Analyzing racial disparities in the United States homeownership: A socio-demographic study using machine learning

Abstract

While increasing homeownership has been a focal point for policymakers in the United States, the distribution of access to homeownership opportunities across all families within the country has not been equitable. The complex relation between social factors and economic conditions has significantly shaped the journey toward homeownership. Of particular concern is the persistent disparity in homeownership rates among Black households in the United States. Effectively addressing these disparities within urban communities necessitates comprehensive policy interventions and social initiatives. However, the success of such endeavors hinges upon a comprehensive grasp of the underlying causes of these inequalities. Drawing upon machine learning techniques, our analysis of national American Housing Survey data reveals clear evidence for the multifaceted and socio-demographic nature of racial disparities in homeownership within the United States. Our findings provide evidence for two previously obscured patterns. First, we observe that race-related risk factors, separate from household characteristics, are heterogeneous, not uniformly affecting all households. Specifically, our analysis highlights: (1) there is a geographical variations in how race-related risk factors contribute to racial inequality, and (2) households with lower educational attainment are potentially at increased risk of discrimination. Second, our findings underscore the potential of policies and social programs aimed at enhancing educational attainment to bridge the racial gap through two mechanisms: (i) mitigating vulnerability by alleviating race-related risk factors, and (ii) suppressing the compounding effects of racial disparities.