fixest Skill

fixest: the canonical R package for fast high-dimensional fixed effects estimation. Covers OLS (feols), Poisson (fepois), GLM (feglm, including logit/probit with FE), negative binomial (fenegbin), and nonlinear models (feNmlm) with multi-way absorbed fixed effects. Supports instrumental variables via three-part formula; difference-in- differences via TWFE and Sun-Abraham (sunab); standard errors including clustered, heteroskedasticity-robust, Newey-West, Driscoll-Kraay, and Conley spatial; and publication output via etable, coefplot, and iplot. Use when execution language is R and the analysis involves fixed effects regression, IV, DiD, or publication-quality regression tables. Python equivalent: pyfixest (which is a port of this package). For panel random/between effects, use plm. For GLM/time series without fixed effects, use base R stats or dedicated packages.

Comprehensive skill for fixed effects regression, instrumental variables, and difference-in-differences estimation with the fixest R package. Use decision trees below to find the right guidance, then load detailed references.

What is fixest?

fixest (Berge, 2018) is the most widely-used R package for fast fixed effects estimation in applied economics and quantitative social science:

• Fast: Multi-way FE demeaning via alternating projections (C++ backend)
• Concise formula syntax: Fixed effects after |, IV after second |, multi-estimation via sw()/csw()/csw0()
• Full GLM support with FE: feglm handles logit, probit, and other GLMs with absorbed high-dimensional FE (unlike pyfixest, which lacks this)
• Sun-Abraham DiD: Built-in sunab() formula function for staggered DiD
• Flexible inference: Switch SE types post-estimation; one-sided formulas for clustering (vcov = ~group)
• Publication output: etable() for regression tables, coefplot() and iplot() for coefficient and event study visualization

Version Notes

This skill targets fixest 0.14.0 (R 4.5.3). fixest 0.13 introduced the breaking changes that pyfixest adopted in 0.40.0 (default SE changed to IID, singleton removal by default, ssc argument renames); DAAF now ships pyfixest 0.60.0, which keeps those fixest-0.13-aligned defaults. fixest 0.14.0 is a stable release building on those defaults.

Key defaults in 0.14.0:

• Default standard errors: "iid" (not cluster-by-first-FE as in pre-0.13)
• Singleton removal: on by default (fixef.rm = "perfect_fit")
• ssc() arguments (canonical 0.14 names): K.adj, K.fixef, K.exact, G.adj, G.df, t.df — the pre-0.13 names (adj, fixef.K, cluster.adj, cluster.df) still work as deprecated back-compat aliases

How to Use This Skill

Reference File Structure

Each topic in ./references/ contains focused documentation:

File	Purpose	When to Read
quickstart.md	feols() basics, formula syntax, multi-estimation (csw, sw, sw0), data requirements	Starting with fixest
fixed-effects.md	Multi-way FE syntax, FE interactions (^), varying slopes, FE recovery via fixef()	FE models and specification
standard-errors.md	Clustered SEs (vcov), HC robust, Conley spatial, Driscoll-Kraay, Newey-West, two-way clustering	Inference choices
iv.md	IV three-part formula, first-stage diagnostics, weak instrument tests	IV/2SLS estimation
did.md	TWFE with feols, sunab() for staggered DiD, event study plots via iplot()	DiD designs
reporting.md	etable() for regression tables (LaTeX, data.frame), coefplot(), iplot(), fixef() for FE extraction	Presenting results
models.md	fepois (Poisson), feglm (GLM with FE), fenegbin (negative binomial), feNmlm (nonlinear), model families	Non-OLS models
gotchas.md	Singleton observations, separation in Poisson, formula parsing pitfalls, SE defaults, panel vs cross-section	Debugging issues

Reading Order

1. New to fixest? Start with quickstart.md then fixed-effects.md
2. Running DiD? Read quickstart.md, then did.md
3. Need IV? Read quickstart.md, then iv.md
4. Making tables? Check reporting.md
5. Non-OLS models? Read models.md
6. Coming from pyfixest? Read quickstart.md then gotchas.md

The reference-file routing in this skill applies to advisory and brainstorming turns as much as implementation. Recommending an SE choice, reviewing a DiD plan, or answering a question that touches a routed topic calls for reading the routed reference file just as much as writing code does — the reference files carry curated caveats and environment-specific constraints (e.g., exact fitstat type names, ssc() defaults) that this overview and general knowledge lack.

