Their trading didn’t take place in a market where very large numbers of transactions produced transparent and public prices through the day, like the stock market. Instead, the traders made bets with derivatives, financial contracts that often trade sporadically and in the shadows of Wall Street. Their trading didn’t take place in a market where very large numbers of transactions produced transparent and public prices through the day, like the stock market. Instead, the traders made bets with derivatives, financial contracts that often trade sporadically and in the shadows of Wall Street. The traders focused on so-called credit derivatives, including one named CDX.NA.IG9, which allow traders to bet on the creditworthiness of a basket of companies.
The federal complaints charge Martin-Artajo and Grout with mis-marking CDX.NA.IG9, and other indices, from March to May of 2012. From December 2011 through June 2012, that contract averaged about 60 trades a day for around $8 billion in notional.2So, I mean: it traded. Sure, sporadically or whatever, and in the shadows of Wall Street or wherever (but really on computers), but it traded.
The reason that the traders' marking process started with mids is that that's where your marking process starts; the complaints quote JPMorgan's accounting policy as saying that the "starting point for the valuation of a derivatives portfolio is mid-market." Because trades are just, like, trades, man; the midpoint of dealer quotes at the end of the day are probably a better indication of where you could exit a round lot at the end of the day than are a bunch of trades done for various reasons and in various sizes at various times throughout the day. But at the same time, having sixty trades a day means those quotes are probably meaningful: if the market is trading at around 100 to 105 and you're quoting 106/109 then you are in trouble.
Now it makes sense that the CDX IG indices would be reasonably liquid because they were designed to be liquid ways to hedge or express a view on credit generally. A particular series of bonds issued by a particular company will be illiquid; "give me a billion dollars of U.S. investment-grade credit" is more generic and, thus, more liquid. Not a ton of bonds trade eight billion dollars worth a day. I mean, Apple stock doesn't trade eight billion dollars a day, on average.
Similarly, while there is some guesswork involved in valuing the portfolio, it's important not to overstate how much. The famous Grout spreadsheet shows the CIO being 2-4bps away from the mids in the CDX.NA.IG.9, and that translating into $143mm of mismarking. Per the Senate report the CIO had about $62bn notional in IG-9 at that point, so their marks there were off market levels by 0.23% of notional.3
Zero point two three percent! The overall portfolio was some $157bn notional, according to that Senate report, and the overall mismarking was $767mm, according to the JPMorgan report that the DoJ more or less endorses, or about 0.49% of notional. The marks were off by nine digits only because they had twelve-digit positions.
Does any of this matter? I mean, I dunno, it's fun to imagine a jury considering:
- Prosecutor: "their marks were wrong by hundreds of millions of dollars!"
- Defense: "no, our marks were within one-half of one percent of the correct number! it's rounding error!"
And I guess both are mostly true?4 The better evidence against Martin-Artajo and Grout seems to me to be evidence of intent - constantly being on the aggressive side of the market, changing marks to get more favorable results, saying that in writing, Iksil's freak-outs, and that stupid spreadsheet - than, like, massive mis-marking. That evidence is really bad! The numbers, though, I'm not so sure. Half a percent, whatever.
I confess that my initial reaction to Peter Eavis's post was "wait 'til he hears how they value loans." Like, those never trade!5 You put them on your books at historic cost, and then you slap them with some somewhat fuzzy loan-loss reserve that you sort of mark to quarterly market conditions, but really nobody expects you to be within 0.5% of the current market value of each of your loans. Really that's why credit indices exist. The CDX.NA.IG.9 index, for all that it is an inscrutable series of letters and dots, is a thing to make banks' credit risk more transparent and liquid and tradeable. You don't know what your $10mm notional loan to some private company in Albuquerque is worth, but you add up all those loans - JPM has $725bn of 'em, of which $2bn are marked at fair value - and hedge them with $150 billion of credit indices and those, y'know, you can value pretty accurately. They trade! There are dealer marks!
It's sort of tiresome to go around saying "stop complaining about derivatives, traditional banking business is riskier," but: right?6 Compared to, like, banking, JPMorgan's CIO portfolio was a model of transparent valuation, even with the fraud. Unfortunately for Martin-Artajo and Grout: that transparency is what made it possible for the DoJ to figure out they were cooking the books.
1.A popular one is "why isn't Iksil being charged since he actually lost all the money!" Well but losing money isn't a crime. The mis-marking is. And he seems to have at least been sad about the mis-marking, and even taken some small steps to stop it.
2.Specifically 64 trades for $10.55bn a day from December 20, 2011 to March 19, 2012 and 56 trades for $6.3bn from March 20 to June 19 of 2012. Plus another yard a day of tranched trades. FT Alphaville has daily Markit data in graph form, though their numbers are a bit smaller; in any case they suggest that the volume was concentrated in January-June.
3.Obviously a much higher percentage of market value. The 10-year traded at around 110bps, according to the complaint, meaning 2-4bps is, y'know, 2-4% of market value.
4.I'm persuadable that the second is not and percentage of market value rather than notional is the right metric for that one. Or, like, if you compared the mis-marking to things like internal loss limits, or net income for that matter, etc., it looks even worse. But still, my go-to analogy is to bonds: the "value" of a credit instrument, loosely speaking, is its notional.
5.I mean, obviously many do, but you know what I mean. It's not like CDX indices.
6.Cf. the thing where losing $6bn on MTM credit derivatives is baaaaaaaaaad but losing $6bn on AFS interest-rate product is no big deal. Because AFS interest-rate product is (1) sort of almost "traditional banking" and (2) not eeeeeeeeevil "derivatives."