Callable Bonds And Option-Adjusted Spread

A company issues a bond it can redeem early, at its own discretion, if rates fall enough that refinancing at a lower rate makes sense — the same reason a homeowner refinances a mortgage. From the investor's side, that's a plain bond with a call option sold to the issuer bolted on, and the investor is paid for selling that option through a higher coupon than an otherwise-identical non-callable bond would offer. The trouble is that if you just compute this bond's yield and compare its spread over Treasuries to a plain bond's spread, you're comparing a number that mixes two very different things — genuine credit compensation, and compensation for having sold an option — as if they were one thing. Option-adjusted spread (OAS) is the fix: value the option separately, subtract it out, and what's left is the spread that's actually comparable to a non-callable bond's.

Splitting the spread

$$\text{Callable bond price} = \text{Straight bond price} - \text{Call option value}.$$

Working this in spread terms: the callable bond's naive yield spread over Treasuries (the "Z-spread") is larger than its OAS by roughly the option's value expressed as a spread — $\text{OAS} = \text{Z-spread} - \text{Option cost (in spread terms)}$. In plain English: part of the extra yield a callable bond pays over Treasuries isn't compensation for credit risk at all, it's compensation for having sold the issuer a call option; OAS removes that option-driven part and leaves the piece attributable purely to credit and liquidity, the only piece that's fair to compare against another bond's spread.

Worked example 1 — decomposing one bond's spread

A callable corporate bond trades at a Z-spread of 180 basis points over Treasuries. A model (say, Hull-White, calibrated to the swaption vol cube) values the issuer's embedded call option as being worth 55 basis points of spread. OAS $= 180 - 55 = 125$bp. If a comparable non-callable bond from the same issuer trades at a spread of 120bp, the two are now genuinely comparable: 125bp versus 120bp says the callable bond is priced almost identically for credit risk once the option is stripped out — the earlier 180bp-vs-120bp comparison would have wrongly suggested the callable bond was 60bp cheap.

Worked example 2 — same Z-spread, different OAS

Two callable bonds from different issuers both quote a 200bp Z-spread. Bond A has low coupon volatility exposure and a call option valued at only 20bp; Bond B is deep in-the-money on its call (rates have fallen a lot since issuance) with an option valued at 90bp. OAS(A) $= 200 - 20 = 180$bp; OAS(B) $= 200 - 90 = 110$bp. Despite identical headline spreads, Bond A is compensating the investor 70bp more for credit risk than Bond B — Bond B's high headline spread is mostly paying for the valuable option the investor sold, not for taking on more credit exposure.

The call the bond investor has implicitly sold has this exact payoff shape from the issuer's side — the issuer "calls" (redeems early) when the bond's value has risen enough that refinancing cheaper is worth it, capping the investor's upside the same way a written call caps a stock investor's.

What this means in practice

Portfolio managers use OAS, not raw yield spread, to compare callable and non-callable bonds on equal footing, and OAS-based effective duration (how much the bond's price actually moves per basis point, accounting for the fact that the call probability itself changes as rates move) is what's used for hedging, not the bond's stated maturity duration.