Can a Fed rate hike bring long-term yields down?

The Fed could raise its policy rate and still see ten-year Treasury yields fall. The key is what investors expect to happen next. A hike that reassures them about future inflation can make long bonds more attractive; one that signals a worsening problem can have the opposite effect. Ahead of the September 16 decision, the message about the years to come matters as much as the immediate rate change.

The decision is approaching in an already uncomfortable market. The Federal Open Market Committee meets on September 15–16. Its target range remains 3.50–3.75%, unchanged at the July 29 meeting. On Friday, September 11, Treasury’s published curve showed 4.63% at two years, 4.96% at ten years and 5.35% at thirty years. Those are nominal constant-maturity market yields, not coupons on a single bond and not live Monday quotes. ^{[1] [2] [9]}

The central bank sets the overnight price of money in a market where lending for longer already requires a higher return. It can raise the former without necessarily raising the latter. It can also leave the former alone and watch the latter climb.

The apparent contradiction begins to disappear once two questions are separated: what does money cost today, and what return makes the next decade’s risks worth taking?

Wednesday’s decision is only part of Wednesday’s information

The CME FedWatch table, timestamped September 14, 2026, at 06:00:36 Chicago time (11:00:36 UTC), assigns an 86.5% probability to a 25-basis-point hike at the September 16 meeting and 13.5% to an unchanged range. The hike outcome is a 3.75–4.00% target range. This is a dated market snapshot and can change before the decision. ^[3]

CME derives these probabilities from federal funds futures prices. Its methodology uses the monthly average effective overnight rate and policy moves in multiples of 25 basis points. The figures reflect market prices and modelling assumptions, rather than a survey of economists or a count of FOMC members’ intentions. ^[25]

In this observation, a hike is widely anticipated. Delivering it would therefore reveal less than an entirely unexpected tightening. Investors would also react to the accompanying explanation, the persistence of inflation and the likely sequence of decisions after September.

One basis point is one hundredth of a percentage point. A 25-basis-point increase would move the range to 3.75–4.00%. It would not mechanically move the ten-year yield from 4.96% to 5.21%. That calculation throws away the very information that matters: the market is pricing far more than Wednesday’s overnight rate.

A well-anticipated decision can still move markets sharply, but the surprise may lie in the policy path or in what the central bank appears to know about the economy. Reading the announcement as a single number risks missing most of the story.

An uneven inflation picture

The Bureau of Labor Statistics release published on September 11 describes prices in August 2026. Headline CPI rose 0.4% month on month, following 0.1% in July, on a seasonally adjusted basis. Its unadjusted twelve-month increase was 3.4%, unchanged from July. Gasoline rose 3.9% in the month and accounted for more than a third of the overall monthly increase. ^[4]

The underlying figures complicate the picture. CPI excluding food and energy increased 0.3% in the month, yet its annual rate slowed to 2.4%, from 2.5%. Monthly momentum strengthened, energy mattered substantially, and annual core inflation eased. Calling that an across-the-board inflation surge would make the narrative tidier than the evidence. ^[4]

The Fed also defines its 2% inflation objective using the PCE price index, produced by the Bureau of Economic Analysis, rather than CPI. The measures overlap but differ in coverage and weights. Comparing annual core CPI directly with a target for headline PCE is not enough to declare either victory or failure. ^[5]

Nor does the Fed produce oil. Raising interest rates cannot immediately restore fuel supplies after an adverse supply shock. Monetary policy works through spending, credit, investment and expectations. The challenge is to stop a sector-specific price increase from becoming a persistent process of repricing across the economy without needlessly damaging demand elsewhere. The July statement already identified supply shocks, including energy, as part of the inflation problem. ^[2]

That creates a genuine trade-off. A temporary disruption can justify patience. Evidence of lasting propagation can justify tighter policy. Neither one monthly print nor the oil price alone determines which description is right.

What a ten-year bond pays investors for

Buying a ten-year bond means comparing it with alternatives. An investor could instead keep rolling short-dated instruments. They could hold a different asset. They could demand a higher return for taking risks that currently seem poorly rewarded.

A standard representation, used by the Fed, divides a long yield into the expected average of nominal short rates over the horizon, plus a term premium. The decomposition is especially natural for a zero-coupon bond, which pays only at maturity, in a term-structure model. When discussing a quoted ten-year par Treasury yield, it is a useful approximation, not an exact accounting identity recovered by subtracting two unrelated published series. ^{[6] [7]}

The expected-short-rate component already contains a view of future inflation. Investors expecting higher inflation for longer may also expect the Fed to keep rates higher for longer. Adding expected inflation all over again to a path of nominal rates would count it twice.

