What Is the Treasury Term Premium?

A 10-year Treasury yield is the market yield on a security with cash flows extending much farther into the future than the next Federal Reserve meeting. Investors can compare that yield with an estimate of the short-term rates they expect over the same horizon. The difference is commonly called the term premium. In a simplified decomposition, a long-term nominal yield is approximately the average expected future short rate plus a term premium.

The term premium is meant to describe the compensation associated with holding longer-maturity interest-rate exposure instead of repeatedly investing at short rates. If yields change unexpectedly, a long-term bond's price can move substantially before maturity. Investors' willingness to bear that risk, the uncertainty around future inflation and rates, and the bond's value as a hedge can all matter to the price of long-term debt. The Federal Reserve's discussion of long-term interest rates describes interest-rate risk as one reason investors may require additional compensation for holding longer bonds.

This decomposition does not mean a Treasury has two separately traded pieces, one for policy expectations and one for a premium. The observed yield is the market price translated into a rate. Expected future short rates and the term premium are inferred components, and a model must define precisely which yields, expectations, maturities, and compounding conventions it uses.

The word “premium” also does not mean that this component is always positive. Some estimates can be negative. A negative estimate says that the model's term-premium component is below zero under its definition; it does not mean the Treasury yield itself is negative, that the investor is guaranteed a loss, or that the model has directly observed a negative payment.

An investor holding a long Treasury faces more price sensitivity to rate changes than an investor rolling over very short bills over the same period. If inflation, policy rates, or the economic outlook change, the bond's market value can move. The holder may also be uncertain about how easily the position can be sold, how much long-duration risk other investors will absorb, and whether bonds will diversify other assets when protection is most needed.

Compensation for bearing that uncertainty is not a fee that the Treasury adds to the coupon. It is a market-implied component of the yield, conditional on the way the risk and expectations are measured. When investors place high value on the predictable cash flows or hedging role of long Treasuries, they may accept a lower yield. When they are less willing to hold duration risk, or when uncertainty increases, they may demand a higher yield. Changes in supply, demand, and investor risk-bearing capacity can therefore matter alongside the expected policy path.

The term premium is distinct from a corporate credit spread. U.S. Treasury securities carry very different credit and contractual risks from corporate debt; a Treasury term premium decomposition is about nominal rates across maturities, while a corporate spread comparison also reflects the issuer, the bond's seniority and liquidity, and expected credit losses. A term-premium estimate should not be read as a forecast of Treasury default or as the same quantity as a corporate bond's spread over Treasuries.

There is also no single observable “true premium” that can be isolated from one day of trades. Investors may have different horizons, liabilities, hedges, and constraints. A model's estimate is a structured interpretation of market prices, not a survey of every holder or a separately quoted market rate.

One approach uses a dynamic term-structure model. The model is fitted to Treasury yields at multiple maturities and represents how interest rates evolve over time. It estimates a path for future short rates and compares the average over the bond's horizon with the fitted yield. A 10-year term premium estimate is then the difference between the fitted 10-year yield and the model's average expected short rate over that horizon.

Some models use only yield data; others also use surveys of professional forecasts to help estimate longer-run rate expectations. The Federal Reserve Board's three-factor nominal term-structure model incorporates survey forecasts of three-month Treasury bill rates alongside yields. The Board explains that those expectations are not directly observable and that survey data are treated as a noisy proxy rather than accepted as perfect market expectations.

The New York Fed's Adrian-Crump-Moench approach is another model-based decomposition. A survey-based method offers a different check: subtract the average short-rate path reported in a forecast survey from the long Treasury yield. That approach is easier to describe, but the survey may not represent the marginal investors setting bond prices. Model and survey estimates answer related questions with different assumptions; neither is an unfiltered observation of expectations.

Some published yield decompositions also include a model residual in one component so that the reported parts add back to the observed yield. The San Francisco Fed's Treasury yield premium estimates explicitly describes a model with expected short rates, a term premium, and a residual; its charting convention assigns the residual to the premium so the displayed components add to the yield. Always read the source's definition before comparing a chart with another model.

Suppose a fitted 10-year Treasury yield is 4.25% and a particular model estimates an average future short rate of 3.75% over the same horizon. Under that model's convention, the estimated term premium is 0.50 percentage points, or 50 basis points:

4.25% fitted yield ≈ 3.75% expected average short rate + 0.50% estimated term premium

This is a hypothetical illustration, not a live market estimate. The calculation shows what “decomposition” means; it does not imply that investors can separately buy a 3.75% expectation and a 0.50% premium.

Now suppose, at a later date, the fitted yield is 4.65%, the expected average short rate is 3.90%, and the estimated term premium is 0.75%. The yield has risen 40 basis points. In the model's accounting, 15 basis points of that change come from the expected-rate component and 25 basis points from the term-premium component. Those two changes sum to the 40-basis-point move, subject to the model's definitions and fitting error.

The pieces can also move in opposite directions. Starting from 4.25% = 3.75% + 0.50%, a later decomposition of 4.30% = 3.95% + 0.35% would show expected short rates up 20 basis points and the term premium down 15 basis points. The total yield would rise only 5 basis points because the estimated components offset one another.

