Does Goldman's "it's perfectly normal" defense hold water?

In a reading of Goldman's September letter of defense against the SEC investigation into the synthetic CDO it created at Paulson's behest, one question stands out: is it really true, as Goldman's claims, that investors in a synthetic CDO would accept as a matter of course that short as well as long participants might have not only an active but a formative role in shaping the portfolio? Paulson initiated the CDO, suggested the initial portfolio to the selection agent ACA, and ultimately approved every bond in the portfolio, more than half of which were in its initial selection. Is that business as usual? Goldman says yes:

There is no industry definition of "Portfolio Selection Agent" that implied that ACA would operate within an ivory tower or refuse to consider suggestions made by interested parties in exercising its independent judgment. In fact, it was a customary feature of the market that participants (including those here) often offered their views on potential securities to be included in referenced portfolios, so no one would have been surprised that Paulson was doing so.

Depending on industry norms, this could either be a particularly brazen bit of sophistry or an effective defense. 


Whether Goldman is legally liable for defrauding investors in the Paulson-initiated CDO, Abacus 2007-ACI, by all accounts remains very much in doubt. But is Goldman ethically liable -- that is, in practical terms, do Goldman's clients have fresh cause to believe that Goldman will willingly trick them to serve its own or another client's interest?

One fixed income analyst I contacted suggests that a role like Paulson's would indeed be at variance with investors' expectations.  According to this analyst, those selling "protection" on the portfolio (i.e., agreeing to pay if the underlying assets stop performing) would assume either that the protection buyers were simply hedging what the buyer perceived to be very small risks of default, or "exploiting a capital structure arbitrage opportunity in a basis trade." That is, the assumption would be that the short sellers were either genuinely buying "protection" -- essentially, insurance -- to reduce their long exposure -- or trying to turn a modest profit because they thought the "protection" they bought was underpriced. This analyst writes:

In other words, the buyers of ABACUS tranches didn't imagine that it was a Paulson, with a profoundly bearish view of mortgage defaults, on the other end, but, perhaps, a big insurance company or mortgage lender with (overall) neutral or even bullish view of mortgage defaults that nevertheless wanted to tinker with its risk exposure at the margins, or a quantitative trading firm that was trying to clip off 10 or 20 basis points of mispricing of the CDO vs. the basket of RMBS that it referenced.

The easy analogy is to insurance.  When you approach an insurance company to ask for coverage, the insurance company doesn't believe that you think a loss is likely.  Indeed, the insurance company believes that you are motivated to avoid a loss, but still want to protect against the contingency.  CDS writers (i.e., synthetic CDO buyers), thought of themselves as selling insruance.  They went wrong, of course, because CDS is not insurance and people were and are free to buy CDS fully expecting a loss to incur and to make far more money on that loss than on the underlying property (indeed, with naked CDS, to make infinitely more since they didn't own a presenatable claim at all) -- and, morever, can perfectly legally even cause that loss to occur.

Sanford Bragg, CEO and President of Integrity Research Associates (which analyzes analysts, helping investors to select the research best suited to their needs), suggests that Goldman's "ivory tower" claim above is  weakly worded.   "It appears that Paulson was doing more than 'offering views,' Bragg notes, adding, "if  'no one would have been surprised' [by Paulson's role] why wasn’t it disclosed?  Nearly half of the offering presentation is devoted to the selection agent (ACA), but there isn’t one whisper about Paulson."  Goldman does address this question in its defense, claiming that "it has never been industry practice for financial institutions to disclose the identities of clients with which they enter into hedging transactions" (p. 36).  The question remains, however, whether Paulson's unusual role -- acting as a kind of de facto portfolio selection agent -- mandated disclosure.  Bragg suspects that "GS is vulnerable in its lack of disclosure of Paulson’s role because it seems that the transaction was essentially commissioned by Paulson."

I have been too slow with this post, collecting info since Wednesday. Today the WSJ's estimable Serena Ng addresses this very question -- whether Goldman met or violated disclosure norms and obligations.  She finds no clear answer:

There wasn't a consistent approach to how these types of CDOs were assembled, or what was disclosed about them, according to lawyers and the Journal's review of documents.

"We're in untested waters here," says Joel Telpner, a partner at Jones Day in New York, who was commenting generally about disclosures in CDO offering documents. Unlike offerings of publicly traded stocks and bonds, "in the private-placement world, the rules about what has to be disclosed and where have never been as clear."

Ng examines similar but not-quite-analogous deals, synthetic CDOs assembled by Deutsche Bank, and CDOs in which the hedge fund Magnetar Capital sought to short risky tranches. In both cases, short sellers involved themselves in the asset selection process. But in neither case did the short sellers initiate the investment vehicles -- nor does it seem that they shaped and dominated the selection process to the extent that Paulson did. As characterized in the Journal article, these deals appear to have been shaped under the conditions claimed as norms by Goldman in its letter to the SEC: " In fact, it was a customary feature of the market that participants (including those here) often offered their views on potential securities to be included in referenced portfolios, so no one would have been surprised that Paulson was doing so."

It would appear that Abacus 2007-ACI was unique in the extent to which the short seller with interests diametrically opposed to the long investors shaped the offering.  The deal's genesis and raison d'etre does seem to be one that would violate even sophisticated investors' expectations and thus would seem to confer on Goldman a unique disclosure duty -- ethically if not legally -- in keeping with its unique nature.

