Social Enforcement of Law

Transcription

1 Social Enforcement of Law Paolo Vanin Ph.D. in Economics European Doctorate in Law and Economics University of Bologna January 2019

2 Questions How do social mechanisms influence crime? Stigma Social norms Social networks Peer pressure Family and neighborhood Abortion, religion, immigration, conscription Social capital and social sanctions Theory and empirics

3 1. Theory Stigma Rasmusen 1996 JLE, Furuya 2002 JEBO, Blume 2003, Funk 2004 EER Social norms Posner 1997 AER, Lindbeck, Nyberg, Weibull 1999 QJE, Weibull, Villa 2005 Social networks Silverman 2004 IER, Calvó-Armengol, Zenou 2004 IER, Calvó-Armengol, Verdier, Zenou 2007 JPubE

4 Rasmusen (1996) JLE Stigma and Criminality Deterrence effect of stigma Static, information-based model of economic stigma (pay lower wages to convicted criminals) Stigma as a cost-effective deterrent (fine levied against labor income) Moral hazard and adverse selection Informal discussion of the trade-off between secrecy (lower recidivism) and stigma (less crime in the first place)

5 Moral hazard model Workers decide crime (C) vs. non crime (NC) Criminals productivity: m Non criminals prod.: m+y>m (don t steal at work) Employers observe conviction and offer wages to convicted and non convicted θ: fraction of criminals in the (mass 1) population a: criminals conviction prob. (1- aθ): unconvicted population Convicted wage: m Unconvicted wage:

6 C exp. payoff: V: crime reward P: punishment u: personal disutility from crime ~ F NC exp. payoff: Attractiveness of C:

7

8 Adverse selection model Crime does not reduce production, but is chosen more often by low skilled workers, so conviction is a signal of low productivity Low skilled workers: Fraction of the population Productivity: m Always commit crime High skilled workers: Productivity: m+y Decide crime vs. non crime

9 Adverse selection model Employers observe conviction and offer wages to convicted and non convicted : fraction of criminals in the (mass 1) population a: criminals conviction prob. (1- aθ): unconvicted population : low skilled unconvicted criminals : high skilled unconvicted individuals Convicted wage: Unconvicted wage:

10 Two equilibria Low crime (separating): High crime (pooling): w= w c Low skilled workers prefer pooling, high skilled prefer separating

11 Stigma: deterrence and recidivism Furuya (2002) JEBO and Funk (2004) EER: recidivism Stigma deters non convicted But induces convicted to recidivism (if legal and illegal markets are separate) High punishment of repeat offenders to reduce the recidivism effect Blume (2003): dynamic effects of stigma Cost incurred for an uncertain duration if crime is detected: stochastic process with forward looking agents (improve on evolutionary models with myopic agents) Static strategic complementarity (the cost of stigma is lower, the more individuals are stigmatized): multiple equilibria, with detection probability and duration of stigma reducing crime Dynamic (social interaction) effects: detection probability and duration of stigma may increase crime, because they increase the number of stigmatized individuals and therefore reduce the discounted cost of stigma

12 Social norms Posner (1997) AER Rule that is neither promulgated by an official source, such as a court or a legislature, nor enforced by the threat of legal sanctions, yet is regularly complied with Four incentives to obey social norms Private benefits from compliance (e.g., language) Emotions (e.g., honor code, blind rage) Stigma (disapproval, ridicule, ostracism, economic discrimination) Guilt or shame (e.g., children s education) Lindbeck, Nyberg, Weibull (1999) QJE Cost of deviating decreases in the share of deviators Potential for multiple equilibria Applied to voting on welfare transfers and labor participation Either low transfers supported by a majority of taxpayers Or high transfers supported by a majority of recipients

