jpube-tables-figures

npx skills add brycewang-stanford/Awesome-Journal-Skills --skill jpube-tables-figures --agent claude-code

Same command for any agent — swap --agent for codex, cursor, copilot.

# Tables & Figures (jpube-tables-figures) ## When to trigger - The identifying variation is buried in a regression table instead of shown in a graph - A bunching, RD, or event-study result has no picture - Tables are dense, under-noted, or not callable in order - You need exhibits that make a policy elasticity legible to a referee ## Why figure-forward at JPubE JPubE's identification often *is* a picture — a spike of excess mass at a tax kink, a jump at an eligibility cutoff, a clean break at a reform date. Because referees are public-finance specialists assessing design credibility, **the headline of a JPubE empirical paper is frequently one transparent graph** that lets the reader see the response before any regression. Build exhibits so the design is self-evident. ## Exhibit norms - **Lead with the design figure.** Bunching: observed density vs. smooth counterfactual around the kink/notch, with the excluded region marked. RD/RKD: binned scatter with the fitted discontinuity/kink and CIs. DID: event-study plot with leads and zero-line. - **Show, then estimate.** A figure that makes the response visible earns more trust than a coefficient; the table quantifies what the figure show