An Identification Algorithm of Systemically Important Financial Institutions Based on Adjacency Information Entropy

Abstract

The identification and supervision of systemically important financial institutions (SIFIs) are crucial for regulators to address systemic risks to the financial system. Prevalent identification methods are biased because less attention is paid to the indirect connections among financial institutions. This study proposes a novel method for measuring the systemic importance of financial institutions. The proposed method is an identification algorithm based on adjacency information entropy that integrates the direct and the indirect connections among financial institutions. The sizes of financial institutions will not be investigated, and each financial institution is regarded as a node in the financial system network. The adjacency degree of every node is calculated. Subsequently, the information entropy of each node can be determined. The magnitude of the information entropy is taken as the importance measure of these financial institutions in the financial network. We conduct empirical analysis by employing the data of stock prices of 16 listed banks in the Chinese securities market with a sample period from October 2007 to December 2019 and adopt the method presented in this study to rank the systemic importance of these banks. Analysis shows that the ranking is in accordance with the reality, and the algorithm will promote the accuracy of identifying SIFIs.