What do we know about recent exchange rate models? : in-sample fit and out-of-sample performance evaluated

Previous assessments of nominal exchange rate determination have focused upon a narrow set of models typically of the 1970's vintage, including monetary and portfolio balance models. In this paper we re-assess the in-sample fit and out-of-sample prediction of a wider set of models that have been proposed in the last decade, namely interest rate parity, productivity based models, and behavioral equilibrium exchange rate models. These models are compared against a benchmark model, the Dornbusch-Frankel sticky price monetary model. First, the parameter estimates of the models are compared against the theoretically predicted values. Second, we conduct an extensive out-of-sample forecasting exercise, using the last eight years of data to determine whether our in-sample conclusions hold up. We examine model performance at various forecast horizons (1 quarter, 4 quarters, 20 quarters) using differing metrics (mean squared error, direction of change), as well as the "consistency" test of Cheung and Chinn (1998). We find that no model fits the data particularly well, nor does any model consistently out-predict a random walk, even at long horizons. There is little correspondence between how well a model conforms to theoretical priors and how well the model performs in a prediction context. However, we do confirm previous findings that out-performance of a random walk is more likely at long horizons.