- Private Experience In Adaptive Learning Models

Here I provide a model that gives some insights regarding questions about actual economic behavior. I take as a source for stylized facts the experiments conducted by Marimón and Sunder as reported in Econometrica, 1993, in which it is shown that people initially do not behave according to the rational expectations assumption, but eventually learn to do so. I propose a slight generalization of the adaptive learning model in order to explain, besides the long run equilibrium observed, the stochastic-like time paths in the aggregate variables. In fact, the introduction of heterogeneity in private experience accumulated over time in a simple adaptive model with fixed decision rules shown is shown to be necessary and sufficient to generate the complex kind of dynamics present in the experiments. In our version of the Marcet- Sargent OLS model, people can not be using useful public information available, but only private experience instead, when they do price forecasting. Otherrwise, we would not be able to explain the data with this model. This result sheds light on the experimental results, in the sense of suggesting a stronger degree of bounded rationality in experimental subjects. In addition, I provide examples within the proposed environment that improve upon the explanatory power of existing adaptive learning models.