Felix Salmon

Is Google becoming a key arm of the law-enforcement complex? It certainly seems to be so with respect to art thefts. I first came across this idea back in November, when Bloomberg Markets profiled Jeff Gundlach, who was hit by art thieves in September:

The cerebral Gundlach also gave investigators a tip for solving the crime. He says that while he was at home in his family room, it dawned on him that thieves would do a Google search using his grandmother’s name to find out more about the paintings and how much they might be worth.

Gundlach told the authorities that they should check the Internet to see who might have googled the name Helen Fuchs. He says exactly two such searches were executed: one by him and one by the thieves.

Now, another man has been arrested for art theft, and was found in much the same way:

In their investigation into the art theft, [officials] found that Mr. Istavrioglou had searched the Internet for reports about the robbery after it took place but before the story became news.

Law enforcement officials, it seems, have pretty easy and routine access to Google’s search-history database, and this is surely only the beginning when it comes to sifting through huge amounts of data to find evidence of crimes. The SEC, for one, has had a large data-mining team in place for a good five years now, going through enormous quantities of data to look for signs of suspicious activity.

Even journalists are getting in on the act of using data to uncover criminal activity. The Sun Sentinel, in Florida, managed to obtain a year’s worth of SunPass toll records for cop cars. That meant that they had data on the amount of time it took cops to drive from one toll plaza to the next. All they needed to do then was measure those distances, divide the distances by the time taken to drive that length of road, and come up with an average speed, for cops who were often just commuting to or from their houses, out of their jurisdiction. The result? The Sun Sentinel found “almost 800 cops from a dozen agencies driving 90 to 130 mph on our highways” — in a state where speeding cops have caused at least 320 crashes and 19 deaths since 2004.

Part of the reason why it has taken so long to bring Libor prosecutions is that going through millions of email and IM records, looking for smoking guns, is still a laborious and time-consuming process. But as data mining techniques continue to evolve, and as databases become increasingly unified and tractable, and our lives are lived almost entirely online, it’s going to be harder and harder for criminals not to leave a discoverable data trail — especially opportunistic criminals, who break the law when they’re given a chance, as opposed to more considered criminals, who spend a lot of time plotting a crime before committing it.

It stands to reason, given advances in computer power and given the size of the networks that we all involve ourselves in every day, that the kind of data crunching that used to be solely the domain of places like the NSA and GCHQ is now going to be available to local police forces and even ordinary citizens, including journalists. The privacy implications are profound, of course: millions of innocent people are going to have their personal data combed on a real-time basis, every day. But that seems to be inevitable, insofar as it isn’t already a reality.

Andrew Ross Sorkin and Peter Lattman have uncovered an interesting wrinkle in the SEC’s case against Mathew Martoma, the most promising part of its huge investigation into Stevie Cohen. The SEC made quite a big deal of the fact that Martoma didn’t just sell his position in two pharmaceutical companies ahead of a big negative announcement; he even kept on selling after that, building up a substantial short position.

But as Sorkin and Lattman have worked out, that’s not really the case: SAC was flat going into the announcement, rather than being short.

The NYT’s spin on this news is that it suggests “a possible line of defense for the portfolio manager”, but it’s not entirely obvious from the report what that possible line of defense is, so let me spell it out.

First, it’s worth stating quite clearly that profits are the same as avoided losses in the eyes of the law. The SEC says that Martoma made $75 million in profits and avoided $194 million in losses as a result of the trading, for a total of $269 million; in the light of the NYT’s new information, that should probably just be $269 million in avoided losses, and nothing in profits. The total amount of money is the same, so the severity of the charges is unchanged.

But here’s the thing: if your trading book is long ahead of a big announcement, you’re basically making a bet on that announcement. Similarly if you’re short. But if you’re flat, that’s the one way of not betting on the announcement. And it now seems that SAC was flat, rather than short.

Of course, if Martoma traded on inside information, then he’s guilty whatever the final position of SAC’s trading book was. But if that position was flat rather than short, it’s no longer circumstantial evidence that SAC thought the announcement was going to be negative.

And there’s another line of defense here, too. As the NYT says, “SAC is well known for its aggressive, rapid-fire trading style, and several former employees say that there is nothing unusual about the fund’s exiting a large position over just a few days.” And this is the defense that has now been opened up. SAC was sitting on substantial paper profits, on its position in Wyeth and Elan. It knew an announcement was coming, and it knew that announcement could move the stocks substantially. If it made the sensible determination that the downside was bigger than the upside, there was every reason for the fund to move to a flat position ahead of the announcement, whether it had any inside information or not.

