TL;DR: Duration is the single most important number on a bond’s risk page. It tells you, in years, how much a bond’s price will move when yields change by one percent. A bond with a duration of eight falls roughly eight percent if yields rise one percent — and rises by about the same amount if yields fall. Long Treasuries lost double-digit percentages in 2026 not because anything was “wrong” with them, but because their duration was doing exactly what duration is supposed to do.
What duration actually measures
Every bond is a stream of future cash flows: coupons every six months, plus the face value at maturity. The bond’s price today is just the present value of those cash flows, discounted at the bond’s yield. When yields rise, the discount factor gets larger, so the present value — and the price — falls. When yields fall, the opposite. FINRA puts it in three sentences: when interest rates rise, bond prices generally fall; when interest rates fall, bond prices generally rise; every bond carries interest rate risk.
Duration formalizes that relationship. There are three flavors you’ll see in practice:
- Macaulay duration — the weighted average time, in years, until you receive the bond’s cash flows. The weights are the present values of each cash flow.
- Modified duration — Macaulay duration divided by (1 + y/n), where y is the yield and n is the compounding frequency. Modified duration is the one you use as a price-change estimator.
- Effective duration — the version used for bonds with embedded options (callable corporates, mortgage-backed securities). Effective duration shocks the whole yield curve up and down and measures the actual price change, which captures option-driven behavior that the modified-duration formula misses.
The rule of thumb every fixed-income desk uses comes from modified duration:
ΔPrice ≈ − Modified Duration × ΔYield
So a bond with modified duration of 8.0 loses about 8% if yields rise 1% (100 basis points). A bond with modified duration of 15 loses about 15%. That’s it — the rest of bond mathematics is refinement on this one idea.
A worked example: the 10-year Treasury
Take the current 10-year Treasury note. On May 22, 2026, the constant-maturity 10-year yield closed at 4.56% per the U.S. Treasury daily yield curve (Treasury.gov). A par bond — one issued today with a 4.56% coupon — would have a Macaulay duration of about 8.14 years and a modified duration of about 7.96.
If yields rise 50 basis points overnight, the rule of thumb says the price falls by roughly 7.96 × 0.50% = 4.0%. A 100 bp move costs about 8%. A 200 bp move — the kind of move 10-year Treasuries actually saw between 2020 and 2023 — costs about 16%, before you adjust for convexity (more on that below).
Notice that the duration (8.14 years) is shorter than the maturity (10 years). That’s almost always true for a coupon bond, because the coupons return some of your money before maturity. The only time duration equals maturity is for a zero-coupon bond, where you receive nothing until the end. A 30-year zero-coupon Treasury at today’s yield has a Macaulay duration of 30.
Why long bonds got crushed in 2026
The long end of the Treasury curve is the textbook duration trade. The 30-year par yield closed at 5.07% on May 22 (Treasury.gov), implying a modified duration of about 15.3 for a freshly issued 30-year. Long-duration ETFs are even more rate-sensitive: the iShares 20+ Year Treasury Bond ETF (TLT) holds older long bonds that, at today’s yields, carry an effective duration in the high-teens.
Between February 27 and May 19, 2026, the 30-year Treasury yield moved from 4.64% to 5.18% — a 54-basis-point rise, per Treasury.gov daily data. The duration rule of thumb predicts that a freshly issued 30-year par bond would have lost roughly 15.3 × 0.54% ≈ 8.3% of its price over that window. Older long bonds with even higher durations lost more. That isn’t a market accident or a credit problem — it’s the mechanical, unavoidable arithmetic of long duration meeting a rising yield curve. We covered the macro of this move in our piece on the May 2026 long-bond selloff; duration is the mechanical reason the price action was so violent.
| Treasury par bond | Coupon / YTM | Macaulay duration (yrs) | Modified duration | Approx loss if yields rise 100 bps |
|---|---|---|---|---|
| 2-year | 4.13% | 1.94 | 1.90 | −1.9% |
| 5-year | 4.27% | 4.55 | 4.46 | −4.5% |
| 7-year | 4.41% | 6.10 | 5.97 | −6.0% |
| 10-year | 4.56% | 8.14 | 7.96 | −8.0% |
| 20-year | 5.06% | 12.80 | 12.49 | −12.5% |
| 30-year | 5.07% | 15.72 | 15.33 | −15.3% |
Visualizing duration: shorter bonds barely budge, long bonds get smashed
The shape is the entire point. A 100-basis-point move costs the 2-year about 1.9% of its price. The same move costs the 30-year about 15.3%. Long bonds aren’t “riskier” in some vague sense — they have eight times the mechanical price response of the short end to the same change in yields. That is what duration measures.
