The objective of this research project – which is financed by the Leibniz-ScienceCampus MannheimTaxation – is to study the causal effect of unemployment insurance benefit levels on the probability to become self-employed (vs. employed) and the duration elasticity. Existing research has neglected the transition channel from unemployment to self-employment and its relationship to payroll taxes. The project will develop a quasi-empirical strategy that deals with the potential bias arising from neglecting the effect from self-employment. To achieve this goal in the project new administrative social security data will be used that allow to describe all relevant labor market flows since the mid-2000s including the whole economic crisis period.