Realized Bitcoin Volatility

Cryptocurrencies are a relatively new phenomenon of the 21 stcentury. Hence, it is as of yet not clear whether they have the capability to evolve to full-fledged currencies accepted world wide, or if they remain a means of speculation. A prominent example for these cryptocurrencies and the focus of this article is Bitcoin. It’scurrent status is not yet clear, albeit most studies tend to classify Bitcoin as an investment (Baur, Lee, and Hong, 2015; Bell, 2013; Glaser, Zimmermann, Haferkorn, Weber, and Siering, 2014). In this article, we conduct a detailed analysis of the volatility in the Bitcoin market. Both aspects—speculative investment or currency—are heavily influenced by the level and nature of the Bitcoin market volatility and we suggest that excess volatility (as compared to mature foreign exchange markets) would classify Bitcoin more as a speculative investment than a currency.

The literature on Bitcoin is relatively recent. Brandvold, Molnár, Vagstad, and Valstad (2015) focus on price discovery between different Bitcoin markets while Cheah and Fry (2015) investigate speculative behavior in Bitcoin trading, in line with the hypothesis that Bitcoin is a speculative asset. There are a number of studies that look into the volatility of Bitcoin, but none of these uses realized volatility albeit the availability of high frequency transaction data. Dwyer (2015), for example, uses monthly standard deviations of Bitcoin prices from Mt. Gox, BTC and Bitstamp. Bouoiyour and Selmi (2016), Bouri, Azzi, and Dyhrberg (2017), and Dyhrberg (2016) rely on GARCH models to estimate daily volatility. All authors conclude that the volatility level is comparatively high, offering different explanations such as cyber attacks, information asymmetry, decentralization, or the absence of regulation. In contrast, the literature investigating foreign exchange (FX) volatility is well-established and diverse. Also, high frequency data and realized volatility are readily used (see, for example, Andersen, Bollerslev, Diebold, and Labys, 2001; Barndorff-Nielsen and Shephard, 2002; Berger, Chaboud, and Hjalmarsson, 2009; Zhou, 1996).

We contribute to the literature by an in-depth analysis of Bitcoin realized volatility. One of the fundamental features of Bitcoin is transparency and the availability of the entire transaction history which allow us the construction of a novel, daily realized variance dataset. We use data for six markets, covering exchange rates with the US dollar,the Chinese renminbi, and the euro. For comparison with well established foreign exchange markets,we use USD/EURandUSD/YEN exchange rates. We find that Bitcoin markets exhibit excess volatility in the sense that the volatility observed there is up to 30 times higher than in FX markets. We understand such a high level of volatility as an obstacle for Bitcoin to perform all task associated with a currency (means of exchange, store of value, unit of account) in a reliable and efficient manner. Also, we find persistence of Bitcoin volatility to be lower than persistence of exchange rate volatility. Furthermore, we suggest that information processing (in the sense of Clark, 1973) works efficiently, but the variability of information is relatively high. Weproposethatthe latteris due tothe fact that the value of Bitcoin is to the greatest extent driven by individual traders’ personal assessment which might be quite diverse, depending on the intended use of Bitcoin.