Not that I have the time or patience to dig through 250,000 pages of documents, but I have a nagging suspicion that the people who are pouring through various AIG-related disclosures may be missing key points or snookered into interpretations that may be unduly flattering to various banksters.
The focus of the recent investigations into Fed secrecy relative to AIG takes up the theme that the Fed was anxious to hid the fact that it had paid out 100% of the value of toxic CDOs to various AIG counterparties. That is because that concern crops up repeatedly in internal communications. A second reason cited is that the Fed wanted to hide who benefitted most from the rescues (as in seeing the various transactions would allow one to see which CDOs had the deepest discounts).
I have trouble with theory 2. First, we now know who the biggest recipients of AIG-related subsidies were, even by exposure (ie, securities lending versus Maiden Lane III), yet the Fed even as of last week was writing LONG and clearly bogus defenses of why it needed to keep transaction level details a deep dark secret. Second, the differences in dreckiness among these CDOs is not all that large. Despite BlackRock valuing this paper at an average of less than 50 cents on the dollar (which they can do, this is all model based, they no doubt have estimated defaults and loss severities on these bonds that all tie out nicely), given how low the ratings are across the portfolio (most of the stuff is rated CCC or lower), market prices are more in the 20 cents and below range. In Japan, the expression for making fine distinctions among things that are probably not worth parsing that neatly translates roughly as “a height competition among peanuts.” This sounds like that sort of exercise. Hence, I suspect there may be some interesting, but not scandalous, information to be gleaned in divulging “who sold the worst turkeys to AIG.” This paper ALL performed horridly, that’s why it was stuffed into Maiden Lane III.
To switch to a particular example of how information is being read incorrectly, consider this example from Bloomberg this evening, in which I deem the reporter to have been spun successfully:
Goldman Sachs Group Inc. was the most aggressive bank counterparty to American International Group Inc. before its bailout, demanding more collateral while assigning lower values to real estate assets backed by the insurer, documents obtained by lawmakers show.
A month before the September 2008 rescue, Goldman Sachs approached AIG about tearing up contracts protecting the bank against losses on collateralized debt obligations, or holdings backed by mortgages, according to a BlackRock Inc. presentation dated Nov. 5, 2008. Goldman Sachs was the only counterparty willing to cancel the credit-default swaps and bear the risk of further CDO losses, provided that AIG make payments based on the bank’s larger-than-average estimate of market declines.
“Goldman Sachs is the least risk-averse counterparty,” according to the presentation, which was prepared by the asset manager for AIG and later given to the Federal Reserve Bank of New York. The firm is “the only counterparty willing to tear up CDS with AIG at agreed-upon prices and retain CDO exposure.”
Yves here. I wouldn’t call Goldman’s actions “least risk averse” as BlackRock did. I’d call it MOST risk averse. They were marking their deals more aggressively than anyone else, which had the effect of allowing them to suck more collateral out of AIG. What would you rather have? A. Cash. B. A contract for possible future payments with a counterparty whose credit quality is falling like a rock. C. Toxic CDOs that no one will buy at this moment, and current conditions may well prove permanent. Tell me, how a strategy to maximize your current cash is NOT risk averse? Marking down your CDOs aggressively and seeing if you can get AIG to cancel the contract at the value you’ve marked it at is a current cash maximizing approach. (And I’d like to know what sexual favors were exchanged to get BlackRock to issue such a Goldman-favoring report).
Goldman’s willingness to tear up the agreements could just as well reflect its willingness to reduce its exposure to AIG if it could settle on its current marks, ie, show no loss on contract cancelation. One of the anomalies we identified in our effort to drill into Maiden Lane III transaction-level detail was that AIG appeared to be somehow getting away with posting less collateral with counterparties than was contractually stipulated. That would make it even more rational for Goldman to be willing to accept a deal in which AIG settled up:
The collateral calls also appear to be a smaller amount than the marks (after taking into account the thresholds) would indicate. Put another way, based on the marks, the collateral calls should have been larger than they were, in aggregate and by counterparty, than they were described in the AIG memo [of 11/07].
Goldman was not only heavily exposed to AIG (unlike other firms, it got CDS guarantees on CDOs only from AIG, while other firms used monolines as well) but it also had a far better idea of total AIG exposures than any other bank. Our look into transaction-level detail suggests that of the AIG ABS CDO transactions on which we found counterparty detail (nearly 80% of the deals identified in an 11/07 AIG memo; all save the strongest appear to have gone into Maiden Lane III), Goldman was either the dealer manager (it originated a CDO guaranteed by AIG, although in many cases another bank had purchased it and was therefore the counterparty) or the counterparty (via having bought an AIG-guaranteed CDO from another dealer) on over 50% of the deals.
Now consider what that means. When it was the bank that had created the CDOs guaranteed by AIG, Goldman was able to generate marks (it knew what the CDO contained); indeed, many of the counterparties would come to Goldman to get marks (although that does not necessarily mean they used them; these trades could be and were marked to model, so a counterparty had latitude to tweak prices). And on deals in which Goldman was the counterparty on deals created by other bank, AIG would ask it what its marks were. Goldman, being Goldman (and per the Bloomberg story, being consistently aggressive) probably provided its own marks. But either way, Goldman had point of view of the value of over 50% of the multi-sector CDO portfolio. And Goldman may well have known that; AIG in some investor presentation discussed that portfolio separately (the total amount as well as an overview of its ratings profile and other portfolio-wide statistics). So Goldman also probably knew or could make an informed stab at how representative the half of the portfolio it saw compared, in crude terms to the other half.
To put it more crudely, Goldman probably at some point had a much better sense of how quickly AIG was decaying than the other counterparties. And then we get into interesting questions. Did Goldman have any conversations with the officialdom prior to AIG starting its terminal decay? How was Goldman positioning itself? Did Paulson and other ex-Goldman officials handle the conflicts appropriately?
There is an thought-provoking albeit somewhat flawed, related story up at Huffington Post by David Fiderer, which argues that Goldman set out to kill AIG and collect a bounty. The big problem I have with it is that he posits that Goldman orchestrated quite a few events in the AIG unraveling to its advantage. Up to a point, that’s plausible; if you have read Roger Lowenstein’s How Genius Failed, on the LTCM bailout. you get an amazing display of how Goldman got its attorney in a key position in the negotiations, then had him act 100% in the interest of Goldman, not the entire rescue group. They were unabashed pigs and did not care at all about playing fairly.
And I’ve seen this first hand. Goldman is adept at understanding a fluid situation and figuring out how to play it to maximum advantage.
But masterful orchestrated multi-step game plans are another matter. That’s not a GS specialty, for an obvious reason: it’s a bad bet. Anything with a lot of moving parts ultimately has high odds of not working out. If you need 7 things to happen for you to get a payoff, and each of them has 90% odds, your odds of winning are less than 50%.
Thus his argument amounts to assuming unmitigated Goldman genius, when buddies of mine have seek big Goldman cock-ups first hand.
So this HuffPo piece, which I recommend, does a very useful job of digging and exposing some interesting (as in suspicious) connections, but pushes them further than I consider plausible.