Essays on Forecast Evaluation of the European Central Bank Survey of Professional Forecasters

This thesis consists of three essays on the evaluation of the European Central Bank Survey of Professional Forecasters (ECB SPF). Chapter 1 serves as an introductory chapter and it describes the European Central Bank Survey of Professional Forecasters and revisions in macroeconomic data. Chapter 2 examines the rationality of ECB SPF aggregate point forecasts using a test that allows less restrictive assumptions on the loss function and fixed-smoothing asymptotics to avoid small sample size distortion. ECB SPF forecasts are shown to be rational. Chapter 3 explores the evaluation of ECB SPF aggregate point forecasts using the Diebold and Mariano test for the null hypothesis of equal forecast accuracy with asymmetric loss functions. I address the small sample size distortion of the Diebold and Mariano test with fixed-smoothing asymptotics. Results show that ECB SPF are slightly more accurate than benchmark models in some cases, especially for short horizons, while their advantage decreases as the horizon extends. Chapter 4 evaluates the accuracy of ECB SPF aggregate density forecasts with the Diebold and Mariano test with loss functions which accommodate the fact that forecasts are reported as histograms. As in the previous Chapters, small sample size distortion of the test is taken into account using fixed-smoothing asymptotics. ECB SPF density forecasts show a good predictive ability, especially for unemployment and real GDP growth forecasts. For inflation, there is no evidence that the ECB SPF forecasts are better than simple benchmark forecasts because inflation was anchored to the ECB target.