensable and superficial and justified the use of violence in order to reach political goals” (3236), why did the urban guerrilla strikes in the 1970s not have similar effects? The violent death rate doubled between 1989 and 1996, reaching 32 per 100,000. We view this early 1990s doubling of the violent death rate as an important part of the puzzle that this book seeks to explain. Of course, a significant (but unknown) fraction of the increase from 1988 to 1989 was driven by deaths associated with the Caracazo; while subsequent political unrest might seem like a straightforward cause of lethal violence in the early 1990s, the direct death toll of the two coups d’etat attempts in 1992 accounted for just 5 percent of all violent deaths that year (Kronick 2020, 1506). That said, a thorough attempt to count the deaths associated with political unrest in these years would likely produce a higher total. In any case, the fact that the violent death rate doubled before Hugo Chávez was elected president in December 1998 bears on theories that point to Chavismo as the source of Venezuela’s lethal violence wave. Briceño-León (2012), for example, describes “the institutional destruction that comes with [Chávez’s] Bolivarian revolution” as an extension of the institutional crisis of the early 1990s and a primary cause of the homicide wave (3233, 3288). A small group of municipalities drove the increase in the violent death rate between 1989 and 1996 (figure 1.4). The increase in violence in this period was geographically confined to the capital region (then called the Federal District), neighboring states Miranda, Carabobo, and Aragua, and select other municipalities (Briceño-León and Pérez Perdomo 1999; Kronick 2020).17 Kronick (2020) views this geographic variation, together with qualitative data, as evidence of rising violence related to drug trafficking. This pattern could also be consistent with criminal violence associated with political unrest. Another plausible hypothesis is that decentralization—which included the proliferation of municipality-level police forces—affected violent death rates. This strikes us as an opportunity for future work. After a lull in 1997–1998, the violent death rate increased sharply nationwide beginning in 1999 and through 2003. The violent death rate jumped from 30 per 100,000 in 1998 to 37 per 100,000 in 1999 and then

The Problem with Venezuelan Homicide Data, and a Solution

47 per 100,000 the following year. By 2003 it reached 65 per 100,000. Unlike the increase of the early 1990s, which was geographically confined, the 1999–2000 increase occurred in nearly every state in the country (figure 1.4; appendix figure A.8). Rebecca Hanson and Dorothy Kronick (2022) suggest that the implementation of a new criminal procedure code in July 1999 might have driven this sharp increase in lethal violence, though the evidence is not definitive. Briceño-León (2012, 3238) suggests instead that the arrival of President Chávez, and in particular his “restriction of police action” and “speeches of uprising,” drove the increase. The violent death rate climbed more gradually after 2003. Between 2003 and 2013 (the last year in our health ministry data), the violent death rate continued to climb, reaching a high of 69.5 per 100,000 in 2012. But the increase during this period was not constant; the violent death rate declined in six of these ten years, such that the overall upward trend was much more gradual than in 1989–1996 or 1999–2003. This pattern bears on all of the explanations mentioned thus far: that Chávez’s rhetoric or “institutional destruction” drove the violence, that oil rents were responsible, or that the conflict stemmed from changes in illegal markets. In all cases, we might ask: is this explanation consistent with slower average growth in the violent death rate after 2003?

Toward a New Measure of Lethal Violence in Venezuela Our aim has been to pull back the curtain shrouding Venezuela’s oftdeplored statistics on lethal violence. The government’s failure to publish data has made it difficult for journalists and the public even to know whether lethal violence rose or fell in a given year, much less for researchers to use quantitative measures to evaluate competing theories about Venezuela’s violence wave, or to inform theories about lethal violence in Latin America more generally. Based on original interviews with officials in Venezuela’s health ministry and in the national investigative police, we document that both agencies do collect useful data on violent deaths. We also find, however, that these data cannot be used to produce a reliable measure of the homicide rate—that is, the rate of violent deaths inflicted unlawfully and intentionally. The problem is that both agencies (the health ministry and the national investigative police) lump many homicides together with nonhomicides—or even nonlethal events—in residual catchall categories that cannot be disaggregated ex post. This problem plagues Venezuelan homicide statistics more than those of other major countries in the Americas.

49

50

Josbelk González Mejías and Dorothy Kronick

We propose that researchers studying Venezuela focus instead on the violent death rate, which includes intentional homicides, firearm accidents, violent deaths of unknown intent, and deaths in legal police action. The disadvantage of this measure is that it may not be the quantity of interest for some studies. The advantage is not only availability, and not only comparability over time and across Venezuelan states and municipalities, but also that the violent death rate may indeed be the quantity of theoretical interest for some studies. We hope that our defense of this measure, as well as our publication of the corresponding data, will serve other researchers who seek to understand the Venezuelan case. In our view, quantitative work leveraging subnational variation can help evaluate the hypotheses set out in the extensive qualitative literature on violence in Venezuela. More than that, or in addition to it, we expect that the Venezuelan case will inform theories of criminal conflict and state violence in Latin America more generally (e.g., Yashar 2018; Lessing 2018; DuránMartínez 2017; González 2020; Castillo and Kronick 2020). To paraphrase Lessing, there are many plausible theories and not enough cases to rigorously test them (2018, 77). One more case, we hope, can make a difference.

