How to bet on boom or bust

This collection of essays by distinguished experts on economic forecasting, aimed at the interested layperson, is from the proceedings of a British Association for the Advancement of Science meeting held in Sheffield in 1999. It contains a great deal to interest both lay and professional readers who wish to understand how economic forecasting works.

The 11 chapters are a mix of primers on the technicalities of forecast construction and evaluation, mainly from the academic contributors, and blow-by-blow accounts of forecasting in practice, by practitioners and commentators. In addition to a joint introduction and epilogue, the editors each contribute a background chapter to answer basic questions such as "What is a forecast?" and "What is forecast uncertainty?". David Hendry's contribution includes an interesting aside on the problem of long-range forecasting, considering the best way to capture a growth trend by looking at UK output since 1700, over 50-year horizons. Paul Turner's neat contribution explains, very clearly, the fundamentals of structural modelling and what it is for. Clive Granger bravely takes on the task of explaining forecast evaluation and contributes most of the equations.

Explaining statistical concepts to students is a tricky task, but the readership of this book, which will include sharp and informed professional and business people, presents, if anything, a tougher challenge. All the contributors have made a valiant effort to explain their subject in laymen's terms, though sometimes leaning too far backwards to keep the discussion familiar.

Although they do not eschew technicalities, the narrative chapters are often the most successful in conveying the flavour of the forecasting enterprise. Neal Hatch's account of the working methods of the Bank of England's Monetary Policy Committee cannot fail to fascinate, given the glamour or, in some quarters, notoriety, that adheres to that body. Lord (Terence) Burns recalls the ups and downs of Treasury forecasting over the Thatcher years, and Ray Barrell gives a similar retrospective of attempts to forecast world economic trends at the National Institute for Economic and Social Research. Diane Coyle provides a revealing and entertaining journalist's view of the comparative performance of published forecasts.

The message that becomes progressively clearer is that the big problem is to predict the turning points of the economic cycle, from boom to recession and back again. As long as growth stays "on course", we can evidently extrapolate its path with some success. It is the timing of the changes that presents the difficulty. Thus, everyone knows that booms and bubbles in the stock market, or the housing market, must come to an end eventually; but it seems that one cannot predict when, as evidenced by the many unfortunates who lose their shirts in these episodes. The most insightful of the academic contributions is that of Denise Osborn and her co-authors, who explain the turning-points problem as a nonlinear phenomenon. In other words, the intervention of a random event can make the economy switch catastrophically from one path to another. Their attempt to identify a leading indicator of the switch mechanism makes a fascinating tale that even the non-technical will be able to appreciate. This is path-breaking economic research in action.

None of the authors spends much time on the most subtle, yet obvious, difficulty that economic forecasters face. Weather and economic forecasts are notoriously imprecise, and one might think from some of these contributions that the reasons were similar - that both systems in question, the atmosphere and the economy, are very large and very complicated. This is true, but there is a crucial difference. An economic forecast, if publicly known and believed, is part of the system being forecast. Indeed, forecasting the economic cycle is arguably impossible, for this reason. If the date of the next downturn could be predicted, like an Atlantic depression, we would all sell our shares ahead of time, and so bring it forward. Ergo , it cannot be predicted.

There is an old joke about two economists walking down a street. One says: "Look, there's a £50 note lying on the pavement." The other says: "Impossible dear chap, you imagined it. If it were really there, someone would have picked it up by now."

James Davidson is professor of econometrics, Cardiff Business School, Cardiff University.