Organized Crime Networks: Theory and Evidence Based on Federal Bureau of Narcotics Secret Files on American Maa preliminary

version, please do not circulate

Giovanni Mastrobuoni June 2011

I would like to thank Theo Diasakos, Matthew Jackson, Claudio Lucifora, Franco Peracchi, Jesse Rothstein, Rocco Sciarrone, Serena Uccello, Aleksey Tetenov and seminar participants at the Economics seminar at University of Arizona, the Labor and Development seminar at Princeton University, the Public Policy seminar at Berkeley, and the workshop on the Economics of Crime and Organized Crime in Palermo, on Institutions, Individual Behavior and Economic Outcomes in Alghero, and on the one in Applied Economics in Petralia for their useful comments. Martino Bernardi, Isabella David, Filippo Maggi, and Dominic Smith have provided excellent research assistance. This research was supported by a Collegio Carlo Alberto grant. I would like to express my gratitude to the Italian Academy at Columbia University for their hospitality. Collegio Carlo Alberto and CeRP, Via Real Collegio 30, Moncalieri, Italy, giovanni.mastrobuoni@carloalberto.org. 2011 by Giovanni Mastrobuoni. Any opinions expressed here are those of the author and not those of the Collegio Carlo Alberto.

Abstract Using a unique data set on criminal proles of 800 US Maa members active in the 1950s and 1960s, and on their connections within the Cosa Nostra network I analyze how the geometry of criminal ties between mobsters depends on family relationships, community roots and ties, legal and illegal activities. I interpret some of the results in light of a model of connections, where more connections mean more prots but also a higher risk of defection. Adding law enforcement to the model allows me to reconstruct the sampling procedure, which I use to build a representative sample of mobsters. I nd that variables that lower the risk of defection kinship, violence, and maa culture among others increase the number of connections. Moreover, there is evidence of strategic endogamy: female children are as valuable as male children, and being married to a connected woman is a strong predictor of leadership within the maa ranks. A very parsimonious regression model explains one third of the variability in the criminal ranking of the men of honor, suggesting that these variables can be used to detect criminal leaders. One of the rather interesting ndings of my simple model is that the distribution of connections is right-skewed, which is remarkably in line with the evidence that maa organizations tend to be extremely hierarchical. Keywords: Maa, Networks, Intermarriage, Assortative Matching, Crime. JEL classication codes: A14, C21, D23, D85, K42, Z13

Introduction

In January 2011, exactly 50 years after Robert F. Kennedys rst concentrated attack on the American Maa as the newly appointed attorney general of the United States, nearly 125 people were arrested on federal charges, leading to what federal ocials called the largest mob roundup in F.B.I. history.1 This paper uses declassied data on 800 Maa members that were active just before the 1961 crackdown to study the workings of the Maa, an organization that has been shown to be very hard to eradicate. For example, record number one (shown in Figure 1) is Joe Bonanno. The les tell us that he was born on January 18, 1905 in Castellamare (Sicily), was 5 foot 5 inches tall, and weighted 190 pounds. He became a US citizen in 1945, resided in Tucson (Arizona), and was married to Filippina Labruzzo. He had one daughter and two sons. He had interests in legal businesses: Grande Cheese Co., Fond du Lac (Wisconsin), Alliance Realty & Insurance (Tucson, Arizona), and Brunswick Laundry Service (Brooklyn, New York). Despite this veil of legal businesses Bonanno was one of the most important Maa leaders in U.S. He attended all top-level Maa meetings, including the 1957 Apalachin meeting and the 1956 Binghamton (New York) meeting. He made trips to Italy to confer with Maa leaders there and negotiate for international narcotic tracking. He also had an extensive criminal history: a record dating from 1930 includes arrests for grand larceny, possession of gun, transportation of machine guns, and obstruction of justice. His closest criminal associates were Lucky Luciano, Francisco Costiglia (Frank Costrello), Giuseppe Profaci, Anthony Corallo, Thomas Lucchese, and Carmine Galante. These records are based on an exact facsimile of the Federal Bureau of Narcotics (FBN) secret les on American Maa members that were active and alive in 1960 (MAF, 2007). I use these les to analyze criminal connections and hierarchies, ranking mobsters based on the number and the quality of their connections.2 I look for empirical evidence
1 2

See The New York Times, January 21, 2011, page A21 of the New York edition. At the end of the 1950s the FBN, which later merged with the Bureau of Drug Abuse Control to form

on organizational rules of the Maa and try to identify the key players. I argue that the geometry of Maa connections is crucial for understanding the activity of the Cosa Nostra (Our Thing) as these connections are the building blocks of the entire Maa, and more generally of organized crime even today. Valachis 1963 testimony and documents found during the 2007 arrest of Salvatore Lo Piccolo, a Sicilian Maa boss, show that the rst rule in the Maa decalogue stays unchallenged: No one can present himself directly to another of our friends. There must be a third person to do it (Maas, 1968).3 Connections are thus necessary for a criminal career within the Maa. Moreover, leadership positions arent simply inherited; soldiers elect their bosses using secret ballots (Falcone and Padovani, 1991, pg. 101). Francisco Costiglia, alias Frank Costello, a Maa boss connected to 34 mobsters, would say he is connected to describe someones aliation to the Maa (Wolf and DiMona, 1974). In 1970 the Organized Crime Control Act dened organized crime as The unlawful activities of ... a highly organized, disciplined association.... The purpose of this research is to shed light on these associations and their organizational structure. The structure has its roots in a world characterized by the absence of legally enforceable contracts. These criminals need to trust each other, and the purpose of this study is to understand how and where this trust emerges. This is the rst study to use network analysis tools on such a detailed set of information on individual Maa members to study the emergence of networks: ranging from their business to their family structure. In the 1960s the total estimated number of members was around 5,000 (Maas, 1968).
the Bureau of Narcotics and Dangerous Drugs, was the main authority in the ght against the Maa (Critchley, 2009). In New York the FBI had just four agents, mainly working in oce, assigned to the area, while in the same oce more than 400 agents were ghting domestic communists (Maas, 1968). 3 The remaining 9 rules are: never look at the wives of friends, never be seen with cops, dont go to pubs and clubs, always being available for Cosa Nostra is a duty - even if ones wife is going through labor, appointments must be strictly respected, wives must be treated with respect, only truthful answers must be given when asked for information by another member, money cannot be appropriated if it belongs to others or to other families, certain types of people cant be part of Cosa Nostra (including anyone who has a close relative in the police, anyone with a two-timing relative in the family, anyone who behaves badly and does not posses moral values).

Since almost all high-ranking members have a record, the 800 criminal proles are clearly a non-representative sample of Maa members. A model with endogenous monitoring highlights the over-sampling of high-ranking members. I devise a way to test the robustness of my ndings to random and non random truncation of the network. Several measures on how central and important members are within the network are employed. A mobster might exert power because he is either connected to a large number of members or connected to a few high caliber gures like Lucky Luciano, Frank Costello, or Joe Bonanno. Mobsters might also be central because they represent bridges that connect dierent clusters of a network. One of the advantages of indices that measure importance based on network-based connections over indices that rely on rank in the organizational hierarchy is that they measure importance in a continuous manner Maa bosses are not all equally powerful, and capiregime often dier in their level of importance, and are more robust to classication errors (Klerks, 2003). Guided by a simple model of connections I try to shed light on several hypotheses about how these networks and the related hierarchies emerge. I evaluate the relative importance of legal and illegal businesses, family ties, and community ties in shaping these networks, contrasting the economic with the social view of Cosa Nostra. Criminals are more likely to be associated with criminals who operate similar illegal businesses if they try to build cartels. Or they may try to diversify the risk of detection by keeping a lower prole and thus associate themselves with criminals who operate dierent kinds of businesses. Carefully chosen marriages might help to establish robust criminal ties. Children may thus prove important, both because of strategic endogamy and because male descendants represent trusted potential associates.4
5

Based on a participant obser-

vation study for a New York based Italian American crime family, Ianni and Reuss-Ianni
Information on the number of male and female children, and on the maiden name of the wife is going to allow me to verify the strategic endogamy theory and to evaluate the importance of descendants. 5 Another instrument for building bonds is the comparatico, a spiritual parentage a la The Godfather. Unfortunately the data do not contain information about these kind of links (Ianni and Reuss-Ianni, 1972).
4

(1972) discuss the importance of family lineage, intermarriage, and kinship. As in the great Maa families of Sicily, the descendants intermarry continuously, for the clan denes who may or may not marry whom. As in clans everywhere, the relationship between the intermarrying pair is dened as strengthening the social structure. (Ianni and Reuss-Ianni, 1972, pg. 192) These marriages would thus have the same function as alliances among European royalty: providing new protectors. Another way children or, more generally, relatives might increase the trust towards a member is because they represent potential targets for retaliatory action. Trust, or better, blind obedience, and the vow of silence, called omert`, are indeed essential for a Maa a clans survival. Maa clans, called Families (as in Bonanno (1983) I use uppercase to distinguish them from the nuclear family), represent societies where social capital produces public bads (Portes, 1998). A Family protects its members and guarantees their monopoly power in exchange for part of their revenues. Large clans will therefore be more powerful but also more exposed. A simple model of sequential dictatorial network formation by a Maa boss formalizes this tradeo and guides my empirical strategy. Some of my economic insights are present in early theoretical analysis of criminal behavior. But most studies have focused on a market structure view of organized crime, where the Maa generates monopoly power in legal (for a fee) and illegal markets. Among others, such a view is present in the collection of papers in Fiorentini and Peltzman (1997), and in Reuter (1983), Abadinsky (1990), Gambetta (1996), and Kumar and Skaperdas (2009). Only two theoretical papers have focused on the internal organization of organized crime groups. Garoupa (2007) looks at the optimal size of these organizations, while Baccara and Bar-Isaac (2008) look at the optimal internal structure (cells versus hierarchies). I borrow some of their insights, mainly that larger organizations are more protable but also more vulnerable. 6

