| name | revacc-tables-figures |
| description | Use when the exhibits are the bottleneck for a Review of Accounting Studies (RAST) manuscript — building self-contained accounting panel tables, event-study and equilibrium figures, and a descriptive-statistics block that survives referee scrutiny. Builds the exhibits; it does not run the analysis (revacc-data-analysis) or polish the surrounding prose (revacc-writing-style). |
Tables & Figures (revacc-tables-figures)
When to trigger
- Tables are dense, inconsistent, or not self-explanatory without the text
- Your descriptive-statistics table is missing or does not let a referee sanity-check the sample
- A DiD or event study has no dynamic/event-time figure
- An analytical paper has propositions but no figure illustrating the comparative statics
- Significance is shown only with asterisks, or coefficients lack the economic-magnitude read
The standard RAST exhibit set
Accounting empirical papers at RAST converge on a recognizable sequence; build it in order.
- Variable definitions — every variable defined with its data source (Compustat item, CRSP field, I/B/E/S measure) so the sample is reconstructable.
- Descriptive statistics — N, mean, median, SD, key percentiles for all variables; this is the table referees use to catch a broken sample (implausible accruals, miswinsorized ratios).
- Correlations — Pearson/Spearman where it informs multicollinearity or the main association.
- Main result — the focal estimating equation with fixed effects and clustering noted in the table notes; report economic magnitude, not only significance.
- Identification diagnostics — pre-trends/dynamic effects (DiD), first stage (IV), bandwidth/density (RD).
- Robustness and cross-section — alternative proxies, subsamples, channel partitions.
For analytical papers, the exhibit set is figures: an equilibrium/comparative-static plot that makes the proposition legible, and a stylized numerical example.
Make every exhibit self-contained
A RAST table must be readable without the body text. The title states what is estimated; the notes give the sample, period, unit of observation, fixed effects, clustering, winsorization, and what significance markers mean. A referee should reconstruct the specification from the table alone. Report coefficients with standard errors (or t-stats) consistently, and translate the headline coefficient into an economic magnitude ("a one-SD increase in disclosure quality lowers the bid-ask spread by X%").
Figures that earn their place
- Event-time / dynamic-effect plot for any DiD — the single most persuasive identification exhibit; show the flat pre-trend and the post-treatment path with confidence bands.
- Event-study CAR plot for information-content/value-relevance claims, with the window and benchmark stated.
- Comparative-static figure for analytical papers — plot the equilibrium object against the key primitive so the accounting reading is visible.
- Avoid chartjunk and dual axes; one figure, one message.
House-style discipline
Follow the journal's formatting (待核实; 检索于 2026-06；以官网为准): consistent decimal places, a stated significance convention, numbered tables/figures referenced in order, and an abstract within the journal's limit (~150–250 words, 待核实). Keep exhibits anonymized for double-blind review (no author-identifying file names or acknowledgements embedded).
Execution bridge (StatsPAI / Stata MCP)
Generate exhibits from the fitted result, not by retyping numbers (the usual source of
body-vs-appendix drift). Full map: execution-with-mcp. RAST is empirical accounting; emphasize identification of disclosure / governance effects and the multiple-testing haircut for mined associations.
- Tables:
etable (multi-model columns) 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 effect size in
interpretable units.
See a full fitted-result → exhibit chain in the JF execution walkthrough.
Checklist
Anti-patterns
- Asterisks-only inference with no standard errors and no economic magnitude.
- Table dependent on the text — notes too thin to reconstruct the specification.
- No descriptive-statistics table, hiding a broken or implausible sample.
- A DiD with no event-time plot, leaving pre-trends unshown.
- Chartjunk / dual axes that obscure the one message a figure should carry.
- Identifying file names or acknowledgements breaking anonymization.
Output format
【Exhibit set】var-defs / descriptives / correlations / main / diagnostics / robustness
【Main table】FE + clustering in notes; economic magnitude reported? yes/no
【Identification figure】DiD event-time / IV first stage / RD plot
【Analytical figure】comparative static illustrated? yes/no
【Self-contained?】each exhibit readable alone; anonymized for double-blind
【House style】decimals/significance convention/abstract limit (待核实)
【Next skill】revacc-writing-style
