Lehman Brothers is not dead. It still roves around, like a ghost of crisis past, to visit former counterparties, remind them of past sins, and demand resolution to derivatives disputes. To date, Lehman has reached closure with all but a handful of the roughly 6,600 derivatives counterparties it once had. Most of the biggest banks – Goldman Sachs and Deutsche Bank among them – signed onto a broad settlement in 2011.
Of the few remaining, the biggest is Citi, which turned its nose up at the industry settlement and accepted the challenge of a $2 billion lawsuit from Lehman. As the trial proceeds – and each party summons its cadres of experts and juicy crisis-era quotes from trading desks – one thing is clear: Should the trial end in a final judgment, it could fundamentally alter the way banks deal with derivatives and defaults. And probably for the dumber.
Nothing like the Lehman bankruptcy happened before 2008 or since. The way Lehman’s massive derivatives portfolio gets resolved, then, will become the precedent for future disputes. And since basically every other bank chose to settle, Citi’s approach could become the approach, at least in the eyes of the courts.
And there’s plenty of daylight between how Citi and Lehman want to derive its derivatives claims. “The parties seem pretty far apart in their perceptions of the value of the claim,” Peter Niculescu, a partner at Capital Market Risk Advisors who has worked on previous Lehman settlements, told me. “There did not seem to be from the public docs any likely avenue to reconcile their perceptions.”
In terms of dollars, the two parties differ by a factor of seven: Citi wants $2 billion; Lehman thinks $266 million is more like it.
When Lehman folded, it had $39 trillion in notional outstanding derivatives bets with others. Citi was on the other side of $1.2 trillion of them. The legal process following bankruptcy of a derivatives counterparty allows the non-defaulting party to come up with a dollar amount representing how much it would hypothetically cost to “close out” those trades and replace the risk they carried.
This exercise leaves plenty of room for interpretation. These derivatives portfolios involved bets going both ways. Lehman and Citi could (and did) have both long and short positions on the same name, at the same maturity, held between them. The portfolio might contain many individual trades, but once you net it all out you’ll have something smaller and simpler to replace. How to account for this fact is the root of the dispute.
Consider the following, extremely simplified book:
There are a few different options Citi had, after Lehman met its maker, to close out these trades and hand Lehman a bill. A brief review:
The What-We-Actually-Did Approach
On September 15, every Lehman counterparty suddenly found itself with a new derivatives risk profile – greater exposure to certain geographies and corporates, reduced exposure to others. So banks sprang into action. Citi’s traders did for the most part what any level-headed trader would: Determine net exposures from Lehman-facing trades on every reference entity – IBM, Venezuela, etc – and then trade as needed to return to where they’d been.
In the simplified example above, that would mean netting the $2o million in long positions on Argentina with the $15 million short (we’re assuming maturities and everything else is equal). Citi, then, would need to find a $5 million long exposure to Argentina to return to its pre-Lehman risk profile. Emails and call transcripts from 2008 show this is essentially what Citi did.
At the end of all that, Citi could tally whatever it cost to do those trades, tack on a few fees for their time, then hand that bill to Lehman. Easy-peasy.
This is not what Citi did. Nor were they required to by law. Moving on.
The Treat-Things-Like-We-Do-For-The-Accountants Approach
Here’s an idea. Every month, Citi traders had to provide a number to accountants that reflects what their derivatives exposures would cost to replace in toto. At Citi, this process was called a market valuation adjustment, or MVA. What if Citi just used its MVA formula to come up with a number for its Lehman exposures? Boy, that would be swell! After all, the way Citi already priced its derivatives, according to internal policy presented in court, sounded a lot like how one would price a closeout according to derivatives contracts:
Mark to market prices should reflect the most recent, reliable price at which the position could be liquidated or, in the case of contractuals, offset...For contractual positions (i.e. derivative instruments), which are typically marked at mid-level curves, this means taking the aggregate net exposure of a given portfolio and adjusting to the bid or offer (depending on if the net exposure is long or short) through a market value adjustment (MVA). In estimating the net portfolio exposure, consideration should be made to the appropriate exposure bucketing, given the portfolio composition and distribution of risk across the tenor of the portfolio.
