Career choice under uncertainty

Contemporary theoretical literature on occupational choice consists mostly of models that treat choice outcomes as either deterministic or risky. This paper proposes taking a slightly more realistic perspective by constructing a general occupational choice model on the basis of the assumption that outcomes are partially uncertain such that some reward distributions are unknown. The change in perspective yields some major advantages: Learning and career trajectories, which in general cannot be generated by models with deterministic or risky rewards, become a natural feature of the dynamic solution of sequential occupational choice problems. Furthermore, earnings-puzzle-like observations can be explained by sufficiently high uncertainty aversion, as uncertainty aversion has a significant impact on learning. In addition, central model predictions are consistent with data on relative choice frequencies and enterprise death rates.