Abstract

Purpose of the article: Most regional macroeconomic processes may not be adequately analyzed without accounting for their spatial nature: regional distances, interactions between neighbors, spill-over effects and interdependencies. This contribution focuses on various factors ruling unemployment dynamics in the Visegrad Group countries and their major economic partners: Germany and Austria. The analysis is performed at the NUTS2 level. Methodology/methods: Spatial econometrics is a unique tool for a broad range of quantitative analyses and evaluations. Spatial econometric models are based on geo-coded data. Spatial econometrics and regional competitiveness paradigms are combined into different types of regression model specifications describing unemployment dynamics. Alternative spatial structures are used for verification of stability in estimated model properties. Scientific aim: We aim to provide a detailed empirical evaluation of spatially determined factors of regional unemployment dynamics, along with insight into the robustness of such approach. Both conceptually and parametrically varying neighbor definitions are used to provide evidence for model evaluation. Findings: We find strong positive spatial dependence patterns in the estimated models, robust against varying neighborhood definitions. Our results strongly support the importance of regional and potentially cross-border cooperation in macroeconomic policies addressing unemployment. The estimated models also underline the importance of using spatial models, by pointing out the bias in OLS-estimated models. Conclusions and limits: Spatial approach to econometric analysis provides important insight and robustness to a broad range of unemployment analyses that may be carried out using regional data. At the same time, it should be noted that this article focuses mostly on the spatial and stability aspects of model estimation, while leaving out other interesting topics such as spill-over effects calculations as based on estimated models. Also, estimations provided in this article might benefit from spatial panel data-based methods - once data availability issues are sorted.