rje-tables-figures

Tables & Figures (rje-tables-figures)

When to trigger

• Turning IO estimates into the article's main exhibits
• Deciding which tables stay in the 40-page main text and which go to the appendix
• Formatting exhibits and equations to the RJE Style Guide

What RJE exhibits carry

As the industrial-organization flagship, RJE expects exhibits that communicate market behavior and welfare, not just regression output:

• Demand / cost parameter tables with standard errors and the estimation method noted.
• Elasticity matrices (own/cross-price) demonstrating sensible substitution.
• Markup / marginal-cost tables tying estimates to conduct.
• Merger-simulation and counterfactual tables reporting price, output, and welfare (consumer/producer surplus) changes, with the maintained assumptions stated in the note.
• Reduced-form work: clean event-study plots (leads and lags, confidence bands) and treatment-effect tables.

Each exhibit should be self-contained: a reader sees the market, the specification, the units, and the assumption set from the note alone.

RJE Style Guide formatting

• Sections numbered consecutively; subsections unnumbered — match table/figure callouts to that scheme.
• Equations numbered flush right, with (1a)/(1b) for multi-part equations.
• Author-date references in notes (no page numbers, no issue numbers).
• Vector output (PDF/EPS) for print; keep figures legible in greyscale; avoid chartjunk.

Page-cap discipline (hard)

Main text <=40 pp, total <=50 pp, double-spaced. Exhibits eat pages fast:

• Keep in the main text the decisive exhibits: headline estimates, the key elasticity/markup table, and the central counterfactual/welfare result.
• Move full parameter dumps, alternative specifications, and robustness exhibits to the appendix (within the <=10-page appendix+references budget).
• Do not offload core exhibits into supporting information, which RJE discourages and may decline.

Exhibit-placement table (main text vs appendix)

Every exhibit must earn its main-text page. Sort by whether it carries the article's core argument or supports it.

Worked vignette: a self-contained counterfactual table

Suppose your central exhibit reports a merger simulation. Built to RJE standards, the table note alone tells the reader everything:

• Rows: merging brands, key rivals, market aggregate. Columns: pre-merger price, predicted post-merger price, % change, consumer-surplus change, producer-surplus change.
• Illustrative content: merging brands +4.2% price, rivals +1.1%, consumer surplus -$31m/year, producer surplus +$44m/year, with a bracketed range across demand specifications.
• Note (self-contained): states the demand model, the conduct assumption (Bertrand-Nash), that the product set is held fixed, that costs are recovered from FOCs, and the units (annualized dollars), plus author-date sources with no page or issue numbers.

A reader who sees only this table should grasp the market, the specification, the maintained assumptions, and the welfare verdict — no hunting through the text.

Referee-pushback patterns and the venue fix

• "The welfare table has no maintained-assumption note." Fix: state the conduct, the fixed-product-set assumption, and the demand specification in the note itself.
• "The elasticity matrix implies implausible substitution." Fix: flag the suspect cells, explain the market reason, or revisit the random-coefficient specification driving them.
• "Main-text tables blow the page cap." Fix: keep one decisive table per claim in the body and route full parameter dumps to the appendix within the budget.
• "Figures are illegible in greyscale print." Fix: use vector output, distinguish series by line style not color, and avoid chartjunk.

Execution bridge (StatsPAI / Stata MCP)

Generate exhibits from the fitted result, not by retyping numbers. Full map: execution-with-mcp. RAND is industrial organization — endogeneity of prices/entry and structural demand; the reduced-form chain for causal claims, structural IO outside it.

• Tables: etable (multi-model) or did_summary_to_latex straight from the result_id.
• Figures: plot_from_result / enhanced_event_study_plot / event_study_table — axis units and the SE/clustering note baked in.
• Every note names the estimator + clustering and states the magnitude in interpretable units.

See a full fitted-result → exhibit chain in the JF execution walkthrough.

Anti-patterns

• A wall of parameter tables in the main text blowing the 40-page cap
• Counterfactual/welfare tables with no maintained-assumption note
• Elasticity matrices that imply implausible substitution, unexplained
• Numbered subsections or inline equation numbers (off RJE style)
• Low-resolution raster figures illegible in print/greyscale

Output format

【Main-text exhibits】[headline estimates, key elasticity/markup, central counterfactual]
【Appendix exhibits】[full params, robustness, alt specs]
【Self-contained notes】units + specification + assumptions? [Y/N]
【Style】sections numbered/subsections not; eqns flush-right (1a)/(1b)? [Y/N]
【Page budget】main __/40, total __/50 — within cap? [Y/N]
【Next step】rje-writing-style