Statistical theory of optimal procrastination

Procrastination is a widespread phenomenon that occurs across multiple domains. Current theoretical accounts invoke specific structures of preference or cost to elucidate the occurrence of procrastination. We propose a theoretical model of procrastination that does not invoke a specific structure of preference, and instead relies on an unavoidable aspect of any projection into the future, namely, that the time it takes to complete a task is variable. This model captures the basic phenomenology of procrastination and predicts a phenomenon that we demonstrate empirically, the "deadline-time-arrival effect": decision makers (DMs) who live closer to their destinations (e.g., schools or offices) tend to arrive later than those who live farther. The same theory explains other instantiations of procrastination, such as repeatedly postponing the initiation of a task as the deadline for completion approaches—e.g., delaying the planned drafting of a grant application. Our framework thus puts forth a unified view of procrastination in which its various manifestations reflect an optimal adaptation to environmental statistics.