Mortgage-Backed Securities Analysis: Prepayment Modeling, OAS, WAL, and Agency MBS Valuation

Mortgage-Backed Securities: The Analytical Challenge

No fixed income sector demands more computational machinery than mortgage-backed securities. A single agency MBS pool combines interest rate risk, prepayment optionality, convexity distortion, and pool-level collateral variation — all changing simultaneously as rates move. The MBS desk at any major dealer runs proprietary prepayment models continuously; buy-side analysts spend hours in Bloomberg's OAX and FIPS functions. Claude cannot replace those models, but it can compress the time between data and insight when you need to understand a structure, evaluate prepayment assumptions, or explain a position to a portfolio manager.

This guide assumes you are working at the practitioner level — you know what a CPR is, you have seen a PSA ramp, and you understand why WAL matters more than stated maturity. The prompts below are structured to give Claude the parameters it needs to do meaningful work.

MBS Structure: From Mortgage Pool to Pass-Through

An agency MBS begins as a pool of residential mortgages originated and delivered to Fannie Mae (FNMA), Freddie Mac (FHLMC), or Ginnie Mae (GNMA). The agency guarantees timely payment of interest and principal, eliminating credit risk from the investor's perspective — the investor is left with interest rate risk and prepayment risk exclusively.

GNMA securities are backed by FHA and VA loans and carry the full faith and credit of the US government. FNMA and FHLMC are GSEs with an implicit (now explicit, given conservatorship) government backing. Private-label MBS — securities without agency wrapping — carry credit risk in addition to prepayment risk and require full credit analysis of the underlying collateral, servicer, and structural subordination.

Pass-through securities distribute each month's collected principal and interest (net of guarantee fee) to certificate holders pro-rata. More complex CMO (Collateralized Mortgage Obligation) structures slice the cash flows into tranches with different WAL profiles — PAC (Planned Amortization Class) tranches absorb prepayment variability within a band, support tranches absorb the excess variability, and IO/PO strips split interest and principal cash flows entirely.

The PSA Prepayment Benchmark: What the Numbers Mean

The PSA (Public Securities Association) prepayment model provides a standard benchmark for quoting and comparing prepayment speeds. The model defines a ramp: CPR starts at 0.2% annualized in month 1, increases by 0.2% per month for 30 months (reaching 6% CPR), then remains flat at 6% CPR indefinitely. This is "100 PSA."

Actual prepayment speeds are quoted as a percentage of this benchmark. A pool paying at 200 PSA is prepaying at twice the benchmark speed. A seasoned pool in a strong refinancing environment might print 400–600 PSA. A pool with geographic, loan size, or loan-to-value characteristics that suppress refinancing might run at 50–80 PSA even in a low-rate environment.

The CPR-to-SMM conversion is essential for cash flow modeling:

• CPR (Conditional Prepayment Rate): annualized percentage of remaining balance that prepays in a given period
• SMM (Single Monthly Mortality): monthly equivalent of CPR. Formula: SMM = 1 − (1 − CPR)^(1/12)
• At 6% CPR: SMM = 1 − (0.94)^(1/12) ≈ 0.514% per month
• At 12% CPR (200 PSA at plateau): SMM ≈ 1.058% per month

Monthly prepayment cash flow = SMM × (Beginning balance − Scheduled principal payment). This is the prepay principal that arrives unexpectedly and must be reinvested.

Weighted Average Life: Your True Duration Proxy

A 30-year mortgage pool does not have a 30-year investment horizon under realistic prepayment assumptions. WAL measures the average time until each dollar of principal is returned, weighted by the dollar amount:

WAL = Σ (t × Principal_t) / Total Principal, where t is measured in years.

Representative WAL benchmarks for a 30Y FNMA 5.5% coupon pool originated in 2024:

• 50 PSA: WAL ≈ 15–17 years (very slow prepayments, extension risk)
• 100 PSA: WAL ≈ 10–12 years (benchmark speed)
• 200 PSA: WAL ≈ 6–8 years (refinancing wave)
• 400 PSA: WAL ≈ 4–5 years (peak refi environment)

WAL is the metric that determines where the MBS sits on the yield curve for comparative purposes. A 10-year WAL agency MBS competes with 10-year Treasuries and corporates for duration-matched portfolios.

