Video: The HMDA Longitudinal Dataset: Tracking Mortgage Lending 1981-2021

Online Event Archive Recorded: October 2, 2024

NCRC Principal Researcher Bruce Mitchell presented the HMDA Longitudinal Data Set (HLD), a comprehensive mortgage lending dataset from 1981 to 2021. Funded by the National Institutes of Health, HLD standardizes mortgage data to 2010 census boundaries, enabling consistent comparisons over four decades. The dataset includes mortgage originations by race, ethnicity, gender, and income, and can be used for research in housing finance, economic development, public health, and policy advocacy. The mapping application visualizes lending patterns, revealing historical redlining and its persistent impact on minority communities. The data is available for download from ICPSR.

 

Speaker:

Bruce Mitchell, PhD, Principal Researcher, NCRC

Transcript:

NCRC video transcripts are produced by a third-party transcription service and may contain errors. They are lightly edited for style and clarity.

Mitchell 0:05
Welcome to this presentation on the HMDA Longitudinal Data Set. The HMDA Longitudinal Data Set is produced by the National Community Reinvestment Coalition, along with our partners at the University of Michigan and the University of Richmond, and partial funding for this research was supported by the National Institutes of Aging of the National Institutes of Health under award number Ro1AG08041 and the content is solely the responsibility the authors and doesn’t necessarily represent the official views of the National Institute of Health. My name is Bruce Mitchell. I’m principal researcher with NCRC, and I want to talk to you today about the HMDA Longitudinal Data Set, a new data set that allows you to track mortgage lending from 1981 to 2021.

HMDA, the home mortgage disclosure act. This is the primary form of capturing mortgage data in the United States. In fact, it captures about 80% of the mortgage market. So it’s very comprehensive. So mortgage lending data has been collected since 1981 and the HMDA mortgage disclosure Act has made this data publicly available. However, because during different eras, there have been many changes to the way that HMDA has been collected and the way that it’s been presented, it’s been impossible to look back over all 40 years of the data consistently. Now the HMDA Longitudinal Data set, or HLD, makes it possible. How does it do this?

Well, first off, it provides the data in a consistent way. Consistent boundaries have been a primary problem, just like HMDA, all data is presented at the census tract level, but unlike the original HMDA data, the boundaries have been consistently adjusted to 2010, census tract boundaries make it easy to compare with other data. You see census tract boundaries often change during the decennial census periods. This is made putting a consistent set of data together impossible up until this point. Now, you can take the HLD and match the data to other data sources, like the LTDB, the longitudinal track database, which provides data on census, the census data over the past 50 years. Additionally, the HLD provides consistent categories, categories for the total number of mortgage originations and amounts, the number of private mortgage originations made by banks or mortgage companies, and then government backed mortgages from FHA and VA. Additionally, the data is comprehensive. This is a national level data set that covers urban, suburban and rural areas across the country. To the right, you can see our mapping application, which presents this data in an easy format, from 1981 up until 2021 and you can look at the national level and individual Metro level and see changes in the amount of lending that’s occurred across the country. Additionally, it’s a comprehensive set of data that provides mortgage data on home purchases, home improvements and refinancing. Finally, it provides data on mortgage originations by race, ethnicity, gender and income level at the census track level.

So what can you do with this data set? Again, you see our mapping application over to the right. This is showing the City of Milwaukee. We’ve zoomed in so that you can see the different layers that are presented or overlays. Here there’s an overlay of the HOLC, homeowners Loan Corporation, city survey maps for Milwaukee, showing their fortune tiered system, the red areas being areas that were graded with the lowest grade of being hazardous all the way up to the green areas that were graded best by examiners 80 to 90 years ago. the red hazardous areas are commonly referred to as red lined areas. These are areas in which lending was greatly discouraged or unavailable to people living in those neighborhoods.

Additionally, what you can see here are the bubbles. The round bubbles indicate the relative number of mortgages that are made at the census tract level within the city. Additionally, the dark purple and light purple indicate the proportion of minorities living in these areas of Milwaukee. The slider on the right hand side enables you to look at the data year by year from 1981 to 2021. Using this, you can see patterns of disinvestment neighborhoods relating to race, ethnicity and income. This mapping interface is very easy to use, but the data is also available for download, where you can look at the individual level data, and that’s available on ICPSR. If you need information on that, you can write to us or go to the ICPSR website and access this data.

So along with the data set, we released a report, “Decades of Disinvestment,” in which we did an analysis of mortgage lending in the HOLC-graded neighborhoods. What you can see over to the right is a histogram displaying decade by decade the lending from the 1980s up through the 2010s on a national level, to ungraded areas in the gray, and then each subsequent HOLC grading level going from green to blue to yellow to the redlined hazardous areas. And you can see this sort of stair step, or monotonic relationship and lending that’s consistent over the decades in which mortgage lending has decreased with each lower HOLC grade. Relationships are consistent even as the decades have passed, and show the persistent problem of redlining even 90 years after the HOLC maps were made. In fact, on average 3034 fewer loans were made in redlined neighborhoods over the past 40 years than in the best-graded neighborhoods.

Here you see the overlay for Milwaukee for 2021 and it shows how the areas, neighborhoods in which people of color live have greatly expanded throughout the throughout the metro area. So this allows you to look at the percentage of people of color living in different areas and see what the amount of mortgage lending occurring there is. This internet active map allows you just look at every city and examine lending for the past 40 years.

So who can use the HLD? Well, first off researchers, researchers in housing finance economic development, can also be used in the fields of epidemiology and public health, whereas there’s increasing interest in the connection between mortgage disinvestment and worse public health so that you can better understand present day health inequities. Advocates and community leaders can use this to precisely connect present day conditions of disinvestment and past structural discrimination. And policy makers. This highlights the importance of programs and policies that NCRC is here to defend like the Community Reinvestment Act or HUD’s Affirmatively Furthering Fair Housing rule. The data set has been created by NCRC, along with our partners. Jason Richardson is a senior director of research. I’m Bruce Mitchell, principal researcher, Jad Edlebi, the data engineer, was largely responsible for doing the coding and creating the maps that you see presented here. And Joseph Dean is our economist. Thank you very much. And if you have any questions, please email, write to us, and we’ll be glad to discuss any of your questions. Thank you.

Transcribed by https://otter.ai

Leave a Comment

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.

Scroll to Top