2

The Diversification of Venezuela’s Violence A Quantitative Exploration José Luis Fernández-Shaw Guerra

Violence is a complex phenomenon dependent on multiple determining, conditioning, and intervening factors. The reduction of this phenomenon to a generic quantification of its volume does not contribute much to our understanding it. Violence is not only complex due to the number of variables involved but also because weights for these factors are not necessarily constant under all conditions, territories, or time. Triggers of violence can be different in different contexts: a crisis, for example, may generate new intervening variables or change the relations established among them. This chapter presents a series of reflections sparked by an exploratory review of data on homicides and violent deaths in Venezuela in recent decades. This is not a final analysis, but a presentation of indicators and evidence that can chart a path for analysis, in the hope that others will join this effort. This is, therefore, a descriptive effort leading to avenues for future research. Violence is generally presented as a unique and undifferentiated phenomenon, which is expressed by a national indicator—such as homicide rates or the number of deaths caused by firearms.1 This creates a methodological illusion in which each point making up a line representing a trend is an expression of the same “substance” expressed by another point. Thus emerges a simplistic representation in which deaths caused 51

Graph 2.1. Venezuela. Deaths caused by firearms and homicides, 1990–2017. Source: MPPS, MPPRIJP, author’s calculations.

Figure 2.1. Venezuela. Homicides and other violent deaths caused by firearms (per 100,000 inhabitants), 2016. Source: MPPRIJP, CICPC, author’s calculations.

The Diversification of Venezuela’s Violence

Graph 2.2. Caracas Metropolitan Area and its periphery. Rates of homicides (2013–2017) and death caused by firearms (1999–2013).

by firearms in 1989 (the year of the famous Caracazo riots) are equal to the 2008 deaths, or to the deaths of victims of militarized police operations in recent years. This reduces any possible difference of magnitude among the phenomena. A useful visualization for the purposes of this chapter is the relationship between homicides and deaths caused by firearms. The general tendency for both indicators is very similar, due to the fact that in Venezuela a very high proportion of homicides are committed with firearms. By 2014, according to data from deaths certificates issued by the Health Ministry (MPPS), 80 percent of violent deaths were the result of firearms. Similarly, in 2017, according to data from the Ministry of People’s Power for Justice and Peace Internal Relations, 85 percent of homicides were committed with a firearm. With the aim of grasping the peculiarities of these forms of violence (which, as can be seen in the graphic and illustration in Graph 2.1 and Figure 2.1, have grown and expanded across the territory), I have used indicators of deaths caused by firearms and homicides in the local context 2 in which they occur in order to understand the fundamental variables that shape violence. To this end the data has been disaggregated to its minimum geographic unit possible and then recomposed into geographical units that account for changes in their dynamics. A methodological observation is important at this point: this analytic strategy

53

54

José Luis Fernández-Shaw Guerra

Graph 2.3. Caracas Metropolitan Area, rates of homicides and deaths caused by firearms by area. Sources: MPPS, MPPRIJP, OVS, author’s calculations. Note that the continuous line indicates firearms, while the dotted line indicates homicides.

requires municipal level data, which is not available from one single source for the entire study period. The systematization of data on homicides by the Venezuelan Security Observatory (OVS), an agency of the Ministry of Interior and Justice (MPPRIJP), started in 2012, and the data on deaths from firearms gathered from death certificates issued by MPPS stops in 2014. Thus, I have joined these two datasets based on the premise that there is no significant difference between the two indicators. 3

Expansion of Violence outside Urban Limits By the end of the 1990s there was broad consensus among specialists that violence in the region, and in Venezuela, was a predominantly urban phenomenon.4 However, in Venezuela there are signs this has changed in recent years. By comparing the evolution of violence in cities with violence in the rest of the country, a significant shift becomes apparent. In Venezuela, violence is no longer a predominantly urban phenomenon. Graph 2.3 shows a comparison of the proportion of deaths caused by firearms for urban versus rural areas. These are the areas I label major urban,5 which are contrasted with the rest of the territory. It is clear that twenty years ago analysts were not wrong to say that violence was primarily an urban phenomenon: more that 70 percent of deaths happened in cities. For the most recent period, however, the proportion of deaths

55

The Diversification of Venezuela’s Violence

Table 2.1. Venezuela, parishes with more than 75,000 inhabitants. Regression model for death rates caused by firearms Nonstandardized coefficients for deaths caused by firearms Models

Model 1

Model 2

Model 3

Model 4

Model 5

Coefficients

Coefficients

Coefficients

Coefficients

Coefficients

 Significance within the model

Constant

124.181

163.131

153.603

152.745

-45.122

0.599

% of men between ages 18 and 24 attending education institution

-1.431

-2.021

-1.921

-1.929

-2.39

0.000

-2.441

-5.008

-7.808

-4.796

0.021

2.614

3.865

4.106

0.000

1.039

2.362

0.001

2.417

0.021

% of extreme poverty (Unsatisfied Basic Needs method) % quality housing deficit % public services deficit % of households not in poverty

Model summaries R

0.484

0.571

0.614

0.647

0.675

R Squared

0.234

0.326

0.377

0.419

0.456

Adjusted R Squared

0.225

0.31

0.354

0.391

0.422

Standard error of estimates

24.9

23.5

22.7

22.1

21.5

t and signficance % of men between ages 18 and 24 attending education institution % of extreme poverty (Unsatisfied Basic Needs method) % quality housing deficit % public services deficit % of households not in poverty

-5.13

-6.41

-6.25

-6.46

-6.82

0.000

0.000

0.000

0.000

0.000

-3.40

-4.17

-4.78

-2.35

0.001

0.000

0.000

0.021

2.62

3.53

3.83

0.010

0.001

0.000

2.45

3.39

0.016

0.001 2.350 0.021

56

José Luis Fernández-Shaw Guerra

is slightly higher in the rest of the country than in major urban centers. In 1999 major urban centers concentrated 44 percent of the population and 71 percent of deaths caused by firearms. By 2017, this situation had changed: major urban areas concentrated only 46.63 percent of homicides in the country. In the areas I have called major urban centers, the murder rate increased by approximately ten points. In the rest of the country it increased by almost thirty points, from 13.86 (deaths caused by firearms for every 100,000 inhabitants) in 1999 to 42.29 (homicides for every 100,000 inhabitants) in 2017. In sum, violence increased outside of the urban areas at a higher pace than in the big cities, which means that today our focus need not be exclusively urban. This finding coincides with the UN Global Study on Homicide report that in Latin America violence is primarily increasing outside big cities (UNODC 2019, booklet 2, 55).

