Developments in Regional Labour Markets in Germany: a Comparative Analysis of the Forecasting Performance of Competing Statistical Models

The aim of this paper is to forecast regional employment patterns in West German regions. After a brief exposition of key labour market issues, we propose Artificial Neural Network (ANN) techniques as a new tool to compute reliable short-term employment forecasts at a regional level. A variety of ANN models are developed and compared. Comparison with methods commonly applied to panel data, such as GMM (Generalised Method of Moments), confirms the ability of ANNs to capture complex data structures in a multi-regional context.