Explosive behaviour in cryptocurrency prices

Abstract

Bitcoin and other cryptocurrencies have enjoyed several well-documented episodes of price exuberance since they emerged on the scene. However, in May 2022, an almost perfect storm shook their world, causing their prices to plummet to unexpected lows. This dissertation aims to contribute to the quest for episodes of exuberance, suggestive of bubbles, in cryptocurrency prices by conducting a comprehensive empirical analysis, which employs state-of-the-art tests for explosive behaviour: the Generalized Supremum Augmented Dickey Fuller (GSADF) test developed by Phillips, Shi and Yu (2015) and, as an original contribution, its panel version proposed by Pavlidis et al (2016). These tests are complemented by an array of other econometric techniques, such as unit root and cointegration tests, and Granger causality tests, to conduct a comprehensive formal investigation of periods of explosiveness in cryptocurrency prices. The formal analysis is complemented and enhanced by a narrative analysis, which aims to shed light on the factors that triggered the episodes of explosiveness. A thorough examinations of the causes of these episodes is timely, as the recent dramatic crash in the prices of cryptocurrencies is evocative of past bubble episodes. To this end, I use data for (at least) the last 5 years (2017-2022) for 7 major cryptocurrencies, which feature in the top 20 cryptocurrencies by market capitalization (Coinmarketcap.com, 18 July 2022). Moreover, after detecting the periods of exuberance in cryptocurrency prices and analysing their causes, I examine whether there have been synchronized periods of explosiveness across the selected cryptocurrencies. Furthermore, I analyse if explosiveness in one or more individual cryptocurrencies led to global explosiveness episodes across all major cryptocurrencies by using a Logit model. Additionally, using Granger causality tests, I establish connectedness across cryptocurrencies, by unveiling some causal links between them. To set the scene for the empirical investigation, I provide a brief introduction about bubbles, types of bubbles, and detection of bubbles in asset markets. Subsequently, I provide a thorough review of the relevant literature on cryptocurrencies as a different asset class, and the existing evidence of bubbles in the cryptocurrency markets. In the empirical analysis, using the GSADF test I detected several explosiveness (bubble) periods in the individual cryptocurrencies and in the panel made of selected currencies. The results reveal that Bitcoin experienced most of the long-lived bubbles. Most of the detected explosivity periods coincide with the ones documented by previous studies. However, given that the data span of my analysis includes the more recent two years, I pay special attention to the period 2020-2021. I unveil evidence of synchronicity among cryptocurrencies, especially during periods of financial turbulence. This dissertation contributes to the ongoing debate about speculative bubbles in cryptocurrency markets, by using an extended the dataset and by conducting not only GSADF tests for individual cryptocurrencies, but also the Panel GSADF in order to detect possible multiple bubbles in the cryptocurrency markets, as well as by demonstrating the existence of synchronised periods of explosivity among cryptocurrencies, that ultimately can help investors in terms of portfolio management.

Date of Award	28 Sept 2022
Original language	English
Awarding Institution	Universidade Católica Portuguesa
Supervisor	Paulo Alves (Supervisor)