Improving economic outcomes via AI-powered bank monitoring and risk management
PI: Nonna Sorokina (Business Division)
Plan for funding tuition for graduate students, or the remainder of the researcher’s salary for postdoc and research faculty: Self funded
Interest rate environment. Interest rate environment is central to the performance of the banking industry. Overtime, interest rates in U.S. varied significantly with a peak in 1981 at nearly 16%. However, for the past fifteen years we have been in a low interest rate environment that can only be compared with the post-WWII period. Such a prolonged time swatch may well represent an entire career for some financial professionals and entire adulthood time for many of their customers. For those who have grown up in the post-financial crisis times, after 2010, rising interest rate environment experience is unprecedented. Although we are nowhere close to the historical highs, a steep and persistent climb of the interest rate brings much uncertainty and particularly for those who have not seen it before firsthand.
Implications of improper interest rate risk management in a high interest rate environment. Banks are in the difficult business of financial illiquid claims (loans) with short-term funding sources (deposits). Not only loans are long-term commitments in many cases, but they are also subject to default and, sometimes, due to unexpected situations. As shown on the below graph from the Federal Reserve Bank of St. Louis, in the recent history, at least one wave of bank failures was strongly associated with the period of the high level and high volatility of interest rates (Figure 2). Thus, understanding the dynamics of bank operations and management are a matter of banking system stability and all consequential outcomes for the deficit units in the economy that rely on banks as liquidity providers.
Access to banking for underrepresented and low-income Americans and bank risk taking. The recent report of U.S. Congress Joint Economic Committee highlights how opportunities of underrepresented and economically disadvantaged U.S. population is disproportionally negatively affected by a lack of access to adequate banking services. Broady et al. (2021) elaborate on the details and implications of the challenges in access to banking services in black communities. At the same time, Federal Reserve Governor Michelle Bowman in her speech at the Wharton Financial Regulation Conference mentions that increase in risk in banking is predicted to further limit availability of banking services to underserved communities.
Costly supervision and AI. Bank failures are detrimental for economic development thus banking industry is heavily regulated. However, bank monitoring is costly and methodologically controversial. See, for example Barth et al., 2004. Banking regulation is ever evolving and received significant boost in the aftermath of the financial crisis of 2007-2009. However, only now effectiveness of the new regulatory framework is getting tested with the four bank failures this year 2023 that started with infamous Silicon Valley Bank (see FDIC’s Bank Failures in Brief – 2023). Michael Barr, the Federal Reserve Board Vice Chair for Supervision, in his recent report confirms that banking system is sound and resilient from the capital and liquidity standpoint, yet it is still vulnerable to the new risks. The newest failures expose the old problem. Supervision is generally unprepared facing new challenges even if those represent a modified version of old issues. Supervision does not have enough capacity and flexibility to act ahead of the coming issues, they act retroactively, at least to some degree. This is where AI-based methods should plug in. Bank call reports with wide array of information along with other bank data are publicly available and provide quite comprehensive real time picture of activities and conditions for all banks in the U.S. Despite the rapid decline in number of banks during most recent decades, there are still over 4,000 institutions in the country. Most of them are small- to medium-sized. Both comprehensive supervision and risk management are cost prohibitive for most of them, but the gap can be closed with an AI-powered monitoring system. Moreover, such a comprehensive approach to monitoring as opposed to one-bank-at-a-time analysis may allow to spot systematic issues more easily and prevent contagion (see for example Kaufman, 1994). I am looking to develop a monitoring system that will aid regulators in their supervisory function and small- to medium-sized banks in their risk management efforts. Increased stability of the banking system in this rising interest rate environment will shield underserved Americans from additional harm associated with economic instability and decline in access to banking services.
Direct Outcomes. This project will be ultimately developed into an impactful academic publication, and will be presented at multiple conferences, seminars and workshops. It will result in applications for the grants of the Center for Rural PA, Russel Sage Foundation, NSF Economics and others. I also plan to reach out to bank regulators and banking communities for implementation of the developed framework in practice.
How is funding for this project critical? The interest rates are high and concerns about stability of the banking system are on the rise. The time to act is now! The seed funding will provide crucial assistance in seeking funding for this research. This project aligns with Penn State’s land grant mission. It will promote economic development and reduce the burden of crises on society as a whole and specifically on the underrepresented and economically disadvantaged population. The project also promotes socially responsible AI practice and study. It will afford developing and application of AI for socially responsible economic practices that nurture welfare of the most vulnerable Americans.