Goldman is trying to diffuse the increasingly harsh light being turned on its dubious practices in the collateralized debt obligation market, with the wattage turned up considerably last week by a story in the New York Times that described how a synthetic CDO program called Abacus was the means by which Goldman famously went “net short” subprime. We’ve mentioned Abacus repeatedly because AIG wrote guarantees on at least some of the Abacus trades.
One of the things that has been frustrating in watching this debate is the peculiar propensity of quite a few observers to defend Goldman and its brethren, and to argue, effectively, caveat emptor. Contrary to the fantasies of libertarians, that is not in fact how markets, particularly securities markets, operate. In virtually every market in the world, when someone represents his wares as being sound and safe and they turn out to be “bad” and dangerous, the seller is considered to have some responsibility for the damage. Remember those Pintos that turned into fireballs when rear-ended? The pets that died from pet food laced with melamine from China? No one suggested that the buyers of those products were at fault.
I’m deliberately using extreme examples, but as we proceed, you will see they are not unreasonable. Those products did not fail by design (well, the faults of the Pinto were known and left uncorrected). By contrast, Goldman wanted its Abacus trades to fail. That was the most profitable course of action for them (note the Times clearly states that that was the role of the Abacus trades and Goldman has not disputed that claim). They were net short, this was no mere hedge of a long position but a trading bet. And those CDOs did turn out to be big turkeys.
If you sell a stock when have inside, non public information that the company was in serious trouble without disclosing this information – you are likely to be held accountable. If you sell a manufactured product and it causes harm, you might be held liable regardless of whether you knew it was faulty (strict liability). In fact, if sellers were not held responsible for product quality to some degree, commerce would become impractical. In a pure “buyer beware”, dog-eat-dog world, every transaction would require extensive and prohibitively costly due diligence.
Now let us turn to the CDO market (and by “CDO” we mean “ABS CDO” or “asset backed security CDO; there are other types), where buyers (contrary to uninformed views otherwise) DID routinely perform extensive due diligence, and nevertheless, lemming-like, went to slaughter. It wasn’t just the investors or the now-derided guarantors; many banks wound up with lots of CDOs on their balance sheets (for not very good reasons, but that is beyond the scope of this post).
But then we have the famed short sellers who used synthetic CDOs as their means for getting short. With a synthetic CDO, no one puts up cash. It is a little corporate entity. The asset side is various credit default swaps, in this case, “referencing” mainly subprime bonds, the BBB tranches. So the cash flow comes from the premium payments on these CDS. The liability side is tranched, so you have an equity tranche, BBB, A, AA, and a junior AAA layers, then say two more AAA layers and the infamous super senior tranche (there could be more tranches than this, but you get the general idea). But the investors do not make cash payments; they are effectively protection sellers, or insurers. But it is important to note that is not how the deals were hawked; they were presented as a convenient, faster-to-launch alternative to the old-fashioned cash CDO. (This may seem like a distinction without a difference, but as you will see soon, it matters).
Now let us turn to Goldman. The New York Times last week published a story that discussed how a Goldman synthetic CDO program (series of deals) called Abacus. Goldman was not merely the structurer of these deals; unbeknownst to its investors, it was usually the ONLY protection buyer in these deals; it occasionally let some favored hedge funds participate in a minor way on the short side. In other words, the sole purpose of these deals was for Goldman to put on a short position.
Goldman issued a disingenuous defense on its website. It isn’t long, but let’s look at the key sections:
Synthetic CDOs were an established product for corporate credit risk as early as 2002. With the introduction of credit default swaps referencing mortgage products in 2004-2005, it is not surprising that market participants would consider synthetic CDOs in the context of mortgages. Although precise tallies of synthetic CDO issuance are not readily available, many observers would agree the market size was in the hundreds of billions of dollars.
Many of the synthetic CDOs arranged were the result of demand from investing clients seeking long exposure.
Synthetic CDOs were popular with many investors prior to the financial crisis because they gave investors the ability to work with banks to design tailored securities which met their particular criteria, whether it be ratings, leverage or other aspects of the transaction.
