Cointegration and long-horizon forecasting

It is widely believed that imposing cointegration on a forecasting system, if cointegration is, in fact, present, will improve long-horizon forecasts. The authors show that, contrary to this belief, at long horizons nothing is lost by ignoring cointegration when the forecasts are evaluated using standard multivariate forecast accuracy measures. In fact, simple univariate Box-Jenkins forecasts are just as accurate. The authors' results highlight a potentially important deficiency of standard forecast accuracy measures--they fail to value the maintenance of cointegrating relationships among variables--and the authors suggest alternatives that explicitly do so.