What do you think Dick Fuld got Lehman to celebrate their five-year anniversary? Photographer: Brendan Smialowski/Bloomberg.
What do you think Dick Fuld got Lehman to celebrate their five-year anniversary? Photographer: Brendan Smialowski/Bloomberg.

As the world nears Peak Lehman Commemoration* it's worth remembering that Lehman Brothers is very much around and, while it is not exactly underwriting any bond deals or advising on any mergers, it sure is litigating fiercely (and expensively!). That's its business these days: fighting over everything to recover as much money for Lehman's unsecured creditors as it can. This fighting can be distasteful -- Lehman is, for instance, suing a bunch of sympathetic nonprofits over swap terminations -- but the $2 billion Lehman's spent on professional fees buys, among other things, a lack of taste.

Perhaps the most interesting of these suits are those against JPMorgan and Citigroup over their derivatives claims, which I'll now proceed to oversimplify drastically. Basically, what happened was this:

  • Every bank had a bunch of derivatives trades on with Lehman.
  • So Bank X, for instance, would have 100 trades where it owed Lehman money, and 125 trades where Lehman owed it money.
  • Also, Bank X would have collateral from Lehman, or vice versa, depending on who owed whom money when they last passed collateral (typically, the Friday before Lehman filed).
  • When Lehman filed for bankruptcy, Bank X would tot up how much it owed Lehman,how much Lehman owed it, and how much collateral there was. Then it would net them, and get some total amount of money.
  • If Bank X owed Lehman $100 million, it paid Lehman $100 million. If Lehman owed Bank X $100 million, Bank X got a bankruptcy claim for $100 million, which is likely to pay out $18 million to $22 million in oh around 2016.

Did I mention that Bank X got to decide how much was owed on each trade? I mean: Lehman could argue about it if it thought that Bank X was wrong. But remember that, in late September 2008, "Bank X" was some people on the Bank X trading desk whose jobs (1) still existed and (2) depended for their continuing existence on losing as little money as possible in turbulent markets. In late September 2008, "Lehman" was an office full of tumbleweed and a bunch of people frantically working the phones for new jobs and/or day-drinking at home. Sharp negotiation of derivatives receivables is not the first priority at a newly bankrupt company.

It is pretty much the first priority of a several-years-bankrupt company, and so Lehman fought fiercely with the Bank X's and other derivatives counterparties of the world, with the result that the Bank X's entered into a big settlement in May 2011 where they all agreed on a uniform approach to valuing all the derivatives claims that the banks had.

But JPMorgan and Citi did not enter into that agreement, so Lehman has spent the last year or so suing them for doing various sorts of naughty things in valuing their derivatives. The suits are pretty fun!** Lehman thinks, for instance, that "JPMorgan artificially inflated the value of its claims arising under derivatives agreements with the Lehman Subsidiaries by more than $2.3 billion," and it is suing to get that $2.3 billion back.***

If you want to believe that JPMorgan and Citi overreached in their claims against Lehman -- and, on first principles, why wouldn't you want to believe that? -- then there are lots of possible ways for them to do so. I mean, for one thing: It was September 2008, lots of things were not trading or were trading at very disjointed levels, you had a certain amount of leeway to just make up numbers and hope for the best.****

But even for things where there was an observable market there's plenty of fuzziness. For instance, JPMorgan and Citi had a certain amount of leeway in deciding when to value those contracts, and in the turbulent markets of September 2008 that gave them a lot of opportunity for mischief. From Lehman's complaint against JPMorgan:*****

Internal email correspondence at JPMCB makes clear that it re-marked its CDS portfolio on September 16 and 17 specifically to increase the value of its claims. On September 15, Jeremy Barnum, a Managing Director at JPMCB with responsibility in the North American Credit Trading business, reported, “I closed out North American Credit Trading as of 12:55 PM EST.” The next day, Barnum reported closing out the same portfolio a second time, writing, “Considering the extremely high level of intraday volatility today and the ongoing news in Credit Markets, I wanted to memorialize that we have closed out the vast majority of the NACT Lehman related risk as of approximately 11 AM this morning.” Finally, on September 17 Barnum disclosed that he was re-marking the portfolio yet again, and explained he was doing so specifically to capture increased profit for JPMCB. He wrote, “Pnl with lehman trades would be up 50 instead of down 50. As a result I am going to change the claim to tonight’s levels.”

