Analyze a file →
FRC Research · Original Research · June 2026

The market that doesn't react.

Data: CFPB HMDA 2018–2025 · the 100 largest lenders · 77 lender-year transitions · 3,075,654 FHA applications
Method: OLS — year-over-year market share change vs. prior-year denial rate
Computation archived and reproducible

In a functioning consumer market, the single biggest cost drives the flow. Airlines with high prices lose passengers. Restaurants with bad ratings lose tables. For a mortgage applicant, the single biggest cost is the probability of being denied — it determines whether the transaction happens at all, and our research shows lender identity explains 63% of denial variance.

So we asked the obvious question nobody had tested: does FHA application flow respond to lender denial behavior?

+0.0047
market-share points gained per point of denial rate.
Statistically zero (p = 0.883, 77 transitions, 8 years).
The flow does not react.

What zero looks like in practice

Group2025 denial behaviorShare 2018Share 2025
Best two lenders (CrossCountry, Guild)6.4% / 7.1%9.4%13.0%
Worst two lenders (NewRez, Wells Fargo)53.3% / 48.6%5.4%6.4%

The two most approval-friendly lenders in America carried a 30–40 point denial advantage for eight consecutive years — and gained 3.7 points of share. The two strictest lenders, one of which denies more than half of its FHA applicants, lost nothing. A lender can deny one in two borrowers and keep its customers coming.

The implication: borrowers are not choosing lenders. They are being assigned — by advertising budgets, broker relationships, servicer transfers, and geographic accident. The assignment process is statistically blind to the one variable that most determines whether they get a home.

Why this matters

This is the mechanism beneath everything else we've measured. It explains why the 9+2 hierarchy never converges, why Shadow Approvals persist year after year, and why a denial says more about routing than about the borrower. Markets self-correct when information moves buyers. This one doesn't — the information exists in public federal data, but no actor in the transaction has the incentive or capacity to bring it into the borrower's decision. So the inefficiency freezes in place, for eight years and counting.

Limitations — read these

This is an aggregate finding, not a borrower-level one. Part of the zero could reflect capacity or channel constraints at low-denial lenders rather than pure information failure, and applicant pool composition is not held constant across lenders. The elasticity describes the market's behavior as a system; it does not prove any individual chose wrongly. A loan-level conditional test (CFPB raw HMDA) is the next step and is planned.

The market won't route you to the right lender.
So route yourself.
Get a file analysis — free →
Source: CFPB Home Mortgage Disclosure Act public datasets, 2018–2025, 11 major FHA lenders. Regression: Δ market share (pp) ~ prior-year denial rate (pp); slope +0.0047, r = 0.017, p = 0.883, n = 77 lender-year transitions; lender-years under 1,000 applications excluded. Data archive: _data/flow_elasticity.json. Statistical research on aggregate market behavior; not financial advice.

Citation: FRC Research (2026). "The Market That Doesn't React." financeratecalc.com/market-that-doesnt-react.html
Related research: Market Blindness Index →  |  Shadow Approvals →  |  71% Not Your Fault →  |  The 9+2 Structure →
⚡ Analyze a file