Duration: how to measure a bond's true risk
The answer
Duration is a single number that tells you roughly how much a bond or a bond fund will lose in price if interest rates rise by 1 percentage point. A duration of 7 means a 1% rise in yields costs you about 7% of your money. It is printed on every debt fund factsheet and almost nobody looks at it.
Why this costs you money
This is the most valuable idea in this cluster, so here it is without decoration.
A debt fund is not a fixed deposit with a better label.
A fixed deposit has a contract: you put in ₹5 lakh and you get back ₹5 lakh plus interest on a stated date. A debt fund has no contract. It holds bonds whose prices change every day, and you receive whatever the units are worth when you sell. If rates rise 2%, a fund with a duration of 8 loses roughly 16%.
Almost every Indian investor in a debt fund could tell you its past return. Almost none could tell you its duration. The past return describes what already happened. The duration describes what can happen next.
The cost is concrete. In 2022 the iShares 20+ Year Treasury Bond ETF, which holds only US government bonds, fell 31.41%. Its effective duration is around 15 years. Long US government yields rose roughly 2 percentage points that year. Multiply 15 by 2 and you get about 30.
Everyone who lost money in long bonds in 2022 was holding a document that told them, in advance, how much they could lose.
How it works
Start with the word, because it is badly chosen and causes most of the confusion.
Duration is measured in years, but it is not how long the bond lasts. It measures when you get your money back on average, weighted by how much arrives at each point.
Take 2 bonds, both maturing in 10 years. Bond 1 pays a 9% coupon, so by year 5 you have collected 45% of the face value. Bond 2 pays no coupon: you get one payment, the full face value, in year 10. Bond 2's money comes back much later on average, so it has a longer duration and will move far more when rates change.
That gives you the 3 rules that generate all the behaviour.
- Longer maturity means longer duration. More payments are further away.
- A higher coupon means shorter duration. More money returns early.
- A zero-coupon bond has a duration exactly equal to its maturity.
Now the number you actually use. Factsheets show 2 versions. Macaulay duration is the weighted average time to receive your money, in years. Modified duration answers the money question:
Price change ≈ − modified duration × change in yield
A fund with a modified duration of 4.5 loses about 4.5% if yields rise 1%, and gains about 4.5% if yields fall 1%. For a 0.5% move, halve it. For a 2% move, double it.
That is the whole calculation, and it is a multiplication. No calculus, no formula sheet, no spreadsheet.
Why "roughly". The relationship is a curve, not a straight line, and the curve is called convexity. It bends in the holder's favour. When yields fall a long way the price rises more than duration predicted; when yields rise a long way it falls less. So the simple multiplication is accurate for small moves and slightly pessimistic for large ones, which is a comfortable direction for an error to run.
What it tells you, and what it does not
Duration tells you about one risk only: interest rates.
It says nothing about default. A fund can have a very short duration and still lose 15% because a borrower failed. Duration and credit quality are 2 separate axes, and SEBI's Potential Risk Class grid exists because they are separate.
Duration also does not tell you direction. It is a sensitivity, not a forecast. A duration of 9 is not a bad thing to hold. It is a large bet, and whether it is a good bet depends on what rates do next. The multiplication is also less reliable for bonds with embedded options, where the issuer can repay early; factsheets for those portfolios show effective duration instead.
One more limit, and it is the useful one. Duration is also a holding period. Hold a bond fund for a number of years equal to its duration, and losses from rising yields and gains from reinvesting at higher yields roughly cancel out. That is why "my bond fund fell, should I sell?" so often comes down to comparing the fund's duration with the years until you need the money.
The decision rule
Compare the fund's duration with the number of years until you need the money.
If your horizon is longer than the duration, a rate rise is not a problem. You will collect the higher yields for long enough to make it back.
If your horizon is shorter than the duration, you are exposed. A rate rise before your date is a loss you cannot wait out.
