Every year, researchers, journalists, and mortgage professionals cite HMDA denial rates as a measure of how hard it is to get a mortgage in a given state. Mississippi is hard. Utah is easy. End of story.
Except it isn't. There's a fundamental problem with this analysis that nobody talks about: HMDA only captures people who applied for a mortgage. It says nothing about the people who bought homes with cash — and in some states, that's nearly half of all buyers.
We ran the numbers. Here's what we found.
When half a state's buyers pay cash, the mortgage applicant pool is not a random sample of that state's homebuyers. It's the other half — the people who needed credit to buy. On average, that group has lower savings, lower incomes, and weaker credit profiles than the cash buyers they're competing against.
So when HMDA tells you Mississippi's denial rate is 30%, it's not measuring how hard it is to get a mortgage in Mississippi. It's measuring the denial rate among the self-selected group of Mississippians who couldn't afford to pay cash — which is a fundamentally different, and more financially strained, population.
"A high denial rate might mean lenders are strict. Or it might mean the applicant pool is weaker. HMDA can't tell you which — and neither can anyone who quotes HMDA without this caveat."
| State | Cash sale % | HMDA denial % | Median income |
|---|---|---|---|
| MS | ~50% | 30.0% | $52,000 |
| WV | ~42% | 29.7% | $55,000 |
| HI | ~45% | 30.7% | $88,000 |
| FL | ~43% | 26.6% | $67,000 |
| MT | ~46% | 22.2% | $64,000 |
| ID | ~45% | 17.5% | $66,000 |
| CA | ~19% | 24.3% | $91,000 |
| WA | ~20% | 20.0% | $92,000 |
| NJ | ~27% | 26.7% | $97,000 |
| MA | ~24% | 25.8% | $96,000 |
Notice the pattern — or rather, the lack of one. Hawaii has a 45% cash rate and a 30.7% denial rate. Idaho has a 45% cash rate and a 17.5% denial rate. Same cash penetration, completely different denial rates. Cash share alone doesn't predict denial rates.
This analysis uses state-level averages, which smooth over enormous within-state variation. Miami's cash rate (~43%) is very different from rural Florida's. And our HMDA data covers FHA loans specifically — a segment already skewed toward first-time, lower-credit borrowers. The selection effect is almost certainly larger in FHA data than in conventional loan data.
We're also working with 34 states where we have matched cash sale and HMDA data. The remaining states may behave differently. This is a hypothesis-generating analysis, not a definitive study. The right next step is county-level matching — which requires data we don't yet have.
A note on conflicting cash-sale figures: NAR's REALTORS® Confidence Index puts the cash share of transactions at 25% (April 2026), notably lower than ATTOM's 39% figure used in this analysis. The difference comes down to scope: NAR's number reflects REALTOR-facilitated transactions only — sales that went through an agent. ATTOM's figure includes the broader market: FSBO sales, investor bulk purchases, and non-MLS transactions, where cash buyers are overrepresented. Both figures are correct for what they measure. We use ATTOM's broader figure because it better represents the full universe of home sales that HMDA's mortgage-only data is being compared against.
But the core point stands: denial rates tell you about the people who applied, not the people who bought. That's a meaningful distinction that almost every HMDA analysis ignores.
Lender choice explains denial rates far better than geography. See which lenders approve your profile.
Route with Zai — 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.
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.