Mafia Inc. : When Godfathers Become Entrepreneurs

Transcription

1 Mafia Inc. : When Godfathers Become Entrepreneurs Marco Le Moglie Bocconi University Giuseppe Sorrenti University of Zurich December 1, 2016 Abstract Although criminal organizations are usually responsible for the deterioration of local economic conditions, they are capable to attract considerable social consensus especially in a context of weak institutional presence. We study one of the main channels used by organized crime to obtain empowerment and territorial control, namely the investment in the legal economy. We focus on the case of Italy, a country historically plagued by a conspicuous presence of mafia-type organizations, and we disentangle for the first time the disruption effect - worsening local economic prosperity - from the entrepreneurial effect - allowing for money laundering and engendering silent social consensus. Our results highlight a strong and sizable investment of organized crime in the legal economy. JEL classification: K42, L26. Keywords: Mafia, Organized Crime, Illegal Enterprises. We wish to thank Guglielmo Barone, Nadia Campaniello, Marina Di Giacomo, Frederico Finan, Sergio Galletta, Guillerme Lichand, Massimiliano Piacenza, Paolo Pinotti, Paola Profeta, David Stadelmann and Gilberto Turati for useful comments. DONDENA, Bocconi University (IT). marco.lemoglie@unibocconi.it Department of Economics, University of Zurich (CH). giuseppe.sorrenti@econ.uzh.ch

2 Mafia organizations are more of a threat than terrorist groups because they modify democracies from within by introducing their illicit earnings into the legal economy. [...] When we think about mafia organizations, we are inclined to see only their illicit activities: drug trafficking, weapons and racketeering. But this is only the surface; behind these lies an enormous and illegally amassed economic power, which is camouflaged and laundered until it becomes legal. [...] The strength of criminal organizations is the lack of attention they receive from governments and their ability to garner social consensus - in cases where the state is absent, the mafia offers services to citizens. Roberto Saviano - The New York Times (20 November 2015) 1 Introduction Criminal organizations are one of the main economic actors in many countries worldwide. Illicit markets managed by criminal organizations in the European Union generate around 110 billion euro per year, an amount close to the 1% of the EU GDP (Organised Crime Portfolio, 2015). According to Demoskopika, the Italian mafia-type criminal organization Ndrangheta was able to raise revenues for 53 billion euro in An amount higher than McDonald s and Deutsche Bank together. 1 Considering that the economic activity perpetrated by criminal organizations is found to be largely detrimental for local development and growth, how are criminal organizations able to obtain empowerment and territorial control? The channels used are multiple. Violence and intimidation are two of the main practices adopted especially in the first stages of the infiltration process in a specific territory. Alongside with violent practices, mafia-type organizations exploit the presence of a weak institutional context to raise different forms of social consensus in the local population through the provision of services and goods usually demanded to institutions. 1 Although these values should always been considered with caution, they provide reliable insights about the magnitude of the phenomenon. 2

3 Social consensus has been one of the fundamental milestones at the basis of the empowerment of many criminal organizations. To provide some examples, Bandiera (2003) describes the raise of the Sicilian Mafia in Italy as a movement fulfilling the need of the population for land protection from predatory attacks. The raise of Pablo Escobar and the Medellin Cartel in Colombia constitutes a similar case. Pablo Escobar and his criminal organization were able to control around 80% of the cocaine trade in the world. To carry on with his business, Pablo Escobar was the instigator of an extraordinary number of killings against members of other cartels, police officers, and government officials. Despite his conduct, Escobar was able to gain the consensus of a considerable fraction of the population of Medellin and Colombia offering services not provided by the Colombian state. New housing projects, hospitals, schools, sport facilities were financed by Pablo Escobar and contribute to the creation of his Robin Hood image guaranteeing to him protection and devotion from a considerable part of the local community. In this paper, we investigate one of the main channels used by criminal organizations to laundry money and raise social consensus, namely the investment in the local entrepreneurial sector. 2 The aim is to disentangle for the first time the disruption effect induced by criminal organizations and responsible for worse local economic conditions from the entrepreneurial effect fostering money laundering and the creation of social consensus. We exploit the massive raise in Italy since the 19th century of mafia-type organizations to pursue our aim and we measure entrepreneurial activity as the number of new enterprises established yearly. 3 The isolation of mafia s entrepreneurial effect from the disruptive one is a challenging task due to, at least, four reasons. Firstly, the detection of activities perpetrated by criminal organizations is particularly difficult due to its illegal nature. Thus, it is very complicated to understand and quantify a behavior that economic agents strive to 2 Money laundering is a vast phenomenon in the Italian economic system. See, for example, Ardizzi et al. (2014) for an estimate of its impact in the Italian financial sector. 3 We refer to this type of criminal organizations with the term mafia. The term mafia often refers to the Sicilian Mafia (Cosa Nostra), while here it also refers to other criminal organizations arisen in Italy in the past, such as the Camorra, the Ndrangheta and the Sacra Corona Unita. In the remainder of the paper we will use mafia, mafia-type organizations and criminal organizations as synonymous, for simplicity. 3

4 hide. Secondly, as mafia s presence usually generates two opposite-in-sign economic effects such as the disruption and the entrepreneurial effect, it is impossible in absence of an exogenous shock to isolate a specific single effect. Thirdly, the definition of mafia, its activity and territorial presence is far from being unequivocal. Lastly, criminal organizations are possibly non-randomly assigned across the territory. We use forensic economics, a Difference-in-Differences estimation strategy, alternative definitions for mafia s presence and a novel instrumental variable approach to overcome each of these issues. Criminal organizations widely perpetrate their businesses through illegal activities, therefore difficult to be detected. Many measures have been proposed to proxy the level of the territorial criminal presence, however these variables suffer from different sources of bias (e.g. degree of corruption in the local public administration, effectiveness of the judicial system etc. ). The forensic economics approach allows to overcome the concern induced by the illegal nature of mafia s activity. It examines how a single outcome, that is potentially the product of both honest and hidden behavior, varies with the incentives for hidden behavior (Zitzewitz, 2012). Precisely, it deals with the use of information about licit markets to highlight different insights on illicit activities. Although criminal organizations mainly operate through illegal activities and markets, their involvement in the legal economy leaves detectable traces. The Italian National Law 580/1993 requires each Italian enterprise to register its activity in the Registry of Enterprises (Registro delle Imprese). This registration is mandatory for all the enterprises operating nationwide, therefore it is undertaken both by legal and illegal enterprises. 4 As a consequence the number of new enterprises established at territorial level represents the optimal outcome to identify mafia s investment in the legal economy as it is an observable measure directly related to such investment. The number of new enterprises is the sum of both those enterprises with no connections with criminal organizations and those with some connections with them. The use of the number of new enterprises is per se insufficient to identify the effect of legal investments of criminal organizations in the legal economy. As mentioned, mafia gen- 4 By illegal enterprises we mean all those enterprises whose activities are carried out by members of criminal organizations or whose capital is raised through the exercise of illegal activities - for example, drug dealing. 4

5 erates two opposite effects when it comes to the analysis of the territorial entrepreneurial activity. On the one hand, a negative (and dominant) effect lowers the entrepreneurial territorial activity, on the other hand the direct investment by mafia is likely to induce an increase in the local entrepreneurial activity. The simple comparison of established enterprises between areas with and without criminal organizations would therefore be ineffective in disentangling these two opposite effects. To isolate the entrepreneurial effect related to mafia s presence, we take advantage of the exogenous shock in the legal credit supply generated by the outbreak of the financial crisis. The financial crisis was responsible for a severe credit rationing started in We implement a Difference-in- Differences (DiD) estimation strategy in which provinces with high mafia presence are compared to provinces with lower mafia rate before and after the outbreak of the financial crisis. The identification relies on the circumstance that the financial crisis sparked a sharp contraction in the legal credit supply provided to entrepreneurs across the Italian territory, while it left almost unaltered mafia s sources of capital (Organised Crime Portfolio, 2015). 5 Moreover, the shock induced by the financial crisis would be considered as exogenous with respect to the presence of criminal organizations. The financial crisis originated in the United States, therefore reducing the possible existence of anticipation effects especially for countries far from the American banking and financial system. The definition of mafia is not unequivocal. Some definitions rely on measures of military occupation to define whether an area might be considered as highly infiltrated by criminal organizations. Other measures privilege the level of economic activities executed by mafia in a specific area to characterize it as highly infiltrated. In this paper, both types of measures are used. We take advantage of the work of Transcrime, a joint research centre on transnational crime, and Fondazione RES, an Italian research centre on local socioeconomic development, that provide different specific indexes for mafia s presence in order to capture the two possible dimensions of the phenomenon. 6 The last concern related to our empirical strategy resides in the possible endogeneity 5 Criminal organizations raise their capital in markets - e.g. drug dealing, racketeering - almost unaffected by the financial crisis. See Section 4.1 for a detailed analysis of one of the main source of profits of criminal organizations, namely the market of drugs. 6 We will explain in detail each measures in the remainder of the paper. 5

