In one email, the person on the structuring desk overseeing the deal wrote, "This is [the CDO trading desk]'s prop trade (don't tell CSAC). CSAC agreed to terms even though they don't get to pick the assets" (SEC complaint against Brian Stoker, paragraph 32.) Half of the eventual CDO was based on securities chosen by Citi's trading desk (paragraph 42). (Yves Smith has more, from the SEC's order against CSAC--which, by the way, is part of Credit Suisse, another big bank.) Of course, the structuring desk didn't do this just as a favor to the trading desk: "On November 14, 2006, Stoker's immediate supervisor informed Stoker that Stoker should take action to ensure that the structuring desk received 'credit for [the trading desk's] profits' on Class V III" (complaint, paragraph 33).
As is common in these cases, the SEC and Citi negotiated a settlement in which the bank would pay $285 million ($190 million for its profits plus a $95 million penalty) but neither admit nor deny the allegations. Judge Rakoff (who previously gave the SEC a hard time over a settlement with Bank of America over the closing of the Merrill Lynch acquisition) refused to approve the settlement, saying that it offered no factual basis on which to even decide whether it was fair, adequate, reasonable, and in the public interest (pp. 13-14).
The absence of a factual basis is a decent legal argument, but I think what Rakoff is really taking aim at is the problem of regulatory capture. For Citi, the settlement boiled down to two things: a $95 million penalty and a promise not to break the law in the future. Anyone who can do basic arithmetic can see that it's worth it to break the law under those terms so long as you have a two-thirds chance of getting away with it (and who thinks the SEC is catching more than one-third of all violations?). And earlier in the case, Rakoff asked the SEC how often it has actually enforced the promise by a bank not to violate the law in the future. The answer: never, at least not in the past ten years (pp. 22-23).
In short, a settlement like this seems to have exactly zero value as a deterrent. The SEC argues that it could lose at trial, but that doesn't justify a settlement that is worth nothing to begin with. It argues that its resources are better spread across more cases, but many times zero is still zero. So the real question is why the SEC wants these settlements so badly. Rakoff's take: after calling the settlement, at most, "a mild and modest cost of doing business" for Citigroup, he continues, "It is harder to discern from the limited information before the Court what the S.E.C. is getting from this settlement other than a quick headline" (p. 11 in his order).
Regulatory capture can occur for many reasons: outright bribery and the hope of future job offers are possibilities, but so are ideological conformity and the desire for good relationships and a peaceful life. The result is the same, however: an agency that cannot or does not enforce the public interest against powerful private actors. In that case, the only hope the public has is private lawsuits--hence Rakoff's desire for facts on the record that can help those suits proceed and his impatience with the "neither admit nor deny" convention. The SEC seems to have taken this construction to the height of absurdity in this week's settlement with Wachovia, which neither admits nor denies rigging the market in municipal bond derivatives--even though it admitted criminal violations in a parallel settlement with the Justice Department.