Does Unemployment Increase Crime?
使用汇率、制造业占比和工会会员率作为工具变量，发现失业率每上升1个百分点，财产犯罪增加4%，比普通最小二乘法估计的1.8%更高，解释了1990年代30%的财产犯罪变化。
OLS may understate the effect of unemployment on crime because of the endogeneity problem (Raphael and Winter-Ember 2001). In this paper, we use changes in the real exchange rate, state manufacturing sector percentages, and state union membership rates as novel instrumental variables to carry out 2SLS estimations. We find a one-percentage-point increase in unemployment would increase property crime by 1.8 percent under the OLS method, but that the elasticity goes up to 4 percent under 2SLS. The larger 2SLS effect has significant policy implications because it explains 30 percent of the property crime change during the 1990s.