Volatility estimation for cryptocurrencies: Further evidence with jumps and structural breaks

Amélie Charles (1) , Olivier Darné (2)

1 Audencia Recherche
2 LEMNA - Laboratoire d'économie et de management de Nantes Atlantique

In this paper we study the daily volatility of four cryptocurrencies (BitCoin, Dash, LiteCoin, and Ripple) from June 2014 to November 2018. We first show that the cryptocurrency returns are strongly characterized by the presence of jumps as well as structural breaks (except Dash). Then, we estimate four GARCH-type models that capture short memory (GARCH), asymmetry (APARCH), strong persistence (IGARCH), and long memory (FIGARCH) from (i) original returns, (ii) jump-filtered returns, and (iii) jump-filtered returns with structural breaks. Results indicate the importance to take into account the jumps and structural breaks in modelling volatility of the cryptocurrencies. It appears that the cryptocurrency returns are well modelled by infinite persistence (BitCoin, Dash, and LiteCoin) or long memory (Ripple) with a Student-t distribution.

Domaines: Sciences de l'Homme et Société; Economies et finances

Amélie Charles, Olivier Darné. Volatility estimation for cryptocurrencies: Further evidence with jumps and structural breaks. Economics Bulletin, 2019, 39 (2), pp.954-968. ⟨hal-03794543⟩