Asymmetric Bitcoin Volatility under Structural Breaks

We show that incorporating endogenously-determined structural breaks into an asymmetric GARCH model reduces volatility persistence in Bitcoin prices. We find that both good news and bad news have less impact on volatility if structural breaks are incorporated. Hence, our results may indicate that previous studies, that ignore structural breaks, might have overestimated the impact of news on Bitcoin volatility. In addition, our results suggest that Bitcoin shocks die out more quickly than previously thought. For example, the half-life of a GARCH model drops from 319 to only 34 days. Our paper provides quantitative evidence of the benefits of including both asymmetric effects and structural breaks in a GARCH model to correctly build Bitcoin volatility models