Commodity Trader Didn't Really Believe in Market Prices
I enjoyed this Commodity Futures Trading Commission action from last week against a trader named John Aaron Brooks. Brooks traded commodities for an unnamed bank* until October 2011, when he "exceeded internal trade limits on cattle futures" and was told to unwind all his positions and go sit in the corner and think about what he'd done. He unwound almost everything, but politely declined to unwind his ethanol futures positions, because he happened to have been fraudulently mismarking those for about a year and "liquidation would expose the losses that Brooks was offsetting and masking." You can imagine how that conversation went.**
The bank fired him and discovered that it was out of pocket to the tune of $42.4 million because of the hidden losses. So it turned him over to the CFTC, which is suing to get the money back and ban him from the industry and do all the good stuff you do to rogue traders.
The weird thing about this case is that Brooks was mismarking Nymex ethanol futures, which trade on an exchange (the Nymex) and have a reported settlement price every day. I mean, they don't trade all that much, but there's definitely publicly reported market data, and from what I can tell a bid-ask spread of at most a couple of pennies. The CFTC complaint helpfully includes an appendix showing, for each trading day:
- The Nymex settlement price for the relevant contracts,
- Brooks's mark in the bank's books, and
- the percentage difference between the two.
The difference starts at 0.89 percent (two cents) and goes up from there, exceeding 10 percent (25 cents or so) every day from March 9 onward and reaching a hilarious peak of 105.16 percent -- Brooks's mark was $4.77, versus a settlement price of $2.325 -- on Sept. 30, 2011, a few weeks before Brooks was caught.***
So: That's not supposed to happen! My colleague William Cohan recently wrote that JPMorgan's "biggest mistake" in the London Whale situation was that the bank relied on traders to mark their own books of over-the-counter derivatives, rather than having an independent valuations group rigorously test valuations against third-party data. You need that rigorous valuation control because valuing structured over-the-counter trades is hard, and leaves some room for judgment. And the Whale traders used that wiggle room in bad faith, mismarking their opaque unlisted positions by something like 2 percent to 4 percent.****
This guy mismarked his transparent listed positions by at least 10 percent for seven months in a row. And you didn't need any fancy third-party valuation service to check it. The exchange's settlement prices are right there on its website. If the bank had had anyone spend a minute a day -- a minute a month, really -- checking Brooks's marks against public data, it would have avoided a $42.4 million loss.*****
A big regulatory push these days is moving over-the-counter derivatives onto exchanges, in part because that will help with pricing transparency. The London Whale episode was among other things good evidence for that argument: These sneaky opaque derivatives allow banks to make up prices; forcing things to exchanges will make banks' books more transparent and more accurate. Well, maybe! This case is a good reminder that, y'know, banks need to actually use that data for all the transparency to do any good.
* Update: It was Citi! Of course.
** Ooh but you don't have to because here's the CFTC complaint:
When Brooks was asked questions under oath about the mismarking of the NYMEX ethanol futures, Brooks repeatedly "decline[d] to answer based on [his] Fifth Amendment and constitutional rights in Ohio versus Reiner."
Probably sensible in the circumstances, though quoting case law sounds a bit like you're trying too hard.
*** The CFTC seems to be overstating the case when it says "In some instances, Brooks's valuations deviated from fair market value, as represented by broker quotations provided to Brooks and/or the daily Nymex ethanol futures settlement prices, by more than two times." I think that's only in the one case. Others are pretty close, though, like off by 80 percent to 96 percent.
**** My rough math on that is in this post.
***** Or a computer! They make computers that can compare numbers against other numbers now! Amazing stuff.