You can tell that leveraged super senior synthetic collateralized debt obligation tranches are fun because they are called leveraged super senior synthetic collateralized debt obligation tranches, and anything with that many words in its name is up to something. And in fact LSS CDOs were popular prior to the financial crisis, got various people in various kinds of trouble, and more or less vanished.

But now Euromoney is reporting that Citigroup is trying to market them again, with a slight modification that might get people into less or at least different kinds of trouble, though it is far from clear that anyone will be interested.

The basic idea of CDOs is that you are a bank (Citi), and you have, say, 100 loans each for $1 million, and you put them in a pot and slice up the default risk and sell it to various people. Someone -- call her the equity investor (it could be you) -- agrees to take the risk of the first 15 defaults; she loses money if even one loan defaults, but in exchange she gets paid a higher interest rate. The 16th through 100th defaults you sell to the senior investor; maybe you call him "super senior" to make him feel special. The senior investor only loses money if 16 or more loans default, which is generally unlikely, so he feels pretty good, and will charge you less for the protection than the equity investor will.

The problem is that it is hard these days to find people who want to put up a lot of money in order not to earn very much on it. So the solution is to pay the super senior guy more, but then you need a trick to come up with the money. The leveraged super senior approach does this as follows:

  1. Okay you are effectively buying $85 million worth of very safe loans from us.
  2. Those very safe loans don't pay very much, maybe 100 basis points over Libor.1
  3. But what we'll do is, we'll loan you $76.5 million to buy those loans.
  4. So you're only really putting up $8.5 million
  5. And we'll charge you even less on our loan to you than those loans pay.
  6. So you'll effectively make a 3.5 percent return on your $8.5 million.
  7. Which is pretty good?

Is it? This guy is unimpressed:

“We looked at the deal but according to our calculations it should be paying around 5.5%, rather than the 3.5% they are offering,” says one fund manager. “For me it’s a bit too expensive, and it seems that Citi are looking to get a cheap hedge and hoping some investors go for it.”

You can sort of see why; my dumb math suggests that Citi is charging 72 basis points for lending you the money, against a loan that pays 100 basis points. So you're borrowing 28 basis points more cheaply than you're lending.2 Whether that is a good business depends on whom you ask,3 but it is not a lavishly good business. (That guy's 5.5 percent gets you a 50 basis point spread.)

The novelty in this deal is that the treatment of the loan is different from the bad old days of leveraged super senior deals. In the old days, the investor would put up the $8.5 million, you'd loan him your $76.5 million, and it would be non-recourse: If you ate through the $8.5 million, he could just walk away rather than putting up more money. (Thus his "super senior" investment in the 15-to-100 defaults tranche looks a lot like a mezzanine investment in a 15-to-23.5 defaults tranche.) That was good for him and bad for you.

On the other hand, you got to eat the $8.5 million based on mark-to-market changes in pricing, not on actual defaults. That was bad for him and good for you. A lot of leveraged super seniors saw no actual losses on default, but big mark-to-market swings when everyone thought everything would default. Investors had to post more collateral and in some cases would walk away rather than posting more -- even though if they'd hung on to the trades until maturity they'd ultimately have gotten most or all of their collateral back.

This led to all sorts of mind-melting, most prominently at Deutsche Bank, because ascribing the appropriate value to the investor's option to walk away was (1) hard, as a matter of math, and (2) somewhat unpleasant, in that it would create large mark-to-market losses for banks in the middle of a financial crisis. Those valuation questions remain controversial to this very day.4

In the new Citi deal, the investor puts up the $8.5 million, but if there are more than $8.5 million in losses, he owes you more money. That's bad for him, and good for you. On the other hand, he only pays you more if there are actual defaults on actual loans, not just widening market spreads. That's good for him -- he's not subject to mark-to-market volatility -- and bad for you. It's bad for you because, if it looks to the market like there'll be a lot of defaults, you can't get any more collateral from him. Instead you just have to hope that, if the market is right and there are tons of defaults, he'll be good for it.5

