US Treasury yields are repricing as Fed policy, fiscal deficits and debt issuance reshape bond markets and challenge traditional portfolio hedges.
Two term-structure models published by institutions within the Federal Reserve System give different answers to a basic question about 2026. Between the last trading day of 2025 and November 6, 2026, the 10-year term premium estimated by the Federal Reserve Bank of New York (New York Fed) with the Adrian-Crump-Moench (ACM) model fell by about 0.11 percentage point. The Kim-Wright estimate maintained by the Federal Reserve Board rose by about 0.30 point over the same days. The term premium is the extra return investors demand for holding a long-term bond rather than repeatedly investing in short-term bills, and both models were estimating term premia from the same Treasury market, where the 10-year Treasury yield climbed from 4.18 percent on the last trading day of 2025 to 4.86 percent over that stretch before reaching 5.15 percent on November 13.
That disagreement offers a way into a widely held account of the bond market, which says persistent deficits, an aging population and a shrinking central-bank balance sheet are keeping long-dated yields high even as policy rates come down. Tested against the decomposition data, the account describes late 2024 well but fits 2026 less comfortably. The term premium did reset upward, by roughly a full percentage point, and it did so while the Federal Reserve (Fed) was cutting rates. In the ACM model, this year's climb in long yields has come mainly from a higher expected path for policy rates after the Fed resumed tightening in September. Fiscal pressure is real and growing, but it is showing up more clearly in how the Treasury finances itself than in any sign that the Fed has subordinated its inflation goal to federal borrowing needs, the condition usually called fiscal dominance. For investors, the more immediate consequence lies in diversification, because the correlation between stock and bond returns has turned positive again and weakened the hedge on which a portfolio of 60 percent stocks and 40 percent bonds depends.
Separating those forces requires splitting a long-term yield into two parts that cannot be observed directly. For a bond of maturity , term-structure models rest on the identity
: yield on an -period bond; first term: average expected short-term rate over the life of the bond; : term premium.
in which the first term is the average short-term rate investors expect over the life of the bond and the second is the term premium. The ACM model builds the expected path from five principal components of Treasury yields in a three-step regression, while the Kim-Wright model adds survey forecasts of the three-month bill rate to pin that path down. That difference is important because the term premium is a model estimate, not a traded price, and the Kim-Wright estimate is a separate model-based measure rather than a component of the ACM decomposition. One possible explanation for this year's divergence is that survey forecasts adjust more slowly than market pricing. If so, a model that incorporates those forecasts could assign more of a sudden repricing towards higher rates to the term premium. The difference between the estimates, however, does not by itself establish that this is what happened.
The table below compares six dates since the Fed began cutting in September 2024, adding yields on Treasury Inflation-Protected Securities (TIPS) and the breakeven inflation rate, the gap between nominal and TIPS yields.
Date | Fed funds target (upper bound) | 10-year yield (ACM fitted) | Expected average short rate (ACM) | Term premium (ACM) | Term premium (Kim-Wright) | 10-year TIPS yield | 10-year breakeven inflation |
|---|---|---|---|---|---|---|---|
September 17, 2024 | 5.50 | 3.62 | 3.90 | −0.28 | 0.06 | 1.53 | 2.12 |
January 13, 2025 | 4.50 | 4.83 | 4.19 | 0.65 | 0.81 | 2.34 | 2.45 |
December 31, 2025 | 3.75 | 4.25 | 3.50 | 0.75 | 0.57 | 1.93 | 2.25 |
September 15, 2026 | 3.75 | 4.98 | 4.27 | 0.71 | 0.96 | 2.62 | 2.38 |
October 7, 2026 | 4.00 | 5.27 | 4.29 | 0.99 | 1.09 | 2.92 | 2.36 |
November 13, 2026 | 4.00 | 5.11 | 4.39 | 0.72 | 0.87* | 2.91 | 2.24 |
The episode that best fits the popular account came first. In the four months from September 17, 2024, the day before the first cut of that cycle, to January 13, 2025, the Fed lowered the upper bound of its target range from 5.50 percent to 4.50 percent, yet the ACM 10-year yield rose 1.21 percentage points. Writing the identity in changes, with for the expected average short rate,
: change in the 10-year yield; : change in the expected average short rate; : change in the term premium.
so about three-quarters of the increase came from the premium. The Kim-Wright estimate, up 0.76 point, agrees on direction and rough size. Inflation compensation contributed as well, with the 10-year breakeven rate up 0.33 percentage point and the real TIPS yield up 0.81 point. These are market-based measures, not components of the ACM estimate. Together, they track the increase in the observed nominal Treasury yield, which differs from the ACM-fitted yield used in the decomposition above.
