Real estate cycles and credit risk in Chinese commercial banking

This paper examines how real estate market cycles shape credit risk in Chinese commercial banks, with an emphasis on residential mortgage lending. Using an unbalanced panel of 145 banks over 2013–2024, we apply stochastic frontier analysis to decompose the non-performing loan (NPL) ratio into an inherent risk component related to loan scale and portfolio composition and an inefficiency component that captures excess non-performance attributable to weaknesses in monitoring and risk recognition. We find that during the housing upswing, higher residential mortgage exposure is associated with lower measured lending inefficiency, consistent with collateral appreciation and delayed recognition suppressing observed non-performance. After 2021, when house prices softened, mortgage growth slowed, and regulatory tightening intensified, this negative association attenuates markedly. The state-dependent pattern is concentrated among state-owned and joint-stock banks with higher real estate exposure, whereas urban and rural commercial banks exhibit comparatively stable inefficiency dynamics. Robustness checks using alternative credit-risk proxies yield consistent conclusions.