Banks appear to be adopting AI faster than the broader finance, insurance, and real estate (FIRE) economy and other industries, according to a recent Economic Letter published by Federal Reserve Bank of San Francisco (FRBSF) economists. The report explores connections between bank size, AI adoption, loan composition, and credit outcomes.
For fund finance lenders, the initial findings create room for discussion on how to implement AI-driven data processing, which excels at crunching the hard data, without displacing relationship component, or soft data, embedded in a repeat-transaction sponsor product. It also raises the question of process changes influencing loan selection.
Large banks (assets over $100 billion) are implementing AI faster than peers, as reflected in a higher share of AI-related job postings. The findings are intuitive: Larger institutions are better positioned to fund upfront investments in technology. Job posting data may be an incomplete metric, particularly for small banks, because of contracts for outsourced vendor services.
AI incorporation into lending processes may tilt credit extensions toward larger, more complex loans that rely heavily on hard data at the expense of relationship driven lending to smaller enterprises where public data may not be available. In addition to loan selection, the report points to interesting insights: AI adoption also appears to be related to higher ROA, but also potentially higher problem loan shares (although this isn’t analyzed across bank size cohorts).
The report also leaves room for further study. Specifically, a time dimension would add depth and could relate changes in ROA to AI adoption for the same institutions. Second, the focus on ROA, a broad indicator, could be supplemented by narrower measures that focus on loans and specific lending sectors. The additional data would help control for differences unrelated to AI, such as funding costs, interest income from securities, trading and asset management income, and operating efficiency that show up in ROA.
Given the size of the NDFI market ($2.0 trillion in the most recent H8 report), the growth rate (18.5% year-over-year), and lender reliance on LP credit scoring and monitoring, AI questions are particularly relevant in fund finance and likely to grow in significance in coming months.