Teaching

Causality is central to much of empirical work in economics. After formally defining the concept of causality through potential outcomes and identification of causal effects, the course introduces several research designs in which identification of causal effects can be established.

The course shows the formal identification arguments, discusses empirical examples, and practical aspects of implementation. Examples of research designs covered are matching, instrumental variables, differences-in-differences, and regression discontinuity designs.

Related Seminar:
Seminar on Causal Inference

Instructor:
Daniel Wilhelm

Examination Format:
Written Exam

Class Format
6 ECTS (statistics), Class meets twice per week, English

This course centers around the generalized method of moments (GMM) as a unifying framework in econometrics. We begin by exploring ordinary least squares (OLS) as a special case of GMM to build a strong foundation. In the second part, we focus our attention on GMM, covering its large-sample properties and key topics like over-identifying restrictions. Finally, we explore various special cases of GMM, including instrumental variables and two-stage least squares, applying these techniques to real-world econometric models.

Instructor:
Mauricio Olivares

Examination Format:
Written Exam

Class Format
6 ECTS (statistics), Class meets twice per week, English

The course covers the foundations of classic nonparametric methods, with emphasis on kernel estimation of density and regression functions. It combines rigorous derivations of the statistical properties of estimators with simulation studies and empirical applications, such as regression discontinuity/kink designs and average treatment effect estimation.

Related Seminar:
Seminar on Treatment Effects: Estimation and Inference

Instructor:
Tomasz Olma

Examination Format:
Grading is based on problem sets

Class Format
6 ECTS (statistics), Class meets twice per week, English

This course is not an in-depth course about machine learning methods, but rather about how such methods can be employed in empirical work in economics. Since the goal of most applications in economics is estimation of causal effects or parameters in economic models, we will discuss how machine learning methods can be useful for such non-prediction tasks.

The course will introduce the concept of double machine learning in which estimators of target parameters are phrased in terms of Neyman orthogonal moment equations. Estimators based on these moment conditions and a suitable version of sample splitting are then shown to possess desirable asymptotic properties like fast convergence rates and asymptotic normality.

After a first part in which some foundations about machine learning methods like the LASSO are established, the course discusses the double machine learning approach and then shows how it can be used in various applications. Examples are inference on regression parameters in the presence of high-dimensional covariates, instrumental variable regressions with high-dimensional covariates and/or instruments, and learning about treatment effect heterogeneity, among others.

Related seminar:
Seminar on Machine Learning in Econometrics

Instructor:
Daniel Wilhelm

Examination Format:
Empirical Project

Class Format
6 ECTS (Statistics), Class meets twice per week, English

Instructor
Klaus Wohlrabe

Examination Format
Written Exam

Class Format
6 ECTS, The course can be credited for:

• Zeitreihen (WP7, BA Statistik, PO 2010)
• Time Series (WP 18, MA Statistics and Data Science, PO 2021)
• Zeitreihen (P 6.0.27, MA Statistik, PO 2010)
• Zeitreihen (P 8.0.32, MA Biostatistik, PO 2010)
• Zeitreihen (P 7.0.5, MA WiSo Statistik, PO 2010)
• Time Series (WP18, MA Statistics and Data Science)

This seminar is designed to be taken together with the course "Causal Inference" as an opportunity to explore topics from the lectures in more depth. Taking the course "Causal Inference" is not required, but background knowledge on causality is strongly recommended.

Because of its link to the Causal Inference course, the topics of the seminar are strongly related: we will read and discuss recent research papers on the identification of causal effects with a focus on applications in economics.

Instructor
Daniel Wilhelm

Evaluation Form
Each student will choose one paper on which they will work throughout the semester.

Treatment effects are the central object of interest in the econometrics of program evaluation. In this seminar, we will cover a series of recent developments in the theory and practice of estimation and inference on treatment effects based on experimental as well as observational data.

Coding skills and background knowledge on probability and statistics at the master's level are strongly recommended. Students who have taken or are taking Econometric Theory might benefit from it.

Each student will choose one paper on which they will work throughout the semester.

Instructor
Tomasz Olma and Vincent Starck

Examination and Class Formats

1. Standard Seminar Course
Statistics Master, 9 ECTS
Economics M12
Presentation, Seminar Paper (15-25 pages), Seminar Paper and Presentations will be graded.

2. Mini Seminar Course
Statistics Master, 3 ECTS
Graded Presentation

3. Attendance
MGSE, No presentation, No grade

In this course, we will learn how to conduct empirical analyses through the lenses of permutation- and bootstrap-based techniques. In particular, we will uncover how these nonparametric methods allow us to make reliable inferences under minimal assumptions in a data-driven way.

This course offers a unique blend of three domains: a solid theoretical foundation, a careful treatment of the computational principles, and a comprehensive revision of key applications that are essential for causal inference. This way, you will understand the basics of nonparametric inference and how to apply it in practice.

Using real-world examples from experimental or observational data, we will gain experience by applying these methods to uncover causal relationships and draw robust conclusions. Crucially, we will engage in hands-on projects where you can put your coding skills to the test.

This course is a stepping stone to modern data analysis. Whether you are planning a career in statistics, data science, economics, psychology, or any field that relies on data-driven decision-making, this course will equip you with the tools, knowledge, and coding expertise to take on contemporary problems in applied causal analysis.

Instructor:
Mauricio Olivares

Examination Format:
Oral Exam

Class Format
6 ECTS (Statistics), Class meets twice per week, English

Module (Statistics)
WP4 Ausgewählte Gebiete der Angewandten Statistik

This topics course in econometrics is tailored for advanced undergraduate students. The first part is devoted to causal inference and aims to give a solid understanding of the main concepts and modern methods.

The second part of the course covers networks, especially network description and representation as well as the study of network formation models.

If time permits, we will discuss causal inference issues in the presence of a network, e.g., the problem of interference, the reflection problem, and related topics.

Instructor:
Vincent Starck

Examination Format:
Written Exam

Class Format
6 ECTS (Statistics), Class meets twice per week, English

Module (Statistics)
WP7 Ausgewählte Gebiete der statistischen Modellierung