The most important rate move of the last cycle happened while the Federal Reserve was standing still. On July 26, 2023 the Fed delivered what turned out to be its final hike, taking the policy rate to a 5.25% to 5.50% range, and then did nothing for over a year. Between late July and October 19, 2023, the 10-year Treasury yield rose by roughly a full percentage point to touch 5%, its highest level since 2007. The cost showed up everywhere except in Fed communications. The 30-year mortgage rate reached about 7.8% in late October, the highest since 2000. The S&P 500 fell roughly 10% from its July high. None of it came from a change in what markets expected the Fed to do.
That episode is the cleanest introduction to the term premium, the extra yield investors demand for holding a long bond instead of rolling short-term bills for the same period. It is the portion of a long yield that expected policy cannot explain. It cannot be observed directly, every estimate of it depends on a model, and it is the variable that decides whether deficits matter for borrowing costs. When it moves, long rates stop taking instructions from the central bank.
What Moved in 2023
The 2023 sequence matters because it isolates the mechanism. On August 1, Fitch downgraded US sovereign debt from AAA. On August 2, the Treasury's quarterly refunding announcement raised the size of its coupon auctions, meaning more long-dated debt would hit the market each quarter. Over the following weeks, expectations for the path of the policy rate barely shifted. Long yields kept rising anyway.
The popular reading at the time blamed "higher for longer," the idea that the Fed would hold rates up for an extended period. That story explains some of the move, but it cannot explain the whole thing, because two-year yields, which track expected policy closely, rose far less than ten-year yields. The gap between them is where the term premium lives. Estimates published by the New York Fed, which had shown a negative term premium for most of the prior decade, swung back toward positive territory during those months.
Then, on November 1, 2023, the Treasury slowed the pace of its coupon increases and leaned more heavily on short-term bills. Long yields fell sharply over the following weeks. Supply news moved the long end in both directions while the Fed sat still.
What this means for you: when long yields rise and the two-year barely moves, the Fed is not the driver, and Fed speeches will not reverse it.
The Decomposition Behind the Number
Any long yield can be split into two pieces:
10-year yield = average expected short rate over 10 years + term premium
The first piece is what you would earn by rolling three-month bills for a decade, given the market's best guess about future policy. The second piece is compensation for committing to a fixed rate for ten years and bearing the risk that your guess is wrong.
The difficulty is that only the left side is observable. As of October 1, 2026, the 10-year yield stood at 5.24% and the 2-year at 4.78%, a 2s10s spread (the 10-year yield minus the 2-year yield) of +0.46 percentage points. To extract the term premium you must estimate the expected short rate, and that is where models diverge.
Consider two illustrative estimates. Model A, built on surveys of economists, assumes the policy rate averages 4.40% over the next decade. Model B, fitted purely to the shape of the yield curve, assumes 4.90%.
- Model A: 5.24% minus 4.40% = a term premium of 0.84 percentage points
- Model B: 5.24% minus 4.90% = a term premium of 0.34 percentage points
Same yield, same day, and one model says investors are demanding two and a half times the compensation the other model sees. These figures illustrate the arithmetic; they are not the readings of any specific published model. Real estimates differ by similar margins, which is why serious practitioners track several at once and focus on the direction they share rather than the level any one of them produces.
The practical consequence of the decomposition is duration math. Duration is a bond's price sensitivity to yield changes; a 10-year Treasury trading near par has a duration of roughly 8. If the term premium rises by 50 basis points with expected policy unchanged, the bond loses about 8 × 0.50% = 4% of its value. Holders absorb that loss with no help from the Fed, because the Fed did not cause it.
What this means for you: treat any single term premium estimate as a range, and pay closer attention when the 10-year moves without a matching move in the 2-year.
Why Anyone Demands the Premium
The term premium exists because someone has to hold the duration the Treasury issues, and holders differ sharply in how sensitive they are to price.
One group is price-insensitive buyers. These are holders who buy for reasons other than expected return: the Fed during quantitative easing, foreign central banks managing reserves, pension funds and insurers matching long liabilities, banks meeting liquidity rules. They absorb duration almost regardless of yield. For most of 2010 to 2021 they absorbed enormous amounts, and term premium estimates sat near zero or below it.
