Essays in online labor markets

Abstract

This thesis explores the economics of online labor markets. The first paper evaluates a market intervention that sought to improve efficiency within the world's largest online labor market. The second paper provides an illustration of how online labor markets can serve as a platform for helping researchers study economic questions using natural field experiments. The third paper examines the role of supervision within a firm using detailed productivity data. In the first paper, we report the results of an experiment that increased job application costs in an online labor market. More specifically, we made it costlier to apply to jobs by adding required questions to job applications that were designed to elicit high-bandwidth information about workers. Our experimental design allows us to separate the effect of a costly ordeal vs. the role of information by randomizing whether employers see workers' answers. We find that our ordeal reduced the number of applicants by as much as 29% and reduced hires by as much as 3.6%. Overall, the applicant pool that underwent the ordeal had higher earnings and hourly wages, but not better past job performance. The ordeal also discouraged non-North American workers. We find no evidence that employers spent more when vacancies were filled, but some evidence that employer satisfaction improved. These improvements were the result of information provision rather than selection. Finally, we did not find any heterogeneity in outcomes across job category, contract types, or employer experience. In the second paper, we conduct the first natural field experiment to explore the relationship between the "meaningfulness" of a task and worker effort. We employed over 2,500 workers from Amazon's Mechanical Turk (MTurk), an online labor market, to label medical images. Although given an identical task, we experimentally manipulated how the task was framed. Subjects in the meaningful treatment were told that they were labeling tumor cells in order to assist medical researchers, subjects in the zero-context condition (the control group) were not told the purpose of the task, and, in stark contrast, subjects in the shredded treatment were not given context and were additionally told that their work would be discarded. We found that when a task was framed more meaningfully, workers were more likely to participate. We also found that the meaningful treatment increased the quantity of output (with an insignificant change in quality) while the shredded treatment decreased the quality of output (with no change in quantity). We believe these results will generalize to other short-term labor markets. Our study also discusses MTurk as an exciting platform for running natural field experiments in economics. In the third paper, we investigate whether greater supervision translates into higher quality work. We analyze data from a firm that supplies answers for one of the most popular question-and- answer ("Q&A') websites in the world. As a result of the firm's staffing process, the assignment of supervisors to workers is as good as random, and workers are exposed to supervisors who put forth varying degrees of "effort" (a measure based on a supervisor's propensity to correct work). Using this exogenous variation, we estimate the net effect of greater supervision and find that a one-standard-deviation increase in supervisor effort reduces the number of bad answers by between four and six percent. By decomposing the total effect into the separate effects on corrected and uncorrected answers, we conclude that supervisor effort tends to lower the number of good answers among uncorrected answers. Interestingly, observable worker behaviors (i.e., answer length and time to answer a question) seemed unaffected by supervision. None of the results vary with worker experience.