Related Skills

Skill	Relationship
pyfixest	Python port of this package — near-identical formula syntax; residual gaps are GLM families beyond logit/probit/gaussian and fenegbin. Load when execution language is Python.
data-scientist	Methodology guidance — load for "why and when" behind methods
r-python-translation	Cross-language mappings for R fixest vs Python pyfixest and statsmodels
plm	Random effects, between estimator, Hausman test — complements fixest when FE-only is insufficient
r-stats	Base R glm(), lm() without FE — use when FE absorption is not needed
gt	Publication-quality tables — use gt/modelsummary for formatted regression output (alternative to etable)

Quick Decision Trees

"I need to run a regression"

What kind of regression?
├─ OLS with fixed effects → ./references/quickstart.md
├─ OLS without fixed effects → ./references/quickstart.md
├─ IV / 2SLS with FE → ./references/iv.md
├─ Poisson (count data) with FE → ./references/models.md
├─ Logit / Probit with FE → ./references/models.md (feglm)
├─ Negative binomial with FE → ./references/models.md (fenegbin)
├─ Multiple models at once → ./references/quickstart.md (sw/csw)
└─ Nonlinear custom model → ./references/models.md (feNmlm)

"I need difference-in-differences"

DiD design?
├─ Simple 2×2 DiD (one treatment date) → ./references/did.md
├─ Staggered treatment timing → ./references/did.md
│   ├─ Sun-Abraham saturated (sunab) → ./references/did.md
│   └─ TWFE (caution with heterogeneity) → ./references/did.md
├─ Event study plot → ./references/did.md + ./references/reporting.md
└─ Parallel trends assessment → ./references/did.md

"I need to choose standard errors"

What inference?
├─ Heteroskedasticity-robust (HC1) → ./references/standard-errors.md
├─ Clustered (one-way / two-way) → ./references/standard-errors.md
├─ Newey-West (HAC) → ./references/standard-errors.md
├─ Driscoll-Kraay (panel with cross-sect dependence) → ./references/standard-errors.md
├─ Conley spatial → ./references/standard-errors.md
└─ Small sample corrections (ssc) → ./references/standard-errors.md

"I need to present results"

Presenting results?
├─ Regression table (multiple models) → ./references/reporting.md
├─ Coefficient plot → ./references/reporting.md
├─ Event study plot → ./references/reporting.md
├─ LaTeX table output → ./references/reporting.md
└─ Extract fixed effects → ./references/reporting.md

"Something isn't working"

Having issues?
├─ Different results from old code → ./references/gotchas.md
├─ Singleton warnings → ./references/gotchas.md
├─ Poisson separation/convergence → ./references/gotchas.md
├─ Formula parsing errors → ./references/gotchas.md
├─ pyfixest vs fixest differences → ./references/gotchas.md
└─ Collinearity with FE → ./references/gotchas.md

File-First Execution in Research Workflows

Important: In DAAF research pipelines, fixest regressions are executed through script files, not interactively. This ensures auditability and reproducibility.

The pattern:

1. Write regression code to scripts/stage8_analysis/{step}_{task-name}.R
2. Execute via Bash with automatic output capture wrapper script
3. Validation results get automatically embedded in scripts as comments
4. If failed, create versioned copy for fixes

Closely read agent_reference/SCRIPT_EXECUTION_REFERENCE.md for the mandatory file-first execution protocol covering complete code file writing, output capture, and file versioning rules. All regression scripts must follow the Inline Audit Trail (IAT) standard — see agent_reference/INLINE_AUDIT_TRAIL.md. For regression code, document model specification choices (why this estimator, why this clustering level, what identifying assumptions) with # INTENT:, # REASONING:, and # ASSUMES: comments.

See:

• agent_reference/WORKFLOW_PHASE4_ANALYSIS.md — Stage 8 (Analysis & Visualization)
• agent_reference/INLINE_AUDIT_TRAIL.md — IAT documentation standard

The examples below show fixest syntax. In research workflows, wrap them in scripts following the file-first pattern.

