Working Paper
Precision Without Labels: Detecting Cross-Applicants in Mortgage Data Using Unsupervised Learning
WP 25-25 – We develop an algorithm to detect loan applicants who submit multiple applications in a loan-level dataset without personal identifiers. Our method detects applicants that submit multiple mortgage applications with 92.3 percent precision.
Featured Work
Working Paper
Recurring-Payment Sensitivity in Household Borrowing
WP 25-22 – This paper provides evidence of payment sensitivity in household borrowing decisions: Mortgage borrowers respond to the size of the recurring payment as opposed to discounted total loan costs when choosing between loan options.
Article
Data in Focus: Mortgage Fairness Explorer
Economic Insights — Fairness is something we all aspire to but struggle to define. When is something fair? This isn’t just an esoteric philosophical question. In the field of mortgage financing, it’s a matter of law.
Conference Summary
Workshop on Changing Demographics and Housing Demand
On October 24 and 25, 2024, the Federal Reserve Bank of Philadelphia’s Consumer Finance Institute (CFI) and Fannie Mae’s Economic and Strategic Research Group (ESR) cohosted a hybrid conference on the topic of changing U.S. population demographics and their influence on housing demand and housing finance.
Featured Data
Advancing Fairness in Lending Through Machine Learning
Showcasing the work of Philadelphia Fed researchers, this interactive data visualization explores an approach to credit lending using machine learning and fairness goals that may help address current disparities in credit access.
Updated: 05 Feb ’24