Sort by

The London Whale, the FBI and Math

The FBI is reported to be looking into phone calls and theactions of four employees in its investigation into events surrounding the infamous “London Whale” trades that lost JP Morgan Chase upwards of $6 billion earlier this year. It has also been reported that the SEC is on the case.

Jamie Dimon has repeatedly diminished the importance of the misguided attempt by the bank’s Chief Investment Office to effectively corner a segment of the credit default swap market. In Congressional testimony, he took full responsibility in a dramatic “buck stops here” statement. But he also characterized the episode as a failure of management to maintain diligence, not a fundamental malfunction of JP Morgan Chase’s vaunted risk management systems.

Are the FBI and SEC investigating management foul-ups? I think not.

As reported previously here, the core issue of the London Whale episode involves arithmetic, or rather the complex Value at Risk algorithm universally used by banks and derivatives clearing houses to measure the risks posed by holding derivatives. VaR has been much maligned, especially following the 2008 financial meltdown in which banks almost universally mis-measured their risk exposures to the mortgage market. But, to paraphrase the pro-gun special interests, VaR doesn’t kill banks, people kill banks.

VaR is a statistical algorithm that calculates risk based on historic price movements. The basic math and its application are driven by three factors: the time it takes to get out of a derivatives position if you have to (market liquidity); the amount that prices might move against you as you get out of the position (historic price volatility); and price movement correlations within your portfolio of derivatives. Price correlations are relevant to measuring how the same market price move might benefit you as well as generate losses so that compensating profits and losses can be offset.

The VaR calculation is mechanical, but the assumptions are a matter of judgment. The three main drivers of the calculation are set by management decisions that can alter the outcome dramatically. It is clear that this process played out in the early part of this year as the VaR calculation used by JP Morgan Chase was altered so that the apparent risk of the London Whale’s massive credit default swap position was reduced. Subsequently, the bank is reported to have reversed that change, compelling the bank to get out of the positions.

The public does not know what caused the change in VaR assumptions, but it is very likely that it was a dispute that is commonplace at trading firms. Risk managers generate the VaR calculations. When the risk measurement is inconvenient for traders who want to take or maintain a position, the traders push back, arguing that the assumptions should be changed. Traders have immense power within their organizations. They generate profits. Risk managers have much less power. They are seen as curmudgeons who kill the dopamine-fueled high of the risk-addicted trader. It is very likely that the London Whale and his colleagues successfully pushed the risk managers to alter the price correlation assumptions to reduce the apparent risk represented by their portfolio. Like an obsessive gambler who can’t bring himself to walk away from the blackjack table, the London Whale altered the truth hoping that he would win if only the game continued. If this is true, he misled management, but also JP Morgan Chase investors.

If this scenario is accurate, the implications are immense, reaching far beyond the fates of those who might be prosecuted and the damage done to the bank and its colorful CEO. Federal regulators are currently crafting the capital and leverage requirements for banks. A central question is how to measure risk so that the capital requirements of a given bank will be proportionate to the risks it bears. Unsurprisingly, the measurement of derivatives risk is the thorniest issue.

There are two levels contemplated in the rules. In one, the banks are given leeway to use “advanced” risk metrics to calculate their own risks. In the second level, simpler metrics are imposed by rule. The highest risk measurement is the one that must be covered by capital. If the London Whale fiddled the VaR calculations, it is a stark illustration of the futility of allowing banks to calculate their own risks. The safety and soundness of the system depends on the metrics imposed by the rules.

I am sad to report that the metrics proposed by the regulators are nothing short of nonsense. It’s not that they are too lax. They simply do not calculate a useful number. The regulators seem to be obsessed with their own lack of resources and sophistication, grasping for an absurdly simple procedure whose main benefit is that it can be implemented using a handheld calculator. This will not work. Whatever it costs, the people who are tasked with monitoring the banking system simply have to use math that approaches the sophistication of the banks they must oversee.