No, the SEC Didn't Collude With Goldman Sachs on CDOs

Felix Salmon and Matt Yglesias are up in arms about how the Securities and Exchange Commission settled its collateralized debt obligation cases with the banks, and I do not get it. 
Poor Fab. Photographer: Peter Foley/Bloomberg

Felix Salmon and Matt Yglesias are up in arms about how the Securities and Exchange Commission settled its collateralized debt obligation cases with the banks, and I do not get it. The theory seems to be either that the SEC should not be allowed to settle cases, or that the SEC should engage in constant public critical self-reflection. Neither seems like a plausible model.

The story is that every bank did some shady stuff in designing and marketing synthetic CDOs, and that that shady CDO stuff may or may not have been fraud under the relevant securities laws. This was an untested proposition to say the least. Generally speaking, the SEC felt that the banks didn't do a great job of disclosing that someone was betting against the mortgages that they were selling to investors. Since a synthetic CDO is nothing but a bet -- no actual mortgages change hands, it's just one side betting on mortgages and another side betting against -- and since the banks actually did disclose that, this was kind of a weird theory, though it had an obvious populist appeal.

So the SEC did what any collection of sensible lawyers would do. It found the most winnable case it could find for each bank, 1 and threatened to sue (or, sued). The most winnable case was, of course, the one with the most embarrassing e-mails attached to it. You don't go to a jury saying, "It turns out that the selection agent for this CDO thought the sponsor was selling protection on the equity tranche, when it was in fact buying protection on a senior tranche," because that is mostly gibberish. You go to a jury saying, "Lookit this e-mail from a French dude talking about himself in the third person as 'the fabulous Fab...standing in the middle of all these complex, highly levered, exotic trades he created without necessarily understanding all the implications of those monstruosities!!!'" If there are three exclamation points, it's gotta be fraud.

The banks had their own sensible lawyers. They did not want to take these cases to trial, because of those e-mails, and because there was a significant risk that they'd (1) lose, (2) annoy their regulator and (3) leave themselves open to other lawsuits on their other CDOs.

The SEC did not particularly want to take these cases to trial either, because there was a significant risk that it would lose. (Remember, its legal theory was weird.) Then it would have nothing to show for its marquee financial-crisis cases.

So both sides wanted to settle. This is not a surprise; regulatory lawsuits against big banks rarely go to trial. This is partly because both sides have too much at risk, but it's also for another reason, which is that the purpose of all of this is regulation. The main goal is to improve future conduct, so a big cash settlement and an agreement to stop doing whatever bad thing you were doing tends to accomplish most of the regulator's goals. 2 So why spend resources on a trial you might lose?

So the banks offered the SEC a bunch of money to make the cases go away, and the SEC said yes. Totally normal. The amount of money was pretty much plucked out of thin air -- by, to be fair, Goldman Sachs 3 -- but you will never get a satisfactory explanation for the amount of money in any settlement. The losses on the CDOs were as much as several billion dollars per bank, but that doesn't mean that that's what the SEC would get if it went to trial. You have to discount that by the probability that the bank would win, the possibility that it would not be held liable for all of the damages, etc. It's a negotiated amount driven more by the sides' confidence in their cases and willingness to go to trial than by the amount of potential damages. The going rate seems to have been a few hundred million dollars per bank.

"Per bank" but not necessarily "per case." The banks had each done, like, a dozen CDOs, but the SEC focused on its most winnable case against each bank. The banks wanted a blanket settlement for all of their cases, so they could put their CDO worries behind them. The SEC did not want that, partly because it wanted to keep the banks on their toes, but mostly because it decided that it would look better with one big win per bank than a dozen small wins. A compromise was, again, reached: Everyone agreed to settle one case per bank, so that the SEC could claim a big win, but the SEC quietly wrapped up its investigations in the other cases, so the banks could mostly sleep at night. American Lawyer describes what then SEC enforcement director Robert Khuzami told then SEC inspector general David Kotz:

Khuzami was determined that Goldman's payment only be linked to ABACUS. "This was not a $550 million settlement for 11 cases," Khuzami told Kotz. "We may tell Goldman that we are concluding our investigations in these other matters without recommending charges, but that doesn't mean we're settling them. And that was an important point for us, because we didn't want them out there saying, you know, they settled 12 CDO investigations for an average of $30 million each, and, you know, didn't [Goldman] get a great deal."

This is what Salmon and Yglesias are upset about. Salmon:

As Khuzami says, if you look at them on a per-CDO basis, the big headline numbers suddenly become much more modest and affordable for Wall Street banks. So there's a real scandal here: firstly, the SEC was not being fully honest with the public about the deals it was cutting. Secondly, the SEC failed to stand up for CDO investors it should have fought for. Thirdly, the SEC tried to make it look as though it was levying massive fines for single deals, when really the settlements were omnibus deals covering some unknown quantity of CDOs.


What is now looking clearer and clearer is that the settlements were not as advertised. The banks paid money -- in Goldman Sachs' case $550 million -- not to settle one CDO suit, but to settle all the CDO suits. So rather than Goldman paying $550 million for wrongdoing around the Abacus CDO and then facing 10 more charges related to 10 other suspicious CDOs, it was paying a price of $55 million per CDO to settle all 11 cases. Except the SEC didn't want to look like it was letting the banks get away with a slap on the wrist, so it worked out an arrangement whereby both sides would publicly act as if only one case had been settled while agreeing under the table that all claims were now resolved.

