Does happiness forecast implied volatility? Evidence from nonparametric wave-based Granger causality testing

Abstract

Past studies on the causal relationship between online sentiment and implied volatility indices report mixed results. In this paper, we employ a daily happiness index extracted from Twitter as a proxy for online sentiment to investigate the relationship between sentiment and the Chicago Board of Option Exchange volatility indices. Applying a nonparametric wave-based Granger causality test, allows us close examination of frequencies and periods of causality. Investigating four different subsample periods, we evidence that daily happiness sentiment consistently manifests short-term causality of implied volatility indices. This is evidenced over our sample period as a whole, and for three out of four investigated subperiods—the exception being a subperiod of three years following the Global Financial Crisis. Additional Diebold and Yilmaz (2012) spillover testing evidences that the causal relationship between happiness and implied volatility is strongest for stable low happiness subperiods. Importantly, we also evidence intervals where spillovers from happiness to implied volatility spike sharply. This suggests that the relationship between happiness and implied volatility is heavily conditioned by exogenous factors. Consequently, while we identify consistent short-term causality of happiness to implied volatility measures, economically meaningful forecasting of implied volatility from happiness remains elusive. Our findings should be of considerable interest to investors and practitioners, as well as to behavioral finance researchers interested in how sentiment shapes the ‘uncertainty of uncertainty.