OAS: Stripping Out the Prepayment Option

The Z-spread (zero-volatility spread) measures the constant spread over the Treasury spot curve that discounts scheduled cash flows to the market price. The problem: MBS cash flows are not scheduled. They depend on prepayment speeds, which depend on rates, which are stochastic.

OAS (Option-Adjusted Spread) uses Monte Carlo simulation — typically 200–1,000 interest rate paths — to generate a distribution of possible cash flow scenarios. For each path, prepayment speeds are modeled using a prepayment engine (PSA-based, or more sophisticated models incorporating burnout, seasonality, loan age, and rate incentive). The OAS is the single constant spread added to each path's discount rates that equates the average present value across all paths to the market price.

OAS decomposition for agency MBS:

• Z-spread: 65–90bps (includes option cost)
• Option Cost: 30–45bps (value of prepayment option held by borrowers)
• OAS: 25–50bps (the "true" spread after removing optionality)

In mid-2025, current-coupon agency MBS (5.5% FNMA 30Y) OAS was approximately 42–48bps over Treasuries. This is the spread that represents compensation for liquidity and residual prepayment model uncertainty — it is not credit compensation, since the GSE guarantee eliminates credit risk.

Negative Convexity and Its Portfolio Implications

Standard bonds exhibit positive convexity: as rates fall, price rises at an accelerating rate; as rates rise, price falls at a decelerating rate. The price-yield curve is convex (bows toward the investor).

MBS display negative convexity because falling rates accelerate prepayments, shortening duration exactly when you want it to lengthen. Rising rates slow prepayments, extending duration exactly when you want it to shorten. The MBS price-yield curve is concave in the rate-falling region — price appreciation is capped by prepayments returning capital at par.

Implications for portfolio managers:

• MBS underperform duration-matched Treasuries in volatile rate environments (convexity drag)
• Hedging with receiver swaptions or Treasury futures requires constant rebalancing as duration shifts
• OAS-positive MBS offer yield pickup vs Treasuries in return for accepting this convexity drag
• The "cheapest to deliver" in a rate decline is often a specified pool with superior prepay characteristics

TBA vs Specified Pools: The Pay-Up Framework

The TBA (To-Be-Announced) market is the forward delivery mechanism for agency MBS. Buyers and sellers agree on issuer, coupon, maturity, and face amount today; pools are allocated two days before settlement. Because the cheapest-to-deliver pools are allocated into TBA, TBA prices reflect worst-case prepayment assumptions for a given coupon.

Specified pools — pools with identifiable collateral characteristics that suppress prepayments — trade at premiums to TBA. Common specified pool categories and typical 2025 pay-ups above TBA:

• Low loan balance pools (<$150K original balance): +20–35 ticks (32nds)
• Geographic spec (NY state, high-cost areas): +8–15 ticks
• High LTV (80%+ original LTV, less equity to extract): +10–20 ticks
• Investor property pools (rental property, less refi-sensitive): +15–25 ticks
• Manufactured housing pools: +5–12 ticks

Pay-up analysis requires estimating the present value of the prepayment benefit over the likely holding period — Claude is useful for working through the mechanics of this calculation with specific speed assumptions.

Claude Prompts for MBS Analysis

The following prompts are designed for fixed income professionals working with agency MBS. They assume you have access to current pool data and are using Claude for computational assistance and structured analysis.

1. PSA-Based Cash Flow Projection

"Model monthly cash flows for the first 12 months of a 30Y FNMA 5.5% pool: original balance $100M, WAC 6.0%, WAM 358 months (2 months seasoned), factor 1.0. Run two scenarios: 100 PSA and 200 PSA. For each month show: beginning balance, scheduled principal, prepayment principal (SMM applied to beginning balance less scheduled principal), interest, and total cash flow. Use SMM = 1 − (1 − CPR)^(1/12) where CPR follows the PSA ramp. What is the accumulated prepayment through month 12 in each scenario?"

2. WAL Calculation at Multiple PSA Speeds

"Calculate WAL for a 30Y FNMA MBS pool with the following characteristics: $1M original balance, 5.5% coupon, WAC 6.0%, 360-month original WAM, current factor 0.98 (2 months seasoned). Calculate WAL at 75 PSA, 150 PSA, 250 PSA, and 400 PSA. Present results in a table showing: PSA speed, plateau CPR, plateau SMM, approximate WAL in years, and what this WAL implies for yield curve positioning. Explain how WAL changes non-linearly with PSA speed."

