Your mortgage rate is the 10-year Treasury plus a number that moves

On Thursday, May 14, the average 30-year fixed mortgage rate in the United States was 6.36%. One week later, on May 21, it was 6.51% — fifteen basis points higher. Over the same stretch the 10-year Treasury yield rose from 4.46% to 4.57%. Both numbers went up. They did not go up by the same amount, and the gap between them — which is the actual subject of this post — widened from 1.90 points to 1.94.

That gap is the subject of this post. It is also, in a sense, the subject of this entire site.

A short, honest note first, because this is distillfin's first post under a new editorial direction and pretending otherwise would be the wrong way to start. This domain previously published general personal-finance content — the kind of thing that gets auto-generated, lightly researched, and quietly optimized for search rather than for readers. That content is being retired. What distillfin does now is narrower and, I hope, more useful: it takes a specific, publicly available data series, charts it, and explains what the chart does and does not mean. No price targets. No "you should buy." Just public data, read carefully. My name is Ravi. The disclaimer at the top of every article is not boilerplate: this is data analysis, not advice you should trade or borrow on.

With that said — the spread.

The number nobody quotes

When mortgage rates are in the news, the number you hear is the headline rate: 6.51%, in this case, from Freddie Mac's Primary Mortgage Market Survey, which FRED publishes as the series MORTGAGE30US. When Treasury yields are in the news, you hear the 10-year: 4.57%, published as DGS10.

What you almost never hear quoted is the difference between them. And the difference is where most of the interesting behavior actually lives.

Here is the relationship, stated as plainly as I can: the 30-year mortgage rate is approximately the 10-year Treasury yield plus a spread. The spread is not a constant. Over the long run — pulling both series back to the early 1970s, when MORTGAGE30US begins — it has averaged somewhere around 1.7 to 1.8 percentage points. Right now it is sitting at 1.94. That is wider than the historical norm, and it has been wider than the historical norm for most of the last three years.

If you take one thing from this post, take that: when you read that mortgage rates went up, you are reading two facts compressed into one. The Treasury side moved, and the spread moved, and they are driven by different things.

Why a 10-year Treasury is even the comparison

A reasonable question: why compare a 30-year mortgage to a 10-year bond? The mortgage is 30 years; the obvious match would be the 30-year Treasury.

The answer is prepayment. A 30-year mortgage is almost never held for 30 years. People sell their homes, they refinance, they pay early. The effective life of a typical mortgage has historically clustered somewhere around 8 to 12 years, which makes the 10-year Treasury the closer behavioral match. The industry settled on the 10-year as the reference rate for exactly this reason, and DGS10 is the series that reference points at.

So the comparison isn't arbitrary. It reflects how long the money is actually out the door.

What lives inside the spread

If the mortgage rate were just the 10-year plus a fixed number, mortgage lending would be a much calmer business than it is. When I first started tracking these two series side by side, the constant-gap assumption is exactly the thing I had to unlearn. The spread moves because it is compensation for a stack of real risks and costs that a Treasury bond does not carry:

• Prepayment risk. This is the big one. When rates fall, borrowers refinance and the lender's high-rate asset disappears right when it became valuable. A Treasury can't do that to you. Investors demand to be paid for that asymmetry, and that payment is part of the spread.
• Credit risk. A homeowner can default. The U.S. Treasury, for practical purposes within this analysis, does not. Even with government-backed mortgage securities, the servicing and guarantee machinery costs something.
• Servicing and origination cost. Collecting payments, managing escrow, handling delinquencies — none of that is free, and it is baked into the rate.
• Liquidity and demand for mortgage-backed securities. Most U.S. mortgages are bundled into MBS and sold. The spread reflects how much investors currently want to hold that paper versus, say, holding Treasuries directly.

When the spread is near its long-run 1.7-to-1.8 level, those costs are being priced about where they have historically sat. When it is near 1.94, as it is now, the market is charging more than usual for that bundle of risks — and the single largest contributor over the past few years has been the last item on the list: weaker structural demand for mortgage-backed securities, partly because the Federal Reserve stopped being a large buyer of them after it began shrinking its balance sheet.

What the last two weeks actually showed

A note on method before the table, because pairing these two series requires one decision. The mortgage figure is published every Thursday; the 10-year trades every business day. Throughout this post I pair each Thursday mortgage reading with the 10-year's close the day before — Wednesday — the last full trading day before the mortgage number lands. It is one defensible convention among a few, and stating it out loud is the difference between a number and a number you can check.

Here is the recent movement on that basis:

Between the May 14 and May 21 readings, the mortgage rate rose 15 basis points. The paired 10-year rose 11 of those. The remaining 4 came from the spread widening, from 1.90 to 1.94. So a reader who saw "mortgage rates jumped this week" and assumed it was entirely the bond market would be roughly three-quarters right — and one-quarter wrong. Most of the move was the Treasury. Not all of it, and "most" is not "all."

There is also a sharper move inside that week worth flagging honestly, because it cuts against a tidy narrative. The 10-year did not climb smoothly. On a daily basis it ran from 4.47% on May 14 up to 4.67% on May 19 — a 20-basis-point rise in three trading sessions — then gave back 10 of those basis points by May 20, landing at the 4.57% the table uses. The weekly mortgage figure, published once and reflecting the days before it, smooths that entire spike away. The chart of DGS10 is jagged; the chart of MORTGAGE30US is a staircase. They are tracking the same underlying force at different resolutions, and knowing that is the difference between reading the data and being read by it.

How to check this yourself

Everything above comes from two free, public FRED series. You do not need a subscription, a terminal, or my chart. You need:

• DGS10 — 10-Year Treasury Constant Maturity Rate, updated daily.
• MORTGAGE30US — 30-Year Fixed Rate Mortgage Average, updated weekly on Thursdays.

Pull both from the St. Louis Fed's FRED site, put them on the same axis, and subtract one from the other. The spread series you get is the thing this post is about. FRED will even compute the subtraction for you if you add MORTGAGE30US and DGS10 to one graph and use the built-in formula field — the expression is just a-b.

I ran exactly that this week. The recent DGS10 observations came back ranging from 4.26% in mid-April to 4.67% on May 19; the MORTGAGE30US readings ranged from 6.23% in late April to 6.58% last August. Those are the raw numbers behind every sentence above, and you can reproduce them in about four minutes.

What this does not tell you

I want to be precise about the limits, because the failure mode of finance writing is taking a real pattern and stretching it into a forecast.

The spread being wider than its historical average is not a prediction that it will narrow. "Wider than average" has been true for roughly three years; mean reversion is a tendency, not a schedule. Anyone who tells you the spread must compress back to 1.7 by a particular date is selling you a timeline the data does not contain.

It is also not advice about whether to lock a mortgage rate, refinance, or buy a house. Those decisions depend on your own situation, and a chart of two national-average series cannot see your situation. What the spread can do is make you a more literate reader of the headline: when the mortgage rate moves, you now know to ask which half moved — the Treasury, or the spread.

That is the whole job of this site, stated in one example. Chart the public data. Explain the mechanism. Stop before the part where someone tells you what to do with it. I would rather publish a post that ends at "here is what the gap measures" than one that pretends the gap is a signal to act on — because the first one is true and the second one is the thing that got the old version of this site retired.

Next week's post stays on FRED, and stays on rates: I want to chart the 10-year against the 2-year, because the gap between those two — the part of the yield curve everyone actually argues about — has its own story, and its own set of things it cannot tell you.

— Ravi