Cointegration and Long-Horizon Forecasting
NBER Technical Working Paper No. 217
We consider the forecasting of cointegrated variables, and we show that at long horizons" nothing is lost by ignoring cointegration when forecasts are evaluated using standard multivariate" forecast accuracy measures. In fact, simple univariate Box-Jenkins forecasts are just as accurate. " Our results highlight a potentially important deficiency of standard forecast accuracy" measures they fail to value the maintenance of cointegrating relationships among" variables and we suggest alternatives that explicitly do so.
Published: Journal of Business and Economic Statistics, Vol. 16 (1998): 450-458.