Search

In this paper, we present a review of various computational experiments concerning neural network (NN) models developed for regional employment forecasting. NNs are nowadays widely used in several fields because of their flexible specification structure. A series of NN experiments is presented in the paper, using two data sets on German NUTS-3 districts. Individual forecasts are computed by our models for each district in order to answer the following question: How relevant are NN parameters in comparison to NN structure? Comprehensive testing of these parameters is limited in the literature. Building on different specifications of NN models-in terms of explanatory variables and NN structures-we propose a systematic choice of NN learning parameters and internal functions by means of a sensitivity analysis. Our results show that different combinations of NN parameters provide significantly varying statistical performance and forecasting power. Finally, we note that the sets of parameters chosen for a given model specification cannot be light-heartedly applied to different or more complex models. [ABSTRACT FROM AUTHOR]

Because of heterogeneity across regions, economic policy measures are increasingly targeted at the regional level and, therefore, require regional forecasts. The data available to compute regional forecasts are usually a pseudo panel of a limited number of observations over time and a large number of regions strongly interacting with each other. Traditional time-series techniques applied to distinct time series of regional data are probably a suboptimal forecasting strategy. Although both linear and nonlinear models have been applied and evaluated to forecast socioeconomic variables, spatial interactions among regions are often ignored. This article evaluates the ability of spatial error and spatial lag models to correct for misspecifications due to neglected spatial autocorrelation in the data. The empirical application on short-term forecasts of employment in 326 West German regions shows that the superimposed spatial structure that is required for the estimation of spatial models improves the forecasting performance of nonspatial models. [ABSTRACT FROM AUTHOR]