Violence as a Complex Phenomenon Can we assume that growing violence in nonurban contexts is caused by the same determinant, conditioning, and intervening variables as that of major urban areas? Can the same interpretative keys be used for both settings? To explore this question I constructed a database with twenty-two variables for every parish of the country with more than seventy-five thousand inhabitants, a size that allows for stability of rates6 and approximates a lower limit for typical urban sectors. This provided eighty-eight parishes, which constitute 8 percent of the country’s parishes, but account for 52 percent of all deaths caused by firearms. The dependent variable was death by gunshot; the group of variables included the general socioeconomic conditions of households (using the “basic unsatisfied needs” method for calculating poverty), access to education among all members of the household, as well as the material conditions in which they live, including housing and access to basic services. I also included variables regarding the territorial space inhabited, such as the percentage of rural population and the percentage of homes that have access to garbage removal service, something that reflects the precarious conditions typical of self-constructed barrios in Venezuelan cities.7 The independent variables were expressed in each case as simple percentages in order to have a prestandardization. No other standardization was carried out based on range or variability. I applied linear regression to this database. In light of the prestandardization of variables, the resulting coefficients of each variable provide a first view of their direct importance. I then used stepwise regression utilizing the probability of F as a criteria, with a beginning value of 0.05 and an ending value of 0.10, such that the algorithm incorporates

57

The Diversification of Venezuela’s Violence

Table 2.2. Venezuela, parishes with presence of deaths caused by firearms. Regression model for the rate of deaths caused by firearms Nonstandardized coefficients for deaths caused by firearms Models

Model 1

Model 2

Model 3

Model 4

Model 5

Significance within model

Constant

83.05

129.81

146.82

144.31

159.14

0.000

% of men between ages 18 and 24 attending education institution

-0.786

-1.322

-1.574

-1.47

-3.211

0.000

-1.201

-1.054

-1.501

-1.383

0.000

-1.632

-1.976

-2.303

0.000

0.86

0.826

0.003

1.596

0.018

% of extreme poverty (Unsatisfied Basic Needs method) % houses with educational deficit % quality housing deficit % of men between ages 18 and 24 attending education institution

Model summaries R

0.165

0.219

0.248

0.271

0.283

R Squared

0.027

0.048

0.062

0.074

0.08

Adjusted R Squared

0.026

0.046

0.058

0.069

0.074

Standard error of estimates

52.0

52.0

51.1

50.8

50.6

t and significance % of men between ages 18 and 24 attending education institution % of extreme poverty (Unsatisfied Basic Needs method) % houses with educational deficit % quality housing deficit % of men between ages 18 and 24 attending education institution

-4.64

-6.22

-7.02

-6.52

-4.17

0.000

0.000

0.000

0.000

0.000

-4.11

-3.59

-4.62

-4.220

0.000

0.000

0.000

0.000

-3.32

-3.95

-4.45

0.001

0.000

0.000

3.14

3.02

0.002

0.003

2.363

2.363 0.184

58

José Luis Fernández-Shaw Guerra

Table 2.3. Venezuela, comparison of two models. Parishes with more than 75,000 inhabitants and total parishes with deaths caused by firearms  

Models

 

> 75,000 Inhab.Total parishes

R

0.675

0.283

R Squared

0.456

0.080

Adjusted R Squared

0.422

0.074

Standard error of estimates

21.5

50.6

in each step the variable with the strongest association on the dependent variable, and successively incorporates the next variable according to its level of significance, given that in each step the significance is recalculated to minimize “redundant” effects. The variables can also drop out of the model, if a new configuration of variables shows a better adjustment. The final order of the variables shows their prioritization. The algorithm stops when it does not find any other variables that significantly add to the relationships already discovered. The rate of deaths caused by firearms8 was regressed on a series of variables described in the list at the end of the chapter. The final model showed an R-squared of 0.456. Since the variables were added stepwise by the algorithm by virtue of their importance, the R-squared can be decomposed by subtracting the R-squared of each model from the previous model, thus, attendance at educational centers for young men between eighteen and twenty-four years of age (0.234), percentage of the population in extreme poverty (0.092), percentage of homes with inadequate quality of life (0.051), percentage of homes lacking public services (0.042), and percentage of non-poor population (0.037). It is also important to analyze the resulting coefficients and their directions. Given the prestandardization described earlier, the magnitude of effects give us a sense of their weight in the estimation. But the sign gives a sense of whether they favor or inhibit the rate of violence. As can be seen by evaluating the coefficients, the percentage of men between eighteen and twenty-four years who attend educational centers and the percentage of extreme poverty are inversely related to the rate of gun violence. In other words, the greater the percentage of men studying, the lesser the violence. As well the greater the extreme poverty, the lesser the violence. This finding is consistent with one of the main points of this book. The other variables are directly related to the rate of violence and underline the importance of adequate physical space. The data suggest an important relationship between education cover-

