Premium
Cryptocurrencies value‐at‐risk and expected shortfall: Do regime‐switching volatility models improve forecasting?
Author(s) -
Maciel Leandro
Publication year - 2021
Publication title -
international journal of finance and economics
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.505
H-Index - 39
eISSN - 1099-1158
pISSN - 1076-9307
DOI - 10.1002/ijfe.2043
Subject(s) - cryptocurrency , autoregressive conditional heteroskedasticity , volatility (finance) , econometrics , economics , value at risk , expected shortfall , markov chain , bayesian probability , stochastic volatility , statistics , mathematics , risk management , computer science , finance , computer security
This paper evaluates the presence of regime changes in the log‐returns volatility dynamics of cryptocurrencies using Markov‐Switching GARCH (MS‐GARCH) models. The empirical study compares the prediction performance of MS‐GARCH against traditional single‐regime GARCH methods for one‐, five‐ and ten‐steps‐ahead volatility forecasting of six leading digital coins such as Bitcoin, Dashcoin, Ethereum, Litecoin, Monero and Ripple. Using a Bayesian approach, different MS‐GARCH structures are estimated considering specifications up to three regimes, three scedastic functions and six error distributions, resulting in a total of 54 models for each cryptocurrency. Forecasts are compared according to an economic criterion, that is, through the estimation of Value‐at‐Risk (VaR) and Expected Shortfall (ES) risk measures. The results support the evidence of regime changes in the volatility process of selected cryptocurrencies and show that MS‐GARCH models do provide more accurate VaR and ES forecasts than their single‐regime counterparts.