Securitization: MBS & ABS

Picture a bank that just wrote 5,000 car loans. On paper it’s a fine business — thousands of borrowers each sending a monthly cheque. But those loans are stuck: each one is a private, illiquid contract that the bank must hold for years, tying up capital and exposing the bank to every borrower who stops paying. The bank doesn’t want 5,000 trickling cash flows on its balance sheet for five years. It wants cash today, and it wants the risk off the books. The machine that grants both wishes — that takes a pile of dull, illiquid loans and stamps out clean, liquid, rated bonds an investor anywhere can buy — is called securitization. It is one of the most important plumbing inventions in all of finance, and it’s also the machine that, badly operated, helped blow up the world in 2008. Let’s open the hood.

Turning loans into bonds

The analogy. Think of a juice press. You can’t sell a warehouse full of muddy, mismatched fruit — every piece is a little different, hard to value, hard to move. But run them through a press and you get a clean, uniform, sellable product. Securitization is the press: illiquid loans in one end, standardized tradable bonds out the other. The trick that makes the juice purer than the fruit — bonds that can be rated higher than the bank that made the loans — is a clever piece of legal engineering.

The mechanism, step by step. A bank (the originator) has a pile of loans. It doesn’t securitize them on its own balance sheet — it sells them. Specifically:

1. Pooling. The originator gathers many similar loans into one pool (5,000 car loans, or 3,000 mortgages, or a basket of credit-card receivables). Pooling is what makes the statistics work: one borrower is a coin flip, but 5,000 of them have a predictable default rate, the way a casino can’t predict one spin but can predict a year.

2. True sale to an SPV. The originator sells the pool — a genuine, legal true sale — to a freshly created shell company built for exactly this purpose: a Special Purpose Vehicle (SPV), also called an SPE or, when structured as a trust, simply “the trust.” “True sale” is the load-bearing phrase: the loans must legally leave the originator and belong to the SPV, not merely be pledged as collateral. If it’s a true sale, the assets are gone from the bank’s estate for good.

3. Issuing the bonds. The SPV needs money to pay the originator for the pool. It raises that money by issuing securities — bonds — to investors. The investors’ cash flows up to the SPV (and on to the originator); the bonds flow out to investors.

4. Servicing the bonds. Now the loans live in the SPV, and the monthly payments from all those borrowers (interest plus principal) flow into the SPV and are passed through to the bondholders. The borrowers’ cash flows are what service — pay — the bonds. A servicer (often the original bank, for a fee) collects the payments and forwards them.

Bankruptcy remoteness — the secret sauce. The SPV is deliberately built to be bankruptcy-remote: a do-nothing shell whose only assets are the loan pool, with no other business, no other creditors, and legal walls that stop it from being dragged into the originator’s bankruptcy. Why does this matter so much? Because of the magic it unlocks: if the originating bank goes bust, the loan pool is insulated — it’s not the bank’s property anymore (true sale), and the SPV can’t be pulled into the bankruptcy. The bondholders still get the borrowers’ payments regardless of what happens to the bank.

That insulation is why the bonds can be rated above the originator. A shaky BBB bank can sell loans into an SPV and have the resulting senior bonds rated AAA — not because the bank got safer, but because the bonds no longer depend on the bank at all. They depend only on the borrowers in the pool and the legal walls around them. Stripping out the originator’s own credit risk is the entire point.

The pass-through. The simplest output of this machine is a pass-through security: every investor owns a pro-rata slice of the entire pool’s cash flows. Whatever the borrowers pay — interest, scheduled principal, and any early repayments — gets divided up and “passed through” to investors in proportion to their holdings. Own 1% of the pass-through, receive 1% of every dollar the pool produces. No carving, no priorities — just a thin, even slice of everything. (Next lesson we’ll carve the pool into tranches with different risks; the pass-through is the un-carved version.)

Now let’s watch the press run. The animation below shows the whole pipeline in three stages — a pool of loans gets sold into the SPV / issuer, which then issues a stack of bonds to investors. Press play and follow the flow left to right. Notice that the bonds come out in a layered stack (a big Senior layer, a thinner Mezzanine, a sliver of Equity) — that layering is tranching, the subject of the very next lesson. For now, just absorb the shape: pooled loans go in, a stack of bonds comes out.

