How does JP Morgan respond to a crisis?

By Felix Salmon
January 16, 2013

If you have a bit of time today, the official JP Morgan post mortem on the London Whale affair is well worth reading. The whole thing is 132 pages long, although the executive summary — which is very clearly written — is only 17 pages.

One thing the report certainly does is reinforce my conviction that you can’t hedge tail risk. The losses all took place in something called the Synthetic Credit Portfolio, which was described as a “Tail Risk Book” — something designed to make money “when the market environment moves more than three standard deviations from the mean based on predictions from a normal distribution of historical prices”. In other words, JP Morgan is well aware that market moves are not normally distributed, and therefore it has a whole derivatives book in place to protect itself against inevitable unexpected events.

The whole point about tail risk, however (a/k/a “black swans”) is that you can’t anticipate exactly what it’s going to look like before you see it. In this case, the biggest tail event was the publication of stories, in the WSJ and Bloomberg, talking about JP Morgan’s positions. Those stories had a massive effect on the mark-to-market valuation of JP Morgan’s positions At the beginning of the first trading day after the stories appeared, it looked as though JP Morgan might be facing a one-day loss of $700 million; in the event, the final official number was $412 million. Ina Drew, the person in charge of the portfolio, sent an email to JP Morgan’s CEO and CFO, in which she observed that the move was “an eight-standard-deviation event”.

The report doesn’t say how many eight-sigma events the CIO has ever seen: my guess is that this is the only one. But here’s an idea of how crazy eight-sigma events are: under a normal distribution, they’re meant to happen with a probability of roughly one in 800 trillion. The universe, by contrast, is roughly 5 trillion days old: you could run the universe a hundred times, under a normal distribution, and still never see an eight-sigma event. If anything was a black swan, this was a black swan. And it didn’t help JP Morgan’s “tail risk book” one bit. Quite the opposite.

Another thing the report does is show just how difficult it is for any large organization to actually implement what managers want. At JP Morgan, for instance — where the CEO has an unusually large degree of power and knowledge of what is going on — the whole firm was meant to be reducing its “risk-weighted assets”, or RWA, since the higher a bank’s RWA, the more capital it needs under Basel III. And yet somehow, by the time this directive trickled down to the London Whale, it had been watered down and misinterpreted to the point at which the office’s RWA actually went up — substantially — rather than down.

What’s more, there’s a constant theme running through the report of managers being told what they want to hear, rather than the truth, especially with regard to substantial losses. When those appear, no one wants to tell Ina Drew about them; instead, the traders do everything they can to try to either fudge the numbers or attempt to trade their way out of the position.

Interestingly, one way that numbers were fudged was to use the favorite tool of quants around the world, the Monte Carlo analysis. After the Bloomberg and WSJ stories appeared, for instance, one trader drew up an analysis of just how bad the position could get. He modeled nine extreme event like a “bond market crash” or a “Middle East shock”, and found that in six of them, the portfolio lost money, with the losses ranging from $350 million to $750 million. This analysis did not go down well:

This trader sent his loss estimates to the other on April 7. According to the trader who prepared the loss estimates, the other trader responded that he had just had a discussion with Ms. Drew and another senior team member, and that he (the latter trader) wanted to see a different analysis. Specifically, he informed the trader who had generated the estimates that he had too many negative scenarios in his initial work, and that he was going to scare Ms. Drew if he said they could lose more than $200 or $300 million. He therefore directed that trader to run a so- called “Monte Carlo” simulation to determine the potential losses for the second quarter. A Monte Carlo simulation involves running a portfolio through a series of scenarios and averaging the results. The trader who had generated the estimates did not believe the Monte Carlo simulation was a meaningful stress analysis because it included some scenarios in which the Synthetic Credit Portfolio would make money which, when averaged together with the scenarios in which it lost money, would result in an estimate that was relatively close to zero. He performed the requested analysis, however, and sent the results to the other trader in a series of written presentations over the course of the weekend. This work was the basis for a second-quarter loss estimate of -$150 million to +$250 million provided to senior Firm management.

In the event, of course, the portfolio ended up losing not $150 million, not even $750 million, but more like $6 billion, with some $800 million of those losses taking place in the six trading days leading up to April 30, long before the decision was made to liquidate the position. Which just goes to show how useful stress tests are. (Remember, the initial worst-case estimates were put together after the WSJ and Bloomberg stories appeared, which means that JP Morgan was acutely aware, at this point, of the risk that the market would move against them just because their positions were public.)

