The Relation Between Overreaction in Forecasts and Uncertainty: A Nonlinear Approach

Abstract

This paper examines if overreaction of oil price forecasters is affected by uncertainty. Furthermore, it takes into account joint effects of uncertainty and oil price returns on forecast changes. The panel smooth transition regression model from Gonz alez et al. (2005) is applied with univariate and multivariate transition functions to account for nonlinear relations. Data on oil price expectations for different time horizons are taken from the European Central Bank Survey of Professional Forecasters. The results show that forecasters overreact for low levels of uncertainty and underreact for increasing uncertainty. Furthermore, returns are found to be more relevant for forecast changes in short time horizons while uncertainty dominates for longer ones.