There have been an increasing number of empirical papers that investigate the bubble price dynamics in cryptoasset markets. Being the first and most famous cryptoasset, the majority of studies focus on Bitcoin prices. For instance, Cheung et al. (2015) use daily Bitcoin data from 2010 to 2014 and adopt the Phillips et al. (2015) methodology to examine whether price bubbles exist in Bitcoins traded on what was previously the largest exchange, Mt. Gox. Estimates from the Supremum Augmented Dickey-Fuller (SADF) test reveal that most of the bubble-like periods over that time span did not last for long, and indeed only a few days. Similarly, Corbet et al. (2018) employ daily data on Bitcoin from 2009 to 2017 and provide evidence of Bitcoin bubble behaviour around the turn of the year from 2013 to 2014. Moreover, Ethereum exhibits bubble behaviour in the beginning of 2016 and in the mid-2017.

Geuder et al. (2019) also look at the existence of bubble-like dynamics in the price of Bitcoin. They look at the period from 2016 to 2019 and investigate the presence of bubbles based on two alternative methodologies: the Phillips et al. (2015) method, which is based on a supremum augmented Dickey-Fuller (ADF) test, and log-periodic power law (LPPL) approach by Filimonov and Sornette (2013), which is based on the alternative idea that prices in a bubble period do not follow an exponential dynamic, but rather a faster-than-exponential price growth. Empirically, they show that there are indeed bubble periods in Bitcoin prices, regardless of the testing methodology implemented. For instance, after December 6, 2017, Bitcoin’s price growth rate changes and bubble deflates as part of the latest bubble period ending in January 2018. Before December 6, 2017, the LPPL test show that the Bitcoin price is characterized by faster-than-exponential growth.

Perhaps on a different note, more recently, Perepelitsa and Timofeyev (2020) argue that the fact that Bitcoin can be infinitely divisible in smaller units makes possible the emergence of self-sustained price bubbles. That is, a group of investors can “pump” the price up at a positive rate, without ever running out of cash. Although the idea of never-ending bubbles can be appealing to any investor, the idea is questionable at the very least. More precisely, the idea of Perepelitsa and Timofeyev (2020) is that, given BTC is infinitely divisible, thus everyone can afford a fraction of it, prices will keep increasing as far as the fraction of optimist investors dominates the pessimistic ones. Although it is reasonable to think that investors acting adaptively can generate long-lasting, positive price dynamics, it is perhaps less likely that optimists will continue to dominate pessimists in the long run. Nevertheless, Perepelitsa and Timofeyev (2020) pointed out a key aspect of the price dynamics of BTC, that is, it can be thought of as primarily driven by investors’ sentiment, an aspect that is constantly referred to in existing empirical research.

All in all, the evidence on bubble periods in Bitcoin prices is somewhat robust across methods and studies. However, neither the timing nor the magnitude of these bubbles is consistent across studies, which raise some question about the actual reliability from a purely market timing perspective. In addition, a natural question that arises from the existing empirical results is what eventually caused these episodes of bubble behaviour. Apart from the rational bubble view, there are many further reasons and determinants that may help to explain Bitcoin price behaviour. Given the global monetary environment for fiat currencies, institutional changes such as the introduction of Bitcoin futures, the use of Bitcoin for web-based transactions, governments restricting the use of cryptoassets and other regulatory changes may have significantly influenced Bitcoin price dynamics. A reduced-form econometric framework is unlikely to provide any meaningful answer in this respect.