Steve Randy Waldman articulates norms that he asserts Goldman violated in terms congruent to those of the analyst quoted above, but on a more theoretical plane. Waldman makes a complex distinction between the norms involved in trading securities and those of swaps -- norms which cross in a synthetic CDO.  While both necessarily involve exchanges between parties taking long and short positions, in securities sales the interests of buyers and sellers can be aligned, while in swaps they're directly competitive.  The investors in 2007-ASI would have assumed that they were purchasing a security -- notwithstanding that a synthetic security is created by parties engaging in swaps, agreeing on one side to provide an income stream as long as the underlying securities continue to pay, and on the other to make large payouts when the securities were downgraded or failed.  Waldman suggests that, foolish as the long investors may have been to rely on multiple intermediaries to produce an income stream based on the peformance of the underlying securities, they would not have expected and had no cause to expect that they were "competing" with a short seller betting on the failure of the whole package (who had in fact shaped its construction to with a view to facilitating its failure). Here is his bottom line:

Synthetic CDOs were composed of CDS positions backed by many unrelated counterparties, not one speculative seller. Goldman’s claim that “market makers do not disclose the identities of a buyer to a seller” is laughable and disingenuous. A CDO, synthetic or otherwise, is a newly formed investment company. Typically there is no identifiable “seller”. The investment company takes positions with an intermediary, which then hedges its exposure in transactions with a variety of counterparties. The fact that there was a “seller” in this case, and his role in “sponsoring” the deal, are precisely what ought to have been disclosed. Investors would have been surprised by the information, and shocked to learn that this speculative short had helped determine the composition of the structure’s assets. That information would not only have been material, it would have been fatal to the deal, because the CDO’s investors did not view themselves as speculators.

Waldman further suggests, in a response to readers' comments, that the dynamics of a transaction built around a custom-made or "bespoke" investment product are fundamentally different from those of standardized products, which multiple buyers trade on an open market:

With a bespoke product, there is generally an initiating party (Paulson for our CDO, perhaps an industrial firm looking to hedge an idiosyncratic risk). The counterparty to a bespoke contract (e.g. an investment bank) usually knows something about the initiator and can divine something about its motivation. The counterparty to a bespoke product should generally not be anonymous. If the counterparty remains anonymous, it should at least be identified that there is an identifiable initiating counterparty. Otherwise, there would be terrible scope for tailoring products based on precise information asymmetries that the non-initiating counterparty wouldn’t suspect.

These diverse voices more or less concur: a security created at the behest and largely to the specs of a short seller and marketed to investors without disclosure of that fact was a malign mutation in a market run amok. As the FT's John Gapper suggests today, its creation marks Goldman's transition from an advisory firm that by creed put its clients first to a self-serving counterparty whose message to clients is “We’re taking care of our balance sheet and you’re on your own.” Goldman would claim that it plays both roles under different circumstances and "manages" any conflict between them. But when the bank simultaneously trades on its own account and offers advisory services, one role corrupts the other. Gapper's verdict:  Goldman "treated IKB like a trade counterparty while Paulson got the privileges associated with a client."

If clients' takeaway matches Gapper's, the case's public airing will hurt Goldman more than any jury verdict or settlement.


* The disclosure question may in fact be moot regarding the largest investor, portfolio manager ACA. (ABN Amro, which lost $900 million on the deal, guaranteed ACA's exposure and so bet on ACA's solvency, not on the performance of the bond portfolio.) CNN reported on Wednesday that  Paolo Pellegrini,, the former Paulson employee who put the deal together, testified that he told ACA that Paulson intended to short the portfolio -- specifically, "That we wanted to buy protection on traunches of a synthetic RMBS portfolio.”  Perhaps this is not the whole story -- Goldman's defense points out that "holders of equity may also hold other long or short positions that offset or exceed their equity exposure." The SEC, which took Pellegrini's testimony, asserts in its complaint that ACA believed that Paulson had a long position "and according, that Paulson's interests in the collateral section [sic] process were aligned with ACA's" (p. 2).  If ACA did have full knowledge of Paulson's intent to short the whole portfolio, that leaves only the German bank IKB as a potential unwitting victim. Regarding IKB, Goldman attempts a swift, vague equivalence, asserting that the bank, along with Paulson and Goldman itself, "offered views on the securities proposed for the Reference Portfolio" (p. 27). In any case, Goldman did attempt to sell its own remaining stake to other (presumably unwitting) investors.

UPDATE: at Propublica, Jesse Eisenger and Jake Bernstein have produced an incredible article about Magnetar, a hedge fund created to short the CDO market, which operated much as Paulson did in the Abacus deal and had a strong hand in creating some 30 CDOs. Eisenger and Bernstein have identified 26 of these deals, 96%  of which were in default by the end of 2008.  Magnetar's tack was to undertake to buy the riskiest, or "equity," tranche of each CDO, and then (by best inference) massively short the entire portfolio via credit default swaps, ultimately earning far more on the swaps than it lost on the equity tranche. Even more elegant: the money it earned before the mortgage meltdown from the high risk/high yield equity tranche investments funded its short investments until the default flood started.  More to our point, by taking the hard-to-sell equity tranche,  Magnetar enabled each deal and thus won major influence over its construction, which it used  in some documented cases and by inference in many others to negotiate aggressively with the portfolio selection agent. Magnetar claims not to have initiated deals or proposed the initial portfolio as Paulson did (per the WSJ article cited above), and the degree to which counterparties knew what it was up to is unclear.  But according to Eisenger and Bernstein it enabled dozens of deals just  as the CDO market was seeming to wane (starting in spring 2005) and thus gave the market fresh impetus. And it generally pressed to make the portfolios as risk-laden as it could, ostensibly to get more yield in the equity tranche but probably to make its shorts pay off.  In some of the deals described by Eisenger and Bernstein, it appears to have had as much of a shaping hand as Paulson in Abacus 2007-ACI.