---

Quick Reference

Essential Setup

library(fixest)
library(arrow)  # For parquet I/O (DAAF convention)

Core Estimation Functions

Function	Purpose
feols(y ~ x | fe, data)	OLS with fixed effects
fepois(y ~ x | fe, data)	Poisson with fixed effects
feglm(y ~ x | fe, data, family)	GLM (logit, probit, etc.) with fixed effects
fenegbin(y ~ x | fe, data)	Negative binomial with fixed effects
feNmlm(fml, data, family)	Nonlinear models with fixed effects

Formula Syntax Quick Reference

Pattern	Meaning	Example
y ~ x1 + x2	No FE	y ~ educ + exper
y ~ x | fe1 + fe2	With FE	y ~ educ | state + year
y ~ x | fe | x_endo ~ z	FE + IV	y ~ exper | state | educ ~ college_prox
i(factor, ref = val)	Categorical with ref	y ~ i(year, ref = 2000) | state
sunab(cohort, period)	Sun-Abraham DiD	y ~ sunab(cohort, period) | id + period
sw(x1, x2)	Stepwise alternatives	y ~ sw(educ, exper) | state
csw0(x1, x2)	Cumulative stepwise	y ~ csw0(educ, exper) | state
c(y1, y2) ~ x	Multiple outcomes	c(wage, hours) ~ educ | state

Post-Estimation Essentials

fit <- feols(y ~ x1 + x2 | fe, data = df)

summary(fit)                             # Print results
summary(fit, vcov = "hetero")            # Re-estimate with robust SEs
summary(fit, vcov = ~state)              # Clustered SEs
coef(fit)                                # Named vector of coefficients
se(fit)                                  # Standard errors
confint(fit)                             # Confidence intervals
predict(fit)                             # Fitted values
resid(fit)                               # Residuals
fixef(fit)                               # List of FE estimates
r2(fit, type = "r2")                     # R-squared
r2(fit, type = "wr2")                    # Within R-squared
fitstat(fit, type = "ivf")               # First-stage F (IV only)

Reporting

etable(fit1, fit2, fit3)                # Console regression table
etable(fit1, fit2, tex = TRUE)          # LaTeX output
coefplot(fit1, fit2)                    # Coefficient plot
iplot(fit)                              # Event study / interaction plot

Topic Index

Topic	Reference File
First regression	./references/quickstart.md
Formula syntax	./references/quickstart.md
Multi-estimation (sw, csw, csw0)	./references/quickstart.md
Multiple outcomes (c(y1,y2))	./references/quickstart.md
Data requirements	./references/quickstart.md
Multi-way fixed effects	./references/fixed-effects.md
FE interactions (^)	./references/fixed-effects.md
Varying slopes	./references/fixed-effects.md
FE recovery (fixef)	./references/fixed-effects.md
Singleton removal	./references/fixed-effects.md
Clustered SEs	./references/standard-errors.md
HC robust SEs	./references/standard-errors.md
Two-way clustering	./references/standard-errors.md
Newey-West (NW)	./references/standard-errors.md
Driscoll-Kraay (DK)	./references/standard-errors.md
Conley spatial	./references/standard-errors.md
Small sample corrections (ssc)	./references/standard-errors.md
IV formula syntax	./references/iv.md
First-stage diagnostics	./references/iv.md
Weak instrument tests	./references/iv.md
TWFE DiD	./references/did.md
Sun-Abraham (sunab)	./references/did.md
Event study plots	./references/did.md
Parallel trends	./references/did.md
etable	./references/reporting.md
coefplot	./references/reporting.md
iplot	./references/reporting.md
LaTeX output	./references/reporting.md
Poisson (fepois)	./references/models.md
GLM with FE (feglm)	./references/models.md
Negative binomial (fenegbin)	./references/models.md
Nonlinear (feNmlm)	./references/models.md
Singleton observations	./references/gotchas.md
Poisson separation	./references/gotchas.md
Formula parsing pitfalls	./references/gotchas.md
SE default changes	./references/gotchas.md
pyfixest vs fixest differences	./references/gotchas.md

Citation

When this library is used as a primary analytical tool, include in the report's Software & Tools references:

Berge, L. (2018). "Efficient estimation of maximum likelihood models with multiple fixed-effects: the R package FENmlm." CREA Discussion Paper 2018-13. Updated as: Berge, L. (2026). fixest: Fast Fixed-Effects Estimations [Computer software]. https://CRAN.R-project.org/package=fixest

Cite when: fixest is used for regression estimation (OLS, Poisson, GLM, IV) or difference-in-differences analysis. Do not cite when: Only loaded but no estimation performed.

For the arXiv methods paper:

Berge, L., Butts, K., & McDermott, G. (2026). "Fast and User-Friendly Econometrics Estimations: The R Package fixest." arXiv:2601.21749.

For method-specific citations (e.g., Sun-Abraham, individual DiD estimators), consult the reference files in this skill and agent_reference/CITATION_REFERENCE.md.