Out-of-Sample Forecast Performance
of Economic Variables
for France, Germany, and Italy

The paper investigates the usefulness of quantitative and qualitative indicator variables as forecasting means of industrial production growth in France, Germany, and Italy. The analysis is carried out for two sets of forecasts whose main difference concerns the way in which projections are defined. Forecasts are obtained from simultaneous equations models that generate predictions from the autoregressive processes of indicator variables. In computing root mean squared error statistics and Theil inequality coefficients, the forecasting performance of the leading indicators is compared against forecasts from the autoregressive structure of the reference series. The empirical evidence points to the usefulness of the sampled indicator variables as forecasting instruments of industrial production growth. Dependent on the set of forecasts, satisfactory predictions are available for the short- or medium-run, i.e., for the period of one month or for the horizon of three- to six-months.