Approximating Fixed-Horizon Forecasts Using Fixed-Event Forecasts

Abstract

Many forecast surveys ask their participants for fixed-event forecasts. As fixed-event forecasts have seasonal properties, users often employ an ad hoc approach to approximate fixed-horizon forecasts based on these fixed-event forecasts. We derive an optimal approximation for fixed-horizon forecasts by minimizing the mean-squared approximation error. Like the ad hoc approach, our approximation employs a weighted average of the fixed-event forecasts. The optimal weights tend to differ substantially from those of the ad hoc approach. In empirical applications, the gains from using optimal instead of ad hoc weights turn out to be sizeable. The approximation approach proposed can also be useful in other applications.