Title: Entropic tilting for macroeconomic variables: The role of asymmetrically distributed survey forecasts
Authors: Anastasia Allayioti - University of Warwick (United Kingdom) [presenting]
Abstract: There is a growing interest in incorporating external information extracted from a survey of professional forecasters into real-time macroeconomic predictions from vector autoregressive (VAR) specifications. The method of entropic tilting achieves this by modifying the baseline VAR distribution such that it matches certain moment conditions. Existing papers adopting this methodology focus on the first two moments of the forecast distribution (mean and variance). By implicitly assuming that the first two moments summarize all required information, these papers restrict their attention to a symmetric environment. We propose a modification to the standard relative entropy approach which allows for asymmetry in the macroeconomic variables and explores the predictive content of higher-order moments. The proposed methodology involves tilting the VAR distribution towards an aggregate survey forecast density that has been appropriately reshaped to match the non-Gaussian features of the sample data. We illustrate this methodology with an application examining real-time forecasts for four U.S. macroeconomic variables. We consider a variety of VAR models, ranging from time-varying volatility to non-Gaussian errors. Results across models indicate meaningful gains in terms of both point and density forecast accuracy relative to individual multivariate specifications and existing forecasting methods that blend model-based forecasts with external judgement.