Poverty Indicators at Local Level: Definitions, Comparisons in Real Terms and Small Area Estimation Methods

The importance of computing poverty measures at sub-national level is nowadays widely attested. Local poverty indicators are relevant both for a detailed planning of the policy actions against poverty and social exclusion, and for the citizens to evaluate their effects. However, there are still open problems to compute adequate sub-national poverty indicators. They refer to: 1) the definition of poverty lines; 2) the methods for accounting the spatial variation of the cost of living to make comparisons in ‘real terms’ between different areas; 3) the use of Small Area Estimation methods when the sample size is not enough to obtain accurate estimates of the indicators at local level. In this paper, we discuss the issues above by presenting some analyses on the impact of using different poverty lines on the value of the poverty rate for the 20 Italian Regions, which represent a planned domain of study in Italy. Then, we estimate the poverty rate for the 110 Italian Provinces, unplanned domains in Italy, by using specific parametric models and SAE methods. The key results highlight strong differences in the territorial distribution of the poverty rate by using national versus sub-national specific poverty lines. The effect of the heterogeneity of the general spatial price indexes on the poverty rates seems instead less important in comparison with the relevant territorial differences in the cost of housing. Moreover, the different methods of estimation of poverty rates at local level provides interesting first results and indicates the route for further research to improve the methods of estimation of poverty at the sub-regional level.