Combination Notes and First-to-Default Baskets
Book: Structured Finance and Collateralized Debt Obligations | Author: Janet M. Tavakoli | Publisher: John Wiley & Sons (2008) | ISBN: 978-0-470-44344-6
The second half of Chapter 10 covers some of the more creative packaging that grew up around CDO equity, plus first-to-default baskets, which deserve their own honest discussion because the pricing is genuinely difficult and a lot of people get it wrong.
Actively Traded Synthetic CDOs
When bank arrangers hold a claim to part of the residual cash flows, they sometimes also get the right to actively trade the portfolio or perform limited substitution. This is especially common in synthetic arbitrage CDOs.
Many investors are skeptical. They argue that a manager would not have avoided Enron or Railtrack. They worry that if the bank arranger has trading rights and once losses exceed the equity investment, the arranger has every incentive to dump deteriorating credits into the portfolio. The arranger pockets the extra spread. The tranche investors up the line absorb the losses.
The defense is good documentation. Limited substitution provisions can cap the number of obligors substituted per year, require rating minimums for substituted names, and restrict substitution unless the exiting name has widened significantly relative to its peer group. Some structures even give outside super senior investors a right of refusal on substitutions.
Interest Subparticipations: When Equity Is Not Really First Loss
In Europe, particularly in Germany and the Netherlands, a structure called the interest subparticipation (ISP) appeared in many balance-sheet CDOs. It exploits a regulatory capital loophole.
Normally, equity attracts a one-for-one capital charge. If a bank holds EUR 10 million in equity, it must hold EUR 10 million in capital against it. But in Germany and some other venues, future interest income has zero capital charge. The ISP structure appears to sell the equity risk to an investor while the bank actually retains first-loss risk.
How? The ISP investor is entitled to an interest subparticipation in the reference portfolio’s future interest income. If a loss is allocated to the equity position, the issuer pays the ISP investor a payment equal to that loss. That payment comes from the issuer’s own funds, not from the reference portfolio’s interest. The calculation is based on cumulative interest earned by the reference pool, which is used as a sizing mechanism to ensure the investor almost certainly gets paid. But the payment source is the issuer.
If the reference portfolio has an 8% gross interest rate and the first-loss tranche is only 2% of the deal, there is four times as much interest available as needed to cover potential losses. Very unlikely the ISP investor does not get paid. Interest subparticipation pieces have received private ratings as high as AAA or AA, sometimes as low as A.
This means the bank effectively retained first-loss risk while appearing to have sold it. European bank portfolio managers bought these as if they were slightly enhanced rated floaters, earning 20 to 40 bps premium over comparable-rated instruments when the structure was new. By 2002 that premium had disappeared as the market understood the structure better.
Participation Notes
Participation notes (PNs) combine a rated CDO tranche with an equity tranche. They give investors rated exposure that also participates in equity cash flows.
An example from the book: combine EUR 8 million of a AAA tranche and EUR 2 million of equity. The equity is returning 15.5% and the AAA is at Euribor plus 50. The combined note gets an Aa2 rating for return of principal and for payment of a coupon of Euribor minus 120 bps. But the expected coupon on the note is Euribor plus 120 bps, because of the equity participation upside. In the same market environment, plain Aa2 notes are trading at Euribor plus 80.
The key structural detail: the equity slice in a participation note is a vertical slice, meaning it participates pro rata with other equity investors. Same seniority. The rated portion provides principal protection, and the equity portion provides income upside.
This lets investors who are restricted to rated securities get equity-like returns. Insurance companies in the US got favorable capital treatment for notes rated above single A. The NAIC would rate these securities NAIC 1, the highest classification, which carries the lowest capital charge.
One cautionary tale: a New York branch of a European bank bought a portfolio of CDO tranche investments, some of which turned out to be participation notes. The notes were rated AA or better, but only for return of principal. The coupon depended on equity performance. When equity tranches deteriorated due to credit losses, income on the participation notes declined. Management was surprised. The portfolio manager had invested within the guidelines as written. The problem was that management expected the AA rating to apply to both principal and interest. It did not.
The bank could not liquidate the positions without taking a loss. Beyond the direct income loss, there was opportunity cost: capital was locked up until maturity.
Capped Participation Notes
Some participation notes use a senior slice of the equity tranche, not a pari passu slice. This means the senior equity slice has a preset yield, and the participation note coupon is capped at that level. Less volatility, but you give up the full upside.
In a standard participation note, the equity investment is pari passu with other equity investors. First loss hits everyone equally. In a capped participation note, junior equity holders absorb losses first, and then the senior equity portion of the participation note. The senior equity investor is protected from early losses. In exchange, the coupon is capped.
Combination Notes
Combination notes (or combo notes) take two rated CDO tranches, not equity, and package them together. For example: EUR 6.66 million of the AAA tranche (Euribor plus 50) and EUR 3.34 million of the Baa2 tranche (Euribor plus 285). Combined, the note gets an A2 rating for timely payment of principal and coupons, with a coupon of LIBOR plus 128. But in that market environment, comparable A2 CDO tranches traded at LIBOR plus 155. The combination note looks unattractive versus the alternative.
