By Yves Smith and Tom Adams, an attorney and former monoline executive
Gretchen Morgenson has a lengthy article tonight at the New York Times, “Testy Conflict With Goldman Helped Push A.I.G. to Edge.” While it provides some useful new tidbits, it peculiarly focuses on an aspect of Goldman’s dealings with AIG that, particularly with the benefit of hindsight, the bank can defend, and gives short shrift to other issues that implicate not only Goldman, but much bigger fish, namely the Fed and the Treasury.
Let us stress that this post is NOT a defense of Goldman. We are no fans of Goldman; the firm does a masterful job of cutting the cake so it gets the biggest piece, and deserves to have its actions scrutinized. But as other commentator have mentioned, the fixation on Goldman often lets other culpable players get off easy (for instance, while roughing up Blankfein at the FCIC hearings was a welcome spectacle, no one laid a glove on Jamie Dimon). The story also at quite a few junctures gives an account that appears unduly in line with AIG’s point of view, which also serves to weaken the impact of some of the legitimate issues it raises.
Let us start with the central thesis which is:
1. Goldman was overly aggressive in marking down the CDOs it had insured with AIG. Remember, the bigger the losses reported on the CDOs, the more cash AIG would have to pony up to Goldman
2. Goldman’s actions contributed to AIG’s demise.
We’ll deal with each of these points, and then discuss the important issues (and we think bigger) issues we think this focus misses.
Shortcomings of the Morgenson Thesis
The story starts, dramatically, with a heated conference call in January 2008 in which Goldman and AIG staff are arguing over the “marks” or prices Goldman is assigning to its CDOs insured with AIG:
A.I.G. executives wanted some of its money back, insisting that Goldman — like a homeowner overestimating the damages in a storm to get a bigger insurance payment — had inflated the potential losses. Goldman countered that it was owed even more, while also resisting consulting with third parties to help estimate a value for the securities.
This remains the subtext of the article: Goldman was overly aggressive, other firms were marking the deals at higher prices.
There is a wee problem with this account. Goldman’s marks were proven correct. With the benefit of hindsight, most players, particularly AIG, were in denial. Moreover, AIG’s assertion that Goldman was not willing to get independent valuations is disputed later in the article. But if you didn’t get that far, or weren’t paying close attention, you would conclude Goldman was using “marks” that would not stand up to scrutiny.
In late 2007 and early 2008, the monolines were facing similar issues to AIG. Rather than getting cash calls, they were facing aggressive marks from their bank counterparties. The monolines argued vehemently with their accountants and their investors (and some of their employees) that the marks overstated the declines in their insured CDOs due to market illiquidity and that the bonds would recover in value. This is where Bill Ackman’s open source document, his detailed analysis of MBIA and Ambac came in – he put forth the analysis in January 2008 that showed that they mark to market write downs would translate into real losses. The rating agencies not long afterwards started downgrading AAA asset backed securities CDOs, verifying the “aggressive” position Ackman and Goldman were taking.
The story also repeats the AIG/Fed flattering claim that these CDOs have “rebounded.” We’ve discussed long form in other posts that given the continued, serious deterioration in the underlying mortgages, this notion is simply not credible. The decay in credit quality across the portfolio is severe, and there has been no “rebound” in prices of severely distressed CDOs.
One monoline insurer, faced with a material discrepancy between what the “market” was saying the CDO bonds were worth and what “management” was saying, had two outside firms to evaluate its portfolio. Both valuations differed in approach, but both confirmed material losses from the CDOs. This was in November 2007, before the AIG/Goldman dispute.
The jig was up for AIG by January of 2008 and the debate was only one of timing, not of what the actual outcome would be. Coincidentally, Ambac, FGIC and XLCA were downgraded in January 2008 directly as a result of high expected losses in their CDO portfolios. Any case against Goldman for aggessive marks against AIG by the SEC or other parties would have take the market environment into consideration. Across the board, CDOs were causing losses and downgrades for the people who insured them. It therefore makes plenty of sense that Goldman would be requesting more collateral for their exposure with AIG.
