Insights of Adaptive Learning Approach to Modeling Expectations: A Numerical Comparison with Adaptive Expectations and Rational Expectations

摘要

This study explores the macroeconomic implications of the main theories of expectations formation, i.e. adaptive expectations, rational expectations and adaptive learning, in the context of the standard growth model that provides the backbone of a lot of macroeconomics models that are used in modern research. It is shown that the adaptive expectations formulation implies a high degree of inertia even when a high correction factor is assumed. In contrast, the rational expectations and the recursive least squares learning algorithms exhibit a much faster return to equilibrium in case of a shock. The paper also emphasizes the importance of the initial conditions for the behavior of macroeconomic variables in case of the learning algorithm: if more preliminary periods are allowed so that the initial values are closer to the coefficients coming from the rational expectations solution, the predicted path of the variables is much closer to the one under rational expectations.