How does attrition affect estimates of persistent poverty rates? The case of EU-SILC

Evidence about poverty persistence is an important complement to information about poverty prevalence at a point in time. The persistent at-risk-of-poverty rate is one of the primary indicators of social inclusion, and the only indicator that is derived using samples from the longitudinal component of EU-SILC. Sample drop-out from the longitudinal samples (‘attrition’) reduces sample size thereby decreasing the precision of estimates of persistent poverty indicators, and may be selective and lead to bias. We examine these issues. We show that rates of attrition from the four-year EU-SILC samples used to calculate persistent poverty rates vary substantially across Member States, and there is also substantial cross-national diversity in the characteristics of individuals lost to follow-up. We provide evidence that application of longitudinal weights does not fully account for the effects of attrition, and that different assumptions about the poverty status of attritors lead to wide bounds for estimates of persistent poverty rates for most Member States