A Tenth of a Point Off Mortgage Rates Moved Nobody

Mortgage rates fell a tenth of a point in August 2026 and purchase applications fell anyway, for the third straight week. If you own a home and are waiting for cheaper money to refill your buyer pool this fall, the demand data already voted. It voted no.

A $20 dip that is still a year-over-year increase

Per Freddie Mac, the 30-year fixed averaged 6.67% for the week of August 13, 2026, down from 6.69% the prior week and roughly a tenth below the late-July 2026 peak near 6.77%. On a $300,000 loan, the move from 6.77% to 6.67% saves about $20 a month. That is not a demand catalyst.

The headline also hides the direction that matters. At 6.67%, the rate sits above the 6.58% Freddie Mac recorded a year earlier, in August 2025. Same loan, the payment today runs about $18 a month more than it did last summer. The weekly trend points down. The annual trend points up. Sellers are being sold the first line and living the second.

Buyers had the discount and stayed home

If a small rate break were enough, applications would show it first. They show the opposite. Per the Mortgage Bankers Association, purchase applications for the week ending August 14, 2026 fell 2% from the prior week and ran about 3% below the same week a year earlier, the third consecutive week of annual declines. Refinance volume sat roughly 18% under its year-ago level.

New construction confirms it. MBA's builder survey showed applications for new-home purchases down 5.7% in July 2026 from a year earlier, per HousingWire, with new-home sales tracking near a 647,000 annual pace, down 3% from June 2026.

The read: at roughly 6.7%, the marginal buyer is blocked by income and payment math, not by a rounding error in the rate. Shaving ten basis points does not widen a seller's buyer pool in any way the seller can feel.

The Treasury built this dip. It can unbuild it.

The decline is not coming from cooling inflation or a Fed cut. The New York Times reported on August 20, 2026 that the Treasury has turned to interventionist tactics to pull borrowing costs down, including expanding its bond buyback program. Bankrate noted in the week of August 19, 2026 that the buyback size doubled.

A rate that falls because the government is buying bonds is more fragile than one that falls because the economy is slowing. It can reverse when the intervention pauses or bond markets push back, and rates already spiked to a more-than-one-year high after the July 2026 FOMC meeting. Do not build your listing timeline around a number that can be un-engineered faster than your house goes under contract.

Price to the buyer who exists at 6.7%

Two conclusions follow, and both cut against waiting.

First, the leverage is drifting toward buyers, not sellers. Freddie Mac's August 2026 commentary put listing prices modestly below year-ago levels while for-sale inventory improves. More competing supply against flat demand means pricing to the market you have, not to the buyer you hope shows up at 6%. Our national housing market report tracks that supply build monthly.

Second, equity growth will not cover the cost of waiting. Our data across 26,274 US ZIP markets shows the typical home worth about $289,803 in July 2026, a 3.0% annual gain that is cooling toward the pace of inflation. Holding a listing through fall means paying carrying costs against appreciation that barely outruns them, in exchange for a demand wave the application data says is not coming. Sellers who need to move on a schedule should weigh the speed-versus-price tradeoffs now rather than bet on a rate reversal.

The rate move is real but small, still worse than a year ago, and manufactured rather than earned. Buyers looked at it and stayed home. Sell into 6.7%, set the price for the buyer who is actually shopping, and stop asking a tenth of a point to do work it cannot do.