Risk 2002 USA: World Bank’s Perlin questions validity of risk model assumptions

The World Bank’s chief financial officer, Gary Perlin, questioned whether some of the broad assumptions used in risk modelling were still valid in the current economic climate today during a keynote address at Risk magazine’s Risk 2002 USA conference in Boston.

Perlin said risk managers are currently operating in the most challenging of environments, adding that it is a time they may have planned for, but possibly one that they most feared. “When stress-testing, we often use scenarios that we barely imagine could happen in reality. Now we are actually working through them,” Perlin said in his keynote address.

Events such as Enron’s collapse, US corporate credit events and Argentina’s sovereign default, have called into question some aspects of current thinking on risk management, according to Perlin. “We are now questioning whether some of the broad assumptions underlying our models are still valid.”

Other open questions included whether or not changes in risk tolerance among investors, and generally higher volatility in financial markets, are part of a structural change, or are simply changing with the economic cycle, he said.

Meanwhile, Harry Markowitz – pioneer of modern portfolio theory – took a different tack, using his keynote address to present his latest research on portfolio management. Using utility theory, Markowitz demonstrated how a heuristic utility function can produce comparable risk-reward characteristics to that of an optimal solution calculated via simulation for a simple small portfolio.

Speaking of the case of a large portfolio including a dozen asset classes and a thousand individual securities, Markowitz said: “If all the world’s current computing power were to have started trying to find the optimal solution - including all state variables and time periods - for a very large portfolio back at the time of the big bang, it would still be calculating now.” Whether or not the similarity between the heuristic and optimal solutions found for a simple portfolio holds for large portfolios is not yet known.