Discover how Option-Adjusted Spread helps investors measure relative value by adjusting for embedded option costs, plus learn practical modeling approaches and pitfalls for callable, putable, and MBS securities.
Option-Adjusted Spread (OAS) can sometimes feel like a puzzle piece you never knew you needed—until you realize how many curves, possible call or put features, and embedded options exist in the bond market. OAS is essentially the spread you’d see on a bond over a specified benchmark yield curve once you factor out the “noise” (or value) coming from any embedded options. The embedded option might be an issuer call feature, an investor put, or more complex forms of optionality such as those in mortgage-backed securities (MBS).
Why does this matter? When a bond includes an option—for instance, a call option that allows the issuer to redeem the bond early—part of that bond’s yield compensates the investor for the likelihood that the issuer will call the bond. This can skew the apparent spread. OAS aims to “adjust” for that, helping you compare bonds on a more level footing.
You might be thinking: “That sounds fancy, but is it really a game changer?” Well, yes. Without removing the embedded option effect, comparing yields across various fixed-income instruments becomes a bit like comparing apples to apple pies—both share some similarities, but the extra ingredients can distort your sense of value.
Let’s say you’ve got a callable bond. The call feature is valuable to the issuer because they can refinance the debt at more attractive rates if interest rates decline. From your perspective, that call option means your bond might get “taken away” just as prices rise from falling interest rates— diminishing the potential upside. The nominal spread that’s quoted on this callable bond will reflect some premium for that risk. But if you want to compare that bond’s pure credit and liquidity spread with a similar non-callable bond, you have to adjust for that optionality. That’s where OAS comes in.
For putable bonds, it’s the investor who has the right to sell the bond back at a predetermined price before maturity. This is obviously a good thing for you as an investor. If yields spike or credit quality deteriorates, you can choose to “put” the bond, limiting your downside. The nominal yield on a putable bond might be lower than a straight bond because the investor owns a valuable put option. But again, to compare apples to apples, you want to adjust for that embedded put. That adjusted spread is the OAS.
Finally—my favorite and probably the most storied example: mortgage-backed securities. Homeowners have the right to prepay their mortgages at any time, effectively embedding a call option in MBS. This phenomenon complicates MBS pricing because it depends on consumer behavior, market interest rates, and a variety of other dynamic factors. OAS models are widely used here to account for all possible prepayment scenarios so you can figure out the “true” spread you’re receiving for taking on MBS credit and liquidity risks.
To visualize exactly what’s going on, you can think of a bond’s yield as being composed of a few building blocks:
(1) The risk-free rate (e.g., from a Treasury curve) (2) A credit spread (compensation for default or downgrade risk) (3) A liquidity or marketability premium (4) Embedded option costs or benefits (call, put, prepayment, conversion, and so on)
When you remove that option cost from (4)—perhaps through a binomial tree valuation or a Monte Carlo simulation—what remains is your OAS. This approach isolates the portion of the yield spread that compensates you for risk and liquidity, rather than for the potential gains or losses from the embedded option.
Modeling OAS can get quite technical, but the concept remains straightforward: you generate many potential interest rate paths, project the bond’s cash flows (considering if the option will be exercised in each scenario), and discount those cash flows back at each node using the appropriate risk-free rate plus the OAS.
A typical binomial tree for interest rates might look like this:
graph LR
A["Starting Rate <br/>Node 0"] --> B["Rate Up <br/>(Node 1U)"]
A["Starting Rate <br/>Node 0"] --> C["Rate Down <br/>(Node 1D)"]
B["Rate Up <br/>(Node 1U)"] --> D["(Node 2UU)"]
B["Rate Up <br/>(Node 1U)"] --> E["(Node 2UD)"]
C["Rate Down <br/>(Node 1D)"] --> F["(Node 2DU)"]
C["Rate Down <br/>(Node 1D)"] --> G["(Node 2DD)"]
At each node, you forecast what the bond’s cash flow would be. For a callable bond, the issuer might call at a certain node if interest rates have fallen sufficiently. For a putable bond, the investor might put the bond back if rates have risen enough or if credit spreads widened. You discount each possible path’s cash flows at the appropriate short rate (plus the trial OAS) and iterate until the present value equals the bond’s observed market price. The OAS that equates the model’s price to the bond’s actual market price is, in fact, the bond’s “option-adjusted” spread.
