Modelling and forecasting regional service employment in Great Britain

This paper attempts to forecast employment growth in the service sector, on a regional basis across Great Britain. Four forecasting models are used: Time-varying parameters, regression, state space, and ARIMA. The empirical results suggest that the regional growth rates in service employment displayed a steady convergence during the 1980s, but deconvergence in the 1990s, and that the composite leading indicator exerted a strong influence on regional service employment. In terms of out-of-sample forecast accuracy, none of the models dominates the others, though the state space model has a slight edge on average, particularly in the case of multiperiod predictions. In addition, the relative forecasting performance of a professional structural model of the GB regions is compared to the estimated models. (C) 1999 Elsevier Science B.V. All rights reserved.