Hahaha that's pretty good. As someone who used to work on a derivatives desk I have trouble mustering any outrage at this: Of course if your counterparty goes and files for bankruptcy on you, you should treat it as an opportunity to get out at the best possible price, with a lookback if at all possible. I mean, the bankruptcy is a pain! You've been put to serious inconvenience, even leaving aside the whole possible-end-of-the-world-financial-system situation arising from this particular bankruptcy. You might as well get paid for it.

But what a derivative trader thinks he deserves is maybe not what his lawyer wants him to put in writing, and this business of changing your valuation date over and over again until you get the very best one for JPMorgan (and the very worst one for Lehman) looks pretty bad.

Maybe the main issue in the suits is how the banks should have netted their Lehman trades. If you have one derivative trade with Lehman, and you get to decide how to terminate it, you will probably close it out at your side of the market: If you are long a thing, you will in effect sell it back to Lehman at the offer; if you are short the thing you'll buy it from Lehman at the bid. That's pretty straightforwardly what's allowed by the contracts,****** and it matches intuitively what would happen in the real world. Outside of bankruptcy, if Lehman just came to you and said "hey we want to get out of this trade," you'd charge them (half) the bid/offer spread to get out.

But what if you have 70,000 trades on with Lehman, as JPMorgan did? Some are long, some are short, and a lot of them partially offset each other. One day you buy one CDX index, the next day you sell a different CDX index. Two separate trades, two separate contracts, but for economic and risk management purposes very close to offsetting.

Presumably if Lehman had just called JPMorgan up one day in, I dunno, May 2008, and said "hey we'd like to get out of every single one of our derivatives trades with you," JPMorgan wouldn't have said "okay, we'll charge you the bid/offer spread individually on each trade," and if they had Lehman wouldn't have taken the deal. Instead they'd tot up all the economic exposures and risks, come to one net mid-market number, and then tack on some overall spread for their trouble. Two essentially offsetting trades don't cost twice as much to unwind as one such trade. They should really cost less than the one trade.

Lehman's claim is that JPMorgan and Citi in effect charged 70,000 separate bid/offer spreads, rather than netting the trades in a commercially reasonable way:

Instead of applying portfolio aggregation to determine the economic equivalent of the material terms of the transactions as a group – as any reasonable trader or risk manager would do, as JPMorgan customarily did, and as commercial reasonableness requires – JPMorgan closed out the vast majority of the trades in isolation so it could maximize the number of charges for hypothetical losses included in its claims. Even where there were perfectly identical offsetting trades, JPMorgan in many instances added hypothetical add-on charges to both sides of the trades. This procedure was not commercially reasonable, and it greatly inflated the amount of hypothetical losses claimed by JPMorgan.

JPMorgan and Citi disagree:*******

Contrary to the Plaintiffs’ contentions, JPMorgan did net out matching trades of opposite direction before computing its claims to the extent practicable. The Plaintiffs assert, however, that JPMorgan should have netted out allegedly similar (but non-matching) risks so that JPMorgan would receive no claim for absorbing those risks. Nothing in the governing agreements or market practice supports any such claim. Plaintiffs’ further assertion that JPMorgan should have netted certain additional trades entered into by different trading desks even when it was impractical and inefficient – and potentially costly to the Plaintiffs – to do so likewise is without substance.

Who's right? I don't know, tell us in the comments etc., though the answer is probably somewhere in between the two positions. The ISDA agreements governing the trades talk about commercially reasonable terminations but don't specifically require netting; the traders and lawyers I've polled on market practice more or less think that the rule is "there's no obligation to net down as much as possible, but you shouldn't be a terrible pig either." Charging 70,000 separate bid-ask spreads is probably unreasonable, but netting as much as Lehman wants probably isn't required either. JPMorgan and Citi should get some sort of windfall for their trouble.

Which in itself is sort of odd! People fret a lot about the interconnectedness of the big banks and investment banks, with an emphasis on how the collapse of one bank can bring down many others. But the banks are in nontrivial part networks of zero-sum derivatives contracts, where one party's loss is the other party's gain, and where one party's bankruptcy is (sometimes) the other party's opportunity for, like, appropriately constrained piggishness. Five years ago next week, in what was otherwise a dark time for the world's financial system, some traders at JPMorgan and Citi probably were gleefully re-marking their positions to get more money from Lehman. Despite the lingering lawsuit I suspect their observance of the Lehman anniversary will be happier than most people's.

