This page serves as a quick check for EC487 Advanced Microeconomics.
明确会考的内容
Lecture 1: 不完全信息静态博弈
- Bayesian Nash Equilibrium 的定义和求解(策略是类型到行动的函数,期望收益最大化)
- Harsanyi 的 Nature 表示法(将不完全信息博弈转化为不完全信息的完全博弈)
- Contribution game 的求解——教授明确说”fair to expect you can solve this one”
Lecture 2: 信号博弈与 Weak PBE
- Weak PBE 的定义:策略 + 信念,optimality + consistency,是全课用得最多的均衡概念
- Beer-Quiche game:所有均衡(无分离均衡、两个混同均衡),包括混合策略均衡——教授说”I would know the beer and quiche game”
- Spence signalling game:分离均衡、混同均衡、均衡的连续统——教授说”these two are two examples you want to know”
- Intuitive Criterion:知道概念(对 off-path belief 的合理性约束),知道如何在 Beer-Quiche 和 Spence 这两个具体例子中应用。教授明确说不会要求你对全新博弈应用 IC
- Off-path beliefs 的角色:为什么 off-path beliefs 的任意性导致均衡多重性
Lecture 3: Cheap Talk
- Crawford-Sobel cheap talk game:教授说”everything that we did in terms of the cheap talk game seems fair play”
- Babbling equilibrium 的存在性证明
- 完全揭示均衡仅当 时存在
- Interval/partitional equilibrium 的构造(猜测→求信念→求最优响应→验证)
- 递推公式、 最大消息数
- 需要知道证明——教授明确说”I’d say yes in this case”
- Problem set 中关于 ex ante alignment vs interim misalignment 的分析
Lecture 5: 道德风险 (Moral Hazard)
- First best(可观察 effort)vs Second best(不可观察 effort)的对比
- First best: 完全保险、participation constraint binding
- Risk neutral agent 的特殊情况:可以”sell the firm”实现 first best
- Risk averse agent + 二元 effort:写 Lagrangian、解 FOC、解释乘子不为零的含义
- MLRP 条件与合同形状的解释(单调性等)
- Jensen’s inequality 论证 second best 成本更高
- LEN model(连续 effort、正态分布、线性合同):first order approach
- Problem set 6 很重要:limited liability 约束、信息精度的比较静态
- 有界分布的特殊情况(effort 完全可推断,first best 可达)
Lecture 6: 机制设计 (单买家 Screening)
- Direct mechanism 的定义——“definitely important, it’s going to come up”
- Revelation Principle:陈述和证明都要知道——“you definitely need to know it”
- 两类型情况的 Lagrangian 求解
- Myerson Lemma(连续类型):IC + IR 等价于 非递减 + 非递减 + 的积分表达式——“you need to know this lemma”(证明不强制要求)
- Virtual valuation、integration by parts 的变换过程(熟悉即可,不需要自己做积分分步)
- 最优机制求解:no distortion at the top, distortion at the bottom, information rent
- Myerson regularity 条件
Lecture 7: 拍卖与多买家机制设计
- First-price auction、second-price auction 的均衡求解——“definitely I would know how to do these examples”
- 从 indirect mechanism 到 direct mechanism 的转换——“definitely something you want to know”
- Interim IC(与 ex post IC 的区别)
- Revenue Equivalence Theorem
- 最优拍卖中的 virtual valuation 与排除规则
- 最后一个 Problem set 非常重要——教授说”this lecture and its problem set, very much, yes”
明确不考的内容
- Higher-order beliefs / Email game:“this is not in the exam”(仅作为类型丰富性的例子)
- Global games 作为一个概念/refinement:“this is not a word that comes up in my test”(但求解不完全信息博弈+取极限的技术可能考)
- Reputation games / Behavioural players:“I want to promise that this is not in the exam”,“the words behavioural player are not in the exam”
- Lecture 4 整体 (Akerlof / Rothschild-Stiglitz / Rational Expectations Equilibrium):教授说”all of lecture 4 is off”
- D1 criterion:教授说”if you paid me $10,000 right now, I couldn’t tell you what D1 was”
- First order approach 何时不成立的一般性讨论(超出课程范围)
- Full implementation vs partial implementation 的区别
考试格式提示
- 去年的考试有 4 道题,混合了不同 topic 的工具(如道德风险 + reputation 的组合)
- 考的是新组合的旧工具,不是完全照搬课堂例子,但会有详细的 sub-question 引导
- 去年有一题就是 Beer-Quiche game,一题是 cheap talk game
- 教授建议参考去年考卷来了解结构