September 2026 Mortgage Survey Gap Explained

The September 2026 mortgage survey gap was nine basis points for 30-year fixed loans. Freddie Mac’s September 24 release reported an average of 7.03%; the Mortgage Bankers Association’s September 23 release reported 7.12% for conforming loans. The lower, later-published figure does not establish an overnight decline in pricing.

The surveys cover different measurement periods and loan populations, and they report fees differently. Those distinctions explain why publication order alone cannot tell borrowers whether pricing improved. Neither average represents an offer a borrower can select.

What the September releases reported

MBA’s rate-and-points figures above assume 80% loan-to-value, meaning the loan equals 80% of the property’s value. Its conforming category covers balances of $832,750 or less.

The 30-year difference is 0.09 percentage points, or nine basis points; the 15-year difference is one basis point. Neither measures the savings available from switching lenders. Our mortgage-rate data and reports provide context for following each series over time.

The measurement periods end five days apart

Applying Freddie Mac’s Thursday–Wednesday schedule, the September 24 Primary Mortgage Market Survey (PMMS) covers September 17–23. MBA’s September 23 release identifies its reporting period as the week ending September 18.

PMMS therefore includes September 19–23, dates beyond MBA’s stated reporting week. Calling these Wednesday’s and Thursday’s mortgage rates obscures that difference: neither number isolates pricing on publication day.

To assess weekly direction, compare each series with its own previous reading. Freddie Mac’s 30-year average increased eight basis points; MBA’s increased 15. Both weekly measures rose, even though the later-published number was lower. Neither weekly change identifies the direction of rates on a particular day.

The surveys cover different application populations

Freddie Mac draws purchase applications from Loan Product Advisor, its automated underwriting system. Its filters include conventional conforming loans, one-unit owner-occupied homes, FICO scores of at least 740 and loan-to-value ratios of 75%–80%. The measure uses application submissions rather than selecting only rate locks. Freddie Mac’s methodology explains those restrictions.

MBA’s dated release describes a survey of U.S. closed-end residential mortgage applications through retail and consumer-direct channels, reporting both purchase and refinance activity. That describes the application survey’s coverage; it does not establish that its rate average uses Freddie Mac’s borrower filters.

The releases do not quantify how much of the nine-basis-point gap comes from timing, borrower mix or sampling. The differing averages alone cannot separate those effects.

Unreported points do not mean zero points

PMMS does not publish average discount points or origination fees under its application-based methodology. An unreported fee is not a zero fee: the average cannot be labeled a zero-point offer. Freddie Mac’s methodology note documents that limitation.

MBA reports the contract interest rate separately from points, with origination fees included in the points figure. Its 0.73 points therefore cannot all be described as money paid to buy down the rate.

One point equals 1% of the loan amount. On a hypothetical $400,000 loan, 0.73 points would equal $2,920 upfront—not an addition of 0.73 percentage points to the interest rate. The rate reduction purchased by discount points varies by lender, product and market conditions, as the CFPB’s points guidance explains. Without comparable fee information, these releases cannot establish which financing package costs less.

What nine basis points means in payment arithmetic

For a hypothetical payment illustration, assume a $400,000 balance, 360 monthly payments and a fully amortizing fixed-rate loan. Using the two reported rates as mathematical inputs produces monthly principal and interest of approximately:

• 7.03%: $2,669.27
• 7.12%: $2,693.52
• Difference: $24.25

These calculations exclude points, other closing costs, taxes, homeowners insurance, mortgage insurance and association dues. They show the payment effect of changing the rate under fixed assumptions, not two available offers or their total borrowing costs.

Use the mortgage payment calculator to change the balance while holding the term and other assumptions fixed.

Use borrower-specific documents for a quote discussion

For an actual comparison, use Loan Estimates for the same loan amount, purpose, property, loan type and term. Compare offers obtained close together, and align the points or lender-credit assumptions. Review principal and interest alongside upfront lender charges, mortgage insurance and cash to close. The CFPB’s comparison guidance explains which figures to examine.

Confirm the lock period and expiration separately. Issuing a Loan Estimate does not necessarily lock the rate; the document indicates lock status, and the lender should explain any extension costs. CFPB rate-lock guidance covers those questions.

Our guide to how mortgage pricing works can support that conversation. When a borrower is weighing points, ask the loan officer to show total costs over plausible holding periods using that borrower’s actual terms.