So I don't get the concern? The actual amount of money that the banks paid is the actual amount of money that the banks paid. Goldman paid $550 million to the SEC. That's $550 million per CDO case that it settled, or $55 million per CDO that it didn't settle, or $183 million per exclamation point in that embarrassing Fab Tourre e-mail. How you count does not seem particularly relevant from a public policy or regulatory perspective. Goldman's CDO practices cost it money, however it's divided.

Of course if you think that the right number was $550 million per CDO, then Goldman wildly underpaid. On the other hand, if you think that the right number was $0 per CDO, then it overpaid. That's what a settlement is. The SEC decided that getting a big fine on its best case, and nothing on its weaker cases, was a better outcome than either getting a small fine on all of its cases, or rolling the dice on a whole lot of trials. Sensible lawyers make judgments like that all the time. 4

So what's wrong with it? One of Salmon's concerns is that investors in the showpiece CDOs got some of their money back from the SEC -- the SEC distributed $250 million of the $550 million it got from Goldman, for instance -- but investors in the non-showpiece CDOs got nothing. But this is America, and they can sue. If the banks defrauded them out of hundreds of millions of dollars, there's probably a lawyer somewhere who will take their case. Sensible lawyers like taking hundred-million-dollar cases, especially against unsympathetic banks. If they can't win those cases, then that suggests that the SEC got too much money out of the banks, not too little.

The bigger concern is that the settlement announcements were an attempt by the SEC to puff up its record and to make the public think that it is more impressive and effective than it actually is. This concern is more compelling, I guess, except that literally every sentence ever uttered by any regulatory or political body in the whole history of the earth is an attempt by that regulator to puff up its record and make the public think that it is more impressive and effective than it actually is. Go to the New York Times home page right now, and you can read an article about "a Rose Garden ceremony that featured several standing ovations for the embattled departing health secretary." Do you think the ceremony focused on how embattled she was?

The SEC's puffery was puffery, yes, but I can't see whom it hurt. The public was pretty adequately informed: It knew that the banks had a lot of CDO controversies, that they paid massive fines to settle some of them, that others just sort of went away, and that the ones that were settled tended to have really embarrassing e-mails attached to them. That still pretty much seems like the story? And the puffery had some beneficial regulatory consequences, too, in that it created a precedent of big price tags attached to individual cases -- a precedent that allows, or might even force, the SEC to push for bigger fines in the next case.

So I don't get why the SEC shouldn't be able to settle cases for the best deal it thinks it can get, or why its consideration of "the best deal" shouldn't include "the deal that makes the SEC look like a big impressive regulator." On the other hand this, from Yglesias, is quite right:

Among other things, to pull something like this off requires an exceptional level of trust between high-ranking SEC officials and high-ranking bank officials.

But everything requires an exceptional level of trust between high-ranking SEC officials and high-ranking bank officials! We talked yesterday about how JPMorgan mostly self-regulates: The job of making sure that JPMorgan follows the securities laws, or the money-laundering laws, or capital requirements, or whatever else, is done by JPMorgan. The tiny outmatched regulators mostly nudge JPMorgan around the margins; the actual work of compliance is done by the bank itself. Of course that requires trust.

So, sure: The SEC and the banks often work together toward common goals; often their interests are aligned. Sometimes that alignment is over getting regulation right, or agreeing on what banks can and cannot do. But sometimes they find common ground in the midst of disagreement: When the SEC thought the banks had committed fraud, and the banks thought they hadn't, they found that it was in both of their interests to resolve the cases in a way that made the SEC look good. That's how compromises are made.

This column does not necessarily reflect the opinion of Bloomberg View's editorial board or Bloomberg LP, its owners and investors.
  1. Except Deutsche Bank! Where Robert Khuzami worked. We'll meet Khuzami in a minute and you'll see why that's weird.

  2. From Goldman's settlement announcement:

    The landmark settlement also requires remedial action by Goldman in its review and approval of offerings of certain mortgage securities. This includes the role and responsibilities of internal legal counsel, compliance personnel, and outside counsel in the review of written marketing materials for such offerings. The settlement also requires additional education and training of Goldman employees in this area of the firm's business. In the settlement, Goldman acknowledged that it is presently conducting a comprehensive, firm-wide review of its business standards, which the SEC has taken into account in connection with the settlement of this matter.
  3. (My former employer.) Salmon:

    It's quite impressive how quickly and accurately Goldman nailed the amount of money that it would have to pay the SEC to settle the case: when it took three months to come to the $550 million settlement, I for one assumed that Goldman had to be dragged kicking and screaming to that point. In fact, however, Goldman was happy to offer half a billion dollars right off the bat.
  4. Also, sensible lawyers drop cases all the time without formal settlements. You say "an arrangement whereby both sides would publicly act as if only one case had been settled while agreeing under the table that all claims were now resolved," I say "a settlement on one case and quiet regulatory death for the rest." But quiet regulatory death is the fate of almost all regulatory actions, many of them meritorious. The bulk of regulation is regulation by inaction. The SEC even has a whole huge body of precedent in the form of "no-action letters," where it tells you what it won't sue you for.

(Matt Levine writes about Wall Street and the financial world for Bloomberg View.)

To contact the author on this story:
Matthew S Levine at

To contact the editor on this story:
Zara Kessler at

Before it's here, it's on the Bloomberg Terminal.