qe-tables-figures

Tables & Figures (qe-tables-figures)

When to trigger

• Tables still mark significance with asterisks or boldface (QE forbids this)
• Exhibits are dense and the key quantity is hard to find
• Figures/tables are collected at the end instead of placed in-text
• You are preparing final files and need the right formats and notes

QE house rules for exhibits (source map refreshed 2026-06-20)

QE applies specific presentation rules that differ from most journals:

• No asterisks, no boldface to denote statistical significance. Instead, report standard errors and confidence / coverage sets. Let magnitude and uncertainty speak; do not encode significance with symbols.
• Place figures and tables in-text on the relevant page, not gathered at the end of the manuscript.
• Final accepted files are preferred as LaTeX source with .eps / .jpg figures; the submitted PDF embeds exhibits in-text.
• The manuscript is 1.5/double-spaced, ≥12pt, ≤32 lines per page — design exhibits to remain legible under those constraints.

Table craft

• Lead with the estimate(s) the paper is about; put the headline quantity where the eye lands first.
• Under each estimate, show its standard error (and a confidence/coverage set where the inference is non-standard, e.g., weak-IV or boundary parameters).
• Self-contained notes: estimand, sample, units, clustering level, estimator, and how uncertainty is computed — so the table stands alone.
• For structural work, separate targeted moments (data vs. model) from untargeted moments used for validation; report parameter estimates with standard errors.

Figure craft (QE is quantitatively visual)

• Event-study plots with leads/lags and confidence bands; model-fit overlays (data vs. simulated moments); counterfactual curves with uncertainty bands.
• Show uncertainty visually (bands / intervals) rather than implying it with significance stars.
• Avoid chartjunk (no 3D, minimal color); vector output for print resolution.
• Caption states what the reader should conclude, with the key number in the text.

Execution bridge (StatsPAI / Stata MCP)

Generate exhibits from the fitted result, not by retyping numbers. Full map: execution-with-mcp. Quantitative Economics spans structural and applied micro; the chain serves its reduced-form lane, structural estimation uses its own toolkit.

• 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.

Checklist

• No asterisks or boldface for significance anywhere in tables or figures
• Every estimate shows a standard error; coverage/confidence sets where inference is non-standard
• Figures and tables placed in-text on the relevant page (not at the end)
• Notes are self-contained (estimand, sample, units, estimator, inference)
• Structural: targeted vs. untargeted moments shown; parameters with SEs
• Final files prepared as .eps / .jpg for the LaTeX source at acceptance
• Exhibits legible under 12pt / 1.5-spacing / ≤32-lines constraints

Anti-patterns

• Marking significance with *** / boldface (QE explicitly forbids this)
• A 9-column regression table with the key estimate buried
• Collecting all figures/tables at the end of the manuscript
• Figures with no uncertainty shown, leaning on stars in a companion table
• Notes that require the body text to interpret the exhibit

Reading an exhibit the way a QE referee does

Worked vignette: rebuilding a structural-fit table (illustrative)

A draft reports a 6-column table marking three coefficients with asterisks and stacking all figures after the references. The QE rebuild: a moment-fit table with two blocks — targeted moments (data 0.31 vs. model 0.30; data 0.18 vs. model 0.19) and untargeted moments used as validation (data 0.44 vs. model 0.41) — and a parameter block showing each estimate with its standard error (e.g., risk aversion 1.8, s.e. 0.3). No stars anywhere; the event-study figure moves in-text beside its discussion with a 95% band on each lead and lag. (Numbers illustrative.) The reader now sees fit, parameters, and uncertainty without decoding symbols.

Referee pushback and the exhibit fix

• "Significance is encoded with stars — remove them." → Replace with standard errors and, where inference is non-standard, a coverage set.
• "I can't tell which moments are targeted." → Split the table into targeted and untargeted blocks; label the validation moments.
• "The figure hides its uncertainty." → Add confidence bands to event-study and counterfactual curves; never imply significance via a companion star table.

Output format

【Significance reporting】asterisk/bold-free; SEs + coverage sets shown? [Y/N]
【Placement】figures/tables in-text on the relevant page? [Y/N]
【Headline quantity】lands first in the main table? [Y/N]
【Structural exhibits】targeted vs. untargeted moments + params w/ SEs? [Y/N or N/A]
【Final files】.eps/.jpg ready for LaTeX source? [Y/N]
【Next step】qe-writing-style