Causal Inference Workshop

Materials for Sustainable Development PhD Spring 2024 Causal Inference Workshop

Based on Claire Palandri's 2022 version of the workshop

Outline

• Week 1: Modeling assumptions
• Week 2: Potential outcomes framework (and DAGs, briefly)
• Week 3: IV and RDD (overview, DGP, assumptions, examples)
• Week 4: IV and RDD coding exercises and SDev examples
• Week 5: Event study, DiD, DiDiD, and the (potential) problem with TWFEs
• Week 6: New TWFE literature, DiD/TWFE coding exercises
• Week 7: Synthetic Control and Synthetic DiD
• Week 8: Pre-estimation, estimation, and post-estimation steps
• Week 9: Inference
• Week 10: Fixed effects, climate regressions, remote sensing
• Week 11: Text analysis, wrap-up

Coding examples and exercises (in R, Stata, and Python)

• Week 1: What happens when errors are spatially correlated? Simulated exercise
• Week 4:
  • 01a_iv_simulated: IV exercise using simulated data (adjusting strength of instrument, exclusion restriction)
  • 01b_iv_card1995: IV exercise using published data (Card 1995), based on this
  • 01c_rdd_simulated: RD exercise using simulated data (different functional forms of X, non-linear DGP, etc.), based on this
  • 01d_rdd_carpenterdobkin2009: RD exercise using published data (Carpenter and Dobkin 2009), based on this
• Week 6:
  • 01_twfe: Implements various new TWFE estimators in R
• Week 7:
  • 01_synthdid: Implements synthetic control and synthetic DiD examples in R
• Week 9:
  • 01_rand_inf: Simple randomization inference example, based on this
• Week 10:
  • 01_fe: Simple examples showing problem with perfectly collinear treatment variable and time/unit FE
• Week 11:
  • 01_nlp_example.ipynb: Example of NLP preprocessing steps (stemming, etc.) as well as simple tasks (sentiment analysis, LDA)
  • See Google Colab notebook here

Other helpful resources for causal inference and research design

• Causal Inference: The Mixtape by Scott Cunningham (contains helpful coding examples in Stata, R, and Python)
• The Effect: An Introduction to Research Design and Causality by Nick Huntington-Klein
• Paul Goldsmith-Pinkham's Yale Applied Empirical Methods PhD Course repo
• Kirill Borusyak's Berkeley ARE PhD Applied Econometrics
• TWFE/DiD with heterogeneous treatment effects resources:
  • Helpful starting point: de Chaisemartin and D'Haultfoeuille review/survey
  • YouTube DiD reading group videos (very helpful!)
  • Jonathan Roth's DiD resources page
• Very cool LLM visualization (thanks to Junho Choi for sharing!)