One of the top political themes in 2026 politics is “affordability”. Most people would argue that affordable housing should be at the top of the “affordability” crisis. The single most important database when it comes to the financing of housing in the United States is extracted from the Home Mortgage Disclosure Act, or “HMDA”. The HMDA requires almost all mortgage lenders to report detailed activity pertaining to the financing of 1-4 family and multifamily mortgage lending annually. The purpose of the HMDA is to provide regulators, community leaders, and the public, information on whether financial institutions are serving and meeting the housing needs of their communities. And one of the most important housing needs is the need for affordable housing. Indeed, within the HMDA data reported by lenders is information pertaining to the number of housing units financed as well as the number of affordable housing units financed by all multifamily mortgages reported annually. But the government agency in charge of publishing mortgage lending data, the Consumer Financial Protection Bureau, or “CFPB”, publishes the data about the financing of affordable housing in a virtually useless format. This is tragic, because the data reported by lenders contains vital information in detailed format that could be extremely useful, but which is manipulated by the Bureau in a way that makes it useless. And this is not a problem recently mandated by the CFPB under the Trump Administration. The problem existed under the Biden Administration too.
Back in 2015 the Bureau significantly revised and added numerous “data points” to the information to be reported under the HMDA. Included in the expanded information were details about the number of individual living units financed by multifamily mortgages. Data fields were added to identify the number of housing units and the number of affordable housing units (called ”multifamily affordable units”) financed by all multifamily mortgages originated during the year. This detailed information is potentially highly valuable for determining the needs of the community for affordable housing and the progress, or lack thereof, made in meeting those needs.
But a funny thing happened on the way to taking that valuable reported information and then reporting it to the public.
The Bureau decided not to publish the data in the format reported by lenders. Rather, the Bureau decided to publish the count of living units in “bins” and the number of affordable units as a percentage of the units in each multifamily structure. Rather than knowing how many living units have been financed by multifamily mortgages, all that is disclosed in the public data is a series of numbers grouped in “bins” as follows: 5-24, 25-49, 50-99, 100-149, and >149. This makes counting the actual number of units financed impossible. Making matters worse, the data reported for affordable housing units is reported as a percentage of the number of units financed by each multifamily mortgage. Since we don’t know the exact number of units financed by each multifamily mortgage, we can’t determine the number of affordable units financed.
What makes the Bureau’s approach even more perplexing (other than that it takes potentially valuable information and makes it worthless), is that the Bureau must go to great lengths to convert the meaningful data reported by lenders to meaningless information reported by the Bureau. It’s as if the Bureau is going out its way to trash valuable data. Again, this can’t be blamed on the Trump Administration because the Biden Administration tolerated this nonsense too.
It’s time for the Bureau to stop the coverup and to release the valuable HMDA data in usable format so that public officials, regulators and the public itself can determine the contributions toward affordable housing made by each lending institution as well as the aggregate progress made within each community. Ironically, it would be easier and simpler for the Bureau to do the right thing rather than go out of its way to manipulate the data and make it worthless.
