Citi among banks banks omitting data about race from home loans—depriving researchers of both positive and negative trends

The significance of the data, which is collected pursuant to the Home Mortgage Disclosure Act, is further reflected in the fact it has, in the past year alone, been cited by the Consumer Financial Protection Bureau, the Federal Financial Institutions Examinations Council, the Office of the Comptroller of the Currency, among others.

The problem is that more than 12% of borrowers don’t volunteer the information that is requested by the law, and that 90% of loans sold to third parties are stripped of the data that is acquired. That’s according to the National Community Reinvestment Coalition, a non-profit that studies issues surrounding racial and social economics.

“The impact is profound,” according to an NCRC report published today, “as these gaps hinder our ability to understand who is receiving loans and under what terms, which is vital for assessing fairness and inclusivity.”

To help combat the problem the NCRC today pledged to never again use any data that doesn’t include demographics on race. “Beginning with this report, NCRC is eliminating records without demographic data from our calculations of the percent of loans made to specific races,” the researchers wrote.

The NCRC and others say the missing data is largely due to loopholes in the HMDA. Passed in 1975 to help ensure more equitable distribution of loans, the HMDA rule requires that in-person and phone applicants provide demographic data. But online applicants can opt out.

“In the past, it was assumed that those choosing not to select a race were more likely white,” said Richardson. “However, in this report, we demonstrate that loans without data likely reflect racial diversity more accurately than previously thought. Thus, the correct approach is to exclude these loans.”

Exacerbating spotty data, third party loan purchasers don’t need to track demographic information at all. Seven of the top 10 loan-purchasing institutions from last year used a loophole that allows them to erase borrower demographic data on the mortgages they bought, according to report co-author and NCRC senior researcher, Jason Richardson, in a conversation with Fortune.

“A few years ago, it was rare for lenders to buy loans and strip demographic data, but Citibank pioneered this practice,” said Richardson. “Now, many lenders who purchase loans use this loophole.” Citi did not respond to a request for comment.

Certainly, many a prejudiced lender can hide behind this data black hole, but some more positive trends are also obscured.

The NCRC report shows “in what might be a sign of a historic point” that Hispanic lending for home loans—16.5% of all home purchases last year—was nearly identical to their overall share of the U.S. adult population. Black borrowers also saw their lending rates improve, though nowhere near to their overall share of the population.

Unfortunately, these seemingly positive trends are difficult to confirm because of the incomplete data.

 “We urgently need more comprehensive data on small business and community investment to effectively craft policies that mitigate the harsh realities of redlining,” according to the report.

Of course, any increase in data collection about borrowers comes with increased risk of invasion of privacy. Though the CFPB says there’s “low, if any, privacy risk” in the HMDA, a 2017  report  by economist Anthony Yezer expressed concerns such data collection could lead to widespread violations of privacy.

That’s of little concern to the NCRC. “The extensive benefits of detailed data collection, encompassing income, race, sexual orientation and gender identity, decisively outweigh any concerns over burden or privacy,” the authors wrote. “It’s imperative that efforts to curtail this essential data collection be recognized as not just misguided but as detrimental to the health and well-being of our communities.”