Adaptive learning in financial markets

We investigate adaptive or evolutionary learning in a repeatedversion of the Grossman and Stiglit (1980) model. We demonstratethat any process that is a monotonic selection dynamic willconverge to the rational expectations asset demands if the proportionof informed traders is fixed. We also show that these learningprocesses have a unique asymptotically stable fixed point atthe Grossman-Stiglitz (GS) equilibrium. The robustness of learningto noisy experimentation is studied using Binmore and Samuelson's(1999) deterministic drift approximation. Conditions on economicand learning process parameters for adaptive learning to leadto the GS rational expectations equilibrium are presented.