led by Dr Bernhard Reinsberg
This training course provides doctoral students with the necessary methodological tools for policy evaluation. While policy-makers are interested in the causal effects of policy interventions, a perennial problem that makes such assessments difficult is endogeneity, for instance due to reverse causality.
This course will first review conventional methodological strategies—matching, regression discontinuity designs, and instrumental-variable analysis—that promise to address endogeneity bias. Focusing on instrumental-variable analysis, we will derive its theoretical properties for the linear case. For more complex problems, for instance if there are two types of policy interventions, the course will introduce students to a more flexible framework—conditional mixed process estimation—which is similar to a control function approach.
While students will be presented examples from recent applied research, they are encouraged to bring their own problems to class and work on them during the second part of the training.
Who’s it for?
Intermediate knowledge of regression analysis is required and interest in causal inference and quantitative methods is recommended.