If I were a defense lawyer here, I’d be coming up with hundreds of previous cases where SAC exited a large position in a short amount of time, ideally ahead of some big announcement. Some of those exits will have been smart, in hindsight, while others will have been silly: SAC would have been better off holding onto its position rather than going flat. But the decision to go flat and take profits (or cut losses) is a common one within SAC, and can happen at any time for any of a million reasons. And as a result, SAC’s trading activity is not in and of itself prima facie evidence of insider knowledge.

Frankly, this isn’t much of a defense. Trading activity is what the SEC uses to try to find possible abusers of inside information; it’s not what the SEC uses to try to prove such cases. In this case, the SEC is relying on the testimony of Sid Gilman, the doctor who leaked the trial results to Martoma before the official announcement.

But the news does help insulate Cohen, even if it doesn’t help Martoma very much. No one knows what Martoma told Cohen before Cohen made the decision to go flat, but SAC’s trading action is entirely consistent with a simple declaration that Martoma wasn’t comfortable being long any more. (The rest of SAC was already making a strong case against being long at this point.) If Cohen knew that an announcement was imminent, and that the one person who wanted to be long no longer wanted to be long, then it would have made sense for him to go flat ahead of the announcement, even if he had no inside information at all. And there’s no particular reason to believe that Martoma would have admitted to Cohen that he had illegal insider information.

As Sorkin and Lattman say, the statute of limitations on this trade is rapidly running out: if the SEC will have to either bring charges against Cohen soon, or not at all. And so long as Martoma is refusing to cooperate with the SEC, it increasingly seems as though the SEC’s best chance yet to nail Cohen is going to slip through its hands.

Call me naive, but after the Barclays revelations, I actually thought that I couldn’t be shocked about the extent of Libor manipulation. Boy, was I wrong. I could quote all 40 pages of the FSA notice fining UBS for Libor fraud: this is far, far worse than simply understating UBS’s borrowing costs so as to make investors think the bank was healthy. In fact, a lot of the fraud was designed to move Libor up rather than down: whatever the traders could make the most money manipulating.

The FSA concludes, quite explicitly:

UBS’s misconduct is, although similar in nature, considerably more serious than Barclays’ because it was more widespread within the firm, being exacerbated by the control failings, in particular the inherent conflict of interest in its submission function. More individuals, including Managers and Senior Managers, participated in or knew about the manipulation and there were more instances of individual manipulation, across more currencies. Furthermore, the extent to which UBS colluded with others was significantly greater and involved financial rewards being paid to Broker Firms.

The latter point is key: UBS didn’t just manipulate its own submissions, but actively attempted to manipulate other firms’ submissions as well. And at points the bribery was so explicit as to beggar belief that anybody would ever communicate such things on the record:

If you keep 6s [i.e. the six month JPY LIBOR rate] unchanged today … I will fucking do one humongous deal with you … Like a 50,000 buck deal, whatever … I need you to keep it as low as possible … if you do that …. I’ll pay you, you know, 50,000 dollars, 100,000 dollars… whatever you want.

A “50,000 buck deal” here does not mean a $50,000 deal: it means a $50 billion deal. If the broker on such a deal siphons off a fee of 0.0001%, that’s $50,000 right there.

The $1.5 billion that UBS is paying in fines here is enormous, but it’s not remotely enough: if the chairman and CEO of Barclays were forced to resign over much lesser Libor fraud, then we’re going to need to see heads roll at UBS too. And, with any luck, some individual criminal prosecutions of UBS executives, to boot.

That said, UBS has already taken the most drastic action it could: it has basically shut down its entire fixed-income business. That unit made enormous profits when things were going well — but it was staffed by rogue traders, who manipulated Libor rates around the world as a matter of course, and who on top of that contrived to lose mind-boggling amounts of money during the financial crisis.

Other fines, for other banks, are sure to follow this one — but if Barclays was dreadful and UBS was much worse than Barclays, it’s hard to imagine that anybody has clean hands here. You want to know why pretty much the entire financial sector is still trading at less than book value? This is why: the number of investors who trust the banks is now zero, and banking seems to have become a game of picking up fraudulent nickels in front of a relentless justice-department steamroller. (And for good measure there are all the civil suits as well: the $1.5 billion that UBS is paying today is just a down-payment on the all-in cost of its Libor fraud.)

The fixed-income department at UBS was the merged product of many storied firms: Swiss Bank, SG Warburg, Dillon Read, Paine Webber, Kidder Peabody, Phillips & Drew, and many others. And that’s the most depressing part of this whole story: there were good and honest managers at all those shops, and they all got pushed out by the fast-buck merchants. The inevitable conclusion: if you’re a senior fixed-income executive in the investment banking world, you’re necessarily suspect. Because this isn’t the kind of world where honest men live long.