Why Mortgage Rates Don't Follow the Fed

On September 18, 2025, the Federal Reserve cut its policy rate by a quarter point. Over the next two weeks the average 30-year fixed mortgage rate went from 6.26% to 6.34%.^[1] Up. Not down. Up, right after a cut, which is the exact opposite of what every headline that week told people to expect.

This is not a fluke, and it is not the Fed being sneaky. It happens because the thing the Fed sets and the thing that prices your mortgage are two different instruments, in two different markets, held by two different sets of buyers. Once you actually see the chain, the whole “wait for the Fed to cut and then buy” plan falls apart. I want to walk through that chain carefully, because it is one of the highest-stakes misunderstandings in personal finance. People delay six-figure decisions on the strength of it.

The Fed sets a one-day rate. Your loan is thirty years long.

The federal funds rate is the interest rate banks charge each other for overnight loans of reserves. One night. That is the entire maturity. In June 2026 the FOMC held that target range at 3.50% to 3.75%.^[6] It is a real lever, and it does move things that reprice fast: credit cards, HELOCs (home equity lines of credit, which float against prime), auto loans, savings yields, and adjustable-rate mortgages once they escape their fixed period.

A 30-year fixed mortgage is not a fast-repricing instrument. It is a loan that gets bundled into a mortgage-backed security (an MBS, a bond made of thousands of pooled home loans whose payments flow through to investors) and sold to a bond buyer who is comparing it to other long-dated bonds. The natural comparison is the 10-year Treasury, because the average 30-year mortgage doesn't actually last 30 years. People move, refinance, or pay off. Effective life lands closer to seven to ten years, so the 10-year is the honest benchmark.

Researchers at the Atlanta Fed put it plainly: for the last twenty years, mortgage rates have tracked the 10-year Treasury far more closely than the fed funds rate.^[1] The Dallas Fed went further and said the quiet part out loud. There is no direct link between the policy rate and mortgage rates. The gap between the 30-year mortgage and the fed funds target has been as narrow as 70 basis points and as wide as 600.^[3] A relationship that ranges from 0.7 to 6.0 percentage points is not a relationship. It is a coincidence with good PR.

The actual chain, start to finish

Here is what really determines the number on your rate sheet. It stacks, and only one layer is even adjacent to the Fed.

The spread is the part almost nobody explains, and it's where the interesting money is. Since the end of the Great Recession the 30-year mortgage has averaged about 1.7 percentage points above the 10-year Treasury. Right now it's closer to 2.0.^[11] On a $500,000 loan, that extra 30 basis points is roughly a hundred bucks a month, forever, and it has nothing to do with monetary policy.

The free option you didn't know you had

When you take out a 30-year fixed mortgage in the US, you get something a bond investor almost never gives away: the right to pay the whole thing off, at any time, for no penalty. Rates drop? You refinance. The investor who was collecting 7% gets handed their principal back and has to redeploy it at 5%.

Rates rise instead? You do absolutely nothing. You sit on your 3% loan and laugh. The investor is now stuck holding a below-market bond for years.

Heads you win, tails they lose. That asymmetry has a name in bond math: negative convexity. A normal bond gains more when yields fall than it loses when yields rise. An MBS does the reverse, because falling yields trigger the prepayments that take the bond away from you. The Boston Fed published a paper in May 2026 that nailed this down: after you strip out cash-flow timing, credit risk, and lender costs, the large and volatile gap that remains between mortgage rates and Treasury yields is essentially the price of the prepayment option.^[2]

The Boston Fed found something else that's genuinely useful for forecasting: the shape of the yield curve moves the spread. A steep curve (long rates well above short rates) signals the market expects rates to rise, which makes refinancing less likely, which makes your free option less valuable, which narrows the spread. Roughly 40 basis points of narrowing per percentage point of steepening.^[2] A flat or inverted curve does the opposite.

The Dallas Fed's model lands in the same place from a different angle. Twenty years of data, three factors: the level of 10-year rates, the slope of the curve, and implied interest-rate volatility.^[3]Notice what's not on that list. The overnight rate that CNBC breaks into programming to tell you about.

Three times the Fed moved and mortgages went the other way

Theory is cheap. Here's the receipt, three times over.

The 2024 episode is the one to burn into memory, because it contained the perfect experiment. The Fed delivered exactly what the “wait for cuts” crowd was praying for. A full 100 basis points. And the buyer who waited for it ended up with a rate roughly 25 basis points worse than if they'd locked before the first cut.^[15]That is what a failed thesis looks like when it's tested with real money.

The market moved on. So did the buyer of last resort.

There's a structural reason spreads have stayed fat, and it has to do with who's actually buying the bonds.

