CAIA® CDO tranches: where do losses break through?

On this page

A collateralised debt obligation, or CDO, channels the cash flows and losses of a credit pool into claims with different seniority. The junior layer takes losses first; a senior layer suffers principal loss only after the protection beneath it is exhausted. Credit enhancement makes a particular claim more resilient to a defined amount of loss, but it cannot make weak collateral or joint defaults disappear.

Start with the pool, then the order of claims

Imagine a vehicle holding loans or other credit exposures. Its assets generate interest and principal, while its issued notes promise payments to investors. Those notes do not all have equal claims on the same cash. A contractual waterfall determines which expenses and noteholders receive cash first, and a loss waterfall determines who absorbs shortfalls. The equity or first-loss position has the greatest exposure to deterioration. Mezzanine claims sit between it and the senior notes. Seniority is therefore a rule for sharing losses, not a statement that the underlying borrowers have become safer.

The Bank for International Settlements comparison of CDOs and CLOs describes this ordering and stresses that the apparent protection of a senior tranche depends on both default risk and loss correlation. In exam work, first identify what is being lost: a borrower may default on its face amount, but recoveries reduce the realised pool loss. Then allocate that realised loss through the structure. Confusing defaulted principal with loss after recovery overstates the hit to every tranche.

An illustrative loss waterfall

Take a wholly invented pool with £100 million of collateral and four funded claims: £10 million of equity, £15 million of junior debt, £15 million of mezzanine debt and £60 million of senior debt. The amounts sum to the pool value at inception. Assume for this simple example that fees, timing differences, interest payments and replenishment are absent. Only principal loss matters, and losses are applied first to equity, then junior debt, then mezzanine debt, and finally senior debt.

Suppose £20 million of the pool defaults and the vehicle recovers £5 million. The economic loss is £15 million. Equity absorbs £10 million and is wiped out; junior debt absorbs the remaining £5 million and still has £10 million of principal. Mezzanine and senior principal remain intact. A statement that the senior note was unaffected is true for this scenario, yet it says nothing about a larger or more concentrated loss.

Now keep the original capital structure but suppose a separate, more severe scenario causes £45 million of realised pool loss after recoveries. Equity takes the first £10 million, junior debt the next £15 million, mezzanine debt the next £15 million, and the senior note takes the final £5 million. Its remaining principal is £55 million. The senior loss starts only after £40 million of claims beneath it have been consumed. The arithmetic shows the attachment point for this simplified structure; it is not a forecast of a real deal.

A useful check is to reconcile every step. The losses allocated across the four claims must equal the pool loss; no pound of loss can vanish in the waterfall. The surviving tranche balances must sum to the remaining asset value, assuming the simplified structure has no other assets, liabilities or charges. If your allocation totals more or less than £45 million in the second scenario, revisit the order of claims before drawing a conclusion about protection.

What credit enhancement actually changes

Subordination is the basic internal enhancement: claims below a note absorb losses before that note. Overcollateralisation adds collateral value beyond the debt claims it is intended to protect. Excess spread is the difference between income earned on collateral and the amounts owed to investors and expenses; subject to deal terms, it may absorb smaller shortfalls or be trapped. A reserve account provides cash already set aside. These mechanisms protect in different ways and on different time scales, so merely saying that a deal is enhanced is incomplete.

An overcollateralisation test compares eligible collateral value with the relevant outstanding claims under the transaction documents. A breach may redirect cash that would otherwise reach junior interests toward senior repayment. An interest coverage test instead asks whether eligible income covers required interest. A deal may appear to have a large principal cushion while weak cash income triggers a payment restriction. Equally, income can look adequate until credit losses erode asset value. Always read the specified numerator, denominator, valuation basis and trigger before interpreting a reported ratio; the simple classroom waterfall does not supply those contractual details.

External credit support is different. A guarantor or protection seller may promise to meet specified losses, moving some exposure from noteholders to that provider. This introduces the possibility that protection fails when it is needed. A strong-looking wrapper does not justify skipping analysis of the provider, the protection scope or exclusions. The BIS research on CDO rating methods shows why recovery and dependence assumptions matter alongside individual borrower default estimates. Enhancement is conditional protection under a contract, not a general guarantee of repayment.

Why correlation changes the answer

A pool of many exposures can still be vulnerable when borrowers respond to the same shock. If losses arrive gradually across unrelated names, the first-loss layers may absorb them while the vehicle receives time to collect cash. If many exposures weaken together, their losses can exhaust junior capital in the same period. The senior tranche is especially sensitive to the tail of the pool-loss distribution: it may be untouched in ordinary scenarios and then suffer once aggregate loss crosses its attachment point.

This is why counting assets is a poor substitute for testing common drivers. Exposures backed by different securities can still depend on one property market, one borrower sector, one funding channel or one model assumption. The BIS account of pre-crisis subprime CDOs describes weak diversification and losses moving together more strongly than investors had expected. That history is a warning about a mechanism, not a claim that every present-day structure has the same collateral or outcome.

Model risk adds another layer. An estimate of how frequently individual credits default does not by itself establish how likely a large pool loss is. Recoveries may be lower during broad stress, and correlations inferred from ordinary periods may fail when assets are sold at once. If an exam vignette gives two structures with identical expected losses but different dependence assumptions, do not assume that their senior claims have equal risk. Work from the loss distribution and the attachment point, not simply the average loss.

How to read an exam description

Use a fixed sequence. Identify the underlying exposures and whether the stated loss is before or after recovery. Write down the order in which claims absorb principal loss. Calculate how much protection sits beneath the tranche you are analysing. Then look for enhancements that add collateral, cash or third-party protection, and ask what event makes each feature operate. Finally consider common exposure, valuation assumptions and the possibility that an external provider fails. This sequence prevents a high rating or a long list of enhancements from becoming a shortcut around the pool itself.

The most common error is to treat seniority as elimination of credit risk. In the £15 million loss scenario the senior note survives because £40 million of subordinate claims was available, not because the pool loss was harmless. Another error is to apply the loss directly to senior notes before exhausting equity and junior claims. A third is to use a collateral coverage ratio as if it were an observed recovery rate. Those quantities answer different questions.

For a broader map of the credential, use the CAIA programme page. The Omni methodology page explains the study approach. When revising this concept, test yourself on a new, invented waterfall: change the recovery amount and recompute the allocation before reading the answer. Being able to explain why a senior loss begins at a specific point is more useful than memorising that senior notes are paid first.