Authors	Hujer, R., Caliendo, M. and Radic, D.
Year	2001
Reference	Working Paper, University of Frankfurt
Keywords	Evaluation, Matching,Difference-in-Differences, Monte Carlo Simulation
JEL-Classification	C13, C14, C15, H43, J68
Presentations	Conference of the Italian Labour Economic Association (AIEL) - Florence 2001.
Download	PDF
Abstract	The need for evaluation of active labour market policies is no longer questioned. Recent overviews of microeconometric evaluation studies for Germany have shown substantial differences regarding the estimated effects of these measures. Whereas most of the differences might be due to different data sets used or different time periods examined, some of the variation is likely to be induced by different evaluation approaches. It is obvious, that nobody knows the real extent of the programme effects, making it hard to assess the quality of different estimation techniques.The aim of this paper is twofold. In the beginning we present four different evaluation strategies and discuss the methodological concepts associated with them. Whilst presenting the methodological concepts of each estimator, we will also discuss the data needed for their implementation. By doing so, we hope to give advice for the construction of datasets in the future. Then we perform a simulated labour market experiment, which allows us to control and estimate all relevant effects. We are going to show how the different estimators perform under different conditions.The experiment is designed to take into account observable as well as unobservable characteristics in determining the labour market success and the participation decision of individuals. This design enables us to show the (dis-) advantages of each estimation technique and present the environments under which they work best.