Morgan’s mess

JPMorgan’s $2B trading loss is pretty big news, but some other big news, if you are interested in ERM, came early in the conference call. (You’ll have to sign up to listen.)

“We’d shown average VaR at $67” million for the chief investment office, CEO Jamie Dimon said. “it will now be $129” million. (Educational aside: VaR is Value at Risk, the amount a company believes is the most it is likely to lose on a bad day in a normal trading environment.)

Morgan had switched VaR models last quarter. Upon further review, the company has concluded, the new one sucks. So it has returned to the prior version; hence the leap in perceived risk.

It sounds like trading strategies that looked solid under the new model were really crappy. In the conference call, Dimon self-flagellates regarding the planning and execution of a flawed trading strategy, which he doesn’t describe but is elsewhere linked to “hedging” by the London Whale, a famous-turned-notorious trader.

“In hindsight, the new strategy was flawed, complex, poorly reviewed, poorly executed and poorly monitored,” he said.

The fact that he doesn’t describe it is ominous. That and other information in the call make me think the trades haven’t been unwound, and if Dimon tipped off what they were, aggressive traders (read: Goldman) would grind the position into dirt.

And Dimon’s applying some sugar-coating, to a banking novice like me.

First, the VaR he cites is a one-day window. (Insurers generally use a one-year window.) That’s standard for banking, so OK.

But not OK: The VaR is a 95% confidence level. Insurers prefer a 99.5% level, or 99.5%. And it takes a lot more capital to support a 99% VaR than a 95% – up to 50% more, based on my back-of-the-envelope playing around with z-tables.

Worse still, the revised VaR is the average for the quarter. But the risky position was built up at the end of the quarter. Quarter-end VaR for CIO was $186M, or about 50% higher than quarter average. Max for the quarter was only $1M higher.

Worse still: Emphasizing CIO’s VaR makes it sound like investments around the rest of the company correlate away CIO’s risk. They do, but not to the degree you might think.

For the whole enterprise, VaR doubled from last quarter, to $170M from $88M. And the company ended Q1 with VaR of $201M. Looking at the ratio of enterprise VaR to CIO VaR at various points in time, it doesn’t look like the hedging strategy really was hedging, but what do I know?

Above, for your enjoyment, I’ve enclosed a screen shot (click to enlarge) of the relevant page from the 10-Q (pdf).

None of this is to suggest the Morgan is imperiled. To my untrained eye, shareholders’ equity is about $190B, seemingly enough to absorb a whole lot of goofy trades and lousy capital models, but not enough to prevent me from reiterating the Agnes Rule: Banks can’t sell anything but money.

As far as the multiple ERM mistakes here, I leave that to the reader as an exercise.

Added bonus: Here JP Morgan explains Value at Risk to the great unwashed.


3 thoughts on “Morgan’s mess

  1. DW says:

    You’d probably want to net out the 60m of intangibles from the equity calculation…

    ah, forget it. Analyzing a TBTF bank’s financial statements is pointless. It’s impossible to figure out the #1 thing that matters: their exposure is to short-term (intra-month) financing, which is what took down Lehman, Bear, etc.

    It’s the shadow bank run that kills these guys.

  2. jimlynch9999 says:

    Yeah, all I did was divide 99th percentile z-score by 95th. Doing more without knowing a lot more is probably futile, as you note.
    I didn’t even assume a lognormal curve, which is standard in insurance. That’s because, at least as of a couple years ago, I’d heard banks were assuming normality.
    Which brings us back to the Agnes Rule.

  3. Hi!

    Based more on your comment on Felix Salmon’s post than on your writing above, it appears that you have a similar position to Frank Partnoy in a recent opinion piece in the Financial Times. That is, more information about the risk would lead to better decision making about the risk. But, as is seen clearly in comparing catastrophe models- their uses, by whom, and for what purpose- the value of risk estimates depends upon desired outcomes. Sort of a situation where ends justify modeling methodology (as seen recently with deciding which AIR model to use to rate Citizens’ cat bonds). It seems that Salmon’s position differs from yours and Partnoy’s because he emphasizes that it is ultimately the human user that makes judgements about which model to use and what the modeled risk means for the decision at hand. More risk information does not appear to be pragmatic solution.
    I wrote about it here:

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s

%d bloggers like this: