Risk in the cryptocurrency markets: the role of structural breaks and fat-tailed distributions in estimating value-at-risk and expected shortfall

Cryptocurrencies have gained much attention in recent times with investors, speculators, and regulators showing a keen interest in the cryptocurrency markets. However, not much attention has been paid to quantifying their risk measures. This paper estimates the risk in the cryptocurrency markets using Value-at-Risk and Expected Shortfall. We use Johnsons Su distribution to model the innovations in the returns and present a comparative analysis of different fat-tailed and skewed distributions used in modeling the returns. The estimation takes into account endogenously determined structural breaks in the data. We employ several backtesting methodologies to test the efficacy of the forecasts. Empirical results show that the Johnson's Su distribution gives exceptional results, and outperforms other fat-tailed distributions and the normal distribution, especially at the 1% (for long positions) and 99% levels (for short positions). Furthermore, our results are robust to different subsamples and the methodology employed (recursive or rolling window). Our results have clear policy implications for various market participants, regulators, and the government.