Career Dynamics Under Uncertainty: Estimating the Value of Firm Experimentation
This paper develops and structurally estimates a learning model in which firms acquire information about workers' ability by observing their performance over time. A firm consists of a collection of jobs which differ in the informational content of performance, as measured by the dispersion in posterior beliefs after output is observed. Ability is general across jobs and firms. Because of the trade--off between learning and short--run profit maximization, a firm's optimal job assignment policy is solution to an experimentation problem---a multi--armed Bandit problem with dependent arms. In presence of firm competition, the job--dependent quality of performance signals can cause distortions in information acquisition within firms, inducing (ex ante) inefficient job assignment and turnover. The model is estimated using longitudinal data from a single U.S. firm on the cohorts of managers entering the firm at the lowest managerial level between 1970 and 1979 (the same dataset used by Baker, Gibbs and Holmstrom [1994a, 1994b]). Estimation results confirm that a theoretically restricted learning model can succeed in fitting the dynamic pattern of separations, promotions and demotions, and individual wage profiles. The estimated model is then used to assess the impact on mobility, between jobs and firms, and on wages of alleviating ex ante (at the beginning of an employment relationship) and interim (during employment) uncertainty, respectively through improved screening and monitoring. Given the estimated divergence between private and social returns to information, alternative policies (e.g., mandatory probationary employment contracts) are evaluated to quantify the potential welfare gains associated with increased experimentation in firms.
In order to set up a list of libraries that you have access to,
you must first login
or sign up.
Then set up a personal list of libraries from your profile page by
clicking on your user name at the top right of any screen.