Is the fight against Mexican drug cartels beneficial to public security? Nils-Hendrik KLANN University of Göttingen & University of Heidelberg New Directions in Welfare 2011 Congress 1. Juli 2016 Motivation “We have 18 months and if we do not produce a tangible success that is recognizable to the Mexican people, it will be difficult to sustain this confrontation into the next administration.” Gerónimo Gutiérrez, Deputy Secretary for Domestic Security (Diplomatic cable from 2009; Wikileaks) New Directions in Welfare 2011 Congress 1. Juli 2016 Motivation The ‘War on Drugs’ has become a global phenomenon as governments in many countries seek to fight the activities of international drug cartels. The growing sophistication of drug gangs as well as their everincreasing affinity to violence against opponents pose a direct challenge to the authority of governments. In many countries, society is caught in the middle between opposing forces in an increasingly brutal conflict between gangs and security forces. New Directions in Welfare 2011 Congress 1. Juli 2016 Motivation Acting both as a producer and transport hub for drugs, Mexico has become the center stage for an extremely violent conflict between gangs competing to deliver drugs to the US market. Since 2006, the Mexican government has significantly stepped up its initiative against drug cartels, relying on police as well as 35,000 soldiers to fight Mexico’s drug cartels. With regards to tackling the activities of Mexico’s drug cartels, the government’s anti-drug initiative is often called into question. New Directions in Welfare 2011 Congress 1. Juli 2016 Motivation On the other hand, no empirical analysis exists to date focusing on the broader implications of Mexico’s drug war for society. Objective of this paper Investigate the effect of Mexico’s anti-drug initiative on the prevalence of non-drug offences such as property crime, assault, rape and murder at the district level. New Directions in Welfare 2011 Congress 1. Juli 2016 Outline 1. 2. 3. 4. 5. Introduction Literature Overview Research Outline Empirical Results Conclusion New Directions in Welfare 2011 Congress 1. Juli 2016 1. Introduction With the ‘War on Drugs’ going on for several years and its effect on cartel activities dubious at best, which effects will this initiative have on the Mexican society? This paper focuses on the prevalence of non-drug related offences in the approximately 2500 municipal districts in the time period from 1998 until 2008 to assess whether or not the intensity of anti-drug efforts impacts on the prevalence of non-drug crime (NDC). New Directions in Welfare 2011 Congress 1. Juli 2016 2. Literature Overview Prior research on the effect of drug enforcement on other forms of crime focus provide contrasting predictions: Theory 1 Intensified drug enforcement increases non-drug crime Welfare of society is reduced as the anti-drug initiative implies negative externalities in the form of rising crime rates. Relies on seminal contribution by Becker (1986) and Ehrlich (1973) predicting the likelihood of criminal activity in the light of expected gains and costs: New Directions in Welfare 2011 Congress 1. Juli 2016 2. Literature Overview Several sources such as Sollars, Benson et al. (1994) Benson, Kim et al. (1994), Benson, Rasmussen et al. (1998) Benson, Leburn et al. (2001) analyze the implications of a marked concentration of police forces to battle drug offenders in 67 Florida counties around the 80s and 90s on non-drug crime. Also, Resignato (2000) and Shepard and Blackley (2005) conduct a similar investigation focusing on New York State. Key finding the concentration of finite law enforcement resources on one predominant type of crime increases all other types of crime (crowding out effect) New Directions in Welfare 2011 Congress 1. Juli 2016 2. Literature Overview An alternative explanation of this positive relationship is discussed in Miron (1999) and with an empirical analysis in Miron (2001). Here, violence is a systemic feature of black markets, in which participants are unable to resort to legal institutions such as the police or courts to resolve disputes or enforce their property rights. This process is exacerbated as enforcement intensifies. Key finding Intensifying enforcement efforts increases the potential for violent turf wars. Classic example: gang crime in the ambit of alcohol prohibition in the United States. (e.g., Asbridge and Weerasinghe (2009)) New Directions in Welfare 2011 Congress 1. Juli 2016 2. Literature Overview Theory 2 Intensified drug enforcement reduces non-drug crime Contrasting opinion see the potential for complementarity between the objectives to fight DC and NDC offenders. In an ideal case, stepping up drug enforcement not only reduces drug offences but beyond that yields additional gains as the measure leads to a simultaneous reduction of non-drug offences e.g., robbery, assault and kidnapping. New Directions in Welfare 2011 Congress 1. Juli 2016 2. Literature Overview Analyzing US crime and incarceration data for the time period of 1983 until 1996 Kuziemko and Levitt (2004) find evidence that stricter punishment of drug offences yields a twofold effect: 1. 