The SEC's case against Goldman sounds really, really bad; "fraud," after all, is just such a strong word. But it's hard to appreciate much more about the case unless you happen to speak the obscure language of derivative structured credit transactions. As you'll see, stripped of the lingo, the case is easy to understand. Its important lesson is not so much that GS can be unpleasant as a counterparty (what a shock); instead, the crucial point is how much of the "financial engineering" of the past few years has rewarded the bankers for doing nothing but introducing risk into the system.
The first thing you need to know is how Credit Default Swaps, CDSs, work. Think: "an insurance policy's evil twin." A CDS can indeed provide protection against the risk that an issuer will default on a bond: if I own a GE bond and am concerned it might not pay, I can buy a CDS against it to cover myself.
What's the problem? That I can buy a CDS that pays off in the event of a GE default even if I don't own a GE bond. To buy a real insurance policy, I have to have an "insurable interest" in the object being insured. That's why I can't buy a policy against your house burning down. A good rule, as otherwise I might be tempted to buy a policy and torch your place. This lack of an "insurable interest" requirement creates some nasty incentives in the real world; just ask Lehman Brothers, or Greece.
Next, let's look at CDOs, Collateralized Debt Obligations. These come in two varieties, "real" and "synthetic". As you might guess, the former are relatively wholesome and serve a genuine economic function; the latter are dangerous and do not.
Real CDOs are bonds backed holds pools of debt obligations. A so-called SPV, a special purpose vehicle, raises money by issuing bonds, uses that money to buy big pools of mortgages, and then uses the mortgage payments to pay off the bonds. These instruments have been central to the housing market for years: banks sell off the mortgages they originate, get fresh capital to lend, and do it again.
Now, we get to the structure at the heart of the Goldman case, synthetic CDOs. These are strange and dangerous beasts, so stay with me. Here, the SPV doesn't buy mortgages or bonds or anything real at all. Instead, synthetic CDOs just provide a way to wager on how someone else's (real) CDOs will perform.
It's like betting on a sporting event, with the SPV as the bookie. An entity known as a "selection agent" picks the pile of CDOs on which wagers will be taken, known as the "reference" securities. Investors who think this group of CDOs will pay off as scheduled (again, to whoever really owns them, not the investor in the synthetic... he's just a spectator) go "long" by buying bonds from the SPV. Folks who want to take the "short" side of the bet buy a CDS from the SPV. The SPV, our bookie, holds the proceeds from the long and short sides, and pays out to the winner.
The beauty of this structure from Wall Street's standpoint is that the deal does not involve the purchase of anything; it's the ultimate "derivative". A single mortgage-backed bond can be "referenced" -- bet upon-- over and over again. So, all a banker has to do is find long and short players to match up. It's a great way to generate fees, but creates no economic value. Worse, in fact: these structures generated a devastating multiplier effect when the real mortgage-backed bonds "referenced" in so many deals began to default.
The Goldman case itself? It's all about that selection agent. Normally, that's an independent expert with a stake in the portfolio's success. But here, the SEC claims that the portfolio was actually selected by the fearsome Paulson & Co. hedge fund, which took the short side of the deal. In fact, says the SEC, the whole scheme was Paulson's idea: the fund had assembled a list of the biggest CDO stinkers it could find, and then approached Goldman with the idea of creating a synthetic CDO just so they could profit when the junk went south.
Even Goldman salesmen would struggle to place the long side of the deal with that history. So, the SEC says, Goldman misled a well-respected selection agent, ACA, into reluctantly approving Paulson's born-to-fail portfolio by claiming Paulson was going long in the deal. The SEC alleges fraud in Goldman's failure to disclose these "facts" to the few very sophisticated institutions that invested, which lost a billion bucks or so to Paulson on the trade.
Goldman denies all this. It says it never told ACA that Paulson was going long, and that ACA did ultimately pick the portfolio. Moreover, it says everyone involved was experienced and capable of making independent judgments about the portfolio (not a bad point).
What should we take from all this? Personally, I just don't care about purported conflicts of interest among players at this level. To me, what the case really illustrates is how often Wall Street "innovation," like synthetic CDOs, are just new casino games that generate big profits for the house, but only huge risks for our economy.