Empirical White Paper (Updated July 2023)
Report Summary:
This empirical white paper is part of a broader research project on the explainability and fairness of machine learning in credit underwriting. The empirical research was conducted in collaboration with Professors Laura Blattner and Jann Spiess at the Stanford Graduate School of Business.
This report considers the capabilities, limitations and performance of proprietary and open-source tools to help lenders manage machine learning underwriting models as required by law. The report focuses on use of the tools in: (1) generating individualized disclosures that state why particular applicants were rejected or charged higher prices; (2) identifying what factors in the model drive disparities in model predictions among different demographic groups; and (3) assessing the overall operation of underwriting models for risk management purposes.
The evaluation analyzes model diagnostic tools from seven technology companies–Arthur, H2O.ai, Fiddler, RelationalAI, SolasAI, Stratyfy, and Zest AI–as well as several open-source tools.
This white paper is an update of research published in April 2022.
Download the Report Download the Fact SheetAcknowledgments
Principal Investigators and Other Contributors:
Laura Blattner, co-principal investigator, is a former Assistant Professor of Finance at the Stanford Graduate School of Business. Laura earned her Ph.D. at Harvard University. She also holds a B.A. in Philosophy, Politics, and Economics and an M.Phil. in Economics from the University of Oxford. At Stanford, Laura teaches an MBA class on Financial Technology (FinTech). Laura’s research focuses on the governance and regulation of algorithmic credit underwriting.
P-R Stark, co-principal investigator, served as FinRegLab’s first Director of Machine Learning Research. In that role, she led the organization’s efforts to develop and execute policy-relevant research on the use of AI in financial services. She is an experienced advisor to financial institutions, helping firms respond to regulatory inquiries and manage adoption of new technologies, among other things. P-R holds an A.B. in Classics from Princeton University, an M.A. (Oxon) in Philosophy, Politics, and Economics from the University of Oxford, and a J.D. from Harvard University.
Jann Spiess, co-investigator, is an Assistant Professor of Operations, Information & Technology at Stanford University’s Graduate School of Business. He is an econometrician working on machine learning and causal inference. Jann is particularly interested in developing methods for transparent, robust, and replicable inferences from complex data.
Contributing Authors:
Sarah Davies
Duncan McElfresh
Sormeh Yazdi
Zishun Zhao
Data Science Team:
Mario Curiki
Georgy Kalashnov
Mashrur Khan
Fiona Sequeira
FinRegLab would also like to recognize the presenters and members of our project Advisory Board who contributed to productive discussion of the development of the design, execution, and interpretation of this research. The Advisory Board consists of subject matter experts from computer science, economics, financial services, and regulatory backgrounds and includes representatives from approximately 30 major institutions including bank and nonbank financial institutions, technology firms, advocacy and civil society organizations, and academic institutions. State and federal regulators participated as observers in Advisory Board meetings.
We would also like to thank the following individuals who provided valuable feedback on this report:
- Alexei Alexandrov
- Marsha Courchane and Adam Gailey, Charles River Associates
- Steve Dickerson, Raghu Kulkarni, and Kate Prochaska, Discover Financial Services
- Patrick Hall, bnh.ai
- Stephen Hayes and Eric Sublett, Relman Colfax PLLC
- Scott Lundberg, Microsoft Research
- Michael Umlauf and Gene Volcheck, TransUnion
We would like to acknowledge the FinRegLab team who worked on convenings and reports related to this project:
Natalia Bailey, Alex Bloomfield, Kelly Thompson Cochran, Colin Foos, Saurab Gautam, Gillous Harris, Tess Johnson, and Kerrigan Molland.
With support from
Mastercard Center for Inclusive Growth
The Mastercard Center for Inclusive Growth advances equitable and sustainable economic growth and financial inclusion around the world. The Center leverages the company’s core assets and competencies, including data insights, expertise, and technology, while administering the philanthropic Mastercard Impact Fund, to produce independent research, scale global programs and empower a community of thinkers, leaders, and doers on the front lines of inclusive growth. The Center has provided funding to support this research.
JPMorgan Chase & Co.
JPMorgan Chase is committed to advancing an inclusive economy and racial equity. The firm uses its expertise in business, public policy and philanthropy, as well as its global presence, expertise and resources, to focus on four areas to drive opportunity: careers & skills, financial health and wealth creation, business growth & entrepreneurship, and community development.
Flourish Ventures
Flourish, a venture of the Omidyar Group, has provided operating support to FinRegLab since its inception. Flourish is an evergreen fund investing in entrepreneurs whose innovations help people achieve financial health and prosperity. Established in 2019, Flourish is funded by Pam and Pierre Omidyar. Pierre is the founder of eBay. Managed by a global team, Flourish makes impact-oriented investments in challenger banks, personal finance, insurtech, regtech, and other technologies that empower people and foster a fairer, more inclusive economy.
Explainability and Fairness in Machine Learning for Credit Underwriting: Empirical Research
FinRegLab is working with a team of researchers from the Stanford Graduate School of Business to evaluate the explainability and fairness of machine learning for credit underwriting. Our focus is on measuring the ability of currently available model diagnostic tools to provide information about the performance and capabilities of machine learning underwriting models. This research will help stakeholders assess how machine learning models can be developed and used in compliance with regulatory expectations regarding model risk management, anti-discrimination, and adverse action reporting.
Learn MoreVirtual Conferences
Webinar: Getting Ahead of the Curve: Emerging Issues in the Use of AI and Machine Learning in Financial Services
January 17, 2024 – 2:30 PM ET – FinRegLab hosted a webinar with senior federal financial regulators to discuss the growing use of artificial intelligence and machine learning in financial services, including credit underwriting.
Learn MoreMachine Learning Explained: Evidence on the Explainability and Fairness of ML Credit Models
FinRegLab hosted a virtual conference in April 2022 featuring research being conducted by FinRegLab and Professors Laura Blattner and Jann Spiess of the Stanford Graduate School of Business on the use of machine learning in credit underwriting, with a particular focus on their potential implications for explainability and fairness.
Learn MoreArtificial Intelligence and the Economy: Charting a Path for Responsible and Inclusive AI
The U.S. Department of Commerce and National Institute of Standards and Technology, FinRegLab, and the Stanford Institute for Human-Centered Artificial Intelligence (HAI), hosted an April 2022 virtual conference, “Artificial Intelligence and the Economy: Charting a Path for Responsible and Inclusive AI.”
Learn MoreRelated Publications
Explainability & Fairness in Machine Learning for Credit Underwriting: Policy Analysis (December 2023)
This policy analysis explores the regulatory and public policy implications of the increasing use of machine learning models and explainability and fairness techniques for credit underwriting in depth, particularly for model risk management, consumer disclosures, and fair lending compliance.
Read MoreExplainability & Fairness in Machine Learning for Credit Underwriting: Policy & Empirical Findings Overview (July 2023)
This paper summarizes the machine learning project’s key empirical research findings and discusses the regulatory and public policy implications to be considered with the increasing use of machine learning models and explainability and fairness techniques.
Read MoreThe Use of Machine Learning for Credit Underwriting: Market & Data Science Context (Sept. 2021)
This report surveys market practice with respect to the use of machine learning underwriting models and provides an overview of the current questions, debates, and regulatory frameworks that are shaping adoption and use.
Read MoreAI FAQs (Series)
To provide a resource for financial services stakeholders, we have prepared FAQs on the use of AI in financial services. Individual entries highlight technological, strategic, and ethical debates relevant to the use of advanced modelling and analytical techniques in financial services, and many explain concepts relevant to our research on machine learning credit underwriting.
Read More