ELECTORAL COMPETITION AND CRIMINAL VIOLENCE IN ITALY ( )

Transcription

1 ELECTORAL COMPETITION AND CRIMINAL VIOLENCE IN ITALY ( ) Salvatore Sberna Istituto Italiano di Science Umane - Florence salvatore.sberna@sumitalia.it Paper presented at the ECPR Joint Session Conference Workshop on Political Institutions and Conflict St. Gallen ABSTRACT Do criminal organizations use strategically violence for electoral purposes? This research aims at analyzing the relation between criminal violence and elections in southern Italy ( ) where four regionally-based organized crime networks operate. In this study, criminal-electoral violence is defined as any organized act or threat by criminal organizations that occur during an electoral process, from the date of nomination for political offices to the date of elections, to intimidate, physically harm, blackmail, or abuse a political stakeholder in seeking to influence directly or indirectly an electoral process. The empirical analysis is drawn from a unique panel data of monthly crimes (incendiary and explosive attacks) reported by police forces in 105 Italian provinces from 1983 to 2003 (Minister of Interior - SDI data set). Through a diff-in-diffs design, the paper finds statistical evidence that there is a positive correlation between mafia violence and elections, which means that as elections approach intimidation attacks increase in Southern Italy. These findings are consistent with a large case study literature documenting the interventions of criminal organizations into the electoral process in southern Italy. All the evidence indicates that criminal groups used a wide variety of strategies to make sure that their preferred candidates got elected. 1

2 1. INTRODUCTION Elections are crucial in democracy and when they are perceived as unfair, unresponsive, or corrupt, democratic legitimacy is compromised (Fisher, 2002). Therefore, it represents a serious threat if criminal resources are absorbed in electoral process than if they were exclusively devoted to illegal activities. Numerous studies on political violence and terrorism show that terrorist groups (Eubank and Weinberg, 2001; Pape, 2003; Kydd and Walter; 2002, 2006; Chenoweth, 2010; Aksoy, 2010) or ethnic groups (Cohen, 1997; Collier, 2009) use strategically violence around elections times, or that political violence can be manipulated by incumbents according to the competitiveness of elections (Wilkinson, 2004; Collier and Vicente, 2009; Acemoglu & al., 2009). However, no empirical works exist on criminalelectoral violence, i.e. that violence employed strategically by criminal organizations as elections approach. This paper shows that electoral timing explains variation both in politics and in crime (Sanchez and Chacon, 2005). This is true not for all types of crime, but especially for those criminal phenomena which are organized and institutionalized over time in specific areas 1. In the case of Italy, many qualitative studies and media report about criminal violence during electoral campaigns but there is no cross-regional empirical exploration of the relationship between the timing of elections and such violence (Gambetta, 1993; Della Porta and Vannucci, 1999). What is the relationship between election times and violence in those countries where strong institutionalized criminal groups operate? Why is violence preferred to non-coercive means to influence electoral competition? Few empirical studies have dealt with this urgent issue due the lack of data and secrecy and the reliability of the materials collected by law enforcement agencies. This paper seeks to overcome these serious methodological problems by focusing on the relationship between elections and criminal violence, arguing that some single points in time (i.e. elections) have an impact on criminal violence. The analysis evaluates this effect in Italy where several criminal groups operate and are commonly clustered in four regional organizations: Camorra in Campania, Sacra Corona Unita in Apulia, Ndrangheta in Calabria and Cosa Nostra in Sicily. Data for the empirical analysis are drawn from a 1 Lack of property rights enforcement or weak monopoly of violence exercised by central government versus its peripheries, together explain the informal institutionalization of criminal enterprises, specialized in providing protection, according to the insightful analysis of Diego Gambetta (1993). This is the case of protection market in Russia lead by the Mafiya (Varese, 2001), or the trouble transitions in some eastern European countries. In Japan, on the contrary, Yakuza criminal groups (Hill, 2003) are more similar to Italian organized crime, both, in fact, operate in established democracies, compared to the environment of some democratic transitions in Latin America, such as in Brazil, or the penetration of drug cartels in Mexico and Colombia. 2

3 unique panel data of monthly crimes (arson and bomb attacks) reported by police forces in Italian provinces from 1983 to 2003 (Minister of Interior - SDI data set). The evidence shows that Italian mafias are most likely to engage in violence in electoral periods, thus as elections approach these organizations strategically use violence and other means in order to influence the electoral process. Approaching elections, I argue, lead to an increase in the volume of attacks in those areas where the presence of criminal organizations is high, while in those areas with lower criminal presence this is not the case. Thus, periods close to elections are periods of heightened criminal activity when areas are controlled by mafias. The logic behind the argument is as follows. In all democracies election times are periods of heightened political competition not only among party candidates and interests groups, but similarly among criminal groups when they are interested in influencing the political process (Aksoy, 2010). Therefore, at first glance, criminal organizations are similar to terrorist groups, but they critically diverge. They share a single tactic the use of violence but criminal organizations have a much wider repertoire of action, including illegal activities, money, and especially collusive ties with political elites (Makarenko, 2004; Sciarrone, 2006; Lupo, 2010). Thus a different strategy moves such action. Unlike ordinary criminal organizations, which are avowedly nonpartisan and have virtually no contact with parties, mafias are structurally integrated within the political systems in which they operate. In fact, although they are industries in illegal markets and driven by profit (Gambetta, 1993), they naturally gravitate toward government, because they seek to influence the direction and content of governmental action to reach organizational goals immunity against law-enforcement, rent-seeking (Harasymiw, 2003). Mafias are, after all, organizations which pay attention to whatever is necessary to the maintenance of the integrity and continuity of the organization itself (Selznick, 1948: 29). Therefore, studying mafia-politics relation means looking at the conditions under which political parties are captured by criminal organized interests (Barro, 1993; Grossman&Helpman, 2001; Golden&Tiwari, 2009; Acemoglu et al., 2009). In exchange of immunity, they can affect elections in many ways by contributing to finance campaigns, or mobilizing voters to provide electoral support to politicians they prefer to favor, or, in extreme cases, being themselves running for elections (Della Porta and Vannucci, 1999). Based on this literature I show that the volume and the timing of organized crime s violence can be explained by the occurrence of elections. The paper is organized as follows. The theoretical setting of the relation between criminal organizations and politics is presented in Section 2. In Section 3 I formulate the hypotheses. Data source and construction are explained in Section 4. Descriptive analysis of data about organized and criminal violence is presented in Section 5. Then the empirical strategy is designed in Section 6, followed by the presentations of main results. 3

