Testing the Rational Expectations Hypothesis with Agent-Based Models of Stock Markets.

Using agent-based models of stock markets, this paper examines the rational expectations hypothesis from a bottomup perspective. We apply standard linear and nonlinear econometric tests to arti cial time series generated from two arti cial stock markets composed of bounded-rational traders. The two arti cial stock markets differs in their architectures: one has a business school, and one does not. While the linear test shows that the market with the business school fails to reject the rational expectations hypothesis quite often, the nonlinear one does not. Therefore, strictly speaking, these two agent-based markets of boundedrational traders do not collectively behave as what rational expectations hypothesis predicts, and hence do not lend support to rational expectations hypothesis.