Abstract This study investigates racial disparities in the U.S. mortgage application process by examining outcomes beyond traditional credit denial. Using over 37 million mortgage applications submitted between 2018 and 2024, we analyze four sequential termination points where borrowers may exit the mortgage process: incomplete applications, applicant withdrawals, lender denials, and borrower rejection of approved loans. Leveraging Home Mortgage Disclosure Act (HMDA) data, we employ logistic regression models toassess how race—particularly Black, Hispanic, and “race not provided” applicants—predicts the likelihood of each outcome, while controlling for applicant, loan, lender, and neighborhood characteristics.We find that Black borrowers were 41% more likely than White borrowers to have incomplete applications, 13% more likely to withdraw applications, 78% more likely to be denied, and 25% more likely to reject an approved loan. Hispanic applicants and those who did not disclose race also faced elevated risks. These disparities remained robust across COVID and post-COVID periods.Importantly, our analysis highlights that FHA loans are associated with reduced odds of incomplete applications among Black borrowers, suggesting potential policy pathways to lower administrative burdens. Furthermore, Black applicants were more likely to withdraw applications with small lenders, indicating the potential role of lender capacity and customer service in shaping borrower outcomes.Our findings suggest that racial inequities in mortgage access are not limited to final credit decisions but are embedded throughout the application journey. Cumulatively, Black borrowers had 61% higher odds of failing to secure financing, reinforcing systemic barriers to homeownership and wealth accumulation. This study expands the discourse on mortgage discrimination by examining early and post-decision termination points and offers new directions for research, policy, and practice—especially regarding mortgage product design and lender practices.
Patrick Meehan (Thu,) studied this question.