Sparrow (1991) and later Coles (2001) propose the use of network analysis to study criminal networks. Morselli (2003) follows their proposal analyzing connections within a single New York based family (the Gambino family), Krebs (2002) analyzes connections among the September 2001 hijackers terrorist cells, Baker and Faulkner (1993) study the social organization of three well-known price-xing conspiracies in the heavy electrical equipment industry, Natarajan (2000, 2006) analyzes wiretap conversations among drug dealers, McGloin (2005) analyzes the connections among gang members in Newark (NJ), Sarnecki (1990, 2001) uses network analysis to study co-oending behavior among Swedish teenagers. These studies highlight the importance of deep ties, but have little or no background information on the individuals. Recent papers on social networks show that an individuals position within a network is indeed crucial in explaining her or his level of activity (Ballester et al., 2006). If individual decisions are related to the structure of social contacts each individual chooses (or is trapped in), understanding the formation of network structure is crucial for anti-crime policies. Given the complexity of social relationships, evidence documenting patterns of association between agents is a rst priority, partly because it can inform the theoretical literature looking at network formation processes. Several models of network formation have recently been proposed. Most rely on some form of pairwise regressions (see, for example, Bramoull and Fortin, 2010). In this e study I focus on the elements that are related to the centrality of mobsters rather than determining single connections. I use historical data to understand the functioning of the Maa, over the last 40 years the Maa has continued following many of the same rules, and is still active in many countries, including the US. According to the FBI6 , in 2005 there were 651 pending investigations related to the Italian-American Maa; almost 1,500 mobsters were arrested, and 824 were convicted; of the roughly 1,000 made members of Italian organized crime groups estimated to be active in the U.S., 200 were in jail. In addition, the Italian Maa no
6

The source is www.fbi.gov.

longer holds full control of racketeering. With the end of the Cold War and the advent of globalization, transnational organized crime organizations are on the risemainly the Russian Maa, the African enterprises, the Chinese tongs, South American drug cartels, the Japanese Yakuza, and the, so called, Balkan Organized Crime groupsand their proceeds, by the most conservative estimates, comprise around 5 percent of the worlds gross domestic product (Schneider and Enste, 2000, Wagley, 2006). Williams (2001) discusses how networks within and across these organizations facilitate their fortunes. Notwithstanding the magnitude of these numbers, the illicit nature of organized crime activities has precluded empirical analysis and the literature has overwhelmingly been anecdotal or theoretical (Reuter, 1994, Williams, 2001).7 Indeed, this is the rst study that uses extensive individual level data to describe the hierarchy of such organizations. I hope that the U.S. experience with Cosa Nostra and this wealth of information may oer clues to promising control techniques in countries where organized crime is on the rise (Jacobs and Gouldin, 1999). This work is also related to the growing literature on trust, family values, and family businesses. Guiso et al. (2006) present an introduction to the importance of culture, dened as customary beliefs and values that ethnic, religious, and social groups transmit fairly unchanged from generation to generation, on economic behavior. I will argue that the same applies to criminal behavior. Bertrand and Schoar (2006) present a macro-type analysis about the importance of family values for economic growth, and conclude with the comment that more research is needed to understand how family values shape the organization of businesses and their eciency. I look at how this applies to illegal business.
Levitt and Venkatesh (2000) use detailed nancial activities of a drug-selling street gang to analyze gang behavior. But most gangs do not appear to engage in crimes motivated and organized according to formal-rational criteria (Decker et al., 1998).
7

The Origin of American Maa

Before presenting this empirical study it is important to discuss its historical context. I will talk about when the so-called made men came to the U.S. and how the Maa operated in the 1960s when the FBN was ling the records I analyze in this study. Historians dene two waves of immigration from Sicily, before and after World War I (WWI). Before WWI immigrants were mainly driven by economic needs. Several Maa bosses, like Lucky Luciano, Tommaso Lucchese, Vito Genovese, Frank Costello, etc, were children of these early immigrants. Even though between 1901 and 1913 almost a quarter of Sicilys population departed for America, many of the early immigrant families were not from Sicily. In that period around 2 million Italians, mainly from the south emigrated to the U.S. (Critchley, 2009). These baby immigrants later became street gang members in the slums; they spoke little Italian, and worked side by side with criminals from other ethnicities, mainly Jewish and Irish (Lupo, 2009). Lured by the criminal successes of the rst wave of immigrants, and (quite paradoxically) facilitated by prohibitionism, the second wave of immigrants that went onto to become Maa bosses were already criminals by the time they entered the U.S.. Charles Gambino, Joe Profaci, Joe Bonanno, and others were in their 20s and 30s when they rst entered the U.S., and they all came from Sicily.8 Another reason for this selection of immigrants was the fascist crack-down of the Maa, which forced some of these criminals to leave Sicily. After the second wave of immigration the Maa became more closely linked to the Sicilian Maa and started adopting its code of honor and tradition.9 In 1930 and 1931 these new arrivals led to a Maa war, called the Castellamare war, named after a small city in Sicily where many of the new Maa bosses came from. The war lasted until Maranzano, who was trying to become the Boss of the Bosses, was killed,
Bandiera (2003) analyzes the origins of the Sicilian Maa, highlighting how land fragmentation, absence of rule of law, and predatory behavior generated a demand for private protection. 9 See Gosch and Hammer (1975)
8

probably by Lucky Luciano who had joined the Masseria Family.10 This war put Lucky Luciano at the top of the Maa organization but also led to a reaction by the media and the prosecutors. In 1936 Thomas E. Dewey, appointed as New York City special prosecutor to crack down on the rackets, managed to obtain Lucianos conviction with charges on multiple counts of forced prostitution. Luciano served only 10 of the 30 to 50 years sentenced. In 1946 thanks to an alleged involvement in the Allied troops landing in Sicily he was deported to Italy, from where he tried to keep organizing the organization. Between 1950 and 1951, the Kefauver Committee, ocially the Senate Special Committee to Investigate Crime in Interstate Commerce, had a profound impact on the American public. It was the rst committee set up to gain a better understanding of how to ght organized crime, and the main source of information was a list of 800 suspected criminals submitted by FBNs Commissioner Anslinger, most likely an early version of the records used in this paper (McWilliams, 1990, pg. 141). But the Committee could not prove the existence of a Maa. After Lucianos expatriation several other Families headed the organization: Costello, Profaci, Bonanno, and Gambino. Family ties were of utmost importance. According to Bonannos autobiography (Bonanno, 1983), he became the Boss of the Bosses in part by organizing the marriage between his son Bill and the daughter of Profaci, Rosalia in 1956. In 1957 Gambino took over the leadership. Throughout the 1950s the FBN continued to investigate the Maa, but until an unexpected event happened in 1957 the Maa was not under media attention. The unexpected event was a lucky raid of an American Maa summit, the Apalachin meeting. Police detained over 60 underworld bosses from the raid. Joe Bonanno managed to ee, and later came to be known as Joe Bananas. After that meeting everyone had to agree with the FBNs view that there was one big and well organized Maa. This meant the beginBefore this event, in order to end the power-struggle between Masseria and Maranzano, Lucky Luciano had oered to eliminate Joe the Boss Masseria, which he did at an Italian restaurant by poisoning Masserias food with lead.
10

10

ning of the end of the American Maa. Robert Kennedy, attorney general of the United States, and J. Edgar Hoover, head of the Federal Bureau of Investigations, joined Harry J. Anslinger, the U.S. Commissioner of Narcotics, in his war against the mob. The same years a permanent Senate Select Committee was formed the McClellan commission. Anslingers FBN conducted the investigative work and coordinated nationwide arrests of Apalachin defendants. Lucky Luciano died of a heart attack at the airport of Naples in 1962. In the 1950s and 1960s Cosa Nostra was governed by a Commissione of 7-12 bosses, which also acted as the nal arbiter on disputes between Families. The remaining 10 to 15 families were smaller and not part of Cosa Nostras governing body. After learning that he had been marked for execution Joe Valachi became the rst and most important informer for the FBN and later the FBI starting in 1962,11 and revealed that the Cosa Nostra was made of approximately 25 Families. Each Family was structured in hierarchies with a boss, Capo Famiglia, at the top, a second in command, called underboss, Sottocapo, a counselor, Consigliere, and several capo, Caporegime, captains who head a group of soldiers (regime) (Maas, 1968). Figure 2 indicates how in 1963, thanks to Valachis testimony, a U.S. Senate commission set up to investigate organized crime, called the McClellan commission, drew the Bonanno Family tree structure. And Valachi provided information on many more families. Since my data represents a snapshot of what the authorities knew in 1960, they do not contain information about the Family each member belongs to. Nevertheless, well see that the pattern of connections is clearly informative about the structure of Cosa Nostra. Joe Valachis testimony conrmed FBNs view (which at the time wasnt FBIs view) that the Maa had a pyramidal structure with connections leading toward every single member (almost the whole network is connected and the average path length, that is the average distance between two randomly chosen members, is just 3.7).
11 Jacobs and Gouldin (1999) provide a relatively short overview about law enforcements unprecedented attack on Italian organized crime families following Valachis hearings.

11

The FBN Records: a non-random sample of mobsters

The criminal les come from an exact facsimile of a huge Federal Bureau of Narcotics report of which fty copies were circulated within the Bureau starting in the 1950s. These les come from more than 20 years of investigations, and several successful inltrations by undercover agents (McWilliams, 1990). Given that in the U.S. there were an estimated 5,000 members active during those years the list represents a clearly non-random sample of Cosa Nostra members. More active and more connected mobsters were certainly more likely to be noticed and tracked, which is probably why most, if not all, big bosses that were alive at the time have a le. Im going to provide two alternative models. The rst model resembles a sampling procedure that has been called Respondent-DrivenSampling (RDS), and is very mechanical. In the second model, instead, law enforcement and mobsters respond to incentives. Both methods to re-weight the data lead to very similar conclusions.

3.1

Markov-chain Monitoring

There are no exact records about how the FBN followed mobsters and constructed the network, though with the use of surveillance posts, undercover agents, etc. agents were probably discovering previously unknown mobsters following known ones. Two photographs taken in 1980 and in 1988 show how these discoveries might have looked like (Figure 3). If one starts with an initial fraction of N mobsters that are observed p0 , a 1 N vector of zeros and ones, called the seed, after k steps the likelihood of observing a mobsters is

p0 T k = pk

(1)

where T is the transition matrix whose columns sum up to one. Given that the seed, the 12

initially observed mobsters, is unknown I use the stationary distribution p, dened as a vector that does not change under application of the transition matrix, or the likelihood that a mobsters has been observed after several steps independently of the seed:

pT = p .