2. longer incarceration of drug offenders impacts negatively on the duration served by NDC felons BUT counter to the predictions of the aforementioned Becker framework has no proliferating effect on NDC crime: significant negative relationship between the incarceration of drug offenders and non-drug crime: New Directions in Welfare 2011 Congress 1. Juli 2016 2. Literature Overview Key finding Incarcerating drug offenders indirectly helps to reduce non-drug crime as drug offenders dedicate a significant share of their activities towards offences such as assault and murder. In line with this, Levitt and Venkatesh (2000) which investigate the daily routine and financials of a Chicago drug gang finds that about one fourth of a gang member’s time is dedicated to violent crime: Shipley (1989) finds similar trends analyzing offences committed by incarcerated drug offenders: New Directions in Welfare 2011 Congress 1. Juli 2016 2. Literature Overview Summary of the two strands of literature Limited Police resources Crowding out effects NDC increase Incarcerating drug traffickers positive secondary effects New Directions in Welfare 2011 Congress NDC decrease July 1, 2016 3. Research Outline and Methodology New Directions in Welfare 2011 Congress 1. Juli 2016 3. Research Outline and Methodology Analysis of district-level data from approx. 2500 municipal districts, spread out over 32 federal states over the time period of 1998-2008; running the following empirical model: Dependent variable NDC log of non-drug crime incidents in a district (5 indicators) Source: Statistical Yearbooks from 32 federal states; National Statistics Institute New Directions in Welfare 2011 Congress 1. Juli 2016 3. Research Outline and Methodology Independent Variables DE share of drug arrests over all arrests in a district DENS district population per km2 Source: Statistical Yearbooks from 32 states; Mexican National Statistics Institute (INEGI) the Mexican Geographic Service POP log district population URATE district unemployment rate Source: Natl. Institute for Federalism and Municipal Development Source: Secretary for Employment and Social Security (STPS) HIGHW Highway Dummy CDET deterrence arrests/offences Source: Author’s calculation Source: Authors calculation based on Statistical Yearbooks from 32 federal states; National Statistics Institute New Directions in Welfare 2011 Congress Source: Authors calculation based on surface data from 1. Juli 2016 3. Research Outline and Methodology Offences implemented as depended variable include - Robbery - Assault - Rape - Murder - Gang murder* *Data on gang murder is provided by the Mexican Interior Ministry from the year 2006 onwards. Victims were categorized as gang victims based on the circumstances of their death, e.g., the use of large caliber weapons, signs of torture New Directions in Welfare 2011 Congress 1. Juli 2016 4 Empirical Results Regression Framework Ordinary Least Squares (baseline) FE (district fixed effects) Negative Binomial (count data) GMM (endogeneity) Robustness checks carried out in the course of regressions: Exclusion of each federal state and year New Directions in Welfare 2011 Congress 1. Juli 2016 [Pooled OLS] Total offences in log Robbery drug enforcement Assault Murder Rape 4. Empirical Results -0.615*** -1.465*** 0.022 -0.304** log of population log of density unemployment rate highway in district robbery | arrests/offences (-7.60) 1.234*** (95.57) -0.032*** (-3.88) 10.548*** (22.53) 0.357*** (11.55) -2.021*** (-29.28) assault | arrests/offences (-19.13) 1.064*** (89.96) -0.006 (-0.86) 5.168*** (12.30) 0.246*** (8.60) (0.29) 0.881*** (69.45) -0.089*** (-11.70) -3.830*** (-8.83) -0.023 (-0.77) -1.802*** (-33.30) rape | arrests/offences -8.672*** (-66.82) R2 0.834 N 5097 * p<0.10, ** p<0.05, *** p<0.01 New Directions in Welfare 2011 Congress 0.996** (2.43) 