How Will Changes to the Commerce Department Disclosure Rules Impact Housing?

Following my post on DAO 216-26 (Commerce's new disclosure avoidance rules), I was asked how these new rules might impact the residential housing market. This is a great question which I thought worthy of its own post as the chain from "statistical methodology" to "your rent and your underwriting" is shorter than it looks!

HUD's Area Median Income and Fair Market Rent calculations – which set eligibility and rent ceilings for Housing Choice Vouchers, public housing, LIHTC, and HOME – are built directly from American Community Survey data. Treasury also uses AMI to set income limits and rents for LIHTC and tax-exempt housing bond properties. Therefore, it stands to reason that when ACS estimate gets “coarser”, so does everything downstream of it.

Here are a few concrete channels:

1.     Smaller metro areas and non-metro counties: already the places falling back to multi-year averages because their ACS samples are thin, are exactly where coarsening bites the hardest. I expect more volatility and more workaround methodologies in published income limits for these markets;

2.     Small Area Fair Market Rents: calculated at the ZIP code level (specifically to open high-opportunity neighborhoods to voucher holders), need fine-grained geography to work. That's precisely what coarsening degrades. It is quite possible that a deconcentration tool that HUD has invested years in could quietly lose precision;

3.     Site selection, LIHTC Qualified Census Tract designations, and pro-forma underwriting: these all lean on tract- and block-group-level ACS data (vacancy, tenure, income distribution, etc.) Coarser estimates mean wider error bands on core underwriting inputs which can push more developers, especially smaller and non-profit ones who can't afford CoStar or RealPage, toward less reliable substitutes; &

4.     Fair housing enforcement and Affirmatively Furthering Fair Housing analyses: these depend on demographic-by-housing cross-tabs using small geographies. Coarsening or suppression here weakens the evidentiary base for disparate-impact claims just as the data gets harder to use for that purpose.

The framing that matters for anyone in housing policy or real estate is that this order does not reduce data distortion, it relocates it. Noise infusion spreads unbiased random error across a dataset. Coarsening and suppression concentrate accuracy loss on small geographies and small sub-populations – which happen to be exactly where housing policy (voucher deconcentration, LIHTC siting, fair housing enforcement) is trying to act with the most precision.

As I see it, the tool built to protect privacy in small populations is being replaced with one that degrades data quality most in those same small populations.