6 of criminal organizations presence. Mafia is likely to establish its activity in geographical areas characterized by the presence of specific factors pushing the establishment of new enterprises. In this case, our estimates would reflect both the pure effect of mafia s investment and the effects of such specific factors, especially those which cannot control for because unobservable. To overcome this issue we will propose a novel instrument for the historical raise in the XIX century of criminal organizations in the Italian territory. The historical raise of mafia-type organizations in Italy was driven by factors completely different with respect to the ones that induced criminal organizations settlement in the last decades. Our instrument exploits the two main drivers that were responsible for the historical establishment of mafia s presence, namely a week institutional framework (Gambetta, 1993; Konrad and Skaperdas, 2012), and the presence of valuable natural resources (Barone and Narciso, 2015; Buonanno et al., 2015). Specifically, we take advantage of the location of active volcanoes across the Italian territory as volcanoes improve the quality and fertility of lands affected by eruptions, turning this latter into an important asset to defend (also privately) in a context of scarce land availability and weak institutions. Our results confirms the relevance of the criminal organizations entrepreneurial role. Provinces with more mafia infiltration experience a less severe drop - around 5 percentage points - in the number of new enterprises established in the period after the outbreak of the financial crisis. The effect is robust to alternative measures for mafia s presence and to different model specifications. Moreover, the effect is particularly strong when areas characterized by higher levels of credit rationing, sectors (Construction) highly attractive to criminal organizations, and favorable legal forms of enterprises (Limited Companies and Partnerships) are considered. We also provide evidence about the existence of spillovers in the labor market induced by the presence of criminal organizations. The policy implications of this research are paramount and general. Our estimates provide a first quantification of the illegal investment in the legal economy highlighting not only the existence of such investment but also its magnitude. This investment is among the main drivers of social consensus, probably the main obstacle to the institutional fight against criminal organizations. The understanding of criminal organizations behavior is 6

7 crucial to preempt their empowerment and to fight the social consensus they raise in the local population. In addition, the results shed lights on mafia s preferences when it comes to laundry money raised through illegal activities. Finally, our research suggests that, in a context of credit contraction, the entrepreneurial sector becomes more vulnerable to criminal infiltration. As a consequence, changes in the credit market conditions such as credit contractions should be used in future as important signals requiring increasing monitoring and vigilance activities by institutions. The remainder of the work is structured as follows. In Section 2 a brief review of the literature about the origin and the activity of criminal organizations in Italy will be provided. Section 3 introduces the database used for the analysis and descriptive statistics of our sample, while Section 4 explains in detail the implemented identification strategy and presents the baseline estimates. In Section 5, we discuss and address the main threats to our identification strategy. Section 6 shed lights on the mechanism behind our findings. Section 7 concludes the work. 2 Organized Crime in Italy After an initial standstill, in recent years economists have devoted growing attention to the empirical analysis of organized crime and its evolution over time. With reference to the US, Miller (2009) studies both theoretically and empirically leniency in cartel enforcement, Levitt and Venkatesh (2000) analyze criminal behavior of drug-selling street gangs, while Mastrobuoni (2015) focuses on the US mafia members to detect the criminal network effect on their economic status. When it comes to the analysis of organized crime, the Italian case represents a rich soil for research as a result of the exceptional heterogeneity across regions and the existence of a complex and well established criminal activity managed by mafia-type organizations. The rise of the Sicilian Mafia - traditionally deemed as the reference term for organized criminal activities in Italy - is historically dated in the 19th century as a response to a weak institutional context engendering the demand for land protection from preda- 7

8 tory attacks (Bandiera, 2003). The role of institutions in the formation of the Sicilian Mafia is confirmed in Buonanno et al. (2015), in which local mafia expansion is related to the presence of weak institutions in a land with valuable natural resources. 7 The rise of the Sicilian Mafia was contemporaneous with the birth of similar criminal organizations in other Italian regions. The most important examples are the rise of Camorra in Campania, Ndrangheta in Calabria and Sacra Corona Unita in Apulia. After an initial setting characterized by immobility, these criminal organizations started to expand their influence and activity to more productive and profitable areas - namely central-northern regions. Buonanno and Pazzona (2014) investigate the determinants of such expansion from southern regions to the rest of Italy. They highlight the importance of two key factors: on the one hand, the Italian economic miracle between the late 50s and the early 70s was responsible for a massive migration phenomenon from the South to the North of the country. On the other hand, a specific law - the Confino law - also contributed to the national spread of criminal organizations. The Confino law was mainly implemented in the 60s and 70s and imposed a resettlement in a different province than the original one for individuals likely to be involved in mafia-type criminal activities. These two factors, plus a series of other contextual conditions, have produced the first evidence of mafia presence in northern Italy during the late 60s. In the 70s mafia s power increased thanks to its role in drug dealing, kidnapping and racketeering crimes. The 80s and 90s are the decades in which these organizations have completed their settlement through the acquisition of power not only in the illegal market but also in the legal one. The economic and social consequences generated by the territorial presence of organized crime are relevant. According to a general perspective, mafia s presence seems to generate a huge loss in terms of economic resources. Pinotti (2015) estimates that the cost imposed by the presence of mafia in terms of GDP per capita is around 16 percentage points. A similar result arises in Peri (2004); the presence of organized crime is associated with lower levels of employment rates and employment growth. Albanese and Marinelli (2013) highlight the negative impact of organized crime on productivity. 7 For further studies on the relationship between weak institutions, natural resources and organized crime see also Gambetta (1993) and Konrad and Skaperdas (2012). 8

9 The negative effect of mafia s presence is the result of a series of factors. Mafia usually adopts violent and intimidating behavior to control the local entrepreneurial activity and to obtain forms of monopolistic power (Falcone, 1991; ARIEL, 2015). A further factor relates to credit availability and cost. In areas characterized by high levels of criminal activities the cost of credit tends to increase. Credit supply is affected by the amount of money spent by the banks on security and protection. Furthermore, in such areas it is more difficult to assess the quality of borrowers; uncertainty about future behavior is likely to consistently reduce propensity to grant a loan not backed by a collateral (Bonaccorsi di Patti, 2009). Also foreign directed investments (FDI) inflows in the Italian provinces are affected by a local criminal presence. Daniele and Marani (2011) show a negative and significant correlation between FDI and organized crime as a proof of the deterrent role for foreign investors played by local presence of organized crime. FDI inflows are crucial in determining future investments from abroad as they summarize the difficulty in setting up new companies, the government and judicial system effectiveness, the level of property rights security (Globerman and Shapiro, 2002; Bénassy-Quéré et al., 2007; Wei, 2000). Although it is undeniable that mafia-type organizations constitute a deterrent for both national and international private investors, Barone and Narciso (2015) demonstrate how mafia s presence is associated with higher probability to receive public funds in form of business subsidies. This result is driven by the level of corruption in the public administration and by the huge amount of criminal infiltration in Italian institutions. Unsurprisingly, anecdotal evidence shows how the mafia-politicians network is seen by civil society as the most important factor shaping mafia s power nationwide (Cayli, 2013). Our contribution to the existing literature is threefold. We exploit the outbreak of the financial crisis to provide quantitative evidence about the existence of both the entrepreneurial and the disruptive effect induced by the presence of criminal organizations. Lastly, based on the same approach of Barone and Narciso (2015) and Buonanno et al. (2015), we introduce a novel instrument for mafia s presence at national level and not only for a specific region. 9

10 3 Data and Descriptives To estimate whether the presence of criminal organizations locally affects the number of new enterprises, we assemble a new panel database at provincial level covering the period from 2003 to We end up with a sample made by 1,133 annual observations. In this section, we provide a description of the variables used in our models together with the motivation for their inclusion. Appendix A.1 provides more detailed information about the data sources and definition of variables. The number of new enterprises established yearly in each province is obtained from the Italian National Institute of Statistics (ISTAT). The choice of new enterprises, instead of alternative measures for entrepreneurial activity such as the total number of operating enterprises or the number of closed businesses, derives from several considerations. New enterprises were more sensitive to the credit crunch which started with the financial crisis of 2007 (the exogenous shock used to identify the effect of interest) with respect to those enterprises established already. 9 Moreover, the number of new enterprises enables us to disentangle the pure mafia entrepreneurial effect. Established and closed enterprises are victims of racketeering, one of the main activity perpetrated by criminal organizations. It is hardly likely that a potential entrepreneur interested in opening a new business decides to start its activity under the threat of racketeering practices perpetrated by organized crime. As a consequence, our identification strategy based on the number of the new enterprises might be able to isolate and identify those enterprises actively run by criminal organizations through a direct control by the organization or a massive inflow of capital often raised with illegal activities. Mafia s presence at provincial level is measured through the Transcrime Mafia Index (TMI). 10 TMI s definition of mafia s presence includes features related both with mafia s territorial occupation and its infiltration in the economic and social life of a particular area. 8 In order to get a balanced panel we use the classification of the Italian provinces in force until The number of provinces was 103 till In 2005, the number rose to 107. Since 2009, the Italian territory is divided into 110 provinces. 9 It is fair to assume that it is easier to deny credit to a potential new enterprise with respect to stopping the credit supply to an established business. 10 As introduced, the TMI is elaborated by Transcrime, a joint research centre on transnational crime. 10

