MORE IN HEARD ON THE STREET »
By DAVID REILLY
Who knew what when? That is a big question still hanging over J.P. Morgan Chase'sJPM +1.08% more than $2 billion trading loss. Also important, though, is the corollary: Who didn't know what and why?
Clues to both may lie in the circumstances surrounding an adjustment to a relatively esoteric gauge of risk within the bank's Chief Investment Office, which managed the losing trades. Sometime in the first quarter of this year, the investment office adjusted the model for measuring what is known as its "value at risk," or Var, according to the bank.
This measure looks to gauge the maximum potential daily loss for a firm,business unit or type of product. There are plenty of flaws with this metric, which, for example, didn't help in the financial crisis. But banks use it as a benchmark for assessing the performance of trades and as a way to spot brewing problems. Ironically, the measure was created in the mid-1990s by J.P. Morgan.
Changes to the model behind the gauge aren't taken lightly within banks. So it is notable that J.P. Morgan changed the model at the beginning of the first quarter, and then in announcing the trading loss, said it was again changing the model—this time back to the original settings.
The result: The bank originally reported that the investment office had an average trading risk of $67 million, $2 million lower than the previous quarter. After reverting to its original model, J.P. Morgan amended the first-quarter measure to $129 million, showing there had been a huge spike in the potential for loss within the investment office.
Beyond the specifics of what exactly was adjusted within the model, the changes raise two far-broader questions. First, why was it tinkered with? Was it because someone thought the model was falsely showing a too-high level of risk for what was deemed to be a safe trade? Or had the trade gone bad and someone didn't want the Var model to start alerting others to rising risk? The latter would indicate the possibility of wrongdoing.
Second, even if the change was made for a seemingly benign reason, who knew of and approved it among bank executives? Notably, the investment office reporting directly to Chief Executive James Dimon?
If top management wasn't properly consulted, this could show a serious breakdown in internal controls and risk systems. Plus, management and regulators may have been deprived of an early warning sign that something was going wrong.
So far, J.P. Morgan has given little detail around the trading losses and has said it won't furnish more until after the end of the second quarter. Mr. Dimon also agreed last week to testify this summer before a Senate Banking committee hearing on the losses.
For Congress, questions about the bank's risk measures are particularly important. That is because issues around them also touch on the way banks manage capital and report risk to investors.
In some cases, value-at-risk models help to determine how much capital banks hold against trading assets. That gives them leeway since these are internally generated models. The danger: Banks have an "incentive to game their model to lower their regulatory capital requirement," says former Federal Deposit Insurance Corp. ChairmanSheila Bair. The risk grows if banks can tinker, as J.P. Morgan's episode shows may be possible.
J.P. Morgan's experience is also a reminder that the risk measure varies among banks. Indeed, the Office of the Comptroller of the Currency, in a quarterly report on derivatives activity, "cautioned" that the risk gauge can mislead.
The OCC said that J.P. Morgan, Goldman Sachs GS -0.68% and Morgan StanleyMS +0.60% calculate the measure with confidence there is a 95% chance that losses will be within the reading. If the firms used a 99% confidence level like "Bank of AmericaBAC -1.75% and Citigroup, C +0.33% their Var estimates would be meaningfully higher," the OCC said.
In light of that, J.P. Morgan is a reminder for both banks and investors that attempts to measure risk, let alone tame it, are fraught with hazard.
Write to David Reilly at david.reilly@wsj.com
No comments:
Post a Comment