(2)

The Perron-Frobenius theorem ensures that such a vector exists, and that the largest eigenvalue associated with a stochastic matrix is always 1. For a matrix with strictly positive entries, this vector is unique.
0 The corresponding resampling weights are going to be simply wi = 1 , pi

with 0 < pi < 1.

3.2

A Model of Connections and Monitoring

Given the non-randomness of the Cosa Nostra network, before describing the data and analyzing the determinants of the network structure I develop, based on historical accounts, a simple model of Maa connections. The aim of the model is twofold: i) show how a simple model of connections can generate a positively skewed distribution of degree and some interesting comparative static results; ii) model the non-random sampling design. In this Section I develop a simple model of connections. Each mobster decides how many connections to build. There are no restrictions on the number of connections. Given that the period between the Castellamare war (1929-1931) and 1960 was a relatively peaceful time inside Cosa Nostra, with an established Commissione resolving disputes between Families, I do not model competition or war between Families (Lupo, 2009).12 Moreover, after wiping out competing gangs, mainly the Irish and the Jewish, the Italian Maa had gained complete control over racketeering. During those years the Maa behaved like a stable cartel, with self-enforcing strict rules to keep the cartel running: markets were
See Liddick (1999) who nds that in the 1960s New York Citys ve Maa Families were jointly running the numbers gambling.
12

13

geographically segmented, Family memberships were held stable to keep relative power unchanged, the status quo of leadership within Families was endorsed and insurrections were repressed collectively (Bonanno, 1983). For simplicity I also dont model the choice of dierent kinds of connections (boss, underboss, captains, and soldiers). 3.2.1 Law enforcement

Law enforcement invests in monitoring m at a cost q and tries to maximize the probability of disrupting the mobsters connections (1 p), or alternatively minimize the mobsters probability of a successful connection p:
N {mi }N i=1

min

pi + qmi ,
i=1

s.t. pi =

p0i , mi 0, q > 0 (1 + mi )

p0 i corresponds to mobsters i initial probability of success, that is without any eort exerted by the investigators. The functional form for the productivity of monitoring implies that there are decreasing marginal returns in monitoring. The optimal enforcement eort turns out to be
p0i q
1 +1

m =

1 if p0i q

(3)

0 otherwise

3.2.2

Optimal number of connections

Mobsters choose their optimal number of connections knowing that additional connections increase their inuence but also the risk of betrayal, 1 p.13 For simplicity all connections are modeled as if their number was continuous, and they are assumed to increase the mobsters utility or inuence by the same amount, normalized to 1. Risks related to each
13

Baccara and Bar-Isaac (2008) have a game-theoretical model of connections based on information.

14

connections are assumed to be independent. Mobster is expected utility is

max sps for i = 1, ..., N , s.t. s > 0 . i

s

The optimal number of connections is s = i 1 1 = ln pi ln p0i ln (1 + m ) < 1 if p0 q 1 ln p0i = . 1 ln p0i + +1 ln q = 1 otherwise +1 ln p0i

The number of connections is increasing in the initial probability of a successful and protable connection p0 . The higher the ability of preventing soldiers, captains, and bosses from becoming informers for the FBI or the FBN the more connections a mobster will have. What do these abilities depend on? Fear of retaliation for breaking the vow of silence, called omert`, a has certainly been a key factor for the success of the Cosa Nostra. Some members were well known for the violence, their bloodshedding attitude, among these is Albert Anastasia also known as the Mad Hatter and Lord High Executioner, killed in 1957 in a barber shop because of an internal war between Vito Genovese and Frank Costello. Social ties, in particular, family ties between the boss and his aliates are also likely to keep p and q high. Bosses with larger families should therefore have more connections. Moreover, captains with family ties might also have a lower intermediary cost .14 In the next Section I present the rst empirical test of these hypotheses. Given the closed form solutions for s one can easily simulate the distribution of connections for dierent levels of p. Figure 8 shows that even if p is drawn from a uniform distribution the model does not predict a uniform distribution of connections, but rather
See Reuter (1985) for a discussion about the optimal size of criminal organizations and, more generally, about providing incentives for loyalty.
14

15

a positively skewed one. Up until now all mobsters have been treated equally and are supposed to be free to choose an optimal number of connections. The main reason is that monitoring needs to be predictable for each mobster in order to control for the non-random sampling design of the FBN. determines the relative importance of this endowment and the cost of monitoring q. If monitoring is very costly the optimal monitoring might be not to monitor at all. 3.2.3 Sampling

The presence of law enforcement reduces the variability of the number of connections, but, as long as the returns to investment in monitoring are decreasing, not all the way to zero. Monitoring increases with inuence, and more powerful mobsters are more likely to be sampled. Solving for the optimal monitoring as a function of the optimal number of connections: exp +1 s1 i q 1 1 + ln q s i ln q
1
1 1+

m = i

= exp

Given that the minimum number of aliates is two,15 monitoring can be set to 0 whenever s 2, or 1 = 0.74 . exp 1/2

q =

In order to weight the data based on the optimal monitoring ideally we would need to know , but Figure 6 shows that once the weights are normalized to sum up to one,
In particular, disregarding the members whose connections have been in part or fully blanked out, Joseph Falcone, Robert Ansoni, Jack Cerone, Frank Napoli, and Mariano Paolacci have just two connections, while none of the mobsters have just one connection.
15

16

wi =

m1 i 1 N i=1 mi

, dierences in generate very dierent levels of monitoring, mi , but not

very dierent weights wi . Not only has little eect on the weighting, Figure 7 shows that Markov chain weights and optimal monitoring weights have a similar hyperbolic shape, with the main dierence lying where the model predicts no monitoring at all and where we put the monitoring at it minimal level.

Descriptive Evidence About Cosa Nostra and Its Members

From now on mobsters are going to be described with and without correcting for the non-random sampling design. The corrections are done weighting the data based on the weights shown in Figure 6 and 7. Non-weighting corresponds to the case when = .

4.1

Individual Characteristics of the Members

Before analyzing how criminals are connected within the Maa it is instructive to describe the members based on the information contained in the criminal records. Let me rst date the data. Given that the distribution of the year of rst arrest has basically full support within the range 1908-1960 (the only year without a rst arrest is 1910) one can infer that the data refers to what the authorities knew in 1960.16 The records do not report any deaths, thus dont include big bosses that were killed before 1960, i.e. Albert Anastasia boss of one of the 5 New York City families, the Gambino family.17 Table 1 shows that the average age is 52 years, with the oldest mobster being 81 years old, and the youngest one 23. Correcting for the sampling with two very dierent s1 and 15decreases the
Additional evidence is the following description in Michael Russos le: Recently (1960) perjured himself before a Grand Jury in an attempt to protect another Maa member and narcotic tracker. 17 His brother Anthony Tough Tony, instead, was killed in 1963 and is in the records.
16

17

average age by very little. Half of the mobsters reside in either New York, or in New Jersey, and probably entered the U.S. through Ellis Island. Indeed, 29 percent were born in Sicily and another 10 percent in other regions of Italy. Properly weighting the data fewer turn out to be born in Sicily, as Sicilians tend to have leadership positions and are thus oversampled, irrespectively of the weighting procedure that is used. Most remaining mobsters were born in the United States but were of Italian origin as this was a prerequisite to become a member. 73 percent of members are married (70 percent when weighting), but only 60 percent of these are reported to have children. The overall average number of children is 1 and is 2.14 among members with children, equally divided between sons and daughters. 18 percent of members are married to someone who shares her maiden name with some other member (Connected wife), though fewer are when weighting (15 percent when the economic model is used, 14 percent when the Markov chain weights are used). These marriages are presumably endogamous within the Maa. Observe that Im understating the percentage of marriages within the Maa as some Maa surnames might be missing in the data.18 The FBN reports an average of 1.97 siblings per member, while the average number of recorded members that share the same surname is 1.62. These numbers tend to be smaller when weighting. The average height is 5.6 feet, the average weight is 176 pounds, no matter how the weighting is done.19 Mobsters criminal career starts early. They are on average 25 years old when they end up in jail for the rst time, and the majority has committed some violent crime. Only 16 percent do not have an arrest record. I dont know the total number of crimes committed by the mobsters but I know in how many dierent types of crime they have apparently been involved. This number varies between 0 and 9 and the average is 2.58. I also know the number of dierent legal businesses they have interest in. This number varies between 0 and 5 and is on average equal to 1.
While it is also possible that some women might have a Maa surname without being linked to any Maa family, this is very unlikely conditional on being married to a Maa associate. 19 As a note, 18 percent of the mobsters are obese and 58 percent overweight.
18

18

Another variable summarized in Table 5 is the interaction index. The index measures the exposure to what Bonanno (1983) calls, in uppercase, Tradition or Hess (1973) calls, in lowercase, maa, or maa culture to distinguish it from Maa the organization. Looking at Figure 9 helps explaining the index. It shows the current distribution at the zip code level of the members surnames in Italys phone directory.20 Each circle is proportional to the number of surnames present within each zip code. Not surprisingly many surnames show up in Sicily, in Naples, and in Calabria. Many of these surnames appear also in large cities that were subject to immigratory ows from the south, like Milan, Rome, and Turin. For each members surname I computed the probability that it shares a randomly chosen zip code located in the South of Italy with other surnames from the list. To be more precise, the index for member i is equal to 100,000 times the sum across zip codes j of the fraction of surnames of member i present in zip codes j times the fraction of surnames of the other members (i) in the same zip code: #surnamei,j j #surnamei,j #surnamei,j . j #surnamei,j

interactioni = 100, 000

j

(4)

The advantage of this index is that I can computed it for all surnames while information about the Italian community of origin would only be available for those born in Italy. The average index is equal to 2.67 per 100,000, though its lower when weighting the data (2.01). Ten percent of the times the index is zero, either because the zip codes do not overlap or because the surname is not in the phone directory. Tables 2 and 3 show the list of legal and illegal activities that at least 5 percent of members were involved in. Weighting does little to the distribution of legal and illegal activities. Most mobsters owned restaurants, drugstores or were otherwise involved with the supply of food. Real estate, casinos, car dealerships, and import-export were also common businesses. Drug tracking is the most common of the illegal activities (43
20

Unfortunately I could not nd the distribution of surnames in 1960.

19

percent), perhaps partly because the information was gathered by the FBN. Twentysix percent of members were involved in robberies and 23 percent in murders. Weapon oences and assaults are also quite common. Some crimes that are typically associated with organized crime, like gambling, extortions, and liquor oences (during prohibition) are highly represented as well.

