Re-Examining the Neighborhood Distribution of Higher Priced Mortgage Lending: Global Versus Local Methods

Yonghua Zou
DOI: https://doi.org/10.1111/grow.12121
2015-01-01
Growth and Change
Abstract:This paper demonstrates the application of the local geographically weighted regression (GWR) method to mortgage studies, using the Philadelphia metropolitan statistical area (MSA) as a case study. Previous related studies widely employed the global, ordinary least square (OLS) method to examine the spatial distribution of subprime/higher priced mortgages. The OLS method, however, masks the spatial variations in mortgage distributions. The innovative GWR method not only provides a significant improvement in model fitness but also reveals that the statistical relationships between neighborhood characteristics and the prevalence of higher priced mortgage shares are spatially varied. In addition, the GWR method can bring forward potential implications for place-based policy making. Improving upon previous methodologies in mortgage studies, this paper shows that the GWR method can advance our understanding of how neighborhood environments are associated with mortgage lending patterns.
What problem does this paper attempt to address?