University prestige, performance evaluation, and promotion: Estimating the employer learning model using personnel datasets

Employers rely on educational credentials to form their initial belief about a worker's ability and update the belief by observing the worker's performance on the job. We study the careers of white-collar university graduates, using personnel data from two large Japanese manufacturers. These data contain information about the university from which the worker graduated, as well as the worker's performance evaluations and positions in the promotion ladder. As employees move up the career ladder, performance evaluations become a more important determinant for promotion than educational credentials. Structural estimates suggest that employers learn workers' ability quickly through observing their performance on the job, with expectation errors halving after about 3 to 6 years.