Treatment Effect Analysis


Estimating a causal effect or "treatment effect" from nonexperminatal data is the aim of much empirical research in economics. This course will cover the most important concepts and methods in this field from an applied perspective. The proposed schedule is (i) Rubin Model of Causality, (ii) Roy Model of Self-Selection, (iii) Causality and Regression Notation, (iv) Experiments, (v) Conditional Independence, (vi) Heckman Switching Regression, (vii) Instrumental Variables and Local Average Treatment Effect, (viii) Difference-in-Differences and Panel Methods, (ix) Regression Discontinuity Design.

The tutorials provide the opportunity to apply the methods covered in class to real data using the software STATA.

The course grade will be primarily based on the final exam. BDPEMS students also need to submit a referee report on a paper that attempts to estimate a causal effect by applying one of the methods presented in class. Many suitable papers can be found on the current and past seminar schedule at

Literature: Mostly Harmless Econometrics by Angrist and Pischke

Technische Universität Berlin