11 Therefore it might be considered as a general index capturing all the relevant activities perpetrated by criminal organizations. It defines mafia as a system characterized by the presence of criminal groups providing illicit goods and services; using violence, threat, or intimidation to pursue their aims; and with a high degree of infiltration in the political and the economic system. Such definition includes characteristics related to mafia s capacity to control the territory through the use of violence and by interfering in the political and the economic system. At the same time, the definition also includes elements related to mafia s capacity to carry out illicit traffics. The details about TMI s construction are provided in Appendix A.2, while Figure 2 graphically shows the territorial distribution of the index. Figure 2 illustrates how mafia s presence is spread nationwide. Some traces of criminal organizations presence appear over the entire Italian territory, although a strong prevalence emerges in areas in the Centre and in the South of the country. 11 The relation between the number of new enterprises and mafia s presence in a province is subject to the influence of many contextual factors. We collect information on a wide set of observable factors likely to influence the relation of interest for this work. We firstly rely on a set of characteristics of the credit market. The number of new enterprises, at provincial level, is related to many characteristics of the local credit market, as the percentage of big banks, the cost of credit and the capacity of attracting foreign investments. Such variables are also likely to affect the degree of mafia s territorial control. When the local supply of credit is constrained, with high costs, and provided by small banks unable to undertake risky projects compensated by external capital, the opportunity cost of turning to illegal credit channels is likely to decrease. The reduced opportunity cost of illegal capital would facilitate territorial control executed by mafia. According to this, we use the provincial number of bank agencies divided by bank size - big banks vs. small/medium banks - to control for the structure of the banking system. Big banks are defined by the Bank of Italy as those with a total value of traded funds greater than 26 billion euro. We also collect information on the provincial average funding risk to account for the cost of credit, while the value of external capital is proxied with the net flow of 11 The distribution of the TMI index ranges between 0 and 1 and it is strongly asymmetrical. The 75th percentile of the distribution corresponds to a value of

12 direct investments from abroad. A second set of factors to be properly considered relates to the local economic and institutional environment. Economic conditions are likely to affect the number of new enterprises either controlled or not by mafia. Sectors that are more mature and less competitive than others, generate a fewer number of new enterprises. Criminal organizations, according to the reports produced by the Italian judiciary authority, have very specific preferences in terms of sectors of activity. To address this aspect, we use the provincial level of employment in each macro-sector - primary vs. secondary (excluding construction) vs. tertiary - to proxy the size of each sector of activity. The average size of the local production unit - in terms of number of workers per unit - captures the degree of competitiveness of the production system at the provincial level. More specifically, some sectors of activity - e.g. the construction, the health care, the waste treatment and the tourism sectors - are usually recognized as more at risk of criminal infiltration. We take into account the territorial economic relevance of those sectors through the collection of data about the employment rate in the construction sector, the number of beds in the public National Health Service (NHS), the per capita quantity of produced waste, and an index of attractiveness of tourism-related consumption. Moreover, since many of mafia s activities are in sectors strongly related to the demand for consumption goods and leisure, we also include per capita taxable income at provincial level to control for changes in the demand-side. 12 The provincial degree of entrepreneurial activity and innovation are two other important features to be addressed. The number of new enterprises is likely to be positively correlated with these two characteristics, and it probably also shapes mafia s presence, especially through the provincial economic competitiveness. We use the provincial level of self-employment and the provincial number of patent applications to the European patent office to proxy these two characteristics. Finally, the size of the infrastructure network plays an important role in affecting both the capacity of generating new businesses (positive correlation) and mafia s ability to 12 In Section 6, we will also discuss a robustness check based on the analysis of the construction sector, the more infiltrated by criminal organizations. 12

13 control a territory (negative correlation). Information about the total length of roads within each province are collected. The literature about organized crime in Italy pointed out how the institutional and social environments are likely to shape the relation between mafia s presence and the birth of new enterprises. The efficiency of the juridical system stands out as the main institutional factor affecting both the number of new enterprises (positive relationship) and mafia s presence (negative relationship). We measure the efficiency of the judicial system with the provincial average duration in days of a bankruptcy trial. To take into account the importance of social environment, we consider the territorial level of social and human capital. Both factors positively influence the birth of new enterprises and negatively affect mafia s presence. Social and human capital is proxied by the regional average number of blood donations and the provincial circulation of newspapers. 13 Moreover, as mafia s presence is positively correlated with the number of 15 to 24 year old males in the province, and negatively with the provincial urban population, we also include these two data in our database. These two variables are likely to be correlated also with the birth of new enterprises through the effect on the level of competition and market dynamism. Table 1 provides summary statistics for our sample. The average number of enterprises is 629 units with high variability when different geographical areas are considered. The minimum number of enterprises at year-province level is 329 units, while the maximum is 1,420. With reference to mafia s presence, the TMI at provincial level is, on average, By looking at the map of TMI (Figure 2) across provinces, it would be noticed that mafia s presence displays the higher values in those areas where criminal organizations were formed and therefore, where the infiltration in the institutional and social contexts is stronger. 13 Data about blood donations are unavailable at the provincial level. 13

14 4 Identification Strategy and Baseline Results 4.1 Identification Strategy In this work we isolate mafia s investment in the legal economy through the creation of new enterprises from the disruption effect induced by the presence of organized crime and lowering economic and entrepreneurial activity. The identification of such investment would be easy under the existence of direct measures of the actual number of new enterprises financed by mafia. The illegal nature of mafia s activity makes this potential source of data unavailable. Despite this limitation, there is still the possibility to get this information through the use of some observable proxies for mafia s entrepreneurial role. The number and the total value of the enterprises seized from mafia by judicial authority are examples of such proxies. Although these variables are undoubtedly connected with the actual level of these investments, they have different shortcomings. Firstly, since they are based on criminal records, they only rely on the enterprises seized by the judicial authority, but they do not take into account all those still to be discovered. Therefore, their use as proxies for how much mafia invests in the legal economy is likely to lead to a permanent and consistent under-estimation of the actual level of legal businesses carried out by organized crime. In addition, the heterogeneity of these records is only partially determined by the real heterogeneity in the level of mafia s business. The quality and the efficiency of the judiciary system are also likely to constitute a considerable fraction of such heterogeneity. The higher is the level and the quality of the contrast to mafia - especially the political willingness to pursue this scope - the higher both the number and the total value of enterprises seized will be. Finally, as a lag exists between the discovery of an enterprise financed by mafia and its actual seizure (mainly due to the standard procedural time) the total number of seizures will also depend on the efficiency of the judiciary system. We overcome these limits by undertaking a forensic-economics approach (Zitzewitz, 2012). Forensic economics deals with the use of information about licit markets to highlight different insights of illicit activities. The idea is that also illicit activities usually 14

15 need complementary legal sectors or market to be carried. The investment in the legal economy by criminal organizations is no exception to this rule. The Italian National Law 580/1993 requires each Italian enterprise to register its activity on the Registry of Enterprises (Registro delle Imprese). This registration is mandatory for all the enterprises operating nationwide, therefore it is undertaken both by legal and illegal enterprises. The choice of the number of new enterprises guarantees that our variable of interest is undeniably containing the subgroup of enterprises with some connections to the illegal credit market. However, a simple comparison between provinces with and without mafia is insufficient to disentangle the two opposite-in-sign effects induced by the presence of organized crime on the number of new enterprises yearly established. Indeed with this approach we are only able to identify the overall effect that is mainly driven by the general (and dominant) disruption effect leading to a negative coefficient for mafia s presence. To isolate the entrepreneurial effect, we take advantage of the mandatory registration along with the exogenous shock in the supply of legal credit generated by the outbreak of the 2007 financial crisis. The shock was not anticipated by the Italian credit market and, especially, by mafia as it was generated in the US market. Moreover, the financial crisis sparked a sharp contraction in the legal credit supply provided to entrepreneurs across the Italian territory, while it left almost unaltered mafia s sources of capital (Organised Crime Portfolio, 2015; ARIEL, 2015). The activity that generates the higher profits for organized crime in Italy is drug dealing. According to the estimates of Transcrime it generates 7.7 billion euro per year, an amount that almost double the second source of revenues, namely racketeering. We analyze the trends of the Italian market for drugs to understand whether the crisis had implied consumption contraction therefore negatively affecting organized crime profits. With all the caveats related to measurement of illicit activities such as drug consumption and market value, Figure 1 shows the evolution of the Italian market for drugs in the period of interest for this study. The analysis is focused on four different drugs (Amphetamine, Cocaine, Cannabis, and Heroin) and it considers as proxy for market activity the total number of drug offenses, the total number of seizures, the total quantity of drugs seized and the average price. All the figures suggest a strong 15

