Local heterogeneities in population growth and decline. A spatial analysis for Italian municipalities.

Spatially unequal demographic dynamics lead to a progressive fragility of a territory and its socio-economic system. In Italy, municipalities in demographic malaise tend to be increasingly small in size and peripheral in location, and their local spatial aggregation increased over time. A spatial approach is here proposed to investigate the dynamics across time and space of the population variations in Italian municipalities. Global and local spatial autocorrelation analysis and several models of regression were run using as study variable the average growth rates at municipality level. The spatial autocorrelation of the study variable is quite high and stable over time. The regression results show that spatial models (SAM and SAR) outperform the non-spatial model (OLS) and that SAR is the best model. The results also underline that the variation of population is significantly affected by its values in the neighbouring municipalities, confirming the spatial nature of the phenomenon. The presence of schools in the municipality emerges as a key factor for the increase/decrease of the population. Moreover, the decomposition of the effects into direct and indirect effects shows that all the independent variables produce their effects almost 70% directly and 30% indirectly