THE dismal scientists have a dismal record in predicting recessions. In 1929 the Harvard Economic Society reassured its subscribers days after the crash that: “A severe depression is outside the range of probability."
Despite huge improvements in data and computing power, forecasters remain in the dark.
In a survey in March 2001, 95% of American economists thought there would not be a recession, yet one had already started.
Why are recessions so difficult to forecast? One excuse is that economists, unlike weathermen, do not know if it is hot or cold today because their data are always out of date. They have to forecast not only the future but also the immediate past. A less good reason is that economists have a tendency to run with the pack. Predicting a recession is unpopular (especially if you work for an investment bank), and predicting one prematurely will prove costly to clients. It may also cost you your job.
Forecasts produced by economic models with hundreds of equations are notoriously bad at predicting recessions because they tend to extrapolate the recent past. This leads to big forecasting errors near turning-points, because recessions are caused by abrupt changes in the behaviour of firms and consumers. A more reliable way to spot a coming downturn is to scrutinise indicators that have given warning signals in the past. Financial indicators have the longest lead times, but a gauge that performs well in one period may do badly in another.
Leading indicators that combine several economic and financial measures seem more promising. The index of leading economic indicators (LEI), originally produced by America's Department of Commerce and now by the privately run Conference Board, is a weighted average of indicators such as share prices, interest-rate spreads, consumer confidence and new orders. Unfortunately, the LEI failed to predict any of the past three recessions.
The Economic Cycle Research Institute, a private research group, has been more successful. It was set up by the late Geoffrey Moore, a pioneer of research into business cycles. ECRI believes that turning-points can be systematically predicted. It tracks no fewer than 14 leading indices for different parts of the economy and with different lead times. ECRI was one of the few firms to forecast both of the past two American recessions. Its leading indicators for other economies have also fared well. It successfully forecast recessions in Japan in 1997 and 2001. Encouragingly, as this survey went to press ECRI was saying there was no risk of a double dip in America.