3. OAS Decomposition Analysis

"I have a 30Y FNMA 5.5% current coupon MBS trading at 101-08 (32nds) in the TBA market. The interpolated 10Y Treasury yield is 4.45%, the 10Y swap rate is 4.62%. Compute the nominal spread to the 10Y Treasury and the nominal spread to swaps. If the Bloomberg OAX model shows OAS of 44bps and Z-spread of 78bps, what is the implied option cost? What does a 44bps OAS imply about this security vs its 5-year history if historical OAS range was 15–65bps? Is this security cheap, fair, or rich on an OAS basis?"

4. Negative Convexity Impact Quantification

"Demonstrate negative convexity for agency MBS. Current data: 30Y FNMA 5.5% at 100 PSA has effective duration 7.2 years and convexity −1.8. A 10Y Treasury has modified duration 8.1 years and convexity +0.9. If rates fall 100bps: (a) what is the estimated price change for each instrument using duration + convexity approximation? (b) If after the rally the MBS accelerates to 300 PSA, what does effective duration become (assume duration shortens proportionally to WAL reduction from PSA change)? (c) Explain qualitatively why an MBS portfolio manager needs to buy more Treasuries/pay fixed on swaps as rates fall — the negative gamma effect."

5. Specified Pool Pay-Up Analysis

"I'm evaluating a specified pool pay-up decision. Option A: buy 30Y FNMA 5.5% TBA at 101-08 (WAL at 200 PSA = 6.8 years). Option B: buy a low-loan-balance specified pool (original balance <$150K) at 101-28, same coupon. The specified pool is expected to print 120 PSA vs 200 PSA for TBA (slower prepayments = better for premium coupon). My holding period is 12 months. Calculate: (a) monthly income differential assuming the spec pool stays at 120 PSA vs TBA at 200 PSA, (b) the breakeven holding period for the 20-tick pay-up to break even, (c) the risk if prepayment assumptions prove wrong and the spec pool speeds up to 180 PSA."

6. CMO Tranche Analysis

"Analyze a PAC/support CMO structure backed by 30Y FNMA 5.5% collateral. PAC band: 100–300 PSA. Collateral balance $250M. PAC tranche: $150M, support tranche: $100M. At 150 PSA (within band): show that the PAC receives its scheduled principal according to a fixed amortization schedule. At 50 PSA (below band): explain why the support tranche receives no principal and the PAC begins to break. At 400 PSA (above band): explain why the support tranche is paid off rapidly and the PAC average life extends. What are the WAL ranges for the PAC and support under these three scenarios?"

7. Prepayment Speed Analysis vs Rate Environment

"Current mortgage rate (30Y FNMA primary mortgage rate): 6.8%. Pool characteristics: 30Y FNMA 5.5% coupon, WAC 6.0%, pool originated when rates were 5.9%. The refinancing incentive (current rate minus note rate) = 6.8% − 6.0% = +80bps (out of the money — borrowers have no incentive to refinance). Expected CPR = 3–4% (turnover-driven prepayments only, no refi). Scenario: rates fall 150bps to 5.3%. Now refi incentive = 5.3% − 6.0% = −70bps (in the money). Estimate CPR increase and its impact on WAL and price. What is the dollar convexity impact? If I own $50M face of this pool at 101, what is the mark-to-market loss from prepayment acceleration?"

8. Agency MBS vs Corporate Bond Comparison

"Portfolio construction decision: compare two fixed income positions for a $10M allocation. Option A: 30Y FNMA 5.5% current coupon MBS at OAS 44bps, effective duration 7.2 years, negative convexity −1.8. Option B: 10Y BBB corporate bond at OAS 125bps, modified duration 7.5 years, positive convexity +0.8. Risk-free rate = 4.45% (10Y Treasury). Compare: (a) total expected return over 1 year assuming OAS is unchanged for both, (b) which wins if rates rise 75bps — account for duration and convexity effects on each, (c) which wins if rates fall 75bps and MBS prepayments accelerate to 250 PSA, (d) what is the credit risk differential, and (e) which is more appropriate for a bank portfolio under current Basel III capital rules?"

