Dynamic Information Acquisition from Multiple Sources∗

Decision-makers often aggregate information across many sources, each of which provides relevant information. We introduce a dynamic learning model where a decisionmaker sequentially acquires Gaussian signals that are related by a flexible correlation structure. Such a setting describes sequential search between similar products, as well as reading news articles with correlated biases. We study the optimal sequence of information acquisitions. Assuming the final decision depends on a one-dimensional statistic, we show that myopic signal acquisitions are nonetheless optimal at sufficiently late periods. In classes of informational environments that we describe, the myopic rule is optimal from period 1. These results hold independently of the decision problem and its (endogenous or exogenous) timing. We apply these results to characterize dynamic information acquisition in games. JEL codes: C44, D81, D83