Crime and Inflation: Preliminary Cross-National Evidence Richard Rosenfeld University of Missouri – St. Louis USA Inflation is the cruelest tax. Anon New Consensus? • Recent US research reveals robust effects of changing economic conditions on crime rates • Beyond the unemployment rate: GDP, wages, consumer sentiment • Direct effects limited to property crime; indirect effects on violent crime Great Recession • Unemployment rose, economic growth stalled, consumer sentiment plunged during 2008-09 recession • Crime rates fell • Resurgence of the “consensus of doubt” Was the Great Recession Different? • Unlike in previous recessions, inflation rates at historic lows – Prices dropped in 2009 for first time in over 50 years • Anecdotal evidence of inflation – crime connection – – – – 1930s: price deflation 1950s: low inflation 1960s: rising inflation 1970s: stagflation Falling crime rates Rising crime rates Prior Research • Scattered US studies of inflation effects on crime • All show significant positive effect • Little theoretical development • No cross-national research Inflation vs. Other Indicators of Economic Adversity • More widespread • More immediate • Closer relationship to demand for stolen goods Logic Model: Inflation and the Market for Stolen Goods • Supply – Offenders sell or trade goods they do not consume or give away – Offenders respond to incentives, including demand for stolen goods • Demand – Consumers “trade down” as prices rise – Stolen goods are “inferior goods”: demand increases as prices rise (or aggregate income falls) – Acquisitive crime rises with increases in demand for stolen goods Inflation and Violent Crime • Trafficking stolen goods risky business • Underground markets “stateless” locations • Violence potent enforcement mechanism Research Issues • Are year-over-year changes in crime related to inflation? • Is the relationship between crime and inflation nonlinear? • Do former communist nations affect the relationship between inflation and crime? • Are the effects of inflation on crime independent of those of GDP and unemployment? • Does inflation condition the effect of unemployment on crime? Data and Methods • DVs: Homicide, Robbery, and Burglary Rates for 20 Nations, 1990 – 2010 – European Sourcebook of Crime and Criminal Justice Statistics – Eurostat (2008-10) • IVs: Inflation, unemployment, GDP per cap, nation and period effects – World Bank – International Monetary Fund • Fixed effects panel models with panel corrected standard errors – Variables first-differenced, except where noted Sample Austria Bulgaria Denmark England & Wales Estonia Finland France Germany Greece Hungary Ireland Italy Netherlands Northern Ireland Norway Poland Scotland Sweden Switzerland United States Summary Descriptive and Bivariate Results Homicide and Inflation Rates for 20 Nations, 1990 - 2010 (Z Scores) 2.50 Homicide 2.00 Mean Sd 1.50 Inflation 2.49 .60 Mean Sd 2.46 .98 1.00 0.50 0.00 -0.50 -1.00 -1.50 r = .41 -2.00 -2.50 -3.00 1990 1995 2000 2005 2010 Figure 2. Robbery and Inflation Rates for 20 Nations, 1990 - 2010 (Z Scores) 2.50 Robbery Mean 110 Sd 21 2.00 1.50 Inflation Mean Sd 2.46 .98 1.00 0.50 0.00 -0.50 -1.00 -1.50 r = .68 -2.00 -2.50 -3.00 1990 1995 2000 2005 2010 Figure 3. Domestic Burglary and Inflation Rates for 20 Nations, 1990 - 2010 (Z Scores) 2.50 Burglary Mean 398 Sd 63 2.00 1.50 Inflation Mean 2.46 Sd .98 1.00 0.50 0.00 -0.50 -1.00 -1.50 r = .61 -2.00 -2.50 -3.00 1990 1995 2000 2005 2010 Table 1. Regression Results for Yearly Change in Homicide, Robbery, a and Burglary Rates in 20 Nations, 1990 – 2010 ___________________________________________________________ Homicide Robbery Burglary ___________________________________________________________ b Inflation .021 .033** .046** (.013) (.012) (.014) 2 R .105 .197 .177 Obs. 400 400 397 ___________________________________________________________ a Fixed effects panel models. Nation and year effects suppressed. Variables in natural logs. Panel corrected standard errors in parentheses. b Yearly percentage change in consumer prices. ** p < .01 * p < .05 Table 2. Regression Results for Residual Change in Homicide, Robbery, a and Burglary Rates in 20 Nations, 1990 – 2010 ___________________________________________________________ Homicide Robbery Burglary ________________________________ ___________________________ Crime rate t-1 .630** .918** .746** (.069) (.040) (.048) Inflationb .029* .032** .034** (.012) (.012) (.013) 2 R .946 .954 .932 Obs. 400 400 397 ___________________________________________________________ a Fixed effects panel models. Nation and year effects suppressed. Variables in natural logs. Panel corrected standard errors in parentheses. b Yearly percentage change in consumer prices. ** p < .01 * p < .05 Crime as a Non-Linear Function of Inflation Burglary Example Estimated Yearly Change in Burglary Rate in 20 Nations by Values and Percentiles of Inflation Rate, 1990 - 2010 2 28.1 0 -2 -4 10.5 -6 Inflation Rate 4.1 -8 -10 -12 .5 .9 5th 10th 1.6 25th 2.4 Inflation Percentile 50th 75th 90th 95th Multivariate Results by Crime Type Table 3. Regression Results for Yearly Change in Homicide Rates in 20 Nations, a 1990 – 2010 _________________________________________________________ ______________ Model 1 Model 2 Model 3 Model 4 ________________________________________________________ _______________ Former communist -.048 -.007 -.019 -.076 nation (.037) (.057) (.058) (.042) b Inflation .021 -.006 -.008 -.015 (.013) (.018) (.018) (.018) Inflationsqu --.008* .008* .006 --(.003) (.003) (.003) GDPpercap ----.229* .260** ----(.107) (.098) Unemployment ----.057 -.050 ----(.063) (.080) Inflation x ------.062** unemployment ------(.022) 2 R .105 .121 .130 .145 Obs. 400 400 400 400 _____________________________________________________ __________________ a Fixed effects panel models. Nation and year effects suppressed. All variables in natural logs except former communist nations. Panel corrected standard errors in parentheses. b Yearly percentage change in consumer prices. ** p < .01 * p < .05 Table 4. Regression Results for Yearly Change in Robbery Rates in 20 Nations, a 1990 – 2010 _______________________________________________________________________ Model 1 Model 2 Model 3 Model 4 _____________________________________________________ __________________ Former communist -.074 -.011 .014 -.078 nation (.046) (.045) (.042) (.049) b Inflation .033** -.005 -.002 -.008 (.012) (.013) (.013) (.013) Inflationsqu --.011** .009** .007* --(.003) (.003) (.003) GDPpercap -----.258** -.235** ----(.092) (.088) Unemployment ----.094 .012 ----(.049) (.062) Inflation x ------.048** unemployment ------(.018) 2 R .197 .240 .270 .282 Obs. 400 400 400 400 _______________________________________________________________________ a Fixed effects panel models. Nation and year effects suppressed. All variables in natural logs except former communist nations. Panel corrected standard errors in parentheses. b Yearly percentage change in consumer prices. ** p < .01 * p < .05 Table 5. Regression Results for Yearly Change in Burglary Rates in 20 Nations, a 1990 – 2010 _______________________________________________________________________ Model 1 Model 2 Model 3 Model 4 _______________________________________________________________________ Former communist -.043 -.107 -.092 -.096 nation (.090) (.075) (.070) (.071) b Inflation .046** -.007 -.004 -.002 (.014) (.016) (.016) (.016) Inflationsqu --.016** .013** .014** --(.004) (.004) (.004) GDPpercap -----.341** -.350** ----(.109) (.108) Unemployment ----.078 .109 ----(.055) (.062) Inflation x -------.017 unemployment ------(.023) 2 R .177 .251 .286 .287 Obs. 397 397 397 397 _______________________________________________________________________ a Fixed effects panel models. Nation and year effects suppressed. All variables in natural logs except former communist nation. Panel corrected standard errors in parentheses. b Yearly percentage change in consumer prices. ** p < .01 * p < .05 Discussion • Crime is a nonlinear function of inflation • Inflation effects are small but robust • Inflation conditions the effect of unemployment on homicide and robbery, but not burglary • Economic growth has sizable and robust effects on crime – Positive effect of growth on homicide needs further exploration Next Steps • Add other economic indicators – Consumer sentiment – Poverty, inequality • Add other controls – Age composition – Divorce rate – Urbanization