📖 《把 AI 当助手——经济实证研究的诚信、方法与工具》双语网页版:每章中文部在前,English Edition 在后,内容与最新合订 PDF 一致。

附录 F · 专硕四类范式速查

本附录以 金融专硕（MF） 与 农村金融专硕 的培养方案为例，提供学位论文四类范式的快速参考，其他经济类专业硕士同样适用。学硕同学请参见第 11 章学术风格四分化（§11.2）。主要素材来源：四川农业大学经济学院《金融学专业硕士研究生学习指南（定稿）》（2023 年 10 月版，46 页）。

---

F.1 选题原则与硬性要求（指南第 6.1 节）

学院对金融专硕学位论文有三条硬性规定：

1. 第一条：学位论文要求运用基本理论阐述、分析、解决与金融实践相关的问题；论文论据充分，逻辑严密，力求创新，严禁抄袭。

2. 第二条（体裁四分）：学位论文选题应紧密结合金融实践问题。论文基本体裁分为：产品设计与金融实践问题解决方案、调研报告、案例分析、学术型论文。学院不鼓励学术型学位论文（学硕走学术型，专硕走前三类）。

3. 第三条（体量）：论文必须有两个主体章节，全文字数一般不少于 4 万字。

△  字数口径：学校《学位论文格式规范》写"硕士学位论文一般 3 万字左右"，学院对金融专硕要求更高（≥ 4 万字）——以学院标准为准（从严）。

---

F.2 四类范式总览（速查矩阵）

表 F.1 四类范式总览（速查矩阵）。

L5|L5|L5|L5|L5

维度 & 调研报告型 & 产品设计型 & 解决方案型 & 案例分析型
适用问题 & 现状描述/规律发现 & 新工具/新产品设计 & 已有问题的对策 & 典型事件深度剖析
典型题目 & XX 市农户数字鸿沟风险 & 基于 XX 原理的信贷产品设计 & 中小银行不良贷款化解 & XX 银行数字化转型案例
核心方法 & 问卷 + 描述统计 + 回归 & 需求分析 + 原型 + 验证 & 诊断 + 文献 + 政策建议 & 三角验证 + 编码
数据来源 & 一手问卷 / 公开统计 & 一手访谈 + 二手数据 & 公开数据 + 调研 & 访谈 + 档案 + 二手
与本手册章节 & 第 7 章调查 + 第 9 章 OLS & 第 8 章工作流 + 第 5 章测量 & 第 9 章 PSM/DID + 第 10 章案例 & 第 10 章案例研究全章\

选范式的判断顺序：① 是否已有清晰"问题" → 案例/解决方案型；② 是否能接触一手数据 → 调研报告型；③ 是否有可设计的新工具 → 产品设计型；④ 都不满足 → 建议改题。

---

F.3 四类范式选题示例（指南精选）

表 F.2 四类范式选题示例（指南精选）。

L4|L4|L4|L4

范式 & 选题示例（指南 6.2 节） & 数据特征 & 常见方法
调研报告 & 农户数字金融使用行为调研 / 中小银行竞争力评价 & 横截面问卷或面板 & OLS / Logit / Probit
产品设计 & 面向新型农业经营主体的信贷产品 / 保险产品创新 & 用户访谈 + 试点数据 & 需求分析 + 原型验证
解决方案 & 农村信用社不良贷款化解方案 / 农险理赔效率提升 & 案例数据 + 政策文本 & PSM / DID / 案例分析
案例分析 & 某农商行数字化转型 / 某产业链金融实践 & 单案例多源数据 & 扎根理论 / 编码 / 三角\

[提示] 选范式前先看目标期刊或答辩老师偏好：经济类期刊偏好实证型；管理类期刊偏好案例型；金融专硕评委对"调研报告 + 解决方案"最熟悉。

---

F.4 学硕 vs 专硕范式对照

表 F.3 学硕 vs 专硕范式对照。

L3|L3|L3

维度 & 学硕（学术型） & 专硕（MF）
主范式 & 实证论文 / 学术论文 & 调研 / 产品设计 / 解决方案 / 案例
核心方法 & 计量识别（OLS/DID/IV/PSM） & 综合方法（视范式而定）
字数 & 一般 3 万字 & ≥ 4 万字（学院要求）
创新要求 & 理论或方法创新 & 应用价值与可行性
对应本手册 & §9 全章 + §11.2 & 本附录 F + §9 + §10\

---

F.5 阅读资源（指南推荐）

期刊与网站资源（与第 6 章 MCP 工具链配套）：

• 外文期刊（12 种）：JF / JFE / JFQA / RoF / EMFT / JMCB / JBE / JIMF 等

• 重要学习网站（15 个）：央行 pbc.gov.cn / 金监总局 cbirc.gov.cn / 证监会 csrc.gov.cn / 外汇局 safe.gov.cn 等

• 数据库（5 类）：CNRDS / 国研网 / 中经数据 / EPS 数据平台 / Wind

• FAO / IMF / NBER：与第 6 章 MCP 工具链（ / / ）一一对应（详见附录 E.1）

软件选择：指南 Python 侧重量化交易工具链；本手册 Stata 侧重因果推断实证（reg → reghdfe → IV → 中介 → DID → PSM → RDD）。

---

English Edition

Appendix F · Quick Reference to the Four Professional-Master Paradigms

This appendix uses the Master of Finance (MF) and rural-finance professional master’s programmes as examples to give a quick reference to the four paradigms of the degree thesis; it applies equally to other professional master’s degrees in economics. Academic-master’s students should see the four-way split of academic styles in Chapter 11 (§11.2). Main source: Study Guide for Master’s Students in Finance (final version), College of Economics, Sichuan Agricultural University, October 2023, 46 pages.

