In Defense of Forecasting

Current actions often beg for forecasts of things that we will know more about later. But making projections based on imperfect information can often result in imperfect — and sometimes, downright wrong — forecasts, a fact that has not made economists terribly popular during the current financial crisis.

Some argue that it is irresponsible for economists to continue making forecasts. But are these critics right? Should policy makers rely on forecasts using less than ideal data, or should they wait and see? These questions on the ethics, and usefulness, of economic projections arise in all sorts of policy matters, from climate change to health care costs to fiscal stimulus to housing prices. In my work at the World Bank, the question arises when considering the impact of the global financial crisis this year and next, including on the world's poorest people, and how to deal with anticipated problems.

It will probably be two years before we will have household surveys with adequate coverage of developing countries even for 2009. And when the surveys become available, assessing the impact requires analytic work to estimate what would have been observed in the absence of the crisis. The crisis will probably (hopefully) be over before we can have a high degree of confidence about its impact on the world's poor using survey data embracing the crisis period. Yet there are perils of inaction, and of poorly informed actions, including lasting consequences, such as when children are underfed or taken out of school by parents hit by the crisis and with no safety net in place to help.

Some observers have questioned whether we should bother making forecasts with less than ideal data. For example, in a recent interview with The New York Times, Prof. William Easterly of New York University called the World Bank's effort to estimate the impact of the crisis on the world's poorest "a political exercise (rather) than an exercise in real analysis." Professor Easterly's main concern was the uncertainty surrounding these estimates.

There is no denying that forecasts reflect political demands. But that does not mean that responding to the demands is purely "a political exercise." In fact the lags in receiving good, accurate numbers call for more analysis, not less. Yes, that means we have to make assumptions in our forecasts, as in virtually all economic analysis. When we make assumptions in the absence of data, we must disclose them, and test them as far as possible.