Regression Discontinuity Designs in Development Economics: Theory and Practice

Causal inference is a key tool for development economics and policy evaluation. One challenge for its practical implementation is the ideal quasi-experimental empirical setting: policies usually target specific groups or regions, based often on arbitrary rules. In this course, you will learn how to overcome this challenge with a prominent method: regression discontinuity designs (RDD).

More specifically, we will learn how to use ad-hoc – discontinuous – implementation rules for designing the ideal empirical setting for causal inference. Examples include, among others, policies targeting specific age groups, income levels, and geographical regions. These discontinuities, combined with the right regression framework, permit inferring causal effects of policies in developing economies.

The course will cover both conceptual and practical aspects of RDD. Theoretical lectures will cover the formal assumptions (and their interpretation) for state-of-the-art RDD settings (e.g., sharp, fuzzy, or spatial RDDs), followed by practical sessions that will expose students to hands-on activities of real-world implementation of these tools. By the end of the course, students will command a powerful toolbox for applied empirical work and policy evaluation in developing settings.

An advantage of the course is its open programming environment: practical sessions will cover frontier computational RDD tools from the rd packages toolbox in both Stata and R. Hence, students can either follow the hands-on sessions of the course with their preferred software or practice with both.

The course targets researchers and professionals interested in broadening their causal inference toolbox, as well as graduate students aiming at deepening their theoretical and practical knowledge of RDD methods.