grad-info-economics
用資訊經濟學診斷逆選擇與道德風險造成的市場失靈，設計訊號和篩選機制
中文翻譯
--- name: "grad-info-economics" description: "用資訊經濟學診斷並修補由資訊不對稱造成的市場失靈。當使用者需要分析逆選擇、道德風險，或訊號傳遞與篩選機制，尤其是在保險、勞動、信用或產品品質市場時，就用這個 skill。" metadata: category: "WP-27 賽局與制度經濟" tags: ["information-economics", "adverse-selection", "moral-hazard", "signaling", "screening", "market-for-lemons", "Akerlof", "Spence", "Stiglitz"] --- # Information Economics：逆選擇、道德風險與訊號傳遞 ## Overview 資訊經濟學研究雙方之間的資訊不對稱，如何造成市場失靈，並塑造制度上的回應。Akerlof 的「檸檬市場」顯示逆選擇可以讓整個市場崩潰；Spence 的訊號傳遞模型顯示有資訊的一方可以透過有成本的行動，可信地傳達品質；Rothschild-Stiglitz 的篩選顯示沒資訊的一方可以設計菜單，誘導自我選擇。這些框架合起來，解釋了為什麼保險、信用、勞動和二手商品市場會系統性地偏離競爭的理想狀態，也解釋了保固、證書和監管這類制度為什麼存在。 ## When to Use - 診斷為什麼某個市場在失靈或萎縮（品質崩壞、信用配給、保險死亡螺旋） - 評估某個訊號（學位、認證、保固）是否可信而且有效率 - 設計篩選機制（菜單、自負額、試用期）來區分異質的行為者 - 評估政策介入（強制揭露、最低標準、補貼保險） ## When NOT to Use - 資訊是對稱的，或能以零成本驗證（沒有不對稱的問題） - 產品是純粹的經驗財，聲譽和重複購買就能完全解決品質不確定性 - 分析需要建模雙邊以外的策略互動（用完整的賽局理論模型） ## Assumptions ``` IRON LAW: Information asymmetry causes market failure — without corrective mechanisms (signals, screens, warranties), bad drives out good. Markets with severe adverse selection can unravel completely (Akerlof's lemons result). ``` - 一方（有資訊的一方）擁有品質、風險或努力程度的私有資訊 - 另一方（沒資訊的一方）無法直接觀察相關的特徵 - 行為者是理性的，會針對資訊結構做出策略性的回應 - 訊號是有成本的，而且成本因類型而異（單交叉 / Spence-Mirrlees 條件） - 在競爭市場中，不同類型之間的交叉補貼無法持續 ## Methodology **Step 1 — 找出資訊不對稱** 分類：(a) 逆選擇（隱藏類型，簽約前），有資訊一方的類型會影響沒資訊一方的報酬；(b) 道德風險（隱藏行動，簽約後），有資訊一方的努力無法觀察；(c) 兩者同時存在。找出誰有資訊、誰沒有。 **Step 2 — 建模市場失靈** 逆選擇：說明混同（提供單一契約）如何吸引不成比例的壞類型，推高成本、拉高價格，讓好類型退出，也就是解體動態。道德風險：說明保險或契約如何降低代理人付出努力或採取預防措施的誘因，提高期望成本。 **Step 3 — 評估矯正機制** 訊號傳遞（有資訊的一方行動）：找出訊號，驗證單交叉條件（高類型覺得訊號的成本較低），並檢查是否存在分離均衡。篩選（沒資訊的一方設計菜單）：設計能誘導自我選擇的契約，通常高類型拿到有效率的契約，低類型則面對數量扭曲。其他機制：保固、聲譽、認證、強制揭露、監管。 **Step 4 — 評估效率與政策** 把結果和完全資訊的基準比較。計算福利損失的來源：(a) 錯失的交易（好類型被定價擠出市場）；(b) 訊號的浪費（花在沒有直接價值的證書上的資源）；(c) 篩選的扭曲（對低類型的無效率契約）。建議市場機制是否足夠，還是需要政府介入。 ## Output Format ```markdown ## Information Economics Analysis: [Market / Context] ### Information Structure - **Asymmetry type**: Adverse selection / Moral hazard / Both - **Informed party**: [who knows what] - **Uninformed party**: [who lacks what information] - **Hidden variable**: [quality / risk type / effort level] ### Market Failure Diagnosis - **Unraveling risk**: [high / medium / low] - **Pooling outcome**: [what happens if all types are treated identically] - **Separating outcome**: [what happens if types are distinguished] ### Corrective Mechanisms | Mechanism | Who Initiates | How It Works | Effective? | |---------------|---------------|-------------------------|------------| | Signaling | Informed | [e.g., education] | | | Screening | Uninformed | [e.g., deductible menu] | | | Warranty | Informed | [e.g., money-back] | | | Regulation | Government | [e.g., mandatory disclosure] | | ### Efficiency Assessment - **Full-information benchmark**: [first-best outcome] - **Welfare loss sources**: [missing trades / signaling waste / screening distortion] - **Net welfare**: [second-best outcome vs. unregulated market] ### Recommendation [Which mechanisms to deploy; whether policy intervention is warranted] ``` ## Gotchas - 訊號傳遞在社會上可能是浪費的，如果教育只是訊號（不是人力資本），花在上面的資源就是純粹的無謂損失 - 當高類型比例很大時，Rothschild-Stiglitz 模型在純策略下可能沒有 Nash 均衡，Wilson 預期均衡或 Riley 反應均衡是替代方案 - 道德風險和逆選擇會互相影響：附自負額的保險（針對逆選擇的篩選）也能減輕道德風險，但這兩個問題可能需要互相衝突的契約設計 - 強制揭露可能適得其反，如果它導致原本穩定的混同均衡解體 - 在重複互動中，聲譽可以取代正式的訊號，但聲譽很脆弱，也會受到終局效應影響 - 數位平台和大數據在某些市場減少了資訊不對稱（信用評分、評論），卻在其他市場創造了新的不對稱（演算法定價、資料隱私） ## References - Akerlof, G. (1970). "The Market for Lemons: Quality Uncertainty and the Market Mechanism." *Quarterly Journal of Economics*. - Spence, M. (1973). "Job Market Signaling." *Quarterly Journal of Economics*. - Rothschild, M. & Stiglitz, J. (1976). "Equilibrium in Competitive Insurance Markets." *Quarterly Journal of Economics*. - Riley, J. (2001). "Silver Signals: Twenty-Five Years of Screening and Signaling." *Journal of Economic Literature*.
