Abstract
Recently numerous studies have maintained that post-1990 data no longer supports a long-run M2 demand relationship. Suggestive of the failure, several authors have shown that the errors associated with forecasting real income fluctuations increase once the M2 demand vector error-correction information is introduced. However, the present study finds that these results are an outgrowth of the Gulf War and that once the sample is extended the cointegrating information significantly improves the income forecasts.