Protests against police brutality and systemic racism have been prevalent in the United States, and most recently hastened by the killing of George Floyd. This paper evaluates how George Floyd protests affect racial disparities in nonfatal police-civilian interactions using police practice data across 17 cities in 12 states and a combination method of regression discontinuity (RD) and difference-in-differences (DiD). The results show that the protests have not impacted the proportion of African Americans in stops, but have reduced the proportion of African Americans in arrests from 30% to 26%. When dividing all interactions into cases in daylight and in darkness, the decreased effect of the protests only holds during the daytime, instead of nighttime when public supervision is absent. It suggests that Black Lives Matter protests did affect nonfatal police-civilian interactions when it comes to race. However, the day-night differences imply that the decrease in police interactions with African Americans may not be due to the change in police attitudes/beliefs. It is possible that it is a temporary change yielding to the strong public attention at that time.

This paper explores the racial differences in politicians’ persistence in elections. Empirical data from California city council elections and a close election regression dis- continuity design (CERDD) suggest that losing an election causes 70% attrition in rerunning for office. After a loss, however, nonwhite candidates are 59% more likely to run for office again compared to white candidates. The possibility of winning the subsequent election remains the same for different racial groups conditional on rerunning. As such, the persistence of losing nonwhite candidates contributes to closing the racial representation gap.

We draw on statewide data from North Carolina to examine the impacts of racial and ethnic representation in city councils impacts policing. Specifically, we focus on outcomes of traffic stops; e.g., whether a driver receives a warning or a citation after being stopped. We first document large Black-white and Latino-white disparities in the likelihood of consequence (arrest or citation) after a traffic stop. We then use a difference-in-differences design, focusing on changes following (narrow) elections of nonwhite (rather than white) councilmembers, and find that heightened nonwhite representation significantly reduces Black-white gaps in stops and actions taken after a stop. The magnitude of the reduction is similar with and without officer fixed effects, suggesting that results are largely driven by individual officer-level behavior change rather than a change in the composition of the police force.

Outcome bias is pervasive and persistent across different environments. In our noisy gift-exchange game, where agents can perform a real effort task to improve principals' lottery win probability, we replicate outcome bias in effort rewarding when effort is only numerically observable. To investigate the role of principals' beliefs on effort cost, we employed a visual treatment in which principals watch a 30-second video of the agents performing the task. We show that visually observing agents' work corrects asymmetry in rewarding effort. The post-experiment survey suggests that the mechanism through which visually observing effort reduces the outcome bias in reciprocating effort is informing principals about the cost of effort.

A startling number of students drop out from junior high school in rural China every year. Little is known about the social aspect of the dropout process in rural China. The goal of this paper is to examine the relationship between student peer relations and dropout behavior in rural China’s junior high schools. Covering 4596 students in 38 schools, our study finds that 6.9 percent of grade 7 students did not finish school, along with 12.3 percent of grade 8 students. Using a “push out and pull out” framework to understand peer relations, results suggest that both push out and pull out factors are strongly associated with student dropout.

Social interactions in infancy have implications for long-term outcomes. This study uses data from a sample of 1412 rural Chinese infants aged 6–12 and 24–30 months to examine the relationship between peer interactions and cognitive development. Over 75% of the infants in this sample had less than three peers and around 20% had no peers in both periods. The prevalence of cognitive delays is high within this sample and increases as infants age. Multivariate analysis reveals that peer interaction is significantly associated with cognitive development. Heterogeneous analysis suggests that peer interactions and mental development may be related to the child’s primary caregiver and the distance from the child’s household to the center of their village.