Convexity: Why Duration Alone Misjudges Big Rate Moves
Duration predicts a bond will lose exactly as much as it gains for equal moves up and down in yield, but real bonds do not behave that way. Convexity is the correction that explains the gap, and it quietly determines which bonds actually protect a portfolio in a severe rate shock.
The core mechanism: why the price-yield relationship curves
Duration, covered in detail elsewhere, is a straight-line approximation: it assumes that a bond's percentage price change is directly proportional to the change in yield, so that a 1 percentage point rise costs exactly as much, in percentage terms, as a 1 percentage point fall gains. If you plot a bond's actual price against a range of possible yields, though, the relationship is not a straight line at all. It is a curve, bowed so that price falls a little less than the straight-line estimate predicts when yields rise, and rises a little more than predicted when yields fall. Convexity is the name for this curvature, and it is the second-order correction that makes a duration-based price estimate more accurate, especially for large rate moves.
The underlying reason for the curve is mathematical rather than mysterious. Bond price is the sum of future cash flows discounted back to the present, and the discounting formula itself is not linear in the yield, it involves the yield raised to increasing powers in the denominator for cash flows further in the future. Duration captures the slope of the price-yield curve at the bond's current yield, a first derivative in calculus terms, which is an excellent local approximation for small moves but increasingly inaccurate the further yields travel from that starting point. Convexity captures the curvature, the rate at which the slope itself is changing, a second derivative, and adding it to the duration estimate produces a far closer match to the bond's true price behavior.
For a conventional, option-free bond, this curvature works entirely in the investor's favor, which is why convexity is generally described as a desirable property. Because the curve bows in the direction that softens losses on the downside and amplifies gains on the upside, two bonds with identical duration but different convexity are not equally attractive: the one with higher convexity will outperform in both a large rate increase and a large rate decrease, since it loses less when rates rise and gains more when rates fall. This is not a free lunch in a strict sense, since higher-convexity bonds typically trade at a slightly lower yield to compensate, but it is a real, measurable difference in how a bond behaves under stress.
The math: two worked examples correcting duration's estimate
Worked example 1: adding the convexity correction to a large rate move. Suppose a bond has a modified duration of 8 years and a convexity of 90 (convexity is typically expressed in units that, when combined with the yield change squared, produce a percentage correction; the exact derivation is a standard fixed income calculation, but the application is straightforward). Yields rise sharply by 2 percentage points. The duration-only estimate of the price change is -8 x 2% = -16%. The convexity correction adds a term equal to 0.5 x convexity x (change in yield)^2 = 0.5 x 90 x (0.02)^2 = 0.5 x 90 x 0.0004 = 0.018, or +1.8%. Combining both terms, the corrected estimate of the price change is -16% + 1.8% = -14.2%. On a $50,000 position, the duration-only estimate predicts a loss of $8,000, while the convexity-adjusted estimate predicts a loss of $7,100, a difference of $900 that duration alone simply misses.
Worked example 2: comparing two bonds with equal duration but different convexity under a rate drop. Bond X has a modified duration of 6 years and convexity of 50. Bond Y has the same modified duration of 6 years but higher convexity of 130, achieved through a different mix of coupon and maturity. Suppose yields fall by 1.5 percentage points, a favorable move for both. The duration-only estimate for both bonds is identical: -6 x (-1.5%) = +9%. The convexity correction for Bond X is 0.5 x 50 x (0.015)^2 = 0.5 x 50 x 0.000225 = 0.005625, or +0.56%, bringing its total estimated gain to 9% + 0.56% = 9.56%. The convexity correction for Bond Y is 0.5 x 130 x 0.000225 = 0.014625, or +1.46%, bringing its total estimated gain to 9% + 1.46% = 10.46%. On identical $30,000 positions, Bond Y gains roughly $3,138 versus Bond X's $2,868, a difference of about $270 attributable entirely to convexity, despite both bonds having started with the exact same duration.
What the evidence shows about convexity's real value
Convexity's practical value shows up most clearly during periods of unusually high interest rate volatility, precisely the environments in which the linear duration estimate is least reliable. Fixed income desks that manage large government and corporate bond portfolios routinely report tracking error between realized returns and duration-only forecasts that widens noticeably during sharp, fast rate cycles, and narrows again once rates stabilize, a pattern entirely consistent with convexity effects being small in calm markets and material in volatile ones. This is also why bond portfolio managers do not treat duration as a complete risk measure on its own; standard fixed income risk reporting includes both duration and convexity precisely because relying on duration alone has repeatedly produced forecast errors large enough to matter during major rate cycles, including the sharp global rate increases of the early 2020s.