The Diversification of Venezuela’s Violence

age and violence. This link could have several policy implications, from the importance of providing young people in school with a safe environment where they feel protected (an idea at the heart of violence prevention programs such as the “open schools” in Brazil), to the importance of providing wellness spaces for young people, removing them from gang activities that compete for their attention. Despite the inverse relationship with “extreme poverty,” the correlation with poverty is evident in the direct relationship to deficiencies in housing and services. However, using the same model with the same method and the same variables for all parishes in the country (only excluding those parishes with no incidence of the phenomenon), the relationships just described become insignificant. Beyond methodological and statistical issues— such as population size or how the inclusion of more observation units results in higher deviations, weakening the ability to establish tendencies—it may be that the logic underlying the phenomenon could be different in this larger universe. As can be seen in this second model, the variables included are different, and their weight and importance within the series of variables vary. This suggests that the variables at play in parishes of more than seventy-five thousand inhabitants (in which the urban logic is supposed to weigh more) have a different configuration and causal relationship to armed violence in the rest of the country. Adding in all the parishes in the country also weakens the model; it loses strength in its capacity to provide a solid explanation, as can be seen in table 2.2. As mentioned earlier, models that find very significant quantitative relations in some contexts can lose all descriptive relevance in territories with a different character.9 It is important to reiterate that with this statistical exercise I do not claim to build a finished model for the estimation or prognosis of rates of deaths caused by firearms—even the best of the two models presented is far from being a useful instrument for this. The purpose here is to demonstrate the necessity of using territorial units of analysis that are relatively homogeneous (in terms of variables we anticipate affecting the phenomenon). The more heterogeneous the units of analysis, the more multiple underlying logics of violence will counteract and nullify each other, undermining explanatory strength. What needs to be discovered, below the diffuse panorama of heterogeneity, is a series of smaller universes within which there are relatively consistent logics. This is also an important point to stress in terms of anti-violence policies: what works in one context will not necessarily work in another. Thus, unpacking units of analysis in order to account for local dynamics, and from there deriving meaningful analytical constructions, seems like the right approach.

59

Graph 2.4. Caracas Metropolitan Area and periphery, rates of homicides (2013–2017) and death caused by firearms (1999–2013)

Graph 2.5. Venezuela. Percentage of homicides and deaths caused by firearms and homicides according to geographic area, 1999, 2005, 2013, 2017. Source: MPPS, MPPRIJP, author’s calculations.

The Diversification of Venezuela’s Violence

Figure 2.2. Caracas and its periphery, deaths caused by firearms in 2013. Source: MPPRIJP, CICPC, author’s calculations.

Figure 2.3. Caracas and its periphery, homicide rates, 2017. Source: MPPRIJP, CICPC, author’s calculations.

Growth in the Peripheries Caracas is often mentioned as one of the most violent cities on the planet given its high homicide rates.10 Thus, we have an image of it as the place with the highest incidence of violent deaths, in particular deaths caused by firearms. Today however, the capitol’s five municipalities are not the ones with highest rates of deaths caused by firearms. I concentrate on this specific aspect, as it is one of the central aspects for trying to understand the growing heterogeneity of violence. The relative de-

61

62

José Luis Fernández-Shaw Guerra

Graph 2.6. State of Bolívar, rates of homicides and deaths caused by firearms according to type of municipality. Sources: MPPS, MPPRIJP, OVS, author’s calculations.

crease in importance of urban areas and intensification of violence at the periphery appears to be a phenomenon that is showing up in several places in the country. The most emblematic case exhibiting this trend is the Metropolitan Area of Caracas, where levels of deaths caused by firearms seem to have remained stable (although high) compared with the periphery areas of Barlovento and Valles del Tuy, where a sharp increase is visible. The images in figures 2.2 and 2.3, corresponding to the years 2013 and 2017, show the city of Caracas and surrounding areas demarcated by their perimeters. They show changes in the rates of deaths caused by firearms and of homicides for the municipalities in each region.11 In 1999 the rate of deaths caused by firearms (60.4 per 100,000 inhabitants) in the Metropolitan Area of Caracas (MAC) was higher than that of its surrounding areas, as can be seen in graph 2.4. In recent years the murder rates of areas such as Barlovento, Valles del Tuy, and even Guarenas/Guatire are far higher than even the highest rates in Caracas. By 2017 the murder rate of Barlovento (155.7 for every 100,000 inhabitants) was triple the rate of the MAC (54.2). Thus, violence has lost its predominantly urban, capital-centered character. This emergent phenomenon opens several questions about the design of citizen security policies. Are the governments of regional municipalities, with fewer resources than the metropolitan municipalities, prepared to face the emergency prompted by high levels of violence? In specifically nonurban contexts, such as the area of Barlovento, are there security agencies trained and capable of facing this situation with due respect for human rights?

The Diversification of Venezuela’s Violence

Graph 2.7. State of Trujillo. Municipalities Andrés Bello and La Ceiba. Rates of homicides and deaths caused by firearms. Sources: MPPS, MPPRIJP, OVS, author’s calculations.

