This dissertation consists of three chapters. Each chapter investigates a question within law and economics, focusing on issues related to the criminal justice system. The first chapter tests for judicial bias in the treatment of racial minority and female attorneys by examining bail decisions. I account for potential omitted variable bias by exploiting the random assignment of defendants, public defenders, and judges to first-appearance hearings in Miami-Dade County. I analyze the differences across judges in the rates at which attorneys of different demographic groups secure a pretrial release. In order to do so, I develop a finite sample test of significance that accounts for the small sample of cases at the individual judge level. I find significant variation in the release rate between Black and White attorneys measured across different judges. However, I do not find evidence of judicial bias in the treatment of female or Hispanic attorneys in my setting. The standard deviation in the success rate of Black attorneys across judges is 3.4 percentage points larger than what would be expected from sampling variation alone. In settings where they can choose their representation, Black defendants hire Black attorneys at higher rates. Leveraging data from an alternative setting where defendants can choose their attorneys, I simulate client-attorney matches corresponding to these preferences. Under these simulated matches, increasing judges' favorability toward Black attorneys by one standard deviation of the estimated relative favorability would decrease the defendant racial gap in pretrial release rates by 54%. As case outcomes affect not only defendants but also attorneys’ productivity and wages, judicial bias may also help explain the continued underrepresentation of minorities in the legal profession, particularly in its higher ranks. The second chapter is joint work with Anastasia Karpova, Artem Kuriksha, and Peter Meylakhs. It develops a structural model of demand for illegal drug varieties to study how consumers substitute between different types of drugs in response to government policies. This chapter uses a unique longitudinal dataset on prices, quantities, and individual decisions obtained by scraping a darknet marketplace that covered a large fraction of the retail illegal drug trade in Russia. The estimation procedure exploits a novel set of micro-level moment conditions to identify correlations in preferences for specific drug types and the degree of attachment to them. The median own-price elasticity of demand for illegal drugs is -3.6, and there is high substitution within two classes of drugs: medium-risk stimulants and cannabis. Estimates are validated using exogenous variation in the price of hashish caused by increased policing. The estimated model is used to evaluate counterfactual drug policies. It is shown that reducing the price of cannabis has the benefit of decreasing the use of riskier drugs while increasing cannabis use. For every 4 additional doses of cannabis consumed, 1 less dose of another drug is consumed. Our estimates show that the recent introduction of a new family of synthetic drugs has increased total drug demand in Russia by 40%, suggesting that governments should allocate resources to prevent the introduction of new drug products. Finally, the model helps identify the optimal drugs to target for interdiction, specifically those without close substitutes, such as alpha-PVP. The third chapter is joint work with David Abrams and Hanming Fang. This chapter provides empirical corroboration of optimizing models of police behavior by examining changes in stops and frisks around two extraordinary events of 2020: the COVID-19 pandemic onset and the nationwide protests following the killing of George Floyd. During these periods, hit rates from pedestrian and vehicle stops generally rose as stops and frisks fell dramatically. Several alternative explanations are ruled out, including changes in street population, crime, police allocation, and policing intensity. The results are robust to a number of different specifications. This chapter's findings provide quantitative estimates that can contribute to the important goal of improving and reforming policing.