Abstract
In this note an alternative to the widely used reduced - form tests of the monetary model of the exchange rate is proposed. It is shown that the reduced form approach rests on implausible parameter restrictions which can be easily avoided by estimating the long-run money demand functions separately. Moreover, the resulting ‘structural’ forecast equation allows an economic interpretation of the various channels affecting the exchange rate in the monetary model. This approach is illustrated with reference to the DM/Dollar exchange rate where the structural model outperforms several alternative forecasting strategies out-of-sample.