Self‐Fundamentals, Cross‐Fundamentals, and Exchange Rate Predictions

In this paper, we propose incorporating both self‐fundamentals, defined as the fundamentals of two economies in one currency pair, and cross‐fundamentals, defined as the fundamentals of other major economies, to forecast exchange rates in line with the theory of “third‐country effects” of Berg and Mark (2015). We utilize the Mallows model averaging method proposed in Hansen (2007) to optimally combine the predictions provided by fundamental submodels. We find that our approach significantly outperforms the random walk and the alternatives for one‐month‐ahead predictions and that both the self‐ and cross‐fundamentals play important roles in prediction. Furthermore, according to our prediction, we obtain meaningful investment profit trading on currencies and bonds.