In most of economics, marketing and business management, we are interested in causal relationships between variables, rather than mere correlations. For example, it is not the correlation between marketing expenses and sales that is of interest, but the effect of increasing marketing expenses for a product on the sale volume of the same product. In this course, we study methods for estimating and identifying such causal effects.
First, the course provides a brief review of basic regression techniques. Second, we introduce the topic of causal analysis. We will define causal effects based on the potential outcomes framework, encounter the fundamental problem of causal analysis, and discuss what separates association from causation. In the third part of the course, we discuss randomized experiments as the predominant way for establishing causality. Last, we use the potential outcomes framework to discuss designs and methods for data from observational studies. In particular, designs and methods covered include instrumental variables, difference-in-difference, event study design, regression discontinuity design, kink design, and bunching. Examples from the literature and step-by-step tutorials offer hands-on experiences in utilizing the methods.
Preliminary course outline:
- Short review of basic regression techniques (inference, asymptotics and dummy variables)
- Causal inference using potential outcomes
- Randomized experiments
- Regression and causality
- Instrumental variables (LATE)
- Fixed effects and panel data
- Differences-in-differences
- Event study design
- Regression discontinuity design
- Kink design
- Bunching