Use when the identification strategy is the bottleneck for a 《经济学(季刊)》 (China Economic Quarterly, CEQ) manuscript — either a structural model with credible parameters, or a clean quasi-experiment (DID / IV / RDD / event-study). CEQ is the strictest Chinese econ journal here; identification assumptions must be made explicit and each defended. Stress-test the design before drafting exhibits.
Scanned 6/4/2026
Install to Claude Code
npx -y skills add brycewang-stanford/Awesome-Journal-Skills --skill ceq-identification --agent claude-codeInstalls into .claude/skills of the current project.
Are you the author of Ceq Identification?
Add the live security badge to your README — it updates automatically with every re-scan.
[](https://www.skillsdirectory.com/skills/brycewang-stanford-ceq-identification)More formats (shields.io, HTML) on the badges page.
---
name: ceq-identification
description: Use when the identification strategy is the bottleneck for a 《经济学(季刊)》 (China Economic Quarterly, CEQ) manuscript — either a structural model with credible parameters, or a clean quasi-experiment (DID / IV / RDD / event-study). CEQ is the strictest Chinese econ journal here; identification assumptions must be made explicit and each defended. Stress-test the design before drafting exhibits.
---
# 识别策略(ceq-identification)
## 触发时机
- 实证主体仅 OLS + 控制变量,把相关性当因果
- 有处理/外生变异,但识别假设没写清、没逐条辩护
- 拿不准走结构估计还是准实验路线
## CEQ 的铁律:要么结构,要么准实验
CEQ 不接受"加一堆控制变量就声称因果"。两条合法路径:
- **结构估计**:微观基础清晰、识别假设显式、参数有经济含义、能做反事实。
- **准实验**:DID / IV / RDD / event-study,外生变异来源可信、识别假设可检验。
**识别假设必须显式写出,并逐条给经验支持**——这是 CEQ 与多数国内刊的最大差异。
## 设计优先级(强 → 弱)
1. 政策冲击 + DID(含交错/连续处理,须过 `ceq-modern-did`)
2. 断点回归(清晰政策门槛)
3. 工具变量(强工具 + 排他性论证)
4. 倾向得分匹配 + DID
5. 合成控制法
6. 双重机器学习 / 因果森林
7. OLS + 严密内生性讨论(仅在结构/理论实证文章中可接受)
## 分支自检
### DID(→ 详见 ceq-modern-did)
- [ ] 交错处理已用 Callaway–Sant'Anna / de Chaisemartin–D'Haultfœuille / Sun–Abraham
- [ ] 平行趋势:事件研究图必画(见 `ceq-figures`)
- [ ] 安慰剂(随机处理时点/对象)已做
### IV
- [ ] 第一阶段 F 报告,弱工具走 weak-IV-robust(见 `ceq-inference`)
- [ ] 排他性论证 ≥ 3 段(理论/制度/安慰剂)
- [ ] 报告了 reduced form;工具本身外生性有论证
### RDD
- [ ] McCrary / 密度操纵检验
- [ ] 最优带宽(CCT)+ ≥3 个带宽稳健性
- [ ] 协变量在断点处平滑
### 结构估计
- [ ] 识别假设逐条列出并辩护
- [ ] 参数有经济含义、估计稳定
- [ ] 提供反事实 / 福利分析
## 通用必查
- [ ] 识别假设**显式成文**,每条配经验支持
- [ ] 标准误聚类层级有理由(见 `ceq-inference`)
- [ ] 回应"被处理者预期/提前反应"问题
- [ ] 安慰剂/平行趋势/敏感性是**标配而非加分**
## 反模式
- "我们认为该政策外生"却无任何证据
- TWFE + 交错处理不讨论异质性偏误
- IV 用"外生事件 × 上一期内生变量",说不清为何上一期不影响当期
- RDD 用截断带宽但不汇报带宽敏感性
- 把稳健性当摆设,不汇报不利结果
## 输出格式
```
【识别策略】结构 / DID / IV / RDD / DML / 仅相关(需补)
【识别假设是否显式】是 / 否 <补>
【已完成检验】[平行趋势, 安慰剂, 弱工具, 密度, ...]
【缺失检验】[...]
【下一步】ceq-modern-did(若 DID)/ ceq-inference / ceq-mechanism
```
Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.
No comments yet. Be the first to comment!