McClatchy has a breathless piece up on CDOs and other “exotic” transactions that Goldman did in the Caymans (hat tip reader John D). The problem is that the author got his hands on some very solid information (prospectuses of 40 deals) but the story itself is a bit of muddle. While it has some helpful new details, it breaks less new ground than its hyperventilating tone would suggest.
Reading this piece is a “find the pony” exercise. Readers are encouraged to comment on, correct or suggest alternative interpretations, and amplify my efforts to parse this article.
The confused reporting starts at the very top. This is the first, then the third paragraphs:
When financial titan Goldman Sachs joined some of its Wall Street rivals in late 2005 in secretly packaging a new breed of offshore securities, it gave prospective investors little hint that many of the deals were so risky that they could end up losing hundreds of millions of dollars on them..
The documents include the offering circulars for 40 of Goldman’s estimated 148 deals in the Cayman Islands over a seven-year period, including a dozen of its more exotic transactions tied to mortgages and consumer loans that it marketed in 2006 and 2007, at the crest of the booming market for subprime mortgages to marginally qualified borrowers.
Yves here. OK, they don’t tell you what the deals are (as in what type), and the time frame is muffed. The piece starts out by talking about late 2005 onward, then a mere two paragraphs later, it talks about 2002 onward. This is significant because the ISDA protocol for credit default swaps on asset backed securities was not created until June 2005, so capital markets firms could not write credit default swaps on mortgage backed securities in any meaningful fashion. It thus appears it has a pretty heterogenous set of deals that all used CDS and were in the Caymans, and is trying to make generalizations across them. It’s pretty certain to be a mix of what most journalists and market participants call CLOs (collateralized loan obligations) which confusingly are considered to be a type of CDO but from a complexity standpoint are not remotely as hairy and opaque as ABS CDOs, which are resecuritizations (major constituent is tranches of asset backed securities, usually mortgage backed securities, while CLOs are made of whole loans).
We learn that at least some of the deals were synthetic, either wholly synthetic or to a large degree:
In some of these transactions, investors not only bought shaky securities backed by residential mortgages, but also took on the role of insurers by agreeing to pay Goldman and others massive sums if risky home loans nose-dived in value — as Goldman was effectively betting they would.
The “investor” in a synthetic deal is effectively a protection seller, acting in the role of guarantor. He receives income from credit default swap “premiums” and if the deal does badly, has to make good on the guarantee. And for CDS related to asset backed securities, many of the contracts required the “protection seller” or guarantor to pony up a payment merely in the event of a downgrade (while for CDS written on corporations, who are the “reference entities” used in hedging corporate bonds, the trigger is typically a default, bankruptcy filing, and sometimes a reorganization).
Yves here. Soon, we get to this confused paragraph:
Some of the investors, including foreign banks and even Wall Street giant Merrill Lynch, may have been comforted by the high grades Wall Street ratings agencies had assigned to many of the securities. However, some of the buyers apparently agreed to insure Goldman well after the performance of many offshore deals weakened significantly beginning in June 2006.
Yves here. UBS was an active buyer of CDO tranches from other banks for its own account, and people in the CDO business have told me that was atypical. Eurobanks that were NOT originators were often active buyers. I would enjoy some reader input here as to how often and why a capital markets firm would be an “investor” in a CDO of another firm. Perhaps this was pursuant to a correlation trade (it was popular for hedge funds and prop desks to trade “correlation risk” which often wound up looking like a simple spread trade, go long one tranche, say the BBB, and approximate a short of the next higher rated tranche). My impression is that this was not common, but “not common” in a market this big could still add up to a fair bit of activity.
This bit is simply frustrating:
Many of Goldman’s winning bets with other large U.S. banks raised the price tags of 2008’s government bailouts of Citigroup, Bank of America, Morgan Stanley and others by sums that no one has yet determined because the contracts are private, according to people familiar with some of the transactions.
However, one billion-dollar transaction that Goldman assembled in early 2006 is illustrative. It called for the firm to receive as much as $720 million from Merrill Lynch and other investors if defaults surged in a pool of dicey U.S. residential mortgages, according to documents in a court dispute among the parties.
Yves here. OK, they have access to the lawsuit…and this is the summary? No context as to what the deal was about? This is lame. Could this have been part of a correlation trade for Merrill? The story also implies Merrill was a significant participant, which may not be the case. I did a quick search and could not find any references save the McClatchy story. Any reader input would be of interest.
The author also gets very excited about the Caymans angle, when that was not a big deal. As Tom Adams, former head of structured finance at FGIC noted, “Offshore deals happened all of the time – its how certain private placements had to be done. The deal documents looked exactly the same as US deals and were drafted by the same lawyers.”
