A Multiple Time Series Approach to Analyzing and Forecasting the Major French Monetary Aggregates

The seasonally adjusted French monetary aggregates M1, M2, and M3 are analyzed and forecasted by a multiple time series model. Two principal features of the analysis are the modelling of the variance-covariance structure of the joint distribution of the aggregates and the allocation of the aggregates to permanent and transient sources by means of an allocation parameter which is estimated from the data. The forecasting performance of the multiple model is compared to that of univariate random walk and ARIMA models.