This paper analyzes the impact of the 2012 legalization of recreational marijuana in Washington State on opioid abuse. Using synthetic control methodology, this paper finds that the legislation prevented 638 overdose deaths and lead to over 3,600 individuals seeking treatment for opioid abuse disorders. Due to the large health, social, and economic impacts of the opioid epidemic, further research should be conducted into ways to reduce the number of opioid prescriptions, the number of opioid overdoses, and opioid abuse generally.

The United States has long waged war against drugs resulting in, amongst other things, theworld’s largest prison population and costing society billions of dollars yearly. As a possible solution to the problem, legalization of recreational marijuana has emerged with Colorado as a pioneering state. The purpose of this study is to examine whether legalization has a casual effect on unemployment and arrests. This was done by performing difference-in-difference estimations with Colorado as the treatment group, Georgia and Arizona as the control group with unemployment and varying offenses as dependent variables. The results indicate legalization leading to a decrease in marijuana possession arrests. However, the non-drug related offenses and unemployment increased, albeit without statistical significance, which means that the null hypothesis cannot be rejected and leaves room for questioning of the results. This may be explained by there being more factors affecting those variables, Thus, using legalization as a mean to decrease unemployment and arrest rates can be challenged.

My doctoral thesis examines the broad question of the effect of some recent health policies on health and also tries to measure socioeconomic inequalities. The first essay investigates the effect of public health insurance on people with vulnerable health. The second chapter analyses the effect of the legalization of marijuana on health, while the third chapter measures socioeconomic inequalities in health. In chapter 1, I study the evolution of access to health care for individuals in vulnerable health before and after the Affordable Care Act. I define leakage of health care as the aggregation of accessibility hurdles for individuals in vulnerable health. However, "being in vulnerable health" is a linguistic concept that does not have a sharp mathematical definition. I draw on the fuzzy sets theory and assume a non-dichotomous membership function to capture the linguistic imprecision. However, the task of choosing the "right" membership function remains an issue. To circumscribe this additional issue, I use a stochastic dominance approach to test for changes in leakage. In order to establish causality, I exploit two quasi-experimental settings offered by the dependent coverage and the states in which medicaid expansion took place. In order to use these quasi-experiments in a stochastic dominance framework, I extend Athey and Imbens (2006) changes in changes approach to a bivariate setting. Using data from the National Health Interview Survey, the results from a before and after analysis show that leakages are much lower in 2015 compared to 2009 in the US. These before and after results hold irrespective of a person's sex or socio-economic status. The causal analysis shows that leakages in not having insurance and access are reduced in medicaid expansion states after the ACA. Chapter 2 analyzes the implications of these recreational marijuana legalization (RML) on Body Mass Index (BMI) and some healthy behaviours. I exploit the quasi experimental nature of marijuana legalization policy in states using changes in changes and difference in difference approaches to identify the effect of these recreational marijuana policies. Using data from the Behavioral Risk Factor Surveillance System (BRFSS), the results show that recreational marijuana legalization reduces BMI for the entire population. The effect is mainly in the mid and top part of the BMI distribution. Subgroup analysis shows that the reduced BMI resulting from RML is significant among women but not among men. For females, the effect is found both at the lower tail (being underweight) and at the upper tail (morbid obesity). While we found evidence of a reduction in being overweight for both whites and non-whites due to RML, the reduction in obesity and morbid obesity was only found for non-whites. In addition, RML reduces obesity for those below 45 years. I also found evidence that RML increases alcohol consumption, has no effect on smoking of tobacco and binge drinking but reduces the probability of doing any physical activity. The final chapter explores the measurement of socioeconomic inequality using ordinal variables. Most measures of socioeconomic inequality are developed for ratio scale variables. These measures use the mean as a reference point which is non-robust in the presence of categorical variables. This chapter extends Allison and Foster (2004) median based approach to measuring inequalities to a bivariate case and provides conditions to robustly rank any two distributions of socioeconomic inequalities in well-being or mental health. Using the Canadian Community Health Survey (CCHS), I provide robust ordering for socioeconomic inequalities in well-being and mental health for different sub-populations in 2015. The results show that there is less socioeconomic inequality in life satisfaction, happiness, mental health, and general health status among employed males and females compared to their respective unemployed groups in 2015.