* If you enjoy (1) reading Twitter and (2) having heart attacks, I highly recommend following @TBTFLive, a Twitter account providing financial-crisis news on a five-year tape delay, for the next couple of days. It's now the weekend before Lehman filed for bankruptcy and things are about to get real.

** Here are some documents:

*** And, yes, to get all of it back -- not just to reduce a claim in bankruptcy. Lehman makes much of the fact that JPMorgan -- and to a lesser extent Citi -- were in a different position from the average Bank X. The typical counterparty would have been passing collateral with Lehman on a more or less mark-to-market basis: If Bank X's trades were $100 million in-the-money on Thursday night, then Lehman would give Bank X $100 million of collateral on Friday. If on Monday, when Lehman filed, the trades were now $150 million in-the-money to Bank X, then Bank X could keep its $100 million of collateral and submit a $50 million unsecured claim.

JPMorgan, though, was Lehman's clearing bank, and demanded billions of dollars of cash collateral to perform its clearing functions. This cash was unrelated to the mark-to-market on its derivatives but could be used to satisfy those derivatives claims. (Lehman tendentiously says: "All the while, [JPM's] eyes were trained on an $8.6 billion slush fund of excess cash collateral that it extracted from Lehman the week prior to bankruptcy, which it planned to use to ensure a 100 percent recovery of its contrived claims.") Citi was also a clearing bank and so overcollateralized, though its eyes were somewhat less trained and it ended up giving back some of the collateral for reasons that I'm going to go ahead and say were dumb.

So: Is Lehman right that JPMorgan, in its position as one of the only over-collateralized Lehman counterparties, had a bigger incentive than Bank X to overstate its derivatives claims? Oh sure, just first principles. If Lehman owes you $10 million and you can make a plausible but dicey case for $20 million, and then you have to go to bankruptcy court, submit a claim, wait forever, and eventually get 18 cents on the dollar, the benefits of being aggressive are far away and uncertain. If you don't have to go to bankruptcy court, don't have to wait for your money, and can take 100 cents on the dollar, then you might as well be aggressive. Maybe Lehman will sue you and eventually get the money back, but that risk is far away and uncertain. You might as well take the money now.

**** Also you can just tack on various spreads, liquidity surcharges, etc. and see what you can get away with. Lehman has a theory (see paragraphs 68-70 of its JPMorgan complaint) that JPMorgan shouldn't be able to charge a bid/offer spread in closing out these trades, which does not seem to be true as a matter of, like, what ISDA derivatives contracts say. On the other hand they also have a theory (paragraph 70) that JPMorgan charged inflated bid/offer spreads versus spreads that were actually observed in the interdealer market, which is perhaps a more sympathetic claim.

***** FWIW JPMorgan's answer denies these allegations.

****** Again Lehman sort of disagrees (see note ****!) but you don't get the sense that they mean it.

******* Oh I know seven asterisks, sorry. Anyway here is Citi:

In fact, Citi appropriately netted offsetting positions, thereby reducing its net derivatives claim by hundreds of millions of dollars. To the extent Citi overlooked any such offsetting positions in calculating the value of its more than 30,000 trades, Citi will correct the error. But plaintiffs are not really concerned with “offsetting” trades; instead, they argue that Citi should have netted entirely dissimilar trades on the theory that their prices moved in a correlated fashion. For example, plaintiffs argue that Citi should have netted credit default swaps referencing two different high-yield indices, on the theory that these indices track 72 common reference entities and the prices of the different indices tended to move in a correlated fashion. But 28 reference entities – or more than a quarter of the names – in each index are different and thus netting the two indices is not commercially reasonable because it would have left Citi with substantial uncompensated risk. Even more ludicrous, however, is plaintiffs' argument that Citi should have netted all of its credit default swaps referencing RMBS, on the theory that all RMBS are “based on pools of mortgages that had similar economic characteristics” – i.e., subprime or Alt-A residential mortgages – and, according to plaintiffs, “the overall price movement between [Citi‟s] buys and sells” of protection on RMBS were “highly correlated.” This argument is specious. Credit default swaps protect against the default of a particular reference entity. The fact that prices for credit default protection on different reference entities might move in a correlated fashion has no bearing whatsoever on whether a party would be left in a neutral risk position if it netted a sale of protection on one reference entity against a buy of protection on another.