16 stability of drug consumption in the analyzed period, therefore suggesting how profits raised in these market were not affected by the outbreak of the crisis. To provide an example, looking at the top-left panel is it possible to observe that the number of offenses related (only) to drug use - thick black line - has been increasing or almost constant in the period after This framework allows to implement a Difference-in-Differences (DiD) estimation strategy in which we define provinces with high level of mafia s presence as the treatment group, while provinces with low level of mafia s presence constitute the control group. We compare the number of new enterprises between the two groups both before and after the strike of the financial crisis. The rule applied to define the group of each province is based on the distribution of the TMI previously shown. Specifically, we consider as province with Mafia the ones belonging to the fourth quartile of the TMI distribution. 14 Figure 3 graphically represents the rate of change of credit supplied to the industrial sector for both the treated and control groups. Since 2007, due to the outbreak of the financial crisis, both groups experienced a consistent contraction in the credit supplied. The trend for mafia s provinces fairly mimics the one of those provinces with low levels of organized crime. This evidence supports the idea of the crisis as a shock strongly affecting credit availability nationwide. The shock exogeneity is confirmed by the analysis of the pre-crisis period. No anticipation effect is detected. In the years before 2007 the Italian industrial sector is characterized by an increasing (at a constant trend) provision of credit. The pre-crisis trends are remarkably similar across the two groups of provinces. Also in the period after the crisis the two groups of provinces behave in a similar way in terms of credit granted to the industrial sector. This similarity is fundamental to guarantee that our effect is not driven by structural differences in both the supply of legal credit between provinces with mafia presence and their counterpart with no criminal organizations. The graphical representation supports this evidence. However, to be even more cautious we will provide a formal test in Section From now on, we will label Italian provinces distinguishing between those with mafia and those without mafia according to this definition. The 75% percentile of the TMI distribution corresponds to a TMI s value of

17 Figure 4 presents a graphical test for the parallel trends assumption, providing a strong support in favor of our identification strategy. In the period the trend for the number of new enterprises for provinces with no mafia fairly mimics the one for those provinces with a strong mafia presence. The effect of the crisis is evident for both groups since its outbreak in 2007, highlighting a dramatic drop in the number of new enterprises. Moreover, with the exception of 2008 and 2009, it is possible to notice how the line for provinces without mafia is steeper than the one for provinces with mafia. Therefore, we estimate the following DiD baseline equation: NewEnterpises i,t = β 0 + β 1 Mafia i Crisis i,t + βx i,t + α i + γ t + ε i,t (1) where i defines the province, while t the year. N ewenterprises is expressed as the logarithm of the number of new enterprises normalized per 100,000 inhabitants. We aim at consistently estimating the coefficient β 1 obtained as the interaction between our treatment Mafia - to be part of territory characterized by high level of presence of criminal organizations - and Crisis - an indicator variable for the period starting from This coefficient would shed lights on the different trends in the number of new enterprises in areas characterized by different level of mafia s presence once the legal source of credit has been constrained. The equation (1) also contains province fixed effects (α i ) to take into account timeinvariant unobserved heterogeneity at provincial level, and year fixed effects (γ t ) to consider common shocks such as the outbreak of the financial crisis. To capture time-varying determinants of the number of new enterprises at provincial level a vector X it is included in this specification. This vector, as previously shown, contains information about funding risk, percentage of big banks, foreign directed investments, employment in the primary sector, employment in the secondary sector excluding construction, average size of the local production unit, employment in the construction sector, number of beds in public hospitals, waste per capita, capacity to attract tourism, per capita taxable income, self-employment, number of patent applications, average duration trial for bankruptcy, 17

18 length of road system, number of blood donations, urban population and male youngster population. 15 Finally, standard errors are adjusted for heteroskedasticity and clustered at provincial level. 4.2 Difference-in-Differences Estimates The estimates of equation (1) are reported in Table 2. The model in column (1) only includes province fixed effects. Column (2) also includes the full set of controls previously introduced, while column (3) adds year fixed effects. We will refer to the model in column (3) as the full model. The results highlight a clear pattern. Provinces with a high level of mafia s presence experience a smaller reduction in the number of new enterprises in the period from 2007 to The coefficient of interest is always statistically significant and remarkably high in magnitude. In terms of size, the effect of mafia s presence is bounded between 3 and 8 percentage points, with the full model displaying a value of 4.8 percentage points. As previously introduced, we also check that this effect is not driven by a different level of credit contraction between the groups of provinces with and without presence of criminal organizations. Table 3 illustrates the results. In column (1), we model the rate of change of credit supplied to the industrial sector as a linear function of a dummy variable indicating the period after the outbreak of the crisis (Crisis), province fixed effects, and an interaction term Mafia Crisis. In column (2), we replicate the same model by considering year and province fixed effects and interacting the variable M af ia with each single year. Both the specifications suggest how the presence of organized crime is not responsible for a different trend in terms of credit granted to the industrial sector in the period following the outbreak of the crisis. In this estimation framework, the (possibly) non-random assignment of mafia s presence might arise objections to the causal interpretation of results. Organized crime is likely to be settled in particular areas because of attracting factors such as weak institutional 15 In the specifications without year fixed effects this vector will also contain the indicator variable Crisis to capture the general effect of the crisis on the number of new enterprises established yearly. 18

19 presence, high economic potential etc. We will address in detail this potential source of bias in Section 5.3 with the implementation of an Instrumental Variable approach capturing the determinants of the historical raise of organized crime in Italy. Here we replicate the approach used in works when similar issues arises such as Biderman et al. (2010) or Galiani et al. (2005). The reaction to the crisis could depend on secular differences across provinces in the local economic environment. To rule out such a possibility, we estimate a further specification in which we allow for time trajectories of different groups of control variables by interacting them with year dummies. The estimates are reported in Table 4. As the sample size does not allow for contemporaneous inclusion of the trends of all our covariates, we split the analysis according to the four category of controls included in our baseline model. In column 1, we test the trends for the variables capturing credit market conditions as in 2006, in column (2) we use test variables for economic conditions, in column (3) we test the institutional context, while in column (4) social related controls are checked. 16 The size of the coefficient for the effect of interest remains remarkably similar to the one of the baseline estimates at a cost of a reduction of estimates precision. To shed more light on the timing of the effect of interest, we estimate an alternative specification based on a DiD including leads and lags: NewEnterpises i,t = β j=2003 (Mafia i Y ear j )β 1,j + βx i,t + α i + γ t + ε i,t (2) The interest resides in the estimates of β 1,j, the coefficients for the interaction of mafia s presence with the indicator variables for each year. 17 The analysis of leads allows to formally test parallel trends assumption (Autor, 2003), while lags show whether the treatment effect changes over time after the treatment. Splitting the overall effect across different years enables to understand the timing of the overall effect shown before. It might be that the credit rationing took some time to be visible in terms of effect on the number of new enterprises formed each year. The estimates are reported in Table 5. We 16 In other words, we assume that 2006 represents the last observable value of the secular trends of each group of covariates before the shock (possibly) induced by the outbreak of the crisis. 17 We use as reference category the interaction between Mafia and the year 2006, the year before the outbreak of the crisis. 19

20 report only the full model containing both province and year fixed effects. 18 Analysis of coefficients shows how provinces with and without a strong mafia s presence were performing similarly in the period before the outbreak of the crisis. All the coefficients are remarkably small in magnitude and never statistically significant. This pattern confirms the validity of the implemented DiD approach. The size of the interaction between Mafia and each Y ear considerably increases starting from Provinces with more presence of criminal organizations experienced a huge increase in the number of new enterprises from 0.02 percentage points before the crisis to 2.5 percentage points registered in the year of the outbreak of the crisis. Such effect slightly decreases in 2008, while it increases and turns to be significant since The effect in the years is bounded between 4 and 11 percentage points. To sum up, the analysis with leads and lags clearly shows an increasing trend in the effect of interest started in The trend highlights how the effect of mafia s presence on the number of new enterprises becomes progressively higher with the persisting of the crisis and the exacerbating of its consequences. 5 Threats to the Identification This section provides a series of robustness checks to validate the results presented in the previous section. We will address a series of potential concerns regarding the functional form uses, mafia s definition, and the (possibly) non-random assignment of criminal organizations across the national territory. 5.1 Functional Form s Sensitivity We consider here the count data nature of the number of new enterprises. In particular, we test whether our estimates are sensitive to the functional form previously introduced. To take into account the count data nature of the number of new enterprises, we estimate a negative binomial regression model. 19 The control variables used are the same as in the 18 A graphical representation of the results is provided in Figure As the likelihood-ratio test of the null hypothesis that the dispersion parameter is zero is strongly rejected, we opt for the Negative Binomial model instead of the Poisson one. 20