---

F.1 Topic-selection principles and hard requirements (Guide §6.1)

The college imposes three hard rules on the finance professional master’s thesis:

1. Rule one: the thesis must apply basic theory to explain, analyse, and solve problems related to financial practice; the argument must be well evidenced and logically tight, striving for innovation, and plagiarism is strictly forbidden.

2. Rule two (four genres): the topic must be closely tied to problems of financial practice. The basic genres are: product design and solutions to financial-practice problems, survey report, case analysis, and academic-style thesis. The college does not encourage academic-style theses (academic master’s students take the academic style; professional master’s students take the first three).

3. Rule three (volume): the thesis must have two substantive chapters and in general no fewer than 40,000 characters.

△  Word-count standard: the university’s Format Specification for Degree Theses says “a master’s thesis is generally around 30,000 characters”, but the college requires more for the finance professional master’s ($\ge$ 40,000)—follow the college standard (the stricter one).

---

F.2 Overview of the four paradigms (quick-reference matrix)

Table F.1 Overview of the four paradigms (quick-reference matrix).

L5|L5|L5|L5|L5

Dimension & Survey report & Product design & Solution design & Case analysis
Suitable question & Describing the status quo / discovering regularities & Designing a new tool or product & Remedies for an existing problem & In-depth analysis of a typical event
Typical title & Digital-divide risk among farmers in city XX & A credit product based on principle XX & Resolving NPLs at a small or medium bank & Digital transformation at bank XX
Core method & Questionnaire + descriptive statistics + regression & Needs analysis + prototype + validation & Diagnosis + literature + policy advice & Triangulation + coding
Data sources & Primary questionnaire / public statistics & Primary interviews + secondary data & Public data + fieldwork & Interviews + archives + secondary data
Related chapters & Ch. 7 survey + Ch. 9 OLS & Ch. 8 workflow + Ch. 5 measurement & Ch. 9 PSM/DID + Ch. 10 cases & The whole of Ch. 10\

Order for choosing a paradigm: (1) is there already a clear “problem”? → case or solution design; (2) can you access primary data? → survey report; (3) is there a new tool you could design? → product design; (4) none of the above → change the topic.

---

F.3 Topic examples for the four paradigms (selected from the Guide)

Table F.2 Topic examples for the four paradigms (selected from the Guide).

L4|L4|L4|L4

Paradigm & Topic example (Guide §6.2) & Data characteristics & Common methods
Survey report & Farmers’ use of digital finance / competitiveness of small and medium banks & Cross-sectional questionnaire or panel & OLS / Logit / Probit
Product design & Credit or insurance products for new agricultural business entities & User interviews + pilot data & Needs analysis + prototype validation
Solution design & Resolving rural credit-union NPLs / raising agricultural-insurance claims efficiency & Case data + policy texts & PSM / DID / case analysis
Case analysis & Digital transformation at a rural commercial bank / a supply-chain finance practice & Single case, multiple sources & Grounded theory / coding / triangulation\

[Tip] Before choosing a paradigm, check the target journal or the defence panel’s preferences: economics journals favour empirical work, management journals favour cases, and finance professional-master’s panels are most familiar with “survey report + solution design”.

---

F.4 Academic master’s versus professional master’s paradigms

Table F.3 Academic master’s versus professional master’s paradigms.

L3|L3|L3

Dimension & Academic master’s & Professional master’s (MF)
Main paradigm & Empirical / academic paper & Survey / product design / solution / case
Core method & Econometric identification (OLS/DID/IV/PSM) & Mixed methods (depending on the paradigm)
Length & Generally 30,000 characters & $\ge$ 40,000 characters (college requirement)
Innovation requirement & Theoretical or methodological innovation & Application value and feasibility
Where in this handbook & All of §9 + §11.2 & This Appendix F + §9 + §10\

---

F.5 Reading resources (recommended by the Guide)

Journals and websites (paired with the Chapter 6 MCP toolchain):

• Foreign journals (12): JF / JFE / JFQA / RoF / EMFT / JMCB / JBE / JIMF and others

• Key learning websites (15): the central bank at pbc.gov.cn / the financial regulator at cbirc.gov.cn / the securities regulator at csrc.gov.cn / the foreign-exchange regulator at safe.gov.cn, among others

• Databases (5 categories): CNRDS / the Development Research Center network / CEInet / EPS / Wind

• FAO / IMF / NBER: matched one-to-one with the Chapter 6 MCP toolchain ( / / ); see Appendix E.1

Software choice: the Guide emphasises a quantitative-trading toolchain in Python; this handbook emphasises causal-inference empirics in Stata (reg → reghdfe → IV → mediation → DID → PSM → RDD).