Loading today’s stories
Invest1 distinct publisher3 min readPublished
A 6.71% mortgage sits 1.97 points above the 10-year Treasury, the same gap as before January's buyback fanfare, which means a 6% handle now needs a 4.03% ten-year that the market has not offered since mid-2023.
The Investor · Invest desk
Compiled by The InvestorSomething wrong?How this is made
The arithmetic of a mortgage rate is a sum, not a ratio: take the 10-year Treasury yield, add the spread, sign the paper. At Wednesday's weekly average yield of 4.74% and a spread of 1.97 points you get the 6.71% Freddie Mac printed [10]. Mark it to the 4.77% the ten-year touched later in the week and the same spread gives 6.74% [1]. For that borrower to see a 6% handle, the ten-year has to fall to 4.03% [2], the very floor of the 4-5% band that has held since mid-2023 [8]. Get there through the spread instead, with the ten-year parked where it closed Wednesday, and the gap has to compress to 1.26 points [3], which is tighter than the sub-1.5 points the Fed achieved while buying mortgage-backed securities outright and mortgage rates went under 3% [16], by about a quarter of a point [6].
That is the problem with the plumbing. Fannie and Freddie fund the buybacks with operating cash flow that would otherwise have gone into Treasuries, and by shedding Treasuries they already hold, swapping government paper for their own [6]; Wolf Richter's reading is that this turns two large holders into sellers and pushes Treasury yields up, narrowing the spread from the wrong end [7]. Since a 30-year mortgage is priced off the ten-year precisely because it typically gets paid off in about 12 years [3], the selling lands on the exact benchmark the program is trying to undercut. The spread is the number the buybacks can move. The rate is the number the buyer pays.
Richter asks where mortgage rates would be without the buybacks, and answers himself with a question mark: over 7%? [15] That is success defined on a counterfactual nobody can price. What can be priced is the sequence: 6.21% before the January 8 expansion, 6.01% by the end of February, 6.71% now [14], which is 50 basis points above the announcement level and 70 above the low [4]. And the spread today is roughly where it stood at the turn of the year, before the announcements [11], having done its full point of narrowing back when the ten-year was going nowhere and mortgage rates were falling on their own [9].
This is probably wrong, but the honest read is that the buyback program is a portfolio decision wearing housing policy's clothes: the GSEs are concentrating balance sheet into their own credit and prepayment risk and withdrawing a bid from Treasuries, and the borrower gets the sum of both effects. The counter-thesis has teeth, mind you. The Fed ended QT in December 2025 but kept running off MBS, which pushes the spread the other way [13], and in a $12 trillion market [12] the GSEs are one hand among many, so the buybacks may be holding a line that would otherwise have slipped back toward the three points of 2023 [4]. I would take that seriously if the spread broke under 1.7 while the ten-year sat in the low fours. Until it does, the 7% mortgage is a Treasury story.
Ranked by verification strength, evidence, and original report placement.
Freddie Mac's weekly measure showed the average 30-year fixed mortgage rate rising to 6.71% for the week through Wednesday September 2, the highest since July of the prior year.
The 30-year fixed mortgage rate has been in the 6.0% to 7.0% range since September 2022, apart from a few spikes to the upside.
The 30-year fixed mortgage rate tracks the 10-year Treasury yield but sits above it, because a 30-year mortgage on average gets paid off in about 12 years as homes are sold or refinanced.
The spread widened in 2022 and 2023 to more than 3 percentage points, the most since the early 1980s, then began narrowing while remaining historically wide through mid-2025.
Fannie Mae and Freddie Mac, under government conservatorship, began buying back MBS they had previously issued in the second half of 2025 and accelerated the buybacks in January 2026.
The 10-year Treasury yield dipped to 4.77% on the day of publication, in the upper part of the 4-5% range that has prevailed since mid-2023.
Distinct publishers with included, body-backed reporting in this cluster.
Follow any of these and your For You feed starts watching them — no settings page required.
invest
A four basis point thaw: mortgage rates dip to 6.65% for a second week1 distinct publisher
invest
Treasury moved $742 billion in a week. The price was a 5.216% thirty-year.1 distinct publisher
invest
Four Years Below Housing Bust Lows Makes the Freeze an Underwriting Assumption2 distinct publishers
invest
Central banks have sold $233 billion of Treasuries since February. Private money did not blink1 distinct publisher
Evidence-backed comparisons of source perspectives and observed adoption signals. Read the methodology
Which Builder, Operator, and Investor concerns the observed source mix emphasized—not a truth score.
Evidence, demonstrated adoption, hype gap, incentives, and confidence are assessed independently, each on its own current evidence. How these are measured.
Public series, private mechanism
Two numbers carry the story and both are checkable by anyone: Freddie Mac's 6.71% weekly print and the 4.74% weekly average on the 10-year. Subtract, and you have the 1.97-point spread the whole argument rests on. The mechanism is another matter — the claim that Fannie and Freddie are funding buybacks by selling Treasuries and thereby lifting yields comes with no purchase volumes, no holdings change and no agency document behind it.
Program running, size undisclosed
This is not a pilot: the buybacks started in the second half of 2025, were expanded on January 8 and have had eight months to work in a $12 trillion market. What we cannot see is scale — not one dollar figure of purchases appears — so the only read on take-up is the outcome, and the outcome is a spread sitting exactly where it sat before the expansion was announced.
Deflates the policy, oversells the mechanism
January supplied the fanfare; September supplies a spread of 1.97 points, unchanged. On the headline question this reporting runs against the hype rather than with it. It then overreaches twice in the other direction: the buybacks-lift-Treasury-yields chain is asserted without a single flow figure, and 'Over 7%?' is a question doing the work of a counterfactual estimate. Add the teased 'looming 7% mortgage rate' that the body itself calls historically unremarkable, and the net tilt is mildly overstated.
House view, plainly worn
Nobody here is selling a bond. Wolf Street is a single-author, reader-supported site with a standing thesis that 2008–2022 was financial repression, and that thesis shapes the vocabulary and the choice of which comparison periods to draw — the QE-era sub-1.5-point spread and sub-3% mortgage as the cautionary benchmark rather than the aspiration. The stance is stated openly, which makes it easy for a reader to discount; the pull toward a rate-alarm frame is real but visible.
Solid numbers, one voice
Confidence sits mid-range for a simple reason: the arithmetic can be reproduced from published rate series in a minute, but every interpretive step — why the spread is stuck, whether the agencies are lifting yields, what would have happened without the program — has exactly one author behind it and no corroborating account in our coverage.