Costly Information Acquisition in Decentralized Markets: An Experiment

Abstract

In a series of controlled laboratory experiments, we test the rationality of the decisions to purchase information, the informational efficiency of prices and the optimality properties of the resulting allocations in a decentralized markets. The theory predicts that markets with dispersed information and identifiable buyers and sellers converge to a fully revealing equilibrium. It is profitable to pay for information and as such, the Grossman-Stiglitz paradox does not emerge. We find statistically significant improvements for both price efficiency and allocational efficiency in the course of the experiment. The participants remain willing to pay for information, in contrast with centralized markets, where information effectively aggregates very fast (Plott and Sunder (1988)), and where the information auction price drops to zero (Sunder (1992)).