Explaining Exchange Rate Volatility With A Genetic Algorithm

Traditional exchange rate models fail to explain stylized facts such as the high volatility or high trading volumes in an adequate way. Contemporary research has therefore increasingly turned to model the foreign exchange market in a more realistic setting, highlighting for instance the high degree of speculative trading caused by different opinions about the future exchange rate. In this paper, we choose a direct empirical microfoundation to model the interactions between heterogenous agents. In a first step, we show on the basis of pychological evidence that agents rely on strong simplifications for decision making purposes and adjust these decisions only slightly if the outcome is not satisfying. In a second step, we identify technical and fundamental trading rules as the most important investment strategies for the foreign exchange market. Consequently, we construct a model where heterogeneous boundedly rational market participants rely on a mix of technical and fundamental trading rules. The rules are applied according to a weighting scheme. Traders evaluate and update their mix of rules by a genetic algorithm. Simulations of the resulting nonlinear dynamic system produce equilibrium exchange rates that appear to circle around some fundamental value without any apparent tendency to converge, thus replicating the stylized fact of high volatility. The dynamics exhibit a complex behavior, and for some values of the coefficients the model even behaves chaotically. Finally, we enrich the dynamics by allowing for stochastic disturbances. Already for a small shock probability the simulated exchange rate fluctuations mimic very closely what is observed empirically. For instance, we find features such as volatility clusters or a fat tail distribution of returns.