There is one big omission in the report — and that’s any discussion of how the ultimate losses in the portfolio grew to be so enormous. Where did the initial $2 billion estimate come from, and how did it grow to $6 billion by the time all was said and done? The report basically ends when the potential losses were made public, and doesn’t spend any time discussing how Jamie Dimon and his senior executives handled everything from there on in.

From the perspective of JP Morgan’s shareholders, there are two big things to worry about in this whole episode. The first is weaknesses in risk management, which the report goes into in great detail. The second is the way that senior management responds to a crisis, and whether it can do so while keeping its head and minimizing losses. On that front, the report is silent. Did a panicked reaction to the early losses result in a “dump everything immediately” response which ended up causing an extra $4 billion in damage? Who was responsible for those $4 billion in losses, and how avoidable were they? Those questions are never asked, let alone answered.

JP Morgan has looked in great detail at its crisis-prevention architecure: it’s time, now, too look at its crisis-response architecture, too. Because sometimes, it seems, the latter can cause more damage than the former.

13 comments

We welcome comments that advance the story through relevant opinion, anecdotes, links and data. If you see a comment that you believe is irrelevant or inappropriate, you can flag it to our editors by using the report abuse links. Views expressed in the comments do not represent those of Reuters. For more information on our comment policy, see http://blogs.reuters.com/fulldisclosure/2010/09/27/toward-a-more-thoughtful-conversation-on-stories/

This is interesting stuff, but I think the marketplace has to reconsider what a tail risk is.
These things that are called tail risk simply are not tail risk.
A newspaper article is not tail risk. And as you point out, even the beginning of time is not an eight-sigma event, so a bond loss certainly doesn’t qualify as one.
Obviously, everything we’re talking about here is far more common than the financiers believe. If that’s the case, they will always bear far more risk than they know.

Posted by RZ0 | Report as abusive

There is something else for JP Morgan’s shareholders to consider. Their TBTF bank has gotten so big that a Ponzi control freak like handsome Jamie Madoff can’t possibly know what the hell is going on.

Posted by williambanazi7 | Report as abusive

There is no reason to assume that these events are normally distributed on the parameters used, or that they have enough data (long enough time span) to really define the variance even if they are normally distributed.

Posted by skeptometric | Report as abusive

How can you have a crisis-response strategy when the crisis is bad literally beyond your wildest imaginings (eight sigma if the original analysis was correct, which it was not)? At that point the whole enterprise has been shown to be invalid; it should never have been undertaken. It should have been realized that the risks of excess leverage are normally underestimated – the high payoffs distort the analysis. This is why there is regulation – these decisions can’t be left up to those who stand to profit.

Posted by skeptometric | Report as abusive

All this just comes down to the basic rule of Wallstreet:

I make a high risk high reward bet where if I win I become fabulously wealthy. But if I lose, it is my employer who takes the loss (and the taxpayers), and I only lose my job, not any of my previous pay which was predicated on the idea of this bet performing.

Posted by QCIC | Report as abusive

Playing and gambling with immense sums of money is inherently risky. That is why the institutions that do it should be made small enough so that if their guesses go wrong, as they often do, the damage is limited. Obama had a chance to get the banks under control at the time of the TARP bailout but Geithner, to whom he listened too much, prevented that. See the recent Frontline documentary on the issue. Obama was, and still is, too fearful of Wall Street. I think he fears being seen as (horrors) a “radical” black. Little good it does him as you can see from incessant and numerous comments on Yahoo News, etc., calling him a “socialist”. Millions of Americans apparently think bowing to Wall Street is what “socialists” do.

Posted by Chris08 | Report as abusive

JP Morgan declares: “This failure is not our fault. It’s cuz Obama!”

Posted by AlkalineState | Report as abusive

“There is no reason to assume that these events are normally distributed on the parameters used…”

It’s much worse than that. There is very strong that these kind of events are NOT normally distributed. Financial markets are known to enter very strong positive feedback loops at the tails and if you have a positive feedback loop you definitely won’t have a normal distribution. (The strongest kind of feedback loop, perhaps, being the predatory trading behavior of other hedge funds once they realize you’re vulnerable and will eventually have to capitulate.) Which, of course, we all know since this kind of event or even larger happen multiple times per decade not once every 10 billion years. The fact that they even measure things with concepts like standard deviation is a huge red flag that they are smart enough to have memorized their way through Statistics 101 but not nearly smart enough to understand what they’re really dealing with.