The other group is price-sensitive buyers: asset managers, hedge funds, and households who compare the bond yield to every other use of their money. They will hold more duration only if they are paid more to do so.
When issuance rises while the price-insensitive buyers step back (the Fed running down its balance sheet, foreign reserve managers diversifying, banks already full), the marginal bond has to find a price-sensitive home. That buyer sets the price, and that buyer demands a premium. A fiscal deficit therefore matters to yields through the term premium, not through Fed policy. A larger deficit with patient buyers costs little. The same deficit with impatient buyers costs a great deal.
A second force runs through hedging value. From roughly 2000 to 2020, bonds tended to rise when stocks fell, so a long bond worked as portfolio insurance, and people pay for insurance rather than demand payment to hold it. Negative term premium was the price of that hedge. When inflation becomes the dominant risk, bonds and stocks fall together, as they did in 2022, and the bond stops insuring anything. Holders then want compensation, and the term premium turns positive.
The Greenspan "conundrum" shows the mechanism running in reverse. Between June 2004 and June 2006 the Fed raised its policy rate from 1% to 5.25%, and the 10-year yield barely moved. Heavy foreign reserve buying of Treasuries compressed the term premium enough to offset most of 425 basis points of tightening.
What this means for you: to forecast long yields, track who is buying the auctions, not just what the Fed is signaling.
Where the Move Lands First
The cross-asset transmission is where the term premium does its damage, and it rarely shows up in the policy headline.
Mortgages are the most direct channel. US mortgage rates price off the 10-year yield plus a spread, so a term premium shock reaches household borrowing costs within days, regardless of the policy rate. The 2023 mortgage peak near 7.8% arrived with no Fed hike behind it.
Equities feel it through the discount rate. A higher long yield raises the hurdle every future cash flow must clear, and long-duration equities (companies whose value sits mostly in distant earnings) reprice hardest. The roughly 10% S&P 500 drawdown in autumn 2023 tracked the 10-year far more closely than it tracked any earnings revision.
The dollar provides the most useful tell. Normally, higher US yields attract capital and strengthen the dollar. When the term premium is rising because investors doubt fiscal discipline, the relationship can break: yields rise and the currency falls together, because the move reflects a demand for compensation to hold the country's assets rather than an attractive return on them. The UK gilt crisis of September 2022 showed the extreme version, with long gilt yields spiking and sterling sliding until the Bank of England intervened on September 28. US markets showed a milder form in April 2025, when the 10-year jumped about half a point in a week as the dollar weakened.
At 5.24%, the 10-year now sits above the 2023 peak, with the 2-year 46 basis points below it. A positive spread of that size, with the policy outlook not pointing to large further hikes, is consistent with term premium carrying a larger share of the long yield than it did through the last decade. The exact share is model-dependent, as the decomposition above showed.
What this means for you: watch the 10-year and the dollar together. Rising yields with a falling dollar is the signature of a term premium driven by fiscal risk, and that is the version that hurts all three channels at once.
When the Read Goes Wrong
The term premium framework fails in three identifiable ways.
The first is misattribution. If markets quietly raise their estimate of the neutral rate (the policy rate that neither stimulates nor restrains the economy), expected short rates rise, and a model that lags that shift will book the move as term premium. The analyst then concludes the Fed cannot help when, in fact, the market is pricing the Fed. Checking survey expectations against curve-implied expectations guards against this.
The second is policy suppression. Quantitative easing and outright yield caps can hold the term premium down by force, as the Bank of Japan's yield curve control did for years. Estimates during such periods measure the central bank's balance sheet more than investor demand, and they can snap higher when the program ends.
The third is speed. Term premium estimates are smoothed outputs. In a disorderly market such as the 2022 gilt episode, forced sellers dominate for days, and the "premium" reflects liquidation pressure rather than a considered demand for compensation.
What this means for you: confirm a term premium read with auction demand and the yield-dollar relationship before acting on it.
The Practical Takeaway
The term premium is the compensation price-sensitive buyers demand to hold duration that price-insensitive buyers no longer absorb, and it is the channel through which deficits, inflation risk, and hedging value reach long yields without passing through the Fed. With the 10-year at 5.24%, the long end is answering to the marginal buyer of Treasury supply. Mortgage rates, equity valuations and the dollar will follow that buyer, and Fed guidance will not override it.
