Publication-Ready Tables in Stata: esttab, outreg2, and collect

You have six regression models. You need a publication-ready table with stars, standard errors, model comparisons, and formatted headers. In LaTeX. And Word, for co-authors who refuse to use LaTeX.

You have three options in Stata: esttab (the modern standard), outreg2 (the legacy option), and collect (Stata 17+’s new framework). Here’s when to use each.

esttab: The Standard

Key options:

• replace — overwrites existing file (essential for reproducibility)
• se — reports standard errors below coefficients (vs. t for t-statistics)
• booktabs — uses LaTeX booktabs package for clean horizontal rules
• label — uses variable labels instead of names
• stats() — adds summary statistics rows at the bottom

Word/RTF Output

outreg2: The Legacy Option

outreg2 appends results model-by-model instead of storing them first. It works but produces less customizable output than esttab. It’s still widely used because many professors learned it first.

collect: Stata 17+

The collect framework is Stata’s official modern approach. It’s powerful and flexible but has a steeper learning curve than esttab. For most users, esttab remains the faster path to a publication table.

Pro Tips

• Add custom rows: Use indicate() to show which FE are included: indicate("Firm FE=1.firm" "Year FE=1.year")
• Fragment mode: Use fragment when embedding in a larger LaTeX document that already has tabular environment.
• Coefficient map: Use order() and keep() to control which variables appear and in what order.
• Regression table automation: Write a loop that runs all specifications and stores estimates, then call esttab once at the end. This ensures your table is always regenerable from code.