An actual empirical yield curve uses coupon securities, fitted zero-coupon yields, specific dates, and conventions. Do not treat these round numbers as observed forecasts, a price target, or a guaranteed return. Their purpose is to make the addition and subtraction transparent.

The main quantities in the decomposition are not directly observable. A model must infer how short rates may behave over a long period from data that cover only a finite history. Treasury yields are persistent, so a different sample period, number of factors, or assumption about how rates eventually return toward a long-run level can change the estimated path and the residual premium.

The Federal Reserve's research note on the robustness of long-maturity term-premium estimates compares the Kim-Wright and Adrian-Crump-Moench models. The models can differ materially at times, in part because they use different sample periods and data; the Kim-Wright estimates incorporate survey forecasts while the original ACM estimates in the comparison use Treasury yields. Adding survey data can reduce some estimation uncertainty, but that does not guarantee a more accurate estimate if the survey is a poor proxy for market expectations.

Definitions matter as well. The Federal Reserve's three-factor model FAQ notes that its yield term premium includes both a “pure” compensation-for-risk component and a convexity premium. Other authors may define the term premium without that convexity adjustment. A second model may also assign a fitting residual differently. Comparing levels without checking definitions can create a false disagreement; comparing changes can sometimes be more informative, but even changes depend on horizon, model, and date.

These uncertainties are not a reason to ignore the measure. They are a reason to treat it as a range of model-based evidence. The Federal Reserve note finds that some features of estimates are more robust across models than others; for example, model estimates may agree on broad historical patterns while differing on exact levels, signs, or short-term volatility. A single decimal should not be presented as a precise fact about what every investor expects.

Uncertainty about inflation and future rate changes can raise the compensation some investors require for holding longer-duration bonds. Uncertainty about growth, public borrowing, or the future supply of duration may also change investors' willingness to hold long Treasuries. If major investors have less capacity or appetite for duration, other buyers may need a lower price, which corresponds to a higher yield. These channels can overlap; the yield curve does not label each source separately.

Demand can push in the other direction. A long Treasury may be valuable to investors seeking reliable cash flows or a potential hedge against economic stress. Strong demand can raise its price and lower its yield, which may reduce a model-estimated term premium. The New York Fed has described historical term-premium estimates moving with uncertainty and disagreement about future yields, while the Federal Reserve discusses safe-haven demand as one factor that has at times lowered long-term yields.

The relationship is conditional, not a one-variable formula. More Treasury issuance does not mechanically raise the premium by a fixed amount: its effect depends on demand, the maturity of the issuance, the outlook, and how much risk the market is already pricing. Likewise, a Fed balance-sheet change may alter who holds duration, but it does not give a one-for-one estimate of the term premium. Read the premium together with yields, policy expectations, inflation compensation, and market context rather than attributing a move to a single headline.

Not by itself. A 10-year yield can rise because the expected short-rate path rises, because the term premium rises, or because both change. A term-premium estimate attempts to separate those explanations; it is not a statement of what the Federal Open Market Committee will decide at its next meeting.

That distinction helps when a long yield moves against a policy-rate change. A short-term cut affects the near end of the curve, but a 10-year yield includes an average expected path extending far beyond that decision and a premium for longer-term exposure. A cut may be anticipated already, later policy may be expected to differ, or the premium may rise enough to offset a fall in expected short rates. See why long-term Treasury yields can rise after Fed rate cuts for that timing question.

Term-premium estimates are also not economic forecasts. A higher estimate does not guarantee that yields will continue rising, and a low or negative estimate does not guarantee that long bonds will outperform. Bond prices can respond to new information in either direction; the model output describes a proposed decomposition of prices observed up to a date.

Start by checking the maturity, date, and measure. A 10-year term premium is not interchangeable with a 2-year estimate or a forward premium for a future 5-to-10-year period. Check whether the displayed yield is a coupon-bond yield, a fitted zero-coupon yield, or a forward rate; confirm whether values are percentages or basis points and whether the model uses annualized or continuously compounded rates.

Next, identify the provider and method. A New York Fed ACM series, a Federal Reserve Board model estimate, a San Francisco Fed decomposition, and a survey-based calculation may differ in their inputs, sample, definitions, or update schedule. Keep one model fixed when measuring a change through time. If a chart shows an estimate from one model, do not silently substitute another model's number to build a single historical series.

Finally, distinguish the observed yield from each estimated component. Ask whether the model reports a fitted yield, how closely it fits observed Treasury yields, whether a residual is included, and whether data or model parameters can be revised. Use several indicators when describing why the market moved. A term-premium chart can help explain one part of long yields, but it cannot by itself prove what investors expected or identify a unique cause.

For a complete reading, pair the premium with the maturity-matched nominal Treasury yield and an estimate of the expected short-rate path. Depending on the question, real yields and breakeven inflation can provide additional context, though they too contain market premiums. The goal is to state what the model estimates, what market prices show, and which interpretation remains uncertain.