Good versus bad information transmission in the cryptocurrency market: Evidence from high-frequency data

Abstract

Using 5-minute high-frequency data, we study realized volatility spillovers in major cryptocurrencies, employing generalized forecast error variance decomposition. We also include COVID19 period observations and report time-varying and asymmetric connectedness across various cryptocurrencies using realized volatilities and semi-variances. Our study provides diverse connections after distinctly considering good- and bad volatilities, which is unique in the related literature. Bitcoin and Ethereum are central to the system and dominant transmitters of positive shocks, while Litecoin propagates negative shocks abundantly. Ripple and Stellar are the least connected currencies with others, whereas Cardano and EOS are isolated in the network. This feature makes these currencies suitable diversifiers in a portfolio with other cryptocurrencies. Further, the majority of these connections are asymmetric in the long- and short-run. The time-varying and asymmetric nature of connections offers potentially unique opportunities for diversification and portfolios strategies. Total volatility connectedness is not only significantly enhanced but also changed in its nature during the COVID19 period. We observe no significant changes in results after the robustness check through varying lengths of the rolling-window. The findings are important to crypto investors and regulatory authorities for better diversification strategies and effective market oversight, respectively.

Keywords: COVID19, Cryptocurrencies, Good vs bad connectedness, High-frequency data