Risk connectedness between cryptocurrency and commodity markets: empirical result from frequentist and bayesian statistic approaches

Abstract

The financial landscape has witnessed a rapid transformation, marked by the appearance of Cryptocurrencies and global geopolitical tensions. In response, an examination of the interplay between established assets like Gold, Oil, Gas, and the emerging realm of Cryptocurrencies becomes paramount. Employing established financial models, including the Multivariate GARCH framework, our analysis, rooted in the DCC-GARCH (1,1) model, substantiates the presence of significant return and volatility spillover effects within the assets. These findings underscore the risk connectedness of financial assets. Furthermore, we leverage the GO-GARCH (2,2) model, fortified by Principal Component Analysis (PCA). This approach illuminates the vital role of Gold as a hedge against Cryptocurrency risk, particularly during adverse market conditions. However, Cryptocurrencies, as elucidated by the GO-GARCH (2,2) model, also exhibit resilience to various market components influenced by geopolitical conflicts, positioning them as potential hedge instruments during political conflicting periods. The diagonal-BEKK (1,1) model with asymmetric terms excels in discerning that Cryptocurrency stands as the only asset class susceptible to excessive volatility offering Cryptocurrency investors early warning signals in the face of rising volatility. The finding challenges traditional economic wisdom in the relationship between risks and rewards of investment, which contributes to the normative and positive economic study. Lastly, Bayesian network estimations, applied to both the DCC and BEKK models, reveal intriguing insights. They indicate that conditional correlations between Cryptocurrencies, notably Bitcoin and Ethereum, can surpass those established through the Frequentist approach. Conversely, the Frequentist approach may potentially underestimate the conditional correlation between Gold and Cryptocurrencies. The Bayesian framework potentially addresses the curse of dimension in time series analysis and enriches the decision-making toolkit for investors and fund managers. Indeed, these findings hold substantial significance for future research endeavors