Will he be good for it? Reading the Euromoney article you do not get the sense that anyone is unduly concerned about that question. You can read that "fund managers report Citi is offering 350bp to take the risk at 10 times leverage," and if this is a product that is being sold then I guess that makes sense: Everyone is buying the same thing. But if it is protection that is being bought from, effectively, investors who are acting as insurers, then it is weird that everyone is getting the same spread. Or, put another way: If Citi is in effect lending these investors 90 percent of their exposure, on a recourse, mostly unsecured basis, then you'd think Citi would charge them different rates for that loan depending on their credit quality.

(Also depending on their correlation to the underlying loans. Presumably the people who might buy this thing are people who buy things like this. If Citi's portfolio is seeing more than 15 percent defaults, those people might be in trouble. In all likelihood, they'll be in more trouble than this portfolio is.)

What can you conclude from that? I think you can conclude that:

  1. Citi is not that worried about actually having to come after these guys for more money, and
  2. These guys are not that worried about actually having to stump up more money.

Which, I mean, is sensible enough. These are leveraged super senior trades. The market value of that uncollateralized exposure today is, to a rough approximation, zero. The only way the investors will have to put up more money is if (1) there are enough defaults to eat through the junior 15 percent tranche and (2) there are enough more defaults to eat through the investor's 10 percent of the senior tranche. So 23.5 percent total defaults I guess. On an investment-grade-y, 150-basis-points-of-spread sort of portfolio. The only way that happens is in a pretty end-of-the-world scenario where the sorts of funds who'd buy this sort of thing have bigger problems than this particular thing.

One thing to think about trades like this is that they're pretty much always motivated by regulatory capital: Regulators require you to hold $X of capital against $100 million of loans, but they will only require you to hold $Y of capital against $15 million of equity tranche and $Z of capital against $76.5 million of leverage on $85 million of recourse leveraged super senior or whatever the particular package is here, and X > Y + Z. You could imaginably do something like this for risk-management reasons -- you don't want to lose money if all your loans default -- but, realistically, you're a bank, you can't go around being like "hmm what if all my loans defaulted?" If all your loans defaulted that would be irremediably terrible. And I suspect this trade wouldn't help much.

1 The pool involved here pays 150 basis points of spread, but that seems to be for the whole pool -- obviously the senior tranche should pay less, since it's cushioned by the junior tranches. I made up the 100 basis points number but it doesn't matter that much.

2 Here's my dumb math:

If you change the 100bps you get different answers but the difference between the tranche spread and the leverage cost is 20-35bps for any reasonable input there.
If you change the 100bps you get different answers but the difference between the tranche spread and the leverage cost is 20-35bps for any reasonable input there.

3 Consider banks.

4 The Deutsche Bank saga is told in these two Financial Times articles. William D. Cohan wrote about it for Bloomberg View this week. I wrote about it once or twice too.

5 Incidentally, I have a lot of respect for the market as a financial prognosticator but sometimes the market is wrong. Predicting the number of defaults in a loan pool where there are a lot of defaults -- say, during a financial crisis -- is a place where it is particularly likely to be wrong at some point. You've got 100 loans and things are fine and there have been no defaults: The market will predict that there will end up being, I don't know, 3 defaults, and that will be right or wrong or whatever. But then things get bad and 5 loans default and then they get worse and 5 more loans default and then they get terrible and 10 more loans default. Now 20 loans have defaulted. What will the market predict about how many of the original 100 loans will ultimately default? Well! Nobody can predict that 19 of them will default, since 20 already have. Your choices are 20 or "more than 20." The probability of "more than 20" is not zero so the market will imply some number of defaults larger than 20. This argument is true for any number of defaults. But at some point all the defaults that will happen have happened already, and the market's implication that there will be more defaults will be wrong. This is non-rigorous but you get the idea.