Essays in applied microeconomics

This dissertation comprises three chapters in applied microeconomics, each covering a different topic. Chapter 2 brings new causal evidence to an important topic in the economics of crime—the effect of plea bargaining on sentencing outcomes. Despite its widespread use worldwide, its effects on sentencing outcomes remain elusive due to identification challenges. Chapter 2 addresses this issue by examining a policy experiment conducted in China. By exploring the staggered rollout of plea bargaining across courts from 2014 to 2020, I find that plea bargaining reduces sentence lengths on average, but a heterogeneity analysis shows wide dispersion in court-level treatment effects. Consistent with the predictions of a theoretical model, sentencing uncertainty at trial, judges’ caseloads, and the local prevalence of privately retained attorneys are important factors driving the dispersion. Chapter 3 studies a novel theme in health and labour economics: the labour market effects of psychedelics. We partner with two ayahuasca centres in Brazil, from which we obtain the full registration records of 2,482 ceremony participants. We then link these records to Brazilian administrative labour market data. Employing difference-in-differences methods, we show that ayahuasca ceremony attendees are 9 percentage points less likely to be employed in the formal labour market three years after their initial visit. Chapter 4 falls within behavioural economics and examines how the outcomes of earlier rounds influence the assessments of subsequent rounds in figure skating competitions, which have two sequential rounds: the short program (SP) and the long program (LP). Skaters are sorted into groups based on their SP rankings to compete in the LP. Using a regression discontinuity design, we find that skaters who just make the stronger group receive significantly higher scores in the LP. This pattern disappears when the intra-group competing order is further sorted by SP rankings rather than determined by a random draw. Furthermore, the bias is stronger in more subjective score components, and it attenuates as the game's stakes increase. These results suggest that inattentive judges rely on temporary labels that encode outcomes of earlier rounds to form prior expectations about skaters' performances in subsequent rounds.