Does risk aversion predict the future real economy?

Abstract

This study evaluates the forecasting ability of various risk aversion measures for future U.S. real economic activity (REA). Recognizing that widely used proxies for risk aversion differ significantly in their construction and behavior, we assess their empirical validity using multiple criteria, including leading-indicator properties, counter-cyclicality, persistence, and volatility. We conduct both in-sample and out-of-sample forecasting exercises, along with subperiod analyses. While most measures exhibit strong in-sample performance, their out-of-sample accuracy varies with macroeconomic conditions. These results underscore the state-dependent nature of risk aversion and highlight its potential usefulness as a forward-looking indicator of real economic activity.