Network analysis of aggregated money flows in stock markets

Abstract

We introduce the formation of a network of money flows between assets in stock markets, which captures directed relations between assets in terms of how investors have re-allocated money in the stock exchange. Our approach is based on identifying a directed link, or money flow, that occurs when an investor funds a purchase of an asset by selling another asset(s). We extract investor-level money flow networks on daily basis from shareholder registration data, which are then aggregated for both financial institutional and retail investors. Overall, we have a time series of 877 daily networks from 2006 to 2009, which is exceptionally long data on temporal networks. Through our analysis of non-reciprocated triadic patterns in the aggregated money-flow networks, we find that these patterns are both recurrent and significant. However, they are not related to the 2008 financial crisis. Additionally, we observe that the counts on different triadic motifs exhibit not only an autoregressive process but are also interconnected contemporaneously and dynamically. These findings suggest the need for further research using sophisticated network models to provide a comprehensive representation of money flows in stock markets.