Persistence, patience and costly information acquisition

Abstract

A forward-looking agent observes signals of a state that follows a Gaussian AR(1) process. He chooses the signals' precisions sequentially, balancing their marginal cost and informativeness. I characterize his optimal learning strategy, and analyze his steady-state posterior beliefs and welfare. Higher persistence can tighten or loosen these beliefs, but always lowers welfare due to endogenously higher information costs. In contrast, higher patience raises welfare because the agent receives more information from his past selves.