0.704*** (11.46) -0.238*** (-6.75) 5.048*** (2.59) -0.174 (-1.35) -1.866*** (-32.67) murder | arrests/offences constant (-2.45) 0.888*** (51.06) -0.035*** (-3.56) -0.900* (-1.77) 0.005 (0.13) Gang Killings -6.846*** (-55.76) 0.844 4976 -5.584*** (-46.47) 0.722 3876 -0.290 (-1.07) -2.020*** (-28.78) -6.238*** (-34.30) 0.778 2330 -5.719*** (-9.80) 0.291 545 1. Juli 2016 [Fixed Effects] Total offences in log Robbery drug enforcement Assault Murder 4. Empirical Results -0.474*** -0.544*** -0.324*** log of population unemployment rate robbery | arrests/offences (-7.16) 0.088** (2.06) 0.291 (0.47) -1.744*** (-24.52) assault | arrests/offences (-8.04) 0.173*** (3.82) -2.481*** (-3.45) (-4.42) 0.032 (0.81) -2.234*** (-3.03) -0.396** (-2.37) 0.135*** (2.99) -0.759 (-0.72) Gang Killings 0.632 (1.13) 0.125 (0.97) 28.767*** (5.38) -1.326*** (-19.24) murder | arrests/offences -1.428*** (-21.22) rape | arrests/offences constant Rape -8.672*** (-66.82) 0.346 4878 -6.846*** (-55.76) 0.220 4714 -5.584*** (-46.47) 0.308 3663 -0.566 (-1.48) -1.420*** (-17.29) -6.238*** (-34.30) 0.384 2091 -5.719*** (-9.80) 0.533 353 R2 N * p<0.10, ** p<0.05, *** p<0.01 All regressions with time and district fixed effects; robust t-values in brackets Standard errors in OLS Fixed Effects regressions adjusted for clustering across districts New Directions in Welfare 2011 Congress 1. Juli 2016 [Negative Binomial] Total offences Assault Murder 4.Robbery Empirical Results drug enforcement log of population unemployment rate highway in district robbery | arrests/offences -0.481*** (-8.91) 0.184*** (9.47) 0.058 (0.14) -0.168** (-2.23) -1.850*** (-33.73) assault | arrests/offences -0.613*** (-10.45) 0.141*** (7.59) -3.097*** (-6.64) -0.461*** (-5.59) -0.268*** (-3.80) 0.005 (0.21) -2.946*** (-4.66) -0.279** (-2.27) -1.452*** (-29.46) rape | arrests/offences New Directions in Welfare 2011 Congress -0.787*** (-5.18) 0.012 (0.44) 0.660 (0.85) -0.046 (-0.30) Gang Killings 1.624*** (5.36) 0.600*** (11.99) 2.879 (1.22) -0.119 (-0.71) -1.362*** (-28.24) murder | arrests/offences N 4878.000 * p<0.10, ** p<0.05, *** p<0.01 Rape 4714.000 3663.000 -0.272 (-1.01) -1.520*** (-20.56) 2091.000 1160.000 1. Juli 2016 [GMM] Total offences Robbery lagged dependent drug enforcement log of population 0.327*** (5.43) -0.306*** (-3.36) 0.796*** (10.20) 4.856*** (5.86) 0.257*** (4.18) -1.568*** (-14.89) Assault Murder Rape (6.45) -0.848*** (-9.69) 0.742*** (13.88) 2.194*** (3.43) 0.166*** (3.05) 0.208*** (4.60) 0.025 (0.31) 0.590*** (15.20) -3.166*** (-4.14) 0.021 (0.35) 0.216*** (3.53) -0.222 (-1.50) 0.638*** (10.60) -1.270* (-1.80) 0.017 (0.28) . 0.291*** 4. Empirical Results unemployment rate highway in district robbery | arrests/offences assault | arrests/offences -1.244*** (-17.12) rape | arrests/offences -5.224*** (-8.88) -4.519*** (-11.28) -3.600*** (-11.57) -0.770 (-1.19) -1.725*** (-16.45) -4.208*** (-8.23) R2 N 3380 3236 2670 1622 Hansen J 13.882 17.279 11.085 13.337 (p-value) 0.178 0.068 0.351 0.148 Arellano-Bond test for AR1 in 1st differences -6.351 -7.494 -7.957 -4.212 (p-value) 0.000 0.000 0.000 0.000 Arellano-Bond test for AR2 in 1st differences -1.047 -0.574 -0.868 -0.32 (p-value) 0.295 0.566 0.385 0.749 Number of districts 670 685 572 471 Number of instruments 27 27 27 26 * p<0.10, ** p<0.05, *** p<0.01 Two-step System GMM with time fixed effects and Windmeijer finite sample correction New Directions in Welfare 2011 Congress 0.837*** (3.67) 0.805 (1.19) 0.202* (1.80) 2.361 (0.84) -0.159 (-0.64) -1.503*** (-16.92) murder | arrests/offences constant Gang Killings -1.914 (-1.35) 195 0.278 0.598 176 10 1. Juli 2016 5. Conclusion Regarding the effect of drug enforcement in general: A greater share of arrests among all arrests seems to reduce the Prevalence of most NDC offences at the district level. Some indication exists regarding the expected positive relationship between drug enforcement and gang murder – results should be Interpreted with caution however, given the short time span of the Data. New Directions in Welfare 2011 Congress 1. Juli 2016 5. Conclusion Regarding the control variables: In line with expectations, population yields a positive effect on the number of offences. Unemployment has a significant effect on crime - however no singular relationship can be derived from the results. Deterrence measures show the expected negative effect on all types of crime except gang murder. New Directions in Welfare 2011 Congress 1. Juli 2016 5. Conclusion Thank you New Directions in Welfare 2011 Congress 1. Juli 2016