Beschreibung

"This paper develops a set of Neural Network (NN) models to compute short-term forecasts of regional employment patterns in Germany. NNs are statistical tools based on learning algorithms that distribute computation on a set of units that work in parallel. They are therefore able to process large amounts of data. NNs are enjoying increasing interest in several fields, because of their effectiveness in interpolating data when the functional relationship between dependent and independent variables is not clearly specified. The paper compares two NN methodologies. First, it uses NNs to forecast regional employment in both West and East Germany. Each model carried out computes single estimates of employment growth rates for each German district, with a two-year forecasting range. Additional forecasts are then computed by combining the NN methodology with shift-share analysis. Since shift-share analysis aims at explaining variations observed among the districts, its results are used as further explanatory variables in the NN models. The data set used in our experiments consists of a panel of 439 German districts, but because of different size and time horizons of the data, the forecasts for West and East Germany are computed separately. The out-of-sample forecasting ability of the models is evaluated by means of several statistical indicators." (Author's abstract, IAB-Doku) ((en))