In most of economics we are interested in causal relations between variables, rather than mere correlations. For example, it is not the correlation between earnings and education that is of interest, but the effect of increasing someone's education by one year on that same person's earnings. In this course we study methods for estimating and identifying such causal effects. We discuss techniques used in the statistical literature, starting with the dominant method for establishing causal effects, namely randomized clinical trials. We then move on to observational studies and discuss what conditions are required for credible inference for causal effects in the absence of randomization.

We discuss theoretical and practical issues arising in causal inference as well as applications in the economics literature where these or other methods have been employed.

There will be one lecture a week, on Tuesdays from noon to 3:00 p.m. There will be regular problem sets involving analysis of real data sets to develop an understanding of a ability to apply the methods discussed in class.
 

Detailed Course Syllabus Here in Adobe PDF format.  (Go here to download the Acrobat Reader.)

A List of the readings for the course and the location of the readings on the Internet are here.

Problem Set #1

Problem Set #2

    LaLonde Data

    Matlab Program for LaLonde Data

Imbens and Rubin, Chapter 4, "Fisher's Exact Test in Completely Randomized Experiments"
Imbens and Rubin, Chapter 5, "Neyman's Repeated Sampling Perspective in Completly Randomized Experiments"
Hotz, Notes from World Bank Lectures
Hotz, Notes on Bounding Treatment Effects