Conference Agenda

This paper studies how language barriers between lenders and borrowers translate into differences in borrower outcomes in the U.S. mortgage market. I use survey data to infer and machine learning techniques to predict borrowers' English proficiency. I document significant descriptive differences in perceptions of mortgages, application experiences, and mortgage rates between limited English proficient (LEP) and non-LEP borrowers. To measure the causal effects of language frictions, I exploit a Federal Housing Finance Agency policy that provided translated mortgage documents in Spanish to mortgage lenders. After the policy change, LEP Hispanic borrowers had a streamlined application process, contacted more lenders, understood mortgage contracts better, and enjoyed lower borrowing costs. Reducing language frictions also led to expanded access to credit, reduced loan risks, and a more competitive mortgage market for LEP borrowers. Overall, my findings highlight a cost-effective way to create a responsible inclusion of well-qualified LEP borrowers in the mortgage market.