Convexity is also related mathematically to a bond's yield level itself, a subtlety worth flagging for anyone comparing bonds across very different interest rate environments. For a given maturity and coupon, a bond's convexity tends to be somewhat higher when the prevailing yield level is low than when it is high, since the curvature of the price-yield relationship is more pronounced near the lower end of the yield scale. This means the same bond can exhibit meaningfully different convexity depending on the overall rate environment it is issued or held in, a detail that matters more to institutional portfolio managers constructing precise hedges than to an individual investor, but one worth knowing exists so that a convexity figure quoted at one point in time is not assumed to be a permanently fixed property of the bond.
It is important to note that not all bonds carry the favorable convexity described above. Callable bonds, which give the issuer the right to repay the bond early if rates fall, and mortgage-backed securities, where underlying homeowners tend to refinance and prepay their mortgages when rates drop, both exhibit what is called negative convexity over some range of yields: their price gains are capped when rates fall, because the issuer or the underlying borrowers exercise their option to prepay, cutting off the very appreciation that a conventional bond would deliver. This is a well documented and economically intuitive result, since the option being exercised against the bondholder has real value to the issuer, and that value is effectively subtracted from the bondholder's convexity benefit. An investor buying a mortgage-backed security or a callable corporate bond for its yield needs to understand that its behavior under falling rates will disappoint relative to a same-duration Treasury bond precisely because of this effect.
Applying convexity in a real portfolio
For most individual investors holding diversified bond index funds, convexity is not something that requires active daily management, since fund-level convexity is disclosed and reasonably stable, and the effect is secondary to the much larger question of overall duration exposure covered elsewhere. Where convexity becomes practically relevant is in two specific situations. First, when comparing two bonds or bond funds with similar duration and similar credit quality, all else equal, the one with higher convexity is the better holding, since it behaves better in both a rally and a selloff, a distinction worth checking when a fund's factsheet reports both figures side by side. Second, and more important for higher earners building meaningful fixed income allocations, understanding negative convexity matters directly when evaluating callable corporate bonds, agency mortgage-backed securities, or structured notes marketed with an attractively high stated yield: that extra yield is frequently compensation for giving up the convexity benefit that a plain Treasury or investment-grade corporate bond of the same duration would provide, and the true risk-adjusted comparison needs to account for that trade rather than looking at yield alone.
Convexity also has a lesser-known but genuinely useful application for an investor deliberately constructing a portfolio to fund a specific future liability, similar in spirit to the duration-matching techniques used in immunization strategies. Because convexity measures curvature, a portfolio can be built with a duration that matches the target horizon while also carrying higher convexity than a single bond of that same duration would offer, achieved by combining a shorter and a longer bond rather than holding one bond in the middle, a technique sometimes called a barbell structure. The barbell approach captures a modest additional cushion against large rate swings around the matched horizon precisely because combining two more extreme maturities produces higher aggregate convexity than one bond sitting in between them, even when both structures start with an identical duration. This is a refinement worth knowing about, though for most individual investors a straightforward duration-matched bond ladder or index fund remains a perfectly reasonable starting point without needing to engineer a barbell explicitly.
A final practical point concerns how convexity shows up, or fails to show up, in everyday bond fund marketing. Yield and duration are almost always prominently disclosed, since they are simple, single numbers that investors reasonably ask about, but convexity is disclosed less consistently and understood even less widely, despite being freely available in most institutional-grade factsheets and increasingly in retail fund literature as well. An investor comparing two similarly yielding bond funds who takes the extra step of checking convexity, not just duration, is doing a genuinely more complete risk comparison than the large majority of retail investors ever perform.
Actionable breakdown
- Reading a bond or fund factsheet
- Check convexity alongside duration, not duration alone.
- Prefer higher convexity between two bonds of equal duration.
- Note whether a security is callable before assuming positive convexity.
- Evaluating yield versus risk
- Treat an unusually high yield as a convexity trade-off signal.
- Compare mortgage-backed securities against Treasuries of equal duration.
- Ask what happens to the bond if rates fall sharply.
- Building the fixed income sleeve
- Rely on duration for everyday sizing, convexity for stress scenarios.
- Diversify across issuers to avoid concentrated negative convexity.
- Revisit convexity assumptions after any major rate cycle.
Common pitfalls
Ignoring convexity entirely for large positions: duration alone is a fine approximation for small moves but can misstate risk meaningfully once rate changes exceed a percentage point or two.
Assuming all bonds have favorable convexity: callable bonds and mortgage-backed securities can carry negative convexity, capping upside precisely when an investor wants it most.
Chasing yield without pricing the option given away: a higher yield on a callable or mortgage security is frequently the market's price for the convexity the investor is sacrificing.
Overcomplicating small portfolios: for a simple diversified bond index fund position, obsessing over fund-level convexity differences is rarely worth the effort relative to getting overall duration and allocation right first.
The bottom line
Convexity is the curvature duration misses, and while it is a secondary consideration for most simple portfolios, it explains why some high-yielding bonds disappoint precisely when an investor needs them to perform.
All articles · Interest rate risk and duration · Active bond management · Bonds fundamentals · Duration (glossary)