Validity of Network-based Measures of Importance

Before moving towards the second part of the paper, where I try to predict mobsters power inside the criminal organization, Im going to validate the use of network-based power measures. In the introduction I mentioned that connections were the building block of the Maa, and that only directly connected mobsters were allowed to openly talk about their secret organization. Each criminal record contains a list of criminal associates. Figure 1 indicates, for example, that Joe Bonanno was associated with Luciano, Costello, Profaci, Corallo, Lucchese, and Galante. There is no evidence about how the FBN established such associations, and why they were restricted to be 5. But it seems apparent that each prole lists its highest associates. Notice that the mere keeping track of connections shows that even the FBN understood how important these connections were. Indirected connections are clearly more numerous, as mobsters can be listed as associates in several records. Hence, I dene two mobsters to be connected whenever at least one mobster lists the other mobsters last name in his record.21 Table 4 shows the list of members with the 10 highest and 10 lowest number of direct connections.22 For each mobster the FBN records contain a paragraph about his activities within Cosa Nostra. In
In other words, I construct a symmetric adjacency matrix of indirected connections. Since connection are based on surnames members with the same surname will share the same degree. This introduces some noise but dealing with the large variation in rst names, i.e. Antonio/Tony/Anthony, would introduce even more noise in graphing the network.
22 21

20

order to extract information on the level of importance of these criminals the variable Top counts the number of times the words boss, highest, most, head, and top are cited and High the number of times high, inuential, important, leader, leading, powerful, and representing are cited. The Apalachin variable indicates whether the mobster attended the important 1957 Maa meeting in Upstate New York. The last column represents the historically reconstructed position within the Maa. Criminals with many connections are more likely to be recognized as high-ranked members, and more likely to have attended the 1957 meeting. Several members in the top distribution of the number of connections are bosses, i.e. Salvatore Lucania, alias Lucky Luciano, Vito Genovese, Antoni Accardo, Joe and Joseph Profaci. Salvatore Santoro and Salvatore Vitale were instead underbosses of the Lucchese and Bonanno Family. Criminals with the lowest degrees, instead, are mostly soldiers. The number of connections, called degree in network analysis, is clearly the easiest way to measure the importance of members, but in recent years social network theorists proposed dierent centrality measures to account for the variability in network location across agents, and there is no systematic criterium to pick up the right centrality measure for each particular situation (Borgatti, 2003, Wasserman and Faust, 1994).23 Before showing the distribution of some of these measures I briey describe them (see Appendix A for a more technical denition). Unlike degree, which weights every contact equally, the eigenvector index weights contacts according to their centralities. The index takes direct as well as indirect connections and thus the whole network into account.24 The closeness index represents the inverse of the average distance between a node (a member) and all the other nodes, and is a good measure for how isolated members are. The betweenness index measures the number of times a node is on the shortest path between two randomly chosen nodes, and is a good measure for the members capacity to
See also Sparrow (1991) for a discussion on centrality indices in criminal networks. As rst noted by Granovetter (1973), weak ties (i.e. friends of friends) are important source of information. See Patacchini and Zenou (2008) for the role of weak ties in explaining criminal activities.
24 23

21

act like a bridge between clusters of members. These measures of individual centrality allow me to consider dierent nuances in the denition of a Maa leader. The data also allows me to construct a qualitative indicator of importance, based on the union of the Top and High variables discussed in this section. Table 5 collects descriptive statistics about various indicators of importance.25 Weighting the average number of connections mobsters have drops from 11 to 7, and more generally all centrality measures decrease once we control for the non-random nature of the sampling design, no matter the weighting procedures that is used. The non-standardized degree varies between 1 and 71, and Figure 5 demonstrates that the corresponding density is positively skewed. The eigenvector index (centrality) has a density that is very similar to that of degree, while the density of closeness is more symmetrically distributed, meaning that most mobsters are neither too isolated nor too close within the network. The density of betweenness, instead, shows that very few mobsters represent bridges between subsets of the network, most likely Families. Finally, fty-ve percent of the mobsters appear to be high-ranked, meaning that their descriptions includes at least a word that implies leadership. The list of the names of mobsters with the highest and the lowest degree provides only limited evidence that connections matter. For this reason I used historical sources to collect information on the highest ranks achieved by the mobsters up until these days. Ideally one would like to reconstruct their rank in 1960, but this tuned out to be impracticable. Table 6 shows that 70 percent of mobsters are soldiers or their rank could not be found. Both type of mobsters have the lowest centrality measures, and degree is close to 10, which happened to be the typical number of soldiers under the control of captains, called capodecina literally head of ten. The third largest group of mobsters are bosses, which suggest that over time several underbosses, captains, and Consiglieri became dons. Bosses and underbosses have the highest betweenness index, which is
25

The individual centrality measures have been normalized to be between 1 and 100

22

consistent with bosses acting as bridges across Maa Families. Consiglieri, instead, have the highest degree, which is consistent with the role of counselor for the entire Family, but with little bridging capacity across Families. Dierent measures of centrality thus seem to capture dierent roles within the criminal organization. Information on the place of residence of the mobsters allows me to perform a nal validity test of centrality measures. One would expect more powerful mobsters to live in richer homes. While ideally one would like to have information on the value of these homes in 1960, I could only collect information on their value today. Figure 4 shows that the value is increasing with network centrality, and Table 7 shows that this increase is robust to the inclusion of state xed eects and other regressors. The Table shows the coecients from a regression of the log eigenvector index on the log of the house value. Estimated elasticities are close to 10 percent. The last three columns replicate the regressions using as a measure of importance a dummy equal to one when mobsters description contains one of these words: boss, highest, most, head, or top. Conditional on state xed eects the coecients are less precisely estimated, suggesting that network based measures might perform better measures based on the description of the mobsters.

Which Characteristics Predict the Importance of a Maa Member?

Now that connections have been shown to matter, the next step will be to highlight the forces that might inuence p, and, therefore, their number. While for the sake of simplicity the model does not take into account the quality of the links, that there are several ways to measure how central, or how connected a member is within the network. I start my regression analysis using the eigenvector index as dependent variable, as it depends both on the number and the quality of its connections, on the whole sample of mobsters.

23

6.1

Descriptive Evidence

Before moving to the regressions a graphical analysis can shed light on some organizational rules of the Maa. Figure 10 shows that all centrality measures grow steadily with age until age 60 and later starts decreasing. Like typical earning proles there is an inverse U-shaped relationship between centrality and age, but while for earnings proles the peak is typically around 45 or 50, in the Maa . This nding conrms Ianni and Reuss-Ianni (1972, pg.130)s anthropologic results about the importance of age in determining the leadership positions. The dierence between the minimum age (23) and the maximum age (60) doubles the degree, triplicates the eigenvector index, and more than triplicates betweenness. Closeness, instead, shows a very steep increase up to age 30 and then attens out, but the overall increase is more modest. Figure 11 presents how the interaction index, the continuous proxy for maa culture inuences centrality. The overall patterns are less clear than for age, but when interactions are very high all centrality measures appear to increase as well. Eigenvector centrality is low in the absence of daughters or sons, but sons and daughters seem to be equally valuable in increasing the fathers centrality (Figure 12). I will say more about this later when I analyze the importance of intermarriage. Figure 13 shows that the number of types of crimes and businesses also positively inuence mobsters centrality within the network, especially up to the 4th type. Above 4 types of crimes allegedly committed the evidence is more noisy as fewer observations are available. I present how categorical variables inuence network centrality in the next Section, where I move to a regression framework.

6.2

The Eigenvalue Index

In Table 9 I use the eigenvector centrality measure as the dependent variable, though more generally I assume the following linear relationship between the k-th index of importance 24

ck of individual i and his observable characteristics Xi : ck = Xi + ei , k = 1, ..., 4.26 i i Each regression controls for the number of mobsters that in the data share the same surname. I call this variable Extended family members despite the possibility that some of these mobsters might not be related to each other. Given that the connections are based on surnames I add that this variable is to control for the mechanical eect that an increase in the number of mobsters might have on the centrality measures. The coecient is around, which is also the average number of associates contained in each criminal le. In the rst column I regress the eigenvector index on individual characteristics that, despite the possibility of selection bias, are less likely to be endogenous. Age enters quadratically. Peak centrality is reached at age 59 and both terms are signicant. The nding of a peak late in life is consistent with Ianni and Reuss-Ianni (1972)s account of hierarchies based on generations (though it might also reect that more able mobsters are more likely to be alive). Weighting does little to the age at which there is a peak. Another variable that has a signicant positive inuence on centrality is being born in Sicily (a 20 percent increase), independent of the weighting used. This means that not only is the American Maa an Italian enterprise, it values direct links with Sicilians more than with people from other parts of Italy. Sicilian kin-centered social system, with its code of honor and vow of silence, forms the building block for the Maa. Bonanno would write that among Italians he felt safe only around Sicilians. Moreover, nativity does not fully capture adherence to the maa code of law as 60 percent of the mobsters were born in the U.S. This is why I use the interaction index, which depends on the geographical distribution of last names, to proxy for community ties and exposure to the Tradition. The index has a positive and signicant eect on network centrality, especially after weighting the data. Members from Sicily and those with stronger ties to the Maa culture are likely to be trusted more, as they are more likely to adhere to the omert`, increasing p. Body a
26

See Appendix A for a detailed description of the dierent indices.

