Measuring Income Inequality and Poverty at the Regional Level in OECD Countries [E-Book] / Mario Piacentini
Piacentini, Mario.
Paris : OECD Publishing, 2014
63 p. ; 21 x 29.7cm.
OECD Statistics Working Papers ; 2014/03
Social Issues/Migration/Health
Urban, Rural and Regional Development
Full Text
The extent to which income inequality and poverty vary within countries across different regions is very relevant for policy decisions and monitoring. However, sub-national measures are scarce, given the complexity of producing indicators at the regional level from the available data and the methodological issues related to cross-countries comparability. This paper presents a set of indicators of income inequality and poverty across and within regions for 28 OECD countries. These indicators were produced through a new household-level data collection based on internationally harmonized income definitions undertaken as part of the OECD project on "Measuring regional and local well-being for policymaking". The data were collected at the OECD TL2 territorial level, corresponding to NUTS2 regions in Europe and to large administrative subdivisions (e.g. States in Mexico and Unites States) for non-European countries. These estimates confirm that there are significant variations in levels of income inequality within countries, and that regional breakdowns are useful for understanding sources and patterns of income disparities and poverty. For most of the countries relying on survey data for measuring income distribution, standard cross-sectional indicators of income inequality and relative poverty at this regional level are estimated with low precision in the smallest regions due to small samples. This has two main implications for data producers and analysts. First, systematic reporting of confidence intervals is needed to make meaningful comparisons of inequality levels across regions and with respect to the national averages. Second, averaged measures for multiple years or small area estimation methods should be considered as means for obtaining more robust measures. The issues related to the estimation of standard errors for three-year averages in rotational panel surveys and to the definition of the computational sampling structure for sub-national estimates are discussed in the paper.