Staring back at me from the front page of Sunday’s New York Times was a headline that promised an answer to a puzzle that had endured for more than a month, and which I have explored here and here. The blame for the multi-billion dollar JP Morgan credit default swap fiasco had been discovered. “Eureka,” I exclaimed over my weekend omelet, drowning out the soothing intonations of the morning NPR news mavens. “Now we know.”
It seems that the bank’s Chief Investment Office was riven by discord and racked by disease, leading me to recall the fate of ancient Athens in the Peloponnesian War (perhaps this was also an echo of the Greek debt crisis that renders the Morgan loss even more ominous). The article portrays the recently-resigned head of the Chief Investment Office, Ina Drew, as a level-headed and competent manager who had successfully guided her operation through the financial crisis, helping to build the reputation of Jamie Dimon and JP Morgan Chase for risk management. The story speculates that Dimon’s confidence in Drew colored his assessment of the massive position that generated the current loss.
But Drew contracted Lyme disease in 2010 and took an extended leave of absence. While she was away, the heads of the New York and London Chief Investment Office operations fell into bickering over the amount of risk being compiled by the London desk, specifically the trades of Bruno Iskil (the so-called “London Whale” or “Voldemort,” depending on the specific tastes of the traders at competing institutions). The head of the London desk, Achilles Macris, supported Mr. Iskil and shouted down the warnings by Althea Duersten, his counterpart in New York. Even when Drew came back from her sick leave, she was unable to re-assert control. As we now know, Ms. Duersten had an excellent point.
If this makes you think of a better dressed, less tanned version of Jersey Shore, you are not alone.
So we are to believe that JP Morgan Chase really lost between $3 and $5 billion thanks to a tick bite? And we are to see Jamie Dimon as a boss whose faithful loyalty to a trusted employee blinded him to the truth. As I plowed through the article, I wondered whether this was a clever way to throw Drew and her crew under the bus in order to control damage to Dimon, the Dapper Don of Wall Street.
More importantly, to say that the blame for the JP Morgan Chase loss lay at the feet of Lyme disease and internal bickering is like saying that gravity is to blame for the death of the fellow hit by a falling piano while walking on the sidewalk. Gravity and dysfunctional interpersonal relationships were causal factors in each case. However, the blame lies with the systems that allowed each to affect the state of affairs. In the case of the piano, the system was a rope hoist. For JP Morgan Chase, it was its risk controls.
Sick leaves and turf wars occur in every organization. The story here is that JP Morgan Chase’s risk controls were so anemic that they were frustrated by the bickering of trading desk heads. Risk control systems lack the dramatic appeal of screaming arguments in transatlantic phone calls, but the effectiveness of these systems is far from trivial. Washington regulators are deep into the process of adopting rules mandated by the Dodd-Frank Act that promise to re-shape the financial services sector for decades to come. The prescriptiveness of these rules is influenced by the reliability of the risk control systems of the banks. The big banks, and especially JP Morgan Chase, loudly claim that their risk controls are so efficient that the rules need only to lay out some broad principles for guidance. There is no need to restrict bank behavior, they proclaim.
The recent episode illustrates that this assertion is nonsense. JP Morgan quickly pointed out that it had tweaked its value at risk metrics, claiming to correct the problem. But this is not an algorithm issue. The value of derivatives is a devilishly complex matter. Value at risk is a statistical evaluation based on historic price observations. It will always be the case that the huge number of factors influencing derivatives value will interact in unpredicted ways to defeat the utility of every statistics-based valuation device.
Corporate and household lending, the major permitted bailiwick of depository banks if the reforms embodied in the Volcker Rule are effectively implemented, is completely different. Default rates are much more accurately predictable using statistical calculations because the factors influencing derivatives values are so much more complex. Notably, recent academic research points out that flaws in statistics-based valuation models employed by the rating agencies to evaluate credit differences among the multiple tranches of mortgage-backed securities. (John Hull and Alan White, “”Ratings, Mortgage Securitization and the Apparent Creation of Value,” November 2011, available here.) These flaws led to the dysfunctional mortgage bond market that melted down in the crisis. The principles are precisely the same as those related to complex credit default swap positions.
The roles played by Lyme disease, turf wars and algorithmic shortcomings were largely irrelevant, though each intriguing its own way. The real story is that a major, too-big-to-fail bank, benefiting from FDIC insurance and access to the Fed window, should not be making speculative proprietary bets of mammoth proportions. Even assuming the accuracy of JP Morgan Chase’s claim to the best risk controls on the planet, the bank’s systems cannot calculate an incalculable outcome. The managers of banks seem unable to avoid believing that the systems measure the worst case outcome. They are seduced by “99.9%” confidence intervals, seemingly unable to comprehend that the math only measures what it is designed to do, and design is subject to the limits of knowledge and human bias. And statistics are not useful for predicting outcomes in complex, fluid systems.
If these risk metrics are going to be employed to justify putting the world economy at risk of another financial crisis and the depression that will likely follow on from it, the world, including Morgan, would be better off if the calculations were never run.
It is comforting to note that prohibiting proprietary trading by insured banks is far more achievable than a ban on getting sick and self-interested bickering over turf.