Tables & Figures (aejmac-tables-figures)

Tables & Figures (aejmac-tables-figures)

When to trigger

• IRF figures lack confidence bands, units, or a clear horizon axis
• Tables are dense, mislabeled, or report coefficients without the macro quantity of interest
• A model-fit comparison (data vs. simulated moments) is asserted in text but not shown
• Exhibits do not stand alone — the reader needs the body text to decode them

The AEJ: Macro exhibits bar

AEJ: Macro follows AEA house style (applied by the AEA at copyediting; the submitted PDF need not pre-conform, but clean exhibits speed review). Macro adds its own conventions: the impulse-response function is the workhorse figure, and the headline quantity (multiplier, peak response, welfare cost, share of variance) must be readable off the exhibit. Every exhibit should be self-contained: title, axis labels with units and horizon, sample, and the inference object (bands / SE) in the note.

Figure conventions (macro-specific)

• IRFs: plot the response with a shaded confidence band (state the level, e.g., 68% and/or 90%); label the horizon axis in the natural unit (quarters/months/years); mark zero; if sign-restricted, show the identified set / median-target, not a spurious point.
• Fan charts / forecast bands: show the full predictive distribution where the claim is about uncertainty.
• Model-fit overlays: plot data vs. model (targeted and untargeted moments) on the same axes; this is the credibility figure for quantitative work.
• Counterfactual figures: baseline vs. counterfactual paths with uncertainty carried through.
• Variance decompositions / contribution plots: stacked or grouped, with a clear legend.
• Vector output (.eps/.pdf) preferred for final files; keep colors legible in grayscale; no chartjunk.

Table conventions

• Report the economic quantity, not just raw coefficients — e.g., the multiplier, the elasticity, the implied share — with its SE/band.
• AEA tables conventionally use significance stars; AEJ: Macro accepts them, but the standard error / band must be present and must carry the inference (do not let stars substitute for a reported SE). Note significance levels in the table note.
• One decisive number per row; align decimals; units in the column header or note.
• Targeted-vs-untargeted-moment tables: clearly flag which moments were matched.
• Notes carry: sample, frequency, estimator, inference (HAC/cluster), and what an asterisk means.

Main text vs. online appendix

• Main text: the decisive IRFs, the headline table, the model-fit figure. AEJ: Macro articles run ~40 pages including exhibits (检索于 2026-06；以官网为准), so exhibits compete for space.
• Online appendix: full robustness IRFs, additional moments, alternative specifications, derivations.

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.