4 2. ORGANIZED CRIME, ELECTIONS AND PARTIES IN ITALY The literature on organized crime provides compelling evidence that criminal organizations are directly involved in electoral campaigns both in consolidated democracies (Allum & Siebert, 2003) and many democracies in transition that experience a dramatic rise in reported criminality and citizen insecurity, such as in Latin America (Bailey & Godson, 2000; Bergman & Whitehead, 2009). A large case study literature has emerged, documenting such interventions. Interaction and exchanged resources between politicians and criminals can vary significantly from electoral finance to voter mobilization, or finally violence in order to intimidate opposition candidates, voters, or criminal rivals. Intimidation is used also to keep opposition parties representatives away from the polls or simply threatening them (Sberna 2010) 2, or as a mean of electoral propaganda in public markets. More recently a growing literature illustrates theoretically the link between crime and politics (Pimental, 2000; Vannucci & Della Porta, 2010), and some empirical studies evaluate the likelihood that criminals can be selected by national parties to run for elections and then to get elected (Golden & Tiwari, 2009). Therefore, what emerges is that organized crime does not attempt to displace the state, even though it is at war with it. It exists side-by-side with the state, in a relationship variously referred to as complementary or symbiotic (Armao, 2003). This relation is based on mutual interests, both of criminal organizations and politicians. Politicians are not venal but captured, and they are captured because they want to win the next election. In fact, even in democratic regimes, elected politicians devise strategies to gain unanticipated control over votes (Kitschelt, 2000). They need an intimate knowledge of the voter, and for this reason they often hold political machines which allow them an electoral advantage (Golden and Tiwari, 2009). The presence of organized crime opens alternative strategies normally not available to candidates in democratic elections, including intimidation and violence, criminal vote buying, fraud. Menu of electoral manipulation (Schedler, 2002) consists of several alternatives for controlling voters, more than the mere use of intimidation and force. Through these means criminal organizations help favored politicians in retaining political support by voters. Therefore, parties relation with organized crime is based on the premise that the latter can control citizens voting behavior. Looking at organized crime, interests are clear: criminal organizations naturally seek immunity from law-enforcement agencies in order to protect their illegal business. However, greed of immunity tells us only a very limited part of the story. More often these organizations control both illegal markets (protection-extortion, drug) and some specific sectors in legal 2 In the municipality of Seminara in Calabria, for example, the local boss decided to confiscate some voters electoral cards, giving them back few hours before poll closed, and providing to illiterate voters of the so called stampino, which is a mould with the name of favored candidate stamped on it (Sberna 2010). 4

5 markets thanks to money laundering (Lavezzi, 2008). In construction, concrete and waste sectors, the control is often overwhelming and quantitatively hegemonic, but also highly dependent on public investments and regulations. Therefore, mafias naturally gravitate around government because they have preferences over public policies, and this element reduces the incentives of the politicians they favor to disrupt criminal networks 3. This explains why mafia-politics equilibriums likely persist along the time (Acemoglu et al., 2009). Italian mafias have been structurally integrated within the Italian political system despite the regime changes that occurred in the country since the last century. Such relations were by no means contingent but regular and systemic. Police and historical records present exhaustive evidence convincingly demonstrating the persistence of a close link between organized crime and politics from the Eighteen century (Franchetti, 1900; Lupo, 2010). During the First Republic ( ) the systemic network of collusion and the special relationship with the Christian Democracy party (DC) was explained by several factors: among all the lasting one-party dominance of DC in almost all legislative districts and in the majority of municipalities in Southern Italy 4 (LaPalombara, 1964; Leonardi and Wertman, 1989; D.Scheiner, 2006; Magaloni, 2006). Even though there was remarkable evidence about collusion, accusations were serious enough to lead to prosecutions and charges. The analysis of voting behavior in some areas exhibited some un-explainable results, such as massive changes in voting patterns from a candidate to another and very high concentrations of votes for some candidates in particular municipalities. The turning point in mafia-parties relation is widely acknowledged to occur in early 1990s. Italy s 1994 national elections were the most volatile elections in the history of the Republic (41,9% Gunther, 2005:255) and marked the end of political dominance by Christian Democracy, broken down by Clean Hands campaign. However, opportunities and constraints of this interaction had already changed in the 1980s, when judiciary investigations uncovered and permitted the arrest of many top bosses (Jamienson, 1999). The heightened intra-bloc volatility favored DC s coalition allies (Partito Socialista Italiano and Partito Repubblicano) to the detriment of DC s electoral support (Gunther, 2005). At local level, party system fragmentation eroded DC s dominance in criminal-political ties with criminal groups that in many cases preferred to colonize DC s coalition parties at local level (Mete, 2009; Sberna, 2010) 5. 3 Evidence of a criminal-political capture can be found in those areas where due the penetration of criminal organizations citizens obtain fewer public goods (and other policies they value), because policies cater to the preferences of criminal families. 4 The DC was the largest party in every coalition government between 1948 and 1992 and, except for and , controlled the prime ministership every year until 1993 (Leonardi, 1993). At regional and municipal level, the DC was also the largest party in local cabinets. 5 In southern Italy many municipalities have been ruled for a long time by criminal families members, and sometimes through a tough transition from father to son, such as the city of Quindici in Campania, where 5