英文原文
--- name: "grad-info-economics" description: "Apply information economics to diagnose and remedy market failures caused by asymmetric information. Use this skill when the user needs to analyze adverse selection, moral hazard, or signaling and screening mechanisms, especially in insurance, labor, credit, or product quality markets." metadata: category: "WP-27 賽局與制度經濟" tags: ["information-economics", "adverse-selection", "moral-hazard", "signaling", "screening", "market-for-lemons", "Akerlof", "Spence", "Stiglitz"] --- # Information Economics: Adverse Selection, Moral Hazard, and Signaling ## Overview Information economics studies how asymmetric information between parties causes market failures and shapes institutional responses. Akerlof's "market for lemons" shows adverse selection can collapse entire markets; Spence's signaling model shows informed parties can credibly convey quality through costly actions; Rothschild-Stiglitz screening shows uninformed parties can design menus that induce self-selection. Together, these frameworks explain why markets for insurance, credit, labor, and used goods systematically deviate from the competitive ideal and why institutions like warranties, credentials, and regulation exist. ## When to Use - Diagnosing why a market is failing or shrinking (quality collapse, credit rationing, insurance death spirals) - Evaluating whether a signal (degree, certification, warranty) is credible and efficient - Designing screening mechanisms (menus, deductibles, trial periods) to sort heterogeneous agents - Assessing policy interventions (mandatory disclosure, minimum standards, subsidized insurance) ## When NOT to Use - Information is symmetric or costlessly verifiable (no asymmetry problem) - The product is a pure experience good where reputation and repeat purchase fully resolve quality uncertainty - The analysis requires modeling strategic interaction beyond bilateral (use full game-theoretic models) ## Assumptions ``` IRON LAW: Information asymmetry causes market failure — without corrective mechanisms (signals, screens, warranties), bad drives out good. Markets with severe adverse selection can unravel completely (Akerlof's lemons result). ``` - One party (informed) has private information about quality, risk, or effort - The other party (uninformed) cannot directly observe the relevant characteristic - Agents are rational and respond to the information structure strategically - Signals are costly, and the cost differs by type (single-crossing / Spence-Mirrlees condition) - In competitive markets, cross-subsidization between types is unsustainable ## Methodology **Step 1 — Identify the Information Asymmetry** Classify: (a) Adverse selection (hidden type, pre-contractual) — the informed party's type affects the uninformed party's payoff; (b) Moral hazard (hidden action, post-contractual) — the informed party's effort is unobservable; (c) Both present simultaneously. Identify who is informed and who is uninformed. **Step 2 — Model the Market Failure** For adverse selection: show how pooling (offering a single contract) attracts disproportionately bad types, driving up costs, raising prices, and causing good types to exit — the unraveling dynamic. For moral hazard: show how insurance or contracting reduces the agent's incentive to exert effort or take