While a binomial tree is relatively intuitive, Monte Carlo simulations can capture interest rate paths in a more flexible and continuous manner. You might specify a volatility parameter σ and a drift μ for interest rates, then simulate thousands of possible rate paths. You estimate bond cash flows (again, embedding optionality logic at each step) and discount them back to present.
If you ever open up a large fixed-income analytics platform—and see it chugging away for a while—it’s probably crunching these thousands (or millions) of paths to figure out how embedded options affect your bond’s fair price, and from that, it extracts the OAS.
One personal anecdote: I recall the first time I tried a Monte Carlo approach for an MBS portfolio. I didn’t realize how computationally heavy it would be—my old laptop nearly melted! But the payoff was that I got to see which MBS tranches offered a higher OAS relative to their credit risk. That’s powerful, especially if you manage a multi-sector bond portfolio and need to fine-tune your exposures.
It’s easy to get lost in the universe of bond spreads, so let’s quickly contrast three common measures:
• Nominal Spread: A simple difference between the bond’s yield and a comparable benchmark yield (like a Treasury or LIBOR-based swap yield), but it doesn’t adjust for shape of the yield curve or embedded options.
• Zero-Volatility Spread (Z-Spread): The constant spread added to each point of the spot yield curve to discount a bond’s cash flows back to today’s price. Z-spread does not specifically remove the value of any embedded option. If you’re dealing with a callable or putable bond, the Z-spread alone can be misleading.
• OAS: Takes a similar idea to the Z-spread but specifically adjusts for the bond’s embedded optionality. OAS is what you’d see if the embedded option did not exist.
In general, for a callable bond in a stable interest rate environment, you might have:
Nominal Spread > Z-Spread > OAS
That’s because the nominal spread might overstate how much compensation you’re actually getting once the call option’s negative effect is stripped out.
OAS is highly sensitive to volatility assumptions—you tweak vol up or down, you’ll tweak the option value, which in turn changes the OAS. Higher implied or expected interest rate volatility usually makes the call option more valuable to the issuer (bad for the investor) or the put option more valuable to the investor (good for the investor). If your volatility assumption is off, your computed OAS might be misleading.
We can express the relationship between price, discount factors, and OAS as follows:
Where:
In practice, you compute \(\text{OAS}\) by solving the above pricing equation for OAS (given the bond’s market price).
If you enjoy a bit of coding, here’s a toy example in Python-like pseudocode that attempts to estimate a bond’s OAS using a two-step binomial interest rate model. Keep in mind, real implementations can be far more complex.
1import math
2
3def bond_price_with_oas(interest_rate_up, interest_rate_down,
4 up_probability, oas,
5 coupon_rate, face_value, periods):
6 # This is a highly simplified approach, ignoring complexities of call/put decisions.
7 # In a real setting, you'd embed logic to check if the bond is called or put at each node.
8
9 # Step 1: Calculate up and down interest rates for each period
10 # Step 2: Discount expected CFs at risk-free + OAS
11 # ... Real code would have branching logic.
