Bounded Rationality, Learning, and Business Cycles in a Standard Neoclassical Growth Model

Bounded rationality is introduced into a standard growth model by assuming that households form one-period ahead least squares forecasts on production factor prices, and expect that future level of consumption and physical capital will be consistent with the balanced growth path. Under those hypotheses, the constrained lifetime utility maximizing problem of an infinitely lived representative household is equivalent to a succession of simple two-period constrained optimization problems. In this setting, closed form solutions for consumption and investment can be easily derived , and competitive equilibrium trajectories can be computed directly from the model's non-linear structure. When it is calibrated for the US economy and for a low value of the coefficient of relative risk aversion, this model exhibits cyclical or chaotic competitive equilibrium trajectories that do not exist under perfect foresight. For a standard coefficient of relative risk aversion of one, the model augmented with random productivity shocks generates more volatile time series for consumption and investment than under the rational expectations hypothesis