Combination notes can be created at deal close or any time before it. If you want a combination note from an already-closed deal, the tranches can be combined in an SPV and a new note issued. That is slightly more expensive than doing it at the original closing.
Investor Motivation
Why do participation and combination notes exist? Two main reasons:
First, investors who cannot buy unrated equity can access equity-like returns in rated form. US insurance companies get favorable capital treatment for notes above single A. Participation notes structured as a single CUSIP number do not get split into rated and unrated components for capital purposes.
Second, banks and funds that require rated securities for their portfolios can also access the upside. Under Basel II, highly rated notes get better capital treatment. The combination note or participation note is a way to get yield without triggering unfavorable capital charges.
Principal-Protected Structures
For investors who want equity exposure but need to protect the original investment, equity can be repackaged with a zero-coupon bond. The SPE issues a note whose proceeds buy the zero-coupon bond, which accretes to par at maturity for principal repayment. The remaining proceeds go into equity, which pays current income on the note.
A longer-dated zero-coupon bond costs less today, which means a larger portion of the proceeds can go into equity. More equity participation, lower protection cost.
For German insurance companies, the Schuldschein version of this structure had the added accounting benefit that it did not need to be marked to market.
An important pricing insight: the equity IRR without protection goes negative at around 3% annual default rate for the example in the book. The IRR of the principal-protected note starts lower than unprotected equity but never goes negative, because principal is guaranteed by the AA1-rated zero-coupon bond. Default of reference credits in the CDO does not affect that bond’s principal.
First-to-Default Basket Swaps
Tavakoli is unusually direct here: “I have no confidence in any first-to-default pricing model I have ever seen.”
That is not an invitation to avoid the market. It is an invitation to negotiate hard and not let a model’s output end a discussion.
A first-to-default basket references a small number of obligors, usually 10 to 20. The protection seller receives a premium and pays only on the first default in the basket. Maximum loss is par minus recovery on the first defaulting name. In exchange, the premium is higher than any single-name CDS in the basket.
For investment-grade names in diverse industries, the Street prices first-to-default premiums at roughly two to three times the average spread. If the average single-name spread is 100 bps, a first-to-default basket price of 220 to 330 bps is typical.
But here is where pricing models go wrong.
Lower bound error: One example in the book calculated the highly correlated case as 155 bps × 15% = 23 bps. That is impossible. The lower bound for the first-to-default premium must be higher than the narrowest spread in the basket, because even if all names are perfectly correlated, the first-to-default is at minimum as risky as the riskiest single name.
Upper bound misunderstanding: The upper bound is not the sum of individual CDS spreads, even though that is a common answer from experienced people. Why? Because the protection seller only pays on one credit event, not all of them. The probability of paying out is not the sum of individual probabilities. It is the probability that at least one name defaults. For investment-grade names with low individual default probabilities, that is much lower than the sum.
For four investment-grade names with two-year cumulative default probabilities of 0.7%, 0.5%, 0.6%, and 0.7%, the two-year probability of at least one default (assuming no correlation) is 2.48%. That is higher than any individual probability but well below their sum of 2.5%.
For non-investment-grade names with individual default probabilities of 11.71%, 14.20%, 13.50%, and 14.20%, the two-year probability of at least one default is 43.78%. Much higher. The magnitude is completely different and directly affects pricing.
Key insight: uncorrelated credits should command a higher first-to-default premium than correlated credits. With uncorrelated credits, any of four independent causes could trigger a default. With highly correlated credits, one systemic factor drives most defaults, so the probability of triggering is lower.
Moody’s will provide ratings on mezzanine risk of default baskets. Tavakoli is skeptical. Small sample sizes, statistical uncertainty, and the track record of rating agencies in recent years do not inspire confidence. Use ratings as a convenience if your mandate requires them. Do not mistake them for accurate predictions.
First-to-Default Notes
An alternative to the basket swap: a credit-linked note with first-to-default basket risk embedded in an EMTN. The investor purchases the note and takes the first-to-default risk. This is purely a repackaging into funded form for investors who need notes rather than swaps.
The Smartest Equity Investment: Protection Money
The final section of Chapter 10 describes what sophisticated investors who understand corporate finance actually do.
They form their own independent view of default probability and recovery value. They handpick the names in the basket they are willing to write protection on. They negotiate the premium based on their analysis rather than accepting whatever the model spits out.
An example from the book: on a basket including Collins & Aikman, investors were paid an up-front premium of 85 cents on the dollar. When Collins & Aikman defaulted, they locked in a 2% gain on the notional amount allocated to that name. Before the default, they had full use of the 85-cent premium in cash.
The market occasionally misprices this risk. Finding the mispricing requires genuine fundamental credit analysis. The structurers and traders who rely on correlation data from Monte Carlo models rather than fundamental valuation often do not know the difference.
That is the edge: knowing when the model is wrong, because you have done the actual work of understanding the underlying business.
Previous: CDO Equity Structures: Where the Real Money Lives Next: CDO Managers: The Good, the Bad, and the Valued Few