Abstract

Although the German economy managed the last economic recession comparatively well, it suffers from a high and stagnating long-term unemployment and benefit dependency. The German system of basic income support for needy individuals capable of working (“unemployment benefit II”) aims at bringing the jobless back into work through job-search requirements and activation measures. Despite this focus on employment integration, the benefit system is designed as a means-tested minimum income benefit not only for unemployed individuals but for all individuals capable of working and their households whose household income is insufficient to meet a minimum standard of living. Therefore, important recipient groups are not only unemployed individuals but also low-wage or part-time workers, single parents, larger families and older individuals with health restrictions.
Given these characteristics of the benefit system, exits from benefit receipt do not have to be rooted in the individual labour market behaviour, but can also lie in the household context, i.e., in changes in its composition or employment situation of household members. Therefore, the analysis of individual benefit receipt cannot be restricted to employment-related events, but also has to consider other types of exits. Previous evidence has shown that a substantial share of exits are not stable: 29% of the individuals leaving welfare receipt in 2012 returned within three months. This recidivism is likely to differ by exit route. In this paper, we analyse individual benefit recipients’ chances to exit from benefit receipt and recidivism with special attention paid to the routes that take recipients out of benefit receipt (with or without own employment). We use rich administrative data on benefit recipients from the German Federal Employment Agency (“Sample of Integrated Welfare Benefit Biographies”). In a first step, we estimate discrete-time hazard rate models for competing risks that control for unobserved heterogeneity for exits from benefit receipt within three years of first benefit receipt. In a second step, we apply similar models to estimate the probabilities to return to benefit receipt within three years after exit.