Modelling economic growth in the UK: An econometric case for disaggregated sectoral analysis

We review recently developed statistical methods for the analysis of disaggregated data and for assessing the adequacy of models developed at different levels of disaggregation. The methods are illustrated in an analysis of UK output growth using disaggregated industrial data over the period 1973 q1-1991 q4. It is found that disaggregated models are superior to their aggregate counterparts in predicting aggregate output growth, primarily owing to the disaggregation of manufacturing sector output. Tests of aggregation bias carried out on measures of the persistence of shocks to aggregate output show the most satisfactory analysis of output growth to be that carried out at the highest level of disaggregation.