Skip to main content
QUICK REVIEW

[論文レビュー] Fast and user-friendly econometrics estimations: The R package fixest

Laurent R. Bergé, Kyle Butts|arXiv (Cornell University)|Jan 29, 2026
Data Analysis with RComputer Science被引用数 0
ひとこと要約

Introduces fixest, an R package for fast, flexible econometric estimation with a focus on fixed-effects models, offering a unified syntax for OLS, IV, GLM, MLE, and Diff-in-Diff, plus on-the-fly inference and publication-ready outputs.

ABSTRACT

fixest is an R package for fast and flexible econometric estimation, providing a comprehensive toolkit for applied researchers. The package particularly excels at fixed-effects estimation, supported by a novel fixed-point acceleration algorithm implemented in C++. This algorithm achieves rapid convergence across a broad class of data contexts and further enables estimation of complex models, including those with varying slopes, in a highly efficient manner. Beyond computational speed, fixest provides a unified syntax for a wide variety of models: ordinary least squares, instrumental variables, generalized linear models, maximum likelihood, and difference-in-differences estimators. An expressive formula interface enables multiple estimations, stepwise regressions, and variable interpolation in a single call, while users can make on-the-fly inference adjustments using a variety of built-in robust standard errors. Finally, fixest provides methods for publication-ready regression tables and coefficient plots. Benchmarks against leading alternatives in R, Python, and Julia demonstrate best-in-class performance, and the paper includes many worked examples illustrating the core functionality.

研究の動機と目的

  • Motivate the need for fast and flexible econometric estimation in applied research, especially with fixed effects.
  • Present fixest as a unified, speed-optimized toolkit for multiple econometric models.
  • Showcase the usability and inference features, including robust standard errors and publication-ready outputs.
  • Demonstrate performance benchmarks against other software and provide worked examples.

提案手法

  • Describe the core architecture: optimized parallel C++ code via Rcpp for speed.
  • Introduce a fixed-effects estimation approach with a novel fixed-point acceleration algorithm.
  • Present a unified formula interface enabling multiple estimations, stepwise regressions, and interpolation in a single call.
  • Explain post-estimation flexibility for on-the-fly standard-error adjustments and various vcov options.
  • Showcase functions feols, fepois, feglm, femlm for different model families and their shared syntax.
  • Demonstrate table and plot utilities (etable, coefplot) for publication-ready outputs.
Figure 1 : Default coefplot graph for split sample estimation.
Figure 1 : Default coefplot graph for split sample estimation.

実験結果

リサーチクエスチョン

  • RQ1How can fixed-effects estimations be performed rapidly across varied data contexts?
  • RQ2Can a unified syntax support multiple econometric models with consistent inference and output tools?
  • RQ3What are the computational advantages of fixest's fixed-point acceleration over traditional demeaning approaches?
  • RQ4How does fixest perform relative to leading alternatives in R, Python, and Julia across model families?
  • RQ5What utilities does fixest provide for on-the-fly inference adjustments and presentation of results?

主な発見

  • fixest achieves fast fixed-effects estimation using a novel fixed-point acceleration algorithm implemented in C++.
  • A common syntax enables OLS, IV, GLM, MLE, and difference-in-differences estimators with fixed-effects in a single framework.
  • Post-estimation tools allow quick on-the-fly adjustment of standard errors and multiple vcov specifications.
  • Estimations with fixed effects are faster when using the fixest specialized routine for nuisance parameters (individual and time effects).
  • The package provides publication-ready tables and plots (etable, coefplot) and supports multi-estimation outputs in a single call.
Figure 2 : Default iplot graph for DiD estimations.
Figure 2 : Default iplot graph for DiD estimations.

この論文を読む

Nubintのビューアで原文を開き、読みながら気になる箇所をすぐに質問できます。

Nubintでこの論文を読む

このレビューはAIが論文の原文をもとに作成しました。引用する前に原文をご確認ください。