25

weight of the mobster does not inuence his status, while his height does but only after weighting the data. In column 4, 5 and 6 I control for additional factors, like the nuclear family structure of the mobster, that might in part be endogenous. The purpose is thus merely descriptive, and useful for predictive purposes. Each additional child increases the eigenvector index by 0.8, but the eect gets weaker when weighting the data. Children thus seem to be more valuable for higher ranked mobsters. It doesnt matter whether the child is male or female. This nding challenges the criminological view that within the Maa male children are more valuable than female ones because they represents potential workforce. The leading explanation for this nding is that in a male only society like the Maa connected girls (probably in excess demand) could be married strategically. The number of siblings, and being married or divorced does not inuence the centrality index. This is in line with Falcone and Padovani (1991, pg. 113)s view that unlike the Maa in Italy the American Maa adopted a more liberal view towards divorce. Being married to a wife who is connected, instead, leads to an almost 50 percent increase in the index (slightly less when weighting). These ndings, again, seem to suggest that higher p, due to more trusted links, increase the optimal number of connections. I cannot rule out the possibility that more connected mobsters are also more likely to nd, or be given, a connected wife. While these alternative interpretations do not matter when the only purpose is to discover the leading gures within the Maa, one has to be careful in giving a causal interpretation to these estimates. The leading mobsters are active in New York and New Jersey. The remaining two variables measure how many dierent types of crime and dierent types of businesses the mobster was involved in. Both variables are positive. This is consistent with more able criminals being more able to diversify risk. But the number of business types (+16 percent) are a better predictor for key players than the number of crime types (+4 percent). The combined regressors explain almost 30 percent

26

of the variability of the eigenvector index. Criminals who were known to have committed violent crimes have a larger index, but the eect is not signicant. If being violent reduces the risk of betrayal this nding is consistent with my model. An arrest increases the importance, though again, the eect is not signicant, and endogeneity might mean that higher ranked individuals might be more likely to be arrested. Age at rst arrest is negative, showing that not only age, but also experience increases the level of centrality in the network (young members are often recruited in jail; Ianni and Reuss-Ianni, 1972, pg. 45), but this eect is also not signicant once I control for all the other factors.

6.3

Other Network Centrality Indices

Eigenvector centrality represents only one way to measure centrality in the network. Other indices capture dierent nuances of centrality. In Table 10 I look at dierent measures of centrality and importance using the same specication used in the last column of Table 9. For brevity I only show the results that control for monitoring with = 15. In order to highlight how dierent measures are able to capture dierent characteristics of the network, and since there is no systematic way to decompose the dierent centrality measures based on the importance of direct and indirect links, I developed a simple statistical way to accomplish the same goal: I simply add as a dependent variable the residual (j ) of a linear projection of the alternative measures of centrality (cj ; j = 1) on i i degree (c1 ): cj = + c1 + j ; j = 1. The coecients on the residuals of a particular i i i i centrality index measure the nuances captured by that particular index with respect to a simple count of direct connections. Column 2 replicates the last column of Table 9 to ease the comparison with the other measures. Comparing columns 1 and 2 shows that the coecients are not very dierent when degree is used to proxy for leadership. The only coecients that seem to be smaller are the ones on residing in New York or New

27

Jersey, and the one on the types of businesses. A simple test for the signicance of these dierences that looks at the residuals conrms that these dierences are signicant. This means that living in New York and New Jersey, and the number of businesses increase the eigenvector centrality not only through the direct links but also through the indirect links, while for all the other variables that are signicant in column 2 but not in column 3 only direct links matter (connected wife, age, and the interaction index). Closeness, an inverse measure of the average distance from the other members, and betweenness, the ability to build bridges, capture dierent aspects in the denition of individual importance or power within a network. Column 4 shows that overall the coecients for closeness do not dier substantially from those seen before. Closeness is more than any other index dependent on kinship. The Sicily coecient, the interaction index coecient, and the Connected wife coecient are all highly signicant. Moreover, column 5 shows that a large part of these eects is driven by the indirect links (cannot be explained by degree alone). Results that use betweenness as a dependent variable shown in column 6 present a very interesting ndings: there is no evidence that bridges are build through marriages. The coecient on the Connected wife dummy is precisely estimated to be close to zero. Strategic marriages are thus conned to happen within friendly clans, and not across clans that wouldnt otherwise be connected. The number of dierent crimes committed also have no bridging capacity, while businesses do. Experience and age, instead, inuence betweenness beyond the increase in degree. And so does being Sicilian, residing in New York or New Jersey, and being active in several businesses.

28

6.4

Qualitative Measures of Importance

In column 2 of Table 11 I exploit qualitative information on the mobsters contained in the records. In particular, I use the same specication as before but the dependent variable counts the number of times the following words are used in describing a mobster: boss, highest, most, head, top, high, inuential, important, leader, leading, powerful, and representing. Overall the results conrm the importance of directly migrating from Sicily, being old and experienced, residing in New Jersey or New York, being active in many types of businesses, and using violence. But marrying a connected wife, having children, having strong community ties (interactions), and the number of crimes committed are no longer signicantly related to leadership. This lack of signicance leads to a lower R-squared (8 versus 28 percent). While there might be superior ways to extract the information know by the FBN (we did try several qualitative measures), network based measures seem to capture additional attributes of leadership.

6.5

Indegree and Random Truncation of the Sample

Table 12 shows two types of robustness checks. In the rst instead of using indirected links I use only the indegree, meaning the number of times someone appears as an associate in criminal proles of other mobsters. Given that outdegree is somehow bounded it is not surprising that the coecients are driven by indegree, and thus are very close to the ones based on indirected connections shown in the rst column of Table 10. The second and more important robustness check simulates random truncations of the sample. Since the initial sample is likely to be truncated as wellsome scholars estimate that there are as many as 5,000 made men across the U.S, I try to address the possible bias that such a truncation might generate (see Borgatti et al., 2006, for a similar simulation). Columns 3 and 4 show the mean and the median coecients obtained in regressions based on 500 dierent samples where I randomly truncated 50 percent of the observations. Indegree is 29

based on 50 percent of the sample only and, as before, standardized to lie between 0 and 100. Truncation tends to reduce the size of the coecient on the dummy connected wife, which is reasonable given that smaller samples lead to an increase in misclassication error of that particular variable. The other coecients tend to be quite stable.

Concluding remarks

This paper presents the rst thorough micro-level analysis of the US Maa network. Beside testing sociological and historical views about the functioning of these criminal networks, I develop a simple model of connections that highlights the trade-o of connections: they increase prots and power but also the risk of detection. More connections mean more potential informers. Monitoring by law enforcement is shown to depend on the number of connections, which allows me to control for the non-random sampling in the empirical section. Family ties, violence, and exposure to Maa culture reduce the probability of defection and increase the number of connections. Unlike economic organizations hierarchies depend crucially on kinship. My results highlight how intermarriage shapes the network. Women are used to foster the Familys network centrality, but only within trusted Families. Woman are not used to bridge Families that are not otherwise closely connected. Trust shapes the network. Where values are shared and the maa culture is strong, connections are more stable and thus more numerous. Coherent with maa culture, how central members are within the network increases steadily with age. And more central members have more businesses, legal and illegal ones. If social connections are the driving forces of the phenomenon under consideration and if their structure is non-random, as in my case, a detailed study of the characteristics of the network reveals some relevant features of social structure that can guide crime prevention policies. A targeted policy identifying key players in a given area may be an eective way 30

to reduce crime (Ballester et al., 2006, Rka et al., 2000). A key player is an individual e belonging to a network of criminals who, once removed, leads to the highest aggregate delinquency reduction. In practice, the planner may want to identify optimal network targets to concentrate (scarce) investigatory resources on some particular individuals, or to isolate them from the rest of the group, either through leniency programs, social assistance programs, or incarceration. A necessary condition for designing such policies is the ability to map and identify a social network structure. Detailed information about the hierarchy might even allow enforcement agents to break the chain of command by arresting sets of mobsters (Farley, 2003). Only detailed empirical studies on real-world social networks can provide guidance in this direction.

31

Centrality Measures

Let N = {1, . . . , n} be a nite set of agents in network g. Let me dene G as the nsquare adjacency matrix of the network g , i.e. the matrix that keeps track of the direct connections in this network, where gij = 1 if i and j are directly linked, and gij = 0, otherwise.27 The simplest index of connectivity is the number of direct links stemming from each agent i in the network, i.e. degree centrality:
n

c1 i

=
j=1

gij .

The denition of centrality is thus based on the number of direct links only. A variant of simple degree is eigenvector centrality, which also takes into consideration indirect links:
n

c2 = 1 i
j=1

gij c2 j

where is the highest eigenvalue of matrix G. The formula implies (recursively) that the centrality of individual i is proportional to the sum of centralities of the individuals she/he is connected to. It thus can be high even if she/he has low degree. The standard measure of closeness centrality of individual i is given by: 1
n j=1 dij

c3 = i

where dij is the geodesic distance (length of the shortest path) between individuals i and j.28 As a result, the closeness centrality of individual i is the inverse of the sum of geodesic
Measurement errors in the denition of the actual connections are clearly unavoidable in this study. Some names are blanked out and many lower-ranked maosi were not led. 28 The length of a shortest path is the smallest k such that there is at least one path of length k from i to j. I can identify such a length by computing G, G2 , G3 , ..., until I nd the rst k such that the (i, j)th entry of Gk is not zero.
27

32

distances from i to the n 1 other individuals and can be regarded as a measure of how long it will take information to spread from a given member to other members in the network. Betweenness indexes derive from the number of optimal paths across (or from) every node. It can be dened as:
n

c4 = i
j,l

ajl,i ajl

where j and l denote two given agents in g, ajl,i is the number of shortest paths between j and l through i, and ajl is the number of shortest paths between j and l.

