EU-SILC and the potential for synthetic panel estimates

Abstract

In the absence of panel data, researchers have devised alternative methods for estimating synthetic poverty dynamics using repeated cross section surveys. These methods are not only salient in the absence of panel data, but also in contexts where there are concerns over the quality of panel data and/or the panel data are of insufficient length to analyse medium- to long-term mobility trends. Both of these issues afflict the longitudinal element of the European Survey on Income and Living Conditions (EU-SILC) (Hérault and Jenkins, J Econ Inequ 17(1):51–76, 2019). Using the longitudinal element of EU-SILC, this paper assesses the accuracy of the synthetic panel approach put forth by Dang and Lanjouw (2021). For most conventional poverty lines, the DL approach is found to be highly accurate when the true \(\rho \) is known. Similar to Hérault and Jenkins (J Econ Inequ 17(1):51-76, 2019) the pseudo-panel approach for estimating \(\rho \) is found to be highly sensitive to cohort definition. The longitudinal element of EU-SILC, however, offers a unique route for overcoming this shortcoming.