Price and Wealth Dynamics in an Agent-Based Model with Heterogeneous Evolving Strategies

We consider a simple asset-pricing model with one risky and one riskless asset in discrete time. In each trading period heterogeneous boundedly rational agents form their individual demand for the risky asset, and then the price of the asset is determined via Walrasian mechanism imposing a market clearing condition. We assume that agents' strategies are consistent with maximizing the CRRA utility, so that the individual demand is proportional to the agent's wealth. Under this assumption, the evolution of the distribution of wealth across the population of agents can be described as a function of the agents' demand. In the same framework with boundedly rational agents who form their demand based on the prediction of future returns, Chiarella and He (2001) found that the resulting price dynamics resemble the one of real markets: in particular, the phenomenon of volatility clustering was reproduced. Within this framework we introduce the agents who follow other behavioral patterns. Instead of predicting the future returns they update their market positions according to different learning mechanisms. We try to model, in a very schematic way, such features of human behavior mentioned in evolutionary, experimental and behavioral economics literature as imitation of more successful strategies and local search (exploration) of better trading rules. Besides we consider noise traders with purely random demand. Based on simulations we compare the resulting aggregate outcomes