Costly Information and Noise Trading

Abstract

Information acquisition is essential to market efficiency. As pointed out by Grossman (1976), prices cannot be fully revealing as they are in a Rational Expectations Equilibrium (REE), because there would not be incentives to acquire information. Grossman and Stiglitz (1980) proposed a solution to this problem in the form of a noisy REE, where market clearing prices are only partially revealing. Without noise, there would not be information acquisition, and there would not be trading (Milgrom and Stokey, 1982). Vives (2010) introduces noise traders, bot informed traders are exogenous to the model. Our paper extends this framework with an endogenous mechanism of information acquisition with noise trading in a dynamics discrete choice model, to be tested in a laboratory experiment. Theory shows that conditions exist under which more noise trading leads to higher rates of information acquisition. For this result it is crucial that informed traders are more risk averse than uninformed traders. Preliminary experimental results confirm this pattern, for different levels of information cost and information precision. However, market efficiency - measured using the difference of fundamental and market price - decreases with noise trading.

JEL classification: C91, G14.

Key words: discrete choice, experiments, information acquisition, market efficiency