Introduction | Background | Package Usage | Simulated Data Example | A "Real Data" Example | Testing the zero-effect null | Per-cohort, per-event-time, and per-(cohort, time) accessors | Plotting the effects | Tidy outputs with broom | Conclusion | References
Package: fetwfe 1.56.18.9000
fetwfe: Fused Extended Two-Way Fixed Effects
Calculates the fused extended two-way fixed effects (FETWFE) estimator for unbiased and efficient estimation of difference-in-differences in panel data with staggered treatment adoption. This estimator eliminates bias inherent in conventional two-way fixed effects estimators, while also employing a novel bridge regression regularization approach to improve efficiency and yield valid standard errors. Also implements extended TWFE (etwfe) and bridge-penalized ETWFE (betwfe). Provides S3 classes for streamlined workflow and supports flexible tuning (ridge and rank-condition guarantees), automatic covariate centering/scaling, and detailed overall and cohort-specific effect estimates with valid standard errors. Includes simulation and formatting utilities, extensive diagnostic tools, vignettes, and examples. See Faletto (2025) (<doi:10.48550/arXiv.2312.05985>).
Authors:
fetwfe_1.56.18.9000.tar.gz
fetwfe_1.56.18.9000.zip(r-4.7-any)fetwfe_1.56.18.9000.zip(r-4.6-any)fetwfe_1.56.18.9000.zip(r-4.5-any)
fetwfe_1.56.18.9000.tgz(r-4.6-any)fetwfe_1.56.18.9000.tgz(r-4.5-any)
fetwfe_1.56.18.9000.tar.gz(r-4.7-any)fetwfe_1.56.18.9000.tar.gz(r-4.6-any)
fetwfe_1.56.18.9000.tgz(r-4.6-emscripten)
manual.pdf |manual.html✨
DESCRIPTION |NEWS
card.svg |card.png
fetwfe/json (API)
| # Install 'fetwfe' in R: |
| install.packages('fetwfe', repos = c('https://gregfaletto.r-universe.dev', 'https://cloud.r-project.org')) |
Bug tracker:https://github.com/gregfaletto/fetwfepackage/issues
Pkgdown/docs site:https://gregfaletto.github.io
Last updated from:d0a61a59ff. Checks:9 OK. Indexed: yes.
| Target | Result | Time | Files | Syslog |
|---|---|---|---|---|
| linux-devel | OK | 564 | ||
| source / vignettes | OK | 378 | ||
| linux-release | OK | 600 | ||
| macos-release | OK | 928 | ||
| macos-oldrel | OK | 785 | ||
| windows-devel | OK | 802 | ||
| windows-release | OK | 1103 | ||
| windows-oldrel | OK | 946 | ||
| wasm-release | OK | 219 |
Exports:attgtToFetwfeDfbetwfebetwfeWithSimulatedDatacohortStudycohortTimeATTsdebiasedATTdebiasedATTWithSimulatedDataetwfeetwfeToFetwfeDfetwfeWithSimulatedDataeventStudyfetwfefetwfeWithSimulatedDatagenCoefsgenCoefsCoregetTessimulateDatasimulateDataCoresimultaneousCIstwfeCovstwfeCovsWithSimulatedData
Dependencies:codetoolsforeachgenericsglmnetgrpregiteratorslatticeMatrixmvtnormRcppRcppEigenshapesurvival
fetwfe: A Package for Fused Extended Two-Way Fixed Effects
Last update: 2026-10-28
Started: 2026-07-06
Standard errors in fetwfe: the variance decomposition and se_type
1. The headline: tight Gaussian CIs by default under (Ψ-IF) | 1.1 Which propensity-score estimators satisfy (Ψ-IF)? | 1.2 What changed in v1.12.0 | 2. The two variance pieces and how the package exposes them | 2.1 $V_1$: regression-coefficient variance | 2.2 $V_2$: propensity-score variance | 2.3 The combined variance: tilde_v_N and the paper's catalogue | 2.4 A running example | 3. Experimental: cluster-robust standard errors via se_type = "cluster" | 3.1 The formula | 3.2 Why we call this experimental | 3.3 Worked cluster comparison | 4. Choosing the bridge penalty parameter lambda | References
Last update: 2026-10-01
Started: 2026-05-12
Simultaneous confidence bands and the zero-effect test
1. Simultaneous confidence intervals via simultaneousCIs() | The multiplier-bootstrap method: method = "bootstrap" | Simultaneous bands are the default: the ci_type argument | 2. Selection consistency as an implicit zero-effect test | Why this is stronger than a confidence interval on an unregularized estimator | The q < 1 caveat | A controlled-truth worked example | Scope: this is the post-treatment zero-effect test | References
Last update: 2026-08-21
Started: 2026-07-09
High-dimensional inference: debiasedATT() when p >= NT
1. High-dimensional inference ($p \geq NT$): debiasedATT() | A worked example | Why debiased, and how well it covers | An asymptotic alternative: the wild bootstrap | Simultaneous bands at $p \geq NT$ | Controlled high-dimensional truth | References
Last update: 2026-07-21
Started: 2026-07-09
Simulation Vignette for FETWFE: From Coefficients to True Treatment Effects
Introduction | Simulation Workflow Using Piping | Step 1: Generate Simulation Coefficients | Step 2: Simulate Panel Data | Step 3: Run the FETWFE Estimator on Simulated Data | Step 4: Extract True Treatment Effects | Combining the Workflow in One Pipeline | Covariate-dependent cohort assignment | Multinomial-logit DGP | Ordered-logit DGP | Effect of assignment strength | Nonlinear propensity via interactions | Truth derivation under non-marginal DGPs | Targeted-sparsity coefficients for high-dimensional DGPs | Conclusion
Last update: 2026-07-09
Started: 2025-03-05
Coming from did or etwfe? Bringing your data into fetwfe
From did | From etwfe | What fetwfe adds | References
Last update: 2026-07-07
Started: 2026-07-07
Choosing a fusion structure: cohort vs. event-study penalties
What a "fusion structure" is | fusion_structure = "cohort" (the default) | fusion_structure = "event_study" | Two meanings of "event study" --- read this first | A side-by-side demo | When to prefer each | Supplying your own fusion matrix D_N (advanced) | The row/column ordering | Worked example | Staying inside the theoretical guarantees | References
Last update: 2026-06-07
Started: 2026-06-04
Comparison Estimators: ETWFE and BETWFE
Introduction | Setup: simulating panel data | etwfe(): extended TWFE without penalty | betwfe(): bridge-penalized ETWFE | No-covariate setting | Visualizing event-time effects | When to use which | References
Last update: 2026-06-06
Started: 2026-05-11
