Week 4 — Causal Inference II upcoming¶
July 13 – July 17, 2026 · Chapters 4.1 – 4.5 · Lab 4 · Mentor Session 4 · Problem Set 4
The modern panel/quasi-experimental toolkit — and the recent literature showing the "obvious" estimators are often wrong.
You will see, with your own simulation, a two-way fixed-effects DiD estimator return +0.40 when the true effect is −2.57 — and then you will fix it with Callaway–Sant'Anna. This is the single most important methodological week of the program.
Daily schedule¶
| Block | What |
|---|---|
| 9:00 – 12:00 | Read Ch 4.1 — Difference-in-Differences & Event Studies (parallel trends, leads/lags). Work nb4.1. |
| 1:00 – 4:00 | Live whiteboard on parallel-trends testing. |
| 4:00 – 5:00 | Office hours |
| Block | What |
|---|---|
| 9:00 – 12:00 | Read Ch 4.2 — The Staggered-Adoption Crisis (Goodman-Bacon, negative weights, CS, Sun–Abraham, BJS). Work nb4.2. |
| 1:00 – 4:00 | Lab 4 — Staggered-DiD bake-off: same data, four estimators, compare. |
| 4:00 – 5:00 | Office hours |
| Block | What |
|---|---|
| 9:00 – 12:00 | Read Ch 4.3 — Regression Discontinuity (sharp/fuzzy, CCT bandwidth, McCrary). Work nb4.3. |
| 2:00 – 3:30 | Mentor Session 4 — "Reading staggered-DiD papers in 2026" |
| 4:00 – 5:00 | Office hours |
| Block | What |
|---|---|
| 9:00 – 12:00 | Read Ch 4.4 — Synthetic Control & Synthetic DiD. Work nb4.4 (Abadie + Arkhangelsky). |
| 1:00 – 4:00 | Read Ch 4.5 — Bartik / Shift-Share Designs. Work nb4.5 (Rotemberg weights reassemble the 2SLS estimate). |
| 4:00 – 5:00 | Office hours |
| Block | What |
|---|---|
| 9:00 – 12:00 | Wrap PS 4.1 – 4.5. |
| 1:00 – 3:00 | Capstone team formation — final 3 teams of 4–5 students each declared. |
| 3:00 – 4:00 | Demo + Week-5 preview |
| 5:00 pm | Deliverable cutoff |
What to read¶
- Ch 4.1 — DiD & Event Studies
- Ch 4.2 — The Staggered-Adoption Crisis
- Ch 4.3 — Regression Discontinuity
- Ch 4.4 — Synthetic Control & Synthetic DiD
- Ch 4.5 — Bartik / Shift-Share
Notebooks: notebooks/week-04/.
Lab: Lab 4 — Staggered-DiD bake-off.
Mentor reading: Mentor 4 — "Reading staggered-DiD papers".
What to submit (due Fri Jul 17, 5:00 pm ET)¶
- Problem Set 4 — PS 4.1 through 4.5.
- Lab 4 notebook comparing TWFE, CS, Sun–Abraham, BJS on the same staggered panel.
- Capstone team declaration — your team's name + your slot in the team in
cohort/teams.md. - Journal entry —
journal/week-04.md. - Pull request —
week-04 — <your name>.
Data we touch this week¶
- CRSP + Compustat — policy-shock event studies.
- USPTO PatentsView — for the Bartik example.
- USA Spending — for shift-share exposure construction.
Next week preview¶
Week 5 — Reading the Frontier I →. One classic paper per day — Fama & French 1992/1993, Jegadeesh & Titman 1993, Petersen 2009, Bertrand–Duflo–Mullainathan 2004.