Over at Matthew Hancock MP and Nadhim Zahawi MP's Masters of Nothing blog, the shock $2billion loss at JPMorgan, and how it has awakened unwelcome memories of the financial crash, is discussed. Masters of Nothing: How The Crash Will Happen Again Unless We Understand Human Nature is available here.
JPMorgan's shock $2 billion trading loss has resurrected unwelcome memories of the crash: flawed risk management, a huge position gone awry, condemnation from politicians and a downgrade from the ratings agencies. In response to last week's disclosure investors sliced $20 billion off the bank's market cap; 40% of JPM shareholders have now voted for the combined roles of Chairman and CEO to be broken up.
The loss orginated at JPMorgan's London-based chief investment office, which trades derivatives to hedge the bank's balance sheet. They sold insurance on one credit index and bought insurance on others. The loss seems to have occurred when the relationship between the two changed, moving by more than their models suggested they should. Moreover, JPMorgan's large position in one particular index left it vulnerable when the market turned, as speculators forced up the price against the bank.
The irony is that until recently JPMorgan was hailed as unusually safe and conscientious – at least by the standards of its peer group. In the years before the crash CEO Jamie Dimon pursued a relatively conservative business model and was known to have bolstered the prestige of his risk managers by increasing their pay relative to traders. JPMorgan emerged from the crisis one of the America's best capitalised banks, cementing its reputation as a safe haven for depositors.
So how was such a costly mistake allowed to happen, especially at a time when many politicians are calling for tighter regulation of the banking sector? There seem to be two factors at work. The first of these is a flawed modelling of risk. The day it announced its $2 billion loss JPM also revealed that the Value-at-Risk (VaR) at its chief investment office was nearly double the amount previously reported.
VaR is a measure of risk used by large banks to calculate the maximum they could expect to lose trading over a given period, commonly a day. Risk managers use the model to produce statements like, 'You can be 95% sure you will lose no more than £32.4 million over the next 24 hours.' The attraction of VaR is that it allows banks to put a single numerical figure on risk across a whole firm – a figure that can be used to guide the decisions of managers at one remove from the daily flux of markets.
The problem with the model is that it relies on assumptions about probability which fail to account for the unpredictability of market behaviour. Extreme events in financial markets are more common than the maths suggests, possibly in view of investors' lemming-like tendency to follow the herd. Flawed VaR models meant the banks entered the 2008 crisis with a fraction of the capital they needed to absorb the losses made on bad debt. Yet VaR persists because the ability to put a single figure on risk is remarkably comforting for those in the business of hedging against uncertainty.
The second factor relates to the way in which models interact with human behaviour. The apparent objectivity of VaR (and models in general) can amplify a successful trader's bias towards optimism. Behavioural research has shown that the outcome of successful trades made in the past can become inscribed on decisions made in present. This is why, in casinos, big money game shows - and banks - winners rarely quit while they're ahead. By all accounts the traders at JPMorgan's chief investment office were highly successful, yet it may have been precisely that winning streak which led them to bet the house on a flawed position.