This displacement of the MAC as the center of the most violent deaths (of which deaths caused by firearms and homicides are an indicator) does not mean that levels of violence have been reduced in the city to tolerable levels, much less that they have decreased throughout the city. Violent deaths inside Caracas can be as heterogeneous as in the rest of the country. The MAC is constituted by sectors with very important variations in their levels of violence. A first general delimitation of the city reveals that the traditional divisions by municipalities are not enough. Some of these municipalities show very similar characteristics, as for example so-called Eastern Caracas formed by Baruta, Chacao, and El Hatillo; on the other hand, Libertador is marked by considerable diversity,12 with some areas having levels of violence similar to the Sucre municipality of the state of Miranda, considered by some as the most violent in the city. At this point, I should make explicit the rationale underlying the methodological delimitation of geo-social spaces in Venezuela (Grusón 2008; Roche 2013). Behind the territorialization of violence there is a differentiated structure of life conditions and opportunities whereby the geography ends up not being a simple homogeneous grid where the interplay of actors connected to illicit economies disseminate or carry violence from one space to another based simply on an operative logic. If that were the case, expansion from any one space to another would be equally likely. To the contrary, different spaces—whether they are municipalities, parishes, or communities—are characterized by specific forms of material life and social stratification that provide a particular

63

64

José Luis Fernández-Shaw Guerra

Figure 2.4. East coast of Venezuela. State of Sucre. Rate of deaths caused by firearms, 1999, 2005, and 2013. Sources: MPPS, author’s calculations.

configuration in which there are greater or lesser possibilities for the expansion of violence. These conditions preexist the expansion of violence and condition its development.

Rent Extraction in a Fragile Institutional Context Recent research carried out in the area called the Venezuelan Mining Arch 13 found a significant increase in deaths caused by firearms since 2008. This rise runs parallel to increases in the price of gold in international markets and to internal migration flows linked to the mineral’s extraction. These factors trigger confrontations for rent-seeking control of the area. Control ends up in the hands of various groups and parastate organizations such as guerrillas, syndicates, armed gangs, and other criminal organizations. This is a territory where the control held

The Diversification of Venezuela’s Violence

Figure 2.5. State of Sucre, homicide rates, 2017. Sources: MPPRIJP, OVS, OVV, author’s calculations.

by official security forces is precarious, and where state officials are at times part of the interest networks at play.14 Battles over the control of profits—in a territory with little formal control by the state and little capacity to manage conflict—shape a setting where violence is repeatedly used as a conflict-solving mechanism by the different groups in the areas, and as a tool for ensuring the continuation of the gold mining business, which is for the most part irregular and informal. High gold prices in the international market (as well as the weakening of the local currency) generate substantial profits that allow these groups to acquire equipment (vehicles, weapons, communication equipment) and recruit young people, enabling operation and control in the area. Researchers (Ruiz 2018) have described the situation in the area this way: “This panorama blurs the fine lines of a very complex network of power relationships among the actors in a territory with very weakened institutional structure and which mismanages the needs of the population. Thus, institutions, transnational companies, large and small businesses, brokers of minerals, machinery owners, legal and illegal commerce retailers, traditional miners, and the ‘order’ imposed by the mining pranato,15 among other actors and processes, strongly impact the established inertia of the process.” Recent homicide data16 show that the dynamics of armed violence17 described here has intensified in the last years. In recent years, goldmining municipalities had higher homicide rates than the urban municipalities of the state of Bolívar—more than 200 deaths for every 100,000

65

66

José Luis Fernández-Shaw Guerra

Graph 2.8. State of Sucre. Rates of homicides and death caused by firearms, Urban Areas and Paria Peninsula

inhabitants. Within the area, the municipality of El Callao is one of the areas with the highest rates of violent deaths in the country.18

Borders and Drug Trafficking Routes Recent research has revealed evidence of the link between the increase in violence in Venezuela and the role played by the country as a drug trafficking route to other destinations (Kronick 2019). This configuration is particularly evident after 1989, when “counternarcotics policy in neighboring Colombia increased the use of Venezuelan transport routes” (Kronick 2020, 1499). We see an illustration of this hypothesis in the context of the Paria Peninsula. Figure 2.4 shows rates of deaths caused by firearms in the state of Sucre at three different moments: 1999, 2005, and 2013. It reveals that the incidence of deaths caused by firearms in the Paria Peninsula is increasing. One of the most accepted hypotheses (based on numerous journalistic reports) is that the area is under the growing influence of drug trafficking to Trinidad.19 Paria has traditionally been an area of irregular commerce with Trinidad, but drug trafficking has changed the nature of illicit commerce. This is a good example of how violence has decreased in urban centers but moved to smaller towns. Note the evolution of the rates of death by firearms and homicides in the state of Sucre. On the one hand, I have aggregated the municipalities with the main urban centers, Sucre (the city of Cumaná, with an estimated population of 430,539) and Bermúdez (the city of Carú-

The Diversification of Venezuela’s Violence

pano, population 160,740). These two municipalities conform the urban sphere of Sucre. On the other, the municipalities of the Paria Peninsula, Arismendi (Río Caribe), Cajigal (Yaguaraparo), Mariño (Irapa), and Valdez (Güiria) have been grouped. These have an added population of 151,424. From 2008 on, violence decreases in urban areas, but continues to increase in the peninsula, although with a small decrease in the last registered year.