The term premium concerns the risk of holding duration. Duration measures how sensitive a bond’s price is to changes in yields. Other things equal, payments farther in the future have present values that are more sensitive to the return investors require. A long Treasury can suffer a substantial market loss without any deterioration in the government’s ability to make its promised payments.

The holder’s money is not literally locked away for ten years. The bond can be sold. The unknown is the sale price. Even someone who holds to maturity may receive the promised dollars only to find that those dollars buy less than expected. Certainty about nominal repayment does not provide certainty about purchasing power.

The premium can respond to uncertainty about inflation and real rates, the amount of long-duration debt investors must absorb, and their willingness to carry that risk. It is not a fixed fee attached to a maturity. It can even be negative: investors may accept a lower expected return because a long bond provides valuable protection when other parts of their portfolio perform badly. Richard Clarida explicitly discussed that possibility in 2019. ^[6]

This is why the ten-year yield minus the current fed funds rate is not the term premium. The gap also contains the difference between today’s short rate and its expected future path. Subtracting the two-year yield from the ten-year yield does not isolate the premium either.

These distinctions matter because the same observed long yield can reflect very different combinations of expectations and risk pricing. A policy decision may change one component without moving the other, or move them in opposite directions.

Tighten now, and perhaps need less tightening later

Consider a hike that convinces investors the Fed will prevent inflation from becoming entrenched. Short rates over the next few months rise. But the rates thought necessary several years from now may fall because the danger of prolonged corrective tightening has diminished. At the same time, investors may require less compensation for uncertainty.

Those changes can outweigh the immediate increase. There is no guarantee that they will.

A deliberately stripped-down calculation helps. In an arithmetic average over ten years, an extra 25 basis points lasting for one year adds only 2.5 basis points to the average. This holds the other years constant and leaves aside the compounding details of real bond pricing. Its purpose is to show why a modest revision to the distant path can dominate a change at the front end.

A second example makes the mechanism explicit. Every number below is hypothetical. This is neither a September forecast nor an estimate of the actual yield decomposition.

Illustrative variable	Before	After
Policy rate used in the example	3.75%	4.00%
Expected average short rate over ten years	4.10%	4.00%
Term premium	0.90%	0.75%
Simplified long yield: average plus premium	5.00%	4.75%

The policy rate rises 25 basis points. The long yield falls 25. The explanation lies in the middle two rows: the policy action has altered expectations for the years ahead and the price of bearing uncertainty.

Now reverse the interpretation. Investors could view a hike as evidence that inflation is more persistent than previously understood, or as too little too late. Expected future rates and the term premium could rise together. Higher real rates or a larger supply of long-dated debt could also lift yields even when investors believe the Fed remains committed to price stability.

The defensible claim is conditional: raising short rates can lower long yields if the resulting change in expectations and risk pricing is large enough. It is not a strategy that mechanically converts tighter policy into cheaper government borrowing.

There is another timing issue. If investors already expect that credible hike, much of its reassuring effect may already be in prices. Delivering the expected decision might then do little, while failing to deliver it could cause a much larger adjustment. The policy move and the policy surprise are different variables.

Estimating the term premium

Bond prices are observable. The division between expectations and compensation for risk is not printed on the trade confirmation. It has to be estimated.

The chart below uses the Kim–Wright ten-year zero-coupon term-premium series published by the Fed and distributed through FRED. The Treasury lines are constant-maturity yields. The policy line is the upper bound of the Fed’s target range, not the effective overnight transaction rate. ^{[7] [8] [9] [10] [11] [12]}

From July 6 to September 11, the two-year yield rose from 4.13% to 4.63% and the ten-year from 4.48% to 4.96%, while the target upper bound stayed at 3.75%. Market rates plainly did not need a fresh rate increase to move. The comparison describes observed changes without attributing each daily move to a particular event. ^{[9] [10] [11] [12]}

The latest term-premium observation available in the consulted extract is September 4: 0.8892 percentage point, or approximately 89 basis points. The series stops there. It cannot establish how much of the September 10–11 selloff reflected a change in the term premium. ^[8]

The Fed describes the model as a staff research product, not an official statistical release. Data and methodology can be revised. Different specifications can assign more of a yield change to expected short rates and less to the premium, or the reverse. ^[7]

Nor should this zero-coupon premium be subtracted from a par Treasury yield and the residual labelled “the expected Fed rate.” The instruments do not match exactly, and the latest available dates differ. The chart places complementary indicators beside each other. It does not manufacture an exact decomposition of the newest market quote.

