This paper proposes a two-countries model for the determination of the exchange rate, mixing the agent-based (AB) and the stock-flow consistent (SFC) methodologies. This artificial economy is populated by heterogeneous traders who can be either fundamentalist or chartist in exchange rate expectations (De Grauwe and Grimaldi, 2003; Lavoie & Daigle, 2011). Chartist traders are further split between trend followers and trend contrarians, consistently with empirical surveys witnessing both types of strategies in the Forex (Schulmeister, 2007). Consistently with the SFC methodology, traders manage a financial portfolio made of four assets: domestic currency, domestic bonds, foreign currency and foreign bonds. Each transaction respects the quadruple entry accounting principle (Godley & Lavoie,2012). As a matter of originality with respect to the standard AB literature, this model allows to analyze simultaneously the impact of exchange rates fluctuations on real economy dynamics as well as the impact of real economy dynamics on exchange rate fluctuations in an accounting-consistent framework. Moreover, as a matter of originality with respect to the standard SFC literature, the agent-based features
of the model allow focusing on the mechanisms generating empirically observed statistical patterns of exchange rate fluctuations.
Exchange rate transactions on the Forex are explained by the interaction among heterogeneous behavioral rules and portfolio decisions under uncertainty. Speculative motives on the Forex and on the bonds market generate endogenous fluctuations, by reproducing empirically observed dynamics. This model allows to reproduce well acknowledged stylized facts: booms and busts, precarious equilibria, volatility clustering and fat tails (Schulmeister, 1987; 2008; Cont, 2005). The role played by fixed threshold on opening and closing positions and the existence of cash in strategies can explain most of the statistical properties produced by the model. In such a framework, the exchange rate can be fully disconnected from the fundamentals of the economy and is based on agents’ expectations, whose behaviors can generate strong instability. Nevertheless, we show that the proportion of the different types of agents and the interaction of their behaviors - despite playing a relevant role on generating aggregate fluctuations - does not affect the general stability of the model. Contrarily to conventional results (De Grauwe and
Grimaldi, 2003), we show that strategy switching, besides being empirically questionable, is unnecessary to explain fluctuations and global stability: fixed strategies and real economy dynamics (the balance of payments sensitivity to exchange rate fluctuations, the demand and supply of currency and assets for real transaction motives) contribute to explain the statistical patterns and the general stability of the model. Moreover, we show that increasing the share of fundamentalist agents does not always provide more stability if fundamentalist agents are allowed to adopt explorative strategies, rather than being endowed with a divine ability to perfectly guess (on average) the equilibrium conditions of this
complex economy. Intuitive economic policy consequences of the model are drawn, by postponing more rigorous economic policy simulations to further developments of the model.