Violence Outside of the Urban Context Violence is not only present in places where there is an expectation of a high rent-seeking economy (for example, the Mining Arch), or in the context of drug trafficking routes or other illicit economies (Paria Peninsula), it is also present in territories with more traditional and local economies. It is even possible to appreciate a sudden expansion of lethal violence in the municipalities of the state of Trujillo that border the state of Zulia and Lake Maracaibo. More specifically, we see this expansion in the area comprising the municipalities of the state of Trujillo, including Andrés Bello, Bolívar, La Ceiba, Miranda, Motatán, Sucre, and the Baralt municipalities of the state of Zulia. This traditionally agricultural area of banana and papaya crops has one of the highest rates of violent deaths in the country. In particular, the Andrés Bello municipality (capital: Santa Isabel), Las Ceiba (Santa Apolonia), and Miranda (El Dividive), show rates exceeding 100 deaths for every 100,000 inhabitants; these three are among the ten municipalities with the most violence in the country. Note the increase in violence in the area since 2008, and for 2014 note a strong increase in the homicide rate, which turned this area into one of the most dangerous in the country. Journalistic sources point to the presence of many criminal gangs in the area, 20 who battle each other for territorial control. Some of these confrontations have resulted in dozens of deaths. These gangs are manned by local youth who regularly carry firearms. This is evidence of something that is highlighted in various chapters of this book: the struggle for control of territory is a spark for violence. It is not simply the presence of illicit economies, but the dispute among different actors for control of them that leads to violence. What is particularly revealing is that in relatively small towns (the municipalities included have around twenty thousand inhabitants) the incorporation of young people into armed gangs reaches such a high proportion that there are up to six groups per municipality, each of them including dozens of members. This magnitude may be linked to the lack of structural opportunities

67

68

José Luis Fernández-Shaw Guerra

within an economically depressed context.21 In addition to the presence of groups linked to illicit economies, a key aspect is the availability of firearms, which functions as an articulating element for the activities of these groups.

The Diversification of Venezuela’s Violence As should be evident from this brief discussion, Venezuela’s violence is a diverse phenomenon whose configuration changes according to multiple contextual factors. The following are some preliminary conclusions. First, the country is facing an increase in violence in nontraditional spaces outside of major urban centers. This fact has implications for citizen security policy and, without a doubt, for the configuration and operations of the agencies in charge of it, specifically those acting in local contexts which tend to be weaker. It should also provoke review of the actions of other forces in rural areas. They are usually policed by the National Guard, or even the army. The officers of these forces rarely have training on issues of citizen security, violence prevention, and human rights.22 Second, one common denominator in the cases reviewed here is the presence of firearms as something that contributes to the formation of groups with the capacity to compete for territorial dominance. The ease with which these groups have access to armed resources gives them the possibility to grow and develop. This fact highlights the need to update the effort to define and execute a policy controlling arms and ammunition. Third, high levels of violence seem to be generally linked to forms of illicit economies, which can be associated to specific types of goods and have different magnitudes and characteristics. Here, we have reviewed several of these factors, such as the exploitation of valuable minerals that generate high rents, drug trafficking on border areas, extortion of traditional farmers, the transportation sector and roadways. Finally, these different forms of illicit economies, mobilized by armed groups that struggle for territorial control, create a situation where the recruitment of young people, particularly men, becomes an object in itself, as they are these gangs’ main assets. It is important to emphasize that one of the key strategic focus in the prevention of violence should be the creation of development opportunities for young people that provide them with alternatives to their inclusion into armed groups and illicit economies.

The Diversification of Venezuela’s Violence

List of variables incorporated in regressions Rates of death caused by firearms Total population (total) Percent of rural population (rural) Percent of poor homes (NBI) Percent of homes in extreme poverty (NBI) Percent of non-poor homes (NBI) Percent of homes with education deficiencies (DefEduP) Percent of population between three and six years old that attend school Percent female population between three and six years old that attend schools Percent male population between three and six years old that attend schools Percent of population between seven and twelve years old that attend school Percent female population between seven and twelve years old that attend school Percent male population between seven and twelve years old that attend school Percent of population between thirteen and seventeen years old that attend school Percent female population between thirteen and seventeen years old that attend school Percent of male population between thirteen and seventeen years old that attend school Percent of population between eighteen and twenty-four years old that attend school Percent female population between eighteen and twenty-four years old that attend school Percent male population between eighteen and twenty-four years old that attend school Percent of homes lacking quality housing Percent of homes with deficit of public services Percent of homes with trash pickup.

69

Part II Causal Processes and Cycles of Violence

3

Violent Crime in Venezuela Economics and Institutions Josefina Bruni Celli and Javier Rodríguez

Homicide statistics have been used to describe Latin America as one of the most violent regions in the world (Briceño-León and PérezPerdomo 2000; Briceño-León 2006; Di Tella et al. 2010; WHO 2002), and violent crime has been identified as an important impairment to development in the continent (Londoño et. al 2000). Consequently, a large amount of research has been devoted to the study of the causes of violent crime in that region. Most of it has taken on a sociological perspective (Aguirre 2000; Bourguignon et al. 2003; Rotker 2002; Londoño et al. 2000). Much less has been said about the causes of violent crime from an economic perspective. A book edited by Rafael Di Tella et al. (2010) has added to knowledge from perspective of economics of crime in Latin America, by offering chapters addressing the effects of deterrent policies—enforcement and punishment—and socioeconomic factors on violent crime (Soares and Naritomi 2010; De Mello and Schneider 2010). But information provided in this book confirms that there is much room left for rigorous inferential analysis with country-level data that has yet to be accessed and processed by Latin American researchers. Furthermore, nothing has been said, from an economic perspective, about the relationship between Latin America´s most important economic activity—the extractive industry—institutional capacity, and criminal violence. This 73