6 However, Italy is not the only case where crime-politics connections have been so tight that criminal groups have colonized parties or been directly involved in electoral campaigns since they are themselves running elections. Pablo Escobar in Colombia is probably the most popular case of drug carteler in politics and it was not an exception. Therefore, the evaluation of elections impact on criminal violence is the first step in the empirical study of politics-crime nexus. In fact, I argue that the analysis of politics of crime can tell us something more about the criminalization of politics even in established democracies. 3. ELECTIONS AND VIOLENCE. HYPOTHESES Criminal organizations use violence in a myriad of ways and for several motives both in legal and illegal markets. Violence clearly arises from producing illegality 6, and it is exploited instrumentally to pursue diverse goals, such as to resolve disputes among mafioso families, or to enforce the monopoly in illegal markets (Schelling, 1967). However this paper is not interested in explaining all several motives, goals and strategies of mafias violence. I look at a specific type of criminal violence, that I define criminal-electoral violence, i.e. any organized act or threat by criminal organizations that occur during an electoral process, from the date of nomination for political offices to the date of elections, to intimidate, physically harm, blackmail, or abuse a political stakeholder in seeking to influence directly or indirectly an electoral process (Fisher, 2002). Therefore, the strategic timing differentiates this violence from other examples. Election times are crucial for at least two reasons. (1) In democracy, elections increase uncertainty in political equilibriums. Around election times many political groups try to influence the political process and they increase the volume of their activities. Political parties seek contributions and resources for campaigns while interest groups increase their lobbying activities providing such resources. Criminal organizations action is also driven by this logic. If this is true, we should observe an increase in the number of intimidation attacks as elections are close since even criminal groups fiercely compete. Homicides or intimidation among criminals in election times are demonstration of the willingness to use violence and of power both to legal and illegal actors. By those acts they can indirectly enforce other illegal deals and their control upon voters and candidates. several bosses of Graziano clan, part of the New Organized Camorra, inherited the office of mayor for more than 30 years. It was not by chance that in the 80s in Quindici the PSI gained 100% of the seats of the city council. 6 First, the criminal victims of violence are disadvantaged in seeking police protection. And this is obvious since the process of providing an informative complaint will yield information to the police about the illegal activities of the complainant. Second, it is endogenous of illegal markets the problem that participants in illegal markets lack recourse to state provided facilities for settlement of disputes. Violence or its threat (intimidation) may constitute the only method of resolving disputes in some situations (Reuter, 1983). 6

7 (2) An additional reason makes elections crucial for criminal lobbying. During elections the saliency of organized crime as policy issue raises inevitably, because electoral campaigns can draw public attention to mafia-politics collusion. Lack of debate about crime is the first goal criminal groups seek to reach, and this is why they would intimidate political candidates. These informative asymmetries between political-criminal networks and voters can be put at risk if opposition parties candidate denounce the existence of such links, or simply move suspicions about people involved. Therefore public attention can lead to criminal violence. Drawing on these propositions, I formulate synthetically the main hypotheses of this study: the first group proposes an explanation about the variation in criminal violence between electoral and non-electoral periods; the last one is about the variation depending on the nature of elections. Hypothesis 1a. In those areas controlled by organized crime intimidation attacks increase when elections are close. According to this formulation, I argue that elections have an exogenous effect on criminal-electoral violence. I would expect an overall increase of intimidation attacks before elections in those areas controlled by organized crime. The null hypothesis is that criminal violence does not vary significantly as elections approach if the area is not controlled by criminal organizations. Many empirical studies about political violence in electoral times have already tested a similar hypothesis (Wilkinson, 2004; Collier and Vicente, Aksoy, 2010). In literature some theoretical studies show that the timing of terrorist attacks is not random, but strategically decided (Kydd and Walter, 2002, 2006). This paper goes further these contributions, in which electoral violence remains confined to developing democracies, or to countries highly fragmented along ethnic lines. I do not refer to spontaneous or terrorist violence, but to that violence employed by criminal organizations. Doing it, this paper is similar to some recent empirical analyses about the impact of criminal/guerilla violence in Colombia. These studies investigate the electoral effect of guerrilla in a sample of municipalities, and also the willingness of federal government in disrupting guerilla networks (Sánchez & Chacón, 2005), and thus lower the intensity of non-state violence (Acemoglu et al., 2009). However, the existing related literature shows that approaching elections do not unconditionally increase violent attacks from terrorist or ethnic groups, other institutional variables can intervene: the permissiveness of electoral system (Aksoy, 2010); the competitiveness of elections (Wilkinson, 2004); cost-benefit calculations depending on being incumbent or challenger (Collier and Vicente, 2009) 7. The nature of elections (national or local) can also 7 In a recent study about elections in Nigeria, Collier and Vicente (2009) come to this result through a field experiment. They found that voter intimidation is effective in reducing voter turnout, and that violence was systematically dissociated from incumbents, contradicting Wilkinson (2004) s results. Interestingly, they established that incumbents have a comparative advantage in alternative strategies as vote buying and ballot fraud, because they control both the electoral process and state resources. 7

8 have an impact as some recent studies suggest. According to the econometric results in Sánchez & Chacón (2005), decentralization process in Colombia has been negatively correlated with the increase of guerilla violence, since they found a strong connection between the intensification of armed activities and local governments greater political independence and fiscal strength. This finding is consistent with qualitative evidence in Italy, and the current paper aims at testing empirically whether criminal organizations are more likely to employ violence during local elections. Therefore I formulate a following hypothesis: Hypothesis 2. Intimidation attacks are most likely to increase when local elections are close. This hypothesis helps me testing the argument that criminal organizations deepen their ties with local society (Sciarrone, 1996), because they are basically single and autonomous criminal groups without a centralized governance. Therefore, the influence of local politics is the first goal to be reached by these groups. 4. DATA SOURCE AND CONSTRUCTION The most important data for this study are on criminal-electoral violence. They are provided by the Servizio di Indagine (SDI), under the Italian Minister of Interior. The SDI collects data about crimes reported by the three main Italian police forces (Arma dei Carabinieri, Polizia di Stato, Guardia di Finanza) to the courts. The system aggregates criminal acts in several categories by type of action and crime. The measure of criminal-electoral violence is constructed by simply aggregating two of these variables: arson attacks and incendiary attacks ( incendi dolosi and attacchi dinamitardi e incendiari ). I employ the number of incendiary and explosive attacks as indicator of criminal violence among other types of mafia offences reported by the police. By doing so, I explicitly restrict my analysis to real cases of mafias violence. Other reported offences such as extortion, criminal association, drugs dealing encounter at least two limits: they capture only the effectiveness of anti-mafia programs, and they are not related to electoral process. This is the reason why I decide also to use the offences reported by the police instead of the crimes prosecuted by district tribunals and gathered by the Minister of Justice. I have variable criminal violence for each month in the period 1983 to The uniqueness of this dataset derives from the temporal disaggregation, which permits to exploit temporal variation in violence to better understand its causes and effects. I selected the period 1983 to 2003 for two reasons: 8

