The Bond Market Is Doing the Fed's Job
Kevin Warsh speaks at Jackson Hole this week. The Fed has held rates steady, but long-term borrowing costs are doing the tightening.
Key Takeaways
- Long rates have moved higher without another Fed hike. The repricing has also been remarkably orderly: 10- and 30-year Treasury volatility fell after the leadership handover, while the shape of the curve changed little. The Fed Chair got tighter conditions without spending the political capital of a rate hike.
- The latest move is not an inflation-expectations story — or a term-premium story. Since May 22, the 10-year Treasury yield is up about 13 basis points, while the 10-year TIPS yield rose 19 basis points and breakeven inflation fell 6. The Kim–Wright term premium barely moved over the same window.
- But the term premium is still elevated, and that matters for where rates settle. Record government borrowing, growing AI-related duration supply and a more price-sensitive investor base give long-term investors reasons to demand more compensation to hold duration.
The 30-year Treasury yield has pushed above 5.3%, its highest since 2007. The 10-year is around 4.7%. And the Federal Reserve has done nothing to its policy rate since the leadership handover, holding the federal-funds target at 3.50%–3.75% in July.
Yet financial conditions have tightened anyway.
That is the sense in which the bond market is doing part of the Fed’s job. Long-term rates have risen without another increase in the overnight rate, raising the cost of mortgages and other long-duration financing.
But there are two different questions here, and separating them matters.
Why did long rates rise recently?
Why are long rates still so high in the first place?
The data give different answers.
1. Rates moved higher — but the bond market didn’t break
Start with what actually happened after the May leadership handover.
Using five-day averages to avoid putting too much weight on a single close, the 2-year Treasury yield rose about 10 basis points through August 17, the 10-year about 9 basis points and the 30-year about 13.
The entire curve repriced higher.
What did not happen was a dramatic change in its shape. The 2s10s spread narrowed by just 1 basis point, while 2s30s widened by about 3. Nor did the move produce a volatility shock. Over matched 58-trading-day windows, realized volatility was essentially unchanged at the 2-year, down about 10% at the 10-year and down about 8% at the 30-year.
(These curve and volatility comparisons use five-day averages; the point-to-point decomposition in the next section, anchored to exact endpoints, puts the 10-year’s rise at about 13 basis points. The two methods differ by a few basis points by construction.)
In plain English: yields moved. The market did not become disorderly.
That distinction matters going into Jackson Hole.
Warsh has already overseen a sharp change in how the Fed communicates. The first two policy statements under his chairmanship are about 55% shorter than Powell’s final eight, and the recurring language the Fed used to describe its reaction function and possible future policy adjustments has largely disappeared.
That does not mean the communication change caused long rates to rise. Inflation, growth, fiscal policy, global developments and expectations for future Fed policy all moved during the same period.
But it does tell us something about Friday.
A Fed that has deliberately said less about its next move has little reason to suddenly restore detailed forward guidance at Jackson Hole. The better expectation is a broad speech about the economy and the policy framework, not a promise about September.
In plain English: the bond market has pushed borrowing costs higher without panicking. That gives the Fed tighter financial conditions without having to raise its own policy rate — and it reduces the pressure on Warsh to use Jackson Hole to prepare markets for an immediate move.
2. The latest increase in the 10-year isn’t an inflation scare
Now look inside the recent move.
A nominal Treasury yield can be split into two observable pieces: the TIPS real yield and breakeven inflation compensation.
Since May 22:
- 10-year nominal yield: +13 bp
- 10-year TIPS real yield: +19 bp
- 10-year breakeven inflation: −6 bp
That is a useful result.
The market is not charging more because 10-year inflation compensation has risen. It has actually fallen. The latest increase in the 10-year is on the real-yield side of the market, not the inflation-compensation side.
There is another distinction that matters just as much.
In a term-structure model, a long-term Treasury yield can be decomposed into the expected average path of future short rates plus a term premium — the extra compensation investors require to hold duration rather than continually roll short-term debt.
The Kim–Wright estimate of the 10-year term premium was about 0.83% on May 22. By August 14, it was about 0.84%.
That is essentially unchanged.
So the latest rise in the 10-year did not run through the term premium, at least in this model.
And the fact that yields moved higher across the curve with little change in slope does not tell us otherwise. A near-parallel shift can reflect changes in expected future short rates, term premia or both.
The interesting thing about the term premium is not what it did this summer. It is where it already was.
In plain English: investors are not suddenly demanding higher yields because they expect more inflation, and the Fed’s term-premium model does not show a new surge in compensation for holding long bonds. The recent increase is coming from the real side of interest rates — which means markets are demanding a higher return after inflation, for reasons that likely include expectations about future real rates.
