Teams of foxes make the best forecasts, but expert hedgehogs can help


A bane of this economist’s life is the belief that economics is clairvoyance. I should, according to this view, be offering prognostications of what gross domestic product growth will be this year and when the central bank will raise interest rates.

Superforecasting , by Philip Tetlock and Dan Gardner, is one of the most interesting business and finance books published in 2015. Tetlock has for decades collected specific predictions of events from a wide range of forecasters.

In his earlier book, Expert Political Judgment, he demonstrated to little surprise that forecasters were not very good. More surprising was his identification of the characteristics of good and bad forecasters.

Tetlock employs a distinction, credited to the Greek poet, Archilochus, but popularised by the philosopher, Isaiah Berlin, between hedgehogs and foxes. Hedgehogs know one big thing; they have an all-encompassing world view and discover facts that confirm what they already know to be true. Foxes know many little things; they are eclectic in their sources of information and nuanced in their judgments. Hedgehogs command more public attention but foxes make better forecasters.

Harry Truman, the former US president, (perhaps apocryphally) sought a one-handed economist who would not say “on the one hand” and “on the other”. The two-handed approach corresponds to the reality of most complex issues. Yet modern business people, politicians and media seek the Truman type and find it in hedgehogs. To grab attention and build a reputation, it is more important to be unequivocal than to be right.

Tetlock’s “superforecasters” are the winners of a contest he organised in which individuals and teams were asked to make predictions about specific events. They are indisputably nerds. They are diligent in pursuing sources of information, and ready to revise their predictions as more data become available. They do not appear on Fox News; nor do they advise Bernie Sanders, the socialist US Democratic presidential hopeful. They do not earn large speaking fees from business conferences.

Tetlock found, however, not only that his protégés tended to make good predictions; but also that they became better with practice and that teams of superforecasters easily outperformed teams of selected experts.

It would be good to report that their services are widely sought by governments, businesses and financial institutions. But they are not. And perhaps for good reason. The questions Tetlock poses are, necessarily, highly specific.

Otherwise it would not be possible to test the validity of the responses. The question of how many Syrian refugees will land in Europe in 2016 is at best a proxy for what we really want to know. The underlying questions are looser and open-ended. How will the political situation in Syria and Iraq evolve? How will Europe respond to the refugee crisis?

These bigger questions cannot so easily be approached by the systemisation produced by the superforecasters, and may lend themselves to the hedgehogs’ narrative approach. What was the right answer on January 1 1989 to the question “will the Berlin Wall be pulled down in 1989?” A shrewd commentator would have said (though few did) something like “almost certainly the Wall will stand but you should understand the potentially destructive forces undermining the Soviet engine and the East German state”. That type of response combines probabilistic and narrative thinking.

But people long for certainties, though they know they cannot have them. I have learnt that few really want answers when they ask me to predict GDP growth or advise whether interest rates will rise in the third quarter. It is usually easy to move the subject on to something more interesting than macroeconomic forecasting.


This article was first published in the Financial Times on January 6th, 2016.

Print Friendly