| Summary: | 碩士 === 國立臺灣大學 === 經濟學研究所 === 92 === In this paper, we introduce the heterogeneous agents and learning behavior into the RBC model, where firms solve their maximization problem according to their own beliefs about the underlying shock, and these beliefs would be updated by each period through certain criterion. The process of how a firm update its beliefs is characterized by the genetic algorithms and the given criteria divides into a shotrun performance evaluation and a long-run one. We find that when the long-run criteria is accepted within the agents, the learning process urges the beliefs of the society into a level higher than the true shock, while the short-run one is accepted, the after-learned beliefs tend to be undervalued about the state. In a situation where learning behaviors are involved, the dynamics of aggregate variables would be determined then by true underlying shock and the biased behaviors which are driven by agents’ beliefs. As a result, some potential equilibrium could be inducted under this model.
|
|---|