Unless you are deliberately taking a view on rates. Then a long duration fund is the right instrument, and you should size it as a bet, not as a savings account.
For money that must be certain: under 6 months wants a duration under 0.5; 6 months to 2 years wants 0.5 to 1.5; 2 to 4 years wants 2 to 3.5, or a target maturity fund; beyond 5 years you can take a longer duration for the extra yield.
Try this now
This is the action that matters most in this cluster. Five minutes, on your own fund.
- Find the monthly factsheet for the debt fund you hold. Search the exact fund name plus "factsheet". US bond funds and ETFs publish the same figures.
- Find the box of portfolio statistics, usually near the holdings. Write down 3 numbers:
- Average maturity (US funds call this weighted average maturity)
- Modified duration (US funds usually show effective duration)
- Yield to maturity or YTM
- Do the multiplication.
Modified duration × 1 = your approximate percentage loss if yields rise 1%.Then do it for 2%. - Convert it to money. Multiply that percentage by the amount you hold. Write down the rupee or dollar figure. Not the percentage. The money.
- Compare the duration with the number of years until you need this money.
What you should see. A liquid fund will show a duration under 0.2, and a 1% rate move barely registers. A short duration fund shows 1 to 2.5. A corporate bond fund often shows 3 to 5. A gilt fund or long duration fund can show 7 to 9, and a constant maturity gilt fund more.
The figure from step 4 is the one to keep. For many readers it is the first time they have seen, in money, what their safe investment can lose in a year when nothing goes wrong. If step 5 shows a duration longer than your horizon, you have found a real mismatch before it cost you anything.
Three real cases
1. Silicon Valley Bank, 10 March 2023 (United States) — a duration failure, not a credit failure SVB took in a very large volume of deposits during 2020 and 2021 and invested them in long-dated US government and government-backed bonds. At the end of 2022 it held $91.3 billion of these in a held-to-maturity portfolio, against total assets of about $209 billion. Classified as held to maturity, their falling market value did not have to appear in reported profits. The unrealised loss was more than $15 billion. In March 2023 the bank sold its available-for-sale securities to raise cash and realised a $1.8 billion loss. On 9 March 2023, $42 billion was withdrawn in a single day. The bank was closed on 10 March 2023. Not one of those bonds defaulted. They were US government obligations. The bank failed because it had bought long duration with short-term money, and the duration was real even though the accounting had hidden it.
2. The United Kingdom, September and October 2022 — duration with borrowed money UK pension funds used liability-driven investment, or LDI, to match their long-dated obligations, borrowing to hold more long-dated gilt exposure than their cash allowed. After the government's growth plan on 23 September 2022, 30-year gilt yields rose roughly 200 basis points from the end of August to 27 September 2022. Mark-to-market losses triggered collateral calls. Funds sold gilts to meet them, pushing yields higher, triggering more calls. The Bank of England began emergency gilt purchases on 28 September 2022 and bought £19.3 billion before the operation ended on 14 October 2022. The gilts were never in danger of default. Duration and leverage together nearly broke a pension system.
3. Long-dated US Treasury funds, 2022 — the number was printed in advance The iShares 20+ Year Treasury Bond ETF fell 31.41% in calendar year 2022. Its effective duration is around 15 years and its weighted average maturity around 26 years. Long-dated US Treasury yields rose roughly 2 percentage points across the year. About 15 multiplied by about 2 gives about 30%, close to what happened. The fund did exactly what its own fact sheet said it would do.
The question that resolves it
Two debt funds have both fallen 4% this month.
A novice asks: which fund manager made the mistake?
An expert asks: what is each fund's duration, and did yields move, or did something default?
If yields rose 0.5% and both funds have durations around 8, neither manager did anything at all. The funds behaved arithmetically. If yields did not move, something in the portfolio has gone wrong, which is a different problem needing a different response.
Duration is what lets you tell those 2 situations apart in under a minute.