9 since 1983 police forces reported separately crimes committed by criminal organizations from ordinary crimes; second, the information system of data collection changed after The sample consists of 103 provinces for the full sample which includes 24 provinces located in 4 southern regions where mafias are deeply rooted in: Campania, Puglia, Calabria and Sicily. In order to increase the number of observations I consider separately the variation in crime both in province principal towns (capoluoghi) and in the rest of the province (which includes all towns in the province except for principal towns). I check the robustness of the results with a more precise measure of criminal organizations presence in the areas computing the number of people arrested for mafia-association in each province and principal towns. In this panel are included those units that have a value of mafia-type association arrests above the 75th percentile. Concerning the reliability of these data, although indicators on crime and delinquency are notoriously fraught with problems, the indicators used in this paper are expected to be well measured. Under-reporting is negligible for arson attacks because we can suppose that they are well reported for insurance purposes, but also because police and fire department necessarily intervene in these circumstances. However, offences reported as arson attacks do not include in their sample those cases of intimidation which are mainly under-reported by victims because they may be unwilling or afraid to report to authorities for a variety of reasons 9. Tactics used by criminals to influence voters, politicians, bureaucrats is much more complex than simple arson attacks. As the qualitative evidences show, mafias intimidate politicians, voters and enemies in a myriad of ways. To summarize, the present source only takes into account what is reported, not what goes unreported. However, information collected by police are a reliable source of criminal violence, thus I can assume that the dependent variable is accurately measured. While temporal disaggregation makes practically feasible the explanation of variation in violence across months, data do not provide information about both victims and offenders of such events 10. First, I do not use victimization data, which means that we don t know the victims of these intimidation attacks. Therefore, the panel includes all violent acts, not only those targeting voters, politicians, bureaucrats. I deal with this inconvenience by coding as electoral only the month in which electoral 8 Under the 165 scheme report, Servizio di indagine e analisi criminale collected data about those crimes reported by three Italian police forces (Polizia di Stato, Arma dei Carabinieri and Guardia di Finanza) in all Italian provinces from 1983 to The system of reporting was not based on a hierarchical rule according to severity of criminal events, but, on the contrary, gather data about specific and single criminal acts and offenses. It included up to 20 separate offenses per incident, without providing elements on victims and offenders. 9 Many examples of intimidation can be not reported to the police, such as when victims receive letters containing bullets, or attacks which do not need the intervention by the police or the firefighters. 10 A further inconvenient is represented by the aggregation over provinces, instead of municipalities. If most violence occurs in a particular town, aggregation to the provincial unit of analysis has the effect of biasing the results in favor of finding a stronger relationship between the explanatory and independent variables. 9

10 process effectively take place and the month prior to elections. Election-related violence is centered around a moment. By isolating it to one month I make more tractable and reliable the measurement of criminal and electoral violence. Second, except for mafioso homicides and mafia-type association crime, I do not get information about offenders, i.e. who actually commit those crimes. Concerning elections, Table 1 shows when regional elections have taken place in Italy from 1983 to 2003 (5 times). Elections are held in different years across regions and it will be used as a source of identifying variation among groups. 5. ORGANIZED CRIME AND VIOLENCE: DESCRIPTIVE ANALYSIS Before proceeding to analyze criminal violence, I do a brief assessment of the criminal presence over time and across regions in Italy. Starting with mafia-association crimes, data show that their distribution varies across provinces even in Southern Italy. Although it is heuristically valid to speak of mafia-type association for the four regionally-based organized crime networks 11, Italian mafia cannot be considered as though it was a uniform and nation-wide entity, or even regionally organized. Several features concerning origins and development deserve attention from a comparative perspective. No single organizational formula is applicable to all Italian mafia groups. Every group has its own criminal formula, degree of institutionalization, strategic objectives, expressed by the different temporal and geographical expansion dynamics. According to the Direzione Nazionale Antimafia, 40 criminal groups are counted in those regions, with ramifications in several provinces of northern Italy and other countries in Europe (Paoli, 1997). Therefore, if we look at the distribution across regions, Southern Italy is the area where criminal organizations mostly operate. Figure 1 illustrates the geographic distribution and evolution of mafiatype association crime and other mafia-related crimes (i.e. extortion and mafia-type homicides) from 1983 to A huge divide exists between the rest of country and four regions in particular: Campania, Apulia, Calabria and Sicily (see Figure 3). Since the inclusion of the mafia-type association crime into the criminal code in 1982, more than 4350 people have been arrested for being a member of a mafioso organization. Data on mafia-type homicides depict again a highly polarized phenomenon. In 11 Among various definitions, I use in this paper the legal one provided by the Italian Criminal Code, relying on the empirical definition adopted by the law-enforcement agencies in reporting mafia related crimes. The criminal code provides a specific definition of mafia-type organized crime as having additional characteristics. According to article 416 bis: the organization is of the mafia type when its components use intimidation, subjection and, consequentially, silence (omertà), to commit crimes, directly or indirectly acquire the management or the control of businesses, concessions, authorizations, public contracts and public services to obtain either unjust profits or advantages for themselves or others. 10

