This course covers advanced topics in econometrics with a particular focus on topics in causal inference for applications in economics. The course is designed for students interested in pursuing research in applied and theoretical econometrics as well as for students with a solid background in econometrics or statistics interested in learning recent advances in causal inference for applied economic research. The first part presents topics in causal inference, starting from selection on observables, non-compliance, and then focusing on experimental design, and inference in experiments. The second part covers most advanced topics for semi-parametric estimation, heterogeneous treatment effects and policy learning, panel data models, spillover effects. Readings for each topic will include theoretical papers in econometrics and papers in applied economic research that provide empirical examples of the method.
Lectures will be on Tuesday and Thursday 9am - 1015am. Office hours will be held every week for 1h on Tuesday 1030 - 1130. You should always feel free to email me if anything is needed. Grading will be based on 5 problem sets, a final research essay (and a final presentation if time permits). Final grade will assign weight 40\(\%\) to the problem sets (equally weighted after removing the problem set with the lowest grade), 60\(\%\) for the final essay.
Students must have take the first year graduate sequence in econometrics (ECON2120 and ECON 2140) or equivalent with the permission of the instructor. Multivariable calculus, linear algebra, concepts in probability theory and asymptotic statistics will be needed extensively.
Module 1: Introduction to Causal Inference
Module 2: Causal Mechanisms and DAGs
Module 3: Sampling and Design based uncertainty in regression analysis
Module 4: Hypothesis Testing Part I
Module 5: Hypothesis Testing and Incentives (Part II)
Module 6: Publication bias, sampling and incentives
Module 7: Shrinkage and robust estimation in observational studies
Module 8: Effects heterogeneity and policy learning
Module 9: Advances in Panel Data Methods for causal inference
Module 10: Networks and spillovers
Models of spillovers
Reflection problem, identification of peer effects and analysis of network data
Experimental Designs in Complex Experiments
Shift-share designs and recentered instruments
Final Lecture: reviews and other topics
40% problem sets; 60% final essay.
The essay may either (or both) focus on studying an econometric method from a theoretical perspective or a study of a method’s properties and applicability for applied economic research.
The essay must contain a short introduction that motivates the method for applied economic research, a section discussing the setup, a section discussing the methodology and presenting the main theoretical analysis, and a section presenting the numerical properties of the procedure and an empirical application using some existing data.
The theoretical analysis must include the main theoretical argument of the original paper and does not need to contain all theoretical results in the original paper. (For instance, students more interested in the applicability of methods in applied research may only report the main identification result or proof of consistency.) Each formal statement reported in the essay must have a proof in an appendix.
Students are encouraged to present novel numerical studies and an empirical application different from the one in the original paper.
The paper will be graded based on correctness, quality of the writing, and novelty of the theoretical, numerical, and empirical analysis.
For writing the final essay, students are strongly encouraged to read
The book introduces students to research writing.