For an analysis about credibility, that limitation is central. An attractive graph can show that policy rates, market yields and a model estimate behave differently. It cannot prove which investors changed their minds, why they did so, or that a specific official’s words caused the change.

Two episodes of policy and market divergence

The phenomenon is not new. In February 2005, Alan Greenspan told Congress that long-term rates had declined even as the Fed had raised its target by 150 basis points. He discussed possible explanations and treated the movement as a puzzle. The episode establishes that the divergence can happen. It does not establish inflation credibility as its sole cause. ^[13]

The autumn of 2024 provides the opposite configuration. Between the day before September’s first cut and the end of December, the Fed lowered its target range from 5.25–5.50% to 4.25–4.50%. Over those comparison dates, the ten-year yield rose from 3.65% to 4.58%. The two-year yield rose as well. ^{[14] [15] [16]}

A cut today does not stop investors from anticipating fewer cuts tomorrow, a higher equilibrium rate, or more compensation for risk. Conversely, falling long yields after a hike can reflect recession fears as much as restored confidence in price stability.

The historical evidence rules out a mechanical “Fed cuts, bonds rally” rule. It does not justify replacing it with the opposite rule. Separating the drivers requires prices, expectations, economic evidence and estimates of risk. The curve is an information source, not a credibility detector.

Even the term-premium change shown in the historical comparison is a current-vintage model estimate, rather than a reconstruction of everything investors knew in real time in 2024. That distinction is essential when using historical data to tell a story about expectations.

How the Fed reads long-term yields

The next temptation is to say that the bond market is forcing the Fed’s hand. Markets shape financial conditions and provide information. But a higher yield does not, by itself, identify the decision best suited to maximum employment and price stability.

At Jackson Hole on August 28, Kevin Warsh argued that progress against inflation remained insufficient. Yet he also described medium-term inflation expectations as broadly stable and stressed the importance of preserving that anchor. His speech set a standard for policy rather than announcing a September decision. Declaring that expectations have already become visibly unanchored would go beyond that dated assessment and the data presented here. ^[17]

Higher long-term yields can also do some of the cooling themselves. Borrowing becomes more expensive for certain households and firms. Some projects no longer clear their financing hurdle. Housing can weaken. In October 2023, Jerome Powell noted that rising long yields had contributed to tighter financial conditions and that persistent changes could affect the appropriate path of monetary policy. ^[18]

The source of the move is crucial. If yields rise because demand is strong and investors expect a firmer Fed, withholding the anticipated tightening can undermine the assumptions already embedded in prices. If a higher risk premium is independently tightening financing conditions, an additional hike can overdo the restraint. If long yields fall because recession looks more likely, the decline is not necessarily good news.

The committee must judge where the shock came from, how long it is likely to last, how quickly it transmits, and how differently it affects individual sectors. The market presents a question. It does not always supply the answer.

Bessent still has a government to finance

The Treasury operates in the same market with a different task: funding the state over time.

Its estimates published on August 3 called for $739 billion of privately held net marketable borrowing in the third quarter of 2026, followed by $628 billion in the fourth. The estimates assumed end-quarter cash balances of $950 billion and $850 billion respectively. They are financing forecasts, not spending newly authorised in September. ^[19]

Net is doing important work. Treasury also has to replace maturing securities, so gross issuance exceeds the net financing requirement. Net marketable borrowing is not identical to the budget deficit because cash balances and other financing operations matter. And the totals span maturities: describing $739 billion as a block of new ten- or thirty-year bonds would be wrong. ^{[19] [20]}

Higher required yields make subsequent borrowing more expensive, other things equal. But the federal interest bill reprices over time. An existing fixed-coupon bond keeps its coupon even if its market value falls. Treasury does not immediately increase the payment to its holder. Higher costs mainly arrive through new borrowing and refinancing, with faster exposure on short-dated and floating-rate instruments.

Multiplying the entire debt stock by a change in the ten-year yield would therefore produce an impressive number, not the current year’s interest bill. A serious calculation needs the maturity schedule, new financing, the applicable rate at each point, and the treatment of inflation-linked securities.

Future debt supply can nevertheless affect today’s yields. Investors price the risk they expect to absorb, not just the bonds being sold this afternoon. Yet a large borrowing requirement is not evidence that nobody wants Treasuries. It may mean buyers require a higher return. That is less dramatic than a buyers’ strike, but economically important.