This policy also happens to resemble the basic approach of Lehman’s settlement framework with all the other big banks (indeed, Lehman usually brought up its counterparties’ internal valuation processes at the negotiating table). According to Lehman’s reckoning, using Citi’s MVA metrics would yield a final closeout tab of $266 million.
Citi felt otherwise. In pretrial motions, Citi fought to have its MVA practices kept out of the discussion, arguing that accounting policies had “no bearing on Citi's contractual rights” (true) and “are not used to price trades in the marketplace” (also true). But the judge decided that neither of those arguments changed the fact that the MVA policy might be useful in determining what is commercially reasonable for closeouts.
As it turns out, some of the Citi traders tasked with the Lehman fallout did indeed take an MVA-like approach to closeouts, at least initially. In an email dated September 18, 2008, the head of Citi’s emerging markets desk Marc Pagano wrote: “We looked at net risk to a credit and then used the bid or offer for the net position.”
But by March 2009 that methodology would change to what Citi eventually landed on…
The Get-All-The-Money Approach
From a derivatives trader’s standpoint, this is the most logical path to take. Why give Lehman the benefit of the doubt when it’s a few last breaths away from the big sleep?
In practical terms, this approach means netting as little as possible. To return to the oversimplified example, it would mean finding replacement costs for all three Argentina positions, rather than one netted position. And Citi has some justification here. Lehman’s derivatives were a motley bunch with lots of dissimilar features. The question is how generous Citi should have been in bucketing together similar-though-not-identical trades.
In Lehman’s telling, Citi could have been a bit fairer. After some desks initially grouped similar trades, Citi went with a trade-by-trade closeout, matching each position to its bid or offer on the market (generally the worst possible spot on the spread for Lehman), then tacking on liquidity charges for odd-sized trades.
Pagano, the guy who originally included a netted methodology in his closeout process, later testified to feeling “silly and ashamed” he hadn’t gone trade-by-trade from the start. Others testified that the Lehman bust was the first and only time they ever did a trade-by-trade valuation on their derivatives books.
Citi’s aversion to netting explains most of the difference between Citi's and Lehman's closeout estimates. Here’s how the Emerging Markets desk closeouts differed between an MVA-like approach and trade-by-trade, according to Lehman’s analysis:
If it holds up in court, it’s understandable why Citi would go with this trade-by-trade method. The ISDA contract that specifies how to close out trades when one party goes under doesn’t mandate netting of similar positions, just that they should be done in a “commercially reasonable” manner. Who’s to say what’s commercially reasonable?
Lehman’s industry agreement, which required participating banks to net along strict and agreed-upon guidelines, could be an indication. “It’s not the law, but it does contain the imprimatur of commercial reasonableness given how many have agreed to it,” Niculescu said.
But a judgment with Citi would create a different precedent. “The question is how to resolve a settlement dispute involving a variety of derivatives with the non-defaulting counterparty [Citi] would never have intended to replace, and whether they should be valued at a line-item replacement cost or whether true economic replacement costs mean something different,” Niculescu said.
If trade-by-trade is ruled the correct approach, then it raises an interesting question. Citi gives a number to its accountants – the MVA – that’s supposed to account for what it would cost to liquidate its derivatives portfolios. This is necessarily a hypothetical – the sorts of situations that would demand replacing trillions of dollars in derivatives are by definition exceptional. But sometimes, occasionally, they get a chance to see their risk management formulae in action.
Lehman was such an opportunity. But now Citi is telling the court that the MVA has nothing to do with an real-live, in-the-flesh derivatives portfolio liquidation.
If Citi is going to take a trade-by-trade approach when it does close out derivatives – the hypothetical situation imagined by the MVA process – what’s the use of the accounting numbers to begin with? If they don’t reflect reality, shouldn’t they change? According to Lehman’s experts, if you scaled Citi’s derivatives collateral for what it actually charged Lehman, it would leave Citi undercapitalized by $50 billion.
Does Citi really need an additional $50 billion capital cushion? Of course not. When they actually replace trades for their own risk management, they do so in a rational (read: netted) way. It did not and would not ever cost $2 billion to replace Lehman-sized load of derivatives. It’s fair to say that the number is a fiction. What matters is whether it can become a legal fiction.