Yves here. This is all background, but notice that “synthetic CDO” includes synthetic collateralized loan obligations. In fact, if you had said “synthetic CDO” prior to the recent focus on synthetic ABS CDOs, most Wall Street types would have assumed you meant synthetic CLOs. It was a longer established, bigger, and less complex product. CLOs are tranched products made from leveraged corporate loans.
As an aside, this bit is simply precious: “… it is not surprising that market participants would consider synthetic CDOs in the context of mortgages.” Ahem. Goldman pushed the ISDA protocol that made synthetic CDOs possible, and along with Deutsche, launched the first deals. And who were these “market participants” who were so keen to get synthetic CDOs done? Not the long investors, but none other than John Paulson, the famed subprime short. To pretend that the growth of subprime ABS CDOs was driven by the long side is a misrepresentation. It was the shorts that pushed the development of this product.
As we stressed in a recent post, the bulk of the value of a deal was in the AAA tranches, which for ABS CDOs represented 70% to 90% of the total value. They did not drive the structure of the deal; it was well cooked by then and they had influence over certain parameters. As someone who did package CDOs told us:
They [the AAA investors] usually couldn’t see the big picture (e.g. they were getting long massive amounts of risk), yet they were keen to nitpick over documentation. On managed transactions they usually didn’t have the right to DK individual collateral items, but they could (and often did) negotiate deal parameters (e.g. max AA CDO bucket, max synthetic bucket, reporting, controlling class provisions, etc.) I would say that they were brought into the deal around mid-way, once the manager and majority of the equity was locked down. The banker/structurer had to be the arbitrator between equity, monoline, the rating agencies, and their own P&L from the structuring fee & warehouse carry.
In other words, the AAA investors were not equal parties in negotiating, despite their economic weight; many parameters were largely set before they were brought in. The main drivers in the structuring were the equity investors, the monoline, the rating agencies, and the bank’s own economics.
Goldman, of course, protests that everyone is a grown-up, and perfectly capable of taking care of themselves:
The buyers of synthetic mortgage CDOs were large, sophisticated investors. These investors had significant in-house research staff to analyze portfolios and structures and to suggest modifications. They did not rely upon the issuing banks in making their investment decisions.
Yves here. I find this “sophisticated investor” bit very amusing. Just because an investor might be “sophisticated” by securities law standards does not mean he is highly skilled as far as ABS CDOs are concerned.
People who were in that business have consistently said that very few people are competent in the product. One estimated 1000, another even fewer. This is in the world, mind you. No matter how you cut and slice it, this is far fewer than were dealing in the product. This is no surprise to me; I did work in the early days of OTC derivatives, and it was widely known then that the number of people who understood the products was a small fraction of the population selling and buying them. So we’ve had nearly 20 years of it being completely acceptable, even normal, for “sophisticated” investors to dabble in instruments that they were at least a bit, maybe a lot, beyond their ken.
As an aside, that was not for dint of trying to become more knowledgeable. In CDO-land, one of the dirty tricks the dealers played was model arbitrage. The customers were given models for assessing the deals that were less sophisticated than the ones used by the packagers/underwriters.
So let’s go to the next bit of Goldman self-defense:
For static synthetic CDOs, reference portfolios were fully disclosed. Therefore, potential buyers could simply decide not to participate if they did not like some or all the securities referenced in a particular portfolio.
Yves here. Hard to know for certain, but my impression is this is largely irrelevant. Many, if not most, deals were managed, meaning they were effectively blind pools. In those cases, the parameters would be specified, and the equity investor would have a veto on particular assets. And the AAA investors could nix certain categories (say no option ARMs or CMBS) but not particular bonds.
And even if the portfolios were specified, how realistic was it to assume they could be evaluated? The CDOs were composed mainly of trances of subprime RMBS, the BBB tranche in “mezz” CDOs. A typical subprime bond issue contained 5000 mortgages. I honestly do not know how many bonds would be in a CDO, but it was obviously more than 100. 100 RMBS tranches means 500,000 mortgages.
In the early days of the crisis, bottom fishers tried looking at CDOs and for the most part gave up on systematic evaluation. It would take too long and cost too much money to assess the value due to the complexity. Those factors were no less operative when the CDO was being launched. In other words, one could argue that the long investors, particularly the AAA investors, who were earning not-sexy spreads, would find it uneconomic to do old-fashioned, granular credit analysis. That isn’t to say they didn’t stress test them; in fact the monolines modeled them assuming six to eight grade downgrades, which was considered catastrophic for AAA paper. It never occurred to anyone that AAA paper could go to CCC. Why? That had simply NEVER happened before, in the absence of massive fraud…..until it happened on a very large scale, with CDOs.
