Design: Instrumental Variables
This session reviews the basics of Instrumental Variable approaches and then dives into some specific aspects of IVs.
This session aims to strengthen your intuition on Instrumental Variables and on some specific IV topics.
Summary
Instrumental variables are ubiquitous in economics, but their underpinnings are not as straightforward as some other identification strategies. They aim to uncover a source of exogenous variation that affects the treatment but not the outcome directly. This session highlights that an IV estimates a Local Average Treatment Effect (LATE) — the average causal effect for compliers — and that the credibility of this estimate critically depends on the strength and validity of the instrument.
Session Outline
- Introductory example: air pollution and health
- Fundamental principles
- LATE
- Weak instruments
- Estimation
- Summaries
Materials
Exercise
Before class next week (October 21st), please submit the assignment presented in this document.
Specific resources for this lecture
Courses
- A large share of the slides are derived from notes by Claire Palandri and related slides by Anna Papp.
- Chapter 19 of Huntington-Klein (2022) gives great intuition on how IV works
- Chapter 7 of Cunningham (2021)
- Chapter 4 of Angrist and Pischke (2009)
Specific points
Example papers discussed in class
R Pacakges
There is a breadth of packages to implement IV analyses (ivreg or fixest for instance).
I have used the 3 mentioned above in various projects and recommend using fixest for its versatility. It allows to implement analysis with FE, IVs, limited outcomes and so on. Paired with modelsummary to compute standard errors, you should be able to run a lot of your analyses.