21 baseline specification. Table 6 reports the estimates of the model. All the results are remarkably similar to baseline estimates. The DiD strategy shows a positive effect of mafia s presence on the number of new enterprises set up in the period after the outbreak of the crisis. The coefficient is statistically significant and with a magnitude of 4.5 percentage points remarkably close to the one shown in the baseline specification. This test provides a reassuring pattern showing how estimates are not sensitive to the functional form chosen. 5.2 Measurement Error The definition of mafia s presence is likely to be affected by two types of measurement errors. A first type relates with the sources used to define criminal organizations activity. These sources are affected by many unobservable factors (e.g. the efficiency of the judicial system) therefore potentially influencing differences in mafia s presence across the country. We will specifically address this measurement issue in Section 5.3. The second possible measurement error derives by the specific rule applied by Transcrime to define the index for mafia s presence. To overcome this concern, in this section we test the sensitivity of our main findings to alternative definitions of mafia s presence. Our baseline model defines mafia s presence by using the fourth quartile of the distribution of the TMI. This means that some provinces with low levels of criminal organizations are classified as control units. To test the sensitivity to this definition we replicate our main analysis with a less restrictive definition for mafia s presence. Specifically, we also include in the treatment group those provinces in the third quartile of the distribution. Table 7 reports the analysis. The effect remains strongly statistical significant, although as expected the inclusion of provinces with a low level of presence of criminal organizations partially reduces the magnitude of the coefficients. The analysis presented so far is based on the use of the Transcrime Mafia Index (TMI) to identify our treatment group. The definition of mafia is far from being unequivocal. We propose here two alternative indexes for the presence of criminal organizations characterized by the specificity of their definition. 21

22 Firstly, we measure mafia s presence through the Power Syndicate Index (PSI). 20 The PSI relies on the concept developed by Block (1980) to classify the presence of criminal organizations by the type and scope of their activities in a given area. Specifically, the PSI maps mafia s degree of control of a territory in terms of military occupation. The index is based on records of specific illicit activities to evaluate the degree of such control. These activities are divided between core and minor activities. Core activities are mafia s criminal association (Associazione Mafiosa) as described in the Law 646 (art.416-bis), murders by mafia members and racketeering practices. Mafia s criminal association is defined as a group of people that by use of intimidating behavior, membership to the organization subjugation and a code of silence, commit criminal activities, to acquire direct or indirect control of economic activities, concessions, authorizations, public contracts or to generate illicit profits or advantages or to impede or obstruct the exercise of the right to vote or to ensure the procurement of votes for them or for others during elections. Minor activities include the number of properties seized from criminal organizations or city councils dissolved because of mafia s infiltration. 21 The distribution of the PSI over the Italian territory is presented in Figure 6, while the estimates obtained with the PSI index to measure mafia s presence are reported in Table 8. We define as provinces with mafia those provinces with a value of PSI greater than 0. This definition is the less affected by possible measurement error as it compares provinces with some trace - also marginal - of mafia s territorial presence with those without any sign of criminal infiltration. The effect is remarkably similar to the one of the baseline model. With reference to the full model in column (3), the drop in the number of new enterprises established in the period after the crisis is 4 percentage points less severe in provinces with a high level of criminal organizations. This result reinforces our previous analysis confirming both the sign and magnitude of the effect of interest for this work. We also introduce a second index to measure mafia s presence. The Enterprise Syndicate 20 The PSI index is elaborated by the Fondazione RES, an Italian research centre on Sicilian economy and society. 21 Appendix A.3 provides further details about these activities, how they are combined to construct the index and the sources of data used. 22

23 Index (ESI) measures criminal infiltration in a different way with respect to the PSI. 22 It defines mafia s presence in a specific province according to its capacity to provide illegal goods and services. Also this index is based on the work by Block (1980). Appendix A.4 provides a detailed description of the index. Figure 7 represents the distribution of the ESI over the Italian territory. The ESI distribution across the Italian provinces appears as more heterogeneous with respect to the PSI, with several provinces characterized by a high degree of mafia s presence also in the centre and northern part of the country. Although most of the provinces with high PSI have also a high ESI, the opposite is not true. The arising difference relates to the fact that the capacity to raise an illegal traffic is usually a forerunner for territorial control. To check whether the mafia-effect on the number of new enterprises is mainly driven by territorial control, we estimate a model with a restricted sample containing only those provinces with no territorial control executed by mafia. Those provinces are the ones reporting a zero value for mafia s presence measured through the PSI. With this framework we are able to (1) recognize whether mafia s presence fosters entrepreneurial activity only in a specific part of mafia s distribution or the effect is general across the entire distribution; (2) show whether the mafia-effect is a general effect or is strongly related to the time distance from mafia s settlement - territorial control vs. business infiltration. The estimates of our usual model are reported in Table 9. The results are not statistically significant when the full model in column (3) is estimated. The positive sign of the effect and its magnitude of 2.8 percentage points suggest how the infiltration into the local market through the provision of goods and services is the preliminary step to reach a wider territorial control. The policy implications are paramount as these results highlight the importance of the timing of interventions of institutions to fight the empowerment of criminal organizations. 5.3 Endogeneity of Mafia s Presence In this section we introduce an Instrumental Variable (IV) analysis to account for the (possible) non-random assignment of mafia s presence across the Italian territory. Let 22 Also this index, as the PSI, is provided by Fondazione RES. 23

24 us assume a framework in which mafia sets its activity in a specific area because of the existence of factors driving the establishment of new enterprises. In this context, our estimates might reflect both the real effect of mafia s investment in the legal economy and the effect of the specific characteristics of an area that are likely to attract criminal organizations. The IV analysis allows us to address this concern and also to correct the possible measurement error - due to the sources of information used to construct indexes for mafia s presence - discussed in Section 5.2. Our instrument aims at capturing the historical process leading to the raise of mafia-type organizations. A weak institutional context together with the presence of valuable assets (i.e. land and natural resources) has been found to be conducive of the establishment and growth of criminal organizations (Gambetta, 1993; Konrad and Skaperdas, 2012). The raise of the Sicilian Mafia is an exhaustive example. In the XIX century, the demise of feudalism (1812) and the crumple of the Bourbons dominance were at the basis of a growing demand for land protection, mainly driven by an increase in the number of landowners (Gambetta, 1993). Giving the incapacity of the new Italian state - founded in to provide a clear legislation protecting property rights, private protection was needed to defend the newly acquired plots. Local armed groups - often referred to as the Picciotteria or the Onorata Società - that were used to provide protection to large landowners, expanded their activities obtaining an increasing power. In this context, the value of lands and the availability of natural resources became one of the main determinants of the demand for protection. The raise of the Sicilian Mafia is explained with sulfur availability in Buonanno et al. (2015). Similarly, Barone and Narciso (2015) instrument current mafia s activity with historical and geographical measures of land productivity, specifically rainfalls shocks, slope and altitude. We propose a novel instrument for the historical raise in the XIX century of criminal organizations in Italy by using the location of volcanoes across the national territory. The Italian territory is characterized by the presence of nine active - with an eruption episode in the last 10,000 years - volcanoes. 23 Their location is mainly in the Centre and in the 23 The Italian National Institute of Geophysics and Vulcanology reports the presence of ten active volcanoes, including also a submarine one (Ferdinandea) located between Pantelleria and Sicily. Given the peculiar 24

25 South of the national territory (Figure 8). Two of them (i.e. Etna and Stromboli) are considered permanently active, while the others have cycles of activity with high variation. All of them are characterized by several phenomena of secondary volcanism such as the emission of different type of gases or thermal water. Mount Vesuvius (at least 50 eruptive events between 1631 and 1944) and Vulcano (continuously active since 1727 to 1890) are among the most active volcanoes. On the contrary, among the least active volcanoes there are Campi Flegrei (one eruption in the last 3000 years) and Colli Albani (last eruption between 6000 and 2000 B.C.). 24 Volcanoes location and activity is completely independent of human control but it remarkably affects the fertility and the value of the lands surrounding the volcanoes or affected by their activity and eruptions. The soil in the areas close to the Italian volcanoes is extremely productive and attractive in terms of agricultural potential. The region around Mount Vesuvius is characterized by particularly rich soils mainly because of two large eruptions 35,000 and 12,000 years ago that left the region blanketed with thick deposits of tephra which has since weathered to rich soils. As a consequence, the region has been intensively cultivated since before the birth of Christ (Fisher et al., 1997; Sheets and Grayson, 1979). Similarly, around one quarter of the entire population of Sicily lives on the slopes of the Mount Etna as the soils have been historically extremely fertile (Chester et al., 1985). In particular, such high level of fertility of volcanic soil becomes even more important in a context where the territory available for agriculture is limited like in Southern Italy. As Delmelle et al. (2015) notice, volcanic soils are among the highest value soils as they have unique properties. One of the most important properties for human agriculture is their capacity to accumulate relatively large quantities of organic carbon. In fact, although the portion of the land surface covered by volcanic soils is very limited, they contain up to 5% of the global soil organic carbon. Organic carbon strongly affects agriculture since it remarkably increases soil fertility (Bolinder et al., 2010) and improves physical and submarine nature of this volcano, we decided to exclude it from the sample, restricting our analysis only to volcanoes above the sea level. 24 Data about eruptions are provided by the Italian National Institute of Geophysics and Vulcanology. 25

