Political and economic forecasting: Accuracy doesn’t matter

McLaughlin_Group.jpg

I learnt during the new economy bubble that people preferred to be told they were right than to be told what would happen. … [P]eople who had not wished to be told they were talking nonsense before the bubble burst did not wish to be told they had been talking nonsense after the bubble burst either. Indeed they did not recall that they had been talking nonsense. Either they had known that it was a bubble all the time or they had been victims of events that could not have been predicted. Frequently the same individual would make both claims. And the same people would make the same false assertions when the credit bubble burst.

That’s John Kay, writing about punditry and its seeming imperviousness to such small matters as being consistently wrong about what’s likely to happen.

What wins acclaim, Kay contends, is one’s television manner, not the accuracy of one’s forecasts. If you look good and sound authoritative, that’s really what matters. And Kay appropriately cites the work of Philip Tetlock, who has found, to oversimplify a bit, that the better known the pundit, the less accurate his or her forecasts.

Several years ago, Jarol Manheim, Susannah Pierce, and I kept tabs on the predictions that self-styled Washington insiders offered on the popular Sunday morning shout show, “The McLaughlin Group” (pictured above). What we found bears out Kay’s point. Most of the pundits’ predictions were either too unclear to be subjected to a truth test or turned out to be simply incorrect—none of which deterred these “inside dopes” from confidently and loudly offering a new batch of predictions week in and week out. In weather forecasting, we realize that achieving accuracy in the short run is impossible, so the local wather show typically has more to do with the personality of the weather(wo)man than with providing a clear and accurate forecast of whether it’s going to rain tomorrow. The same thing holds for political and economic prognostication, though we prefer to pretend otherwise.

[Hat tip to Phil Young]

Trackbacks/Pingbacks

  1. Pundit Accountability at Jacob Christensen - September 21, 2011

    […] shown that pundits are terrible forecasters – but then again the quality of forecasts may never have played too big a role in the assessment of talking […]