What Treasury buybacks do

On August 19, Treasury announced larger liquidity-support buybacks in the nominal 10–20-year and 20–30-year sectors. Effective September 9, the $2 billion operation cap would become at least $4 billion, through the November 4 quarterly refunding. The announcement concerns maximum operation size, not a guaranteed minimum amount of purchases. ^[21]

The schedule updated on September 9 went further for one transaction: the September 10 operation in the 10–20-year bucket had a $6 billion ceiling, settling September 11. Its minimum remained zero. A published operation limit must not be mistaken for an executed purchase amount. ^[22]

Why would a government that needs to borrow also buy back its old bonds? Because the quality of the market matters alongside the amount it must raise.

The most recently issued securities, known as on-the-run bonds, attract substantial trading activity. Older off-the-run issues can be harder to sell in size and more costly to intermediate. A regular opportunity to sell those securities can help the market function. Treasury’s FAQ describes the current liquidity-support programme as a tool for routine market functioning, not a mechanism designed to counter acute episodes of stress. ^[23]

Consider a fictional operation. A dealer holds older bonds that are difficult to distribute. It offers Treasury a quantity at a price. Treasury can accept offers it finds attractive within the cap, or buy less. The dealer recovers balance-sheet capacity, and accepted bonds are retired at settlement. Transaction frictions may ease. None of this fixes the yield investors will require at the next auction.

The distinction from quantitative easing, or QE, is fundamental. In a monetary purchase programme, the Fed acquires assets for its own balance sheet as part of monetary policy. Here, Treasury repurchases its debt using cash resources and financing. The New York Fed may execute the trade, but as Treasury’s fiscal agent, not as a monetary buyer for its own portfolio. ^{[23] [24]}

Treasury’s borrowing release explicitly states that buybacks are not expected to materially change privately held net marketable borrowing, because new issuance replaces repurchased securities. The operation reorganises debt and its market. It does not provide a lasting way to fund deficits by cancelling the bill. ^[19]

The refinancing mix still matters. Replacing long bonds with short bills can reduce the duration investors must absorb, while exposing the government to more frequent refinancing. Funding the buyback with other long bonds has a different effect. Settlement flows can also temporarily move Treasury cash and banking-system reserves. None of those qualifications turns the transaction automatically into QE.

A yield cap would require a commitment to defend a price or interest rate. The programme in these documents instead provides capped, discretionary purchases in selected securities. It can improve liquidity, affect relative prices and reduce intermediation costs without preventing long yields from rising because of inflation, real rates or debt supply.

Bessent can therefore seek a better-functioning bond market while Warsh considers a more expensive overnight rate. Those policies need not conflict. Improving the mechanics of trading does not require subsidising the price at which everyone hopes to borrow.

AI joins the queue for capital

The stakes extend beyond the government’s interest bill. Treasuries help benchmark private borrowing. A fixed-rate corporate borrower typically faces a maturity-matched reference yield plus a spread compensating investors for credit risk and other features. A floating-rate loan depends more directly on its short-rate reference and reset schedule.

The build-out of AI infrastructure makes that transmission unusually visible. A January 7, 2026 BIS Bulletin argued that the scale of projected investment would require more financing to migrate from operating cash flow to debt, with private credit taking a growing role. Its authors also described macrofinancial risks as moderate at the time while emphasising dependence on strong future earnings. That dated assessment is not a September crisis forecast. ^[26]

Simple arithmetic illustrates the exposure. For $10 billion of financing fully subject to a 0.50-percentage-point increase, the annualised cost rises by $50 million. This is a fictional example, before hedges, amortisation, fees and taxes. It identifies no particular company’s actual exposure.

Credit spreads can also move against the Treasury benchmark. If the ten-year yield falls because recession risk has risen, a corporate borrower’s default risk may be repriced upward. Its all-in funding cost need not decline. Cash-rich companies, leveraged infrastructure projects and labs dependent on new funding rounds therefore face different versions of the interest-rate problem.

It would be equally misleading to add government financing and AI investment as though both were competing for a sealed box containing a fixed stock of dollars. Saving, credit creation, international flows and prices adjust. The relevant competition is over the terms on which someone is willing to take the risk.

For long-lived projects, the timing matters as well as the level. Short financing against assets that will earn revenue for years exposes a borrower to repeated repricing. Locking in long financing removes part of that uncertainty but can be expensive at today’s yields. A successful Fed communication cannot eliminate either trade-off.