26 biological properties of the soil (Hati et al., 2007) by decreasing the bulk density, by improving soil structure and their water-holding capacity and by enhancing microbial activity (Yang et al., 2011). We provide qualitative evidence about the relation between volcanoes presence, value of the lands and the origins of mafia by reporting in Figure 9 the level (in tons per hectare) of organic carbon in the first 30 centimeters soil across the Italian territory. 25 The map shows an intense presence of organic carbon in the areas closed to active volcanoes. 26 The areas surrounding the Mount Etna, Mount Vesuvius and the part of Calabria closed to the volcanoes are extremely rich in organic carbon, providing a direct evidence of the high intrinsic value of these territories in terms of land fertility. This makes reasonable to retain these lands being a primary asset to protect against predatory attacks during the transition from the Bourbons reign to the Italian state. 27 A threat to the validity of our instrument can be represented by the possibility that the high value of volcanic lands before the Italian unification would have a direct effect on current agricultural levels, thus on the total number of new enterprises within this sector. This might invalidate the exclusion restriction assumption of the IV. We retain that such a possibility is unlikely to happen in our case. Even if the spatial distribution of organic carbon is time and spatial-persistent, modern agriculture is less dependent on organic carbon because of the massive use of chemical fertilizers to make the soil more productive. Moreover, as noted by Barone and Narciso (2015), the current role of the agricultural sector in the Italian economy is marginal. 28 In addition, in our regression we address possible differences in the importance of the agricultural sector across provinces 25 The data is not available for the regions of Liguria, Friuli-Venezia Giulia and Apulia. Admittedly, the level is registered in 2014 so it is likely to be slightly different with respect to the period characterized by the raise of mafia. However, the level of organic carbon is highly persistent over time ensuring that the distribution in 2014 is similar to the one of the XIX century. 26 Organic carbon is also abundant in mountain areas such as the Alps and the Apennines. This is due to the way in which these areas originated hundred of millions of years ago, arising directly from the sea in response to the extreme compressive stresses and pressure produced by the collision between the African and the Eurasian plates. 27 This is especially true given the important role still played by agriculture within the economy of Southern Italy at that time. 28 According to the Italian National Institute of Statistics, the share of employment in the agriculture sector was about 70% in 1861, while it represents only 4% in 2009 (ISTAT, 2011). 26

27 by including as a control variable the percentage of employed people within the primary sector in each province. Another possible concern about the validity of the exclusion restriction assumption regards the possible indirect effects that the distance from the volcano might play on the current number of new enterprises through channels other than mafia s presence. One possible channel might be tourism and especially the one directly connected with the volcanic activity (e.g. the presence of thermal water). Also in this case, our empirical models always include a specific variable capturing the general capacity of each province to attract tourism-related consumption. The empirical strategy implemented mirrors the one proposed in the DiD setting. In particular, as volcanoes are located only in the central and southern part of the countries it is extremely important to disentangle the geographical effect with respect to volcanoes presence. In this regard, we adopt the most cautious approach by including in all our analysis province fixed effects both in the first and in the second stage. Specifically, we estimate the following first stage equations: Mafia i Crisis i,t = δ 0 + δ 1 DistanceV olcano i Crisis i,t + δx i,t + α i + γ t + ε i,t (3) where X i,t is a vector containing the same control variables as in equation (1). As external instrument, we use the geodetic distance between each province - we use the province capital (Capolugo di Provincia) - and the closest volcano, weighted by the time distance between the year of the Italian Unification (1861) and the year of the last eruption at that time. The rational behind the use of this composite measure relies on the idea that not only the closeness to a volcano matters but also the frequency of its activity. 29 As the endogenous regressor Mafia i Crisis i,t is the interaction between the timefixed dummy Mafia i and the exogenous time-varying dummy Crisis i,t we interact the 29 As Mount Etna and Mount Stromboli are permanently active, we assigned them a unitary weight. The date of the last eruption before 1861 for the other volcanoes is: 1855 (Vesuvio), 1831 (Vulcano), 1783 (Pantelleria), 1538 (Campi Flegrei), 1302 (Ischia), 600 (Lipari) and between 6000 and 2000 B.C. (Colli Albani). 27

28 external instrument with the dummy Crisis i,t (Baltagi, 2011). 30 This transformation of the instrument, without inducing further endogeneity in our estimates, allows us to control for time-fixed heterogeneity also in the first stage (as for instance the specific geographical distribution of active volcanoes only in the Central-Southern part of the country). The estimates of the first stage - bottom panel of Table (10) - confirms the relation between volcanoes location and the raise of mafia-type organizations. The coefficient of the external instrument is highly significant and with the expected negative sign. Provinces more distant from volcanoes are less likely to be within the last quartile of the TMI distribution. The tests reported in the bottom part of the table rule out the possibility of under or weak identification. The upper panel of Table (10) reports the results of the second stage. The results are slightly larger than the ones obtained in the baseline analysis and suggest a decrease in the number of new enterprises of 8 percentage points less severe in those areas with high mafia infiltration. The larger point estimates are likely to be driven by several factors such as the fact that the correction for the possible endogeneity of mafia s presence aims at capturing only those areas where mafia raised in the XIX century. Moreover, IV also corrects the possible attenuation bias of our previous estimates due to both the measurement error in the sources used to construct the mafia index and the presence of unobservable timevarying heterogeneity. The non-random assignment of mafia s presence seems to not affect the results presented so far. By considering an IV approach capturing the historical determinants of the raise of criminal organizations our previous results are confirmed in magnitude and significance. 30 See Section 4.2 for a discussion about the exogeneity of the credit contraction induced by the outbreak of the crisis. 28

29 6 Mechanism 6.1 Heterogeneous Effects of the Credit Contraction This section analyzes the heterogeneity of results according to the territorial intensity of the credit contraction. Such analysis reinforces the use of credit contraction to disentangle the mafia s entrepreneurial effect from its disruption effect. Once the average effect played by criminal organizations is taken into account as in the baseline estimates, one might expect to find a stronger mafia s entrepreneurial effect in those areas characterized by high levels of legal credit contraction. The credit contraction in Italy caused by the outbreak of the financial crisis has been particularly severe for the case of banks with headquarters particularly distant from Italy (Presbitero et al., 2014). These banks are typically part of big and international banking groups. 31 To verify our hypothesis, we implement a triple DiD estimator based on the following specification: NewEnterpises i,t = β 0 + β 1 Mafia i Crisis i,t + β 2 Crisis i,t BigBanks i, β 3 Mafia i Crisis i,t BigBanks i, βx it + α i + γ t + ε i,t (4) where the number of big banks is interacted with both Crisis and Mafia Crisis. In order to avoid possible endogeneity among the variables Crisis, Mafia and the number of big banks, we use the provincial number of big banks in Moreover, since the number of big banks is also among the set of controls of our specification, to avoid multicollinearity we restrict the sample to the period The results - Table 11 - shed lights on an interesting pattern related to mafia s investment in the legal economy. The coefficient of the interaction between Mafia and Crisis shifts to a negative value. In turns the coefficient for the triple interaction is always positive and statistically significant. With focus on the complete model in column (3), it is straightforward to notice how the net impact of mafia s presence on the number of 31 As said, the Bank of Italy defines as big banks those characterized by a total value of traded funds greater than 26 billion euro. 29

30 new enterprises is positive and it confirms the empirical strategy implemented so far. In areas characterized by higher mafia s presence and more big banks - and therefore by a higher level of credit contraction - the drop in the number of enterprises set up after the outbreak of the crisis in 2007 is reduced. The analysis of the local banking system structure reinforces our empirical setting showing how the effect induced by mafia s presence on the number of new enterprises is stronger in areas characterized by higher levels of credit contraction due to the outbreak of the financial crisis. 6.2 Mafia s Investment Sector-Specificity Criminal organizations are usually recognized as more active in specific sectors of the economy. In particular, Italian mafia-type organizations are particularly involved in the construction sector for several reasons. 32 The construction sector is characterized by high movement of capital and high levels of profitability. This allows criminal organizations to laundry money raised through illegal activities. Moreover, the investment in the construction sector represents the first step of an integrated production cycle. Through the investment in the construction sector, mafia-type organizations would also monitor and be involved in complementary markets such as the ones of stone-pits, storage of materials etc. This way, all the other actors involved in the whole process would risk being absorbed by criminal organizations, therefore conveying the market to a monopoly managed by mafia (Falcone, 1991). The monopoly arises as a consequence of the competitive advantage (deterrence of competitors, salary compression due to the use of illegal labor and consistent capital flows from the illegal economy) typical of criminal organizations. In this section we look for further evidence of mafia s investment in the legal economy using the construction sector as a reference sector. We replicate our DiD estimates using the number of new enterprises in the construction sector as the outcome variable. Firstly, we verify the reliability of the parallel trends assumption comparing provinces with high 32 Around 30% of firms seized to mafia-type organizations by the the Italian judicial authority were operating in the construction sector (Ministry of Interior, 2013). 30

