Adaptive Learning in Financial Markets
研究重复博弈的Grossman-Stiglitz模型中适应性学习能否收敛到理性预期均衡，发现单调选择动态会收敛，并给出学习过程参数条件。
We investigate adaptive or evolutionary learning in a repeated version of the Grossman and Stiglit (1980) model. We demonstrate that any process that is a monotonic selection dynamic will converge to the rational expectations asset demands if the proportion of informed traders is fixed. We also show that these learning processes have a unique asymptotically stable fixed point at the Grossman–Stiglitz (GS) equilibrium. The robustness of learning to noisy experimentation is studied using Binmore and Samuelson's (1999) deterministic drift approximation. Conditions on economic and learning process parameters for adaptive learning to lead to the GS rational expectations equilibrium are presented.