Dissecting Time-Varying Risk Exposures in Cryptocurrency Markets
67 Pages Posted: 30 May 2020 Last revised: 5 Jun 2020
Date Written: May 1, 2020
In this paper we take an empirical asset pricing perspective and investigate the dominant view (possibly, an instinctive reﬂection of the media hype surrounding the surge of Bitcoin valuations) that cryptocurrencies represent a new asset class, spanning risks and payoﬀs suﬃciently diﬀerent from the traditional ones. Methodologically, we rely on a ﬂexible dynamic econometric model that allows not only time-varying coeﬃcients, but also allow that the entire forecasting model be changing over time. We estimate such model by looking at the time variation in the exposures of major cryptocurrencies to stock market risk factors (namely, the six Fama French factors), to precious metal commodity returns, and to cryptocurrency-speciﬁc risk-factors (namely, crypto-momentum, a sentiment index based on Google searches, and supply factors, i.e., electricity and computer power). The main empirical results suggest that cryptocurrencies are not systematically exposed to stock market factors, precious metal commodities or supply factors with the exception of some occasional spikes of the coeﬃcients during our sample. On the contrary, crypto assets are characterized by a time-varying but signiﬁcant exposure to a sentiment index and to crypto-momentum. Despite the lack of predictability compared to traditional asset classes, cryptocurrencies display considerable diversiﬁcation power in a portfolio perspective and as such they can lead to a moderate improvement in the realized Sharpe ratios and certainty equivalent returns within the context of a typical portfolio problem.
Keywords: Cryptocurrencies, predictability, portfolio diversiﬁcation, dynamic model averaging, time-varying parameter regressions
JEL Classification: E40, E52
Suggested Citation: Suggested Citation