In a recent phone interview, Jody Shenn, reporter on the securitized products beat at Bloomberg, asked me why Goldman would have created a deal that let the manager choose to pay off subordinated bonds while simultaneously leaving senior bonds very likely to default.
He had me stumped. The obvious reason – that there was a benefit to Goldman, just didn’t suffice.
First of all, the deals dated to 2006, and the very idea of arbitrarily choosing to pay off junior bonds ahead of senior bonds three years later seemed to stand the entire concept of senior/junior on its head.
A couple of obvious advantages to Goldman or their best clients came to mind, as did the benign concept that they were simply getting rid of the expensive debt first, the way a family would pay off its most expensive credit cards, then car loans, if one of the breadwinners’ paycheck seemed at risk.
That kindest explanation fell flat as soon as I refreshed my memory of the Abacus deals. It had taken all of 30 seconds for me to decide I wasn’t interested back when they came out. It was a virtual kennel — dogs everywhere.
The deal was all “synthetic”, so it had a whopping $1.4 billion in Credit Default Swaps (CDS) in it, written to emulate the performance of what was probably a much smaller universe of bonds in the real world. But that’s the way it worked in 2006…. if you couldn’t buy or sell enough bonds in the market, you just ginned them up out of thin air using swaps. The best part was, you had a rock solid, black letter law protecting you from any prying eyes or weenie regulators. You know, those people who thought that risk should involve capital reserves and disclosure.
Today I got a call from a friend who saw the Bloomberg article, and we took a little stroll down memory lane. He recalled a bizarre conversation with the Goldman trader who was anxiously trying to sell a new subprime bond deal issued by one of Goldman’s special purpose vehicles.
My friend had already been burned by that particular set of documents in an earlier issue in the series, and he went after the Goldman trader big-time over a deal that had been closed a month or two earlier with new loans in it that were already defaulting when Goldman sold them to the deal.
One thing is sure — if a borrower is already two months’ late on a second mortgage that is only three months old, the chances of ever seeing a dime from that loan are even slimmer than that very same dime.
Since the usual industry practice was for the seller to buy back any any loans that defaulted right out of the gates and replace them with performing loans, my friend started leaning on the Goldman trader to do just that. The Goldman trader said the documents for their “shelf” – the shell company that issued the bonds – did not require a buy-back on early period defaults, or EPD’s.
Needless to say, there was no sale for the new bonds under the same documents. That didn’t stop Goldman from issuing tens of billions in subprime bonds each year quite successfully, right up until the music stopped in 2007.
Not every buyer was that careful. Of course, when the buyer was a CDO asset manager created and controlled by that same dealer, you know they were going to buy the bonds.
But Goldman was also playing the other side of the trade at the very same time (2006 onward). They were buying insurance on their own deals, and on deals from the rest of the Street. That’s how they got paid more than $10 billion of taxpayer money by AIG last year, money Goldman will never have to pay back.
They also had a group of very savvy hedge fund customers who were on the short side of the trade along with them in 2006 and 2007. For the most part, Goldman was buying insurance on the top-ranked bonds, and then holding and swapping them, a trade that created near-zero capital charge when the counterparty (AIG) was AAA, and allowed a sliver of a few basis points per annum to fall out of the deal machine and into Goldman’s pocket if the borrowers managed to pay off their loans and all went well.
If the borrowers didn’t pay, that was why you bought insurance, especially when it was really really cheap.
But what about those hedge funds who were also buying CDS insurance on subprime deals? Most of them didn’t bet that the world would blow up by buying protection on the AAA bonds. Instead, they would buy insurance against lower-rated tranches from these subprime CDO deals so their payoff would come if defaults and losses doubled. Those AAA and super-senior AAA bonds were so much cheaper to insure because they didn’t get hit unless losses came in five to ten times as bad as history.
My friend recalled discussing new bonds coming into the market with other portfolio managers. While we were doing our best to make distinctions between the issuers — avoiding most Street deals, Wamu’s Long Beach, Ameriquest and others, some investors were simultaneously buying insurance.
The classic hedged strategy is to find a poorly underwritten deal and buy insurance on it while simultaneously going long on bonds from the better lenders.
Since the Abacus deals were huge, and Goldman was a major player (and the manager), it made sense to buy a better quality loan underwriter’s issue while simultaneously buying insurance on a lower-rated slice of Abacus. Then when the lousy underwriters’ loans went down the tubes first, the insurance could kick in first on the Abacus deal.
Now, imagine this:
You bought insurance on the lower- rated Abacus tranches, and did your due diligence to buy into deals from more careful underwriters. Three years later, and the losses are piling up so high that even the “good” deals are headed for doom. You feel OK about it, because you hedged by synthetically shorting a lousier deal – Abacus.
Suddenly, the lower-rated tranches in Abacus are paid off at par, even though they are not due to be paid yet, and the senior bonds haven’t been paid. You’ve just lost your shirt on your short hedge, and the losses on your long position are right around the corner.
The Rating Agency lowered the rating on the highest priority bonds to very low junk because it’s clear there won’t be enough money to pay them their principal, so it’s more than simple timing here.
So how did $66 million of lower-rated bonds get paid without a penny of loss? Well, what if the dealer that controlled that CDO manager wrote far more insurance than $66 million? I can’t say it’s true, since the CDS are private contracts protected by law from disclosure. But what if?
As my friend pointed out: There were more ways to get screwed in this deal than there are positions in the Kama Sutra.