On September 16, the Federal Reserve raised its federal funds target range by 25 basis points to 3.75%–4.00%. On October 2, the Treasury’s published 10-year par yield was 5.28%. A 10-year bond prices a longer horizon: the path of future short rates and the compensation holders require for interest rate risk.1 2
Our earlier quarter-end commentary recorded rising long yields and a steeper curve. The New York Fed’s ACM model separates the yield into two components. In the same data, changing the starting date changes which component has the larger net increase.
ACM divides a fitted yield into an expectations component and a term premium. The first estimates the future short-rate path over the relevant horizon; the second measures compensation for the interest rate risk of a longer fixed-rate bond. Both come from a model of the same yield curve. The market does not quote them separately.3
From September’s first observation to the latest observation in the workbook, the fitted 10-year yield increased by 35.0 basis points. The model’s expectations component rose by 22.4 basis points and its term premium by 12.6 basis points.
| Same ACM daily model | Change in fitted 10-year zero-coupon yield | Change in expectations component | Change in term premium |
|---|---|---|---|
| September 1–October 1, 2026 | +35.0 bp | +22.4 bp | +12.6 bp |
| September 17–October 1, 2026 | +31.3 bp | +1.5 bp | +29.8 bp |
Source: New York Fed, ACM Daily worksheet. Endpoint differences are calculated at source precision, then rounded to 0.1 basis point. The first row includes October 1 and is not a calendar-September month-end change. The second window is nested within the first. They share an endpoint and cannot be added. These are net endpoint changes; they show neither continuous increases within each window nor an identified switch in mechanism.3
The second row begins on September 17, when the new policy range took effect, and shows a larger net increase in the term premium. This starting date checks a shorter window tied to the policy calendar. Economic information, demand and other policy developments during the period remain in the result; the hike’s separate effect has not been isolated. Applying this allocation to the whole rise would omit the earlier movement in expectations.4
On October 1, ACM’s fitted 10-year yield was approximately 5.217%, comprising an expectations component of 4.312% and a term premium of 0.905%. These values can be added within the same day’s fitted zero-coupon curve. The Treasury’s opening 5.28% is a par yield from the following day. Subtracting the ACM premium would leave a mismatch of dates and curve conventions, not a comparable short-rate expectation.3 2
The New York Fed states that these data are not official estimates of the New York Fed, the Federal Reserve System or the FOMC. Federal Reserve staff research comparing ACM with Kim–Wright explains one source of unstable allocations. A model estimated from yields alone may assume too rapid a return to the long-run mean, making long-horizon short-rate expectations look too stable. More of the yield’s movement then falls into the term premium. Including survey expectations can change that division.5
Surveys have their own limits. The average respondent need not represent the marginal participants setting bond prices, and the historical sample can affect estimates. This mechanism does not establish that the current ACM results are misallocated. It shows why the proportions depend on estimation choices. No other current same-date model or survey decomposition was obtained here to confirm them independently.
Fiscal conditions can enter the curve through more than one channel. They may change risk compensation, or feed into expectations through future economic conditions and policy. ACM’s two components do not separately measure the contributions of fiscal conditions, inflation or growth. Reading its premium as a fiscal shock would add an inference the model has not identified.
On August 3, Treasury projected third-quarter privately held net marketable borrowing of $739 billion, $68 billion above its May estimate, assuming a quarter-end cash balance of $950 billion. It projected $628 billion for the fourth quarter. These numbers were forecasts at publication, rather than completed borrowing totals.6
The third-quarter revision primarily reflected lower projected net cash flows, partly offset by a higher-than-assumed opening cash balance. Refinancing maturing securities, fiscal cash flows and changes in the cash balance each affect borrowing figures. Net borrowing includes cash-management assumptions and cannot be read directly as the quarterly deficit; gross issuance also includes refinancing maturing debt.
The quarterly refunding statement two days later supplied the maturity plans. Treasury then expected to maintain nominal coupon and floating-rate note auction sizes for at least the next several quarters. Seasonal or unexpected variations over the next quarter would be met through regular bill auction sizes or cash management bills.7
The August statement’s 10-year schedule called for an August new issue of $42 billion and September and October reopenings of $39 billion each. All three were anticipated sizes in that statement, following the previous quarter’s new-issue and reopening pattern; they were not final auction results. The monthly differences in that pattern do not show continuously expanding long-coupon supply.7
A higher net borrowing forecast can therefore coexist with unchanged individual long-coupon auction sizes. Bills and long fixed-rate bonds expose the market to different horizons of interest rate risk. Net issuance across maturities, the redemption structure and remaining terms are needed to measure the duration Treasury adds. The auction-size table does not capture the full change in risk the market absorbs.
The longer fiscal path can still enter expectations ahead of issuance. Markets can revalue debt and financing arrangements years into the future, even with this quarter’s auction sizes unchanged. Demand also sets the price: which maturities and how much risk can be absorbed help determine the rates associated with a large funding need.
