Analyze a file →
FRC Research · Cross-Product Analysis · 2025 HMDA

The Product
Trap

Data: CFPB HMDA loan-level 2025 · major deep-mapped lenders · FHA + Conventional originations
At least some lenders behave very differently across products — a single "lender quality" number can hide it

Ask "which lender is best?" and you'll get a single answer. But that question assumes lender quality is one number. It isn't. The same institution can be excellent at one loan product and terrible at another — in the same year, under the same roof.

n = 6
Correlation between a lender's FHA denial rate and its Conventional denial rate
across major deep-mapped lenders, 2025 — statistically indistinguishable from zero

The clearest case: Mr. Cooper

Mr. Cooper — same lender, same year
FHA
8.9%
denial rate
vs
Conventional
47.0%
denial rate

Mr. Cooper is one of the best FHA lenders in this dataset — 5th out of 6, with a denial rate under 9%. On Conventional loans, it's the single worst performer of the group, denying nearly half of all applicants. A borrower who heard "Mr. Cooper is great" from an FHA-focused friend would walk in with the wrong expectation entirely.

All 6 lenders — FHA vs Conventional, 2025

LenderFHA DenialConventional DenialGap
Mr. Cooper8.9%47.0%38.1pp
PennyMac24.1%11.7%−12.4pp
Rocket30.2%25.3%−4.9pp
Freedom16.1%15.9%−0.2pp
Guild8.0%7.4%−0.6pp
CrossCountry5.1%8.4%3.3pp

Only Freedom and Guild show roughly consistent behavior across products — within a point or two either way. Everyone else shows a meaningful swing. CrossCountry and Guild are the only lenders performing well on both products simultaneously.

This breaks a common assumption. Borrowers — and even loan officers — often treat "lender quality" as a single reputation. It isn't. A lender's FHA underwriting team, risk appetite, and overlay structure can be entirely disconnected from its Conventional operation. They may even run on different automated underwriting paths internally.

Why this happens

FHA and Conventional loans serve different risk profiles by design — FHA permits higher DTI (up to 57% with compensating factors) and lower credit thresholds, while Conventional loans (especially through Fannie Mae and Freddie Mac) often carry tighter automated underwriting defaults. A lender's overlay strategy on each program reflects different capital requirements, investor agreements, and risk tolerance — set independently, often by different teams.

The practical result: a lender's reputation in one product says close to nothing about the other. "Ask around" advice — common in real estate and lending circles — implicitly assumes consistency that the data does not support.

What this means for your search

If you're shopping for an FHA loan, FHA-specific denial data is what matters — not a lender's general reputation, and not their Conventional performance. The reverse is equally true. Mr. Cooper's strong FHA position offers no information about its Conventional behavior, and PennyMac's weaker FHA position doesn't predict its comparatively better Conventional outcomes.

Know which product you need before judging a lender.
Get product-specific routing for your exact profile.
Get a file analysis — free →
Data: CFPB HMDA loan-level 2025. FHA = derived_loan_product codes 2/32. Conventional = derived_loan_product codes 1/21. Denial = action_taken code 3, denominator = originated (1) + denied (3) applications. Pearson correlation calculated across the 6-lender FHA denial rate vs Conventional denial rate pairs. 6-lender sample only — broader market patterns may differ. Statistical research on aggregate institutional behavior, not individual loan predictions. Not financial advice.

Related: Decision Surfaces →  |  Lender Personality →  |  Test Your Lender →
⚡ Analyze a file
The part almost nobody publishes free

Prices and rates are widely reported. Whether a lender says yes is not. In the complete 2025 federal record, denial rates across the 100 largest FHA lenders ran from 1.8% to 78.7% — same programme, same year.

And it is not simply who applies where: standardizing on state, loan amount, income, debt-to-income and loan-to-value, applicant mix explains only a 2.7× range in expected outcomes.

Which lenders approve most → Your metro → Denied? →

CFPB HMDA 2025, computed by FinanceRateCalc. Covers the highest-volume lenders published per market, not all lenders. Historical observations, not predictions. CC BY 4.0, not independently reproduced.

The Denial Dispatch
One finding a week from the federal mortgage record.
One chart, three paragraphs, every Saturday. Measured, not assumed.
Get the Dispatch →
AI Accuracy Index The Door Effect The Denial Map Open Data About 184 Metro Gaps Evidence Navigator Hallucination Files Press Newsletter
FinanceRateCalc · Independent analysis of the complete federal HMDA record · Measured, not assumed. · No lender or AI vendor funds or previews this work.
Correction, 2026-09-06. This page previously reported a cross-product correlation of r = 0.056 and described lender quality as “product-specific, not a single reputation.” That estimate rests on six lenders and is not robust: Spearman is 0.543, the 95% confidence interval spans -0.79 to +0.83, and excluding one institution moves Pearson to 0.835. The correlation claim is withdrawn. Note the direction of the instability: deleting the most divergent lender does not weaken a relationship, it creates one — so the sample is dominated by a single observation and cannot support any cross-product claim. What survives is an existence result: at least one large lender (Mr. Cooper: FHA 8.9% vs Conventional 47.0%, both with tens of thousands of decisions) behaves very differently across products in the same year — enough to show that a single lender-level denial rate should not be assumed to be program-invariant, but not enough to estimate how common that is. Logged in the public corrections log.