MBS & Structured —
Mortgages, passthroughs & prepayment.
Mortgage-backed securities pool home loans and sell the cashflows to investors. The complexity comes from those cashflows being unpredictable — homeowners prepay early, refinance, or default. Managing that uncertainty is what MBS trading is about. The 2008 financial crisis was fundamentally an MBS crisis, making this product uniquely important both mechanically and historically.
MBS is niche — most S&T interviewers will only probe it if you express specific interest or interview for a structured products desk. The questions that always come up regardless: what is an MBS conceptually, what are the two main risks, and how does it differ from a traditional bond. Master those three and you have covered 80% of what will be asked.
GFC context: The 2008 crisis began in MBS — specifically in subprime passthroughs and the CDOs built on top of them. Interviewers from structured products desks often ask what you know. The core failure: rating agencies mis-rated tranched MBS as AAA, underestimating correlated default risk when US house prices fell nationally for the first time since the Depression.
An MBS is a form of ownership of a pool of mortgage loans and the cashflows those loans generate over time. When a bank originates mortgages, it can sell those loans to a government-sponsored enterprise (GSE) or private issuer, which pools them and issues securities backed by the mortgage cashflows. Investors who buy MBS are the ultimate recipient of monthly principal and interest payments from thousands of homeowners — without the homeowners knowing it.
Each month, MBS holders receive a payment comprising both principal repayment and interest — unlike a traditional bond that pays only interest until a lump-sum principal at maturity. The issuer deducts fees first: a servicing fee (~50bps annualised) paid to whoever administers the mortgages, and a guarantee/agency fee (~15bps). So a pool of 4.00% mortgages generates roughly a 3.35% coupon to MBS investors after fees.
The fundamental difference from a corporate bond: with a traditional bond, cashflow timing is certain — you know every coupon date and the maturity. With MBS, cashflows are uncertain. Homeowners can prepay at any time by refinancing, selling their homes, or paying down extra principal. This timing uncertainty is the central analytical challenge of MBS investing.
A passthrough is the most common form of MBS. The issuer pools mortgages, collects monthly payments, and passes the cashflows — minus fees — pro-rata to investors. Each investor holds a fractional share of the entire pool's performance.
The dominant issuers are the three GSEs (government-sponsored enterprises):
- Fannie Mae (FNMA): Federal National Mortgage Association. Buys conventional conforming mortgages and issues MBS.
- Freddie Mac (FHLMC): Federal Home Loan Mortgage Corporation. Similar mandate to Fannie. Both entered conservatorship during the 2008 crisis and remain there.
- Ginnie Mae (GNMA): Government National Mortgage Association. Issues MBS backed by government-insured loans (FHA, VA mortgages). The only one with an explicit US government guarantee — Ginnie MBS are the safest form of MBS from a credit perspective.
Fannie and Freddie's backing is implicit (government conservatorship) rather than statutory. In practice, markets treat all three as effectively government-backed, though Ginnie's spreads are tightest.
The fundamental difference is cashflow certainty. A traditional bond pays fixed coupon amounts on fixed dates, with full principal returned at a known maturity. The cashflow schedule is completely predictable (assuming no default).
An MBS pays combined principal and interest monthly — but the amounts vary unpredictably. Each month, homeowners may pay extra principal, refinance, or sell their homes, generating prepayments that return principal to investors faster than scheduled. The investor cannot predict the timing or size of these cashflows.
The practical consequences:
- MBS have no fixed maturity — only an estimated weighted average life (WAL)
- When rates fall, prepayments accelerate (refinancing boom) — principal returns faster but must be reinvested at lower rates: prepayment risk
- When rates rise, prepayments slow — the MBS extends in duration, amplifying price declines in a rising rate environment: extension risk
- Getting hurt in both directions is negative convexity — MBS lose value relative to conventional bonds in both rate environments
Prepayment risk: homeowners pay off mortgages faster than expected. The primary trigger is falling rates — homeowners refinance into cheaper mortgages, paying off the original loan early. The MBS investor gets principal back sooner but must reinvest at lower prevailing rates (reinvestment risk). Home sales also generate prepayments regardless of rate conditions.
Extension risk: homeowners prepay more slowly than expected — principal returns later than anticipated. The primary trigger is rising rates — nobody refinances into a higher-rate mortgage. The MBS extends in duration beyond the original model. In a rising rate environment, this is doubly painful: the bond loses value (all bonds do when rates rise) and it has longer duration than expected, amplifying the price decline.
Both risks together create negative convexity: the MBS investor is hurt whether rates fall (prepayment) or rise (extension) relative to expectations. Unlike conventional bonds which benefit from rate volatility through positive convexity, MBS lose value in both directions. Investors demand an extra yield premium — the option-adjusted spread (OAS) — to compensate for bearing this risk.
Convexity measures how a bond's duration changes as rates move. Most conventional bonds have positive convexity — as rates fall, prices rise more than duration predicts; as rates rise, prices fall less than duration predicts. Positive convexity means you benefit from rate volatility.
