Regional patterns of Romanian emigration. A Geographically Weighted Regression Model

Analyses of emigration flows and their factors of influence are usually conducted at aggregate country level, thus ignoring the existence of large regional inequalities. Since regions are very diverse, the key drivers of emigration are likely to change in space and act with variable intensity, creating different regional patterns. The presence of broad regional inequalities and the concentration of high emigration flows in a few regions, as in the Romanian case, require specific investigation tools, such as the geographically weighted regression (GWR) which is able to highlight the spatial variations in regression parameters. This method provides coefficient estimaties that change from one region to another, according to spatially defined weights; therefore it lets the model to fluctuate territorially, capturing the real spatial patterns of the phenomenon under investigation. Starting from these considerations we analysed the main determinants of the emigration flows in Romania with a GWR model, using new regional migration data issued by the National Institute of Statistics following the last census. We explored the counties’ vulnerability to emigration in relation to several important influence factors: local development level, earnings, employment, and education. Our findings provide some new insights into this topic, revealing sizeable territorial variation in emigration determinants and pointing to the counties that are more strongly afected by certain factors of influence. These results represent an improvement compared to those provided by the traditional global models, which are unable to distinguish the spatial variation in estimations. The county-specific estimates resulting from the GWR model represent useful information for shaping appropriate migration policies, specifically tailored to meet the local needs.