The more logical question then becomes: why weren’t AIG’s other counterparties, such as UBS and Merrill, requesting more in collateral when their CDOs were clearly losing value? Perhaps these other institutions had other reasons – and other exposures – which prevented them from marking down their CDO exposures and demanding more collateral. For instance, if they downgraded these CDOs to reflect downgrades and deteriorating market pricing, were there other CDOs, and perhaps even mortgage backed bonds they’d have a hard time NOT downgrading? Or was it that, unlike Goldman, they had hedges with monolines too. With the monlines being downgraded, they’d be having to mark down the value of the CDS with them. So even if the other banks were fully insured on their other CDOs, but they were insured with monlines, they probably would not be able to report hedge gains that corresponded to CDO losses due to the monoline downgrades (ie, the insurance was proving not to be as effective as it had been assumed to be at the time it was booked).
Many people, with vested interests, were arguing losses would not be that large. They couldn’t really imagine that the losses would be that large – it was beyond previous imagination, but they turned out to be wrong. Much of 2008 was a battle between these forces in the financial industry – bear, lehman, aig, the monolines and even, for a while, the rating agencies – arguing that the losses would not be as terrible as market pricing was suggesting.
Remember the dead body in the room: the portfolio of bonds that AIG insured went from AAA to CCC in less than a year. Goldman looks to have been the only AIG ounterparty on top of the crappy fundamentals.
In a very extreme interpretation of events, someone might be able to argue that the aggressive marks caused the value of the bonds to fall further and pushed the downward spiral. But that betrays a misunderstanding of the CDO deals.
The underlying MBS in the CDOs were already downgraded and declining rapidly in value by 4Q 2007. The credit market had already frozen. The die had already been cast. It was inevitable that this would lead to the CDOs collapse, regardless of their original ratings.
Now let’s get to point 2, that Goldman’s “overly aggressive” collateral calls helped push AIG over the edge. Narrowly of course, this is correct, all the collateral calls together (well, plus the $20 billion black hole of the securities lending portfolio, can’t sweep than under the rug, now can we?) pushed AIG over edge. Goldman was one of the counterparties, the biggest in aggregate, so this seems to stick, right?
Well, you need to dig deeper to prove this assertion, and the article fails to do that. First, per a November 2008 BlackRock memo, Goldman and AIG had reached a settlement of sorts re Goldman’s marks. AIG was applying a 12% haircut.
Now you would then have to look at:
The cash Goldman was getting AFTER the haircut, versus an average across the other banks (adjusted for credit quality, which BlackRock was capable of doing). In other words, you need to see the EXCESS of what Goldman was getting relative to what it would have gotten if it had marked its bonds like everyone else, and see if that differential was big enough to have made a difference in AIG’s demise. We don’t have that analysis, and ex that analysis, this may or may not be the case.
Possible Productive Lines of Inquiry That Get Short Shrift
The focus on Goldman’s marks with AIG largely bypasses what we think is a more serious issue: the role of all synthetic or heavily synthetic CDOs, which allowed Goldman to go net short. The usual vehicle for that was a “mezz” CDO, because the CDS would be on BBB subprime trances, the layer that would go “boom” first. The bulk of Goldman’s AIG-related CDOs were older vintage “high grade” CDOs, meaning the synthetic component was not large (on the deals we looked at, a maximum of 20%) and they would be on AA bonds, which were not the slice you’d be eager to use if your strategy was to go net short. So the fixation on the marks has the unfortunate effect of diverting attention away from what we think was the much more troubling activity: the use of heavily/all synthetic CDOs to establish a short position.
Even though the deal documents allowed for the possibility that Goldman would keep the short interest created by these deals, anyone who invested in them or acted as a guarantor would have thought very differently, and probably have asked for much higher returns if it had understood Goldman was acting as a principal rather than as a middleman (and how Goldman influenced the deal parameters to assure that its short position worked out). The story indicates that $5.5 billion of Abacus trades (a Goldman synthetic short program of 26 deals in total) were insured by AIG. Using the AIG Abacus trades as an entry point into the entire Abacus program would be a very useful exercise.
Then we have this bit:
Mr. [Ram] Sundaram [of Goldman] used financing from other banks like Société Générale and Calyon to purchase less risky mortgage securities from competitors like Merrill Lynch and then insure the assets with A.I.G. — helping fatten the mortgage pipeline that would prove so harmful to Wall Street, investors and taxpayers.
This may mean a lot, or it may mean very little. By our tally. $5.9 billion of Goldman’s CDOs that were insured with AIG had been structured by Merrill. Also note that it is pretty likely that this insurance was entered into at or within days of the closing on the CDOs.
