We develop a structural theory of beliefs and behavior that relaxes the assumption of time consistency in beliefs. Our theory is based on the trade‐off between optimism, which raises anticipatory utility, and objectivity, which promotes efficient actions. We present it in the context of allocating work on a project over time, develop testable implications to contrast it with models assuming time‐inconsistent preferences, and compare its predictions to existing evidence on behavior and beliefs. Our predictions are that (i) optimal beliefs are optimistic and time inconsistent; (ii) people optimally exhibit the planning fallacy; (iii) incentives for rapid task completion make beliefs more optimistic and worsen work smoothing, whereas incentives for accurate duration prediction make beliefs less optimistic and improve work smoothing; (iv) without a commitment device, beliefs become less optimistic over time; and (v) in the presence of a commitment device, beliefs may become more optimistic over time, and people optimally exhibit preference for commitment. This paper was accepted by Neng Wang, finance.
