Measuring the AI-Governance-Performance Nexus in Financial Decision-Making

Abstract

Research on artificial intelligence in finance often reports model accuracy without explaining how institutional governance shapes real decision outcomes. This review examines the analytical relationship among artificial intelligence integration, data governance, and financial performance using the quantitative patterns reported in a large systematic evidence base. It summarises the distribution of research across time, countries, methods, technologies, sectors, and governance themes, then interprets the reported results from supported vector regression, quantile-sensitive analysis, structural equation modelling, multi-group comparison, and sensitivity testing. Machine learning, regulatory compliance, fraud detection, risk analytics, and credit scoring showed particularly strong associations. The structural model indicated that artificial intelligence contributes to financial outcomes directly and indirectly through data governance, supporting a partial mediation interpretation. Results also varied by governance maturity: basic data quality and compliance were more important in early implementation, while hybrid models and compliance automation became stronger in advanced environments. Temporal comparison suggested that direct relationships strengthened after 2020 while the mediating role of governance remained stable. The article evaluates the strengths and limits of these methods and proposes a research agenda based on multi-database evidence, longitudinal modelling, causal designs, and context-sensitive validation.