Discover Regional and Size Effects in Global Bitcoin Blockchain via Sparse-Group Network AutoRegressive Modeling

Bitcoin blockchain has grown into an active global virtual money network with millions of accounts.We propose a Sparse-Group Network AutoRegressive (SGNAR) model to understand the dynamics of its cross-border transactions. It describes the money flows of virtual funds, with a focus on the regional and size effects in the Bitcoin network at a global level. In particular, we develop a regularized estimator with two-layer sparsity, which enables discovering 1) the active regions with influential impact on the global network and 2) the size of the groups which lead the dynamic evolution of the Bitcoin transaction network. Our study considers the up-to-date Bitcoin blockchain, from February 2012 to July 2017, with all the transactions being classified into 60 groups according to region and size. We found that mostly the users with the smallest and largest sizes of transactions from North America, Europe, South America, Africa and Asia were driving the Bitcoin transactions, while the other groups and all the groups in Oceania were either followers or isolated.The global connectivity remained low in the period from 2013 to 2015, although it was high in 2012 and enhanced in the recent years of 2016 and 2017