MBS have negative convexity because of the embedded prepayment option homeowners hold:
- When rates fall: prepayments accelerate, shortening the MBS's effective duration. This limits price appreciation — the investor gets principal back early and must reinvest at the now-lower rates, capping the benefit of the rate rally.
- When rates rise: prepayments slow, extending duration. The MBS behaves like a longer bond, amplifying price declines.
The practical implication: MBS tend to underperform conventional bonds of similar stated duration in both rallies and selloffs. Investors require additional yield compensation — the option-adjusted spread (OAS) — to hold MBS versus Treasuries. Historically this spread has been 50–200bps over equivalent-duration Treasuries, depending on market conditions.
The 2022–23 experience illustrated this vividly: banks holding long-duration MBS (including Silicon Valley Bank) suffered enormous unrealised losses as rates rose, and those losses were amplified by the extension of their MBS holdings beyond original duration assumptions.
MBS cannot have a stated maturity because principal is returned continuously and unpredictably depending on prepayments. Weighted average life (WAL) is the estimated average time until pool principal is fully repaid, weighted by the amount of principal returned at each date.
The WAL is a probability-weighted estimate driven by prepayment assumptions — not a contractual date. It changes as rate environments change.
Counterintuitively, a pool of 30-year fixed-rate mortgages does not have a 30-year WAL. Historically, WAL for a standard 30-year pool has averaged just 4–6 years — because most homeowners refinance, sell, or prepay within that window. During 2020–21's low-rate refinancing boom, many pools shortened to 2–3 year WALs. In 2022–23 as rates rose sharply, WALs on those pools extended significantly as refinancing stopped.
WAL is how MBS investors compare their holdings to other fixed income instruments — a 4-year WAL MBS competes with 4-year Treasury notes — and the MBS must yield more to compensate for the cashflow uncertainty.
PSA (Public Securities Association, now SIFMA) Prepayment Standard Assumptions is the industry benchmark for modelling mortgage prepayment speeds. It provides a standardised ramp-up curve:
- In the first 30 months, the annualised prepayment rate increases at 0.2% per month — from 0% in month 1 to 6% by month 30.
- After month 30, prepayments plateau at a 6% annual rate — this is defined as 100% PSA.
A pool prepaying at 200% PSA is prepaying at twice the standard rate (common in low-rate refinancing environments). 50% PSA means slower than standard (common when pool coupon rates are below current market rates — little incentive to refinance).
PSA assumptions directly drive WAL calculations: higher PSA speeds shorten WAL; lower speeds extend it. Quants and traders calibrate PSA speeds using four drivers: the refinancing incentive (how far market rates are below the pool coupon), seasoning (how old the pool is — newer pools prepay slowly), seasonality (spring/fall see more home sales), and burnout (pools that have already experienced a refinancing wave will prepay more slowly in subsequent rate rallies, as the most rate-sensitive borrowers have already left).
A CMO (collateralised mortgage obligation) takes passthrough cashflows and redirects them across multiple tranches with defined principal payment priority, rather than distributing everything pro-rata to all investors.
In the simplest sequential-pay structure: all monthly principal (scheduled amortisation plus prepayments) flows entirely to Class A until Class A is fully paid. Only then does principal flow to Class B, then Class C. All tranches receive interest throughout.
The result: Class A has the shortest and most predictable WAL (it receives all principal first); Class C has a longer, more stable WAL (it is sheltered from early prepayments because it only receives principal after A and B are exhausted).
The benefit over a passthrough: better-defined duration profiles. Investors with specific duration needs — a pension fund wanting 2-year paper, an insurance company wanting 7-year paper — can find a tranche that matches. The passthrough gives everyone the same uncertain average WAL that shifts with rates; the CMO carves up that uncertainty into more predictable pieces.
Beyond sequentials, structures include PACs (planned amortisation classes with tighter WAL ranges), interest-only (IO) and principal-only (PO) strips, and inverse floaters — each redistributing prepayment risk differently across different investor bases.
The 2008 crisis was fundamentally a non-agency MBS and structured products failure. Four interlocking causes:
Origination standard collapse: Mortgage originators (Countrywide, WaMu, etc.) dramatically loosened standards in 2003–2007, issuing subprime and Alt-A loans to borrowers who could not service them. They had no incentive to maintain quality because they immediately sold loans rather than holding them — the classic 'originate to distribute' problem.
Rating agency model failure: Rating agencies rated CDOs built on subprime MBS as AAA. Their diversification models assumed regional housing markets would not fall simultaneously. When US house prices fell nationally for the first time since the Depression, mortgage default correlations shot to 1.0 — wiping out the diversification benefit entirely. Senior tranches that were supposed to be protected by subordination were not.
Complexity and opacity: MBS were packaged into CDOs, which were packaged into CDO-squareds. Nobody — including the institutions holding them — understood the actual risk exposures. When subprime started defaulting, the market froze because nobody knew which institutions held which risks.
Mark-to-market amplification: Banks holding MBS in trading books had to mark them to distressed market prices even if they intended to hold them to maturity. This created apparent capital shortfalls that forced selling, which pushed prices lower, which created more apparent shortfalls — a doom loop.