33

References
MAFIA: The Governments Secret File on Organized Crime. By the United States Treasury Department, Bureau of Narcotics. HarperCollins Publishers, 2007. Howard Abadinsky. Organized Crime. Wadsworth Publishing, 1990. Mariagiovanni Baccara and Heski Bar-Isaac. How to Organize Crime. Review of Economic Studies, 75(4):10391067, 2008. Wayne E. Baker and Robert R. Faulkner. The Social Organization of Conspiracy: Illegal Networks in the Heavy Electrical Equipment Industry. American Sociological Review, 58(6):pp. 837860, 1993. URL http://www.jstor.org/stable/2095954. Coralio Ballester, Antoni Calv-Armengol, and Yves Zenou. Whos who in networks. o wanted: The key player. Econometrica, 74(5):14031417, 09 2006. Oriana Bandiera. Land reform, the market for protection, and the origins of the Sicilian maa: theory and evidence. Journal of Law, Economics, and Organization, 19(1):218, 2003. Marianne Bertrand and Antoinette Schoar. The role of family in family rms. Journal of Economic Perspectives, 20(2):7396, Spring 2006. Joseph Bonanno. A Man of Honor: The Autobiography of Joseph Bonanno. St. Martins, 1983. Stephen P. Borgatti, Kathleen M. Carley, and David Krackhardt. On the Robustness of Centrality Measures Under Conditions of Imperfect Data. Social Networks, 28(2): 124136, 2006. Steve P. Borgatti. The Key Player Problem, pages 241252. R. Breiger, K. Carley and P. Pattison, Committee on Human Factors, National Research Council, 2003. 34

Yann Bramoull and Bernard Fortin. The Econometrics of Social Networks. New Palgrave e Dictionary of Economics, steven n. durlauf; lawrence e. blume edition, 2010. Nigel Coles. Its Not What You Know-Its Who You Know That Counts. Analysing Serious Crime Groups as Social Networks. British Journal of Criminology, 41(4):580, 2001. David Critchley. The Origin of Organized Crime in America: The New York City Maa, 18911931. Routledge, 2009. Scott H. Decker, Tim Bynum, and Deborah Weisel. A Tale of two Cities: Gangs as Organized Crime Groups. Justice Quarterly, 15(3):395425, 1998. Giovanni Falcone and Marcelle Padovani. Cose di Cosa Nostra. Rizzoli, 1991. Jonathan D. Farley. Breaking Al Qaeda cells: A mathematical analysis of counterterrorism operations (A guide for risk assessment and decision making). Studies in Conict & Terrorism, 26(6):399411, 2003. Gianluca Fiorentini and Sam Peltzman. The Economics of Organised Crime. Cambridge University Press, 1997. Diego Gambetta. The Sicilian Maa: the business of private protection. Harvard Univ Press, 1996. Nuno Garoupa. Optimal law enforcement and criminal organization. JourURL

nal of Economic Behavior & Organization, 63(3):461474, July 2007. http://ideas.repec.org/a/eee/jeborg/v63y2007i3p461-474.html.

Martin A. Gosch and Richard Hammer. The Last Testament of Lucky Luciano. Little, Brown, 1975.

35

Mark S. Granovetter. The strength of weak ties. American Journal of Sociology, 78: 13601380., 1973. Luigi Guiso, Paola Sapienza, and Luigi Zingales. Does Culture Aect Economic Outcomes? The Journal of Economic Perspectives, 20(2):2348, 2006. Henner Hess. Maa and Maosi: The structure of power. Ashgate, 1973. Francis A.J. Ianni and Elisabeth Reuss-Ianni. A Family Business: Kinship and Social Control in Organized Crime. Russell Sage Foundation, 1972. James B. Jacobs and Lauryn P. Gouldin. ter? Cosa nostra: The nal chapURL

Crime and Justice, 25:pp. 129189, 1999.

ISSN 01923234.

http://www.jstor.org/stable/1147609. Peter Klerks. The Network Paradigm Applied to Criminal Organisations. Transnational Organised Crime: Perspectives on Global Security, page 97, 2003. Valdis E. Krebs. Mapping Networks of Terrorist Cells. Connections, 24(3):4352, 2002. Vimal Kumar and Stergios Skaperdas. Criminal Law and Economics, chapter On The Economics of Organized Crime. Edward Elgar Pub, 2009. Steven D. Levitt and Sudhir Alladi Venkatesh. An economic analysis of a drug-selling gangs nances. The Quarterly Journal of Economics, 115(3):755789, August 2000. Don Liddick. The Enterprise Model of Organized Crime: Assessing Theoretical Propositions. Justice Quarterly, 16(2):403430, 1999. Salvatore Lupo. Quando la Maa Trov` lAmerica. Einaudi, 2009. o Peter Maas. The Valachi Papers. Putnam, New York, 1968.

36

Jean Marie McGloin. Policy And Intervention Considerations of a Network Analysis of Street Gangs. Criminology & Public Policy, 4(3):607635, 2005. John C. McWilliams. The Protectors: Harry J. Anslinger and the Federal Bureau of Narcotics, 1930-1962. Univ of Delaware Press, 1990. Carlo Morselli. Career opportunities and network-based privileges in the Cosa Nostra. Crime, Law and Social Change, 39(4):383418, 2003. Mangai Natarajan. Understanding the Structure of a Drug Tracking Organization: A Conversational Analysis. Crime Prevention Studies, 11:273298, 2000. Mangai Natarajan. Understanding the Structure of a Large Heroin Distribution Network: A Quantitative Analysis of Qualitative Data. Journal of Quantitative Criminology, 22 (2):171192, 2006. Eleonora Patacchini and Yves Zenou. The strength of weak ties in crime. European Economic Review, 52:209236, 2008. Alejandro Portes. Social Capital: Its Origins and Applications in Modern Sociology. Annual Review of Sociology, 24:124, 1998. Albert Rka, Hawoong Jeong, and Albert Lszl Barabsi. Error and Attack Tolerance e a o a of Complex Networks. Nature, 406:378382, 2000. Peter Reuter. Disorganized Crime: the Economics of the Visible Hand. MIT press, 1983. Peter Reuter. The organization of illegal markets: An economic analysis. 1985. Peter Reuter. Handbook of organized crime in the United States, chapter Research on American Organized Crime. Greenwood Press, Westport, CT, 1994. Jerzy Sarnecki. Delinquent Networks in Sweden. Journal of Quantitative Criminology, 6 (1):3150, 1990. ISSN 0748-4518. 37

Jerzy Sarnecki. Delinquent Networks: Youth Co-oending in Stockholm. Cambridge Univ Pr, 2001. ISBN 0521802393. Friedrich Schneider and Dominik H. Enste. Shadow economies: Size, causes, and consequences. Journal of Economic Literature, 38(1):77114, March 2000. Malcolm K. Sparrow. The Application of Network Analysis to Criminal Intelligence: an Assessment of the Prospects. Social Networks, 13(3):251 274, 1991. John R. Wagley. Transnational Organized Crime: Principal Threats and US Responses. Congressional Research Service Report for Congress RL33335, March 2006. Stanley Wasserman and Katherine Faust. Social Network Analysis. Methods and Applications. Cambridge: Cambridge university press edition, 1994. Phil Williams. Transnational Criminal Networks. Networks and netwars: the future of terror, crime, and militancy, page 61, 2001. George Wolf and Joseph DiMona. Frank Costello: Prime Minister of the Underworld. Morrow, 1974.

38

Figure 1: Record Number One: Joe Bonanno

39

Figure 2: The Bonanno Family.

Source: Valachis testimony in the McClellan Hearings.

40

Figure 3: Surveillance pictures of Nick Giso in 1980, and Sammy Gravano in 1988

700000

600000

400000

300000

4 p_sdeigen

10

12.2 0

12.4

logHouse value 12.6 12.8

House value 500000

13

13.2

4 p_sdeigen

10

Figure 4: Housing values and centrality

Notes: The gures show house values by eigenvector deciles together with the 95 percent condence interval.

41

.08

Density .04 .06

.02

20

40 Degree

60

80

0 0

.02

Density .04 .06

.08

.1

20

40 60 Centrality (std.)

80

100

Density .02 .03 .04

.05

.01

40

60 80 Closeness (std.)

100

0 0

.05

Density .1 .15

.2

20

40 60 Betweenness (std.)

80

100

Figure 5: Densities of Centrality Indices

.003

.0025

.3

.0015 .002 weight

.1

.001

.0005

20

40 Degree

60

80

20

40 Degree

60

80

Figure 6: Monitoring, Density (left axes) and Weighting (right axes)

Notes: equal to 1, 10, and 15. The weighting functions are decreasing, the monitoring ones increasing in degree.

42

.0005

.001

.02

.0015 .002 weight

density .04 .06

monitoring .2

.0025

.08

.003

.4

.1

0 0

.002

.004

.006

20

40 Degree

60 gamma=2 MC weights

80

Optimal monitoring weights: gamma=15 gamma=1

Figure 7: Weighting based on MC and Optimal Monitoring

Notes: are equal to 1, 2, and 15. MC weights indicate the Markov Chain based weights.

Density 1

1.5

.5

.5

.6

.7 .8 Probability of success p_0

.9

0 0

.1

Density .2

.3

.4

10

20

30 Connections

40

50

Figure 8: Simulations of Degree

Notes: is equal to 0.5, p and q are set to be the same.

43

36 5

38

40

longitude 42

44

46

10 latitude

15

20

Figure 9: Geographical Distribution of Maa Surnames.

Notes: Each circle represents a zip code. The size of the circles is proportional to the number of US Maamembers surnames found in todays Italian phone directory. The plot shows only 20 percent of the distribution of surnames.

44

Degree
20 20

Eigenvector

15

20

40 Age

60

80

10 20

Centrality (std.) 10

Degree (std.) 5 10

40 Age

60

80

kernel = epanechnikov, degree = 0, bandwidth = 2.66, pwidth = 3.99

kernel = epanechnikov, degree = 0, bandwidth = 2.84, pwidth = 4.25

Closeness
70 8

Betweenness

60

30

20

20

40 Age

60

80

2 20

Betweenness (std.) 2 4

Closeness (std.) 40 50

40 Age

60

80

kernel = epanechnikov, degree = 0, bandwidth = 2.42, pwidth = 3.64

kernel = epanechnikov, degree = 0, bandwidth = 3.53, pwidth = 5.29

Figure 10: Centrality and Age

Notes: The gures show Kernel-weighted local linear regression and the corresponding 95 percent condence interval.

45

Degree
60 50

Eigenvector

40

20

5 Interaction index

10

15

0 0

10

Centrality (std.) 20 30

Degree (std.) 20

40

5 Interaction index

10

15

kernel = epanechnikov, degree = 0, bandwidth = .64, pwidth = .96

kernel = epanechnikov, degree = 0, bandwidth = .74, pwidth = 1.11

Closeness
80 40

Betweenness

50

5 Interaction index

10

15

10 0

40

Betweenness (std.) 10 20 30

Closeness (std.) 60 70

5 Interaction index

10

15

kernel = epanechnikov, degree = 0, bandwidth = .92, pwidth = 1.38

kernel = epanechnikov, degree = 0, bandwidth = .73, pwidth = 1.09

Figure 11: Centrality and Tradition

Notes: The gures show Kernel-weighted local linear regression and the corresponding 95 percent condence interval. The interaction index has been truncated at the 95th percentile.

46

4060 percentile range

25 5 0 10 15 20

2 40/60 sons

4 40/60 daughters

Figure 12: Centrality, Sons and Daughters

Notes: Each line represent the 40-60 percentile range.

47

4060 percentile range

20 5 0 10 15

4 40/60 crimes

8 40/60 businesses

10

Figure 13: Centrality, Types of Crimes and of Businesses

Notes: Each line represent the 40-60 percentile range.

