An Application of Spatial Econometrics to the Vulnerability Index for Italy

This research develops a Social Vulnerability Index (VI) using nine socio-economic indicators that inuence regional capacity to cope with natural hazards, enabling spatial identi cation of areas with varying resilience. In addition, we examined spatial heterogeneity and spatial dependence across and within Italian regions, evaluating the impact of three socio-economic variables police-recorded robberies, life expectancy at birth, and households with ve or more members on vulnerability for the years 2020- 2022. To complement, we analyze spatial income inequalities over the same period. Our ndings reveal that vulnerability indices are consistently lower in northern Italian regions and higher in south ern areas and islands. Spatial autocorrelation analysis demonstrates moderate clustering tendencies among north ern regions. Cluster analysis reveals a signi cant relationship between network connectivity and vulnerability levels. For 2020, the Geographically Weighted Regression (GWR) demonstrated superior performance to the Spatial Er ror Model (SEM), indicating that while robberies are more prevalent in northern regions, they show no signi cant relationship with vulnerability in those areas. Conversely, this association proves signi cant in southern regions. Similar patterns emerged for the life expectancy at birth and households with ve or more members. For 2021 and 2022, we explored other approaches and observed that the Fay-Herriot (FH) model seemed to present less uncertainty on posterior results for the parameters, as indicated a low value of the coe cient of variation (CV). Contrarily, the Conditional Autoregressive (CAR) model showed to be a better t for our data by the Deviance information criterion (DIC) measure. Complementary, the 20:20 ratio indicates that the wealthiest 20% of the population earn 6.57 times more than the poorest 20% of the population in 2021, with similar ratios in other years. Gini coe cient calculations con rm signi cant income inequality. However, spatial analysis reveals no signi cant geographic structure inuencing these income disparities.