Why Long-Term Interest Rates Do Not Fall Even When Bad Economic Data Is Released — Breaking Down the 10-Year Yield into 'Expectations' and 'Term Premium' (October 2026)
What you will learn in this article
-
The 10-year government bond yield can be thought of as the sum of the ‘average of expected short-term interest rates over the next 10 years‘ and the ‘premium for locking in money for 10 years (term premium)‘.
-
About half of the rise in long-term interest rates in September 2026 was due to an increase in the term premium. This is one of the reasons why long-term interest rates were slow to fall even when bad economic data was released.
-
What drives the term premium (uncertainty about prices, supply and demand for government bonds) and the limitations of it being an estimated value.
Why we are covering this now
On September 29, 2026, a series of bad economic figures were released in the United States.
Usually, when the economy looks like it is going to worsen, the view that the central bank will stop raising rates or start cutting them strengthens, making interest rates more likely to fall. However, on this day, the yield on the U.S. 30-year Treasury bond reached 5.613% during trading, the highest level since June 2002 (CNBC). The 10-year Treasury bond also continued to rise, reaching 5.24% on September 28 according to official final figures (FRED).
Even though bad economic data was released, long-term interest rates did not fall. To understand this ‘movement contrary to expectations,’ it is necessary to look at long-term interest rates by dividing them into two components.
Basics: Buying a 10-year government bond versus rolling over shorter-term bonds
The comparison is against the ‘path of continuous rollover’
Suppose you have 1 million yen and want to invest it in government bonds for 10 years. There are two paths.
-
Path A: Buy a 10-year government bond and receive the same yield for 10 years
-
Path B: Buy a short-term government bond and, every time it matures, buy another one at the prevailing interest rate (rollover)
The 10-year performance of Path B is determined by how short-term interest rates move from here on. If the market expects that ‘the average short-term interest rate over the next 10 years will be around 4.12%,’ it seems reasonable that the 10-year government bond in Path A should also be around 4.12%.
However, in reality, the yield on the 10-year government bond on September 25, 2026 (calculated using an estimation model by FRB researchers) was 5.14%. The difference from the expected average of 4.12%, which is 1.02%, is the term premium.
Comparing with 1 million yen
Assuming interest rates move as expected, we calculated the difference over 10 years using ‘principal × coefficient’.
Why is an extra premium needed: The ‘patience fee’ for a 10-year lock-in
If Path A yields 153,400 yen more over 10 years, everyone would likely choose Path A. The reason a premium still remains is that Path A has the disadvantage of not being able to change the terms midway.
To use an analogy, it is a 10-year fixed-rate plan. The lender (the buyer of the government bond) fixes the yield they receive for 10 years. Even if market interest rates rise during that period, the yield on the bond they hold does not increase. If they try to sell it midway, they can only sell it at a lower price.
Calculation example: This is the price when you buy a 10-year government bond with a 5.14% yield and, one year later, the interest rate for the remaining 9 years moves by 1 point (calculated with a face value of 100 and annual interest payments).
The ‘patience fee‘ demanded in exchange for taking on such uncertainties—that interest rates might move significantly, inflation might rise more than expected, or one might be forced to sell midway—is the term premium.
Breaking down long-term interest rates into two parts
Method of breakdown
10-year government bond yield = Average of short-term interest rate expectations (for 10 years) + Term premium
Since neither is directly visible, they are estimated using a model. Here, we use the estimates (published on FRED) from a model created by researchers Kim and Wright at the FRB (the U.S. central bank). Because the calculation method for the 10-year yield in this model differs slightly from the 10-year government bond yield we usually see (FRED’s DGS10), the values deviate slightly (5.14% vs. 5.17% on September 25, 2026).
Analyzing the rise in long-term interest rates in 2026
Breaking the rise down into two components, it is as follows.
The rise in 2026 was roughly half ‘expectations’ and half ‘patience fee’. Looking from the beginning of 2025, the rise in the patience fee is greater. The term premium of 1.02% on September 25 is the highest since April 2010.
What moves the expectations component: The central bank
Expectations for short-term interest rates are almost entirely determined by the outlook for central bank policy. The FRB raised the policy rate by 0.25 points in September 2026 (CNBC reported on September 28 that FRB Governor Cook stated she supported this decision). The 2-year government bond yield (which easily reflects policy outlooks) also rose from 4.71% on September 22 to 4.92% on the 28th (FRED).
When bad economic data is released, this ‘expectations component’ usually falls. This is because the view that rate hikes will stop or that the policy will shift to rate cuts becomes stronger.
What moves the term premium: Uncertainty and supply/demand
In general, the term premium is considered to rise in the following situations.
On September 28, 2026, Mohamed El-Erian of Allianz in the U.S. stated on CNBC, ‘Even if oil prices were falling, we would still be having this interest rate discussion. The balance of supply and demand for long-term government bonds is broken,’ and expressed the view that the 10-year government bond yield would remain around 5%. This is a view that seeks the cause not only in prices but also in the supply and demand of government bonds.
How to interpret ‘not falling even with bad data’
By lining up the two components, it becomes easier to interpret movements like those on September 29.
-
Poor economic data → A downward force acts on the ‘expectations component’
-
Anxiety over inflation or supply and demand for government bonds → An upward force continues to act on the ‘term premium’
-
If the latter offsets the former, long-term interest rates will not fall. The longer the bond’s maturity, such as with 30-year bonds, the greater the impact of the ‘patience premium’.
However, the breakdown for September 29 itself has not yet had its estimates published as of the time of writing (the latest is for September 25). It has not yet been confirmed which component caused the movement on this day.