48

49

Table 1: Summary Statistics of Individual Characteristics

Variable Weighting Born in the U.S. Born in Italy (except Sicily) Born in Sicily Age Height in feet Weight in pounds 0.59 0.19 0.29 52.17 5.61 176 Mean Std. Dev. = 0.49 0.39 0.45 10.04 0.20 27 Mean Std. Dev. Mean Std. Dev. Min Max Obs

= 15 MC sampling Variables related to the PERSON 0.60 0.49 0.59 0.49 0.21 0.41 0.22 0.42 0.24 0.43 0.25 0.43 51.58 10.02 51.98 10.41 5.60 0.20 5.60 0.20 175 28 175 28 Variables related to the FAMILY 2.01 4.39 1.80 3.74 0.70 0.46 0.70 0.46 0.05 0.23 0.06 0.24 0.15 0.36 0.14 0.35 0.94 1.40 0.97 1.40 0.50 0.38 0.49 0.38 1.93 2.09 1.87 2.07 1.39 0.86 1.34 0.77 Variables 0.42 0.06 0.61 25.14 0.17 2.53 1.01 related to the ACTIVITIES 0.49 0.43 0.50 0.24 0.06 0.24 0.49 0.62 0.49 9.29 25.19 9.50 0.37 0.17 0.37 1.70 2.52 1.76 0.94 1.03 0.93

0 0 0 23 5 95

1 1 1 81 6.25 365

801 801 801 801 790 790

Interaction index Married Divorced Connected wife Number of children Fraction of daughters Siblings Extended family members

2.67 0.73 0.05 0.18 1.02 0.49 1.97 1.62

5.83 0.44 0.22 0.38 1.44 0.37 2.11 1.05

0 0 0 0 0 0 0 1

53.91 1 1 1 8 1 11 6

801 801 801 801 801 352 801 801

50

Resides in NY Resides in NJ Violent crimes Age of rst arrest Never arrested Types of crimes committed Types of businesses

0.43 0.06 0.63 25.02 0.16 2.58 1.07

0.50 0.24 0.48 9.06 0.37 1.70 0.97

0 0 0 8 0 0 0

1 1 1 67 1 9 5

801 801 801 688 801 801 801

Table 2: Summary Statistics of Legal Businesses Variable Drugstores Restaurants Food companies Manual laborer Casinos Real estate Import export Car dealer Mean 0.18 0.09 0.09 0.07 0.07 0.05 0.05 0.05 Std. Dev. Mean 0.17 0.09 0.08 0.09 0.05 0.05 0.05 0.05 Std. Dev. Mean Std. Dev.

= 0.38 0.31 0.28 0.26 0.25 0.23 0.22 0.22

= 15 0.37 0.30 0.28 0.28 0.23 0.22 0.22 0.21

MC sampling 0.17 0.37 0.09 0.29 0.08 0.28 0.09 0.29 0.05 0.23 0.06 0.24 0.05 0.22 0.04 0.20

Table 3: Summary Statistics of Crimes Variable Drug oenses Robbery Murder Weapon oenses Simple assault Larceny Burglary Gambling Liquor oenses Extortion Counterfeiting Mean Std. Dev. Mean Std. Dev. Mean Std. Dev.

= 0.43 0.50 0.26 0.44 0.23 0.42 0.22 0.42 0.21 0.41 0.20 0.40 0.13 0.34 0.13 0.33 0.13 0.34 0.07 0.25 0.07 0.25

= 15 0.46 0.50 0.27 0.44 0.19 0.40 0.21 0.41 0.21 0.41 0.20 0.40 0.13 0.34 0.13 0.34 0.12 0.32 0.06 0.23 0.07 0.26

MC sampling 0.44 0.50 0.27 0.44 0.20 0.40 0.21 0.41 0.22 0.42 0.20 0.40 0.14 0.34 0.14 0.35 0.11 0.31 0.06 0.24 0.08 0.27

51

Table 4: List of Criminals with the Highest and the Lowest Degree Lastname Lucania Ormento Accardo Accardo Genovese Genovese Coppola Coppola Coppola Strollo Profaci Profaci Santoro Vitale Vitale Name Salvatore John Settimo Antonio Vito Michael Michael Frank Stephen Antonio Frank Joseph Salvatore Vito Salvatore Degree Top High Apalachin Rank (Family) Criminals with the highest 15 degrees 71 1 1 0 Boss 71 0 2 1 Caporegime (Lucchese) 64 1 2 0 Caporegime (Lucchese) 64 0 1 0 Boss 55 1 0 1 Underboss (Luciano) later Boss 55 0 1 1 Boss 54 2 1 0 Caporegime (Genovese) 54 0 5 0 Deported to Italy 54 0 1 0 Soldier (Maggadino) 47 2 2 0 Caporegime (Genovese) 44 1 1 0 Boss 44 1 2 1 Boss (son of Frank) 42 0 0 0 Underboss (Lucchese) 40 1 1 0 Boss in Sicily 40 1 0 0 Underboss (Bonanno)

Criminals with the lowest 15 degrees Castorina Vincent 1 0 0 Kornhauser Max 1 0 2 Bibbo Nicholas 1 0 0 Virusso Santo 1 0 1 Mandala Nicholas 1 0 0 Roberto Dominick 1 0 1 Simoni Pierre 1 0 0 Candelmo John 1 0 1 Colombo Frank 1 0 0 Bongiorno Frank 1 0 1 Amari Philip 1 0 1 Peloso Antonio 1 0 0 Labarbara Joseph 1 0 0 Pine Grace 1 1 1 Valle Alarico 1 1 0

0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

52

Table 5: Summary Statistics of Network-related Variables

Variable Mean Std. Dev. = Degree Degree (std.) Centrality (std.) Closeness (std.) Betweenness (std.) Top ranked Top rank citations 11.14 14.49 12.57 52.53 5.10 0.55 0.85 9.55 13.64 14.16 15.25 9.46 0.50 1.02 Mean Std. Dev. Mean Std. Dev. Min Max

= 15 MC sampling Measures of centrality 7.28 7.37 6.46 5.79 8.98 10.54 7.80 8.27 7.82 10.70 6.93 8.57 45.52 14.86 44.39 14.33 2.78 6.59 2.01 4.45 0.51 0.50 0.52 0.50 0.75 0.94 0.77 0.95

1 0 0 0 0 0 0

71 100 100 100 100 1 7

Table 6: Average Centrality by Rank rank unknown soldier captain consigliere underboss boss N. obs. 336 232 79 13 24 118 degree 12.99 14.44 19.11 24.16 22.24 21.29 eigenvector 9.90 11.44 17.35 24.02 16.61 17.20 closeness 71.11 72.19 75.52 78.65 75.89 75.33 betweenness 4.06 3.80 6.48 6.76 9.88 8.50

Table 7: Housing value regressions

(1) (2) (3) (4) (5) log House value 0.085* (0.043) 0.248** (0.104) 0.899*** (0.104) 0.385** (0.154) yes no 610 0.222 0.220** (0.094) 0.930*** (0.104) 0.447*** (0.135) yes no 610 0.217 0.096 (0.093) 0.157* (0.089) (6) (7) (8)

log-centrality index boss highest most head top Resides in NY Resides in NJ Other Xs State FE Observations R-squared

0.157*** (0.044)

0.096** (0.048)

0.108*** (0.028)

no no 610 0.040

no yes 609 0.373

yes yes 609 0.399

no no 610 0.010

no yes 609 0.359

yes yes 609 0.394

Notes: The additional regressors (Other Xs) are the extended family members, being born in Sicily, or in the rest of Italy, age, age squared, the interaction index, height, weight, marital status, whether the wife is connected, the number of children, the fraction of daughters, the number of siblings, violent crimes, age at rst arrest, never arrested, the types of crime committed, and the types of businesses. Clustered (by family) standard errors in parentheses: *** p<0.01, ** p<0.05, * p<0.1.

53

Table 8: Positive Assortativity Among Crime and Business Types Crimes Murder Robbery Simple assault Burglary Larceny Counterfeiting Drug oenses Gambling Liquor oenses Weapons oenses Extortion SD() 0.440 0.083 0.478 0.091 0.339 0.087 0.232 0.099 0.068 0.094 0.361 0.148 0.939 0.053 0.608 0.117 -0.001 0.072 0.160 0.086 0.403 0.135 Businesses SD() Restaurants 0.158 0.095 Drugstores 0.184 0.098 Food companies 0.447 0.111 Real estate 0.310 0.134 Import export 0.403 0.155 Manual laborer 0.215 0.129 Casinos 0.348 0.111

Notes: Estimated coecients and clustered standard errors by surnames from a regression of crime type dummies on the fraction of associates who perpetrated the same crime type. I restrict the data to businesses held and crimes perpetrated by at least ve percent of the sample.

54

Table 9: The Determinants of the Eigenvector Index

(1) Weighting: Extended family members Born in Italy (except Sicily) Born in Sicily Age Age squared/100 Interaction index Height in feet Weight in pounds Married Divorced Connected wife Number of children Fraction of daughters Siblings Resides in NY Resides in NJ Violent crimes Age at rst arrest Never arrested Types of crime committed Types of businesses Observations R-squared Mean value: 801 0.195 12.57 4.672*** (0.905) -1.042 (1.128) 2.578* (1.466) 0.899*** (0.304) -0.765** (0.300) 2.512* (1.407) 2.049 (2.580) 0.001 (0.020) (3) (4) Eigenvalue index MC 15 3.541*** 4.085*** 5.041*** (0.817) (1.084) (0.764) -0.872 -0.686 -0.599 (0.653) (0.728) (1.056) 1.827** 2.584** 3.387** (0.822) (1.034) (1.381) 0.341** 0.338 0.909*** (0.169) (0.218) (0.315) -0.289* -0.257 -0.768** (0.168) (0.214) (0.311) 2.487** 3.188** 2.388* (1.144) (1.333) (1.316) 2.715* 3.553* 2.534 (1.596) (1.864) (2.423) 0.004 0.002 -0.004 (0.011) (0.013) (0.018) -0.737 (1.057) -0.507 (2.369) 4.505*** (1.546) 0.808** (0.347) 1.778 (1.903) -0.115 (0.218) 5.297*** (1.123) 2.734* (1.650) 1.337 (0.931) -0.057* (0.034) -2.176* (1.199) 0.773** (0.299) 1.915*** (0.527) 801 801 801 0.171 0.194 0.287 6.92 7.81 12.57 (2) (5) MC 3.991*** (0.737) -0.264 (0.663) 2.183*** (0.814) 0.315* (0.178) -0.243 (0.177) 2.363** (1.099) 2.137 (1.483) 0.004 (0.009) 0.288 (0.596) -1.137 (1.400) 2.925*** (0.861) 0.150 (0.212) 1.022 (1.087) -0.093 (0.132) 3.485*** (0.626) 2.376* (1.346) -0.036 (0.492) -0.032 (0.023) -0.741 (0.794) 0.219 (0.203) 0.666** (0.283) 801 0.239 6.92 (6) 15 4.512*** (0.928) -0.022 (0.739) 2.904*** (1.021) 0.425* (0.228) -0.332 (0.225) 3.081** (1.261) 2.963* (1.727) 0.002 (0.011) -0.139 (0.737) -0.903 (1.574) 3.679*** (1.096) 0.450* (0.253) 0.585 (1.219) -0.161 (0.152) 4.027*** (0.753) 2.550** (1.257) 0.466 (0.578) -0.041* (0.025) -1.297 (0.840) 0.334 (0.245) 1.178*** (0.369) 801 0.269 7.81

Notes: The regressions include also missing dummies for year of birth, height, and weight. Clustered (by family) standard errors in parentheses: *** p<0.01, ** p<0.05, * p<0.1.

