A bond with a modified duration of 6.5 years rises about 6.5% when market yields drop by one percentage point, and falls about 6.5% when they climb by the same amount. The "10-year" label on the product name tells you none of this. What determines how violently your bond moves when rates shift is duration, not maturity.
Duration, as Wikipedia defines it, is the weighted average maturity of a bond's cash flows and a measure of price sensitivity to interest rate changes. Frederick Macaulay devised it in 1938; dividing that figure by (1+r) gives modified duration. Three things decide the number: maturity, coupon rate, and market yield. That is why duration belongs ahead of maturity on your checklist.
Why identical maturities carry different durations
Duration falls short of maturity because of the coupons paid along the way. A three-year bond with an 8% coupon does not return your money in one lump at year three — it hands back interest in years one and two. Quarterback Asset Management calculates the duration of exactly this bond at 2.78 years, shorter than its three-year maturity. Only a zero-coupon bond, which pays nothing until the end, has duration equal to maturity.
The higher the coupon, the earlier the recovery, and the shorter the duration. Holding market yield at 8% and varying only the coupon on a three-year bond, the Macaulay and modified durations work out as follows. Each cash flow is discounted at 8% to get its present-value weight, and the years are averaged using those weights.
| Coupon (%) | Macaulay duration (yrs) | Modified duration (yrs) | Price at +1pp yield (%) |
|---|---|---|---|
| 0 (zero-coupon) | 3.000 | 2.778 | -2.78 |
| 3 | 2.907 | 2.691 | -2.69 |
| 5 | 2.853 | 2.642 | -2.64 |
| 8 | 2.783 | 2.577 | -2.58 |
| 12 | 2.705 | 2.505 | -2.51 |
All five rows share the same three-year maturity, yet price sensitivity spreads from 2.78% to 2.51%. The 0.27 percentage point gap looks trivial only because the maturity is short. Stretch the maturity and the same logic amplifies several times over.
Maturity tells you when it ends. Duration tells you how much it shakes until then.
How far does your bond move on a 1pp shift?
The working approximation is simple: price change ≈ -modified duration × yield change. The minus sign is there because yields and bond prices move in opposite directions. Quarterback calculates that a bond with duration 2.735 loses about 2.49% when market yields rise from 10% to 11%.
Feed durations and yield moves into that formula and you get a table showing what each product type asks you to stomach. The values below cross five representative duration buckets with four yield scenarios.
| Modified duration (yrs) | Yield -1.0pp | Yield -0.5pp | Yield +0.5pp | Yield +1.0pp |
|---|---|---|---|---|
| 2 (short) | +2.0% | +1.0% | -1.0% | -2.0% |
| 4.5 (intermediate) | +4.5% | +2.3% | -2.3% | -4.5% |
| 6.5 (10-year government type) | +6.5% | +3.3% | -3.3% | -6.5% |
| 9 | +9.0% | +4.5% | -4.5% | -9.0% |
| 17 (ultra-long government type) | +17.0% | +8.5% | -8.5% | -17.0% |
The two right-hand columns are the point of the exercise. Anyone buying ultra-long bonds on a rate-cut thesis reads only the left half. But that same product sits on an 8.5% loss the moment yields move half a point the other way. Duration is a multiplier on gains and losses alike. This is also where individual-investor treasury bonds, whose coupon locks in if you hold to maturity, part ways with tradable bond products. Hold-to-maturity never needs this table; sell-before-maturity is nothing but this table.
Where the approximation breaks: convexity
The formula is an approximation. The real relationship between bond price and yield is a downward-convex curve, not a straight line, and that curvature is called convexity. Convexity means price gains when yields fall exceed the straight-line estimate, while price losses when yields rise fall short of it.
So the table above holds up practically for yield moves of 0.25 to 0.5 percentage points, and the error grows visibly past one percentage point. Longer duration brings greater convexity, so the absolute error is largest in exactly the ultra-long bucket. That curvature is why, during sharp rate moves, holders report the drop was "smaller than the math said" or the rally "bigger than expected."
What actually happened in past easing cycles
Realized returns confirm the same structure. Economist Korea reported that during the easing cycle from September 2007 to December 2008, US 30-year Treasuries returned 41.4% against 23.8% for 10-year Treasuries. Same issuer, same instrument class — only the maturity bucket differed, and the results diverged by more than 1.7 times.
The same article notes that longer duration means larger price gains in a falling-rate environment for Korean bond products too. Read in reverse: the moment the rate-cut scenario is delayed or reversed, that multiplier runs in the other direction at full strength. The conditions that break this view are specific — cuts smaller than what the market has already priced, cuts arriving later than expected, or fiscal expansion lifting long-bond supply so that long yields decouple from the policy rate. None of the three does much to short duration; all three hit long duration head-on.
What to watch
- The product's stated duration — find the duration figure in the fund report or product disclosure, not the maturity number in the product name.
- Holding period versus duration — if you plan to hold to maturity, price swings are an accounting figure; if early redemption is possible, the table above is your actual P&L.
- The realistic range of yield moves — separate the policy rate, which moves in 0.25pp steps, from long-end yields, which do their own thing.
- The short-long yield spread — whether the curve flattens or steepens changes how long duration performs under the same cut.
- Costs beyond price — for bond funds, premium/discount to NAV and tracking error eat into realized returns independently of the duration math.