74

Josefina Bruni Celli and Javier Rodríguez

chapter seeks to address these two voids in the literature by exploring the case of Venezuela, a country featuring high levels of criminal violence and an important extractive industry—namely, oil. The dataset used in this study, covering the period 1970–2013, represents a multi-year effort in data hunting and harmonization of dispersed data, which the researchers needed for the purpose of rigorously analyzing violent crime in Venezuela from the perspective of the economics of crime. Violent crime may be generally defined as the use of physical force or threat of physical force upon a victim. According to Morgan Kelly (2000, 535, 536), “Economic theory of crime is uncomfortable with violent crime,” for “why should individuals hurt other for no financial return unless they have some taste for violence?” We don’t agree with this position because even though certain forms of violent crime may be motivated by personal passions or psychological disturbance, in Latin American countries where civil war is not being waged, violent crime tends to be instrumental, and at the minimum, indirectly motivated by economic gain. Statistics in Venezuela indicate that the most extreme form of violent crime, homicidal crime, occurs in the of context turf wars, kidnapping, robbery, drug traffic, and hired settling of scores, all of which are instrumental in nature involving high economic stakes or returns. In chapter 7 Luis Gerardo Gabaldón notes that a 2012 evaluation of violent crime by Venezuela’s General Police Council found that 45 percent of violent crime was linked to robbery. Thus, we contend and found this study on the premise that the utility-logic of Becker’s (1968) economics of crime framework is relevant for the study of violent crime in Latin America.1 This study only analyzes homicidal crime. Homicidal crime has two characteristics that make it a reliable indicator: it is subject to a high rate of reporting, and it is the only type of crime that is systematically recorded by hospitals, morgues, and the police. Although authors like Roberto Briceño-León and Rogelio Pérez-Perdomo (1999) indicate that there is a high correlation between homicidal crime and other forms of violent crime, there is no overwhelming evidence in the literature of a correlation between homicide and common crime. In this sense, the hypotheses discussed in this chapter should not be extended to other forms of crime than homicide.

Economics of Crime Framework In Crime and Punishment: An Economic Approach, Gary Becker (1968) proposes that a person (assumed to be a rational actor) decides to commit a crime if the expected utility is greater than the utility that could be ob-

Violent Crime in Venezuela

tained through the use of time and the application of resources to other activities. According to Becker, the decision to commit a crime (Oj) at a given moment is a function of the probability of conviction (Pj), of the severity of punishment if convicted (Fj), and of a variable (Uj) representing all additional factors that could influence the decision to commit a crime: Oj = (Pj; Fj; Uj) One of the key features of this framework is that it views crime as a type of work; that is, criminal activity is an alternative to legal work. To that extent, crime is assumed to be led by an instrumental motive. This basic model has been developed and elaborated upon in a large body of literature (see Ehrlich 2010). It is important to point out that Becker was principally concerned with the influence of Pj and Fj, as he was seeking to develop criteria for formulating optimal public policy determining levels of Pj and Fj. Thus, Becker left the field wide open for the study of Uj. Concern with Pj and Fj has been taken up over and over again in the literature (see three decades of literature reviews in Cameron 1988; Nagin 1998; Ehrlich 2010), where the negative relation between crime and law enforcement activity has generally been confirmed. A portion of the body of empirical research has been especially concerned with the problem of measurement (Levitt 1998) and/or the problem of endogeneity emerging from simultaneity between law enforcement activity and levels of crime (Ehrlich 1973; Levitt 1996, 1997; Lin 2009). Other research has focused on establishing the differential effects of various forms of deterrence, and the effectiveness of deterrence relative to incapacitation (Ehrlich 1975; Levitt 1998; Drago et al. 2009; Paternoster 2010; Zimmerman 2010). There exists a vast literature developing Uj as the set of socioeconomic factors determining the decision to commit a crime (Buonanno 2003). This development derives from Belton Fleisher’s (1966) and Gary Becker’s (1968) argument that poverty or low income creates an incentive to enter criminal activities because it affects the cost-benefit ratio faced by the individual: “The principal theoretical reason for believing that low income increases the tendency to commit crime is that it raises the relative cost of engaging in legitimate activity” (Fleisher 1966, 120). Fleisher’s (1966, 120) specific reasoning is as follows: “The probable cost of getting caught is relatively low, since they (low-income individuals) view their legitimate lifetime earnings prospects dismally they may expect to lose relatively little earning potential by acquiring criminal records; furthermore, if legitimate earnings are low, the opportunity cost of time actually spent in delinquent activity, or in jail, is also low.”

75

76

Josefina Bruni Celli and Javier Rodríguez

Based on the aforementioned reasoning, the literature has sought to establish the relationship between crime and indicators of levels and distribution of income. Isaac Ehrlich (1973) tested the relation in a study of American families, and found a significant negative relation between levels of family income and crime, and also a significant positive relation between inequality and crime (states with higher inequality show greater crime). Inequality, according to Ehrlich, leads to crime through a twofold process: (1) it raises the opportunities for the richer to be potential victims of crime, and (2) it lowers the opportunity cost of the poorer to engage in crime. Morgan Kelly (2000) confirms this relation between inequality and crime in a later study using data from North American urban counties. More recent empirical studies confirm the existence of a relationship between levels of income and crime, and between inequality and crime. In a worldwide study, Pablo Fajnzylber et al. (2002) found a significant negative relation between economic growth and violent crime, as well as a significant positive relation between inequality and violent crime. Paolo Buonanno and Daniel Montolio (2008) show a negative and significant relation between gross national product (GNP) per capita and crime, and also between rate of change in GNP and crime in Spain. Luiz (2001) found similar results between levels of income and crime in South Africa, using average family income. The literature has also seen unemployment as a possible indicator of economic distress and lack of opportunity in legal economic activity, and thus as a possible predictor of crime. Classical empirical studies by Fleisher (1963) and Ehrlich (1973) in the United States, as well as more recent studies by Chor Foon Tang (2011) in Malaysia, and Buonanno and Montolio (2008) in Spain, point to a significant relation between unemployment and crime. Another socioeconomic variable considered in the literature is level of education. The general argument is that education increases return to engaging in legitimate work and raises the opportunity cost of engaging in illegal activities (Becker 1968). The “civilization effect” of education has also been put forward (Usher 1997). Buonanno and Montolio (2008) and Kausik Chaudhuri et al. (2015) found a negative relation between education and violent crime in Spain and India, respectively. Nonetheless, in their global study of violent crime, Fajnzylber et al. (2002) did not find a significant relation. Youth and crime have also been repeatedly associated in the literature (Freeman 1991; Grogger 1998; Waiselfisz 2007; de Mello and Schneider 2010), leading to the prediction—and finding—that crime rates are higher where the proportion of male youth is higher in a population.

