Risk Mismanagement

How risky is a nuclear plant, an offshore drilling platform or an investment bank? When disaster strikes, we invariably hear that an unlikely confluence of events (the proverbial perfect storm) led to the catastrophe. Economist Nancy Folbre, writing in the New York Times Economix blog, sees a common thread in the risk management mistakes that led to disasters on Wall Street and at the Fukushima Daiichi nuclear plant in Japan:

Early warnings about potential nuclear dangers in Japan and about Wall Street’s propensity for risk-taking without liability were both dismissed as paranoid anticipation of low-probability events. With both disasters, regulatory failures set the stage, and taxpayers will end up paying most of the social costs.

Effective risk management is central to economic efficiency. Yet major players in both crises have insisted that they should not be held accountable for risks they underestimated.

A common fallacy is thinking of the contributing causes of a disaster as unrelated when they are often related. The investment bankers and traders thought the simultaneous collapse of their various investments and derivatives couldn't happen, but in a market meltdown, they all became highly correlated and crashed in unison. Likewise the Japanese engineers imagined that the simultaneous breakdowns of diesel backup generators, cooling valves and pumps was highly unlikely, but the cascading failures were all caused by one event. In each case, the engineers and managers never understood how so many components could fail at once.

This fallacy can lead to a tragic underestimation of risk. Take three components of a complex system. If the chance of failure for any one component is 1 percent, then the odds that all three could fail at the same time would be one in a million—if the three events are unrelated. But if breakdown of one component contributes to, or is correlated to, the failure of another, the chance that all three could fail would be closer to one in a hundred, rather than one in a million.

Floyd Norris, also writing in the Times, points out that managers who underestimate risk can end up taking on even larger risks as a results:

Four years ago, there were fears of a financial meltdown — a term borrowed from the nuclear power industry. Now there are fears of a real meltdown.

Comparing the two events may risk seeming insensitive to the rising human toll in northern Japan, but there are similarities in causation. In each case, overconfidence born of experience led to increased risks once a disaster unfolded.

These errors in risk management are compounded and repeated when the engineers and managers allow themselves to believe that they were overcome by extremely unlikely events. And thus, we hear over and over again that yet another disaster was caused by yet another perfect storm.