Is the shift in the uncertainty level provoked by the Great Recession? A non-linear approach to the survey uncertainty
Seminar of the Department of Economics and Finance
Wednesday, 11 March 2020
16:00–17:00 in room SOC-417
Tallinn University of Technology, Estonia
Eesti Pank, Estonia
ABSTRACT: At the time of the Great Recession, there occurred a noticeable shift in the level of survey uncertainty measured as the mean individual variance of density forecasts. Meanwhile, the disagreement of forecasters, or cross-sectional variance of point forecasts derived from the same dataset, has shown completely different and much more expected countercyclical dynamics. Seeking to explain a puzzling regime shift in the mean individual variance, the paper applies smooth transition regression analysis to the data from European Survey of Professional Forecasters and shows that this measure of uncertainty has a long-run nonlinear relationship with the share of non-rounded responses in the survey. Results suggest that the mean individual variance, often referred to as a direct measure of uncertainty, is a “noisy” variable directly associated with the changes in the methodology used by forecasters and as such absorbs information irrelevant for measuring uncertainty. The results are robust after having accounted for potential endogeneity.
Keywords: survey uncertainty; disagreement of forecasters; density forecasts; surveys of professional forecasters; Great Recession; smooth transition; instrumental variables.
JEL codes: C25, C32, C83, D81, E32, E37