11 twenty years, more than 5570 mafia-related homicides have been committed until 2003, mainly in the same regions. Calabria is the one with the highest per-capita rate (see Figure 2). Figure 5 shows the evolution of mafias homicides and intimidation attacks in twenty years ( ). It reveals that a huge peak in mafias homicides was at the beginning of nineties due mafia wars and the instability created by anti-mafia legislation. In contrast, despite the recent decrease in homicides, intimidation attacks are progressively increasing and counter-intuitively it might prove that these organizations can relay anymore on non-coactive resources in lobbying and illegal activities, but they need to engage in violence in order achieve their criminal goals. Moreover, although the concentration in few regions, the distribution of these crimes varies across provinces even within regions under organized crime s control. Figure 2 and 3 show that some southern provinces have a lower crimes rate, such as in Sicily (Ragusa, Enna) or in Campania (Avellino, Benevento). As we will see in the empirical section, the variation within regions will be useful to identify more homogenized control groups. Among the provinces Figure 6 shows the raw correlation between the number of mafia members per 100,000 inhabitants and provincial levels of violent and non-violent crimes. It appears that while the estimated correlations between mafia indicators and crimes are only mild for non-violent crimes, they are clearly strong and positive for violent crimes. Organized crime s penetration is therefore correlated with the outcome this study seeks to explain. Criminal violence, measured as intimidation attacks, is positively associated with the penetration of organized crime in the same area. This is important because in this study I argue that criminal-electoral violence is mostly organized and strategically employed by these organizations. A further proof of this relation is well illustrated in Figure 7. The plot shows that in Southern Italy intimidation attacks evolve differently in electoral and non-electoral provinces. In contrast, trends in the rest of country show an increase during the months after the elections, and this can be easily explained by the fact that the right part of the plot displays summer months. Therefore, in line with Figure 7, pre- and post-elections average comparison suggest that elections increase criminal violence. In contrast, no marked pre-post difference arises for other type of crimes, robbery or homicides. I argue below that one should not expect these categories to be affected by elections. In fact I will use them as falsification tests. Moreover, an overview of the summary statistics on electoral and nonelectoral provinces in Table 2, suggests that only in organized crime territories the difference between electoral and non-electoral is relevant. In the pre-electoral period, the average two-months rate was 11,6 in electoral provinces. This is 10% higher than in the post-electoral one. In all cases expect for southern electoral areas, intimidation attacks are lower value in pre-electoral periods. Finally, while intimidation attacks are slightly higher in electoral areas during elections, in non-electoral areas these attacks are markedly higher in the post-elections periods. Concerning other crimes (rubbery, extortion, homicides), 11

12 they show similar trends simultaneously in electoral and non-electoral areas, suggesting that the variation is explained by other cross-areas time invariant effects. This is the case of rubbery which is slightly higher in pre-electoral periods but in both groups. 6. EMPIRICAL STRATEGY This empirical section aims at identifying the effect of elections on criminal violence. This impact can be estimated by computing for every province the change in criminal violence in electoral and nonelectoral time periods, i.e. in the months immediately prior and after elections. Given the high frequency of data (monthly data at a sub-provincial level), I argue that a significant variation in violence during election times can reliably be explained mostly by the approaching of elections. However, many other factors unfolding over times, besides elections, might have caused the change in violence over time. In order to exclude the presence of confounding factors, I adopt a difference-in-differences identification strategy, which is usually used in economics to study the impact of some treatments (i.e. policy reforms) on economic issues such as unemployment, income convergence, etc. etc. This estimation model compares changes over time for treated units with the change experienced by a control group in the same time period. Applied to this study, elections can be considered as a suitable form of treatment. In fact: (a) elections are clearly exogenous since organized crime cannot affect their timing; (b) they also occur at a single point in time because they have clear starting and stopping dates thus it permits to observe the variation in criminal violence both before and after elections are held; (c) and finally we have more elections over time (i.e. more pre-/post- time periods) and they do not occur necessarily across provinces at the same moment (as regional elections do). In this setting the control group is composed by areas which do not experience elections in the same time period. Thus, the impact of elections on criminal violence can be estimated by computing a double difference, one over time (before-after elections) and one across units (between electoral and nonelectoral areas). The crucial assumption is that in the absence of elections the trend among the two groups would have been the same. In order to be more precise, the outcome variable is some measure of criminal violence. For each province I measure monthly arson and bomb attacks ( attacchi dinamitardi e incendiari and incendi dolosi ). The selection of electoral periods is constrained by data disaggregation, since the dataset provides monthly crimes per province. Therefore, in this study electoral periods are defined as the month prior and the month when elections take place; otherwise non-electoral periods are defined 12

13 as the two months after elections 12. By this selection I cover the entire time period of electoral campaign, and for symmetry, I define the non-electoral period to be as long as the former and it effectively covers the time in which the new elected representatives take up office. The strategy to shortening the span of the electoral period is justified by the need to capture those reported crimes that I assume are connected to the electoral process rather than to other illegal goals (Angrist and Pischke, 2008). Concerning the selection of control groups I use a simple method. Table 1 lists two groups of areas in which regional elections are held in different years 13. Control areas have to be very similar to the treatment group but for the treatment (Bertrand et al., 2004). For this reason, I select those areas that do not receive the treatment, i.e. elections, during the same period. In this control group, for defining electoral and non-electoral months I follow the same strategy used for the treatment group as described above. To summarize, I estimate several versions of the following model: Violence pi,t = α + β ELECT_PROV pi + γpre_elect t (1) + δelections pi,t + D p + D i + D t + ε pi,t In this specification, VIOLENCE pi,t are intimidation attacks in period t and area i of province p. The variable is constructed by summing reported crimes about arson and bomb attacks ("Incendi dolosi" and "Attentati dinamitardi e incendiari"), by area of the province and time period; ELECT_PROV pi is defined as a dummy variable assuming value 1 if in areas p and i elections take place; otherwise it will be 0; PRE_ELECT t is defined as a dummy variable assuming value 1 if month t is the one prior or the same of elections; otherwise it will be 0; Elections pi,t is the interaction term i.e. the product of the two binary variables (ELECT_PROV pi* PRE_ELECT t) and it is equal to 1 only in those months and provinces when and where elections are held. D p province fixed effects; D i area fixed effects; D t time fixed effects ε pi,t is the error term of the regression with variance σ 2 p indicates province; i indicates the area of the province (every province is composed by two areas: one coincide with the municipality of the main city, the other include the rest of the province); t indicates the month of a certain year; α, β, γ, δ are the regression parameters to be estimated In this equation δ is the key coefficient which identifies the causal effect of elections. It is obtained by calculating the difference in differences equal to the change in mean outcomes of violence 12 I also tested a model in which I define as electoral month only the month prior to elections and results are still significant. 13 Information on both dates and provinces are taken from the Anagrafe degli amministratori locali, Minister of Interior ( ). 13