From 2020 through early 2022 the Fed was buying agency MBS by the tens of billions each month. That's a price-insensitive buyer. It does not care about option-adjusted spread or convexity. It just buys. That bid compressed the mortgage spread to historically thin levels and is a big part of why anyone got a 2.75% mortgage.

That buyer is gone. Quantitative tightening formally ended in December 2025, but the Fed is not reinvesting MBS paydowns back into MBS. It's rolling them into Treasury bills, with roughly $180 billion of MBS paydowns projected for calendar 2026, which means the portfolio keeps shrinking passively.^[14][17] The bonds have to be absorbed by private investors: banks, money managers, insurers, REITs. People who very much do care about profits, and who are pricing in the risk of softening home values.^[14]

And the other side of the bond market is getting crowded. Federal deficits mean the Treasury has to issue more debt, and more supply at a given level of demand means higher yields. Fortune ran the arithmetic in June 2026 and pointed at the $3.4 trillion added to deficits through 2034 by the 2025 tax bill as a live reason the 30-year sat around 6.48%.^[12]You can argue about the magnitude. You can't argue with the direction. Your mortgage rate has a fiscal-policy component, and Jerome Powell doesn't vote on appropriations.

So what actually moves your rate?

Two buckets. Market stuff you can't control, and borrower stuff you can. Most people obsess over the first and ignore the second, which is backwards, because the second one is worth more.

• Inflation prints. CPI and PCE releases move the 10-year more in an hour than an FOMC decision usually does in a day. The Fed decision is mostly priced in before it happens. A surprise inflation number is not.
• Treasury supply and deficits. Auction results, refunding announcements, debt-ceiling theater. Boring, and they matter.^[12]
• Rate volatility. When the bond market gets jumpy, the prepayment option gets more valuable and the spread widens. Calm markets are cheap markets.^[3]
• Your credit score, LTV, points, and loan type. The difference between a 680 and a 780 FICO, or between 5% down and 20% down, is often bigger than anything the Fed will do to you in a year. Nobody writes headlines about this. It's the part you actually control.
• Shopping around. Lender margins are a real component of the spread. Getting three or four quotes on the same day is the single highest hourly-wage activity in the entire homebuying process.

What this means if you're buying or holding property

I'll be direct, because the hedge-everything version of this advice is useless.

“Waiting for the Fed to cut” is not a strategy. It is a bet on an indirect, two-step, historically unreliable transmission channel, and you are making it with a leveraged, illiquid asset. Anyone who sat out the fall of 2024 waiting for the cuts got both a worse rate andanother year of price appreciation to pay for. That's the whole trade, and it lost twice.

The demand you should be afraid of is the other buyers.This is the piece people miss. If mortgage rates genuinely drop a full point, every buyer who was sitting on the sidelines shows up on the same Saturday. You don't get the low rate and the calm market. You get the low rate and a bidding war. Cheap money capitalizes into the price. Buying into a high-rate, thin-competition market and refinancing later is, in most markets, the better-shaped trade. Not because it feels good. Because the rate is refinanceable and the purchase price is not.

If you're a landlord, underwrite to the spread, not to the Fed.Your going-in cap rate has to work at today's debt cost with today's rents. A deal that only pencils if you refinance into a rate that requires the 10-year at 3.25% is not a deal, it's a directional bet on the bond market wearing a spreadsheet. And if you're on a bridge loan or a floating-rate facility, then congratulations, you are one of the few real estate people for whom the Fed genuinely is the main event. Those float off SOFR, which does track policy.

The mental model to keep

Stop watching the Fed to predict your mortgage. Watch three things instead, in order.

1. The 10-year Treasury.It's free, it's public, it updates all day.^[7] This is your base layer.
2. The mortgage spread. Take the Freddie Mac weekly survey number and subtract the 10-year.^[5] If that gap is running near 2.0 points against a post-2009 norm around 1.7, you have upside from the spread compressing that has nothing to do with the Fed.^[11]
3. Inflation expectations. They set the 10-year. Everything else is downstream.

In July 2026 that stack reads: fed funds at 3.50% to 3.75%, a 10-year around 4.6%, and a 30-year fixed at 6.49%.^[4] The Fed has been holding. The 10-year has been the story. It has always been the story. The Fed just has a better press office.

Which brings me to the actual point of all this. The number on your loan estimate is not handed down by a committee in Washington. It is assembled, in real time, by strangers who are pricing the risk that you will refinance out from under them the moment it's convenient. Once you see it that way, you stop waiting for permission from the Fed and start doing the two things that actually work: buy when the math works at today's rate, and be the kind of borrower a bond investor wants to lend to.