Strike One To Educate One Hundred *: The Effects of Organized Crime on Political Selection Gianmarco Daniele * Unum castigabis, centum emendabis Roman Empire Quote
Pinotti (2015) Motivation
Motivation Organized Crime: 1. Economic relevance The United Nations (2011) estimate the profits of transnational organized crime at $1.6 trillion accounting for 70 percent of all criminal proceeds 2. Detrimental economic effects (e.g. Pinotti, 2015; Bonaccorsi; 2009; Schneider, 2010) 3. Establishes ties with political institutions to reach their goals (e.g. Dal Bó et al., 2006; Acemoglu et al., 2013; De Feo & De Luca, 2015)
Motivation Political Selection: Modern political economy has tended to focus only on the incentives faced by politicians for good or bad behavior, while neglecting the importance of selection. But no society can run effective public institutions while ignoring the quality of who is recruited to public office and what they stand for Besley, 2005, Journal of Economic Perspectives
Motivation Determinants of Political Selection: Wage (Besley, 2004; Gagliarducci & Nannicini, 2013) Reservation Quota (Beaman et al., 2010, Baltrunaite et al. 2014) Institutional Framework (Mattozzi & Merlo, 2008, Daniele & Geys, 2015) Electoral competition (Galasso & Nannicini, 2011) Organized Crime? (Dal Bó et al., 2006) Effects on: female representation in politics, electoral competition, politicians ability (e.g. human capital)
Motivation THIS PAPER: Focus on the effects of organized crime on political selection Data from Southern Italian municipalities TEST whether a negative shock to politicians payoffs due to mafia affects political selection MAIN RESULT: the murder of a politician by organized crime reduces future politicians ability (proxied by education)
Outline Hypothesis Data Estimation Strategy Results Robustness Tests Spillover Effects Additional Results Conclusions
Hypothesis H1: Ability of organized crime to offer bribes to, or impart punishments on, politicians is inversely related to elected politicians quality (Dal Bo et al., 2006) Politicians payoff goes down when organized crime is politically active, why? Organized crime offers bribes and punishments Politician can: accept the bribe and pay the risk of jail Politician can: reject the bribe and pay the risk of punishment
Hypothesis According to H1, if politicians payoffs, politicians ability will TEST: negative shock to individuals expected payoffs (from politics) due to murders of local Italian politicians by the mafia Murders represent an extreme case of punishment (increase the perceived risk of being a politician) We expect a decrease in politicians ability after the murder
Data Data from 1900 municipalities in Southern Italy (1985-2014) on: Politicians education (proxy of ability) Other politicians characteristics Murdered politicians City level data Local city councils have powers in terms of local taxes and provision of public goods (e.g. transport, economic development, education, waste management, local police)
Data Elections take place every five years, and the mayor is directly elected We consider only electoral years (i.e. when selection takes place) year desc_comune AFT_KILL edu_city_year 1985 CALTANISSETTA 0 16.33333 1990 CALTANISSETTA 0 15.5 1993 CALTANISSETTA 0 15.27273 1997 CALTANISSETTA 0 16.57143 1999 CALTANISSETTA 0 16.75 2004 CALTANISSETTA 1 16.33333 2009 CALTANISSETTA 1 15.5
Frequency Data 1974-2014: 97 local politicians were murdered by criminals (excluding terrorist attacks) 14 12 10 8 6 4 2 0 2011 2010 2009 2008 2007 2006 2005 2004 2003 2002 2001 2000 1999 1998 1997 1996 1995 1994 1993 1992 1991 1990 1989 1988 1987 1986 1985 1984 1983 1982 1981 1980 1979 1978 1977 1976 1975 Year
Frequency Data 1974-2014: 97 local politicians were murdered by criminals (excluding terrorist attacks) 35 30 25 20 15 10 5 0 Sicilia Campania Calabria Other Regions Region
Estimation Strategy Difference-in-differences (DiD) framework We compare municipalities with and without a murder before and after the murder Murders take place in different years Separate the effect of the treatment from possibly unobserved time-specific events
Estimation Strategy
Estimation Strategy Politicians Education Proxy for ability commonly used by political economists e.g. De Paola and Scoppa (2010); Besley et al. (2011); Galasso and Nannincini (2011); Daniele and Geys (2015) Strong evidence of positive effect of education on firms performance Some evidence of the same effects in politics (e.g. Besley et al., 2011) In this case, high variation in politicians human capital (in 1985, 25% of the politicians had only elementary degree)
Estimation Strategy
Estimation Strategy T=0 T=1 T=2
Results (1) (2) (3) After -0.739-0.741-0.733 (3.98)** (3.73)** (3.69)** Municipality FE YES YES YES Year FE YES YES YES Controls NO NO YES Province-Year FE NO YES YES R 2 0.31 0.38 0.36 N 12,964 12,964 11,447 (1) (2) (3) After +1-0.640-0.581-0.505 (2.36)* (2.89)** (2.68)** After +2-0.818-0.872-0.886 (2.16)* (3.65)** (3.59)** After +3-0.626-0.696-0.738 (1.74) (2.88)** (2.91)** After +4-0.904-0.859-0.886 (3.46)** (3.57)** (3.61)** Municipality FE YES YES YES Year FE YES YES YES Controls NO NO YES Province-Year FE NO YES YES R2 0.31 0.38 0.36 N 12,955 12,964 11,447 Negative meaningful effect on politicians education (about 40% s.d.)