14 for the electoral provinces minus the change in mean outcomes for the control group, i.e. not electoral provinces. Assuming that the only treatment pre- and post-elections between the two groups are elections, δ identifies its effect. If elections tend to increase (decrease) violence among the electoral provinces then δ is positive (negative). Any cross-regime time effects on criminal-electoral violence over periods are captured by γ, and any time-invariant differences in criminal-electoral violence between groups is captured by β. Observations are clustered at the city level, hence all estimated standard errors are robust to within province correlation. I run the baseline regression for full sample (all Italian provinces). Then the analysis is restricted to explore the casual mechanisms among areas of the country. I examine Center-North Italy, Southern Italy and finally, in a further specification of the model, as robustness check, I restrict the sample to those provinces that have a value of mafia-association arrests above the 75th percentile. I expect that the relation would be not significant for Northern Italy (2.a) due the lack of organized crime penetration. In the last two models (2.b and 3) I expect a significant and positive correlation even though I restrict the sample. The diff-in-diffs estimation for organized crime provinces (3) is crucial because groups within this panel are very similar, and therefore it is more valid the assumption that in the absence of elections the trend among the groups would have been the same. 7. MAIN ESTIMATES Table 3 shows the main estimates of several versions of model (1). All models include a full set of time, province and principal town dummies. Column (1) in panel (I) shows the estimates of the linear regression for full sample. The estimated coefficient on the variable Elections pi,t (δ) is +1,1 and it is statistically significant as we expected. Results show that as elections approach there is an increase of 1.1 intimidation attacks compared to the case in which elections had not occurred. More interestingly, even though the coefficient of PRE_ELECT does not answer directly to our research question, it shows clearly that in general, without considering the effective occurrence of elections, there is a significant and positive increase in criminal violence in post-electoral months (+2,5). This result is plausible if we consider that in our panel post-electoral months usually correspond with summer when temperatures are naturally higher. Figure 8 clearly shows the evolution of intimidation attacks to and from elections in Northern and Southern regions. The figure graphically confirms these first results. In contrast, Figure 9 proves that other crimes (robbery) do not evolve differently in electoral and nonelectoral periods both in electoral and non-electoral areas. 14

15 Therefore, this first estimation confirms the hypothesis that criminal organizations are most likely to engage in violence during elections. However, the magnitude and the significance of the relation is slightly weak. One can argue that organized crime is not equally distributed across regions in Italy. If this is true we should not expect an increase in criminal violence as elections approach in those provinces not controlled by criminal organizations. In order to test it, in the following columns (2.a and 2.b) I break down the full sample in two panels Center/North Italy and Southern Italy (Campania, Puglia, Calabria, Sicily). Results clearly change. In panel 2.a, the relation is not significant anymore and weaker in magnitude. On the contrary, results in column (2.b) for Southern Italy show that as we move to some areas controlled by organized crime the relation between elections time and criminal violence is stronger. By restricting the panel I identify a more homogenous sample for those unobserved characteristics (political culture, social capital) across provinces that can affect the estimation. For Southern provinces, the estimated coefficient on the variable Elections (β 3) is 2.1, and it is well estimated (t-value +2,57). Given that on average in every electoral province there are 9,3 attacks per electoral period, according to our estimates elections produce an increase in intimidation attacks by 22%. Moreover in the last column, I test the robustness of the previous finding, by looking at those areas that have a value of mafia-association arrests above the 75th percentile. Restricting the attention to organized crime areas involves a variance-bias trade-off. On the one hand, excluding non-organized crime discards relevant variation and increases variance. On the other hand, restricting the sample to organized crime areas I reach two goals reducing potential bias: concentrating on them helps to homogenize the control and treatment groups; second, it reduces the risk of capturing potential unobserved characteristics across provinces. The results of column (3) clearly show that elections timing explains variation in criminal violence. In this model the estimated coefficient is a much bigger in magnitude (+3.4), moreover, the relation is still significant at 5% although a sharp reduction in the number of observations (640 obs.). Moreover, I test the same models but weighted by population (per 100,000 inhabitants), which serves to emulate a regression at the individual level. The weight is in population in 1991 (National 2001 Census, Istat). Estimates should be compared within a column. Results in columns WLS show that the estimated impact of elections is robust to weighting procedure. An alternative way to test the models is the estimation of the effect of elections on common crimes. I suppose, in fact, that common crimes are not significantly related with the electoral months since they are not related to criminal intimidation. Columns (Rubbery Extortion Homicides) show that no significant correlation exists between common crimes and electoral timing. Similar fixed-effects to model (1) are used here to test the robustness of this specification, and the panel is restricted to organized crime provinces. Moreover, even thought the sign of this relation is not relevant for this study, the negative coefficient indicates a decrease in the volume of those crimes before elections. 15