But what were the economics to the shorts? Aha, if you are buying at 140 basis points, and you in the end close out your position at 10,000 (which happened in some cases) you have vastly more upside than the long side. That means you can afford to do much more due diligence, or if you are Goldman or someone packaging a synthetic CDO to suit your needs, you can afford to do the work to devise something that suits you well but will not be obviously toxic to those on the other side of the trade.
This is the part the defenders of Goldman, Deutsche, and others are missing. Heretofore, the packagers had been…packagers. They had product in their warehouses from mortgage originators that they were putting into deals. Just as in traditional underwritings, they were assumed to be balancing the needs of the issuer against those of the buyer. It is more than a bit disingenous for Goldman to protest the everyone knew it could be on the other side of the trade:
It is fully disclosed and well known to investors that banks that arranged synthetic CDOs took the initial short position and that these positions could either have been applied as hedges against other risk positions or covered via trades with other investors.
Yves here. Please. If these deals were regulated by the SEC, Goldman’s role in the deal would have had to be disclosed and it would have made a CONSIDERABLE difference in the pricing. To pretend otherwise is an insult to a reader’s intelligence.
And a lawsuit filed against Morgan Stanley suggests that synthetic CDO buyers, when an investment bank was using the deal for its own purposes, are advancing credible theories as to how they were had From Business Week (hat tip reader Barbara):
Morgan Stanley was accused in a lawsuit of defrauding investors in a collateralized-debt obligation, called the Libertas CDO, by collaborating with ratings companies to place triple-A ratings on the notes.
Morgan Stanley, the sixth-biggest U.S. lender by assets, arranged the offering as it was short-selling almost the entire $1.2 billion worth of assets in the CDO, according to a complaint filed today in federal court in New York.
“Morgan Stanley was betting the entire investment it was promoting would fail,” the public employees’ retirement system of the Virgin Islands said in the complaint. “The firm achieved its objective.”
Yves here. The fact is that pretty much every investment bank in the second half of 2007 was trying to offload its subprime exposures into CDOs. But look at the party suing, the employees’ retirement system of the Virgin Islands. Do you think they have lots of fancy models and skilled experts to assess these deals? I’d bet not. And I’d also bet that at least some of the Abacus investors were no more sophisticated than this Morgan Stanley CDO investor. Back to the article:
“By collaborating with major credit-rating agencies to place Triple-A ratings on the rated notes, Morgan Stanley intentionally or recklessly misled investors in the Libertas CDO,” according to the complaint. “But for Morgan Stanley’s violations of law, the rated notes never would have been issued.”
CDOs typically repackage bonds and other assets into new securities. The Libertas CDO didn’t buy the mortgage-backed securities, according to the complaint. Instead, it entered into credit-default swaps that referenced mortgage-backed securities. Credit-default swaps are financial instruments that function as insurance for bondholders. As the credit-protection buyer, Morgan Stanley was shorting the securities, or betting they would fall in value.
“Morgan Stanley was highly motivated to defraud investors,” the retirement system said.
The bank also had an unfair advantage because it had access to information on the assets the investors didn’t have, according to the complaint. The securities were riskier than the bank let on and in fact were impaired at the time Libertas was created, the retirement system said.
Yves here. I am not a lawyer, but the idea of pursuing the investment bank for fraudulent AAA ratings could be a promising line of attack. Now this is only one suit, but if this theory prevails, this could do considerable damage to industry balance sheets. And right on the heels of record looting, um, 2009 bonuses.
An editorial in the New York Times today indicates that the pursuit of the CDO dead bodies could prove costly to other powerful parties:
To be thorough, investigations of these and other questions would have to reach into the Obama Treasury Department. One of the most aggressive creators of the questionable investments was a firm called Tricadia, whose parent firm was overseen by Lewis Sachs, now a senior adviser to Treasury Secretary Timothy Geithner.
We might finally get some answers, as well as some good theater.