The three stages map exactly onto what we just walked through. The pool is step 1 (pooling). The SPV / issuer in the middle is steps 2–3 (true sale into the bankruptcy-remote vehicle, which then issues). The tranches on the right are the bonds investors buy — here drawn as roughly 80% Senior (AAA), 15% Mezzanine, 5% Equity, the canonical capital-structure split you’ll meet properly when we cover the waterfall. The cash flows from borrowers run backwards through this picture, from the pool up to the bondholders, every single month.

The originate-to-distribute model

Two business models. For most of banking history, a lender ran a hold-to-maturity (or “originate-to-hold”) model: make a loan, keep it, and live with the consequences. If you lent badly, you ate the default. That arrangement aligns incentives beautifully — the person deciding whether to lend is the same person who suffers if the borrower doesn’t pay. You underwrite carefully because the risk is yours.

Securitization makes a second model possible: originate-to-distribute (OTD). Here the lender makes loans intending to sell them — package them into an SPV and distribute the bonds to investors — rather than hold them. The lender becomes a conveyor belt: originate, sell, repeat. Its profit no longer comes from loans performing over time; it comes from fees on volume — the more loans you push through the belt, the more you earn, more or less regardless of whether they ultimately pay.

The analogy. Imagine a chef who cooks every meal they serve and eats the leftovers themselves — they’ll be fastidious about freshness. Now imagine a chef paid per plate sent out the door, where someone else eats the food and the chef never hears the complaints. Quality is going to slide. Originate-to-distribute turns careful chefs into volume cooks.

The incentive rot. The problem is structural, not about bad people. When the loan’s risk leaves your books the moment you sell it, your motivation to underwrite well — verify the borrower’s income, check the credit, demand a real down payment, say no to a dicey applicant — erodes. “Not my problem after I sell it” becomes the silent logic of the whole assembly line. And the loans most likely to be waved through are exactly the ones that should have been refused.

Worked example — the volume trap. Suppose a lender earns a $3,000 fee per mortgage it originates and sells, and it can sell essentially everything it writes. Writing 1,000 careful loans earns $3 million. Loosening standards lets it write 4,000 loans — $12 million — and since the loans are sold within weeks, the lender pockets the fees before any of those weaker borrowers miss a payment. The defaults land two years later, in the investors’ laps. From the lender’s seat, the rational move is to maximize volume and minimize friction. Multiply that logic across an entire industry and you get the mortgage machine of 2005–2007.

MBS: pass-throughs & prepayment risk

When the loans in the pool are mortgages, the securities are mortgage-backed securities (MBS) — the original and still-largest corner of the securitization universe. MBS add a wrinkle no corporate bond has: the borrower can prepay.

Agency vs non-agency. First, a crucial credit distinction:

• Agency MBS are guaranteed by a US housing agency. Ginnie Mae carries the literal full faith and credit of the US government — as close to default-free as a bond gets. Fannie Mae and Freddie Mac (Government-Sponsored Enterprises, or GSEs) provide their own guarantee, implicitly backstopped by the government (a backstop that became explicit when they were taken into conservatorship in 2008). Agency MBS carry minimal credit risk; what’s left is mostly interest-rate and prepayment risk.
• Non-agency (private-label) MBS have no government or GSE guarantee. They carry real credit risk, and to make the senior bonds safe they rely on internal credit enhancement — overcollateralization, reserve accounts, and especially tranching (next lesson). This is the corner where subprime lived, and where the 2008 losses concentrated.

Pass-through vs CMO. Take an agency mortgage pool. The simplest security is the pass-through: investors get a pro-rata share of all the pool’s cash flows — interest, scheduled principal, and every prepayment — sharing the timing uncertainty evenly. But many investors hate timing uncertainty, so the same pool can instead be carved into a Collateralized Mortgage Obligation (CMO): the cash flows are sliced into tranches with different maturities and risk profiles. Classic structures include sequential-pay (tranche A receives all principal first until it’s retired, then B, then the Z-tranche last), PACs (Planned Amortization Classes, which get a stable, scheduled principal payment while companion tranches absorb the prepayment swings), and IO/PO strips (Interest-Only and Principal-Only — splitting the pool so one bond gets only the interest stream and the other only the principal). A pass-through shares prepayment risk; a CMO redistributes it to whoever wants more or less of it.