Violent Crime in Venezuela

Economics of crime reasoning relating the two include historical drop in real earnings and earning opportunities for unskilled youth, making it more profitable for them to engage in criminal activity and increasing the opportunity cost of engaging in legal activity, and the young are less risk averse than adults. In sum, the extant literature on the economics of crime has fundamentally concentrated on two sets of factors: repression or punishment (Becker’s Pj and Fj) and socioeconomic factors. Youth has also been treated as Uj, but only inasmuch as unskilled youth seems to be suffering a historical drop in earnings and earning opportunities. Not much research has been done in Latin America with regards to the relationship between the above variables and violent crime. A paper by Rodrigo Soares and Joana Naritomi (2010) is unique in exploring in a very preliminary way the relation between the above two central categories of variables and homicide in seven Latin American countries. Using national average statistics from 2010 for homicide, income statistics, youth statistics, and deterrence indicators, they analyze patterns and put out the hypothesis that low economic performance and inequality, presence of youth, and low levels of repression (“timid repression”) explain the high levels of violent crime in Latin America. This is what we set out to verify with more abundant data and inferential methods for the case of Venezuela. It is important to point out that in contrast to Becker (1969), Ehrlich (1973, 2010), other researchers from industrialized countries, and researchers concerned with punishment in less developed countries, including Latin America, focus more on the problem of institutional capacity or institutional weakness than on optimal policy decisions. This is why in addition to considering indicators of government investment in law enforcement activity (number of police, number of judges, rate of imprisonment), in their analysis of violence in Latin America, Soares and Naritomi (2010) address the capacity of government to combat crime with a Rule of Law Index, 2 which reflects the strength of state institutions.

Methodological Note In this chapter, violent death rates are the focus of analysis, and no distinction is made between violent deaths caused by robbery, burglary, kidnaping, extortion and others. The 2006 Encuesta National de Victimización (National Survey of Victimization) shows that crimes are grossly underreported by victims (Gabaldón 2006). According to that survey, among the reasons for underreporting were distrust of the police, and the belief that the police would do nothing if the victimization were

77

78

Josefina Bruni Celli and Javier Rodríguez

reported (Gabaldón, Benavides and Parra 2007; Encuesta Nacional de Victimización 2006). Though results of the Encuesta National de Victimización are the most valid, we were unable to use these numbers, because the survey was administered only in 2006 and 2009, making it impossible to use is in time series analysis. Homicide rates have the advantage that the number is based on administrative records of deaths, which are quite complete, and thus best reflect the level of violence that is present in the country. The homicide data used in this study corresponds to a dataset that was further corrected by Josbelk González Mejias and Dorothy Kronick for biases in the use of the label homicide by the Venezuelan government (see chapter 2); González and Kronick define their variable as violent death rate. The sources, descriptions and time period of each socioeconomic variable that was used in estimations are detailed in the appendix. In order to examine the relation between socioeconomic variables and violent death rates, and its consistency with theory, individual estimations were made as follows: Ht = β0 + β1Rt + εt (2) Where: H: Violent death rate R: Socioeconomic variables (GDP per capita, growth rate per capita, inequality, unemployment, informality, and Rule of Law Index) e: Error term t: Time in years In joint estimations—that is, those estimations that include more than two socioeconomic explanatory variables in (2), serial correlation was detected with the least squares method (LSM), which complicates statistical inference. To solve this, estimations were made with feasible generalized least squares (FGLS), using a Prais-Winsten transformation. In contrast with other methods, such as Cochrane-Orcutt, FGLS estimations with a Prais-Winsten transformation has the advantage that it does not lose the first observation in the iteration process. This is an advantage when the time series is relatively short (Wooldridge 2010).

Socioeconomic Variables and Violent Crime in Venezuela Socioeconomic conditions in Latin America, especially inequality, are commonly invoked to explain criminal violence in the region. In this section we analyze the effects of socioeconomic conditions (Uj) on vi-

79

Violent Crime in Venezuela

Table 3.1. Individual regressions of violent crime on socioeconomic variables, Venezuela, 1970–2013 Variables

Violent death rate per 100,000

GDP per capita (1997)

-0.003

Growth rate (change in real GDP per capita)

0.312

Inequality (Gini) Unemployment rate

(-0.14)

(0.52) -249.8043** (-2.97) 2.38366** (3.62)

R-squared

Observations

Years

Theory

0.0006

44

1970–2013

Contradicts theory

0.0068

44

1970–2013

Contradicts theory

0.1552

43

1970–2012

Contradicts theory

0.1368

44

1970–2013

Consistent with theory

t statistics in parentheses; *** p