What happened in those four months has proved durable, and durability is the part of the "structurally elevated" claim that the evidence supports. The daily ACM term-premium estimate averaged −0.31 percent from 2014 through 2023, years of large central-bank bond holdings and low, stable inflation. It has averaged 0.70 percent from January 1 through November 13, 2026, and on October 7 it reached 0.99 percent, its highest reading since June 2014. The chart below traces a swing of about one percentage point that has persisted through a pause, three more cuts in late 2025 and a hike in 2026.
Chart 1: ACM decomposition of the 10-year Treasury yield, daily, January 2014 to November 13, 2026. Source: Federal Reserve Bank of New York, ACM Treasury term premia; Federal Reserve via FRED (DFEDTARU).
This year's rise has a different engine, and the Fed's own documents explain much of it. After the cuts of late 2025 took the target range to 3.50–3.75 percent, the Federal Open Market Committee (FOMC) voted 12–0 on September 16, 2026 to raise it to 3.75–4 percent. Staff estimated in the meeting minutes that inflation measured by the personal consumption expenditures (PCE) price index had edged up to 3.8 percent in August, and the Summary of Economic Projections (SEP) raised the median projection for the policy rate at the end of 2027 to 4.1 percent, from 3.6 percent in June. At the October meeting, two members dissented in favor of a further increase. Applied to 2026 so far, the ACM identity gives
which means, in the ACM model, the higher expected path of short rates accounts for almost the entire increase in the ACM-fitted 10-year yield this year, while the term premium has changed little. The Kim-Wright model tells a different story, with its estimated term premium rising by about 0.30 percentage point between December 31, 2025 and November 6, 2026. The difference matters: the two models agree that long-term yields have risen, but disagree on how much of that increase reflects the term premium.
The shape of the curve points in the same direction. The 2-year yield has risen 1.41 percentage points and the 30-year yield 0.63 point, narrowing the gap between them from 1.37 to 0.59 point. A sell-off driven mainly by an excess supply of long bonds would normally put more upward pressure on long yields and steepen the curve, although other forces can offset that effect.
Set side by side, the two climbs look alike on a yield chart and almost opposite in their composition.
Chart 2: Change in the ACM-fitted 10-year Treasury yield and its model components, in basis points (one basis point is 0.01 percentage point). Source: Federal Reserve Bank of New York, ACM term premia; Federal Reserve Board, Kim-Wright model via FRED; calculations by IndraStra Global Research Team.
The split between real yields and inflation compensation offers one way to assess the fiscal-dominance argument. If investors increasingly believed that the Fed would tolerate higher inflation to keep federal borrowing cheap, market-based inflation compensation might rise, although breakeven rates also reflect inflation risk premia and differences in liquidity between nominal Treasuries and TIPS. Instead, the 10-year breakeven rate stood at 2.24 percent on November 13, a shade below its level at the end of 2025, while the real 10-year TIPS yield rose almost a full point to 2.91 percent. The Treasury Borrowing Advisory Committee (TBAC), the panel of dealers and investors that advises the Treasury on debt management, reached a compatible conclusion on November 3, attributing higher yields primarily to growth, policy expectations and energy prices and observing that the term premium had risen less than yields this year.
Markets have not ignored the politics around the Fed, but they are pricing in an institution that has so far acted against the White House's stated preference for lower rates. Kevin Warsh, nominated by President Donald Trump, took office as chairman on May 22, 2026. Four months later he presided over the hike, after which Trump described a call in which he told Warsh "you might as well vote with the board because it's not going to matter". On June 29, the Supreme Court had denied the administration's application to remove Governor Lisa Cook while her lawsuit proceeds, ruling 5–4 that the government had not shown it was likely to prevail, and Cook was one of the two officials who voted for a further increase in October. Pressure of this kind would warrant the same scrutiny under any administration. On the record so far, the Fed has continued to raise rates in response to elevated inflation despite political pressure for lower rates. That is evidence against an immediate subordination of monetary policy to federal borrowing needs, although the longer-term fiscal-dominance question remains open.
