Empirical Analysis of Unemployment Insurance Take-up: the Importance of Sampling Scheme and Censoring
1, François Fontaine2, David Margolis3
1Aarhus School of Business, Denmark, 2Université de Strasbourg, BETA-CNRS, France, 3CES, CNRS, CREST, France
This paper summarizes the empirical literature on unemployment insurance take-up and aims at explaining the diversity of estimated take-up rates. We highlight two important specificities of unemployment benefits take-up which may have dramatic consequences on the validity of the estimated take-up rates if ignored: UI claiming is a dynamic process and display endogenous links with unemployment duration. Using surveys and administrative data, we show that take-up rates are highly sensitive to sampling schemes. Stock sampling tends to overestimate take-up rates because it oversample longer unemployment spells. Moreover, endogenous right-censoring is likely to underestimate non take-up rates if not taken into account. Lastly, we find that survey-based and administrative-based data give comparable estimates for the determinants of UI take-up but they are biased if endogenous right-censoring is ignored.
View full paper