12
13 # For demonstration, assume one period to maturity, no complicated option logic:
14 expected_next_period_value = (up_probability * (face_value + face_value*coupon_rate)) + \
15 ((1 - up_probability) * (face_value + face_value*coupon_rate))
16 discount_factor = 1 / (1 + (oas + (interest_rate_up if up_probability > 0.5 else interest_rate_down)))
17 return expected_next_period_value * discount_factor
18
19def main():
20 # Example parameters
21 up_rate = 0.04
22 down_rate = 0.02
23 p_up = 0.5
24 test_oas = 0.01
25 cpn_rate = 0.05
26 fv = 1000
27 T = 1
28
29 estimated_price = bond_price_with_oas(up_rate, down_rate, p_up, test_oas, cpn_rate, fv, T)
30 print("Estimated Bond Price with OAS of {:.2f}%: {:.2f}".format(test_oas*100, estimated_price))
31
32if __name__ == "__main__":
33 main()
Obviously, this is a toy. Real OAS modeling is more robust, especially when factoring in early exercise decisions for calls and puts at each node or simulating mortgage prepayments. But hopefully it gives a small sense of how the logic might look in code.
Mortgage-backed securities throw in the extra complexity of homeowner prepayments. You’re not just modeling an issuer’s rational decision to call; you’re modeling individuals who may (or may not) refinance their mortgage. As a result, MBS bond analytics can incorporate prepayment models that rely on historical data, projected rates, seasonality, and even borrower credit scores.
Often, MBS is quoted in terms of “spread to swaps” or “spread to Treasuries,” but you might prefer an OAS measure to see how richly or cheaply an MBS might be trading, once that prepayment option is dissected. When you see an MBS with a wide nominal spread, it doesn’t always mean the bond is a bargain—maybe it’s because everyone suspects faster prepayments that reduce your effective yield. OAS can help cut through that confusion.
• Overlooking Volatility: Underestimating interest rate volatility can lead you to overestimate OAS on callable bonds or underestimate it on putable bonds.
• Model Complexity: Too coarse a binomial tree or too few paths in Monte Carlo can produce inaccurate results.
• Implementation Errors: It’s easy to slip up in the accounting for partial-year coupon accrual or ignoring real-world constraints (like call notice periods).
• Inconsistent Data Sources: Using different yield curves or mismatch between your model’s discount rates and the bond’s currency region can distort OAS analysis.
A best practice is to continually validate your models with market quotes on more liquid bonds with embedded options. If your model’s OAS for a well-traded callable bond is significantly off from what dealers imply, then you might need to revisit your assumptions (especially volatility and path-dependence logic).
I still remember my first “aha” moment with OAS. While working on a small team analyzing corporate bonds, we had two similar names in the same rating category. One was a plain-vanilla bond, the other was callable. At first glance, the callable bond was priced cheaply and carried a pretty tempting nominal yield. But after we plugged it into an OAS model, it turned out that the embedded call option essentially gobbled up a good chunk of coupon potential. The bond wasn’t the “steal” it initially seemed. That was a lesson I never forgot: always check optionality.
• Be ready for both essay-type and item set questions asking you to calculate or interpret OAS.
• In practice problems, pay attention to the underlying interest rate model assumptions. If the question states a certain volatility level, that’s a hint you should incorporate it in your approach.
• Keep track of partial-year periods, frequency of coupon payments, and how the call/put schedule is structured—these details matter on a real exam.
• If you’re short on time in an exam scenario, a quick conceptual approach is to recall how OAS < Z-spread for a callable bond and OAS > Z-spread for a putable bond (generally). This can help eliminate wrong answers fast.
• For MBS, watch out for questions about prepayment assumptions and the differences between nominal spread and OAS-based valuation.
• Fabozzi, F. J. (2016). The Handbook of Mortgage-Backed Securities. Oxford University Press.
• Kalotay, A. J. (1982). “A Bond Valuation Formula That Incorporates Option Features.” Financial Analysts Journal.
• CFA Program Curriculum (Level I), Readings on OAS and Embedded Options in Fixed Income.
Important Notice: FinancialAnalystGuide.com provides supplemental CFA study materials, including mock exams, sample exam questions, and other practice resources to aid your exam preparation. These resources are not affiliated with or endorsed by the CFA Institute. CFA® and Chartered Financial Analyst® are registered trademarks owned exclusively by CFA Institute. Our content is independent, and we do not guarantee exam success. CFA Institute does not endorse, promote, or warrant the accuracy or quality of our products.