16 Therefore we can argue that only mafia crimes can be considered politics-oriented in the sense that they are correlated with institutional variables, such as elections. Finally, I test separately the impact of national elections (column National Elections). The results show that legislative elections are not significantly correlated with mafia violence. The estimated coefficient on the national elections is also smaller in magnitude -0.2 and more interestingly it is negative. Evidence shows that criminal organizations do not use violence during legislative elections as they do for local elections due the different nature of the electoral race. To summarize the robustness checks, difference-in-differences estimations using different panels and types of crime, provide the basis for validation and falsification tests and support the estimation strategy. A further step of this analysis would be to test whether this violence can decisively affect electoral outcomes in Italy, but due the availability of data no statistical evidence about that can be offered at this moment. 8. CONCLUSIONS According to the estimation, elections cause monthly intimidation attacks to increase by almost 2,1 in Southern provinces, and by 3,4 in organized crime areas, which means a 22% increase in the first case and a 31% in the second one. To the best of our knowledge, this is the first estimate of the impact of elections on criminal violence accounting for cross-areas and secular trends. These findings are consistent with a large case study literature documenting the interventions of criminal organizations into the electoral process in Southern Italy. Moreover, even though the goal of the study was not to evaluate empirically the effects of criminal-electoral violence upon the electoral process, we can reasonably argue that in some parts of the country an impact on electoral turnout or on voters preferences might be discovered. The effect of criminal-electoral violence still needs to be estimated carefully. In fact, by moving beyond the existing work, I argue that the increase in criminal violence during electoral campaigns is not linked with stronger criminal control upon electorate, but, in contrast, it is associated positively to higher degree of opposition and electoral accountability. If criminal organizations had a direct control both of voters and agenda-setting, violence would be unnecessary. In contrast, the growing incidence of violence may be an indicator of greater levels of opposition, and a lessening of voter mobilization control. Violence is costly, especially when it is employed too liberally. In that case, criminal organizations may attract unwanted attention from law enforcers or produce a counter-mobilization by civil society. There is less 16

17 violence in organized crime than international cinema and journalism suggest. It is often the threat of violence, rather than its exercise, the source of mafias power to compel others to do what they otherwise would not. Diego Gambetta makes the point in his study of the Sicilian mafia: «violent action, while crucial, is only occasionally demanded of a mafioso (although it must always be perceived as a potential threat)» (Gambetta, 1993:273). The likelihood to use intimidation depends on an evaluation of likely payoff from relying on alternative non-violent resources (money, reputation, membership) and the capacity to effectively influence candidates and voters (Collier and Vicente, 2009). Violence is only one, and probably the most costly resource to be exploited in order to provide mafioso protection, compared to non-coercive means which are equally strategic, such like reputation and information (Gambetta, 1993). When non-violent means become useless, violence is the only available tactic to influence elections. Motivated by this evidence, I argue that an increase of criminal violence could be interpreted as an unexpected effect of political accountability and of the end of one-party dominance in many provinces in Southern Italy since the 80s. The availability of victimization data would help us scrutinizing this mechanism. These data can reveal which side of the political market is particularly favored and targeted by criminal organizations. In fact, criminal groups may choose between two different channels to affect representation: the demand side or the offer side. In the first case, the attempt to change preferences of restive voters through intimidation is not only the most costly strategy, but also the less effective because it might lead to unforeseeable outcomes, such as the increase of antimafia opposition. Only in small municipalities where criminal groups can control directly voters, we can suppose that it would be rational to use violence to convince them. In other conditions, the alternative way is more attractive and cheap. Criminal organizations can target the offer side of political market, by contributing to candidate nominees or creating a climate of fear and terror that would raise the cost of political campaigns and, especially at local level, it would avoid people from standing as candidate cause the risk of being victim of such violence. This becomes the first-best option if criminal groups are capable to capture the nominee of that candidate with preferences very close to theirs, and most likely to win such elections. When criminal groups are not able to capture candidates, therefore they tend to more often resort to violent means to exert influence at a time when the degree of electoral uncertainty is higher. Increased level of competition around election times motivates criminal groups to use violence to influence electoral results. The findings of the paper confirm the abundant judiciary evidence about criminal-electoral violence. More importantly the paper illustrates a previously unexplored dynamic between electoral institutions and the strategic timing of organized crime s intimidation attacks. However, a full analysis on electoral behavior and turnout should be conducted in order to assert confidently that criminal- 17

18 electoral violence has an effect on them. A similar study would help researchers evaluating whether organized crime can only mobilize voters or also change their preferences. 18

19 Table 1 Regional Elections in Southern Italy ( ) Source: Archivio Elettorale, Ministero dell Interno Region Province Regional Elections National Elections Sicilia Agrigento Caltanissetta Catania June 24, 2001 Enna June 16, 1996 Messina June 16, 1991 Palermo June 22, 1986 Ragusa Siracusa Trapani Calabria Campania Puglia Catanzaro Cosenza Crotone* Reggio Calabria Vibo Valentia* Avellino Benevento Caserta Napoli Salerno Bari Brindisi Foggia Lecce Taranto April 16, 2000 April 23, 1995 May 6, 1990 May 12, 1985 May 13, 2001 April 21, 1996 March 27, 1994 April 5, 1992 June 14, 1987 June 26,

20 Kkkkkk Figure 1 Fig. 1.1 Evolution of mafia-type association rate (c.p 416 bis) in Southern Italy and in the rest of country from 1983 to 2003 (Left: total number/right: Crime Rates per inhabitants) Source: SDI Min. Interno Fig. 1.2 Evolution of extortion rate reported in Southern Italy and in the rest of country from 1983 to 2003 (Left: total number/right: Crime Rates per inhabitants) Source: SDI Min. Interno Fig. 1.3 Evolution of mafia-type homicides in Southern Italy and in the rest of country from 1983 to 2003 (Left: total number/right: Crime Rates per inhabitants) Source: SDI Min. Interno 20

21 Fig. 2 Mafia-type homicides in inhabitants ( ) Source: SDI Min. Interno Fig. 3 People arrested for mafia-type association (c.p 416 bis) per inhabitants Source: SDI Min. Interno Fig. 4 City Councils dissolved by central government due mafias penetration ( ) Source: Criminal-Political Capture Dataset, Sberna

