Forecasting recessions under the Gramm-Rudman-Hollings law

Abstract

The targeted deficit reductions of the Gramm-Rudman-Hollings (GRH) law are to be temporarily suspended in case of an official determination that real economic growth either (a) has been less than one percent in the two most recent reported quarters, or (b) is projected to be less than zero in any two consecutive quarters out the next six. This amounts to a particular definition of recession. But business cycles are best identified by the consensus of movements in the principal economic aggregates. Not all recessions are associated with real GNP declining or growing less than 1% for two successive quarters. Also, GNP estimates are subject to long sequences of revisions that are often large. We show that, for these reasons, conditioning a suspension of deficit cuts upon specific changes in preliminary data for real GNP involves very long lags in recognizing recessions. The recessions would be largely over before they were identified. We also show that forecasts of real GNP, based on the consensus among groups of professional forecasters, can reduce these lags considerably. This is so despite the fact that early and accurate predictions of business cycle peaks are rare, and false warnings occur.