This project showed that African Americans in Virginia are far more likely than Whites to be denied a loan to purchase a home. As a result, Black homeownership rates are far below White homeownership rates. Those disparities help explain the wealth gap between Blacks and Whites because owning a home is key to building wealth. The project used extensive data analysis and expert interviews to document the disparities while also putting a human face to the issue.
The project triggered a community discussion in Richmond and other areas of Virginia about how to address disparities in home loans and homeownership. The National Association of Real Estate Brokers, Virginia Bankers Association, Richmond Metropolitan Habitat for Humanity and other organizations circulated the story among their members. Virginia REALTORS, an association of 35,000 real estate agents across the state, created a Presidential Advisory Group dedicated to expanding opportunities for diversity and inclusion. In its materials, the group highlighted the project as a driving force. The article also was a focus of the 2020 Virginia Governor’s Housing Conference, which included a session titled, “Making it Right: How Housers Can Address Racial Inequalities and Close the Homeownership Gap.”
I downloaded data collected by the U.S. government under the Home Mortgage Disclosure Act. The HMDA data for Virginia for 2019 contained 505,456 records — one for each loan application handled by each financial institution that year. Using Microsoft Access, I joined the data file with other tables to translate codes and include the names of lending institutions.
I then filtered the data for home-purchase loans that had been approved or denied, giving me a final data set of 127,860 records. I ran a succession of group-by and crosstab queries in Access to calculate how often applicants from each racial or ethnic group were denied home loans, and why. I calculated the statistics statewide, for each metro area and by locality (city and county). I also computed the denial rates by ethnicity-race for applicants with similar incomes ($40,000-$49,999, $50,000-$59,999, $60,000-$69,999 and so forth).
For comparison, I conducted similar analyses of HMDA data for Virginia as far back as 2007 (available from the Consumer Financial Protection Bureau).
I performed most of the analysis with Microsoft Access and refined the results with Microsoft Excel.
I also downloaded homeownership data from the Census Bureau’s data portal, using the American Community Survey estimates. Using Excel, I computed homeownership rates overall and by ethnicity and race for the nation, for Virginia, for each Virginia metro area and for each city and county in the state. I worked initially with the most recent data (2018) and then compared the results with previous years.
I created the online graphics with Datawrapper, the data visualization tool preferred by Virginia Mercury. For the tooltips for the maps, this required extensively manipulating the HTML coding.
Finally, for transparency and trust, I publicly shared all of the data on the project’s Google Drive.
What was the hardest part of this project?
The hardest part of the project was finding people to humanize the data — to put a human face to statistics. This was complicated by the Covid-19 pandemic, which shut down in-person services at organizations that help prospective homeowners and thus undercut what would have been my strategy for finding people to interview. However, working through social media and with experts I had contacted online and by telephone, I managed to connect with people who represented the story’s key data points (i.e., African Americans who had been denied home loans).
The data analysis was critical to forging those personal connections. The national and local experts I had contacted trusted me — and helped me find “real people” sources — because they knew I had done my homework. I had crunched more than 10 years of HMDA data, sifting through as many as a half-million records for each year. Moreover, I followed a time-tested methodology used by other journalists — most notably Bill Dedman, who won a Pulitzer in 1989 for his analysis of such data. The data analysis enabled me to approach experts with information they were eager to know.
A crucial component of this project was to look at the big picture — not just at racial discrimination in home loans but also at American history, from slavery to redlining. I didn’t shy away from discussing racist mortgage brokers, but I explained that institutional racism and other factors also contribute to the higher loan denial rates for Black applicants. Moreover, I connected homeownership patterns to the wealth gap and described the vicious circle underpinned by data: Because they are less likely to own a home, African Americans have less wealth; and because they have less wealth, African Americans are less likely to own a home.
What can others learn from this project?
My project’s biggest lesson for other journalists is the value of open data — especially massive sets of microdata (like HMDA) that are updated regularly by government agencies. I focused on Virginia, of course, because I was writing for a Virginia-focused news outlet. But this story could be replicated in any state.
It was crucial to jump on the issue quickly. The Federal Financial Institutions Examination Council released the 2019 HMDA data on June 24, 2020; I published my article in less than a month.
The project involved a lot of numbers, and that can intimidate readers. But I was judicious about which numbers to weave into the text of the story, I offloaded most statistics to data visualizations and tables, and I used narratives, quotes and telling details (“She lives in the three-bedroom, two-bath home — ‘white with red shutters’ — with her special-needs son”) to keep readers engaged.
A final lesson for journalists would be to examine solutions as well as the problems highlighted in the story. I devoted a section of my article to strategies to boost Black homeownership. After the project was published, those strategies helped foster discussion among government officials, business leaders and fair-housing advocates.
bit.ly/va-hmda-methodology — “Nerd box” explaining where I got the data and how I analyzed it.
bit.ly/hmda19_va — Google Sheet with summary statistics from my analysis of the HMDA data.
bit.ly/hmda19_va_db — Compressed file containing a Microsoft Access database, which has the tables and key queries from my HMDA analysis.
bit.ly/homeownership_va — Google Sheet with national, state, metro and locality data on homeownership by race.
www.datawrapper.de/_/zZ1aU/ — Bar chart showing mortgage loan denial rates by race for the nation, state and each Virginia metro area.
www.datawrapper.de/_/18V0n/ — Map showing homeownership rates for Blacks and Whites in each Virginia city and county.