Threshold sensitivity of income poverty measures for EU social targets

In this chapter, we have examined poverty evaluation in the EU by focusing on two critical features of head count measures: the arbitrariness of the choice of the poverty threshold and the insensitivity to the seriousness of the poverty condition. We have compared empirical results for the AROP rate, a constituent of the official EU poverty target, with those for 10 (in fact, 9, as the primal and dual measures coincide for k = 1) alternative indices belonging to a class of threshold-free distributionally sensitive measures characterised by Aaberge and Atkinson (2013). We have reported estimates for all the 28 Member States in 2018 based on EU-SILC data. Our main findings can be summarised as follows. First, the correlation of country rankings based on the different selected measures is high and statistically significant, although movements by one to three positions up or down are frequent. The rankings are more robust for dual than for primal measures. Second, accounting for income distribution below the poverty threshold impinges on the evaluation of poverty levels: the extent to which measured poverty is higher in one country than in another depends heavily on the choice of the poverty index. Third, the optimal allocation of an anti-poverty budget may change considerably if a different poverty index is chosen as a target, revealing their ethical views underlying the choice.