On August 19, Treasury also announced larger liquidity-support buybacks in long-dated nominal sectors, effective September 9. Their purpose was to support older securities’ liquidity; the announcement did not measure an achieved effect on yields. The borrowing announcement explains that new issuance replaces bought-back securities and that buybacks are not expected to significantly change privately held net marketable borrowing.8 6
Issuing new securities to buy back old ones changes the composition, while Federal Reserve purchases through its own balance sheet use a different funding chain. Comparing the maturities of the bought-back and newly issued debt is one necessary step in assessing a net change in duration. The buyback announcement alone does not make the operation quantitative easing.
The overnight policy range sets today’s policy boundaries for funding, while the expectations component covers short rates throughout a bond’s horizon. The odds of a move at the next meeting cannot replace that path. Holders of fixed long-term cash flows also face the risk of rates exceeding prior expectations: higher future short rates can lower existing bonds’ market value. The term premium is the model’s estimate of compensation for this risk.
Whether a bond provides protection in a bad economy affects the compensation required. Weaker demand can lower inflation and policy rates, allowing long fixed-rate bonds to serve as a hedge. An adverse supply shock can weaken growth while raising prices. If the central bank raises rates to contain inflation, long-bond prices may keep falling as the economy deteriorates. The same maturity offers different protection in those two states.
In a February 2026 study of far-forward rates, Federal Reserve staff linked risks of future adverse supply shocks and debt sustainability to rising far-forward real risk compensation. That explanation uses a particular historical sample and pricing framework. Its far-forward real risk compensation also differs from the ACM total 10-year term premium used here. It offers a mechanism for changing hedge value, without establishing the cause of September’s yield moves.9
Growth, policy responses, risk preferences and bond supply and demand can change the term premium; inflation expectations can also affect the future short-rate path. Higher expected growth or a higher long-run neutral rate can raise long yields as well. In the main September 1–October 1 window, the larger rise in ACM’s expectations component leaves room for those channels. The premium cannot account for the whole long-end story.
Fiscal supply’s contribution to this rise remains unmeasured. The issuance documents describe institutional plans, while ACM provides one decomposition of the curve. The available evidence has not connected them through an identified causal relationship. A larger net change in the premium over the shorter window is still insufficient to establish that fiscal supply dominated the rise.
Research as of October 4, 2026. The latest Treasury curve observation is October 2. The ACM workbook download completed at 12:13:37 UTC on October 4; its daily worksheet ends on October 1. The results refer to this downloaded data version. Changes use source precision, with table values rounded to 0.1 basis point and yield levels to 0.001%. No missing observation was filled from another day or curve convention. Quarterly financing figures are cited as forecasts at their publication dates.
This article represents KSINQ’s personal research and opinion for informational purposes only and does not constitute investment advice. Model decompositions do not guarantee future yields or asset performance. Past performance does not indicate future results.
Federal Reserve, September 16, 2026 FOMC statement, policy action paragraph. ↩
U.S. Treasury, Daily Treasury Par Yield Curve Rates, October 2, 2026 row, 10-year column and methodology notes. This is a quotation-based interpolated constant-maturity par rate, rather than the yield of a particular executed trade. ↩ ↩2
New York Fed, ACM source workbook and model overview. Uses ACMY10, ACMRNY10 and ACMTP10 from ACM Daily, in percent. Endpoint differences in percent are multiplied by 100 to obtain basis points. The components are model estimates, not independently observable market quotations. Model authors: Tobias Adrian, Richard K. Crump and Emanuel Moench. © 2026 Federal Reserve Bank of New York. Content from the New York Fed subject to the Terms of Use at newyorkfed.org. Endpoint calculations, rounding and this article’s interpretation were prepared by KSINQ, and were neither produced nor endorsed by the New York Fed. The source content and this table’s numerical adaptations of it remain subject to the same New York Fed permissions and conditions; KSINQ’s general usage restrictions do not impose stricter restrictions on either. ↩ ↩2 ↩3
Federal Reserve, September 16 implementation note, directive effective September 17. ↩
Federal Reserve staff, Robustness of long-maturity term premium estimates, April 3, 2017; discussion of samples, survey information and estimation uncertainty. ↩
U.S. Treasury, August 3 marketable borrowing estimates, third- and fourth-quarter estimates and footnote 1. ↩ ↩2
U.S. Treasury, August 5 quarterly refunding statement, nominal coupon and FRN financing, auction-size table and bill issuance sections. ↩ ↩2
U.S. Treasury, August 19 long-end liquidity-support buyback announcement, implementation and purpose paragraphs. ↩
Covitz and Engstrom, Federal Reserve staff, Why have far-forward nominal Treasury rates increased so much in the past few years?, published February 12, 2026, with a figure-source correction on July 15. This is staff research expressing the authors’ views. ↩