Chapters

• Why do we need “causal inference” tools?

• Exogenous vs. endogenous treatments

• The Average Treatment Effect

• The Counterfactual

📝 Main references: Cameron and Trivedi (2022), Cunningham (2020), Gordon et al (2019).

• Inverse Probability Weights

• Propensity Score Matching (PSM) & Nearest-Neighbors

• Coarsened Exact Matching (CEM)

📝 Main references: Almond et al (2005), Cattaneo (2010), Cunningham (2020), Hitt and Frei (2002), Huntington-Klein (2021).

🖫 Datasets: birth_weight from Cattaneo (2010), training_example from Cunningham (2020)

• Brief reminders on the DiD approach

• Unobserved individual heterogeneity

-          Repeated cross sections vs. panel data

-          Bias of the TWFE and the Bacon decomposition

• Heterogeneous treatment effects

-          Over time and across cohorts

-          Challenges associated with staggered DiD

📝 Main references: Baker et al (2022), Callaway and Sant’Anna (2021), Cunningham (2020), Goodman-Bacon (2021), Kretschmer and Peukert (2020).

🖫 Datasets: castle_doctrine_law from Cheng and Hoekstra (2013), callaway from Cunningham (subset from Callaway and Sant’Anna (2021)). 

• Event Studies with Continuous treatment

• Triple difference models

• Contamination, (spatial) spillovers: challenges and possible solutions     

• Serial correlation and mismeasurement of standard errors

📝 Main references : Bertrand et al (2004), Butt (2021), Boehmer et al (2020), Callaway et al (2024), Cameron and Trivedi (2022), Cunningham (2020), Olden and Moen (2022), Zuo (2021).

🖫 Datasets: management_style from Lo et al (2022), organs from Kessler and Roth (2012), michelin from Favaron et al (2022), fracking from Bartik et al (2016) adapted in Callaway et al (2024)

• Comparative case studies

• Data requirements

• Estimation procedure

• Advantages and limitations of the method

📝 Main references: Abadie et al (2010), Abadie et al (2015), Abadie (2021), Cameron and Trivedi (2022), Cunnigham (2020), Zohrehvand et al (2023).

🖫 Datasets: texas from Cornwell and Cunningham (2016), repgermany from Abadie et al (2015), selfdriving_cars from Courthoud (2022).

• The RD design: running variable, cutoff, bandwidth

• Data requirement

• Estimation procedure

• Sharp and Fuzzy RDD

• Regression Discontinuity in Time (RDiT)

• Regression Kink Design

📝 Main references: Cameron and Trivedi (2022), Cunningham (2020), Huntington-Klein (2021), Bradley et al (2017), Flammer and Bansal (2017).

🖫 Datasets: gov_transfs from Manacorda et al (2011), mortgage from Feller (2013), diplomas from Clark and Martorell (2014).