Influencing factors of the risk correlation of financial institutions:Evidence from random forest fusion
Being"too interconnected to fail"has made the risk correlation of financial insti-tutions and its influencing factors a crucial issue in maintaining financial stability.Drawing inspiration from gene regulatory research using random forests,this paper proposes a method to construct a network that captures the relations between different indicators,for the purpose of exploring the influences between risk correlation and its related factors.It is achieved by inte-grating forest fusion and random permutation.The proposed method overcomes the limitations of traditional regression analysis,Granger causality test,and Bayesian networks,while the intro-duction of random permutation enhances the model's capability to handle variable heterogeneity.Empirical results based on 46 listed financial institutions in China from 2012 to 2022 demonstrate that the constructed network can identify the direct or indirect impact of different factors on risk correlation and reveal the influence paths of factors.This provides more comprehensive em-pirical evidence of complex relationships,highlighting the applicability of the proposed approach in addressing this issue and potentially offering a useful tool for financial regulation and risk management.
NETL
NSTL
万方数据
