earticle

영어

As blockchain is developed and introduced to the world, requirements of bitcoin’s functions have been fulfilled, which are recording and validating transactions. For the past several years, cryptocurrencies including bitcoin have drawn various types of people’s attention. Although majorities’ views on cryptocurrency are still controversial, it cannot be contradicted that cryptocurrency is now a global phenomenon. In addition, more than 2800 cryptocurrencies are actively traded at a number of cryptocurrency exchanges and the current total market capital of all cryptocurrencies is still above $200 billion besides its dramatic fall in the beginning of 2018 (Coin Market Cap, 2019). In the long-term, high annual growth of cryptocurrency market is predicted reaching the total market capital of approximately $40 trillion (Toshi times, 2018). However despite the phenomenal attention given to cryptocurrency, not many people, even investors and researchers, are not aware of what cryptocurrency really is and how the cryptocurrency market works. Even though there has been a steady progress in the field of research, most of research are focused on particular topics such as investors’ sentiments on social media and predicting price fluctuation of a certain cryptocurrency (Garcia et al., 2014; Kondor et al., 2014; Kristoufek, 2015; Yelowitz and Wilson, 2015). On top of that, majority of research mainly focus on the fluctuation and price prediction of bitcoin and ethereum, or litecoin and ripple for the most. Apart from price prediction or fluctuation prediction Recently, there have been more and more attempts to accurately predict price fluctuation of cryptocurrency market using similar approach as used in stock market (Lamon et al., 2016; Kim et al., 2016; 2017). Price prediction for stock market is a traditional field of research with various approaches and methodologies to achieve higher level of accuracy (Masulis and Shivakumar, 2002; Zhai et al., 2007; Kim et al., 2014). Large portion of past research generally adopted sentiment analysis based on the dataset extracted from online news articles or social media platform such as facebook and twitter. As mentioned above, methodology-wise identical approach has been adopted to predict price fluctuation of cryptocurrency market. However, from past literature, it can be pointed out that past research overly relied on sentiment of investors to describe unpredictable fluctuation of cryptocurrency market. Stock market and cryptocurrency market differ in several characteristics. For instance, cryptocurrency market operates 24 hours whereas investors cannot trade on weekends and after certain period. Moreover, cryptocurrency market is based on global cryptocurrency exchanges where everyone can trade their cryptocurrencies regardless of time and space. In terms of trading time, trading market, number of shares, influence of issuer etc., and cryptocurrency market has indeed comparable characteristics with stock market. One distinct feature cryptocurrency market has compared to stock market is that cryptocurrency’s price volatility is considerably higher than stocks in terms of price (Reid and Harrigan, 2013; Boehme et al., 2015). This research aims at revealing the working mechanism of cryptocurrency market with the help of event study approach, especially focusing on articles related to regulations on cryptocurrency market. Until now, general tendency is that cryptocurrency market does not follow any rules and no one can accurately predict the fluctuation of the cryptocurrency market. Additionally, based on observations that several cryptocurrencies are less affected by bad news and tend to drift to a less extent, this research attempts to figure out underlying cause of the phenomenon. The data for this research are collected from two main sources, which are ‘coinmarketcap’ and ‘coindesk’. Since event study methodology requires both list of events that occurred and daily price of each cryptocurrency, two cryptocurrency platforms were chosen as main data sources. ‘Coinmarketcap’ has been frequently cited as an important platform, where researchers can crawl necessary data such as daily price, current rank according to its market capital, trading volume of each individual cryptocurrency. Cryptocurrency related articles are posted at ‘coindesk’ and further classified into various domains according to their topics. Among different domains, articles in ‘regulation’ domain are crawled for this research. In order to compute cumulative abnormal return (CAR) due to regulation related articles, estimation period to calculate abnormal return (AR) is set to either (-203, -3) or (-153, -3) based on previous literature and event window is set to either (-1, 1) or (-2, 2). In the pilot study with top 50 cryptocurrencies, a number of cryptocurrencies were significantly influenced by both positive and negative regulation articles generating abnormal returns, but not every cryptocurrencies were significantly affected. In addition to computing CARs of individual cryptocurrencies, further analysis will be conducted using CARs as the dependent variable in order to figure out features that cause abnormal returns. As explanatory variable investors’ sentiment crawled from twitter’s tweets, trading volume of each cryptocurrency and additional constructs will be used in order to determine the most influential factor. With this approach, it is possible to explore factors affecting the entire cryptocurrency market. After conducting the analysis, cryptocurrencies used in the current research can be further classified into few groups: 1) a group, which is not significantly, affected, 2) a group, which is significantly affected, but to a less extent, and 3) a group, which is both significantly affected and to a greater extent. Through the classification process, it might be possible to identify characteristics of each group and those characteristics can be implemented for further analysis and prediction. Expected contributions can be summarized as below. Unlike majority of past research which only focused on few well-known cryptocurrencies (i.e. bitcoin, ethereum, ripple etc.), this research comprehensively analyzes cryptocurrency market considering even minor cryptocurrencies. Since comprehensive approach has not been attempted yet, this research can provide basis for understanding underlying working mechanism of cryptocurrency market confirming that cryptocurrency market also follows certain rules and do not fluctuate randomly. Among various factors, this research helps to figure out which factor has the greatest impact on price fluctuation of cryptocurrency market. Event study methodology followed by additional regression with a number of explanatory variables can determine the most influential factor for fluctuations of cryptocurrency market.