| Article ID: | iaor19932484 |
| Country: | United Kingdom |
| Volume: | 12 |
| Issue: | 1 |
| Start Page Number: | 63 |
| End Page Number: | 67 |
| Publication Date: | Jan 1993 |
| Journal: | International Journal of Forecasting |
| Authors: | Coulson N.E., Robins R.P. |
The authors examine the implications of allowing lags into forecast combination regression, thereby extending previous models. The practical conclusion is that lagged dependent variables, but not lagged forecasts, improve forecast combination procedures. Also, improvements are obtained when nonstationarity of the data is recognized.