What HMDA data can — and can't — tell you about an individual loan officer's fair-lending record

Our explainer on the CFPB complaint database covers a tool that's company-level only. The Home Mortgage Disclosure Act's data is different in one specific, genuinely important way — and still limited in another.

What HMDA actually requires

The Home Mortgage Disclosure Act, implemented through the CFPB's Regulation C, requires most mortgage lenders above certain size thresholds to report detailed data on nearly every mortgage application and loan they handle — property location, loan amount, applicant demographic information (where provided), action taken (approved, denied, withdrawn), and dozens of other data points — compiled annually into a public Loan/Application Register through the FFIEC and CFPB.

The genuine change in 2018: an individual NMLS ID, at the loan level

Starting with data collected in 2018, HMDA reporting requirements were expanded to include, for the first time, the NMLSR ID of the specific mortgage loan originator on each reported loan — not just the company. This is a real, meaningful difference from the CFPB Consumer Complaint Database (see our companion explainer), which has no individual-originator field at all. In principle, this means an individual originator's lending patterns — approval rates, pricing, which neighborhoods or applicant groups they originate for — can be reconstructed from the public HMDA data, filtered down to their specific NMLS ID.

Why "in principle" is doing real work in that sentence

Actually doing this isn't remotely like a NMLS Consumer Access search. HMDA's public data is released as large, loan-level files covering every reporting institution nationwide, accessed and filtered through the CFPB's own HMDA Platform tools — genuinely public and free, but requiring real data work, not a name-search box. Certain fields in the public version of the data are also modified or excluded specifically to protect applicant privacy, which limits exactly how precisely a given loan can be tied back to specific circumstances without careful, deliberate analysis.

A raw pattern still isn't proof of discrimination

Even with individual-level data in hand, HMDA's own data doesn't include every legitimate factor a lender is allowed to weigh in an underwriting or pricing decision. A regulator or researcher looking at a disparity in an individual originator's HMDA-reported outcomes has to account for those legitimate factors before concluding anything about discrimination; a raw statistical gap is a flag worth investigating, not a finding on its own. The CFPB's own review of HMDA data separately flagged a data-quality problem directly tied to individual originators — thousands of loan officers nationwide whose files showed applicant demographic information marked "not provided" at unusually high rates, a red flag about data completeness rather than about lending outcomes themselves.

What this means for checking a specific originator

HMDA data is a real, public, individual-level resource that didn't used to exist before 2018 — but it's a research tool for someone willing to do genuine data analysis, not a consumer-facing lookup the way NMLS Consumer Access is. Our own standard reflects this distinction directly (see our standard, point 3): a company's aggregate complaint pattern is reported only as background, and a specific, named finding against an individual originator is checked and cited on its own evidentiary basis — a raw HMDA statistic alone, without that further analysis, doesn't clear that bar either.