Abstract
This paper provides a microfounded information acquisition technology based on a simple framework with information search. When searchable information is limited, an agent encounters increasingly more redundant information in his search for new information. Redundancy slows down the learning process and generates decreasing returns. Furthermore, as multiple agents search for information from the same source, limited searchability leads to covariance as the acquired information becomes increasingly more overlapped among agents. Using an asymptotic approach, we construct a tractable mapping from resource (attention) allocations to the precision and the correlation of agents' information under varying degrees of searchability of information. We study two economic applications with endogenous information acquisition using our model: (i) a beauty contest coordination game, and (ii) a noisy rational expectations equilibrium.