The increasing complexity of financial networks and the expansion of interconnections among banks have made systemic risk one of the most critical threats to financial stability in modern economies. The 2008 global financial crisis demonstrated that the failure to identify systemic vulnerabilities at an early stage can lead to severe consequences for both the banking system and the macroeconomy. In this context, stress testing, as one of the key risk management tools, has been widely used by central banks and supervisory authorities to assess banks’ resilience to economic shocks. However, traditional stress testing approaches are largely based on predefined scenarios and exhibit limited flexibility when dealing with dynamic and nonlinear financial environments.
The emergence of generative artificial intelligence technologies, particularly large language models, generative adversarial networks, and diffusion models, has created new capabilities for generating complex scenarios, capturing nonlinear relationships, and simulating rare events. This study aims to propose a novel framework for stress testing and scenario generation of banks’ systemic risk based on generative AI models. In this framework, macroeconomic data, banking stability indicators, financial market variables, and interbank network information are used as model inputs, enabling generative models to produce and evaluate a wide range of plausible crisis scenarios.
Theoretical analysis indicates that the application of generative artificial intelligence models can significantly improve the accuracy of banking crisis prediction, enhance the detection of hidden patterns, increase the capability to generate emerging scenarios, and improve the flexibility of the stress testing process. Furthermore, this technology enables the assessment of risk contagion effects within the banking network and the identification of systemically important banks. The findings suggest that integrating generative AI with existing regulatory frameworks can enhance financial stability, improve supervisory decision-making, and increase the resilience of the banking system against economic shocks.