Why it doesn’t ‘always’ happen that way
-
The term premium is an estimate. It is not a price traded directly in the market, and the value changes depending on the model and assumptions. In models from other research institutions, the levels can be quite different. It is safer to look at the direction of whether it is rising or falling, rather than the level itself.
-
It can also be negative. From 2012 to 2023, there were 58 months where the monthly average was negative (the lowest was -0.61% in July 2020). Generally, this is attributed to central banks purchasing large amounts of government bonds and strong demand for government bonds as safe assets. A negative patience premium means that there were many people who said, ‘I want to hold government bonds even if it’s a 10-year fixed rate.’
-
Long-term interest rates often fall smoothly when poor economic data is released. In phases where anxiety over inflation is low, the decline in the ‘expectations component’ takes full effect. In 2026, anxiety over inflation is high, so offsetting effects are more likely to occur.
-
Breakdowns for short periods fluctuate. Even over the three business days from September 22 to 25, the rise was split roughly in half, but the ratio changes significantly depending on the day. It is necessary to look at the trend over several weeks.
Common Misconceptions
Misconception 1: ‘If the economy worsens, long-term interest rates will definitely fall’
What economic weakness lowers is mainly the ‘expectations component.’ If the patience premium rises due to anxiety over inflation or government bond supply and demand, long-term interest rates may not fall. The movement on September 29, 2026, is an example that illustrates this possibility.
Misconception 2: ‘A rise in long-term interest rates = expectations for rate hikes have strengthened’
Of the 0.88 percentage point rise from January to September 2026, the ‘expectations component’ accounted for 0.44 points, and the rest was the term premium. Half of the rise in long-term interest rates was for reasons other than rate hike expectations. that is the calculation.
Misconception 3: ’10-year government bonds are more profitable than short-term government bonds because the yield is higher’
The reason the yield on 10-year government bonds is high is that it includes a patience premium for accepting the risk that the price will fall during the holding period. If interest rates rose by 1 point one year later, the price was calculated to fall by 6.76. The high yield is the reward for the risk being accepted.
Practical Application: How to use this model
When long-term interest rates move significantly, check in the following order.
-
Compare with 2-year government bonds. If the 2-year bond is moving in the same way, it is highly likely that the policy outlook (the expectations component) is moving. If only the 10-year or 30-year bonds are moving, there is a possibility that it is the patience premium.
-
Look at the term premium estimates. Use FRED’s THREEFYTP10 (published with a few days’ delay) to verify which component is driving the rise.
-
Look at news regarding supply and demand for government bonds. Auction results, fiscal deficit outlooks, and the actions of central banks and overseas buyers.
-
Compare inflation outlooks as well. If expected inflation is rising, it is consistent with an increase in the ‘patience premium’.
-
Judge based on trends over several weeks. Do not draw conclusions from a single day’s breakdown.
Summary
-
10-year government bond yield = Average of short-term interest rate expectations + Term premium (10-year fixed patience premium)
-
On September 25, 2026, it was 5.14% = 4.12% + 1.02%. The patience premium is at its highest level since April 2010.
-
The 0.88 percentage point rise in 2026 is roughly half due to expectations and half due to the patience premium.
-
Bad economic data lowers ‘expectations,’ but if inflation anxiety or government bond supply and demand raise the ‘patience premium,’ long-term interest rates may not fall.
-
The term premium is an estimate. Look at the direction rather than the level, and look at trends over several weeks rather than a single day.
Terminology
-
Term premium: The additional interest rate demanded for locking up money for a long period. A ‘patience premium’.
-
Average of short-term interest rate expectations: The forecast for what short-term interest rates (rates close to the central bank’s policy rate) will average in the future.
-
Rollover: Reinvesting in short-term bonds as they mature at the prevailing interest rate.
-
Consumer Confidence Index: An index created by surveying households about their outlook on the economy and employment. Published monthly by the Conference Board, a private U.S. research organization.
-
Job Openings (JOLTS): The number of job openings posted by U.S. companies. Published monthly by the U.S. Department of Labor.
-
Fiscal deficit: The amount by which government spending exceeds tax revenue, etc. Covered by issuing government bonds.
-
Expected inflation rate: The rate of future price increases that people anticipate
-
Estimated value: A value calculated using a model for something that cannot be measured directly
Sources
-
Consumer optimism slides to lowest since 2014 as fears escalate over rising prices and jobs (CNBC) — Published 2026-09-29T14:36Z (publication date extracted mechanically from metadata). Consumer Confidence Index 81.9, -6.7 month-over-month, forecast 89, basis for the lowest level since 2014
-
Dow posts back-to-back losses as Treasury yields continue their ascent: Live updates (CNBC) — Published 2026-09-28T22:02Z, updated 2026-09-29T20:43Z (same as above). Basis for the 30-year Treasury intraday high of 5.613% (since June 2002)
-
Dow slides more than 300 points to start week as Treasury yields pressure stocks: Live updates (CNBC) — Published 2026-09-27T22:03Z, updated 2026-09-28T20:24Z (same as above). Basis for Mr. El-Erian’s remarks and Governor Cook’s support for a 0.25 percentage point rate hike in September
-
FRED (St. Louis Fed) — THREEFY10 (10-year yield from Kim-Wright model) / THREEFYTP10 (same, term premium) / DGS10 / DGS2 / JTSJOL (Job Openings). Retrieved on September 30, 2026. Kim-Wright estimates are as of September 25, and Treasury yields are as of September 28.
-
‘Average of short-term interest rate expectations’ is calculated as ’10-year yield – term premium’, along with breakdowns, ratios, historical values, a 1 million yen table, and price calculation examples.
This document is a summary of facts and a presentation of general interpretations; it is not investment advice or a recommendation to buy or sell. The figures stated are as of the time of writing.