Securitisation
Turning streams of loan payments into tradable bonds — the machine behind MBS, ABS, CLOs and CDOs, its 2008 failure, and its disciplined afterlife.
The idea in one sentence
Take thousands of loans — mortgages, auto loans, credit cards, corporate loans — pool their payments, and sell bonds backed by that stream: lenders get their capital back today, investors get exposure to borrowers they could never reach one loan at a time. That is securitisation, and roughly $13 trillion of it is outstanding.
The assembly line, step by step
- Originate — a bank or lender writes the loans.
- Isolate — the loans are sold to a special purpose vehicle (SPV): a shell company whose only assets are the loans. The point is bankruptcy remoteness — if the originator fails, the loans (and the bondholders) are untouched.
- Tranche — the SPV issues bonds in layers of seniority against the pooled payments (next section).
- Service — a servicer collects payments, chases arrears, forecloses; its quality quietly drives realised losses.
- Distribute — rating agencies grade each tranche; investors buy by rating, yield and layer.
Tranching and the waterfall
- Cash flows down, losses flow up: interest and principal pay seniors first; defaults eat the equity tranche first. One pool, many risk profiles.
- Credit enhancement stacks protections under the seniors: subordination (the layers below), over-collateralisation (more loans than bonds), excess spread (loan interest above bond coupons), and reserve accounts.
- Triggers police the structure: if collateral tests fail, cash is diverted from junior to senior tranches — the covenants of the securitised world (CLO OC/IC tests are the living example).
- The layer names recur across the atlas: MBS, ABS, CLOs, CDOs & synthetic tranches.
Why it exists — the honest case
- Funding: lenders recycle capital instead of warehousing 30-year loans — more credit supplied per unit of bank balance sheet.
- Risk transfer: concentrated local risk (one region's mortgages) becomes diversified, distributable paper.
- Tailoring: one pool serves money funds (short seniors), insurers (long seniors) and hedge funds (equity) at once.
- Done conservatively, it works: European covered-bond-adjacent structures, auto ABS and post-crisis CLOs sailed through both 2008 and 2020 with minimal senior losses.
2008: how the machine failed
- Originate-to-distribute rotted incentives: lenders who sell every loan stop caring if it repays — underwriting collapsed exactly where volume grew fastest (subprime).
- Correlation was mispriced: pooling only diversifies independent risks. US house prices were one national bet; tranching one repeated bet protects nobody (the CDO page runs the arithmetic).
- Re-securitisation multiplied labels, not safety: CDOs of mezzanine MBS, then CDO-squared — AAA stamps stacked on the same underlying risk.
- Ratings failed at the tails: models calibrated to decades without a national house-price fall met one.
The disciplined afterlife
- Risk retention ("skin in the game"): originators must keep ≥5% of what they securitise — the direct fix for rotten incentives.
- The EU's STS regime (simple, transparent, standardised) and loan-level disclosure rebuilt the boring end of the market.
- CLOs became the flagship: actively managed corporate-loan securitisations with hard collateral tests — senior tranches took no principal losses through 2008 or 2020 (the leverage migrated to the loans themselves; see cov-lite).
- SRT / synthetic risk transfer is the growth frontier: banks buy mezzanine protection on their own loan books from funds — securitisation as a regulatory-capital tool, now a >$1tn reference market regulators watch closely.
Reading any securitisation like a practitioner
- Ask the correlation question first: what single macro variable hits the whole pool at once? That variable, not the rating, is the senior tranche's true risk.
- Check who keeps the equity — an originator holding first loss believes in the loans; one selling it is renting its underwriting.
- Read the triggers: OC/IC tests and cash-diversion rules decide who gets paid in stress — the document is the instrument.
- Structure ≠ alchemy: tranching sorts risk, it cannot reduce it. If the pool is bad, someone owns that badness — the structure only decides who, and how politely they're told.