TVP-VAR Based Carr-Volatility Connectedness: Evidence From The Russian-Ukraine Conflict

Abstract

This paper aims to examine the spillover between volatilities obtained from the Conditional Autoregressive Range (CARR) process with the Time-Varying Parameter Vector Autoregressive (TVP-VAR) based Diebold-Yilmaz approach. We apply Gumbel distributed CARR (1,1) to estimate the volatilities. The summary statistics for the volatility series indicate that the series are not normally distributed, and innovations fit the Gumbel distribution. Also, the obtained volatility series are stationary. We also observe that a significant autocorrelation emerges in all series and the square series. Therefore, using a TVP-VAR model with a time-varying variance-covariance structure is a proper econometric framework to capture all these empirical properties. Moreover, we investigate the impact of the Ukraine-Russia Conflict on global markets as an example. For this purpose, we consider the Russian stock market index and indices selected from among the twenty largest stock exchanges by asset size to perform the connectedness analysis. In TVP-VAR based connectedness approach, we calculate averaged connectedness measures of two panels, without and with the Russian stock exchange. The findings show that the total connectedness index is 79.91% in the first panel, and it increases to 81.44% with the addition of Russian market.