On the use of cross-sectional measures of forecast uncertainty

Abstract

This paper investigates the role of cross-sectional dependence among private forecasters, assessing its impact on the measurement and use of the forecasting uncertainty. We determine the circumstances under which cross-sectional measures of uncertainty (such as the disagreement across forecasters) are valid proxies for private information, and analyse the impact of distributional assumptions on private signals. In particular, we explore the role played by cross dependence among forecasters, arising from factors such as partially shared private information. We validate the theory through a Monte Carlo exercise, which reinforces our findings, as well as through an application to US nonfarm payroll data.