This dissertation consists of three independent chapters that study how to improve public policies and reduce the level of social injustice through the lens of microeconomics and the innovative use of new data sets. In the first chapter, I test the impact of neighborhood heterogeneity on the private contribution of local public goods. Using a panel data set containing over two million non-emergency service requests and detailed census-tract level data on socioeconomic characteristics from the American Community Survey, I find that, contrary to the prevailing view in the literature, racial and linguistic heterogeneity have little to no negative effect on private voluntary contributions to local public goods. Income inequality, on the other hand, reduces private contributions by a significant margin. In the second chapter, my coauthors and I examine how job transfer rules and preferences affect labor market efficiency and access to quality teachers. To do so, we recover teacher and school preferences using data from Minneapolis Public Schools’ web-based internal teacher labor market. Overall, we find that the average teacher prefers schools serving already-advantaged students and the average school prefers applicants who are more effective, hold an advanced degree, and not in their early-career. These preferences help explain why we observe the troubling sorting patterns among teachers and suggest that further liberalizing the teacher labor market may exacerbate the inequitable distribution of quality teachers. Finally, the third chapter evaluates a hidden social cost of air pollution beyond hospital admissions and premature deaths: student achievement. Given the strength of evidence linking academic performance to long-term life outcomes and the fact that disadvantaged and marginalized communities tend to get more exposure to air pollution, this additional cost should be identified and quantified correctly. Using an exogenous source of variation in the levels of air pollution from the closure of an airport terminal, I find that the closure led to a roughly 2 percent of a standard deviation increase in high-stakes test scores.