9. IO Strip Valuation

"Value an IO (Interest-Only) strip on a 30Y FNMA 5.5% pool with $200M notional. The IO receives only the interest portion of each payment — no principal. At 100 PSA, estimate the IO cash flows for months 1–12 and total PV at a discount rate of 8%. Now at 400 PSA (aggressive prepayment), redo the PV calculation. How much IO value is destroyed by prepayment acceleration? What is the effective duration of an IO strip — explain why IOs are one of the few fixed income instruments with negative effective duration (price rises when rates rise). How does this make IOs useful as a hedge against prepayment extension?"

MBS in a Fixed Income Portfolio

Agency MBS occupy a specific and important role in institutional fixed income portfolios. Their attributes are well-suited to certain mandates and poorly suited to others:

Best suited for: bank portfolio management (favorable risk-weighted assets vs corporates), insurance company general accounts (long-duration liability matching with specified pools), core plus strategies seeking OAS pick-up over Treasuries, central bank reserve management.

Challenges for: total return mandates in volatile rate environments (convexity drag), LDI (liability-driven investing) where duration stability is essential, unconstrained mandates that can access better risk-adjusted alternatives.

The analytical toolkit for MBS — PSA speeds, WAL, OAS, prepayment model assumptions — creates a knowledge barrier that Claude can help practitioners navigate efficiently. Whether you're explaining prepayment risk to a board, evaluating a TBA position against a specified pool, or modeling the impact of rate scenarios on your MBS allocation, structured AI prompts can compress hours of spreadsheet work into minutes.

For related fixed income analytics, see our guides on bond pricing and OAS, credit spread analysis, and fixed income analysis overview. For portfolio-level risk aggregation including MBS positions, see portfolio VaR methodology.

Frequently Asked Questions

What does 200 PSA mean in practice?

200 PSA means the pool is prepaying at twice the PSA benchmark speed. The benchmark (100 PSA) ramps from 0.2% CPR in month 1 to 6% CPR by month 30 and stays there. At 200 PSA, the ramp reaches 12% CPR by month 30 and remains at 12% CPR. For a seasoned 30-year FNMA pool in a refinancing environment, 200 PSA is a moderate-to-high speed that significantly shortens WAL vs the benchmark assumption.

How is OAS different from Z-spread for MBS?

Z-spread assumes static, unchanged cash flows discounted over a single yield curve. MBS cash flows change with every rate scenario because prepayments respond to rates. OAS uses Monte Carlo simulation — typically hundreds of rate paths — to model prepayments on each path. The OAS is the constant spread that, when added to each path's discount curve, makes the probability-weighted average present value equal to the market price. Z-spread for agency MBS always exceeds OAS by the option cost (the value of the prepayment option held by borrowers), typically 25–45bps for current-coupon agency MBS.

Why do MBS have negative convexity?

Mortgage borrowers can prepay their loans at any time without penalty (for most standard US residential mortgages). When rates fall, borrowers refinance en masse — the MBS investor receives principal at par and must reinvest at lower prevailing rates. This means that as rates fall, the MBS duration shortens rather than lengthening (as it would for a standard bond). This negative gamma is negative convexity: the price-yield profile curves away from the investor, capping upside in rallies.

What are typical agency MBS OAS levels?

Agency MBS OAS varies significantly with the rate environment and supply/demand dynamics. In normal markets (2015–2019), current-coupon FNMA 30Y OAS ranged roughly 20–40bps. During the COVID shock in March 2020, OAS widened to 70–80bps briefly. Post-Fed quantitative tightening (2022–2024), OAS settled in the 40–65bps range as the Fed reduced its MBS holdings. As of mid-2025, current-coupon agency MBS OAS was approximately 42–48bps — wider than pre-COVID norms, reflecting reduced Fed balance sheet support.

How does the TBA market work and who participates?

TBA (To-Be-Announced) is the forward market for agency MBS. Buyers and sellers agree on four parameters: agency (FNMA, FHLMC, or GNMA), coupon, maturity (15-year or 30-year), and face amount. Settlement date is typically in the following month. Actual pools are "announced" two days prior to settlement. The market allows originators to hedge their pipelines before loans close (selling TBA forward locks in the execution price) and allows dealers and investors to establish positions without identifying specific pools. Daily volume exceeds $200 billion, making TBA second only to US Treasury markets in liquidity.