Disparity or Not: An Exploration of Mortgage-Lending Practices and Solutions For Minorities

Despite the implementation of various housing policies such as the Fair Housing Act (FHA), Home Mortgage Disclosure Act (HMDA), and Equal Credit Opportunity Act, disturbing reports of mortgage lending disparities among minorities persist, particularly African-American communities. This study investigates the prevalence of such disparities in recent mortgage lending practices through quantitative research. Initially hypothesizing no disparity for minorities, sophisticated statistical methods like logistic regression to analyze a vast dataset comprising 14 million nationwide mortgage applications were utilized. Many believe racial disparities in mortgage-lending are due to differences in income level and loan amount. To examine whether there is merit in this belief, this study employs a multivariate method by taking 10 related economic factors into consideration which exposes a higher declination rate for African-American borrowers even after accounting for these factors. Consequently, the study substantiates the existence of significant systemic racial disparities within mortgage lending. In a step towards addressing these disparities, an application was developed that integrates insights from the abovementioned advanced machine learning model and the research of Harvard Professor Kosuke Imai using fully Bayesian Improved Surname Geocoding (fBISG). The overarching goal of this application is to empower borrowers, particularly those from African-American communities, by providing assistance in navigating mortgage expectations and enhancing their probability of success in securing loans.