Who Moves First? Information Flows in Cryptocurrency Exchange Markets

We analyze the price discovery process in several cryptocurrency exchanges. Our primary purpose was to investigate the direction of information flows among different exchanges in order to reveal informationally dominant exchanges. It is a common feature for assets and their underlying derivatives to be traded on different trading venues. That is the case for cryptocurrencies too: trading activities are highly decentralized among thousands of exchanges. Therefore, in order to evaluate the market properties, it is crucial to know where the new information is initially incorporated. Our hypothesis is that some venues primarily contribute to the price discovery, while the others mainly follow them. This work aims at determining these informationally dominating exchanges. We apply econometric methods proposed in the work of Hasbrouck (1995), where author introduces the concept of Information Shares. We obtained data from nine exchanges on BTC/USD and ETH/USD pairs in the form of execution trades using parser. We followed the estimation procedure described in Hasbrouck (1995). In order to resolve potential problems with interpretation of results we applied modification of the model proposed in the work of Lien and Shrestha (2009). Most of the calculations were made in R. After data adjustment, the final database consisted of ten days of observations. For each of these days information shares were calculated. Our results showed that, indeed, there are two dominant markets for Bitcoin, each having about 30% information share, and one dominant market for Ethereum with more than 50% information share.