我国金融机构尾部风险影响因素的非线性研究——来自面板平滑转换回归模型的新证据

“十三五”期间,我国防范化解金融风险攻坚战取得决定性成就,而在“十四五”规划开局之际,我国的金融风险形势面临新的挑战,防范风险仍是金融业的永恒主题。在此背景下,本文采用相对重要性分析技术方法,考察机构规模以及相关基本面因素对我国上市金融机构尾部风险的贡献程度。接着,本文结合边际效应分析技术考察机构规模对风险的异质性效应,深入分析“太大而不能倒”假说在中国的适用性。在此基础上,进一步运用前沿的面板平滑转换估计模型,研究机构规模与尾部风险的非线性关系,并分析基本面因素对该异质性效应的影响力度。研究结果表明,我国上市银行等金融机构规模的增加能够有效缓释我国金融系统的尾部风险,但该影响效应将随着特许权价值、资产质量、杠杆水平、成本水平、收入结构、贷款结构等基本面指标的变化而出现显著的非线性转变。在此基础上,对强化我国金融系统中的风险防控薄弱环节、提高金融机构的风险吸收能力提出建议,以期为我国深化金融业改革开放、推动高质量发展提供理论分析与实证检验的参考依据。

Nonlinear Analysis of the Determinants of Tail Risk:New Evidence from the Panel Smooth Transition Regression Model

Small financial institutions frequently encounter tail risk events such as insolvency and significant decline in asset quality in the post-crisis period. These events challenge the traditional supervisory concept of “too big to fail.” There is currently growing uncertainty in the capital market and increasing economic downward pressure. The Chinese capital market is also undergoing accelerating reform. It is therefore academically and practically important to investigate the intrinsic tie between bank size and tail risk and to explore the determinants of tail risk.This paper complements and expands the literature with a high level of originality. First, most domestic literature addresses risk contagion between financial institutions. There is little discussion of whether the “too big to fail” theory can be applied under China's actual economic conditions. Second, there is currently little consensus regarding the direction of effect of bank size on tail risk. The literature suggests that the fundamental variables of financial institutions actually play an important role in this relationship (Buch et al., 2019). Research highlights the need to include fundamental variables in the model to evaluate the heterogenous impacts of institution size on risk-taking more efficiently. Third, linear baseline regression models are often used when researching driving factors of tail risk. However, examining the relationships among variables under the traditional linear empirical framework may result in great bias, as indicated by Acemoglu et al. (2015) and De Vita et al. (2018). This bias makes it difficult to identify the risk sources in the financial system. Finally, research is likely to overlook the fact that the economic reform process exhibits an incremental trajectory in China when analyzing the nonlinear interconnectedness among variables. It is therefore more appropriate to discuss the smooth evolution of tail risk in China under the panel smooth transition regression (PSTR) model. Our sample consists of 44 Chinese listed financial institutions, comprising 11 banks and 33 non-bank institutions. The sample period runs from January 2008 to June 2020. MES is constructed to represent the tail risk level in this paper. All the data come from the Wind database.Our paper uses the dominance analysis method developed by Israeli (2007) and Givoly et al. (2019) to investigate the contributions of bank size and other fundamental variables to banks' tail risks. We find that bank size is not the only main determinant of bank risk; variables such as the non-performing loan ratio, personal housing loan ratio, and non-interest income ratio are also significant in the model. We next introduce the marginal effect analysis technique and provide strong evidence of the heterogeneous effects of fundamental variables on tail risk conditional on bank size. Using the PSTR model proposed by Cheikh and Zaied (2020) and González et al. (2017), this paper further discusses the nonlinear impact of bank size on tail risk and the roles of other fundamental variables in this relationship. The result indicates that an increase in the size of banks reduces the tail risk of the financial system in a highly nonlinear way. The reduction of tail risk depends on fundamental variables such as franchise value, asset quality, leverage, cost, income structure, and loan portfolio. The conclusions remain consistent and robust even when we extend our sample to 44 financial institutions. We also find that the evolution of tail risk is more volatile in financial institutions than in the banking sector. Our findings yield three important policy implications. First, the tail risks of small financial institutions deserve stronger supervisory attention and differentiated regulatory responses, especially at the level of cost management. Second, it is more appropriate to deleverage the financial sector gradually than in a rush. Finally, stronger integrated financial supervision is urgently needed to meet the emerging trend of cross mixed operation in the Chinese financial market. This paper thereby enhances insights into how to deepen financial reform and achieve high-quality economic development in China both theoretically and empirically.