Convexity: the second-order correction
Duration is a straight-line approximation. The true price-yield relationship is curved, or convex: as yields move further from the starting point, the linear duration estimate gets less accurate. The full second-order Taylor expansion is:
ΔPrice/Price ≈ − Dmod × Δy + ½ × Convexity × (Δy)2
Convexity is always positive for a plain-vanilla bond, and it always works in the holder’s favor: when yields fall, the price rises by more than duration alone predicts; when yields rise, the price falls by less. For small yield changes (under 25 bp or so), you can ignore convexity. For the kind of 100–200 bp moves the Treasury market has produced in recent years, convexity meaningfully softens the duration estimate and is worth including.
There are two real-world cases where convexity flips negative, and they matter:
- Mortgage-backed securities. When rates fall, homeowners refinance, and the MBS investor gets cash back at the worst possible time (when rates are lower). That capped upside makes MBS negatively convex over a range of rates.
- Callable corporate bonds. When rates fall, the issuer calls the bond. Same problem. This is one reason effective duration matters more than modified duration for credit portfolios.
Common mistakes
- Confusing maturity with duration. A 10-year coupon bond has a duration well under 10 years — closer to 8 at today’s yields. Maturity tells you when the principal comes back; duration tells you the cash-flow-weighted “center of gravity.”
- Thinking high coupons mean high duration. Higher coupons actually shorten duration, because more of your money comes back early. A high-yield bond with the same maturity as a Treasury usually has lower duration.
- Calling a bond fund “safe” without knowing its duration. An intermediate Treasury fund with 6-year duration and a long-Treasury fund with 17-year duration are not the same instrument. Their drawdown profiles in a rising-rate environment are nearly three times apart.
- Ignoring convexity for big yield moves. For 100–200 bp shifts, duration alone is too pessimistic on price losses and too pessimistic on price gains. Pros build the convexity term into their risk reports.
- Forgetting that duration changes over time. As a bond ages, both its maturity and its duration drop. As yields change, modified duration also changes (because the yield is in the denominator).
Where duration fits in the bigger picture
For a portfolio manager, duration is the lever that translates a view on rates into position size. If you think yields will fall 50 bps, a 10-year duration portfolio will gain about 5%; a 2-year portfolio will gain about 1%. Duration is also how risk teams convert disparate fixed-income positions into a single number — the DV01 (dollar value of a basis point) is simply the dollar P&L from a 1 bp parallel shift, and DV01 = price × modified duration × 0.0001.
For a retail investor, the practical takeaways are simpler. Check the duration on any bond fund you own. If you want low rate-sensitivity, stay short (BIL, SHV, SHY-style products have durations under 2). If you want maximum upside on a rate-cutting cycle, go long (TLT, EDV-style products have durations in the high teens to mid-20s). Don’t be surprised when those long products move 10% in a quarter — that’s the contract you signed when you bought duration.
Related concepts to learn next
- Key-rate duration: a vector of durations against each point on the curve, used when the yield curve doesn’t move in parallel.
- DV01 and PV01: the dollar-and-basis-point translation of duration, used by trading desks for hedging.
- The yield curve: how rates differ across maturities, and what shape changes (steepening, flattening, inversion) imply. See our walk-through of the Japan 40-year JGB record for a real example of duration meeting a curve repricing.
- Convexity hedging: how MBS portfolios buy options or Treasury futures to offset their negative convexity exposure — one of the recurring sources of bond-market volatility.
Sources
- FINRA — Bonds and Interest Rates (duration as a measure of sensitivity to a 1% change in rates)
- U.S. Department of the Treasury — Daily Treasury Par Yield Curve Rates, 2026
- Federal Reserve — Open Market Operations (FOMC target range, currently 3.50–3.75%)
- iShares 20+ Year Treasury Bond ETF (TLT) — fund page with current effective duration
- ECMSource — 30-Year Treasury Yield 5.13%: Long-Bond Selloff (May 2026)
- ECMSource — Japan 40-Year JGB Hits Record 4.04%: Why It Matters Globally
Disclosure: This article was produced with AI assistance and reviewed before publication. It is for informational purposes only and is not investment advice.