FinRegLab presented a research proposal for research on the use of federated machine learning in Bank Secrecy Act/Anti-Money laundering compliance to the Central Bank of the Future Conference, which was hosted by the Federal Reserve Bank of San Francisco and the University of Michigan’s Center on Finance, Law & Policy. The 2020 edition of this conference explored the changing role of central banks and their potential to foster more inclusive economies in the United States and around the world. We also convened a panel conversation to highlight the potential inclusion and efficiency benefits of using federated learning for BSA/AML.

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.

This overview highlights key themes from our longer market & data science context report, including the state of machine learning adoption for credit underwriting and key issues in developing and managing machine learning models with regard to explainability and fairness.

This fourth edition of our FAQs focuses on emerging techniques to explain complex models and builds on prior FAQs that covered the use of AI in financial services and the importance of model transparency and explainability in the context of machine learning credit underwriting models.

FinRegLab worked with a team of researchers from the Stanford Graduate School of Business to evaluate the explainability and fairness of machine learning for credit underwriting. We focused 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 helps 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.

This third edition of our FAQs considers the technological, market, and policy implications of using federated machine learning to improve risk identification across anti-financial crime disciplines, including in customer onboarding where it may facilitate more accurate and inclusive customer due diligence.

FinRegLab is working with teams at The Ohio State University and Charles River Associates to evaluate new workout structures and data and technology applications for consumers who are struggling with unsecured credit. The project will use data from pilots organized by the National Foundation for Credit Counseling and other sources.

FinRegLab partnered with the Financial Health Network, Flourish Ventures, and Mitchell Sandler to provide a working paper summarizing the current U.S. federal legal framework governing consumer financial data with the goal of laying a foundation for future policy analyses and discussions.

This update catalogues recent initiatives involving the use of non-traditional credit data, including cash-flow information. It considers how dramatic shifts in economic conditions due to the Covid-19 pandemic and mass movements for racial justice have increased incentives to adopt new data sources and models, but also created new market and policy challenges.

This second edition of our FAQs considers more deeply issues and debates about model transparency, explainability, and the implications of using machine learning for credit underwriting.