Econometrics

See TUMonline
Note: Registration until April 9, 2021, via Moodle.

Participants shall be able to select the appropriate econometric method given a certain problem and data set; to apply this method proficiently using STATA and/or R; to know the advantages and pitfalls of each method; and to judge if the econometric approach in published studies is correctly chosen and well executed.

Econometric analysis aims at uncovering economic mechanisms, their causes and effects. Understanding the mechanisms behind a phenomenon is indispensable if one is to give advice to managers or policy makers, or to build theory. Simple regressions on cross-sectional data show associations, but not causality, so we need more sophisticated methods. This course shall convey econometric methods that allow causal inference, or at least to come closer to uncovering causal effects. The focus will be on applicable knowledge, less on details of the theory. Topics comprise various methods to address selection issues and come close to causality: 1. Randomized controlled trials and natural experiments 2. Matching 3. Regression discontinuity design 4. Instrumental variables 5. Panel data 6. Differences-in-Differences 7. Heckman selection models

Doctoral students only. Basic knowledge of econometrics. Ideally, participation in an introductory course on econometrics.

Learning methods are a mix of seminar presentations by the participants, group discussions, application of econometrics software, and lectures. We will use STATA, though if you prefer you may use R instead. Participants are expected to prepare each session and in particular read the assigned material and run the regression examples provided by Cunnningham such that we can have a discussion in class.

Participants will be assessed based on their seminar presentation (70%) and oral contributions to the course (30%). Seminar presentations will be held by groups of two. The presentation of 90 min to 120 min shall introduce and explain the respective method as well as applications. Presenters will suggest an article in which this method is applied. The group will bring a dataset with which participants will apply the respective method during the course. The lecturer will meet with each group beforehand to aid in the preparation. The course is pass/fail, not graded. In order to pass the course, participants must take part in all classes. In case of excused absence due to illness they need to hand in a written assignment about the content of the class they have missed.

We will use the textbook by Cunningham including the data examples, plus select papers on the respective method. Cunningham, S. (2021): Causal Inference: The Mixtape. https://mixtape.scunning.com/index.html.