Posted by TGDC | Report as abusive

You keep using that word ‘Monte-Carlo’. I do not think it means what you think it means.

Posted by seanmatthews | Report as abusive

The point of all this is that calling this event an “eight sigma” event, or anything else to do with a normal distribution, is folly. The risks associated with banks’ assets are not normally distributed. A risk simulation (including a Monte Carlo) that generates a bunch of random scenarios in which parameters are varied normally may be useful in some circumstances, but it should absolutely not be your guide to what the “worst case scenario” is. Not even the lowest percentile, or tenth of a percentile, in your Monte Carlo is the worst case scenario. If you want the worst case, you sit down with some creative pessimists — like the guy who did the nine scenarios for things like an oil shock — and draw up what you think are the very worst plausible cases. Then you simulate something that’s much worse than even that. And even then you’re probably underestimating the potential damage.

Posted by Auros | Report as abusive

I wonder if the next post-crisis report will be better than this one?

Posted by f.fursty | Report as abusive

To this mid-level rube banker from the deepest part of the Maine woods the “task force report” reads like a company trying to make some very simple mistakes look complex. The “whale trades” out of the CIO have always been explained / justified as a hedge against the investment portfolio.

Hedging a risk does not ever under any circumstance generate a profit. Hedging reduces profit or losses. You have a position A which you think will make money, (in this case the 300 billionish in completely sensible bonds that were entirely appropriate for JPM to be holding with excess short term deposits.) To offset the risks of position A you have hedge B, the net very long CDS position. Nothing remotely hard to understand about that. Large bond portfolio that I am required to mark to market so I want to hedge the risk and I know that this will cost me some of the money I earn holding the bonds.

You don’t have to go any further than page 7 of the executive summary:

“…the traders hoped that the combined effect of these additions would allow them, among other things, to EARN premiums…” you can stop right there knowing everything you really need to know. JPM wanted the cake and to eat it too. Their completely appropriate hedges were costing them many hundreds of millions of dollars. They knew this going in… but at some point they grew tired of paying 9 or 10 figures for insurance.

If only they could use their clearly superior risk management models, trading acumen, and sterling reputation to earn just a little of their premium back…

…if you want to read further (and the report really is porn for wantabe wonks like me) then continue on to page 37:

“e-mails on January 30, the same trader suggested to another (more senior)trader that CIO should stop increasing “the notionals,” which were “becom[ing] scary,” and takelosses (“full pain”) now; he further stated that these increased notionals would expose the Firm to “larger and larger drawdown pressure versus the risk due to notional increases.” While the documentary record does not reflect how, if at all, the more senior trader responded to these concerns, the traders nonetheless continued to build the notional size of the positions through late March.”

Kid Dynamite has covered this better than I ever could. When you find yourself suddenly deep in a hole and you have no idea how you got there the very first thing you should do IS PUT DOWN THE FREAKING SHOVEL.

So JPM did the opposite on two obvious decisions:

A.) they tried to shuffle exposure so cute that they could somehow make money instead of paying money to insure an asset.

B.) when some of them realized the totally, glaringly, stupidly, obvious flawed strategy… rather than SLAM on the breaks and tell mom and dad the bad news… they put the petal to the metal thinking that somehow they could trade their way out of the unholy mess they got themselves in.

Posted by y2kurtus | Report as abusive

Wow, wow, wow, wow…

Using a Monte Carlo simulation and averaging the outputs for a non-linear system is kind of missing the whole point of Monte Carlo simulations in non-linear systems.

Monte Carlo simulations are useful for, say, nuclear reactors. A nuclear reactor is a linear system from the POV of neutron transport. So Monte Carlo simulations can yield an accurate picture of the state of a running reactor as a superposition of many single particle simulations (google LANL MCNP).

In a non-linear system, Monte-Carlo simulations are useful to quickly explore the parameter space and sniff around for non-obvious, mmm, trouble, based on the assumption that the state space of most systems, even strongly non-linear systems, tends to be continuous. But you never, ever superpose states in a non-linear system. And a financial model is never linear (for they all have at least one very strong, very stateful non-linearity called ‘insolvency’).

And yes, the unfortunate conclusion is that financial models are more dangerous than nuclear reactors. Wall Street urgently needs a NRC of its own.

Posted by Frwip | Report as abusive