| name | jf-tables-figures |
| description | Use when finalizing tables and figures for a The Journal of Finance (JF) manuscript for the AFA flagship. Enforces accessible, self-contained, general-interest exhibits with Roman-numeral tables, economic-magnitude reporting, and matching Internet Appendix numbering; it does not decide which results to run. |
Tables & Figures (jf-tables-figures)
When to trigger
- Tables/figures are drafted but notes, units, or formatting are inconsistent
- A reader cannot interpret an exhibit without hunting through the text
- You are aligning main-text exhibits with Internet Appendix exhibits
JF norm: self-contained, accessible exhibits
JF is general-interest and prizes accessibility. Each exhibit must be self-contained: a reader from another subfield should understand it from its caption and notes alone. Because the manuscript faces a 60-page limit (≥1.5 spacing, 12-pt font), keep only decisive exhibits in the body and send the rest to the Internet Appendix (bundled at the end of the same PDF, labeled IA.*, not counted against 60 pages; see jf-internet-appendix).
Exhibit standards
- Numbering: JF uses Roman numerals for tables (Table I, II, …) and Arabic for figures (Figure 1, 2, …); the Internet Appendix mirrors this as
IA.I, IA.1. Keep the schemes consistent.
- Notes state the inference: sample, period, units, and the standard-error/clustering convention belong in every table note.
- Economic units: report magnitudes in interpretable units (bps, % of market cap, Sharpe gain), consistent with JF's general-interest voice.
- No orphan exhibits: every table/figure (body or IA) is referenced from the text.
Checklist
What a JF exhibit must carry (calibration anchors)
JF exhibits answer to a general-interest reader, so the table itself, not the surrounding prose, must deliver the economics. Useful calibration anchors for the flagship, hedged where conventions vary by subfield:
| Exhibit element | JF expectation (illustrative anchor) |
|---|
| Body exhibit count | Typically a lean set of decisive tables/figures; the exhaustive grid lives in the Internet Appendix |
| Magnitude column | A column or note giving the effect in bps, % of market cap, or Sharpe gain — not stars alone |
| Inference convention | Every table note names the SE estimator and clustering dimension(s) |
| Numbering | Roman tables (I, II, …), Arabic figures (1, 2, …), IA mirrors as IA.* |
| Self-containment | Caption + notes let an out-of-subfield AFA reader interpret the exhibit |
Exact body-exhibit counts are not a fixed rule; confirm against the journal's current author guidelines and recent issues, but the cultural signal at JF is "few, decisive, self-explaining in the body; everything else online."
Worked vignette — formatting a momentum horse-race table
Illustrative numbers. A corporate-finance reader picks up a Table II reporting a long-short anomaly. The raw spread is 0.62% per month. The amateur version prints only coefficients and three stars. The JF-grade version, by contrast:
- Labels the row in economic units — "long–short return, % per month" — and adds a note: "55 bps after Fama–French five-factor + momentum adjustment (illustrative), annualized Sharpe gain ≈ 0.4."
- States inference in the note: "Newey–West (12 lags) t-statistics in parentheses; sample 1972–2023 (illustrative); NYSE breakpoints, value-weighted."
- Sends the equal-weighted, alternative-breakpoint, and decile-by-decile versions to the Internet Appendix as
IA.II, IA.III, cited from the text.
The reader now understands the magnitude, the inference, and where to find more — without leaving the page. That self-containment is the JF house standard.
Referee-pushback patterns and the JF-specific fix
| Pushback on an exhibit | JF-specific fix |
|---|
| "I can't tell what 0.62 means economically" | Add a magnitude note (bps/month, Sharpe gain, % market cap) |
| "Which standard errors are these?" | Name the estimator and clustering in every table note |
| "Table III and IA.3 use different definitions" | Reconcile; keep one variable definition across body and IA |
| "The body has fifteen tables" | Move all but the decisive ones to the Internet Appendix |
Execution bridge (StatsPAI / Stata MCP)
Generate the exhibit from the fitted result, don't transcribe numbers by hand (the
most common source of body-vs-IA inconsistency). Full map:
shared-resources/empirical-methods/execution-with-mcp.md.
- Regression tables:
etable (multi-model columns) or did_summary_to_latex /
did_summary_to_markdown straight from the result_id — one variable definition,
one set of numbers, body and Internet Appendix in sync by construction.
- Event-study / coefficient figures:
plot_from_result, enhanced_event_study_plot,
event_study_table — emit the figure with axis units and the SE/clustering note baked in.
- Every exhibit note must name the estimator and clustering (pull it from the
result's diagnostics) and state the economic magnitude in interpretable units —
the JF self-containment standard, now machine-generated rather than retyped.
For a full chain producing JF-ready tables/figures from a fitted DiD result, see
resources/worked-examples/02-execution-walkthrough.md.
If StatsPAI/Stata are not connected, format from the vendored resources/code/ outputs.
Anti-patterns
- A table whose meaning is unclear without three paragraphs of text
- Significance stars with no economic magnitude
- Missing SE/clustering convention in the note
- Body crowded with exhibits that belong in the Internet Appendix, breaching the 60-page limit
- Inconsistent numbering between body and appendix
- A figure with no axis units or a caption that assumes subfield-specific notation
Output format
【All exhibits self-contained?】yes / no
【Notes state SE/clustering + units?】yes / no
【Magnitudes in economic units?】yes / no
【JF numbering consistent (Table I.. / IA.*)?】yes / no
【Decisive-only in body, ≤60 pp?】yes / no
【Next step】jf-writing-style
