Predicting the money multiplier and velocity of circulation in the U.S. economy: Direct versus indirect forecasting methods

This paper compares the direct and indirect methods of predicting the money multiplier and velocity of circulation in the U.S. economy. Forecasts are generated using the autoregressive <(AR) model and Harvey's structural time series model. In addition to point forecasts, prediction intervals (calculated by using the recerltly proposed bootstrap-after-bootstrap) are used as a means of evaluating forecasting accuracy. The results turn out to be mixed with respect to the superiority of one method over the other. However, some evidence is found indicating that the direct method produces more accurate forecasts when multi-step forecasting based on AR fitting is used. When Harvey's model is used, the indirect method is found to be generally superior