Skip to content

New to Omni?·Get 20% off your first purchase with code WELCOME20 at checkout.See plans

FRM® duration estimates: when does convexity matter?

Duration gives a local bond price estimate. Learn when curvature, changing cash flows or curve shifts make that shortcut unreliable, with a worked calculation.

ConceptOmni Curriculum Team6 min read

A blank bond-style certificate lies below a curved metal arc on a navy desk.
On this page

Duration is a useful first estimate of how a bond price responds to a small yield change, but it is a straight-line estimate of a curved relationship. Convexity improves that estimate when the yield move is larger and cash flows are fixed. If cash flows can change or different maturities move by different amounts, a simple duration-plus-convexity calculation may still answer the wrong risk question.

What duration estimates

For a conventional fixed-cash-flow bond, modified duration measures the local slope of price against the bond’s yield. A positive duration pairs with the familiar inverse relationship: higher yield implies a lower price, all else equal. The word “local” is essential. The estimate assumes the slope observed near the starting yield can be carried over the contemplated move. The CFA Institute refresher reading on duration distinguishes measures based on a bond’s own yield from measures based on a benchmark curve, and explicitly warns that price and yield are not linearly related.

ΔPP≈−DmodΔy\frac{\Delta P}{P}\approx-D_{\mathrm{mod}}\Delta y
Approximate proportional price change using modified duration and a decimal yield change.

In that expression, the yield change is a decimal, not a number of percentage points typed without conversion. An increase of one percentage point is 0.01. The result is a proportional price change; multiply by the starting price if the question wants currency. Duration also needs a matching definition of yield. A shift in the bond’s own yield is not automatically the same as a parallel shift in a benchmark curve plus an unchanged credit spread. The GARP FRM learning objectives identify the official current syllabus; candidates should use it to check the scope of the relevant Part I reading.

Why a tangent line becomes inaccurate

As yield changes, the present value of each future payment changes. For an option-free bond, the price–yield curve bends, while the duration estimate continues along a tangent drawn at the initial yield. Following a yield fall, the curve normally lies above that tangent, so duration alone understates the gain. Following a yield rise, the curve also lies above the tangent, so duration alone exaggerates the loss. This is a geometric statement about a positive-convexity bond, not a promise about the realised total return, which can include coupon income, spread changes and passage of time.

Convexity measures that curvature around the starting point. The usual second-order approximation adds a term based on the square of the yield change. Because the square is positive after either a rise or a fall, positive convexity lifts both duration-only price estimates. The CFA Institute Research Foundation treatment presents the duration and convexity terms together. It also makes clear that both are approximations: they summarise the curve near a starting point rather than repricing every cash flow at the new yield.

ΔPP≈−DmodΔy+12C(Δy)2\frac{\Delta P}{P}\approx-D_{\mathrm{mod}}\Delta y+\frac{1}{2}C(\Delta y)^2
Second-order price estimate for modified duration, convexity and a decimal change in yield.

Work through a move in both directions

Assume a hypothetical option-free bond is priced at £100, has modified duration of 5 and convexity of 40. For a one-percentage-point yield rise, the decimal yield change is positive 0.01. Duration alone estimates a 5% price fall, or a price of £95. The convexity correction is one half times 40 times 0.01 squared, which is 0.002, or 0.2% of the starting price. The two-term estimated decline is therefore 4.8%, giving an estimated price of £95.20. The exact repriced bond could differ because the formula stops after the curvature term.

Now use an equally sized yield fall. The duration component becomes a 5% gain because the sign of the yield change reverses. The convexity component remains positive 0.2% because the change is squared. The two-term estimated price becomes £105.20, compared with £105 from duration alone. The asymmetric outcomes are the point: the estimated gain after a fall is larger than the estimated loss after an equal rise. If a candidate changes the sign of the convexity term on the down move, the result loses the defining effect of positive curvature.

The value of the correction depends strongly on the size of the shock. With the same hypothetical duration and convexity, a yield move one quarter as large gives a duration effect one quarter as large, while the convexity effect is only one sixteenth as large. That follows from the squared yield change. It explains why duration can be serviceable for a very small local move even though the price curve is never literally straight. It also explains why no single universal basis-point cutoff says when convexity “must” be included; the tolerance for approximation error depends on the bond and the decision.

