Findings & checks · 0 flagged
Every scanned point with the score it earned and what moved between them.
First recorded scan — no prior version to compare against.
The primary manifest — the file an agent reads to learn what this artifact does.
Economic-Research Skills
<p align="center"> <img src="assets/cover.jpg" alt="《经济研究》 journal cover" width="220"> </p>
English | 简体中文
Agent skill stack for manuscripts targeted at 《经济研究》 (Economic Research Journal) — the top economics journal in China (published by the Institute of Economics, CASS; ISSN 0577-9154; monthly; founded 1955). 18 skills span the full lifecycle from topic selection to R&R rebuttal, shipping a runnable Stata + Python code library.
The stack is dual-mode — it both diagnoses a manuscript's bottleneck (templates, checklists) and generates publication-grade draft passages (introduction, mechanism, abstract, response letter) for you to refine.
It is not a generic Chinese-economics writing toolbox. It is an Economic-Research-specific methodology stack.
Why a Separate Economic-Research Skill Stack?
《经济研究》 imposes constraints that differ materially from AER / AEJ / top finance journals (rows marked [official] are verified against the journal's own sources):
| Constraint | Economic Research | Implication |
|---|---|---|
| Discipline | Economics (macro / institutional / econometric / industry) | Pure-management or pure-case papers are off-fit |
| Topic | Theory + China problem; anti "quant-for-its-own-sake" [official] | "Policy evaluation only" reads as a working paper |
| Contribution sentences | 3–5 explicit "边际贡献" bullets in intro | Cannot be a single paragraph |
| Literature review | Bilingual; canonical theory citations required | Missing canonical theory refs is a desk-reject signal |
| Identification | Quasi-experimental; modern estimators expected | TWFE-only / OLS + controls often desk-rejected |
| Mechanism | Near-mandatory; step-wise mediation retired (江艇 2022) | Estimate D→M, argue M→Y from theory — no endogenous mediation regression |
| References | Author-year (作者，年份) [official] | Not numbered [1][2] / [J][M] style |
| Abstract | ~300-char 中文提要 + 5-element English Summary [official] | English Summary needs a Chinese mirror for editing (not published; official sets no length) |
| Keywords | 3–5 + JEL codes | Not 3–8 |
| Policy implications | "Significance" level, not action memos | Different from 《管理世界》 actionable-policy style |
| Authorship | ≤5 authors, ≤2 affiliations, one corresponding [official] | Author info on a separate page; body anonymized |
Every fact above is verified and sourced in resources/official-source-map.md. Where the journal does not publish a hard limit (word/table/formula counts), the tools treat numbers as "rules of thumb — defer to the current author guidelines," never as hard assertions.The 18 Skills
Framing: er-workflow (router) · er-topic-selection · er-introduction · er-literature-review · er-theory-hypotheses
Empirical core: er-data-sample · er-identification · er-mechanism · er-heterogeneity · er-robustness · er-tables-figures
Polish & submission: er-policy-implication · er-abstract · er-style · er-reviewer-lens · er-reproducibility · er-submission · er-rebuttal
| Skill | Purpose |
|---|---|
er-workflow | Router — decides which sub-skill to invoke next |
er-topic-selection | Theoretical-grounding fit + contribution-sentence template |
er-introduction | Five-part funnel intro, hook patterns, contribution teaser |
er-literature-review | Bilingual structure + canonical-theory checklist + conversational review |
er-theory-hypotheses | Theory section + testable hypothesis derivation |
er-data-sample | Data note, formula-based variable table, descriptive stats |
er-identification | Modern quasi-experimental design (staggered DID / IV / RDD / DML) |
er-mechanism | Mechanism analysis — post-江艇(2022): D→M + theory for M→Y |
er-heterogeneity | Heterogeneity cuts: five-dimension priority + theory |
er-robustness | Robustness system organized by identification threat |
er-tables-figures | Three-line tables, variable table, figure aesthetics (code-exported) |
er-policy-implication | Significance-level policy framing (not action-level) |
er-abstract | 中文提要 + 5-element English Summary + blacklist scrub |
er-style | Language polish: empty-significance → concrete contribution |
er-reviewer-lens | Pre-submission self-audit of high-frequency reviewer critiques |
er-reproducibility | One-click replication package + data-availability statement |
er-submission | Pre-submission checklist + manuscript template |
er-rebuttal | R&R response-letter structure |
Runnable Code Library (Stata + Python)
resources/code/ is a copy-and-run, one-click-reproducible empirical skeleton; command syntax is verified:
code/
├── stata/00_master.do one-click master + fixed seed
├── stata/01_clean.do merge, sample filtering, winsorize
├── stata/02_descriptive.do summary stats + balance (three-line)
├── stata/03_did_modern.do TWFE → Bacon → CS/BJS/SA/dCDH → event study
├── stata/04_iv.do KP rk F + effective F + AR weak-IV-robust CI
├── stata/05_rdd.do rdrobust bias-corrected CI + rddensity
├── stata/06_dml.do ddml + pystacked multi-learner
├── stata/07_mechanism.do D→M only (江艇 2022 stance)
├── stata/08_robustness.do by-threat checks + placebo dist + wild bootstrap
├── stata/09_tables.do esttab three-line tables + event-study figure
└── python/ clean_panel / dml_doubleml / event_study_plotWorked-Example Library (publication-grade, one case end-to-end)
resources/worked-examples/ threads one coherent empirical case ("Golden Tax Phase III and firm TFP") through every section — introduction, theory & hypotheses, identification, mechanism (江艇 2022 stance), heterogeneity, robustness, abstract, policy, and an R&R response letter — with internally consistent numbers you can hold up as a target when revising a draft.
The policy is real; all coefficients are illustrative and come from no published paper (see the directory's README disclaimer). It demonstrates the "generation" mode's quality bar: "revise my draft to look like this."
Quick Start
/plugin marketplace add https://github.com/brycewang-stanford/Awesome-Journal-Skills
/plugin install economic-research-skills
/reload-pluginsOr copy manually: mkdir -p ~/.claude/skills && cp -R skills/er-* ~/.claude/skills/
First prompt:
Use er-workflow to tell me which skill I should use next for my Economic-Research manuscript.Methodological stance (what sets this apart from a generic writing assistant)
- Mechanism follows 江艇 (2022): drops step-wise mediation / Sobel / bootstrap indirect-effect shares; estimate D→M and argue M→Y from theory.
- Staggered DID must report a heterogeneity-robust estimator: TWFE-only is flagged high-risk; ready-made commands for CS / BJS / SA / dCDH + Bacon decomposition + event study.
- Strict author-year references: scrubs
[1][2]/[J][M]/[27,28,29]numbered-citation residue. - Exemplars only from verified papers: avoids misattributing 《管理世界》 / 《中国工业经济》 classics to 《经济研究》 (see the source-map's pitfall table).
What this repo does NOT do
- It does not judge whether your theoretical contribution is genuinely original — the researcher's call
- It does not simulate a specific reviewer (
er-reviewer-lensonly distills high-frequency, structural critiques for self-audit) - It does not catalog rejection rates / impact factors
- Generated passages are drafts — verify and own them
Related
- Awesome-Journal-Skills — Index of journal-specific skill packs
- AER-skills — American Economic Review
License
MIT