Volatility expectations and disagreement

Abstract

This paper examines the use of survey-based measures in volatility forecasting. We argue that an aggregate volatility forecast built up from individual forecasts should be the sum of individual expected volatilities and the dispersion in mean return forecasts. We use data coming from a repeated survey to capture volatility expectations and mean returns of investors, and to produce aggregate volatility forecasts. Our survey-based volatility forecasts are consistent and quantitatively similar with forecasts based on GARCH and implied volatility models. This result is robust to both in-sample and out-of-sample comparisons and in response to news.