A new dataset will give researchers an unprecedented look at how long-term disinvestment and redlining affects public health.
Funded by a grant from the National Institutes of Health (NIH), the National Community Reinvestment Coalition (NCRC) and Institute for Social Research (ISR) have collaborated to create the Home Mortgage Disclosure Act Longitudinal Dataset (HLD). Covering activity since 1981, the HLD combines more than 40 years of data from the only available set of home mortgage information.
The HLD effectively bridges the gap between the different eras of HMDA data collection and clarifies it for researchers, using the same census tract boundaries to organize the information. This accessible format should open new doors for investigating the historical effects of redlining and disinvestment through new, geographic alignment across multiple decades of mortgage reporting.
“The HMDA longitudinal dataset allows for new opportunities for analysis of neighborhood level disinvestment for the research community including of public health, redlining, gentrification, and equity in general,” said Bruce C. Mitchell, who leads the NCRC team. “The HLD combined with NCRC and the University of Michigan’s prior work on a Historic Redlining Score is a useful tool for assessing patterns of structural racism, like the association of redlining and prolonged neighborhood disinvestment.”
The finished product is something entirely new. Previously, researchers had been unable to present this data comprehensively. Home Mortgage Disclosure Act (HMDA) data was collected in four distinct phases, which affected the type and organization of the data and also created a degree of unresolved errors and omissions from the reporting institutions and federal regulators. The resulting datasets involved production and integration of datasets with national coverage of mortgage lending. This will link with other datasets encompassing historic redlining, locations of industrial sites, and public health.
Read a first major report on redlining utilizing the HLD here.
“The temporally and geographically normalized HLD will facilitate studies of the drivers and consequences of modern lending discrimination in the US. The HLD is easy to link to any other data resource with 2010 census tract identifiers, including administrative data and longitudinal cohort studies,” said Helen Meier, an assistant research scientist at ISR.
The NCRC research team, utilizing years of experience using HMDA data, was able to cross-reference existing data through the decades to establish common indicators of the number and amounts of mortgage loan originations, the type of lending activity, and the income and race of borrowers. This involved the development of SQL scripts to organize the data accurately and in an understandable way.
Researchers then applied Brown University’s Longitudinal Tract Database (LTDB) in order to establish common geographic boundaries for the data, representing the same neighborhoods consistently year over year. The HLD can be matched with a broad range of US Census data available through the LTDB, facilitating comparisons of the income, race, housing status, and other data available to the research community. This, in turn, should yield new research on crucially important subjects.
“Already, NCRC is working with U-M researchers to develop a Minority Disparity Index of mortgage lending, and preparing a report on disparities in mortgage lending to historically redlined neighborhoods,” Mitchell said.
The HMDA Longitudinal Dataset will be hosted by the National Neighborhood Dataset (NaNDA), the preeminent source of neighborhood-level data available to researchers. NaNDA is an archive at ICPSR, part of the University of Michigan Institute for Social Research.
This mapping tool seems very compelling however Im having trouble understanding it, and cant seem to find a map key that explains all the data one sees when clicking on a census tract. I’d love to better understand what this tool is telling me.