31 level of mafia with those with low level of mafia s presence both before and after the outbreak of the crisis. The results shown in Figure 10 highlight how the trends are parallel when the two sub-groups are compared. The graphical representation also shows how the decreasing trend for provinces with no mafia presence is steeper after 2007 than the one for provinces with mafia. Table 12 reports the estimates of the DiD model. A significant (although only at the 10% level) positive effect of mafia s presence on the number of new enterprises in the construction sector is detected. On average after 2007, provinces with higher levels of mafia s presence experienced a less severe decrease - 6 percentage points - in the number of new enterprises in the construction sector. The results are useful according to a dual perspective. On the one hand, the size of the effect of mafia s presence in the construction sector - recognized as one of those of more interest for criminal organizations - reinforces the evidence about its role of investor in the legal economy. On the other hand, estimates confirm that the construction sector would require a strong monitoring activity from institutions as this constitutes one of the first options used by criminal organizations to invest their capital. 6.3 Legal Form of New Enterprises In this section we analyze mafia s investment in the economy in terms of legal form chosen for new societies. The Italian Ministry of Interior states what are mafia s preferences in terms of legal forms of enterprises for its investment (Ministry of Interior, 2013). Limited Companies are by far the first-best option for criminal organizations (46.6%) when it comes to invest their capital. The preference for Limited Companies is driven by two factors. Firstly, these societies are particularly easy to be established with a minimum required initial capital of 10,000 euro. Secondly, these societies guarantee limited patrimonial responsibility of business partners. Other commonly used legal forms adopted by mafia are Individual Companies (25.8%) and Partnerships (23.3%). To understand whether the results reported by the Ministry of Interior are confirmed also in our research framework, we analyze three different legal forms of societies: Limited Companies, Individual Companies and Partnerships. The idea is to verify possible changes 31

32 with respect to the standard forms of investment by criminal organizations occurred during the spread of the financial crisis. As usual, we firstly graphically show the parallel trends for the three groups of legal forms - Figure 11. The trends are parallel in the three cases, although it should be mentioned how in the case of individual companies provinces with mafia experienced a huge increase in 2004 not reflected by an increase of the same entity for the no mafia counterpart. However, the trends are remarkably similar for the other years pre outbreak of the financial crisis. The formal analysis by legal form of society is presented in Table 13. The coefficient for the DiD estimator is positive and statistically significant when the number of Limited Companies is considered as dependent variable. The result is particularly high in magnitude (9 percentage points) and significance and it is in line with the evidence provided by the Italian Ministry of Interior. The effect is also positive and significant, when Partnerships are considered. On the contrary, according to the results in column (3), the relation between the presence of criminal organizations and the number of Individual Companies, does not stand out. This evidence appears motivated by the fact that typically Individual Companies are smaller in terms of size, therefore requiring lower levels of external capital. This makes more difficult to detect mafia s entrepreneurial effect by using the exogenous variation induced by the credit contraction. Indeed, Figure 11 shows how a strong effect appeared when the credit contraction became more severe (i.e ). The analysis by legal form of society strongly supports the theory of a conspicuous investment in the legal economy by criminal organizations. Moreover, it confirms evidence suggesting mafia s preference for specific typologies of societies. 6.4 Closed and Registered Enterprises Up to now we have only considered the number of new enterprises as dependent variable of interest. A consistent part of the literature - e.g. Pinotti (2015) - highlights the detrimental effect on the local economy induced by the presence of criminal organizations. To provide evidence of the overall effect of mafia s presence on the entrepreneurial local activ- 32

33 ity we focus here on the alternative outcomes, namely the number of closed and registered enterprises. 33 This analysis aims at understanding the size and the possible contemporaneous existence of effects on the local entrepreneurial activity other than those detected in terms of established enterprises. The parallel trends assumption is graphically tested in Figure 12. Both closed and registered enterprises are characterized by similar patterns across mafia presence in the pre-crisis period. However, closed enterprises display a clear converging trend in the postcrisis period. Table 14 confirms the graphical evidence. The interaction term M af ia Crisis is indeed positive and statistically significant. Areas with high mafia s presence experienced during the financial crisis a higher number - around 7 percentage points - of closed enterprises. The number of registered enterprises is not statistically significant although positive in sign. These results are extremely relevant as they testify the persistence of the detrimental effect on the local economy played by criminal organizations. This detrimental effect has not been corroded by the financial crisis started in 2007 and it is still very significant and sizable. Despite this detrimental economic effect lowering local development and growth, criminal organizations are still able to create social consensus and operate on a local scale as they are able to exploit weak institutional settings to reach empowerment. 6.5 The Effect of Mafia s Investment on Local Labor Market Criminal organizations exploit the investment in the legal entrepreneurial sector not only to raise profits, but also to increase social consensus through, for example, the improvement of local labour market conditions. In this section, we investigate whether mafia s investment in the legal economy is also responsible for a change in the local employment or it is only an instrument to foster the informal labor market and to laundry capital raised through illegal activities. We replicate our DiD estimation strategy using separately as dependent variable the 33 The number of registered enterprises is a stock measure indicating the number of official enterprises operating in a territory in a specific point in time. 33

34 number of employed, unemployed and inactive people. As always, we firstly verify the reliability of the parallel trends assumption comparing provinces with high level of mafia with those with low level of mafia s presence both before and after the outbreak of the crisis. The results are shown in Figure 13, and graphically highlight how the trends are parallel across different groups of individuals when Mafia and No Mafia provinces are compared. Estimates of the model are in Table 15. According to the results in columns (1), (2) and (3), mafia s presence explains a consistent fraction of the after-crisis drop in employment and unemployment, and an increase in the number of inactive people. These results are consistent with two possible explanations. On the one hand, mafia s presence and its investment in the legal economy might be directed to accomplish money laundering rather than a proper productive investment. In this case, the benefits in terms of employment would be likely to be limited, whereas their cost in terms of deterioration of the local labor market are possibly high. This implies a reduced number of people having or looking for a job, and an increased number of discouraged individuals who give up searching for one. 34 On the other hand, criminal organizations are acknowledged as promoter of illegal jobs. The Ministry of Interior states that salary compression - mainly obtained through irregular jobs - is used by criminal organizations to obtain a considerable competitive advantage with respect to legal enterprises (Ministry of Interior, 2013). This means that the decrease in the number of employed and unemployed people and the relative increase in the number of inactive individuals is not only generated by a worsening of local labor market conditions, rather it might also be due to a part of the local population switching from a regular job to a non regular one. To clarify this possible mechanism, we replicate our estimates in columns (4), (5) and (6) using a triple DiD approach where the percentage of non regular jobs at provincial level as in 2003 is interacted both with Mafia and Crisis. 35 Concerning employed people, once the regional relevance of the illegal labor market is taken into account, the coefficient for 34 As a result, they pass from the unemployment status to the inactivity one. 35 Admittedly, the percentage of irregular jobs at provincial level is a measure difficult to be precisely quantified. To this aim we use the information in ISTAT (2005). More details are provided in Appendix A.1. 34

35 the effect of mafia s presence turns to a positive and significant value. At the same time, the triple DiD coefficient is negative and significant; the effect on employment is higher in areas where the informal labor market is less widespread. On the contrary, the effect of mafia s presence on the number of inactive individuals turns to a weakly significant negative value, while the coefficient for the triple DiD is now weakly positive. People are more likely to switch to the inactive status where the informal labor market is a more accessible alternative. This evidence seems to suggest the following pattern; mafia s investment in the legal economy is likely to generate a positive effect on local employment. However, this effect strongly depends on the level of informality of the local labor market. When the local labor market is characterized by high levels of informality, criminal organizations tend to privilege irregular jobs, therefore reducing the overall impact on regular employment and increasing the number of inactive people. 7 Conclusion In this work we shed lights on mafia s investment in the legal economy exploiting a forensic economics approach. Such approach is based on the mandatory registration required by each enterprise operating in the Italian territory, both legal and illegal. A Difference-in- Differences (DiD) research design allows us to identify the effect of mafia s presence on the number of new enterprises set up each year at the provincial level. The DiD exploits the outbreak of the 2007 financial crisis to disentangle the positive mafia entrepreneurial effect from its negative disruption effect. The analysis provides several evidence concerning mafia s role as entrepreneur. A positive-in-sign and significant effect of mafia s presence on the number of new enterprises stands out. In particular, the comparison before and after the outbreak of the crisis highlights that provinces with a high mafia presence experienced a drop in the number of new enterprises reduced with respect to those provinces with no mafia presence. We highlight on the timing of the effect of interest, finding an increasing in magnitude effect 35

36 in the period after the outbreak of the crisis. The peak was reached during the years characterized by a more severe credit contraction. Measuring criminal organizations empowerment is never free of concerns and subject to possible measurement error. The results are robust to a different functional form specification and alternative mafia presence definitions. We find very similar results with the use of alternative indexes for mafia presence and local empowerment. The use of additional indexes also allows to infer some insights about the procedure that mafia adopts to achieve local empowerment. These insights are especially useful for institutions when it comes to fight criminal infiltration before they obtain a high degree of territorial control. Finally, we also provide a set of additional estimates shedding important lights on specific features of mafia s investment in the legal economy. Our paper confirms how specific sectors and forms of society, for e.g. construction and limited companies, are particularly at risk of mafia s infiltration or what is the real employment effect due to the establishment of illegal enterprises. The policy implications of this research are paramount and general. Firstly, our findings pinpoint how in a context of credit contraction the entrepreneurial sector becomes more vulnerable to criminal infiltration. As a consequence, credit market instability should be used in future as an important signal requiring increasing monitoring and vigilance activities by institutions. Secondly, this work provides a rigorous quantification of the illegal investment in the legal economy. It also sheds lights on mafia s preference when it comes to reinvest money raised through illegal activities. The understanding of criminal organizations behavior is a fundamental step to preempt their empowerment and to fight the social consensus they raise in the local population. 36