Prepayment risk and how we measure it. A mortgage borrower can repay early — they refinance to a cheaper rate, sell the house and move, or just pay extra. That makes the timing of an MBS’s cash flows uncertain, which is the defining risk of the product. Two standard measures:

• CPR (Conditional / Constant Prepayment Rate) — the annualized fraction of the remaining pool expected to prepay this year. A 6% CPR means roughly 6% of the outstanding balance prepays over the year.
• PSA (the PSA benchmark) — an industry standard prepayment ramp. 100% PSA means the CPR starts at 0 and ramps up by 0.2% per month for the first 30 months, then levels off at 6% CPR for the remaining life of the loan. “200% PSA” means twice that speed (levels at 12% CPR); “50% PSA” means half (levels at 3%). It captures the real-world fact that brand-new mortgages rarely prepay, but seasoned ones settle into a steady churn.

Negative convexity — the killer feature. Here’s the cruel part. Prepayment isn’t random — it’s correlated with interest rates in exactly the wrong direction for the investor. When rates fall, borrowers rush to refinance, so prepayments spike: your high-coupon MBS gets repaid early, and you must reinvest the returned principal at the new, lower rates. When rates rise, borrowers cling to their cheap mortgages and stop prepaying, so your principal stays locked up in a now-below-market bond exactly when you’d love it back to reinvest higher.

So the MBS investor loses both ways: rates down → called away early → reinvest low; rates up → stuck long in a cheap bond. This is negative convexity — the bond’s price underperforms a normal bond when rates move in either direction, the mirror image of the friendly positive convexity a plain Treasury enjoys. It’s the reason MBS yield more than agency credit risk alone would justify: investors demand extra compensation for being short that optionality (the borrower owns a free prepayment option, and the investor is short it).

Worked example — the call-away. You hold a 100% PSA agency pool throwing off a juicy 6% coupon; you bought it expecting it to amortize down at a steady ~6% CPR. Now market rates drop sharply. Refinancing becomes a no-brainer for borrowers, and prepayments leap — say the pool starts running at 30% CPR instead of 6%. A huge chunk of your principal floods back over the next year, years ahead of schedule. You wanted to keep collecting 6% for a decade; instead you’re handed a pile of cash to reinvest at the new, lower market rate of maybe 3.5%. Your high-coupon bond got called away at the worst possible moment. The same prepayment that delights the homeowner gut-punches the investor.

ABS: everything else

The press doesn’t care what you feed it. Once you can pool loans, sell them to an SPV, and issue bonds, you can do it with any stream of receivables — and the market does, prolifically. When the collateral is not mortgages, the result is an asset-backed security (ABS). (Strictly, MBS is a subset of ABS, but in practice “ABS” means non-mortgage and “MBS” means mortgage.)

The ABS menu. The mainstream categories:

• Auto loans — the bread-and-butter of ABS. Predictable amortization, short maturities, deep history.
• Credit-card receivables — pools of card balances. These are revolving, not amortizing (more on that below).
• Student loans — pools of education debt (in the US, often tied to government programs).
• Equipment leases & loans — tractors, aircraft engines, IT hardware.
• Dealer floorplan — the financing dealers use to stock inventory (the cars sitting on the lot).

And then the esoteric end, which proves the machine eats almost anything with a cash flow: aircraft leases, music and film royalties (David Bowie famously securitized his catalog — “Bowie Bonds”), solar panel lease payments, even data-center revenues. If it produces a reliable stream of payments, someone has tried to securitize it.

