Convexity of Credit Spreads

What Is Convexity of Credit Spreads?

Convexity of Credit Spreads describes the second-order, nonlinear sensitivity of a credit instrument's price to changes in its credit spread, analogous to the convexity concept in rates markets but applied to the spread dimension of a bond's yield. In standard fixed income analysis, spread duration (also called DV01 per basis point of spread) captures the first-order — linear — price sensitivity to spread movements. But credit instruments exhibit meaningful nonlinearity: a 100bps widening from tight levels (say, 50bps to 150bps) has a materially different price impact than a 100bps widening from distressed levels (say, 500bps to 600bps). This is because as spreads widen, the expected loss component embedded in the bond's pricing accelerates in a concave fashion relative to the bond's residual value, and the effective duration of the instrument shortens as it approaches distressed pricing — compressing price sensitivity at very wide levels. The convexity profile is particularly pronounced in high-yield bonds, leveraged loans, CDS tranches, and CLO equity and mezzanine tranches, where the transition between par-pricing and distressed-pricing regimes is rapid and nonlinear.

Why It Matters for Traders

For credit portfolio managers, ignoring spread convexity leads to systematic underestimation of tail loss in stress scenarios. A portfolio reporting a spread duration of 4 years at current tight spreads may face dramatically higher dollar losses per basis point of widening during a credit selloff than the linear duration estimate implies, because the portfolio composition itself shifts as weaker credits transition to distressed pricing. In structured credit, CLO tranches exhibit extreme spread convexity: equity tranches can move from par to near-zero quickly once default rates breach model thresholds, while senior AAA tranches exhibit near-linear behavior. During credit cycle turns, this convexity creates asymmetric P&L profiles — short credit positions through CDS carry a built-in positive convexity profile (limited loss premium paid, uncapped gain on defaults), while long credit positions in tight-spread environments embed significant negative convexity risk.

How to Read and Interpret It

Practitioners measure credit spread convexity by computing the second derivative of price with respect to spread — often expressed as the change in spread DV01 per 100bps of spread movement. A practical rule: investment-grade bonds at 80–120bps spreads typically exhibit modest convexity; high-yield bonds at 400–600bps show substantially higher convexity; distressed bonds at 1,000bps+ exhibit lower convexity as prices compress toward recovery value floors. Traders should watch: (1) spread-duration ladders that show how portfolio DV01 shifts as spreads widen 100bps, 200bps, and 300bps; (2) option-adjusted spread (OAS) changes relative to model predictions during volatile periods; and (3) the CDS basis relative to cash bond spreads, which can signal convexity mismatches between hedges and exposures.

Historical Context

During the March 2020 COVID shock, US investment-grade corporate spreads (as measured by the ICE BofA IG index) widened from approximately 100bps in late February to over 370bps by March 23 — a 270bps move in under four weeks. Portfolio managers running linear spread-duration models severely underestimated actual mark-to-market losses, as the convexity of spread widening at those magnitudes was substantial. The IG index lost approximately 12–14% in price terms, far exceeding what a simple DV01 × spread-move calculation would have projected. Similarly, during the 2008 GFC, high-yield spreads reached 1,800–2,000bps, at which point the convexity relationship essentially collapsed as pricing transitioned fully to recovery-value analysis rather than spread-duration arithmetic.

Limitations and Caveats

Spread convexity models are highly sensitive to the assumed recovery rate, which itself is uncertain and procyclical — recoveries tend to fall precisely when convexity matters most, in broad credit selloffs. Liquidity convexity can dominate spread convexity in stress periods: bid-offer spreads widen dramatically, making model convexity estimates irrelevant for execution purposes. Additionally, embedded call options in investment-grade and high-yield bonds introduce negative convexity at tight spreads through the call feature, complicating the pure spread convexity analysis.

What to Watch

• HY spread levels relative to historical convexity inflection points around 400–500bps, where spread duration sensitivity accelerates
• CLO tranche pricing divergence as a leading indicator of nonlinear credit stress
• Excess bond premium spikes that signal convexity-driven overshooting
• CDS-bond basis dislocations that indicate hedges are mispricing the convexity profile of underlying cash instruments