37 References Albanese, G., Marinelli, G., Organized Crime and Productivity: Evidence from Firm-Level Data. Rivista Italiana degli Economisti 28(4), Ardizzi, G., Petraglia, C., Piacenza, M., Schneider, F., Turati, G., Money Laundering as a Crime in the Financial Sector: A New Approach to Quantitative Assessment, with an Application to Italy. Journal of Money, Credit and Banking 46(8), ARIEL, Organized Crime Infiltrations of Legitimate Businesses in Europe: A Pilot Project in Five European Countries. Final Report of Project ARIEL, Transcrime. Autor, D., Outsourcing at will: The contribution of unjust dismissal doctrine to the growth of employment outsourcing. Journal of Labor Economics 21(1), Baltagi, B. H., Econometrics. New York: Springer. Bandiera, O., Land Reform, the Market for Protection, and the Origins of the Sicilian Mafia: Theory and Evidence. Journal of Law, Economics, and Organization 19(1), Barone, G., Narciso, G., Organized Crime and Business Subsidies: Where Does the Money Go? Journal of Urban Economics 86, Bénassy-Quéré, A., Coupet, M., Mayer, T., Institutional Determinants of Foreign Direct Investment. World Economy 30(5), Biderman, C., De Mello, J. M. P., Schneider, A., Dry Laws and Homicides: Evidence from the São Paulo Metropolitan Area. The Economic Journal 120(543), Block, A., East Side-West Side: Organizing Crime in New York. University College of Cardiff Press. Bolinder, M. A., Katterer, T., Andren, O., Ericson, L., Parent, L. E., Kirchmann, H., Long-term soil organic carbon and nitrogen dynamics in forage-based crop rotations in Northern Sweden. Agriculture, Ecosystems & Environment 138(3-4),

38 Bonaccorsi di Patti, E., Weak Institutions and Credit Availability: The Impact of Crime on Bank Loans. Questioni di Economia e Finanza (Occasional Papers) - Bank of Italy 52. Buonanno, P., Durante, R., Prarolo, G., Vanin, P., Poor Institutions, Rich Mines: Resource Course and the Origins of the Sicilian Mafia. The Economic Journal, 125(586), Buonanno, P., Pazzona, M., Migrating Mafias. Regional Science and Urban Economics 44, Calderoni, F., Where is the Mafia in Italy? Measuring the Presence of the Mafia across Italian Provinces. Global Crime 12(1), Cayli, B., Italian Civil Society Against the Mafia: From Perceptions to Expectations. International Journal of Law, Crime and Justice 41(1), Chester, D., Duncan, A., Guest, J., Kilburn, C., Mount Etna. The Anatomy of a Volcano. Stanford University Press. Daniele, V., Marani, U., Organized Crime, the Quality of Local Institutions and FDI in Italy: A Panel Data Analysis. European Journal of Political Economy 27(1), Delmelle, P., Opfergelt, S., Cornelis, J., Ping, C., Volcanic Soils. In: Sigurdsson, H., Houghton, B., Stix, J., McNutt, S. (Eds.), Encyclopedia of Volcanoes, pp , Academic Press. Falcone, G., Cose di Cosa Nostra. Fabbri Editori. Fisher, R., Heiken, G., Hulen, J., Volcanoes; Crucibles of Change. Princeton University Press. Galiani, S., Gertler, P., Schargrodsky, E., Water for Life: The Impact of the Privatization of Water Services on Child Mortality. Journal of Political Economy 113(1),

39 Gambetta, D., The Sicilian Mafia. The Business of Private Protection. Harvard University Press. Globerman, S., Shapiro, D., Global Foreign Direct Investment Flows: the Role of Governance Infrastructure. World Development 30(11), Hati, K. M., Swarup, A., Dwivedi, A., Misra, A., Bandyopadhyay, K., Changes in soil physical properties and organic carbon status at the topsoil horizon of a vertisol of central India after 28 years of continuous cropping, fertilization and manuring. Agriculture, Ecosystems & Environment 119(1), ISTAT, L economia Sommersa ed il Lavoro Non Regolare. ISTAT, Italia in cifre. Konrad, K. A., Skaperdas, S., The Market for Protection and the Origin of the State. Economic Theory 50(2), Levitt, S. D., Venkatesh, S. A., An Economic Analysis of a Drug-Selling Gang s Finances. The Quarterly Journal of Economics 115(3), Mastrobuoni, G., The Value of Connections: Evidence from the Italian-American Mafia. The Economic Journal 125(586), F256 F288. Miller, N. H., Strategic Leniency and Cartel Enforcement. American Economic Review 99(3), Ministry of Interior, PROGETTO PON SICUREZZA Gli Investimenti delle Mafie. Organised Crime Portfolio, From Illegal Markets to Legitimate Businesses: The Portfolio of Organized Crime in Europe. Final Report of Project OCP, Transcrime. Peri, G., Socio-Cultural Variables and Economic Success: Evidence from Italian Provinces Topics in Macroeconomics 4(1). 39

40 Pinotti, P., The Economic Costs of Organized Crime: Evidence from Southern Italy. The Economic Journal 125(586), Presbitero, A., Udell, G. F., Zazzaro, A., The Home Bias and the Credit Crunch: A Regional Perspective. Journal of Money, Credit and Banking 46(1), Sheets, P. D., Grayson, D. K., Volcanic Activity and Human Ecology. Academic Press. Wei, S.-J., How Taxing is Corruption on International Investors? Review of Economics and Statistics 82(1), Yang, X., Li, P., Zhang, S., Sun, B., Xinping, C., Long-term-fertilization effects on soil organic carbon, physical properties, and wheat yield of a loess soil. Journal of Plant Nutrition and Soil Science 174(5), Zitzewitz, E., Forensic Economics. Journal of Economic Literature 50(3),

41 Figures Figure 1: The Market of Drugs in Italy Total drug offences Year Amphetamine Heroin Drug use Cocaine Cannabis Total number of seizures Year Amphetamine Heroin Cocaine Cannabis Total seizures (Kg) Mean price Year Year Amphetamine Heroin Cocaine Cannabis Amphetamine Heroin Cocaine Cannabis Note: The figure shows four indicators of the Italian market of illicit drugs (Amphetamine, Cocaine, Heroin, Cannabis). The top-left panel reports the time trend in the total number of drug offences (the black thick line represents the total number of drug offences related to drug use only). The top-right panel reports the time trend in the total number of seizures. The bottom-left panel reports the total number of kilograms seized. The bottom-right panel reports the estimated average price. 41

42 Figure 2: Geographical Distribution of the TMI Note: The map shows the geographical distribution of the Transcrime Mafia Index (TMI) over the Italian provinces. The more intense is the filling color of a province, the higher is the TMI in that province. 42

43 Figure 3: Rate of Change in the Supply of Legal Credit Rate of change in the number of credits (ref. T-1) Year No Mafia Mafia Note: The figure shows the dynamics of the yearly rate of change in the total number of credits granted to the industrial sector. The comparison is between provinces within the first three quartiles of the TMI distribution (No Mafia, continuous line) and those belonging to the last quartile (Mafia, dashed line). Figure 4: Parallel Trends Assumption Average number of new enterprieses Year No Mafia Mafia Note: The figure shows the trends of the average number of new enterprises (per 100,000 inhabitants). The comparison is between provinces within the first three quartiles of the TMI distribution (No Mafia, continuous line) and those belonging to the last quartile (Mafia, dashed line). 43

44 Figure 5: Testing Parallel Trends Assumption MafiaX2003 MafiaX2004 MafiaX2005 MafiaX2007 MafiaX2008 MafiaX2009 MafiaX2010 MafiaX2011 MafiaX2012 MafiaX Point estimates with 95% confidence intervals Note: The figure shows point estimates and the 95% confidence intervals of the model in Table 5. The omitted category is the interaction between Mafia and the dummy for the year 2006 (the year before the outbreak of the crisis). 44

45 Figure 6: Geographical Distribution of the PSI Note: The map shows the geographical distribution of the Power Syndicate Index (PSI) across the Italian provinces. The more intense is the filling color of a province, the higher is the PSI in that province. 45

46 Figure 7: Geographical Distribution of the ESI Note: The map shows the geographical distribution of the Enterprise Syndicate Index (ESI) across the Italian provinces. The more intense is the filling color of a province, the higher is the ESI in that province. 46

47 Figure 8: Active Volcanoes in Italy Note: The map shows the geographical distribution of active volcanoes over the Italian territory. A Volcano is considered active if it showed some activity in the last 10,000 years. An approximate date of the last eruption for each volcanoes is presented in parenthesis. A red triangular indicates a volcano with persistent activity. Source: Italian National Institute of Geophysics and Vulcanology (INGV) 47

48 Figure 9: Geographical Distribution of Organic Carbon Note: The map shows the geographical distribution of organic carbon in tons per hectare in the 0-30 centimes soil over the Italian territory. The more intense is the filling color of a territory, the higher is the presence of organic carbon within that territory. Source: Italian National Research Institution of Environmental Protection (ISPRA) 48