Investor trading networks are gaining rapid interest in financial market studies. In this paper, we propose three improvements for investor trading network analyses: investor categorization, transaction bootstrapping and information aggregation. Each of these components can be used individually or in combination. We introduce a tractable multilayer aggregation procedure to summarize security-wise and time-wise information integration of investor category trading networks. As an application, we analyze the unique dataset of Finnish shareholders throughout 2004-2009. We find that households play a central role in investor networks, having the most synchronized trading. Furthermore, we observe that the window size used for averaging has a substantial effect on the number of inferred relationships. However, the relative node centrality in the networks is rather stable. We would like to note that the use of our proposed aggregation framework is not limited to the field of investor trading networks. It can be used for different non-financial applications, with both observable and inferred relationships, that span over a number of different information layers.
↧