Descriptif

This course presents an overview of econometric methods used for causal inference, i.e., methods designed to estimate the impact of a potential cause (usually a policy intervention or other institutional change) on an outcome of interest. Selection effects can impede attempts to infer causality. Causes and consequences are discussed relying on the counterfactual framework used in the program evaluation approach. The course will be largely based on critical reading of empirical articles, putting emphasis on the identification issues. 

The course covers a variety of identification designs, including randomized experiments, matching, difference-in-difference, instrumental variables, and regression discontinuity designs. In addition, it presents recent advanced statistical methods such as quantile regression analysis. It also discusses the appropriateness of the underlying assumptions of these estimators, as well as the interpretation of the results obtained by those methods. 

By the end of the course, students should be able to:

-    explain the counterfactual framework, and use it to interpret the concept of selection. 
-    understand the leading quantitative methods for causal inference, and apply them to a variety of policy designs and available data
-    compare the strengths and weaknesses of these estimators in a given research context
-    recognize and interpret the conditions under which these estimators possess desirable statistical properties
-    explain the consequences of the violation of their identifying assumptions

Prerequisites

The course assumes a good knowledge of basic statistics and linear econometrics (linear regression model, estimation and testing).