The article goes through some background, including how much Goldman originated in RMBS and related instruments ($40 billion in 2006 and 2007). This is where the piece goes a bit off the rails:
In 2006 and 2007, as the housing market peaked, Goldman underwrote $51 billion of deals in what mushroomed into an under-the-radar, $500 billion offshore frenzy, according to data from the financial services firm Dealogic. At least 31 Goldman deals in that period involved mortgages and other consumer loans and are still sheltered by the Caymans’ opaque regulatory apparatus.
Yves here. That number is most assuredly NOT all RMBS CDOs. In fact, have looked at a couple of the Abacus deals, one has some RMBS, but also included other exposures. There are industry databases that do show collateral compostion; Dealogic is not one of them. Back to the article:
Goldman’s wagers against mortgage securities similar to those it was selling to its clients are now the subject of an inquiry by the Securities and Exchange Commission, according to two people familiar with the matter who declined to be identified because of its sensitivity.
Yves again. We have another sloppy bit here:
Goldman’s subprime dealings burned taxpayers a second way…three foreign banks — France’s Calyon and Societe Generale and the Bank of Montreal — bought protection against securities they purchased in Goldman’s Caymans deals, using AIG as a backstop.
Those banks got a total of $22.6 billion from AIG (Societe Generale $16.9 billion, Calyon $4.3 billion and Bank of Montreal $1.4 billion), though not all of the money was related to investments in deals underwritten by Goldman.
Yves here. Look, I am as fond of taking on Goldman (or any securities firm engaging in bad behavior) but this is simply irresponsible. Three foreign banks that hedged some Goldman deals with AIG got bailouts. Got that part. The article then cites the GROSS AMOUNT of the rescue, and cutely throws in “though not all the money” was related to the Goldman deals.
The author evidently has NO IDEA how big these exposures are, and they could have been comparatively small ($50 million? $100 million?) But putting big number of the total exposure first creates the impression that the amount related to Goldman was large (Google “anchoring”), particularly since the disclaimer is so half-hearted. “Not all” implies “most” when it appears the author has no idea of the actual numbers.
Now having said that, there are still some useful bits.
ACA was deeply involved, and I don’t recall seeing numbers this large before:
The insurance unit of ACA Capital Holdings Inc. wrote $65 billion in swaps coverage, mostly on the Caymans deals called collateralized debt obligations, or CDOs, before it folded and turned nearly all its assets over to the banks that had thought ACA would backstop them.
And this section was also worthwhile:
_ Goldman’s Caymans deals were riddled with potential conflicts of interest, which Goldman disclosed deep in prospectuses that typically ran 200 pages or more. Goldman created the companies that oversaw the deals, selected many of the securities to be peddled, including mortgages it had securitized, and in several instances placed huge bets against similar loans.
_ Despite Goldman’s assertion that its top executives didn’t decide to exit the risky mortgage securities market until December 2006, the documents indicate that Goldman secretly bet on a sharp housing downturn much earlier than that.
_ Goldman pegged at least 11 of its Caymans deals in 2006 and 2007 on swaps tied in some cases to the performance of a bundle of securities that it neither owned nor sold, but used as markers to coax investors into covering its bets on a housing downturn.
There is a particularly interesting contention:
The Wall Street figure who insisted upon anonymity said that despite all the hoopla, there were few private investors in CDOs, and that banks have suffered most of the losses, one reason “why you haven’t seen a lot of complaining.”
Yves here. Again, it is hard to know for sure who the investors are, since only the packager/underwriter knows for sure. But based on reports from some who were active in that business, plus various press reports, this statement is narrowly true, but a lot of these banks were second and third tier European institutions. Remember “banks” covers a multitude of sins. And reader Crocodile Chuck, via e-mail, reminds us that CDOs were sold to Australian town councils and fire brigades:
The councils here are under continual barrage of ‘cost shifting’ from the state and fed govt’s. On top of this, liability cover for parks, public places is increasing. The population is aging, and consumes more services (many delivered locally). Everyone has to do more with less.
Of course the ‘officers’ for investment went for yield-AS LONG AS IT
WAS ‘AAA’. A lot of these ppl haven’t been to university.
Yves here. Having said that, the councils would not be targets for a pure synthetic deal, but for a cash or hybrid (one with bonds and CDS both as the assets). But the McClatchy piece makes clear that some of the deals were hybrids.
I wish organizations like McClatchy would put stuff like the deal documents it has on line once an article like this has been out for two weeks. They are unlikely to refer to them again, and if so, only in a cursory way. Fresh sets of eyes could get a great deal more out of them.