Methodology
Corrections log

When a figure changes, it says so here.

Derived statistics can be wrong in two ways: the arithmetic can be wrong, or the universe can be wrong — the second is harder to notice and more consequential. This page records every correction to published figures, what caused it, and how much it moved. Nothing is quietly edited.

26 July 2026 · Universe correction

Reverse mortgages (HECM) removed from the FHA universe

What prompted it

An AI assistant, asked to review the filter logic in the abstract, raised two categories of risk that had not been checked: whether business-purpose records and whether FHA-insured reverse mortgages (HECM) were being included in a universe described as "mortgage applications." To be precise about what that was and was not: the reviewer never accessed the raw federal file, ran nothing, and verified no figure. It identified a blind spot in the stated method. The measurement against the raw file, and everything that followed from it, was done here — which means this correction is self-found and self-published, prompted by an outside observation about logic rather than confirmed by an outside check of the data.

That distinction matters enough to state twice: nothing on this page constitutes independent verification. What it records is that a gap was pointed out, measured, and corrected in public.

What was found

Business-purpose records were negligible: 3,493 applications, 0.29% of the universe, moving the national rate by 0.006 points. They remain included and the effect is disclosed here.

Reverse mortgages were not negligible. HECM is an FHA-insured product with a fundamentally different underwriting process — no income qualification in the conventional sense, age-based eligibility, different failure modes. The universe contained 29,691 such applications, 2.4% of the total, and they were denied at a materially lower rate (12.3%) than the rest, pulling every aggregate downward. Including them under a headline described as mortgage denial was a definitional error, not an arithmetic one.

What changed

FigurePublished (to 25 Jul)Corrected
Decisioned universe1,217,2971,187,606
National FHA denial rate21.7%22.1%
Top-100 lender spread1.8% – 78.7%1.8% – 78.7% (unchanged)
"Incomplete application" median share1.2%1.8%
Carrington incomplete share73.5%75.2%
Metro range (1,000+ apps)8.9% – 31.8%9.0% – 32.9%
Cleveland inside-metro spread73.8 points73.7 points
El Paso small-loan penalty3.61×3.94×
Idaho small-loan penalty3.19×4.45×
State range14.8% – 29.0%15.0% – 31.4%
Applicant mix explains2.9×2.7×

The central finding — a 44× spread across the 100 largest FHA lenders, which survives standardization for borrower profile — is unchanged. Every other published figure moved by between 0.1 and 1.3 points, in the direction of showing more denial, not less.

A second error found in the same pass

While rebuilding, an error in the peer-adjustment code was found and fixed: observed denials were counted over all of a lender's applications while expected denials were computed only over applications with a complete profile, inflating ratios for lenders with sparse fields. Both are now computed over the same subset, and institutions whose profile coverage falls below 50% are excluded from the adjusted measure rather than shown with an unstable ratio. This error had not appeared in any published figure — it was caught in the rebuild — but it is recorded here because a correction log that only lists errors someone else found is not a correction log.

What is now in place to prevent recurrence

The rebuild script validates its own column selection (it verifies, for example, that values read as metro codes are five digits before using them), excludes purchased loans, withdrawals, incomplete closures and the preapproval track explicitly rather than by numeric range, and prints the universe composition on every run. A pre-publication audit script scans the site for figures that no longer match the current data.

Deposited versions

Cite 10.5281/zenodo.21575105 for the dataset as such — it always resolves to the current version. Cite a specific version DOI below when a figure needs to be pinned to the exact form in which it was published.

VersionDOIStatus
2.0.0 · 26 Jul 202610.5281/zenodo.21590145Current — HECM excluded, adjustment coverage fixed
1.0.0 · 25 Jul 202610.5281/zenodo.21575106Superseded — remains citable and visible

The earlier version was not withdrawn. Anyone who cited it can see precisely which figures changed and by how much, which is the point of versioned deposit rather than silent replacement.

Standing practice

Corrections are published rather than applied silently, with the previous figure shown alongside the new one. Where a correction affects a dataset deposited under a DOI, a new version is deposited rather than the original being replaced — the superseded version stays citable and visible. Where it affects a working paper, a revision note is added rather than the figure being changed in place.

If you find an error in anything published here, please write: [email protected]. The methodology is published precisely so that this is possible, and a correction found by a reader is worth more than one we find ourselves.

What would count as verification — and has not happened

Everything on this site is computed and published by a single independent researcher. It has not been peer reviewed, replicated, or checked by any party without a stake in the result. A dataset correcting itself is evidence about how a source behaves under scrutiny; it is not evidence that a figure is right.

The check that would settle it is specific and cheap: pull the 2025 HMDA loan/application register, filter loan_type = 2, keep action_taken in {1,2,3}, exclude reverse mortgages, count action 3, and see whether the national figure lands on 22.1% and the top-100 spread on 1.8%–78.7%. If your run disagrees, that is a finding and it will be published here with your name on it if you want it.

Until someone unaffiliated does that and reports back, the accurate way to cite anything here is "FinanceRateCalc reports X" rather than as an established figure. That framing is not modesty; it is the correct epistemic status of a single self-published source, and readers who apply it are reading correctly.

Nothing here has been independently checked. How to check it →