Essays in Applied Microeconomics

Abstract

This dissertation includes three essays. The first essay is entitled "Do Peers Influence Occupational Preferences? Evidence from Randomly-Assigned Peer Groups at West Point" and is coauthored with Michael Kofoed. We study the effect of an individual's peers on his occupational preferences among cadets at West Point. We exploit unique institutional details at West Point, including random assignment of cadets to peer groups and the repeated elicitation of occupational preferences to overcome common biases that arise in the study of peer effects. Overall, we find little evidence of peer effects in this setting. The second essay, coauthored with Ronald Ehrenberg, is entitled "Are High-Quality PhD Programs at Universities Associated with More Undergraduate Students Pursuing PhD Study?" This paper uses restricted-access Survey of Earned Doctorates merged with other datasets to study which attributes of a doctoral university are associated with a higher share of its undergraduate BAs who proceed to earn a PhD. We consider four fields: humanities, physical sciences, life sciences, and social sciences. We use truncation correction methodology to correct for PhDs earned after the data end. In our main specification, we find that across all fields PhD production is positively related to student test scores and the number of high-quality PhD programs an institution has. It is negatively related to the total number of students and the share of total BAs that are received in the field. The third essay is entitled "Information and the Beauty Premium in Political Elections." It is coauthored with Joseph Price. We study the beauty premium in U.S. political elections. We collect subjective beauty data on candidate photos from participants on an internet survey. We find that more beautiful candidates receive a higher vote share and are more likely to win, but that this is moderated by situations in which the voter is more likely to have less information on the candidate. The beauty premium is smaller—or zero—for U.S. congressional races than for state congressional races. It is also smaller for incumbents and in elections in which there is more campaign spending.