What would make this wrong
If duration did not measure interest rate sensitivity, funds with high and low duration would fall by similar amounts when yields rise. They do not. In 2022, short-duration US bond funds fell low single digits while long-duration funds fell 25% to 32%. The gradient tracked duration closely.
The honest limits are 4.
Duration only works for parallel shifts. It assumes all yields move by the same amount. Real curves twist, and then a single duration number predicts poorly.
Duration ignores credit entirely. A short-duration credit risk fund can lose far more than a long-duration gilt fund in the same month.
Published duration is a snapshot. A manager can change it between 2 factsheets. If the mandate allows a wide range, last month's number is a description, not a promise.
Convexity means the line is only approximate. For a 3% move the simple multiplication overstates the loss. It is a guide, not a guarantee.
In India
Every open-ended debt scheme publishes a monthly factsheet showing average maturity, Macaulay duration, modified duration, yield to maturity and rating profile. That is a SEBI requirement, not a courtesy. SEBI also defines debt fund categories by duration, which makes the labels unusually informative.
| Category | Macaulay duration of the portfolio |
|---|---|
| Overnight fund | 1 day |
| Liquid fund | Up to 91 days maturity |
| Ultra short duration | 3 to 6 months |
| Low duration | 6 to 12 months |
| Short duration | 1 to 3 years |
| Medium duration | 3 to 4 years |
| Medium to long duration | 4 to 7 years |
| Long duration | Over 7 years |
| Gilt fund | No limit, government securities only |
The category name is doing real work. "Short duration fund" is not marketing language. It is a rule the fund must follow.
The Potential Risk Class grid, in force since 1 December 2021, adds the second axis. Class I means Macaulay duration stays at or below 1 year, Class II at or below 3 years, Class III no limit. A fund labelled A-I is the most conservative cell on both axes.
One important Indian detail. Gilt funds have no credit risk and can have enormous duration. A constant maturity 10-year gilt fund holds long government bonds permanently and never shortens as it ages. It is the purest interest rate bet available to an Indian retail investor, and it is sold on the phrase "no credit risk" — true, and not the risk that will hurt you.
In the United States
US bond funds and ETFs publish effective duration rather than modified duration, because so much of the US market contains callable bonds and mortgage-backed securities where the borrower can repay early. For plain Treasuries the 2 numbers are nearly identical.
American investors have 3 tools for controlling duration that Indian investors largely do not. Individual Treasuries, at a $100 minimum on TreasuryDirect, with a known maturity date. Bond ladders, where one rung matures each year and is reinvested at current rates. And defined maturity bond ETFs, holding bonds that all mature in a stated year, then returning the money and closing.
One American trap is worth naming. Mortgage-backed securities lengthen their duration exactly when that hurts. When rates rise, refinancing stops and the bonds last longer, so duration grows into a falling market.
Where they differ, and what that tells you
The information is equally available in both countries. Only one set of investors can act on it without buying a different product.
An American who finds a fund's duration too long and wants certainty on a date can sell the fund and buy a Treasury maturing on that date, from the government, at a $100 minimum. The duration problem disappears.
An Indian has fewer routes. RBI Retail Direct allows direct G-Sec purchase with a ₹10,000 minimum, but very few use it and the secondary market for a small holding is thin. Target maturity funds and G-Sec or SDL index funds are the practical middle path.
What that tells you is a specific instruction for Indian readers. Your duration decision is mostly made at the moment you choose the fund category, because switching later costs tax. In India, moving between debt funds is a taxable event: for units bought on or after 1 April 2023, gains on specified mutual funds are treated as short-term capital gains and taxed at your income slab rate, however long you held them.
An American switching inside a retirement account pays nothing to correct a duration mistake. An Indian pays tax for it. So the check belongs at the beginning, not after the loss.
Carry this
- Duration × the change in yield = your approximate percentage gain or loss.
- Duration is measured in years and is not maturity. It is when your money comes back, on average.
- Compare the fund's duration with the years until you need the money. That one comparison answers most debt fund questions.