55

Table 10: The Determinants of Importance Indices

(1) Degree Extended family members Born in Italy (except Sicily) Born in Sicily Age Age squared/100 Interaction index Height in feet Weight in pounds Married Divorced Connected wife Number of children Fraction of daughters Siblings Resides in NY Resides in NJ Violent crimes Age at rst arrest Never arrested Types of crime committed Types of businesses Observations R-squared 4.596*** (0.840) -0.259 (0.748) 1.572** (0.789) 0.429** (0.176) -0.378** (0.172) 2.402** (1.095) 2.661* (1.593) 0.006 (0.011) 1.051* (0.630) -1.441 (1.313) 2.848*** (0.790) 0.254 (0.209) 0.534 (1.007) 0.033 (0.130) 1.182** (0.588) -0.196 (1.190) -0.341 (0.491) 0.011 (0.025) 0.087 (0.885) 0.328 (0.207) 0.405 (0.250) 801 0.282 (2) (3) Eigenvalue Total Residual 3.991*** -0.248 (0.737) (0.344) -0.264 -0.025 (0.663) (0.415) 2.183*** 0.732* (0.814) (0.445) 0.315* -0.081 (0.178) (0.118) -0.243 0.105 (0.177) (0.115) 2.363** 0.148 (1.099) (0.588) 2.137 -0.317 (1.483) (0.909) 0.004 -0.002 (0.009) (0.006) 0.288 -0.682 (0.596) (0.424) -1.137 0.193 (1.400) (0.687) 2.925*** 0.298 (0.861) (0.468) 0.150 -0.084 (0.212) (0.113) 1.022 0.529 (1.087) (0.589) -0.093 -0.123* (0.132) (0.072) 3.485*** 2.395*** (0.626) (0.378) 2.376* 2.556*** (1.346) (0.669) -0.036 0.278 (0.492) (0.341) -0.032 -0.042** (0.023) (0.017) -0.741 -0.821 (0.794) (0.580) 0.219 -0.084 (0.203) (0.109) 0.666** 0.293* (0.283) (0.168) 801 0.239 801 0.079 (4) (5) Closeness Total Residual 6.236*** 2.201*** (1.194) (0.407) -2.427 -0.053 (1.640) (0.290) 2.705 0.668** (1.710) (0.312) 1.294*** 0.221*** (0.456) (0.080) -1.173*** -0.208*** (0.447) (0.078) 3.868*** 2.455*** (1.379) (0.838) 5.111 0.455 (3.410) (0.614) 0.017 0.000 (0.029) (0.004) 0.694 -0.019 (1.371) (0.253) -4.229 -0.222 (3.291) (0.448) 3.887** 1.211*** (1.607) (0.318) 0.207 0.176** (0.520) (0.076) 1.498 -0.180 (2.639) (0.373) 0.249 0.048 (0.312) (0.051) 6.704*** 0.144 (1.334) (0.227) 7.541** 0.169 (3.434) (0.463) -0.530 -0.175 (1.161) (0.204) 0.001 0.002 (0.055) (0.010) -2.122 0.249 (2.009) (0.361) -0.103 0.105 (0.482) (0.073) 1.130* 0.193* (0.597) (0.102) 801 0.271 801 0.272 (6) (7) Betweenness Total Residual 2.230*** (0.835) -2.201 (1.371) 1.335 (1.265) 0.920** (0.391) -0.843** (0.377) 1.774* (1.045) 2.791 (2.791) 0.012 (0.025) -0.222 (1.170) -2.973 (2.470) 1.405 (1.203) -0.014 (0.385) 1.033 (2.009) 0.220 (0.240) 5.674*** (1.033) 7.712*** (2.626) -0.234 (0.958) -0.009 (0.049) -2.199 (1.806) -0.389 (0.370) 0.777 (0.480) 801 0.178 0.031 (0.042) 0.162 (0.107) -0.046 (0.097) -0.002 (0.030) 0.019 (0.029) -0.073 (0.070) -0.020 (0.202) 0.002 (0.002) -0.125 (0.099) 0.190 (0.193) 0.206* (0.117) -0.028 (0.028) 0.218 (0.142) 0.041** (0.020) 0.104 (0.087) 0.274 (0.183) 0.126 (0.087) -0.001 (0.003) 0.150 (0.141) -0.002 (0.029) 0.052 (0.041) 801 0.101

Notes: All regressions use weights based on = 15. The regressions include also missing dummies for year of birth, height, and weight. The leader variable counts the number of words that describe mobsters as leaders. Clustered (by family) standard errors in parentheses: *** p<0.01, ** p<0.05, * p<0.1

56

Table 11: Robustness Checks

(1) (2) Eigenvalue Weighting: Extended family members Born in Italy (except Sicily) Born in Sicily Age Age squared/100 Interaction index Height in feet Weight in pounds Married Divorced Connected wife Number of children Fraction of daughters Siblings Resides in NY Resides in NJ Violent crimes Age at rst arrest Never arrested Types of crime committed Types of businesses Observations R-squared 5.041*** (0.764) -0.599 (1.056) 3.387** (1.381) 0.909*** (0.315) -0.768** (0.311) 2.388* (1.316) 2.534 (2.423) -0.004 (0.018) -0.737 (1.057) -0.507 (2.369) 4.505*** (1.546) 0.808** (0.347) 1.778 (1.903) -0.115 (0.218) 5.297*** (1.123) 2.734* (1.650) 1.337 (0.931) -0.057* (0.034) -2.176* (1.199) 0.773** (0.299) 1.915*** (0.527) 801 0.287 15 3.991*** (0.737) -0.264 (0.663) 2.183*** (0.814) 0.315* (0.178) -0.243 (0.177) 2.363** (1.099) 2.137 (1.483) 0.004 (0.009) 0.288 (0.596) -1.137 (1.400) 2.925*** (0.861) 0.150 (0.212) 1.022 (1.087) -0.093 (0.132) 3.485*** (0.626) 2.376* (1.346) -0.036 (0.492) -0.032 (0.023) -0.741 (0.794) 0.219 (0.203) 0.666** (0.283) 801 0.239 (3) (4) Leadership -0.009 (0.037) 0.039 (0.093) 0.153 (0.099) 0.015 (0.027) -0.002 (0.026) -0.056 (0.058) -0.104 (0.206) 0.001 (0.001) -0.051 (0.087) 0.083 (0.163) 0.107 (0.104) 0.012 (0.029) 0.167 (0.142) 0.037* (0.019) 0.166** (0.084) 0.323* (0.189) 0.084 (0.082) -0.002 (0.003) -0.053 (0.112) 0.014 (0.024) 0.087** (0.040) 801 0.071 15 0.031 (0.042) 0.162 (0.107) -0.046 (0.097) -0.002 (0.030) 0.019 (0.029) -0.073 (0.070) -0.020 (0.202) 0.002 (0.002) -0.125 (0.099) 0.190 (0.193) 0.206* (0.117) -0.028 (0.028) 0.218 (0.142) 0.041** (0.020) 0.104 (0.087) 0.274 (0.183) 0.126 (0.087) -0.001 (0.003) 0.150 (0.141) -0.002 (0.029) 0.052 (0.041) 801 0.101

Notes: The dependent variable Leadership counts the number of times the words boss, highest, most, head, top, high, inuential, important, leader, leading, powerful, and representing are cited. The regressions include also missing dummies 57 for year of birth, height, and weight. Clustered (by family) standard errors in parentheses: *** p<0.01, ** p<0.05, * p<0.1.

Table 12: Regressions of Indegree with Random Truncation of Half of the Sample
Extended family members Born in Italy (except Sicily) Born in Sicily Age Age squared/100 Interaction index Height in feet Weight in pounds Married Divorced Connected wife Number of children Fraction of daughters Siblings Resides in NY Resides in NJ Violent crimes Age at rst arrest Never arrested Types of crime committed Types of businesses 5.54 -0.44 2.58 1.11 -1.00 2.77 3.09 0.00 0.04 -1.54 4.78 0.81 0.53 0.12 2.11 0.69 0.69 -0.02 -1.70 0.64 1.34 se(beta) 0.86 0.97 1.29 0.31 0.30 1.26 2.33 0.02 0.99 1.96 1.50 0.33 1.58 0.19 0.95 1.89 0.86 0.03 1.12 0.27 0.47 simulated average median 6.03 5.92 -0.29 -0.26 2.97 3.08 1.15 1.14 -1.03 -1.02 3.02 2.76 3.07 3.24 0.01 0.00 0.63 0.54 -0.58 -0.60 4.28 4.02 0.82 0.80 0.41 0.45 0.09 0.09 2.53 2.56 0.71 0.65 0.63 0.56 -0.02 -0.02 -1.78 -1.70 0.67 0.67 1.48 1.52

Notes: The rst two columns show the coecients and the standard errors based on the whole sample. Columns 3 and 4 show the mean and the median coecients of 500 regressions based on randomly truncated samples.

58