The Challenge Of Low Probability, High Impact Risks

BP is looking for ways to better manage "low-probability, high-impact" risks like the Gulf of Mexico oil spill, the company's CEO tells the FT.

But those kind of Black Swannish risks are, by their nature, remarkably difficult to manage.

Sure, BP and other oil companies can cut fewer corners. And they can prevent the exact failures from the BP spill from recurring. But what about all the other low-probability, high-impact risks they face?

It's tempting to rely on some combination of market forces and government regulation to push companies to better manage risks.

But, as the Harvard economist Ken Rogoff wrote this week, "Economics teaches us that when there is huge uncertainty about catastrophic risks, it is dangerous to rely too much on the price mechanism to get incentives right."

In other words: If both the magnitude of the risk and the magnitude of its impact are essentially unknown how can the market appropriately price those risks?

For that matter, Rogoff says, it's not clear how "to adapt regulation over time to complex systems with constantly evolving risks." As a result, he argues, "we may be doomed to a world of regulation that perpetually overshoots or undershoots its goals."

Regulation of offshore drilling will certainly get a lot tighter because of the spill. The industry will face more scrutiny from government watchdogs, and higher legal liability for oil spills.

But the problem of low-probability, high-impact risks goes far beyond oil. And, as the NYT's David Leonhardt writes, "it would be foolish to think that the only risks we are still underestimating are the ones that have suddenly become salient."

He cites as possible examples the risk that investors will grow wary of lending money to the U.S. government, or the risk that climate change will be more drastic than expected, leading to flooded cities.

Why stop there? You could add to the list a nuclear blast, a deadly new virus or just about any other radically destabilizing but fundamentally unquantifiable risk.