Dynamic connectedness between COVID-19 news sentiment, capital and commodity markets

Abstract

This study investigates the dynamic transmission mechanism between COVID-19 news sentiment (Google Trends Index), and S&P100, crude oil and gold volatility indices using the recently developed time-varying parameter vector autoregressive (TVP-VAR)-based extended joint connectedness approach. This framework corrects for the Generalized Forecast Error Variance Decomposition (GFEVD) normalization problem. The obtained empirical results suggest that dynamic total connectedness is heterogeneous over time and severely affected by COVID-19. More importantly, we identify COVID-19 news sentiment to be the main driver of spillover shocks indicating that it is indeed an important predictor of the volatility indices employed in our study. Thus, our findings have important implications for policymakers, private investors, as well as for portfolios and risk managers.