Essays in labor economics

Abstract

This thesis sheds light on two cutting-edge topics in Labor Economics, peer effects in the workplace and non-cognitive skills, and makes a methodological contribution to the related literature. The literature on peer effects in the workplace seeks to better understand co-workers' effect on an individual's productivity through the interactions among workers beyond the production technology. In the first essay, titled Learning When It Counts: Evidence from Professional Bowling Tournaments, I test the hypothesis that a worker can improve productivity by learning from peer co-workers in high-skill jobs. While demand for high-skill workers has been increasing, high-skill jobs often require workers to make a decision, facing uncertainty underlying their tasks. Highly skilled professionals have deep insights to pick up meaningful patterns of information. Therefore, if they are in an environment that allows them to learn additional information from co-workers, their productivity can improve. In this paper, I examine the productivity effects of learning among high-skill peers about uncertain conditions underlying their tasks with variations in the "space of ideas," exploiting a unique, novel dataset from professional bowling competitions. Specifically, a bowler learns about lane conditions in part by watching his competitor bowl on the same lane. A right-handed bowler learns more relevant (to his task) information from competing with another right-hander than with a left-hander, as the used part of the lanes (the proximate space of ideas) varies with handedness. I compare the probabilities of bowling a strike of bowlers matched with like-handed competitors versus opposite-handed competitors. I find a large impact of the same ideas space on learning, e.g, being paired with a like-handed bowler increases strike probability by 14 percentage points. This finding adds evidence for the existence of peer effects in high-skill jobs. I also show that learning curves exist only when bowlers are in same-handed match-ups, by examining how these differences change from one frame to the next over a game. Another calculation is determining how much total scores could be increased by pairing bowlers to raise the proximity in the space of ideas. These results are suggestive of how much workplaces might increase productivity by optimally pairing workers based on the proximity of the space of ideas. The second topic of this dissertation is non-cognitive skills such as conscientiousness, self-control and social skills. Conventionally, economists have assumed that measures of cognitive skills such as IQ were sufficient to represent the role of human capital in production. However, a growing body of research suggests that non-cognitive skills are important factors in educational attainment and labor outcomes. Recent research in psychology shows that bilingualism can help strengthen social skills and self-control. In the second essay, joint with Tracy Regan and titled Bilingual Advantage in Non-cognitive Skills, we examine the causal relationship between bilingualism and non-cognitive measures, exploiting a large dataset from Education Longitudinal Study of 2002 (ELS:2002). To isolate the causality, we use an indicator for whether either parent was foreign-born as an instrumental variable for bilingualism. We find that raising the degree of speaking a language other than English to parents from none to all of the time can increase a student's percentile in the U.S. national distribution by 36 percentage points for conscientiousness (being well organized and working hard) and by 39 and 50 percentage points for instrumental motivation (academic motivation to achieve external goals such as better job opportunities) and persistence (keeping working even in difficulties). In particular, the bilingual advantage in persistence turns out to be significant only for disadvantaged children (the lowest socioeconomic status quartile) but insignificant for the others. These results suggest that bilingualism can be promoted as a policy tool to reduce inequality and call for further research on the relationship between bilingualism and non-cognitive skills. In the final essay, titled Simple Transformation for Finding a Maximum Weighted Matching in General Digraphs, I propose a novel, simple procedure using an existing efficient algorithm to find an optimal pairing that can produce the maximum output. As shown in the first essay, this algorithm can be useful for the optimal deployment of workforces with the consideration of peer effects. Particularly, the procedure is applicable to cases in which the order in a pair matters. The order can complicate the problem of finding optimal pairings, because a pair can have two orders. To address this ordered pairing problem, I devise a simple transformation of a general directed graph to a proper (undirected) graph. Using the transformed graph, a maximum weighted matching can be found, using any existing polynomial-time algorithm for undirected graphs. By recovering orientations in the found matching, a maximum weighted matching for the original directed graph can be found. I prove the matching from the suggested algorithm is always a maximum weighted matching in the directed graph. This thesis contributes to Labor Economics by adding evidence in newly-rising topics. The first chapter shows evidence of peer effect--learning from competitors--among high-skill workers. The second chapter suggests that bilinguals have an advantage in forming non-cognitive skills. The third chapter proposes an algorithm for finding an optimal pairing to maximize the aggregate productivity in the consideration of the learning effect found in the first chapter. I hope that the findings in the thesis will meaningfully contribute to the developing literature of Labor Economics.