precautions, increasing expected costs. **Step 3 — Evaluate Corrective Mechanisms** Signaling (informed party acts): identify the signal, verify the single-crossing condition (high types find the signal less costly), and check whether a separating equilibrium exists. Screening (uninformed party designs menu): design contracts that induce self-selection — typically, high types get efficient contracts while low types face quantity distortion. Other mechanisms: warranties, reputation, certification, mandatory disclosure, regulation. **Step 4 — Assess Efficiency and Policy** Compare the outcome against the full-information benchmark. Calculate welfare loss from: (a) missing trades (good types priced out); (b) signaling waste (resources spent on credentials that produce no direct value); (c) screening distortions (inefficient contracts for low types). Recommend whether market mechanisms suffice or government intervention is needed. ## Output Format ```markdown ## Information Economics Analysis: [Market / Context] ### Information Structure - **Asymmetry type**: Adverse selection / Moral hazard / Both - **Informed party**: [who knows what] - **Uninformed party**: [who lacks what information] - **Hidden variable**: [quality / risk type / effort level] ### Market Failure Diagnosis - **Unraveling risk**: [high / medium / low] - **Pooling outcome**: [what happens if all types are treated identically] - **Separating outcome**: [what happens if types are distinguished] ### Corrective Mechanisms | Mechanism | Who Initiates | How It Works | Effective? | |---------------|---------------|-------------------------|------------| | Signaling | Informed | [e.g., education] | | | Screening | Uninformed | [e.g., deductible menu] | | | Warranty | Informed | [e.g., money-back] | | | Regulation | Government | [e.g., mandatory disclosure] | | ### Efficiency Assessment - **Full-information benchmark**: [first-best outcome] - **Welfare loss sources**: [missing trades / signaling waste / screening distortion] - **Net welfare**: [second-best outcome vs. unregulated market] ### Recommendation [Which mechanisms to deploy; whether policy intervention is warranted] ``` ## Gotchas - Signaling can be socially wasteful — if education serves only as a signal (not human capital), the resources spent on it are pure deadweight loss - The Rothschild-Stiglitz model may have no Nash equilibrium in pure strategies when the proportion of high types is large — the Wilson anticipatory equilibrium or Riley reactive equilibrium are alternatives - Moral hazard and adverse selection interact: insurance with deductibles (screening for adverse selection) also mitigates moral hazard, but the two problems may require conflicting contract designs - Mandatory disclosure can backfire if it causes unraveling of previously stable pooling equilibria - In repeated interactions, reputation can substitute for formal signals — but reputation is fragile and subject to end-game effects - Digital platforms and big data are reducing information asymmetry in some markets (credit scoring, reviews) but creating new asymmetries in others (algorithmic pricing, data privacy) ## References - Akerlof, G. (1970). "The Market for Lemons: Quality Uncertainty and the Market Mechanism." *Quarterly Journal of Economics*. - Spence, M. (1973). "Job Market Signaling." *Quarterly Journal of Economics*. - Rothschild, M. & Stiglitz, J. (1976). "Equilibrium in Competitive Insurance Markets." *Quarterly Journal of Economics*. - Riley, J. (2001). "Silver Signals: Twenty-Five Years of Screening and Signaling." *Journal of Economic Literature*.