Money demand instability and the performance of the monetary model of exchange rates

The paper tesis the ability of the monetary approach to explain the long-ron behavior of the exchange fale in the G-7 bilateral relationships during the floating period. I use Johansen's (1988,1991) approach to test fot fue existence of a cointegraling relationsbip between the exchange rate, the money supply and real output. Consistent with previons results I find little evidence in support of the monetary model in the full sample (1973-1991). Further analysis reveals that one of the principal building blocks of the monetary model, the money demand equation, exhibited a significant degree of instability during the period. I use the tesis proposed by Hansen (1992) to test for parameter stability in the money demand equation and to estimate potential break points. The estimated break points are then used to define stable subsamples ror each bilateral relationship. The subsample cointegration results provide strong support for the monetary approach in most non US bilateral models, indicating that the exchange rate responds with a lag, primarily to monetary shocks, and that long-run exchange rate homogeneity cannot be rejected in most instances. In addition, I compare the predictive performance of the error correction model agaist the predictions of a random walk model. In four of me six bilateral models for which the forecasting comparison can be undertaken the error correction model produces more aceurate one-quarter-ahead forecasts.