Two gilts, the same £100 of nominal, wildly different risk. Yield only tells half the story. The 27 August gilt report priced 104 gilts, and two of them make the point: the 0 3/8% Treasury Gilt 2026 (gross yield 3.37%, clean price 99.55) and the 0 5/8% Treasury Gilt 2050 (yield 5.76%, price 33.49). The 2050 pays 2.4 percentage points more — but “how much risk am I carrying” is a different question from “how much does it pay”. This post is about the two numbers that answer it — modified duration and DV01 — and what the gilt book actually looks like through them.
Duration: the percentage answer
Modified duration is the percentage price change for a 100bp move in yield — the first derivative of price with respect to yield, scaled. It is the standard risk measure because it is comparable across bonds: it says “if the yield on this bond moves 1%, how much does my money move?”
The book spans the full range. From the 27 August report:
| Gilt | Price | Modified duration | 100bp move ≈ |
|---|---|---|---|
| 0 3/8% Treasury Gilt 2026 | 99.55 | 0.15 | 0.15% |
| 4¼% Treasury Gilt 2046 | 82.84 | 12.57 | 12.6% |
| 0 5/8% Treasury Gilt 2050 | 33.49 | 19.94 | 19.9% |
| 0 1/8% Index-linked Treasury Gilt 2073 | 44.20 | 43.80 | 43.8% |
The average modified duration across the 104 gilts is 11.88 years. Duration is driven by two things: time to maturity and coupon size. Low-coupon long bonds carry the most — the 0 5/8% 2050 has the longest duration in the conventional book, and the index-linked 2073 stretches past 43 years. A 100bp move in yields rarely happens to the whole curve in a single day, but the long end routinely does more in a week, which is why “long duration” is the first thing a book gets asked about.
DV01: the pounds-and-pence answer
DV01 is the change in the value of £100 nominal for a 1bp move in yield. It is approximately price × modified duration / 10000. The same three gilts:
- 0 3/8% 2026: 99.55 × 0.15 / 10000 = 0.0015 — the report’s figure
- 4¼% 2046: 82.84 × 12.57 / 10000 ≈ 0.104 — the report says 0.1054
- 0 5/8% 2050: 33.49 × 19.94 / 10000 ≈ 0.067 — the report says 0.0672
The counterintuitive bit is right there: the 2050 has the longest duration in the conventional book yet a DV01 barely half the 2046’s. Duration says “20% per 100bp”; DV01 says “£0.067 per £100 nominal per basis point”. Both are true — they answer different questions. The 2050 trades at 33.49, so there is less money at risk per £100 of nominal; the 2046, priced at 82.84, has far more capital exposed.
Which one do you look at?
- Duration when the question is proportional risk — “how many percent do I move per percent of yield” — or when the position is measured in market value.
- DV01 when the question is pounds — sizing a hedge or adding up book exposure. A hedge must match the DV01 of the position, not its duration; matching duration with a bond at a different price level hedges the wrong amount of money.
- DV01 per £100 of market value (dv01 ÷ clean price) when comparing risk per pound actually deployed. Highest in the book: the 2073 index-linked gilt at roughly 0.44 per £100 of market value per bp — almost £1 of risk per £100 of value for a 2bp move. Lowest: the 2026 conventional at 0.0015, barely a rounding error.
That ratio is the honest “how fast does each £100 I actually have at risk move” — it removes the price-level distortion and shows why the ultra-long index-linked tail is where the book’s real volatility lives.
The book in one number
Sum the DV01 column and holding £100 nominal of every one of the 104 gilts leaves you with a book worth about 8.45 per basis point of parallel shift: a 1bp move moves it £8.45, a 10bp move £84.45. That is not a market-value-weighted portfolio number — it is the arithmetic of the curve’s aggregate sensitivity, and it is a useful sanity check on “is this book big-risk or small-risk” that survives the fact that long bonds trade at very different prices than short ones.
Putting it to work
The engine computes modified duration and DV01 for all 104 gilts every day — they are part of the same analytics pipeline that produces gross and post-tax yields, sitting alongside the wider gilt analytics story. Two habits make the numbers useful rather than decorative: rank the book by DV01 per market value, not by duration, when asking which gilts dominate your risk; and express hedges in DV01 terms, so the size of the hedge falls out of the arithmetic instead of a guess. Duration tells you where the risk is; DV01 tells you how much of it there is — the daily report carries both, live in the gilt archive.
If you’re building a fixed-income analytics stack and want the risk numbers right the first time, get in touch.
