Gillian Tett has a good column on the super-senior game, in which she chastises banks for losing sums "larger than the gross domestic product of many countries" by entering into "a humongous, misplaced bet on a carry trade that was so simple that even a first-year economics student (or Financial Times journalist) could understand it". Super-senior tranches of CDOs, she explains, were zero risk weighted by dint of their triple-A credit rating, and the banks never stopped to ask why they yielded 10bp more than risk-free funds:
But, last August it became clear why this 10bp spread existed: namely, because these assets are not as liquid as government bonds in a crisis. Indeed, the prices of some tranches of debt have fallen by 30 per cent in recent months, to the shock of senior managers.
I don’t think that bonds fall by 30% just because they’re illiquid in a crisis: there was definitely a lot more credit risk buried in these tranches than the ratings agencies ever imagined.
But more to the point, there’s a strong case to be made that the super-senior tranches of CDOs were never worth what the banks booked them at, not even before the market crashed. Here’s the bit of the story that Tett only hints at when she talks of banks buying these instruments "simply to keep the CDO machine running".
Many banks, up until last summer, were making enormous sums of money buy packaging bonds into CDOs. They would buy up a bunch of debt, roll it all up into a CDO, and then slice it into tranches ranging from the super-senior all the way down to equity. The riskier the tranche, the higher the yield. And somehow, by the magic of securitization, the value of all a CDO’s different tranches, in aggregate, would be a little bit higher than the cost of all the bonds which went into it. And that difference was profit for the bank.
Except. When you slice up a CDO, you end up with a relatively small amount of relatively high-risk debt, and a very large amount of triple-A debt. The riskiest triple-A tranches could get sold off to yield-hungry investors who were only allowed to buy debt with a triple-A rating, but no one was interested in the "super-senior" tranches which yielded only 10bp over government bonds but which were more or less impossible to sell since there was never a market in them.
If the banks had sold the super-senior debt at a market yield – if they’d bumped up the coupons on the super-senior tranches to the point at which some investor would have been willing to buy them – then the magic profits of securitization would have been eradicated, and the efficient markets hypothesis would have held: you can’t make a pie bigger by slicing it laterally rather than radially.
But the bankers’ bonuses were all tied to the idea that you could make a pie bigger by slicing it laterally. And so they found the only buyer they could for the underpriced super-senior tranches: themselves.
Think about it this way. I buy the ingredients for an apple pie: apples, sugar, flour, that sort of thing. Add them all up, and they come to $10. I then cut the pie into four unequal slices. I sell one for $1, another for $3, and a third for $5. The fourth slice I price at $2, but I can’t find any takers.
Now in theory, if I could sell that fourth slice for $2, then I’d have a nice little business going. But I can’t, so instead I keep a hold of that fourth slice myself, telling everybody that it’s worth $2, and booking my $1 profit. Then, one day, the game stops, people start asking awkward questions about how much that slice is really worth, and it turns out that when I test the market, no one’s willing to pay even $1 for it, let alone $2. And that’s when I write down the value of my slice and take a big loss for the quarter.
But hey, at least I got nice big bonuses so long as the game was going.