The topic du jour is apparently Goldman's synthetic CDOs. Gretchen Morgenson wrote a typically ill-informed and absurd front-page article on this subject last week (can the NYT please stick a fact-checker on her?), which got a considerable amount of attention. Yves Smith also wrote a long post denouncing Goldman's "deceptive synthetic CDO practices." Most of Yves' arguments are frankly baseless, but after an exchange with Yves in the comments section, I think I've identified the nub of the issue.
Yves' argument is that Goldman should have disclosed to investors that it was planning to retain the short position in the synthetic CDO, and not simply act as an intermediary. Arranging banks necessarily take the initial short position in a synthetic CDO, but most would then sell off the short position, pocketing just the arranging fee. Goldman, however, arranged some synthetic CDOs (emphasis on some) where it kept the short position for itself — and when the housing and structured finance markets collapsed, it was Ferraris for everyone! Yves thinks Goldman should have disclosed to investors that it would be in a position to profit if the synthetic CDO declined in value. I disagree.
First of all, investors don't have a right to know what Goldman's internal positions are, and they never have. Every single investor understood that Goldman also invested for its own account, and no one would have expected them to disclose their proprietary positions.
More importantly, what you have to realize — and where I think Yves goes wrong — is that Goldman wasn't necessarily placing an independent bet against the synthetic CDO market; rather, it was using synthetic CDOs to bet against the housing market. There's a big difference. Goldman was betting that the housing bubble would burst, and that the resulting decline in the housing market would be reflected in synthetic CDOs referencing mortgage-backed CDOs. (The mechanism was this: declining housing prices → higher default rates → reduced cash flows to mortgage-backed securities → lower RMBS/CDO prices → higher value of CDS protection on RMBS/CDOs → ca-ching!)
Goldman wasn't betting that liquidity would vanish and the CDO market would completely collapse — that was an unexpected windfall. In that sense, they got lucky (more on that later). Think about it like this: Goldman was betting that RMBS/CDO prices would fall from 100 to 75 because of weak fundamentals in the housing market, so they put themselves in a position to benefit from a decline in RMBS/CDO prices; in reality, the first hints of weakness in housing caused liquidity to completely dry up and RMBS/CDO prices to collapse from 100 to 50. They definitely weren't expecting that.
If Goldman's plan all along was for the synthetic CDO market to collapse, then why were they consistently the biggest liquidity provider (by far) in structured products and structured finance CDS (i.e., ABX tranches)? The point is that Goldman didn't need to artificially drive down the synthetic CDO market. They set themselves up to profit from the decline in synthetic CDOs that would follow naturally from weakening fundamentals in the housing market. By the same token, Goldman didn't need to manipulate the structure of the synthetic CDOs they arranged; any run-of-the-mill synthetic CDO referencing subprime-backed cash CDOs (to the extent there was such a thing) would've suffered steep price declines once housing prices started plummeting. What's more, later investors sometimes made the deal conditional on Goldman retaining the equity tranche, or required that the senior or super-senior tranches contain put options, so it's not like investors were defenseless — they had tools they could use to protect themselves, and they sometimes did.
It's also important to examine the timeline. Goldman's Abacus shelf was started in 2004, and Goldman didn't start going short the housing market as a firm until December 2006. When they did decide to start shorting the housing market in December 2006, they were initially just hedging the long exposure to CDOs they had built up, and they primarily executed this hedge by shorting the lower-rated ABX tranches (an index CDS referencing 20 subprime-backed CDOs) rather than retaining the short position in synthetic CDOs. This makes sense: their long CDO exposure in early 2007 would have been concentrated in 2005-2006 vintages, and the ABX was the only product that would have allowed them to hedge earlier-vintage CDOs (by, for example, buying protection on the ABX BBB- 06-2 tranche, which references 2006-vintage CDOs).
It wasn't until late April 2007 that Goldman started unloading its CDO inventory and took a real directional short position on housing, and when they did, they again primarily used the ABX rather than retaining the short position in synthetics they arranged. From Charles Ellis' The Partnership:
In late April, Dan Sparks, head of mortgages, and two traders, Josh Birnbaum and Michael Swenson, met with a small group of senior executives and warned of a major problem with the firm’s inventory of $10 billion in CDOs: It was heading south. Sparks wanted the firm to cancel underwriting any pending CDO issues, sell all the inventory it could, and make major bets against the ABX index. Sparks’s recommendation was accepted and implemented. (emphasis added)I know Goldman underwrote a few more CDOs after April 2007 — it was probably too late to cancel the May issues — but to my knowledge, not many of them were residential mortgage-related (they were mostly backed by CMBS and bank loans).
So let's sum up: Goldman arranged synthetic CDOs off its Abacus shelf from 2004 to 2007; they started betting against housing via synthetics and the ABX in December 2006, and began to take a directional short position on housing in late April 2007; they arranged — at most — only a handful of cash and synthetic CDOs backed by RMBS after they decided to actively short the housing market; in those deals, they fully disclosed that they would be taking the initial short position in the synthetic deals, and that they might choose to retain the short position; every investor understood that Goldman actively invested for its own account, and none of the investors had any reason to expect Goldman to disclose its proprietary positions; and Goldman had publicly stated on numerous occasions that it was bearish on housing. Yeah, it's safe to say that critics of Goldman are grasping at straws.
If you ask me, Goldman was almost as much lucky as they were good. Like I said earlier, they were expecting serious declines in the CDO market, but the utter implosion of the CDO market was an unexpected windfall. Also, the only reason they started to take a closer look at housing — as Lloyd Blankfein has stated on several occasions — was because they took mark-to-market losses on their MBS holdings for a few days in a row in late 2006. That prompted them to dig deeper, and they subsequently realized that the housing market was in serious trouble. That's all well and good, but Goldman got lucky: what if the MBS market had collapsed over a weekend, without any prior warning? Why did Goldman need mark-to-market losses to tip them off about the horrid state of the subprime market? Shouldn't they have learned how bad things were when their RMBS and CDO underwriting teams did their due diligence? They definitely deserve credit for investigating further after the mark-to-market losses, but that's not how it should have happened.