Autopsy of the JPM Whale
Last month JP Morgan released the report of a task force set up to explore the huge trading losses suffered last year in its CIO group. The facts, in brief, appear to be as follows:
CIO stands for Chief Investment Officer, and the original purpose of the CIO group was to manage JPM’s liquid asset portfolios. In 2010, an additional objective was established of seeking to manage the credit risk exposure of the bank. That is, the purpose of a bank is to make loans and manage the interest rate, credit and liquidity risk of doing so. Given the existence of derivatives markets in corporate credit (chiefly, the Credit Default Swap or CDS), it is now possible for a bank to lend heavily in commercial credits and then hedge or reduce that risk by appropriate transactions in CDS or other credit derivatives. As of the end of 2011, the CIO group had positions in CDS that had the effect of reducing JPM credit risk. In particular, they were “short high yield” which means they had taken positions in derivatives that would pay off if credit spreads on high yield (risky) credit rose. Then, the feeling among top management was that credit conditions were improving so that the CIO group was directed to reduce its hedge positions. Well, it began to do so but found that the exit prices were not attractive. That is, they were only able to offset existing positions at prices that meant losses, and the feeling was that further selling would drive prices lower and create more losses (the accounting for hedge positions, as opposed to loans, is that they are marked to market).
So, instead of reducing the short high yield position, the CIO traders decided to build a long investment grade position. The idea, apparently, was that the high yield CDS and investment grade CDS would move in tandem, and if the combined position was short high yield and long investment grade, the longs and the shorts would more or less offset one another, so even though the gross amount of positions was rising, the net risk was falling. Between January and March 2012, the traders kept building both the long and short positions. In the report there are references to the traders attempting to “defend” their positions. I take that to mean that by expanding an existing position a trader would tend to support the pricing or valuation of that position (and improve that trader’s profit/loss statement). In just a few months these positions got to be enormous.
By the end of March 2012, rumors were circulating in the market of a “Whale” (very large trader, who was affecting market prices by the size of his positions) operating in the CDS market. This “Whale” was rumored, rightfully so as it turns out, to be a member of the JPM CIO team. Having established the existence of the whale, other traders took opposite positions in the market, on the theory that the whale would eventually have to trade out of its positions (to “surface” so to speak). By the time of the earnings release for the first quarter of 2012, the JPM management team were fully apprised of the situation, but still thought that it was manageable. It was a “tempest in a teapot” according to Chief Executive Officer Jamie Dimon.
The task force report carefully tracks the assessment of actual and prospective loss through the days of March and April 2012. The expected loss, and range of possible losses, turned out to be woefully inaccurate. Ultimately, JPM suffered a trading loss of $6.2 billion on the Whale position. Yet, risk reports as late as mid-April showed a worst-case loss of a few hundred million dollars. What is disturbing about this is that JPM no doubt has the best trading and analytical people and systems in the universe. And yet they were massively wrong.
One take is that there were mistakes made and there now have been improvements in the risk management process to address those mistakes, so that the system is now stronger. This is the point of view taken by the task force report.
A different take is that we continue to place undue confidence in quantitative models that are based on historical data. Yet again we see in this report mention of a large standard deviation (SD) event (in April the head of the CIO Group, Ina Drew, asserted that they had just experienced an “eight SD event”). The implication is that this event was so unlikely as to be out of the realm of reasonable consideration. Yet, huge SD events seem to happen all the time. We should become disabused of the idea that high SD events are remote.
It is impressive that JPM has sufficient capital and earnings power to absorb a $6.2 billion trading loss and still report positive earnings in every quarter of 2012 (it is also testament to the skill and dexterity of the accounting staff at JPM). However, in my view, this episode does undermine the notion, most forcefully pushed forward by Jamie Dimon himself, that bank capital requirements are adequate and should not be raised.
Three percent of JPM capital was wiped out in this episode, with little or no contemporaneous understanding by senior management. Fortunately, JPM had a capital cushion, but had they been operating at the current capital standard, they would have become capital deficient and might have needed yet another government bailout. Let’s agree on a level of capital that banks need to operate, say 10% of assets, and then require a capital cushion above that to absorb losses for 10 (or at least eight) SD events.