Which Uncertainty Measure is Most Informative? A Time-varying Connectedness Perspective

Abstract

We investigate the relationship between the three most popular uncertainty measures with the means of the state-of-the-art connectedness frameworks applied to the time-varying parameters vector autoregression model with stochastic volatility. We find marked increases in uncertainty connectedness during major economic turmoil and hostile events. VIX turns out to be the most forward-looking uncertainty measure that persistently transmits shocks to the remaining uncertainty proxies at lower frequencies. In turn, GPR, approximating specific information related to geopolitical risk, transmits shocks to other measures at short-term frequencies, while the EPU index is largely replicating unanticipated movements in the VIX or GPR. We also present implications of these findings for economic modelling.