An embedded option changes the shape itself

The positive-convexity argument relies on fixed promised cash flows. If an issuer can call a bond, falling yields can make early redemption more likely. The bondholder then may lose some of the expected price gain because cash flows end earlier than the original maturity would suggest. CFA Institute research on callable bonds explains that an embedded call can lower duration and can produce negative convexity in a relevant yield region. It would be wrong to add a positive option-free convexity number mechanically to a callable bond simply because the word “bond” appears.

A callable bond can also have a duration that changes markedly as rates move. If the option is far from exercise, its behaviour can resemble an ordinary bond more closely. If a rate fall brings exercise into view, the cash-flow pattern changes and the same local slope no longer describes the next move well. A risk measure that allows the cash flows to respond to interest rates is needed for that setting. The exam skill is to inspect the instrument before selecting a formula: identify who has the option, when it might be exercised and whether the bond’s projected payments can remain fixed in the scenario.

A curve move is not necessarily one yield move

A separate limitation appears when short- and long-maturity yields move differently. A single duration calculated for one yield cannot fully explain a steepening, flattening or twist of the term structure. The relevant cash flows may be discounted by different parts of the curve, so a bond with the same headline duration can respond differently depending on which maturities change. Convexity improves a calculation for curvature in one price–yield relationship; it does not tell you which points on a non-parallel yield curve moved. This is why the shock description belongs beside the risk measure, not as an afterthought.

Download CSV
Article data
Small move in a fixed-cash-flow bond’s yieldModified durationConfirm yield units and price base
Larger move in a fixed-cash-flow bond’s yieldDuration plus convexityReprice if precision matters
Callable bond near an exercise regionCash-flow-sensitive risk estimateTest how exercise changes payments
Non-parallel term-structure shiftCurve-specific sensitivitiesIdentify maturities that actually moved

Do not confuse the scale of a position with a new source of risk. A duration expressed as a percentage sensitivity can be multiplied by market value to estimate a currency change. A desk may instead use a money duration or value per basis point. That conversion does not repair a bad assumption about curved prices, optionality or the shape of the rate shock. The question “how much money?” comes after “which yield and which cash flows?” For Part I practice, describe the shock in words before choosing the calculation.

The candidate’s decision rule

On an unfamiliar FRM vignette, identify the bond’s cash-flow certainty, the initial and shocked yield, and whether the move is small enough for a local approximation to meet the question’s requested precision. For fixed cash flows and a larger stated move, calculate the duration effect, then add the convexity effect with the decimal yield change squared. For an embedded option, stop and consider cash-flow changes. For a curve twist, ask whether a single yield shock is defined at all. This sequence prevents a correct formula from being applied to an unsuitable instrument.

The FRM programme page sets this topic in its exam context, and our method offers a way to practise such classification. The useful lesson is that duration is a local slope, convexity describes how that slope changes, and neither statistic eliminates the need to state the scenario. Candidates who can explain the assumptions before calculating are better placed to spot an answer that is numerically neat but economically mismatched.

Sources

  1. 2026 FRM Learning Objectives | GARP · garp.org · Retrieved
  2. Yield-Based Bond Duration Measures and Properties | CFA Institute · cfainstitute.org · Retrieved
  3. Appendix C: Price Behavior of Fixed-Income Securities | CFA Institute Research Foundation · rpc.cfainstitute.org · Retrieved
  4. Using Duration and Convexity in the Analysis of Callable Bonds | CFA Institute · rpc.cfainstitute.org · Retrieved

Explore FRM study resources

Educational content, not investment advice. This article is written to help candidates prepare for professional exams. It is not a recommendation to buy, sell or hold any security, and it does not take your personal circumstances into account.

Omni Finance Academy is an independent exam-preparation provider and is not affiliated with or endorsed by CFA Institute, GARP, CAIA Association or CFP Board. CFA®, FRM®, CAIA® and CFP® are trademarks of their respective owners. Full disclaimer.