The Heterogeneous Treatment Effect of Marijuana Legalization on Crime Rates

This study examines the heterogeneous treatment effect of marijuana legalization on crime rates in USA based on the crime data reported by institutions and provided by Uniform Crime Reporting (UCR) programming from 1994 to 2016. Four types of crime are analyzed, including drug-related crime, violent crime, property crime, and sex-related crime, where drug-related crime being the main dependent variable and the others being estimated as the spillover effect. There are two types of treatment to be considered: medical legalization and recreational legalization. For medical legalization, I first applied baseline Difference-in-Differences (DID), then conducted event study to investigate if the treatment effect varies across time, and finally used the Dynamic Treatment Effect to formally test it. For recreational legalization, I applied Generalized Difference- in-Differences (GDID), a new method proposed by Kedagni and Ban (2022) to create a bound for the treatment effect, which is robust to the violation of Parallel Trend Assumption (PTA). As a result, the study finds a significantly negative effect for both recreational and medical legalization on drug-related crime rate. The spillover effect shows that recreational legalization has a negative impact on other categories of crime, but medical legalization has an insignificant effect.Bachelor of Scienc