OAS (Option-Adjusted Spread)

OAS calculation involves Monte Carlo simulation with these steps:

1. Generate thousands of interest rate paths using short rate model (typically Hull-White or Black-Karasinski). Model must be calibrated to current term structure and swaption volatilities.

2. For each rate path, project cash flows using prepayment model. If rates fall, prepayment speeds increase (refinancing); if rates rise, speeds slow (extension). Apply prepayment model's response function to each scenario.

3. Discount cash flows on each path back to present using the risk-free rate plus a trial spread. Average the present values across all scenarios.

4. Iterate on the trial spread until average present value equals market price. This spread is the OAS.

The calculation requires: prepayment model, interest rate volatility assumptions, base curve (Treasury or SOFR), and pricing model for discounting. Changes to any input materially affect calculated OAS.

Three spread measures with different uses:

Nominal spread: Simple yield to maturity minus Treasury yield. Ignores both option risk and term structure. Fast to calculate but misleading for securities with embedded options or non-parallel cash flows. Example: 30-year MBS yielding 5.5% vs 30-year Treasury at 4.5% = 100bps nominal spread. But this doesn't account for prepayment risk.

Z-spread (Zero-volatility spread): Constant spread over spot rate curve that equates present value to price. Accounts for term structure but NOT options. Uses single deterministic cash flow projection. Example: same MBS has 85bps Z-spread when properly discounted using Treasury curve. Still ignores prepayment uncertainty.

OAS: Spread over spot curve accounting for embedded options via Monte Carlo. Example: same MBS has 45bps OAS after removing prepayment option value. The 40bps difference (85 - 45) represents the value of the prepayment option given to borrowers.

Rule of thumb: For securities with significant option risk (MBS, callable corporates), OAS most relevant. For securities with minimal options (CLO debt tranches, non-callable corporates), Z-spread sufficient and more stable.

OAS models also generate option-adjusted duration and convexity measures:

Effective duration: Measures price sensitivity to parallel yield curve shifts, accounting for changing cash flows as prepayments respond to rates. Calculated by shifting entire curve up/down 25bps, recalculating price via OAS model. Effective Duration = (Price if rates -25bps) - (Price if rates +25bps) / (2 × Price × 0.25%).

Example: 30-year MBS has 6-year effective duration despite 30-year maturity because high coupon means heavy prepayment expected, shortening average life. Modified duration would show 12+ years (wrong). Effective duration captures the prepayment response.

Effective convexity: Measures how duration changes as rates change (curvature). MBS exhibit negative convexity—when rates fall, duration shortens (prepayments accelerate); when rates rise, duration extends. This makes MBS underperform in both rally and sell-off scenarios, explaining persistent OAS compensation.

These metrics are critical for hedging. Hedging MBS with Treasury futures requires effective duration, not modified duration, or hedge ratio will be wrong and PnL will swing with rate moves.

OAS is not constant—it varies with market conditions and rate levels:

Falling rate environment: OAS tends to widen as prepayment risk increases. Investors demand more compensation for the increased likelihood of receiving principal at par when security is trading at premium. 2020 example: when 10-year Treasury fell to 0.50%, agency MBS OAS widened to 60-80bps from 30-40bps normal, reflecting prepayment fears.

Rising rate environment: OAS may tighten as prepayment risk diminishes and extension risk increases. Investors accept lower OAS because security unlikely to prepay. However, extension risk creates its own premium, so OAS tightening is limited. Also, rising rates often coincide with Fed QT and reduced MBS demand, offsetting the prepayment benefit.

High volatility environment: OAS widens because option value increases with volatility (options are worth more when uncertainty is high). March 2020: rate volatility spiked, MBS OAS blew out to 150+ bps even for agency paper, reflecting uncertainty about Fed actions and prepayment behavior.

Understanding these dynamics prevents confusion when OAS widens while credit quality improves—often it's reflecting increased option value or technical factors, not deteriorating fundamentals.

OAS accuracy depends entirely on model quality, creating significant model risk:

Prepayment model risk: If model underestimates prepayment speeds in falling rate environment, calculated OAS appears attractive but security will underperform. During 2003 refi boom, many prepayment models failed to capture speed of response, causing losses for MBS buyers relying on 'cheap' OAS signals.

Interest rate model risk: Choice of short rate model (Hull-White vs Black-Karasinski vs others) affects results. Must be calibrated to current swaption market. If calibration is stale, OAS calculation is wrong.

Path-dependency: MBS prepayment behavior is path-dependent (burnout effect—loans that didn't refi when rates initially fell are less likely to refi later). Most OAS models don't capture this well, overestimating prepayment speeds for seasoned pools.

Liquidity and technical factors: OAS reflects liquidity premium and technical supply/demand, not just credit and option risk. When Fed buys heavily (QE), OAS compresses below 'fair value.' When Fed sells (QT), OAS widens. Model can't distinguish credit deterioration from technical widening.

Best practice: Use OAS as one input among many. Cross-check with Z-spread trends, static yield analysis, and prepayment-to-assumption sensitivity. Never rely solely on model-generated OAS for investment decisions.

OAS application varies by sector based on embedded options:

Agency MBS: Heavy use. Prepayment option is primary risk. OAS typically 30-80bps depending on coupon and market conditions. Models very sophisticated due to deep market and extensive data.

Non-Agency RMBS: Less reliable. Credit risk dominates, prepayment models less tested. OAS often unreliable indicator. Z-spread or credit spread over LIBOR more commonly used.

CMBS: Limited use. Most CMBS have prepayment lockouts or penalties (yield maintenance/defeasance), reducing option risk. When used, focuses on extension risk from troubled loans. OAS spreads are wide (200-400bps for BBB) reflecting credit, not options.

ABS (auto, credit cards): Minimal use. Prepayment risk exists but less rate-sensitive than mortgages (autos too small to refinance, credit cards revolve). Z-spread or nominal spread more common. OAS adds little value.

CLO: Not applicable. No embedded options (loans can prepay but CLO manager can reinvest, and reinvestment period structures handle this). Credit spread or discount margin used instead.

Know when OAS is meaningful vs when simpler spread measures suffice.