Skip to content

MultiPeriodDiD: full event-study specification with pre-period coefficients #125

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
igerber merged 8 commits into from
Feb 2, 2026
Merged

MultiPeriodDiD: full event-study specification with pre-period coefficients #125

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
8 commits
Select commit Hold shift + click to select a range
842eefc
MultiPeriodDiD: full event-study specification with pre-period coeffi…
igerber Feb 1, 2026
a2931e6
Address PR #125 review feedback: validation hardening and docs fix
igerber Feb 1, 2026
9440e09
Address PR #125 review round 2: NaN propagation and docs fixes
igerber Feb 1, 2026
8207057
Address PR #125 review round 3: D_it warning and HonestDiD NaN filtering
igerber Feb 1, 2026
d1de9cb
Address PR #125 review round 4: HonestDiD post-period NaN guard
igerber Feb 1, 2026
66b2754
Address PR #125 review round 5: avg_se Inf guard and HonestDiD ordering
igerber Feb 1, 2026
3cb1c85
Address PR #125 review round 6: stale comment and FutureWarning text
igerber Feb 1, 2026
ecddfb2
Address PR #125 review round 7: fix sorted() ordering for HonestDiD
igerber Feb 1, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
34 changes: 34 additions & 0 deletions CHANGELOG.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -7,11 +7,45 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0

## [Unreleased]

### Changed
- **MultiPeriodDiD: Full event-study specification** (BREAKING)
- Treatment × period interactions now created for ALL periods (pre and post),
not just post-treatment
- Pre-period coefficients available for parallel trends assessment
- Default reference period changed from first to last pre-period (e=-1 convention)
with FutureWarning for one release cycle
- `period_effects` dict now contains both pre and post period effects
- `to_dataframe()` includes `is_post` column
- `summary()` output now shows pre-period effects section
- t_stat uses `np.isfinite(se) and se > 0` guard (consistent with other estimators)

### Added
- Time-varying treatment warning when `unit` is provided and treatment varies
within units (guides users toward ever-treated indicator D_i)
- `unit` parameter to `MultiPeriodDiD.fit()` for staggered adoption detection
- `reference_period` and `interaction_indices` attributes on `MultiPeriodDiDResults`
- `pre_period_effects` and `post_period_effects` convenience properties on results
- Pre-period section in `summary()` output with reference period indicator
- `ValueError` when `reference_period` is set to a post-treatment period
- Staggered adoption warning when treatment timing varies across units (with `unit` param)
- Informative KeyError when accessing reference period via `get_effect()`

### Removed
- **TROP `variance_method` parameter** — Jackknife variance estimation removed.
Bootstrap (the only method specified in Athey et al. 2025) is now always used.
The `variance_method` field has also been removed from `TROPResults`.

### Fixed
- HonestDiD: filter non-finite period effects from MultiPeriodDiD results
(prevents NaN propagation into sensitivity bounds; raises ValueError
when no finite pre- or post-period effects remain)
- HonestDiD VCV extraction: now uses interaction sub-VCV instead of full regression VCV
(via `interaction_indices` period → column index mapping)
- MultiPeriodDiD: `avg_se` guard now checks `np.isfinite()` (matches per-period pattern;
prevents `avg_t_stat=0` / `avg_p_value=1` when variance is infinite)
- HonestDiD: extraction now uses explicit pre-then-post ordering instead of sorted period
labels (prevents misclassification when period labels don't sort chronologically)

## [2.2.0] - 2026-01-27

### Added
Expand Down
2 changes: 1 addition & 1 deletion CLAUDE.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -60,7 +60,7 @@ cross-platform compilation - no OpenBLAS or Intel MKL installation required.

- **`diff_diff/estimators.py`** - Core estimator classes implementing DiD methods:
- `DifferenceInDifferences` - Basic 2x2 DiD with formula or column-name interface
- `MultiPeriodDiD` - Event-study style DiD with period-specific treatment effects
- `MultiPeriodDiD` - Full event-study DiD with treatment × period interactions for ALL periods (pre and post). Supports `unit` parameter for staggered adoption detection. Default reference period is last pre-period (e=-1 convention). Pre-period coefficients enable parallel trends assessment. `interaction_indices` maps periods to VCV column indices for robust sub-VCV extraction in HonestDiD/PreTrendsPower.
- Re-exports `TwoWayFixedEffects` and `SyntheticDiD` for backward compatibility

- **`diff_diff/twfe.py`** - Two-Way Fixed Effects estimator:
Expand Down
45 changes: 29 additions & 16 deletions README.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -561,31 +561,37 @@ results = twfe.fit(

