Disruptive innovation or systemic resilience? Investigating the impact of artificial intelligence on banking stability

This study examines the impact of artificial intelligence (AI) on banking stability in 37 OECD countries from 2010 to 2022, guided by Neo-Schumpeterian innovation theory which frames AI as both a stabilizing and disruptive force. AI is operationalized using annual venture capital funding directed toward AI-focused startups, covering domains such as machine learning, generative AI, natural language processing, and predictive analytics—data sourced from Crunchbase. We further validate this using an alternative proxy: the number of AI startups. Banking stability is measured by the Z-score, capturing the risk of insolvency. Using Panel-Corrected Standard Errors (PCSE) for baseline estimation, and two-step System Generalized Method of Moments (GMM) as the core identification strategy to address endogeneity, along with Heckman selection models to correct sample selection bias, the findings reveal that increased AI funding significantly enhances banking stability—particularly in technologically advanced and well-regulated OECD economies. Furthermore, mechanism analysis confirms the partial role of financial development in this relationship, while government effectiveness shows no significant mediation. Cross-country heterogeneity analysis reveals stronger effects in Western Europe and countries with advanced financial infrastructure. By integrating AI investment dynamics with institutional and technological contexts, this study provides actionable insights for designing responsible AI strategies that promote banking resilience while managing systemic risks.