The balance-sheet strand of that account has meanwhile fallen out of date. The FOMC decided in October 2025 to conclude the reduction of its securities holdings on December 1, and in December it began purchasing shorter-term Treasury securities to keep reserves ample. By the week ended November 11, 2026, the Fed's bill holdings had risen by US$382 billion from a year earlier to US$578 billion, while its mortgage-backed securities (MBS) fell by US$186 billion as principal ran off and was reinvested in bills. The Fed's securities portfolio is now growing while shifting towards shorter-term Treasury holdings. As mortgage principal is repaid, the Fed's holdings of mortgage securities decline, leaving a larger share of mortgage-market duration outside its portfolio. Its separate purchases of shorter-term Treasury securities are shifting the composition of its assets towards bills. At the same time, the Fed is becoming a steady buyer at the short end of the Treasury market. These purchases are intended to maintain ample reserves in the banking system, not to finance the federal deficit.
Fiscal pressure is unmistakable in the arithmetic of the budget itself. The Congressional Budget Office (CBO) projected in February a deficit of US$1.9 trillion, or 5.8 percent of gross domestic product (GDP), for fiscal year 2026, which ran from October 2025 to September 2026, with debt held by the public rising from 101 percent of GDP to 120 percent by 2036 and net interest climbing from 3.3 percent to 4.6 percent of GDP. The outturn was close to that estimate, as the Monthly Treasury Statement records a fiscal 2026 deficit of US$1.99 trillion against US$1.78 trillion a year earlier. Demographics enter mainly through this channel, since CBO attributes the rise in outlays to greater spending on Social Security and Medicare alongside interest; this analysis does not attempt to identify a separate, direct effect of aging on the term premium, and none of its figures depend on one. The same baseline projected that the quarterly average 10-year yield would rise from 4.1 percent in the fourth quarter of 2025 to 4.3 percent in the fourth quarter of 2027, reflecting higher term premiums. It also assumed that the Fed would cut rates in 2026. The observed 10-year market yield on November 13 was already about 0.85 percentage point above that projected level. If the gap persists, federal interest costs could end up higher than the baseline.
Faced with those numbers, the Treasury has chosen to rely more heavily on short-term bills rather than increase the size of its long-term bond auctions. Its November 4 refunding statement kept every nominal coupon auction size unchanged, including US$42 billion for the 10-year note and US$25 billion for the 30-year bond in refunding months, and said the department expects to maintain those sizes "for the next several quarters" while it monitors Fed bill purchases and growing private demand for bills. It also plans to buy back up to US$70 billion of older securities this quarter for liquidity support. The chart below shows the result in the composition of the debt: bills made up 23.0 percent of marketable debt held by the public at the end of October, up from 14.5 percent at the end of 2019, according to the Monthly Statement of the Public Debt.
Chart 3: Bills as a share of marketable Treasury debt held by the public, month-end, January 2019 to October 2026. Source: US Treasury, Monthly Statement of the Public Debt (FiscalData); Quarterly Refunding Statement, November 4, 2026.
That choice blunts the mechanism on which the supply story relies. The Treasury has so far held coupon auction sizes steady rather than add more long-term debt to the market. The term premium's plateau in the ACM model is consistent with that restraint, although the data cannot show how much of the movement is due to issuance policy rather than the other forces moving yields. The restraint also has a horizon. At the TBAC meeting, Treasury staff reported that the median primary-dealer forecast implies a US$2.14 trillion funding shortfall over fiscal 2028 and 2029 at current coupon sizes and bill supply. The committee judged that larger coupon auctions will be warranted later in fiscal 2027, recorded "considerable debate" over how much to add at the long end, and noted that money market fund assets have leveled off and that the bill market "has started to show some signs of price sensitivity." A heavy bill share carries a second cost in a tightening cycle. Because bills mature and must be refinanced frequently, higher policy rates feed into federal interest costs faster than they would with a longer-dated debt stock.
A serious objection to reading the bill strategy as a warning is that demand for Treasuries has held up. The TBAC report found no evidence of a persistent deterioration in auction demand, described foreign demand for US fixed income as having accelerated this year and argued that households and private pensions still hold relatively little fixed income. That view deserves weight, and on current data it is correct. The holdings data below, from the Treasury International Capital (TIC) system, compare August 2026 with a year earlier and show why the test still lies ahead.
Holder | August 2025 (US$ billion) | August 2026 (US$ billion) | Change |
|---|---|---|---|
All foreign holders | 9,249.1 | 9,331.7 | +0.9% |
Of which foreign official | 3,919.8 | 3,755.6 | −4.2% |
Japan | 1,183.9 | 1,068.3 | −9.8% |
China, mainland | 699.7 | 590.7 | −15.6% |
United Kingdom | 901.3 | 1,009.5 | +12.0% |
Marketable Treasury debt held by the public | 29,388 | 31,807 | +8.2% |
Foreign holdings as a share of that debt | 31.5% | 29.3% | −2.2 percentage points |
Sources: US Treasury, TIC Major Foreign Holders of Treasury Securities (latest data for August 2026); Monthly Statement of the Public Debt. Country figures reflect where securities are held in custody, not necessarily the beneficial owner. The share is approximate because TIC totals include a small amount of nonmarketable debt. Calculations: IndraStra Global Research Team.