22 Figure 5 Evolution of mafia-type homicides and intimidation attacks from 1983 to 2003 Source: SDI Min. Interno 22

23 Figure 6 Violent and Non-Violent Crime Correlation by Mafia Density ( ) Source: SDI, Ministero dell Interno INGIURIE OMICIDI COLPOSI PERCOSSE DELITTI INFORMATICI SFRUTTAMENTO PROSTITUZIONE VIOLENZE SESSUALI LESIONI DOLOSE FURTI OMICIDIO PRETERINT. INTIMIDATION ATTACKS 0,5 0,4 0,3 0,2 0,1 0-0,1-0,2 TENTATI OMICIDI ESTORSIONI ROBBERY STRAGE HOMICIDES RICETTAZIONE INCENDI VIOLAZI. PROPRIETÀ INTELLETT. RICICLAGGIO ASSOCIAZ. DELINQUERE ALTRIDELITTI ATTENTATI STUPEFACENTI MINACC INFANTICIDI SEQUESTRI PERSONA CONTRABBANDO TRUFFE CORRUZIONE MINORENNE DANNEGGIAMENTI USUR ATTI SESSUALI CON MINORENNE CONTRAFFAZIONE Lllllllllllllllllllllllllllll Figure 7 Intimidation attacks for months to and from elections Note Source: SDI, Ministero dell Interno 23

24 Table 2 Summary Statistics, Electoral and Non-Electoral Areas North Electoral areas Non-electoral areas Pre-Elect Post_Elect Pre-Elect Post_Elect Intimidation Attacks 2,84 3,04 2,43 3,05 (9,03) Extortion 0,75 (1,88) Robbery 8,22 (27,09) Homicides 0,15 (0,51) (7,55) 0,88 (3,73) 8,18 (26,40) 0,16 (0,53) (4,98) 0,83 (1,71) 8,59 (28,34) 0,16 (0,56) (6,49) 0,75 (1,72) 8,47 (27,04) 0,16 (0,51) South Electoral areas Non-electoral areas Pre-Elect Post_Elect Pre-Elect Post_Elect Intimidation Attacks 9,43 9,16 8,52 9,98 (10,9) Extortion 2,59 (4,25) Robbery 29,51 (67,62) Homicides 0,63 (1,31) (9,66) 2,61 (3,60) 28,85 (65,45) 0,625 (1,16) (9,21) 2,84 (3,94) 31,61 (77,46) 0,71 (1,28) (11,11) 2,49 (3,87) 28,22 (66,98) 0,77 (1,38) Organized Crime Areas Electoral areas Non-electoral areas Pre-Elect Post_Elect Pre-Elect Post_Elect Intimidation Attacks 11,66 10,93 9,625 12,70 (12,81) Extortion 3,21 (5,49) Robbery 53,99 (95,36) Homicides 0,89 (1,72) (10,30) 3,38 (4,31) 52,79 (92,12) 0,97 (1,49) (10,19) 3,275 (4,36) 59,28 (110,91) 1,04 (1,56) (13,34) 3,10 (4,97) 52,44 (95,55) 1,01 (1,65) Note: Northern Italy, 158 areas and obs.; Southern Italy, 48 areas and 1456 obs; OC Areas, 20 areas and 640 obs. Source: SDI, Ministero dell Interno. 24

25 Table 3 Main Estimates (I) Dependent variable: Monthly intimidation attacks (1) (2.a) (2.b) (3) Italy Center/North South Organized Crime Prov. PRE-ELECT (0.610)*** (0.321) (1.627)*** (2.957)*** ELECT_PROV (0.431) (0.751) (0.571) (0.981)* Elections (0.655)* (1.088) (0.817)** (1.388)** CONSTANT (0.715)*** (0.193)*** (1.486)*** (2.692)*** OBSERVATIONS PROV. FIXED EFF. YES YES YES YES TIME FIXED EFF. YES YES YES YES PRINC. TOWN FIXED EFFECTS YES YES YES YES Note: Robust standard errors in parentheses *** significant at 1% ** significant at 5%; * significant at 10% ; Elections are all regional elections from 1983 to 2003; Source: SDI, Ministero dell Interno (II) Dependent variable: Monthly intimidation attacks per inhabitants (WLS.1) (WLS.2.a) (WLS.2.b) ( WLS.2.b ) Italy Center/North South Organized Crime Prov. PRE-ELECT ELECT_PROV Elections CONSTANT (0.052)*** (0.183)** (0.131) (0.257)* (0.053)*** (0.241)* (0.136) (0.127)*** (0.244)*** (0.322) (0.393)* (0.490)** (0.289)*** (0.354) (0.472)** (0.490)** OBSERVATIONS PROV. FIXED EFF. YES YES YES YES TIME FIXED EFF. YES YES YES YES PRINC. TOWN FIXED EFFECTS YES YES YES YES Note: Robust standard errors in parentheses *** significant at 1% ** significant at 5%; * significant at 10% ; Elections are all regional elections from 1983 to 2003; Source: SDI, Ministero dell Interno (III) Dependent variable: Other crimes per inhabitants PRE-ELECT Extortion Rubbery Homicides Intimidation Attacks (0.233) (0.619) (0.059) ELECT_PROV (0.231) (0.556) (0.063)) Elections (0.333) (0.810) (0.083) (0.610) CONSTANT (0.542)*** (0.939)** (0.161) (0.605)*** OBSERVATIONS PROV. FIXED EFF. YES YES YES YES TIME FIXED EFF. YES YES YES YES PRINC. TOWN FIXED EFFECTS YES YES YES YES Note: Robust standard errors in parentheses *** significant at 1% ** significant at 5%; * significant at 10% ; Elections are all regional elections in sequence; Estimations for Organized Crime Provinces Panel (75th percentile); Single-difference estimation for legislative elections; Source: SDI, Ministero dell Interno 25

26 Figure 8 Intimidation attacks for months to and from elections Note Source: SDI, Ministero dell Interno 26

27 Figure 9 Robbery crimes for months to and from elections Note Source: SDI, Ministero dell Interno 27