Systematic Job Search : New Evidence from Individual Job Application Data

We use novel high-frequency panel data on individuals' job applications from a job posting website to study how job seekers direct their applications over the course of job search. We find that at the beginning of search, applicants are sorted across vacancies by education. As search continues, education becomes a weaker predictor of which job a job seeker applies for, and an average job seeker applies for jobs that are a first-week choice of less educated job seekers. In particular, between week 2 and 26, the correlation between a job seeker's education and our measure of the type of job he applies for drops by 33 percent, with half of the drop happening by week 5. We interpret these findings to suggest that search is systematic, whereby a job seeker samples high wage opportunities (conditional on his belief about the probability of meeting the job requirements) first and lower wage opportunities later. The findings are consistent with the literature that documents declining reservation or desired wages, and provide evidence in favor of theories of job seekers' learning.This paper was originally circulated under the title "Sorting by Skill Over the Course of Job Search."