Volatility estimation for cryptocurrencies using Markov-switching GARCH models

Abstract

In the 21st century, digital currencies have become a disruptive technology that is shaking up both financial markets and academic environment. Investors, politicians, companies, and academics are attempting to improve their understanding of these currencies for future investment possibilities and technological applications. This study aims to evaluate changes in different volatility states of eight digital currencies (BTC, ETH, LTC, XRP, XMR, NEM, LISK, and STEEM) that showed the highest liquidity and market capitalisation from 2013 to 2017. The methodology involved the MSGARCH model, using SGARCH, EGARCH, GJRGARCH, and TGARCH models. Our study demonstrated that two volatility regimes, that is, one with a larger volatility and another with a smaller one, clearly exist for all the analysed cryptocurrencies. What differs between the currencies is the probability of a second regime occurring. Moreover, we concluded that for both the first and second state, the asymmetry coefficient (gamma) is positive for all currencies.