AAA CDOs were eligible repo collateral. According to the IMF, the haircuts on AAA ABS CDOs (the type we are discussing here) were a mere 2-4% as late as April 2007. That means that when Goldman bought these CDOs, it could have financed as much as 98% of the purchase price in the repo market. If SocGen and Calyon were providing repo financing, this is not newsworthy. If there was a different, non-standard arrangement, that could be a very different story.
Also unexamined is the fact that most of the Goldman’s deals appear not to have been on the firm’s behalf. Goldman’s non Abacus trades would have been considered to be “cash” CDOs. Consider this section of the November BlackRock memo:
Goldman’s exposure to AIG is limited to the differential between collateral requested (what they are likely posting to swap counterparties) and collateral requested from AIG….Goldman’s swap counterparties are exposed to Goldman Sachs risk rather than AIG counterparty risk. and are therefore less likely to be receptive to deep concessions.
Goldman has said it does not hold the cash CDOs, but has back to back swaps on most of the positions.
Now why would Goldman not be holding the CDOs and have entered into back to back swaps? Goldman probably sold the CDOs to customers, and also provided CDS on them too. This was a pretty common arrangement. The banks preferred that the insurer did not “face” an end customer directly; the originating bank would take the CDS position with the monoline or AIG and then write swaps against it (this enabled the bank to earn more in fees).
This idea is also confirmed by Maiden Lane III disclosure:
AIGFP, the LLC and the New York Fed have entered into agreements with AIGFP’s credit derivative counterparties to terminate approximately $53.5 billion notional amount of credit derivatives and purchase the related multi-sector CDOs. Of these, CDOs with a principal amount of approximately $46.1 billion settled on November 25, 2008. Settlement on the remaining $7.4 billion is contingent upon the ability of the related counterparty to obtain the related multi-sector CDOs and thereby settle with the LLC and terminate the related credit derivative contracts with AIGFP.
Notice the mention of a single counterparty that had to settle later because it had not rounded up all its CDOs. That counterparty is believed to be Goldman. That does not preclude Goldman (and other banks that had sold CDOs and were acting as swap counterparties) from having better luck in getting their customers to do whatever they needed to do to settle with Maiden Lane III.
But that raises yet another issue that has yet to be examined: to the extent that the banks had sold CDOs to customers, and were merely acting as CDS middlemen, the purchase of the CDOs themselves by Maiden Lane III constituted a bailout of customers. And it is a near certainty that these were not, say, US pension funds, parties that the Fed might have a policy interest in assisting. Goldman’s CDO customer list is closely guarded, but it is believed to have a heavy representation of Middle Eastern investors. If true, why did the Fed extend a backdoor bailout to them?
Diverting Attention from the Fed, Treasury, and Paulson
Doubts about the monolines were becoming serious in the first acute phase of the credit crisis, September-October 2007. They were a daily focus in the business press in January-February 2008. Bear went under and was rescued because its failure would have rocked the CDS market.
There might have been a defensible case for denial of the seriousness of the problems afflicting the CDS market up through March of 2008. The Bear meltdown, the fact that the ratings agencies were starting systematic, aggressive downgrades of CDOs, and the belief that Lehman and Merrill were next to go meant understanding the risk of mortgage-related CDS exposures was imperative to understanding systemic risks. Even a cursory examination would have led straight to AIG. The “oh were weren’t their regulator” is an implausible excuse. The Fed and the SEC most assuredly WERE regulators of the parties exposed to AIG.
And this inattention raises further issues. Virtually all of the Goldman CDO exposures with AIG were entered into when Paulson was CEO of Goldman. Why has there been no inquiry into his role in overseeing, and ultimately profiting, from these deals?
AIG’s failure to understand the implications of the current market values and downgrades underlying its CDOs sheds light on its CDO underwriting process (or more accurately, lack thereof). How widespread was this problem at other companies? Did anyone that invested in (or insured) these deals really understand them? It suggests that only reason any of these parties were involved in these deals was because they were rated AAA. Yet to date, the SEC, Treasury and Fed have done nothing to address the essentially misleading nature of AAA ratings on such bonds.
Although details are emerging bit by bit on CDOs and credit default swaps, there is still a great deal that is not out in the open. The failure of regulators to push for much larger-scale inquiries suggests at best a troubling complacency, or at worst, the knowledge that a full bore investigation will reflect very poorly on the powers that be.