The Nonlinear Impact of Climate Physical Risk on Bank Non-Performing Loan Ratios: Evidence from Chinese Listed Banks

Abstract

The frequent occurrence of extreme weather exacerbates the threat of climate physical risks to financial stability. However, whether its impact on bank credit risk follows a simple linear pattern remains unclear. Based on the unbalanced panel data of 28 listed banks in China from 2010 to 2023, this paper tests the nonlinear effect of extreme precipitation on the non-performing loan rate. The results indicate a significant inverted U-shaped relationship between extreme precipitation and the non-performing loan rate, that is, the initial physical loss pushes up the risk of default, and the non-performing rate falls after exceeding the threshold, which may be affected by post-disaster policy buffering and other factors. The U-test formally confirms that the inverted U-type relationship is established within the data range (p=0.015). The effect is most significant in the two lag periods and concentrated in small banks. Large banks did not show a significant nonlinear effect due to the high degree of business dispersion. Further analysis found that small banks in areas with higher insurance coverage exhibit a more pronounced inverted U-shaped pattern, which is consistent with the theoretical expectation that insurance compensation plays a buffering role. After multiple verification of hybrid OLS, fixed effect, wild cluster bootstrap and multi-phase lag combined regression, the conclusion is robust. This study provides new micro-level evidence for the nonlinear boundaries of climate risk transmission to the banking system.