Limited participation and exchange rate dynamics : does theory meet the data ?

Meese and Rogoff [1983] show that macroeconomic models of the Seventies fail to outperform the random walk exchange rate forecasts. Macroeconomics thus provides useless information as far as out-of-sample exchange rate forecasting is concerned. However, since Meese and Rogoff's seminal paper, advances have been made in the theoretical modeling of international macroeconomic dynamics. New Open Economy Macroeconomics, developed in the wake of Obstfeld and Rogoff [1995]'s work, shows that intertemporal general equilibrium models capture international stylized facts. In particular, Hairault, Patureau and Sopraseuth [2002] have shown that a small open economy model based on credit market frictions successfully accounts for exchange rate volatility. This paper aims at assessing the empirical performances of this model. First, the model statistically matches a large set of features observed in the data, in addition to exchange rate dynamics. Credit market frictions are found to be critical in making the model consistent with the data. Secondly, we find that, in the recent period only, the limited participation model succeeds in outperforming the random walk exchange rate forecasts in the medium run.