Applied Empirical Research: Methods and Practical Implementation

Most practical managerial decisions as well as discussions in the business sciences evolve around questions such as “What happens to Y if we change X?”, “Is the new business strategy X the reason for increases in revenue Y?”, or “Is the change that we see in Y caused by changes in X or is the change in Y driven by coincidence or some other factor?”. In other words, both practical decision-making and academic research on business decisions require knowledge about cause and effect. However, identifying causalities is usually not straightforward. For example, if a manager implements some new tax-planning strategy and the firm’s profit increases in the subsequent year, it is not clear if the new strategy was the cause for increased profits or if profits would have increased even in the absence of the new strategy. That is, the correlation between the new strategy and subsequent profits does not necessarily reflect a causal effect. A serious evaluation of the new business strategy will, however, need to identify if the change in profits was indeed caused by the new strategy.

Such an analysis of causal effects requires knowledge of both practical data analysis (using statistical software) and methods and strategies to identify causal effects. This course equips students with the skills related to both these components: it provides i) an introduction to causality and an overview of the most important methods and approaches for causal inference, and ii) a hands-on practical introduction to data analysis.

Overall, students learn how to apply the most important methods and how to use statistical software (including the handling of “big data” and common business data bases) in the context of empirical work and causal inference. In general, these skills are very valuable for work both in industry and academia.

The course is generally suited for students with and without prior knowledge of, or particular interest in, taxation: Examples will be from taxation, but the taught methods and empirical applications generalize beyond tax topics.

In line with the objectives of the class, one part of the course focuses on hands-on empirical applications and students learn how to conduct their own empirical analysis. For this purpose, students are introduced to the usage of a statistical software package (R) and to the access and analysis of large data sets (in particular firm databases sets such as Amadeus or Compustat). The introduction to software R starts from scratch and no prior knowledge is necessary. This part of the class will mostly be taught in the Tuesday meetings.

The other part of the course teaches the concept of causality and the most important methods to estimate causal effects. These include randomized experiments, linear regression, difference(s)-in-difference(s), instrumental variables, regression discontinuity design and bunching. The focus is on understanding the advantages and disadvantages of the available methods and less on a highly technical presentation. This part of the class will mostly be taught in the Monday meetings.

To receive a grade, students conduct an empirical analysis using a statistical software package and ‘real-world’ archival data. This empirical project will comprise either the replication of an existing research paper or the analysis of an independently developed research question. This applied data project reflects the objective of the class that students learn to conduct their own empirical analysis.