Speculation, Heterogeneity and Learning: A Simulation Model of Exchange Rates Dynamics

Luigi Marengo and Hélène Tordjman
University of Trento, Italy
Marengo@risc1.gelso.unitn.it

Abstract

This paper presents a model of a speculative determination of exchange rates. Our basic claim is that speculation is intrinsically a disequilibrium phenomenon, mainly because the imperfect rationality of economic agents engenders heterogeneity of their beliefs, representations, and ``models of the world". Hence, agents trade in non-equilibrium and changing environments, where the properties of individual and collective learning as well as the institutional characteristics of markets' arrangements may affect the dynamics of asset prices. We illustrate this point by developing a model of exchange rate dynamics, the foreign exchange market being one of the most speculative markets, and as such quite close to a market of ``pure speculation" (no fundamental determination of the asset price). For that matter, we used classifiers systems and genetic algorithms (Holland) to simulate the behaviour of learning agents and studied the aggregate results, in terms of prices and volume of transaction, of their interaction. Preliminary results show that series of simulated exchange rates exhibit some of the distinctive properties characterizing real series.