Convexity - The Second-Order Curvature of Bond Prices

By EC Assets Research Team, Fixed Income Research · Published · Updated

Convexity: Convexity is the second-order measure of a bond's rate sensitivity - the curvature that duration's straight-line estimate misses. Positive convexity helps the holder symmetrically (more gain when yields fall, less loss when they rise); negative convexity, as in callable bonds and mortgages, does the reverse.

What Convexity Measures

The relationship between a bond's price and its yield is a curve, not a line. Duration is the slope of that curve at the current yield; convexity measures how quickly the slope changes. Together they estimate the price change for a change in yield Δy:

$$\frac{\Delta P}{P} \approx -D_{\text{mod}},\Delta y + \tfrac{1}{2},C,(\Delta y)^2, \qquad C = \frac{1}{P}\frac{d^2P}{dy^2}$$

For an ordinary bond without options the second term is positive whichever way yields move. Convexity therefore adds to the gain when yields fall and subtracts from the loss when they rise. It is the bond counterpart of an option's gamma.

Redington (1952) made convexity part of immunisation theory: an insurer whose assets match the duration of its liabilities is protected against small parallel moves only if the assets are also more convex than the liabilities.

How It Works

Convexity rises steeply with maturity, because the curvature of discounting grows with the square of time. For a zero-coupon bond with annual compounding it equals T(T + 1) / (1 + y)². At a 4 percent yield a 2-year zero has a convexity of 5.55, a 10-year zero 101.70 and a 30-year zero 859.84. Coupons lower convexity, as they lower duration, because they bring cash flows forward; a 10-year 4 percent coupon bond at par has a convexity of 80.75.

Spreading cash flows across maturities raises convexity for a given duration. That is the logic of a barbell, which combines short and long bonds, against a bullet concentrated at one maturity.

For bonds whose cash flows depend on rates, convexity is measured numerically. The bond is repriced with yields shifted down and up by a small amount, and effective convexity is (P down + P up - 2P) / (P × Δy²). Option-adjusted models use the same recipe with the embedded call or prepayment option revalued at each shifted yield.

Convexity is not free. Because a convex position benefits from volatility in yields, investors accept a lower yield on more convex bonds. Ilmanen (1995) shows this convexity bias pulls long-dated yields below where expectations and term premia alone would put them, and it grows with rates volatility, as measured by the MOVE index.

Worked Example

The 10-year bond paying an annual coupon of 4 and yielding 4 percent has a modified duration of 8.11 and a convexity of 80.75. For a 100 basis point move, the convexity term adds ½ × 80.75 × 0.01² = 0.40 percent. Repricing the bond exactly at each new yield:

Change in yield Actual price change Duration only Duration and convexity
+100 bp -7.72% -8.11% -7.71%
-100 bp +8.53% +8.11% +8.51%
+200 bp -14.72% -16.22% -14.61%
-200 bp +17.97% +16.22% +17.84%

Duration alone predicts symmetric moves; the convexity term recovers most of the asymmetry.

Now compare two portfolios with the same modified duration of 9.62. A bullet holds a 10-year zero with convexity 101.70. A barbell holds 71.4 percent of its value in a 2-year zero and 28.6 percent in a 30-year zero; its convexity is 249.63, about two and a half times as much. For a parallel move of 100 basis points either way, the barbell's extra convexity is worth about ½ × (249.63 - 101.70) × 0.01² ≈ 0.74 percentage points over the bullet. In return it usually yields less, so it lags if yields stay put.

When It Applies (and Limitations)

It is still an approximation. Duration plus convexity is a second-order expansion. It is very good for moves of a couple of hundred basis points on ordinary bonds but drifts for larger shocks and long maturities, where full repricing is needed.

Parallel moves only. Like duration, a single convexity number assumes the whole yield curve shifts together. A barbell's advantage can vanish if the curve twists against it.

Negative convexity. Callable bonds and mortgage-backed securities embed options sold by the investor. When rates fall, issuers call bonds and homeowners refinance, capping price gains and returning cash early, so the price-yield curve flattens or bends the wrong way. Their effective convexity must be computed with option-adjusted models.

Hedging feedback. Holders of large negatively convex portfolios hedge by buying duration as rates fall and selling it as rates rise, which can amplify moves in the government bond market itself.

Why It Matters for Institutional Investors

Accurate risk for large moves. In a sharp rate shock a duration-only estimate can be materially wrong, so risk systems for bond portfolios carry both measures.

A priced characteristic. Positive convexity costs yield; negative convexity earns it. Investors in mortgage-backed securities and callable bonds are paid a spread for selling convexity, and should know they are effectively short an option.

Barbell versus bullet. Choosing between them at the same duration is an explicit bet on rates volatility: the barbell wins if yields move a lot and loses if they stay still.

Liability-driven investing. Pension and insurance hedges must match convexity as well as duration, or the hedge drifts as rates move and leaves a residual exposure exactly when markets are volatile.

References

  1. Redington, F. M. (1952). Review of the Principles of Life-Office Valuations. Journal of the Institute of Actuaries, 78(3), 286-340.
  2. Ilmanen, A. (1995). Understanding the Yield Curve, Part 5: Convexity Bias and the Yield Curve. Salomon Brothers.
  3. Tuckman, B., & Serrat, A. (2011). Fixed Income Securities: Tools for Today's Markets (3rd ed.). Wiley.
  4. Fabozzi, F. J., & Fabozzi, F. A. (2021). Bond Markets, Analysis, and Strategies (10th ed.). MIT Press.

Frequently asked questions

What is the difference between duration and convexity?

Duration is the slope of the price-yield curve, the first-order estimate of how a bond's price responds to a change in yield. Convexity measures how that slope changes, the curvature. Adding the convexity term to the duration estimate gives a much more accurate price change for large moves in yield.

Why is positive convexity good for a bondholder?

Because it bends both outcomes in the holder's favour: when yields fall the price rises by more than duration predicts, and when they rise it falls by less. Investors value that asymmetry and accept a slightly lower yield to obtain it, particularly when rates are volatile.

What is negative convexity?

The unfavourable curvature of bonds with embedded short options, such as callable bonds and mortgage-backed securities. When rates fall, issuers call the bonds or homeowners refinance, capping price gains and returning cash early, so the holder gains less when yields fall than it loses when they rise. Investors demand extra yield for bearing it.

How is convexity like an option's gamma?

Both measure the curvature of a price response and both benefit from large moves. A positively convex bond, like a long-gamma option position, gains from volatility in yields whatever the direction; a negatively convex bond behaves like a short-gamma position and suffers from large moves.

When does convexity need to be taken into account?

For small moves duration is accurate enough. Convexity matters for large rate moves, where the duration estimate becomes materially wrong, for long-dated bonds, whose convexity is large, and for liability hedges, which must match convexity as well as duration to stay balanced when rates move.

Stay informed

Market commentary, firm news and research from EC Assets - direct to your inbox.