Overconfidence Test in Cryptocurrencies Markets Using VAR Analysis

Abstract

The rapid technological development in finance in general, and in BTC in particular, has continued to gain traction worldwide since its emergence in the last decade. Initially, this growth was largely due to the work of tech enthusiasts; however, this changed in 2017, when BTC and various other CCs became main selling points in the financial world. Since the early peak of these electronic-cash developments in 2018, Google searches for ‘Bitcoin’ and ‘cryptocurrencies’ have continued to reach record highs. Most recently, on the 2nd of April, 2019, this trend in Google searches for cryptocurrencies, particularly ‘BTC, reached 90 percent of the peak observed on the 19th of November, 2018. This market for cryptocurrencies (hereinafter referred to as ‘crypto-market’) has gained substantial attention from theorists and empirical researchers, but the inherent complexities in this phenomenon are yet to be fully explored. Motivated by overconfidence theory, in the current study, we question whether the recent degree of volatility and trading activity in cryptocurrency behaviour shares any overlap with the momentum of the Dot-Com bubble in the 1980s, where tech-based infrastructure gained power and became overconfident very quickly, yielding spurious share prices for these firms and their products. Following this, it is expected that an identifiable pattern in price behaviour will emerge after correction to such significant growth (Caporale and Plastun, 2018). Moreover, it is worth noticing that the majority of crypto-participants are overwhelmingly male (Coindesk, 2018) and psychological research has established that, in the area of finance, men are more overconfident compared with women (Barber and Odean, 2001). Despite studies on overconfidence originating in the field of psychology, such investigation has migrated into the literature on economics and finance. Particularly, overconfidence is considered as the key behavioural factors needed to understanding the trading puzzle (De Bondt and Thaler, 1995). Indeed, overconfidence theory had taken its place in the growing list of behavioural studies which were used to be on the fringes, but are now occupying mainstream research in relevant fields. Accordingly, although our analysis was initially motivated by the overconfidence hypothesis, the findings related to the dependence of trading volume and overconfidence are considered an essential empirical principle which should be duly acknowledged by both theorists and empirical researchers. This thesis investigates the lead-lag relationship between the turnover and return of the crypto-market in general, and the three largest market cap coins (BTC, ETH and XRP) in particular. Alternatively stated, we attempt to support notions of the existence of overconfidence bias in this market. The study utilised the Vector Autoregression (VAR) technique, as well as the Granger-causality and Impulse Response Function (IRF) to produce the three key findings of this study. The first key finding is that the activity in the crypto-market suggests the explanatory power of overconfidence during its tremendous volatility in 2018 is statistically significant but economically subtle. Accordingly, this study found evidence of overconfidence bias in Bitcoin and Ripple investments, as well as in the crypto-market, wherein the trading activities were responsive to lagged-market returns. The relatively pronounced dependence of ETH turnover on its lagged-return presents our second finding, which suggests ETH 3 participants trade under the disposition effect. We also perceive that BTC, ETH and XRP differ in terms of fundamental drivers and investment intentions, which may account for the economically nonsignificant findings when performing joint interpretations. In addition, we illustrate a high degree of similarity in price movement between BTC and the Dot-Com bubble, as they follow the same psychologically-based market cycle. Therefore, theoretically speaking, it is predicted the significant movement in the price of BTC, as well as the performance of cryptocurrencies will track the market crash of the 1980s due to the overconfidence bias. The conclusion is consistent with a working paper by Obryan (2018). Practically speaking, however, we more consider differences in fundamental underlie cryptocurrencies and its specific long-term use case, which make the price momentum should be interpreted differently. The level of adoption of these tech-integrations then should become the central focus of market participants.