Revolving vs amortizing — the structural fork. This is the key way ABS differ from one another and from MBS:

• Amortizing collateral (auto loans, most ABS) pays down a fixed schedule of principal + interest over a known term, like a mortgage. The pool’s balance shrinks predictably.
• Revolving collateral (credit cards) has no fixed schedule — balances bounce around as cardholders borrow and repay. To keep a card-ABS bond alive for a set term, the structure uses a revolving period: incoming principal is recycled to buy new receivables (keeping the pool topped up) for a few years, followed by an amortization period where principal is finally paid out to bondholders.

Why ABS is usually tamer than MBS. Two reasons. First, shorter maturities — a five-year car loan simply can’t wander as far as a thirty-year mortgage. Second, and bigger: prepayment is far less rate-sensitive. People refinance a $300,000 mortgage the instant rates dip a point, because the savings are huge; almost nobody bothers refinancing a $22,000 car loan over a quarter-point, because the dollar savings are trivial. So auto-ABS cash flows are shorter-dated and more predictable, with little of the negative-convexity drama that haunts MBS. That predictability is exactly why auto and card ABS are workhorse, investment-grade products even in stressed markets.

Why slice it into tranches?

A finished pool of loans produces one river of cash flows with one risk/return profile. That’s a problem, because the world’s investors have wildly different appetites. A pension fund or insurer is legally and temperamentally restricted to very safe assets. A hedge fund is hunting high returns and will happily take the first-loss risk to get them. Sell a single pass-through and you can satisfy only one of them — leaving demand (and money) on the table.

Tranching solves this by carving the same cash flows into priority layers — tranches — that get paid in a strict order. The senior tranche gets paid first and absorbs losses last, so it’s ultra-safe (and low-yield): perfect for the insurer. The equity (or “first-loss”) tranche gets paid last and eats the first defaults, so it’s risky and high-yield: catnip for the hedge fund. The mezzanine sits in between. The crucial insight: tranching doesn’t change the total risk in the pool — it redistributes it, concentrating safety in the senior layers by piling the danger onto the junior ones. You’re selling safety to those who crave it and risk to those who hunt it, all manufactured from a single pool.

That’s why the bonds came out of the animation as a layered stack rather than one block. The mechanics of how the losses and cash flows cascade down those layers — the waterfall, the over-collateralization, the credit enhancement that lets a pool of merely-OK loans produce a chunky AAA senior tranche — is the entire subject of the next lesson.

Putting it together

Securitization is the press that turns a pile of illiquid loans into liquid, rated bonds. The originator pools similar loans and true-sells them to a bankruptcy-remote SPV, which issues securities backed by the pool’s cash flows; because the loans have legally left the originator and the SPV is insulated from its bankruptcy, the bonds can be rated above the originator itself. The simplest output is a pass-through (a pro-rata slice of every cash flow). The same machine enables originate-to-distribute, where lenders make loans intending to sell them — collecting fees on volume while the default risk walks out the door — the moral-hazard accelerant that, unrestrained, fed the 2008 crisis (now curbed by risk-retention rules). When the collateral is mortgages you get MBS: agency (Ginnie / Fannie / Freddie, minimal credit risk, mostly rate and prepayment risk) or non-agency (no guarantee, real credit risk, where subprime lived). Mortgages add prepayment risk — measured by CPR and the PSA ramp (100% PSA levels at 6% CPR) — which gives MBS negative convexity: when rates fall, borrowers refinance and your high-coupon bond is called away to be reinvested at lower rates. Pools can be sold as pass-throughs or carved into CMO tranches (sequential, PAC, IO/PO). When the collateral is anything else — autos, cards, student loans, equipment, or esoteric royalties and aircraft — you get ABS, generally shorter-dated and more predictable than MBS, split between amortizing (autos) and revolving (cards) structures. And the reason to tranche at all: one pool, one risk profile, can’t satisfy both the safety-craving insurer and the yield-hungry hedge fund — so you slice the same cash flows into priority layers and sell each appetite its own bond. Exactly how those layers absorb losses — the waterfall — is where we go next.

Next — tranching and the waterfall: we’ll take the layered stack of bonds you watched come out of the press and follow the cash and the losses down through senior, mezzanine, and equity, building the credit enhancement that turns a pool of merely-OK loans into a chunky AAA tranche.