Measured against the growth of the debt itself, the foreign figures show a market leaning on new buyers outside the official foreign sector. Foreign holdings grew 0.9 percent while marketable debt grew 8.2 percent, and holdings of foreign official institutions, mostly central banks and other public bodies, fell by US$164 billion. Private foreign buyers and domestic investors appear to have absorbed much of the additional Treasury debt, helped by higher yields. That is how a market is meant to clear. The unanswered question is how much higher yields will have to go to attract buyers for the additional bonds that the TBAC expects the Treasury will eventually need to sell.
Whether a repricing comes from growth, inflation or the term premium also decides whether bonds still protect a stock portfolio. The relevant measure is the correlation between stock returns and bond returns ,
where monthly S&P 500 price returns stand in for stocks and the 10-year bond return is approximated from the constant-maturity yield with duration . The calculation is intended to show how the changing correlation affects portfolio volatility, not to reproduce the performance of an actual 60/40 fund. The rolling 36-month correlation of monthly returns stood at +0.34 in October 2026, compared with an average of −0.32 across rolling 36-month windows during 2000–2021. A separate calculation using daily returns gives a correlation of +0.26 for 2026 so far. That pattern matches a Federal Reserve staff note published in October, which found that the hedging value of government bonds depends on inflation pressures and the stability of inflation expectations and that adverse supply shocks arriving amid above-target inflation can push stocks and bonds down together. An earlier analysis by the Bank for International Settlements (BIS) tied the 2021 sign switch to high and volatile inflation and argued that a positive correlation leads investors to require a higher term premium. The long history in the chart below shows that the negative correlation of 2000 to 2021 was one regime among several.
Chart 4: Rolling 36-month correlation of monthly S&P 500 price returns and approximate 10-year Treasury returns, 1967 to October 2026. Source: S&P 500 closes (S&P Dow Jones Indices); Federal Reserve H.15 10-year yield via FRED (DGS10); calculations by IndraStra Global Research Team.
The cost of that shift can be expressed in portfolio terms. For weights 0.6 and 0,
Here , : volatilities of stocks and bonds; : stock–bond
and with the annualized volatilities of the past 36 months, 12.3 percent for stocks and 6.4 percent for the bond proxy, a correlation of +0.34 gives a 60/40 volatility of about 8.6 percent. Applying the 2000–2021 average correlation of −0.32 to the same volatility assumptions would reduce the estimate to about 7.0 percent. A mix of 60 percent stocks and 40 percent cash would carry estimated volatility of about 7.4 percent, assuming zero cash volatility. On these assumptions, replacing bonds with cash would have lowered estimated portfolio volatility from 8.6 percent to 7.4 percent. This is a volatility comparison, not a comparison of total returns or a claim that cash would necessarily have delivered better investment performance. The hedge has not vanished. When technology shares fell in late October, the TBAC noted that the equity pullback, together with wider credit spreads, helped restore a safe-haven bid for Treasuries. Bonds still offset growth scares, but they offset inflation and rate shocks poorly, and those shocks have dominated 2026. For allocators, that argues for treating duration as a position chosen for the shock it is meant to hedge, rather than as automatic ballast.
Several scheduled releases will show whether the plateau in the term premium holds. The FOMC meets on December 8–9 with new projections, and the September median implied one more quarter-point increase this year; another hike with stable breakeven rates would extend the 2026 pattern. The Treasury's next refunding documents are scheduled for February 1 and February 3, 2027, and any change in forward guidance on coupon sizes would be the first concrete sign that the duration test is arriving. New ACM and Kim-Wright readings will show whether the two models converge, and CBO's next baseline will show how much of the higher yield path it builds into interest costs. On current evidence, the likeliest near-term outcome is that long yields keep tracking the expected path of policy rates, although the Kim-Wright model suggests that the term premium may be playing a larger role than the ACM estimates indicate. The bigger risk is that the premium starts rising again once the Treasury needs to sell more long-term debt into a market where bonds no longer reliably hedge stocks. The fiscal story behind long yields has not gone away; in 2026 it has been waiting in the issuance calendar.
Apple News, Google News, Feedly, Flipboard, and WhatsApp Channel
COVER IMAGE: The cover image is AI-generated and is used for illustrative purposes only. It does not depict actual people, places, or events.






