Posted on October 26, 2012 by Ben Connard

On May 10th, Jamie Dimon, CEO of JPMorgan Chase & Co, announced a trading loss of about $2 billion in the bank’s Chief Investment Office (CIO) and noted that it may increase to $3 billion in the month ahead. The stock fell 9.3% on May 11th, the most in almost nine months. Speculation has the loss increasing to $5 billion or even $9 billion and it may take the rest of the year to liquidate the trades.

The previous month, Bloomberg News and the Wall Street Journal reported a JP Morgan trader named Bruno Iksil, aka the London Whale, was taking derivative positions so large they were distorting the market. At the time, Dimon called the matter a “tempest in a teapot.” The bank states that its CIO is focused on hedging and is not involved in trading for short-term profits.  

Using these events as a guide, can an investor learn to spot the potential risk based on data the company releases? We can start with the income statement, balance sheet and cash flows reported quarterly. The detail in these numbers is shockingly limited. The balance sheet should have some information, but $2.3 trillion in assets is broken into only 7 categories, including a generic $351 billion in Trading Assets.

JPMorgan files more detailed reports with the Board of Governors of the Federal Reserve System. Here, the assets are listed in more detail. For example, trading assets are broken down into treasury obligations, loans, etc., but there is still an “other” category which is almost one third the trading asset’s value (or well over $100 billion). 

In addition to being limited on details, these top-line reports give us very little hint of the risks the company is taking. In more detailed disclosures, banks report Value at Risk (VaR).  The VaR is modeled given expected returns, their volatility and the correlation of those returns among the various assets. VaR is usually given at the 95% confidence level and is supposed to represent the maximum loss 95% of the time. For example, a daily VaR of $1 million means in 19 of 20 days the loss will be less than $1 million.

The value of VaR can be debated. It tells us nothing about the magnitude of those losses on 5% of the days even though those losses are what are most important. However, given that it’s what we have, we can at least hope the banks report it accurately and faithfully.

JPMorgan’s history with VaR is complicated. In 2009, questions arose when its VaR rose from under $100 million in 2007 to $336 million in Q1 2009. At that time, Dimon stated that he doesn’t pay much attention to VaR and it’s tied to hedge positions. In 2010 and 2011, the banks VaR appears to fall. But in 2012, JPMorgan changed the VaR model used in its CIO so it showed half the VaR that would have been recognized in the old model—and the VaR fell to $67 million.

This change was short-lived as the old model was reinstituted after the trading loss was announced on May 10th so that VaR increased back to $129 million. In other words, JPMorgan dismisses VaR as a measure of risk, retools the formula to artificially lower risk, and then, only after a $2 billion (or more loss), decides to revert to the old, supposedly more accurate model.

On top of JPMorgan’s VaR games, reports came out at the end of May that the bank’s CIO valued some of its trades at different prices than its investment bank. This news means the inaccuracy of the VaR model was further exacerbated by the poor inputs.

JPMorgan’s risk metrics are poorly calculated at best and intentionally misleading at worst. The bottom line is JPMorgan, and most other bank conglomerates, give us very little hint to their day-to-day risks. The only hints we get is when the loss rears its ugly head, by which time the stock has already fallen.