13 Weibull, Villa (2005) Crime, Punishment and Social Norms Anti-crime social norm of endogenous intensity Guilt and shame The expected cost of shame (not of guilt) depends on detection probability Both can depend on the share of compliers Legal and criminal returns may be Non correlated (e.g., drugs, weapons, auto-theft, robbery) Correlated (e.g., tax evasion, economic extortions, theft within firms and organizations) S.c. for equilibrium existence and uniqueness Discrete jumps in crime rates in response to continuous changes in norm sensitivity or crime premium

14 Networks Relevance of the structure of social interaction for crime rates Reputation incentives: Silverman (2004) Information transmission: C-A, Zenou (2004), C-A, Verdier, Zenou (2007)

15 Silverman (2004) IER Street Crime and Street Culture Reputation based theory of street culture Small fraction of the population that directly benefits from violent crime (street types, s) Probability to be victim of an assault decreasing in the reputation of responding with violence (being s-type) Incentive to use violence to establish reputation Violent crimes with no apparent benefits Especially valuable to youths Small variations in the fraction of s may trigger large variations in violence Negative social capital and surprising policy implications Social interaction intensity may foster a street culture Stiffer penalties may increase the reputational value of violence and hence raise it Stricter enforcement of property crime may reduce violence by reducing s

16 Calvó-Armengol, Zenou (2004) IER Social Networks and Crime Global competition among criminals Local know-how sharing among linked criminals Crime is higher in denser social networks Connections have a multiplier effect Connections may create multiple equilibria Network formation Pairwise stability of the complete network

17 C-A, Verdier, Zenou (2007) JPubE Strong and Weak Ties Individuals learn about Crime opportunities: by interacting with active criminals Job opportunities: through employment agencies Network: strong (permanent) and weak (transitory) ties Unemployment benefits raise u. duration and Increase crime if wages are low (from u. to crime) Reduce crime if w are high (from u. to employed) A higher frequency of weak ties Raises crime and reduces both employment and unemployment Reduces the effectiveness of crime deterrence (from crime to u.), because it raises the transition from u. back to crime

18 2. Empirics Family, peers, neighbourhood and network Case and Katz (1991) WP, Kling et al. (2005) QJE, Glaeser et al. (1996) QJE, Galbiati, Zanella (2012) JPubE, Glaeser, Sacerdote (1999) JPE, Levitt and Venkatesh (2000) QJE, Donohue, Calvó-Armengol et al. (2005) WP, Patacchini and Zenou (2005), Mastrobuoni (2015) EJ Abortion, religion, immigration, conscription Levitt (2001) QJE, Heaton (2006) JLE, Bianchi, Buonanno, Pinotti (2011) JEEA, Pinotti (2017) AER, Galiani, Rossi, Schargrodsky (2009) AEJAE Social capital and social sanction Buonanno, Montolio, Vanin (2009) JLE, Buonanno, Pasini, Vanin (2012) PIRS, Buonanno, Vanin (2017) JEBO

19 Case, Katz (1991) NBER Family, Neighborhood and Disadvantaged Youths Effects of family background and neighborhood char s on youths (Boston, 1989) Family Those with family members in jail, with drug problems, with teenage mothers or non married parents, and with more educated parents Have a higher probability of crime, of drug use, of out of wedlock birth, and get more schooling Neighborhood Residing in a neighborhood in which many youths commit crime, use drugs, are out of work or out of school Increases the probability of analogous outcome Even controlling for family and personal characteristics

20 Kling, Ludwig, Katz (2005) QJE Neighborhood Effects on Crime Moving to Opportunities randomized housing experiment First two years after random assignement Less violent-crime arrests for both males and females For females less arrests also for other crimes Several years after relocation More property-crime arrests for males

21 Glaeser, Sacerdote, Scheinkman (1996) QJE Crime and Social Interactions Theory: social multiplier effect of imitation Imitation magnifies the effects of crime determinants Empirics: estimate the degree of imitation in each crime by the ratio of observed to predicted crime variance across cities Control for observables and make assumptions on variance predicted by unobservables Evidence of social interaction and peer imitation in different crimes Strong for auto theft and larceny Moderate for assault, burglary and robbery Weak for arson, murder and rape

22 Galbiati, Zanella (2012) JPubE Tax Evasion Social Multiplier Theory: social multiplier from congestion in audits Strategic complementarity in income reporting Identification: contrast of the variance of tax evasion at different aggregation levels Method: Manski (1993) ReStud, Graham (2008) Ectca Data on audits to self-employed workers (small business and professionals) in Italy Estimated multiplier: 3 The equilibrium response to a shock that reduces reported income (indip. across individuals) is 3 times higher than the average direct response

23 Glaeser, Sacerdote (1999) JPE Why is there more crime in cities? Why is there more crime in cities? Higher returns to crime: explains 13% to 1/3 High transport costs for crime Lower search costs in cities Lower arrest prob.: explains at least 10% Much more if high elasticity of crime to deterrence Attraction (or creation) of crime prone individuals: 30-50% Unexplained residual: 15-50% Conjecture: weaker social sanctions?

24 Levitt, Venkatesh (2000) QJE A Drug-Selling Gang s Finances Average wage only slightly above legal alternatives Street level soldiers earn around the min wage High rank members earn much more Gang wars ¼ of the time High cost of injuries, deaths (7% per year), lost profits Individual incentive to prove tough, but collective cost Possible explanations Unusual preferences Miscalculation of risk Non economic considerations

25 Donohue, Levitt (2001) QJE Legalized Abortion and Crime Legalized abortion accounts for as much as 50% of crime drop Abortion legalization in 1973 in the US Big crime rates drop in the US in the 90 s 40% for homicides >30% for property and violent crime Earlier drop in states that legalized earlier Higher drop in states with higher abortion rates in the 70 s and 80 s Only arrests of those born after legalization fall in high relative to low abortion states Channels Smaller cohort size Aborted children would be particularly prone to crime Abortion allows to delay childbearing in suboptimal conditions

26 Heaton (2006) JLE Religion and Crime Negative relation between religion and crime in OLS estimates County-level data on religious membership and (property and violent) crime rates IV: instrumenting religious membership with historical religious adherence Negligible causal effect of religious participation on crime

27 Bianchi, Buonanno, Pinotti (2011) JEEA Immigration and Crime Negative correlation between share of immigrants and crime in Italy, Might be driven by immigrant demand-pull factors correlated with crime IV: instrument immigrants with supply-push factors in the origin country Measured by flows to the rest of Europe No significant causal effect of immigrants on total crime Significant for murders, robberies and (less) for theft

28 Pinotti (2017) AER Clicking on Heaven s Doors Effect of immigrant legalization on crime Problem: legal and illegal migrants are different Regression discontinuity design (RDD) Residence permits in Italy allocated on click days on first come, first served basis, until reaching fixed quotas Cutoff time unknown to applicants Compare male applicants who applied just before and after the cutoff Legalization halves crime rate in subsequent year

29 Galiani, Rossi, Schargrodsky (2009) AEJAE Conscription and Crime What effect (if any) of conscription on crime? Fire arms training and violent environment Order and obedience Causal effect: lottery for eligibility for conscription of young males in Argentina Based on last digits of national ID Intention to treatment estimation Conscription increases the probability of developing a criminal record in adulthood In particular for property and weapon-related crimes

30 Social capital and social sanction BMV (2009) JLE Civic norms (Blood donations) and associational networks (instrumented with historical measures of social capital) significantly reduce property crime rates in Italy BPV (2012) PIRS The strength of social sanctions, as captured by the density of social interactions (measured through demographic and geo-morphological information) significantly lowers property crime rates in Italy

31 Social closure and crime BV (2017) JEBO Measure social openness by the diversity of surname distribution History of migration and inbreeding All else equal, social closure leads to less crime and more tax evasion Social closure strengthens social sanctions and social control, leading to more cooperative outcomes in local interactions, but it reduces cooperation on a larger scale