### Multi-Period DiD (Event Study)

For settings with multiple pre- and post-treatment periods:
For settings with multiple pre- and post-treatment periods. Estimates treatment × period
interactions for ALL periods (pre and post), enabling parallel trends assessment:

```python
from diff_diff import MultiPeriodDiD

# Fit with multiple time periods
# Fit full event study with pre and post period effects
did = MultiPeriodDiD()
results = did.fit(
panel_data,
outcome='sales',
treatment='treated',
time='period',
post_periods=[3, 4, 5], # Periods 3-5 are post-treatment
reference_period=0 # Reference period for comparison
reference_period=2, # Last pre-period (e=-1 convention)
unit='unit_id', # Optional: warns if staggered adoption detected
)

# View period-specific treatment effects
for period, effect in results.period_effects.items():
print(f"Period {period}: {effect.effect:.3f} (SE: {effect.se:.3f})")
# Pre-period effects test parallel trends (should be ≈ 0)
for period, effect in results.pre_period_effects.items():
print(f"Pre {period}: {effect.effect:.3f} (SE: {effect.se:.3f})")

# Post-period effects estimate dynamic treatment effects
for period, effect in results.post_period_effects.items():
print(f"Post {period}: {effect.effect:.3f} (SE: {effect.se:.3f})")

# View average treatment effect across post-periods
print(f"Average ATT: {results.avg_att:.3f}")
print(f"Average SE: {results.avg_se:.3f}")

# Full summary with all period effects
# Full summary with pre and post period effects
results.print_summary()
```

Expand Down Expand Up @@ -951,10 +957,10 @@ Create publication-ready event study plots:
```python
from diff_diff import plot_event_study, MultiPeriodDiD, CallawaySantAnna, SunAbraham

# From MultiPeriodDiD
# From MultiPeriodDiD (full event study with pre and post period effects)
did = MultiPeriodDiD()
results = did.fit(data, outcome='y', treatment='treated',
time='period', post_periods=[3, 4, 5])
time='period', post_periods=[3, 4, 5], reference_period=2)
plot_event_study(results, title="Treatment Effects Over Time")

# From CallawaySantAnna (with event study aggregation)
Expand Down Expand Up @@ -1413,14 +1419,15 @@ Pre-trends tests have low power and can exacerbate bias. **Honest DiD** (Rambach
```python
from diff_diff import HonestDiD, MultiPeriodDiD

# First, fit a standard event study
# First, fit a full event study (pre + post period effects)
did = MultiPeriodDiD()
event_results = did.fit(
data,
outcome='outcome',
treatment='treated',
time='period',
post_periods=[5, 6, 7, 8, 9]
post_periods=[5, 6, 7, 8, 9],
reference_period=4, # Last pre-period (e=-1 convention)
)

# Compute honest bounds with relative magnitudes restriction
Expand Down Expand Up @@ -1488,14 +1495,15 @@ A passing pre-trends test doesn't mean parallel trends holds—it may just mean
```python
from diff_diff import PreTrendsPower, MultiPeriodDiD

# First, fit an event study
# First, fit a full event study
did = MultiPeriodDiD()
event_results = did.fit(
data,
outcome='outcome',
treatment='treated',
time='period',
post_periods=[5, 6, 7, 8, 9]
post_periods=[5, 6, 7, 8, 9],
reference_period=4,
)

# Analyze pre-trends test power
Expand Down Expand Up @@ -1764,23 +1772,28 @@ MultiPeriodDiD(
| `covariates` | list | Linear control variables |
| `fixed_effects` | list | Categorical FE columns (creates dummies) |
| `absorb` | list | High-dimensional FE (within-transformation) |
| `reference_period` | any | Omitted period for time dummies |
| `reference_period` | any | Omitted period (default: last pre-period, e=-1 convention) |
| `unit` | str | Unit identifier column (for staggered adoption warning) |

### MultiPeriodDiDResults

**Attributes:**

| Attribute | Description |
|-----------|-------------|
| `period_effects` | Dict mapping periods to PeriodEffect objects |
| `avg_att` | Average ATT across post-treatment periods |
| `period_effects` | Dict mapping periods to PeriodEffect objects (pre and post, excluding reference) |
| `avg_att` | Average ATT across post-treatment periods only |
| `avg_se` | Standard error of average ATT |
| `avg_t_stat` | T-statistic for average ATT |
| `avg_p_value` | P-value for average ATT |
| `avg_conf_int` | Confidence interval for average ATT |
| `n_obs` | Number of observations |
| `pre_periods` | List of pre-treatment periods |
| `post_periods` | List of post-treatment periods |
| `reference_period` | The omitted reference period (coefficient = 0 by construction) |
| `interaction_indices` | Dict mapping period → column index in VCV (for sub-VCV extraction) |
| `pre_period_effects` | Property: pre-period effects only (for parallel trends assessment) |
| `post_period_effects` | Property: post-period effects only |

**Methods:**

Expand Down
Loading