The Fed sets the overnight rate. It does not set the ten-year yield. In 2026 that distinction stopped being academic — long yields have risen while policy sat still, and the gap between what the Fed controls and what the bond market demands has a name.
This article is for informational and educational purposes only. It is not financial advice, investment advice, or a recommendation to buy, sell, or hold any security, cryptocurrency, or financial product. Always verify data with official sources before making financial decisions.
Short answer: the compensation for holding duration
Short answer: Term premium is the extra yield investors require to hold a long-dated bond instead of rolling short-term bills over the same horizon. It compensates for the risk that rates, inflation or the supply of bonds turns out differently than expected. It cannot be observed directly — it has to be estimated by subtracting a model’s path of expected future short rates from the actual yield, which means every term premium figure you see is a model output rather than a market price.
Decomposing a 10-year yield
The standard framework splits any long yield into two parts.
| Component | What it represents | What drives it |
|---|---|---|
| Expectations component | Average expected overnight rate over the bond’s life | Fed policy path, growth and inflation outlook |
| Term premium | Compensation for bearing the risk of being wrong | Bond supply, inflation uncertainty, demand from price-insensitive buyers |
The practical implication is that a rising ten-year yield has two possible explanations with opposite meanings. If the expectations component is rising, the market is pricing a stronger economy or a more hawkish Fed. If term premium is rising while expectations are flat or falling, the market is demanding more compensation for risk — which can happen for reasons that have nothing to do with the growth outlook, such as a heavier issuance calendar or reduced foreign demand.
The ACM and KW models, and why they disagree
Two estimates dominate the discussion, and they routinely differ by a meaningful margin on the same day.
The Adrian, Crump and Moench model, published by the New York Fed, extracts term premium from the shape of the yield curve using a statistical factor approach. The Kim-Wright model, published by the Federal Reserve Board, uses a different specification and incorporates survey forecasts of future rates.
They disagree because term premium is not identified by the data alone — the split between expectations and premium depends entirely on assumptions about how expectations are formed. A model that anchors expectations to surveys will attribute more of a yield move to term premium than one that lets expectations float freely. Neither is wrong; they are answering the question under different assumptions. Anyone quoting a precise term premium figure without naming the model is quoting one estimate as though it were a measurement.
What pushes term premium up
- Supply. More Treasury issuance, particularly at long maturities, requires clearing at a higher yield. Quarterly refunding announcements are the scheduled version of this; unexpected deficit deterioration is the unscheduled version.
- Inflation uncertainty. Not the level of expected inflation — that sits in the expectations component — but uncertainty about it. Wider disagreement about the inflation path raises the risk of holding a fixed nominal cash flow for a decade.
- Reduced price-insensitive demand. Foreign official buyers and a central bank running quantitative easing both buy without regard to yield. When those buyers step back, price-sensitive investors have to clear the market, and they demand more.
- Correlation regime. When bonds reliably rally in equity sell-offs, duration is a hedge and commands a lower premium. When stocks and bonds fall together — as they do when inflation is the shock — duration stops hedging and the premium rises.
2026 in context: a Fed on hold, a steepening long end
The current configuration is the textbook case. The Fed has held at 3.50%–3.75%, so the expectations component for the front end is pinned. Yields further out have risen anyway, reaching 4.69% on the ten-year by late July on tariff-driven inflation expectations and supply concerns.
A curve that steepens while the policy rate is unchanged is close to a definitional term premium story — there is no policy move to attribute it to. That does not make it benign. Term premium-driven steepening tightens financial conditions through mortgage rates and corporate borrowing costs without the Fed having done anything, and it is not obviously responsive to Fed communication.
What this means for mortgages, equities, and the dollar
Thirty-year mortgage rates track the ten-year yield plus a spread, so term premium passes into housing affordability directly. A borrower does not care whether their rate rose because of expected policy or because of issuance — the payment is the same.
For equities the channel is the discount rate. Long-duration growth companies, whose value sits in distant cash flows, are mechanically more sensitive to a higher long yield than companies with near-term earnings. This is the standard explanation for why technology underperforms on days when the long end sells off.
The dollar effect is ambiguous and worth flagging as such. Higher yields normally attract capital and support the currency, but term premium rising because investors demand more compensation to hold US assets is a different signal — one that has historically coincided with dollar weakness rather than strength. Which effect dominates depends on why the premium is rising, and that is exactly the question the models disagree about.
Mini glossary
- Duration. Sensitivity of a bond’s price to a change in yield, expressed in years.
- Expectations hypothesis. The proposition that long yields equal the average of expected future short rates. Term premium is the measured deviation from it.
- Quantitative tightening. Reduction of the central bank’s balance sheet, which increases the supply of duration the private market must hold.
- Quarterly refunding. The Treasury’s scheduled announcement of borrowing plans, including the maturity mix. A shift toward longer maturities raises the duration supplied.
What this article does not conclude
Term premium is an estimate, not an observation, and the leading models disagree about its level. A rising estimate is consistent with several different underlying causes — supply, uncertainty, changed demand — and the decomposition cannot tell you which one is operating.
Current model estimates are published and updated by the New York Fed and the Federal Reserve Board. Where the level matters, take it from the source and note which model produced it.