This paper compares the properties of a structural model-the London Business School model of the U.K. economy-with a time series model. Information provided by this type of comparison is a useful diagnostic tool for detecting types of model misspecification. This is a more meaningful way of proceeding rather than attempting to establish the superiority of one type of model over another. In lieu of a better structural model, the effects of inappropriate dynamic specification can be reduced by combining the forecasts of both the structural and time series models. For many variables considered here these provide more accurate forecasts than each of the model types alone. KEY WORDS Structural models Time series models Composite predictors Comparative methodsdata aggregate, disaggregate Comparative methods-macro models Comparative methods ; ARI MA, causal (simultaneous system) Ex ante Combining forecasts--causal, time series Time series-ARIMA This paper compares the properties of a structural econometric model (SEM) with a time series model.
There is an extensive literature on comparing the predictive performance of SEMs with time series models. The early findings suggested that time series models performed better than the SEMs. This has provided an important stimulus to the development of econometric techniques which give greater recognition to the dynamic structure of error processes. The widespread use of these more sophisticated techniques in the specification of macroeconomic models suggests that the relative merits of SEMs and time series models need to be re-examined.
The predictive performance of the London Business School model of the U.K. economy-a large non-linear structural model-is compared with a time series model. Both expost and ex ante predictions are generated for periods from one to eight quarters ahead. For many variables the structural model outperforms the time series model though for some economic variables the time series model is better. This suggests some possible areas of misspecification in the structural model.
The proper response in these circumstances is to respecify the model. But this can be an expensive, time-consuming activity and there is no guarantee that a better model specification can be found. In the interim it is possible to use composite predictors which combine the time series