3. Why the long end can still stay high
This is where the longer history matters.
My earlier decomposition of the 10-year Treasury looked over several years rather than several months. On that horizon, the result was very different. In the ACM model, the 10-year yield had risen about 107 basis points over three years even though the model’s expected-short-rate component had fallen about 36 basis points. The term premium had risen roughly 143 basis points. Kim–Wright showed a smaller increase, but the same direction.
So there are two stories:
- The latest move: mostly a real-yield move, with little change in the Kim–Wright term premium.
- The longer-run backdrop: a term premium that has already risen substantially from the unusually depressed levels of the previous era.
That is important because the amount of duration investors are being asked to absorb is growing.
Governments and companies are expected to borrow roughly $29 trillion from bond markets in 2026, double the amount a decade ago. U.S. federal debt has moved above $40 trillion. At the same time, the AI investment boom is creating another source of long-duration financing demand. Dallas Fed researchers estimate AI-related investment-grade issuance could be around $300 billion this year alone, with additional duration entering markets through private credit and swaps.
The investor base is changing too.
The share of total U.S. federal debt held by foreign and international investors peaked at roughly 34% in 2013 and is now around 24%. That does not mean foreign investors are dumping Treasuries. Their holdings simply have not kept pace with the extraordinary growth in the stock of federal debt.
Meanwhile, central banks have stepped back from the enormous bond purchases of the previous decade, leaving more issuance to be absorbed by private investors who are generally more sensitive to price.
This is a better way to think about the supply story.
It is not simply that the world has “run out of savings.” More government and corporate borrowing can affect long rates through several channels. It can raise the equilibrium real rate. And, by increasing the amount of duration markets must absorb, it can increase the compensation investors demand to hold long-dated securities.
Recent Federal Reserve research finds exactly that: higher expected government debt can raise both the longer-run neutral real rate and the Treasury term premium.
In plain English: there is a lot more debt for investors to buy, and the buyers absorbing that debt increasingly need to be paid more to hold it. That can keep long-term rates high even if the Fed eventually cuts the overnight rate.
4. Why a Fed cut may not deliver much mortgage relief
This is where the distinction between the policy rate and the long end becomes tangible.
Different borrowing rates respond to different parts of the yield curve. Credit-card rates, for example, are closely connected to prime and short-term rates. Mortgages are different.
The 30-year mortgage rate is much more closely tied to longer-term Treasury and mortgage-backed-security yields than to the overnight federal-funds rate.
The divergence is already visible in the data. Mortgage rates have remained elevated even as the policy rate has moved lower from its previous peak.
Recent Dallas Fed research puts numbers around that relationship. Holding other factors constant, the mortgage rate has an estimated partial beta of about 85% to the 10-year Treasury yield, compared with less than 20% to the federal-funds rate.
That doesn’t mean the Fed is powerless over mortgages. A rate cut can pull down expectations for future short rates, and if the 10-year follows, mortgage rates should fall too.
But there is no one-for-one relationship.
If long Treasury yields remain elevated because investors expect higher real rates or continue to demand substantial compensation to absorb duration, mortgage rates can remain sticky even while the Fed eases.
That is the risk for housing.
The bottom line
The recent rise in Treasury yields is not an inflation-compensation story. Nor does the Fed’s own Kim–Wright model show a meaningful increase in the term premium since May.
But the level of the term premium remains elevated relative to the world investors became accustomed to after the financial crisis. And record government borrowing, new AI-related duration supply and a more price-sensitive investor base are reasons the long end may remain stubborn even if the Fed eventually cuts.
The bond market is doing the tightening calmly without a politically costly hike for a new Fed Chair appointed by a President who openly asked for lower rates. That outcome is also consistent with Warsh’s longstanding preference for a smaller Fed footprint.
That is the backdrop Warsh takes to Jackson Hole. Expect him to stick to that script on Friday.
Sources and methods: Treasury calculations use FRED DGS2, DGS5, DGS10, DGS30, DFII10 and T10YIE. The point-to-point 10-year decomposition begins May 22, 2026; curve and volatility comparisons use the matched five-day-average windows described in the accompanying analysis. Term premium is the Federal Reserve Board’s Kim–Wright THREEFYTP10 estimate; the longer-horizon comparison also references the New York Fed ACM model used in the earlier Treasury decomposition. Structural borrowing figures are from the OECD Global Debt Report 2026. Mechanism checks draw on 2026 Federal Reserve research on debt, the neutral rate and term premia, and Dallas Fed research on AI-related duration supply and mortgage-rate transmission. Foreign holdings are FDHBFIN as a share of GFDEBTN and should be interpreted as foreign and international investors’ holdings of federal debt relative to the total debt stock, not as foreign-official Treasury holdings.
