| This dissertation contains three chapters. Each applies the tools of applied microeconomics to questions in labor economics, the economics of education, and social economics, respectively. In the first chapter, which is joint work with Amanda Pallais, we present the results of a series of field experiments in an online labor market designed to test whether workers referred to a firm by existing employees perform differently from their non-referred counterparts and, if so, why. We find that referred workers have higher performance and lower turnover than non-referred workers. We demonstrate a large role for selection: referred workers perform better and persist longer even at jobs to which they are not referred at a firm where their referrers do not work. Team production is also important: referred workers are much more productive when working with their own referrer than with someone else's referrer.;In the second chapter, I examine growth in educational attainment over the past thirty years by gender and demographic characteristics. I show that both the rise in educational attainment and the rise of the female advantage in educational attainment occurred relatively similarly across socioeconomic status (SES). I also demonstrate how a prior result showing an increased gradient of education by SES used incorrect sampling weights and is not robust to a more permanent measure of SES.;In the third chapter, which is joint work with Duncan Gilchrist, we exploit the randomness of weather, and the relationship between weather and movie-going, to test for and quantify network externalities in movie consumption. We instrument for opening weekend viewership with unanticipated weather shocks when a movie first opens and estimate the effects of early viewership shocks on later viewership. Given the large set of potential weather measures, we implement Least Absolute Shrinkage and Selection Operator (LASSO) variable selection methods. We find large momentum from network externalities in movie consumption. Neither a supply response nor information dissemination plays a significant role in our estimates. Network externalities appear to be stronger for females than males, and for youth than adults. |