Four possible readings of the same press conference

The September 16 outcome will not be interpretable from “plus 25” or “unchanged” alone. Four configurations remain possible.

A hike reassures. Near-term rates rise, but investors lower the distant expected path or the premium for uncertainty. Long yields fall. This would illustrate the paradox, without promising easier credit for every private borrower.

A hike fails to reassure. The move seems late, inadequate, or informative about more persistent inflation. Long yields rise too. Greater debt supply or higher real rates could generate the same outcome without any collapse in the Fed’s credibility.

A hold is understood as a response to economic weakness. Investors expect less inflation pressure and lower future policy rates. Long yields can decline. Leaving policy unchanged would not then amount to surrender.

A hold creates doubt. Investors infer more tolerance for inflation or less ability to control it. They demand a higher return for long lending. The Fed has left its rate alone, yet financing conditions tighten anyway.

These are conditional scenarios, with no assigned probabilities. Distinguishing them after the decision requires the expected short-rate path, real yields, inflation-expectation indicators, credit spreads and, when published, term-premium estimates. Even breakeven inflation, the yield gap between nominal and inflation-linked bonds, contains risk and liquidity components. It is not a pure survey of future prices. ^{[27] [7]}

An immediate fall in the ten-year yield would not, by itself, establish that credibility had been restored. The announcement effect, other incoming information and the persistence of the move would all need examination. A clean causal claim requires more than a before-and-after chart.

The answer to the headline is therefore yes, it is possible. Higher short rates can lower long yields by changing expectations and the risks investors are willing to carry. They cannot give the Fed unrestricted control over the ten-year yield or erase Treasury’s financing constraints.

Wednesday’s press conference must help investors understand what comes next. Money can become more expensive overnight while investors require less to lend for a decade. That only works when the tightening makes the future look sufficiently less uncertain.

Further reading

Our guide to the US Treasury market explains auctions, the yield curve and liquidity risks. Pacing the AI frontier explores infrastructure funding and the allocation of financial risk.

Sources

1. FOMC calendars, Federal Reserve.
2. FOMC statement, July 29, 2026, Federal Reserve (2026-07-29).
3. CME FedWatch, CME Group (2026-09-14).
4. Consumer Price Index, August 2026, Bureau of Labor Statistics (2026-09-11).
5. Why does the Federal Reserve aim for inflation of 2 percent?, Federal Reserve.
6. Remarks on the term premium and long-term interest rates, Federal Reserve, Richard Clarida (2019-11-12).
7. Three-factor nominal term structure model, Federal Reserve.
8. THREEFYTP10: Term Premium on a 10 Year Zero Coupon Bond, Federal Reserve / FRED.
9. Daily Treasury Par Yield Curve Rates, 2026, U.S. Treasury.
10. DGS2: 2-Year Treasury Constant Maturity Rate, Federal Reserve / FRED.
11. DGS10: 10-Year Treasury Constant Maturity Rate, Federal Reserve / FRED.
12. FOMC statement, June 17, 2026, Federal Reserve (2026-06-17).
13. Monetary policy testimony to Congress, Federal Reserve, Alan Greenspan (2005-02-16).
14. FOMC statement, September 18, 2024, Federal Reserve (2024-09-18).
15. FOMC statement, December 18, 2024, Federal Reserve (2024-12-18).
16. Daily Treasury Par Yield Curve Rates, 2024, U.S. Treasury.
17. Keynote remarks at the 2026 Jackson Hole Economic Policy Symposium, Federal Reserve, Kevin Warsh (2026-08-28).
18. Remarks at the Economic Club of New York, Federal Reserve, Jerome Powell (2023-10-19).
19. Treasury Announces Marketable Borrowing Estimates, U.S. Treasury (2026-08-03).
20. Sources and Uses of Financing, August 2026, U.S. Treasury (2026-08-03).
21. Increased long-end liquidity support buybacks, U.S. Treasury (2026-08-19).
22. Tentative Buyback Schedule, Q3 2026, U.S. Treasury (2026-09-09).
23. Buyback FAQs, TreasuryDirect.
24. Treasury Debt Auctions and Buybacks as Fiscal Agent, Federal Reserve Bank of New York.
25. Understanding the CME Group FedWatch Tool Methodology, CME Group (2023).
26. Financing the AI boom: from cash flows to debt, BIS, Aldasoro, Doerr and Rees (2026-01-07).
27. Tips from TIPS: Update and Discussions, Federal Reserve, Kim, Walsh and Wei (2019-05-21).