Studying the Influence of Income Differences and Credit History on Racial Disparities in the Mortgage Market using Machine Learning

Abstract

A mortgage loan is required to purchase a house, but racial disparities make obtaining this loan harder for Black Americans. According to the National Association of Realtors, debt-to-income ratio and credit score are potential reasons for higher denial rates, so this work examines the role of income and credit on mortgage lending. Since renters make up a higher percentage of Black households compared with White and Asian households, Black Americans disproportionately make up the credit invisibles according to the CFPB. Graphical analysis of ACS Housing Data indicated a positive relationship between rental rates and denial rates, so sparse history is a potential reason for African Americans having higher denial rates. Graphing ACS Income Data using Seaborn demonstrated the effect of income differences on denial rates for African Americans, as the graph demonstrated that Low income correlated with high denial rates and vice versa. To validate data analysis, we used the Random Forest approach to model denial rates based on rental rates and income characteristics. Our computational results from the study suggest that to combat the racial disparities in the housing market, it is important to encourage more loan companies to count rent payments in credit history.