Theséance of economics

Year-end forecasting is iffy at best, so it pays to curb your exuberance

We’re about to enter the white-heat period of the year when attention shifts to producing and analyzing year-end economic forecasts. By the end of December, scores of financial outfits in Canada and around the world will have cranked out estimates of economic performance in the New Year: Growth will be up so much during 2014, unemployment will be down, inflation will be low, the dollar will be stable, trade will rise — all of it measured to decimal points that imply fine-tuned precision generated by economic models.

It is not news that the forecasting business isn’t all it’s cracked up to be. Despite the annual outlook ritual, everybody knows forecasts have every chance — maybe a 5050 chance — of being wrong, perhaps even spectacularly off the rails. But recent comments and developments paint an even bleaker picture of the predictive capabilities of economic models. The dismal science of economics, it turns out, looks more like a séance than the science it claims to be.

Few economists would go that far, although Jonathan Wright of Johns Hopkins University came mighty close. “I think economics is in about the same state as medicine in about the 18th century,” the professor told The Wall Street Journal in October. In other words, it’s moved beyond leeching, but not much. Wright is an impressive numbercruncher and producer of modelling and analyses, so he has hands-on experience.

Even more experienced is Alan Greenspan, former head of the U.S. Federal Reserve. In his new book, The Map and the Territory: Risk, Human Nature, and the Future of Forecasting, Greenspan essentially blames the 2008 global economic crisis and the current period of stagnation on the failure of economic forecasting. Economic modelling suffered “an existential crisis” in 2008, he says, and it has never recovered.

In the book’s opening chapter, Greenspan nimbly traces the mostly Keynesian backstory to what he calls “the discipline of model-based economic forecasting.” By the 1980s, the Fed seemed to have landed on a model that worked. It combined “elements of Keynesianism, monetarism, and other more recent contributions to economic theory, [and] seemed particularly impressive.”

But leading up to the universally unanticipated crisis of 2008, Greenspan writes, the world experienced a new phenomenon. “Macromodeling inequivocally failed.” The Fed’s model failed, the IMF’s model failed. Three days before the crisis hit in 2008, J.P. Morgan predicted U.S. growth would be accelerating into the first half of 2009. Greenspan asks: “What went wrong?”

His answer is far from satisfying. The models failed because they were not programmed to take into account the craziness of people in the markets — financial players, investors and consumers. Forecasters had never thought it necessary to account for the “wild and even deranged mood swings that are uncoupled from any underlying rational basis.”

Greenspan now believes these mood swings can be measured and “made an integrated part of the economic forecasting process.” So far, though, no such integration has taken place. It’s far from clear, moreover, that Greenspan’s plan to map the territory of behavior offers a real solution to the modelling problem.

In an upcoming paper in the Journal of Economic Literature, Jonathan Wright concludes that massive changes in the economy have left existing business cycle models in the dust. “We have a long way to go to achieve scientific consensus,” he told me in an interview. If his medicine analogy is right, that long way requires a 100- year leap.

The jacket cover of The Map and the Territory describes it as, “A master class in the alchemy of economic decision making.” Alchemy, 18th century medicine, séance — whatever the analogy — the message is clear: When you hear the year-end forecasts, curb your exuberance.