Robustness Test: Pre-Trends (1) (2) (3) Before -2-0.041 0.110 0.177 (0.14) (0.37) (0.60) (base category Before-1) After +1-0.651-0.552-0.459 (2.34)* (2.57)* (2.30)* After +2-0.829-0.844-0.840 (2.12)* (3.34)** (3.21)** After +3-0.637-0.667-0.695 (1.71) (2.74)** (2.71)** After +4-0.915-0.833-0.844 (3.32)** (3.40)** (3.38)** Municipality FE YES YES YES Year FE YES YES YES Controls NO NO YES Province-Year FE NO YES YES R2 0.31 0.38 0.36 N 12,955 12,964 11,447 No Pre-Trends: the drop in ability follows a political murder
Robustness Test: Type of Politician Mayor Alderman Councillor After -0.424-1.500-0.914 (0.53) (2.11)* (2.41)* Municipality FE YES YES YES Year FE YES YES YES Controls NO NO YES Province-Year FE NO YES YES R 2 0.12 0.17 0.34 N 8,068 7,253 11,095
Spillover Effects
Spillover Effects T=0 T=1 T=2
Spillover Effects (1) (2) (3) (4) (5) (6) NEIGBefore -2 0.057 0.148 0.141 0.068 0.148 0.137 (base category NEIGBefore-1) (0.45) (1.09) (1.03) (0.52) (1.09) (0.99) NEIGAfter +1-0.247-0.201-0.202-0.217-0.206-0.206 (2.76)** (2.02)* (2.01)* (2.35)* (2.00)* (1.99)* NEIGAfter +2-0.229-0.171-0.167-0.196-0.163-0.159 (2.24)* (1.54) (1.48) (1.87) (1.42) (1.37) NEIGAfter +3-0.312-0.246-0.257-0.277-0.247-0.253 (2.96)** (2.19)* (2.22)* (2.56)* (2.15)* (2.15)* NEIGAfter +4-0.384-0.241-0.294-0.320-0.228-0.280 (3.66)** (2.05)* (2.46)* (2.96)** (1.90) (2.30)* Municipality FE YES YES YES YES YES YES Year FE YES YES YES YES YES YES Controls NO NO YES NO NO YES Province-Year FE NO YES YES NO YES YES R 2 0.30 0.38 0.36 0.31 0.38 0.36 N 12,732 12,732 11,312 12,491 12,491 11,120
Additional Results: Voting What about the demand side? Voters might prefer extremist parties after a political murder, which in turn might have less educated politicians Test winning party after a murder Controlling for the winning party
Additional Results: Voting Civic Left Right Centre After Murder -0.069 0.070 0.034-0.080 (1.21) (1.03) (0.84) (1.20) Municipality FE YES YES YES YES Year FE YES YES YES YES Controls YES YES YES YES Province-Year FE YES YES YES YES R2 0.50 0.21 0.23 0.39 N 11,601 11,601 11,601 11,601 Dep. Var. dummy=1 (otherwise=0) depending on the winning party No change in winning party
Additional Results: Voting (1) (2) (3) Before -2-0.046 0.095 0.160 (0.16) (0.32) (0.54) (base category Before-1) -0.650-0.549-0.456 (3.49)** (2.55)* (2.28)* After +1-0.843-0.847-0.846 (3.43)** (3.31)** (3.20)** After +2-0.627-0.659-0.683 (2.53)* (2.70)** (2.66)** After +3 After +4-0.894-0.821-0.833 (4.07)** (3.35)** (3.35)** left -0.012 0.006-0.004 (0.29) (0.13) (0.08) right -0.083-0.072-0.078 (1.72) (1.41) (1.39) centre 0.014 0.007-0.006 (0.29) (0.14) (0.11) Municipality FE YES YES YES Year FE YES YES YES Controls NO YES YES Province-Year FE YES YES YES R2 0.31 0.38 0.36 N 12,964 12,964 11,447
Conclusions Politicians ability (measured by their levels of education) is strongly reduced after a shock represented by the murder of a local politician by organised crime Evidence of spillover effects in nearby cities, where we observe a significant reduction in politicians education New finding, representing an unsettling effect of criminal organizations on political selection