Super-seniors are not easy things to understand, as you’ll know if you managed to trudge through my attempted explanation. I got some good questions in the comments, here’s my attempt at the answers.
Eli and fresnodan both bring up the issue of counterparty risk: after the bank has bought insurance on its CDOs, how does it know that its counterparty will have the money to pay up if and when there is an event of default?
It doesn’t always know for sure. But remember that synthetic bonds are structured so that the collateral payment gets invested up-front, so if the bank hedged its exposure by issuing synthetics, it’s probably fine.
On the other hand, as Noel notes, some banks ended up buying cheap protection on their super-senior tranches from AIG Financial Products. Which, as we’ve seen, was a very good deal for the banks, and a very bad idea for AIG. And, thanks to Uncle Sam, AIG is still around to pay out on those contracts.
Matthew asks a couple of questions. Firstly:
What’s the difference between this scenario than the bank just lending money to the top 86.6% of companies by creditworthiness and letting other lenders – perhaps specialists – lend to the other 13.4%? And aren’t the profits the same for the bank in both cases?
The answer is that it’s not the bottom 13.4% of companies by creditworthiness which always default. You don’t know which companies are going to default: you just know that some but not all companies are likely to. So you lend to them all and then protect yourself against the first 13.4% of losses. It’s a bit like saying "whoever the losers are, those companies, in hindsight, we’ll choose not to lend to".
His second question is this:
Why would the income from the CDOs go ‘to zero’? This would mean all 100% of mortgage payers (in this example) default. If this is the case then the problem is massive failure to understand the riskiness of an asset- but it would have to be absolutely massive to get it 100% wrong? In fact so massive, fraudulent, instead?
I tried to explain this here, in words, and here, in pictures. But in a nutshell, these CDO weren’t simple pools of mortgages. Instead, they were pools of junior tranches of mortgage-backed securities. And the junior tranche can go to zero even if quite a lot of homeowners continue to pay their mortgages in full and on time.
Anon asks whether "at least one factor in the ongoing ‘success’ of these instruments was based on expectations of a declining USD" — no. But he or she is quite right that the bankers
remained focused on short term profit pressures in the absence of ‘new’ ideas, confident both in apparent limits to their own personal liability and understanding that their firms would be too big to fail when the music stopped – and protected by bonuses and severance packages which would see them into very comfortable retirements IF proprietary knowledge of their firms did not keep them in demand for the rest of their careers.
And finally Kevin Drum reads me as saying that the banks were "creating a synthetic version of the subprime market that was even bigger than the original". This isn’t really true: they created a synthetic version of the subprime market that was actually smaller than the original — that was the problem, that it didn’t fully hedge the subprime assets they held on their books.
Kevin also says that the banks kept synthetic subprime CDOs on their own books — which happened in a few cases, notably at UBS, but was actually pretty rare. Normally they kept the real CDOs on their own books, and hedged by creating and selling the synthetics.