The ability to assess risk and uncertainty is critical for investment banks and businesses. While some may advocate the use of complex models, INSEAD Dean of Faculty and Professor of Decision Sciences, Anil Gaba, believes that if you’re looking to forecast risk, you’d do well to keep it simple.
He’s on a search for convenient rules of thumb – or what he calls ‘frugal parsimonious heuristics’—to help managers and investors make better decisions more easily.
There’s been enormous theoretical progress in the past fifty or sixty years in understanding how people should make decisions in business, politics and economics.
The Black-Scholes equation revolutionized the world of derivatives. Cheap computers have empowered model builders and statisticians.
The gap between theory and practice
But Gaba, currently the Orpar chaired Professor of Risk Management at INSEAD, says despite all that progress, there is a growing gap between theory and practice.
“You take young people, you train them very hard in these theories—public policy makers, investment bankers, and general managers,” Gaba says. “But when they go out into the real world, they still tend to rely a lot on their gut feeling.” And that’s not good, because people fall into predictable psychological traps when they use instinct to make decisions.
“They always underestimate the amount of uncertainty that may exist in a situation. The stock market is a perfect example.” Investors are often overconfident in their abilities to pick winners. They set their low guesses too high and their high guesses too low. They don’t diversify enough—and they buy and sell too often, taking profits too early, and eat into their gains by running up unnecessary trading costs.
Such overconfidence is rife. “We also see this in weather forecasting, with medical doctors, in personal judgments and so on,” Gaba says.
The problem is that what may make sense in a university lecture theatre may be half-forgotten by the time the theory might need to be applied many years later in the messy reality of business.
“If you give a very complicated formula to a practitioner, it’s almost like a black box to them. They tend to avoid using it. And that’s not surprising, as sometimes they don’t understand the mechanics of the formula. Sometimes they don’t use it because they don’t have enough information to use the formula.”
Even worse, complicated complex models don’t actually predict real-life events very well, according to research by one of Gaba’s colleagues at INSEAD, Spyros Makridakis. “[He’s] spent his entire career looking at forecasting.” Gaba says. “And the main result of his work has been basically that simple methods do at least as well as some of most complicated methods”— a very disappointing result for people who spend a lot of time developing complicated models.
“But at the same time it’s very good news for practitioners because it’s easier for them to use the simpler models.”
Take a common task, Gaba says. As a manager, you need to take a stand on what will happen to a particular market—how much will demand grow or shrink next year.
“There are five people in the management team. Each one has an opinion. These opinions might be different. The question is how do you combine these opinions?
“You can use an extremely complicated model—Bayesian statistics, different weights for different experts, you revise it, and so on. But research suggests the most effective solution might be far simpler.”
Simple averaging works well
“Simple averaging of subjective opinions can at times do at least as well as some of the most sophisticated normative models we've come up with over the past two hundred years.”
“So taking four or five opinions and taking an average is a very parsimonious way of dealing with the problem.”
For Gaba and his colleagues, the research now is about finding out which simple rules work best under what conditions by building on existing understanding of psychological traps. He’s also working on how people from different cultures assess risk differently and the way it affects their decisions. But there’s still a lot of work to do. As the saying goes, “It’s tough to make predictions—especially about the future.”