Shocks show limits of financial models

By Matt Scroggins

ORLANDO, Fla.—Low-frequency, high-severity risks—the kind that keep insurance executives awake at night—are a significant threat in the financial sector because of an over-reliance on inherently limited mathematical models, according to one risk scholar.

Speaking to a room full of insurance industry executives Monday morning at the Property Casualty Insurers Assn. of America meeting in Orlando, Nassim N. Taleb said that people in insurance “know what they are talking about when they talk about risk.” Those in the finance world, however “don't know much about risk,” said Mr. Taleb, author of the best-selling book “The Black Swan: The Impact of the Highly Improbable.”

He recounted information from a Wall Street Journal article, in which an employee of a so-called quant fund—one that bases investment decisions on quantitative analysis—explained that in August 2007, such funds experienced three straight days of what had been considered 1-in-10,000 year events.

The central problem, Mr. Taleb said, is that “the more remote the event, the less data I have and the harder it is to estimate its probabilities from experience, from empirical data, from observations.” That necessitates the use of models to project the probability of adverse events, he said. But the underlying theories those models are based upon “can be wrong. And they were wrong,” Mr. Taleb said.

“The world is getting more complicated and, as you will see, much harder to forecast.”

He contrasted two domains of probability and understanding. One, he termed “Mediocristan,” where events are predictable and follow statistical rules such as the law of large numbers. In the other domain, “Extremistan,” extremes—known as fat-tail events—dominate, Mr. Taleb said.

“Welcome to Extremistan,” he said, noting that the modern economic world “is dominated by extremes.”

One key force that adds complexity—and volatility—to the economic realm is technology, he said. Because of technology, “everything is over-optimized,” Mr. Taleb said, which increases what he terms the “fragility” of systems.

He noted, for example, that Internet technology has greatly accelerated the pace of communication and exacerbated movement in financial markets.

“Iceland was bankrupted by BlackBerrys,” he quipped, noting that such technologies allow catastrophic financial developments, such as runs on banks, to unfold much more rapidly than they could in the past.

He noted that large financial institutions make money by using assets to bet against extreme events. But because those bets are made using limited, flawed statistical models, they occasionally “blow up,” resulting in huge losses, which the companies then characterize as “unexpected.”

“Stop exposure to tail events,” he said. “Banks did not lose money on their regular business.”

He then discussed the role of insurance in boosting the robustness—or ability to withstand shocks—of business activities.

Optimization and insurance “are antagonists,” he said. “Something over-optimized is fragile.”

However, “Mother Nature likes inefficiencies, because they ensure your survival,” he said, noting for example that the human body has “spare parts,” such as two eyes and two lungs.

“Redundancy is insurance,” he said.