build-tables

name: build-tables description: Combine saved Stata estimates into publication-ready tables via esttab. Produces both .tex (for paper) and .csv (for audit) with consistent formatting. disable-model-invocation: true argument-hint: "[table-name or estimates-list]" allowed-tools: ["Bash", "Read", "Edit", "Write", "Grep", "Glob"]

Build Publication-Ready Tables

Take a set of saved estimates and produce a single table in .tex (for the paper) and .csv (for audit / sharing) with the project's standard formatting.

When to Use

• After running dofiles/03_analysis/*.do that produces est store m_<name> results
• When assembling a multi-spec table (main + alt outcome + alt cluster + alt FE)
• Before rendering the report — tables must exist in output/tables/ first

Steps

1. Identify the estimates in $ARGUMENTS:

   • If a table name (e.g., main_regression): search dofiles/03_analysis/ for est store m_main* and assemble
   • If an explicit estimates list (e.g., m_ols m_iv m_did): use those
   • If empty: ask the user which table to build

2. Locate the producing do-file that has the est store calls. The do-file should also do the esttab export. If it doesn't, write a helper do-file in dofiles/04_output/<table>_assemble.do.

3. Compose the esttab call with project conventions:

   estimates restore m_main
   estimates restore m_alt_cluster
   estimates restore m_alt_fe
   
   esttab m_main m_alt_cluster m_alt_fe ///
       using "output/tables/<name>.tex", replace ///
       se star(* 0.10 ** 0.05 *** 0.01) ///
       booktabs label collabels(none) ///
       stats(N r2_within mean_dep, ///
             labels("Observations" "Within R-sq" "Mean of dep var") ///
             fmt(%9.0fc %9.3f %9.3f)) ///
       drop(_cons) ///
       title("<table title>") ///
       addnotes("Standard errors clustered at <level>." ///
                "Significance: * p<0.10, ** p<0.05, *** p<0.01.")
   
   esttab m_main m_alt_cluster m_alt_fe ///
       using "output/tables/<name>.csv", replace ///
       se star(* 0.10 ** 0.05 *** 0.01) plain ///
       stats(N r2_within mean_dep)
   

4. Run the do-file via /run-stata.

5. Verify outputs:

   • Both .tex and .csv exist in output/tables/
   • Read the .csv and spot-check coefficients are sensible
   • Confirm the .tex includes N, R², mean dep var, cluster info, significance stars

6. Report: path of new .tex and .csv, the spec each column represents, a one-line summary of the headline coefficient.

Examples

• /build-tables main_regression → assembles output/tables/main_regression.{tex,csv} from m_main-prefixed estimates.
• /build-tables m_ols m_iv m_did → assembles a 3-column table from those specific saved estimates.

Troubleshooting

• "estimates ... not found" — est store m_<name> was never run. Re-run the producing do-file.
• Missing mean_dep — add estadd ysumm after each reghdfe call in the producing do-file.
• esttab not installed — ssc install estout, replace.
• Long table names break LaTeX — use the label option and define short labels via label var.

Notes

• Tables ALWAYS go to BOTH .tex and .csv — .tex is for the paper, .csv is for reviewers / coauthors who don't speak LaTeX.
• Never hand-edit the produced .tex; the next pipeline run will overwrite it. Adjust esttab options instead.
• Significance stars: * p<0.10, ** p<0.05, *** p<0.01 is the project default. Override only if the journal requires different.