# 现代研究阅读：产能约束与通胀 从互补松弛走到同冲击反事实与非线性交互，分清论文重建和教学模拟。 Entry: zh-ei17 | Node: EI-17 | Language: zh | Editorial revision: 2026-09-21 ## Teaching instructions 你正在教授EI-17《现代研究阅读：产能约束与通胀》，采用内容版本2026-09-21-review-v3，当前范围：本篇完整学习单元。 先使用工具实际读取当前required_readings列出的公开原文完整单元，核对标题、版本、表头、单位、脚注和适用范围。来源清单或作者先前读过不等于你本次读过；runtime_reading_log初始为空，按实际结果填写，不预填成功。网页太大时用按页/文本范围的工具；不要用搜索节选顶替完整单元。读取失败时只说明具体缺口并尝试同机构/作者对应版本的开放正文，不能声称已完成依赖该原文的讲解。 完成读取后，按学习者已有水平选择直接讲解、推导或互动，不强迫重学已掌握的四则运算。诊断任务：先解a=100,K=50的乘子边界，再区分“保持同冲击、放松约束”与“联合冲击减容量单独路径”。 沿随包完整正文推进一个完整推理任务：先结构和变量，再真实材料/模型设定，再推导和比较，最后迁移与反馈。真实观察、作者估计、本站教学参数必须保持各自身份。通过标准：互补松弛、恰触边μ=0、同冲击对照、四状态交互公式全部正确；并能区分Figure 7c的2021–2022总体通胀高约1–2pp与Figure 7a的2020Q2–2021Q2商品通胀加速约一半，不能把两者当同一分母，也不能将教学价格称后验通胀或相加为100%。 使用随包实验的相同输入、算法和静态结果。你可以改变教学参数，但先声明改变哪一条件；不改写冻结观察值，不编造未取得的论文曲线。解释题和迁移题的完整解析已随正文提供，先让读者作答，再对照推理而非只报分数。 额外约定：EI-10年化不是付息流水；EI-11没有未定义衍生品；EI-12信用额度不得计入30天来源；EI-13基期产出权重；EI-15分支来源仅随所选分支读取；EI-16需求与发电覆盖不同；EI-17 Figure7/9不是同一实验；EI-18来源尺度疑点不替作者静默修正。只适用当前篇相关项。 Before substantive teaching, actually retrieve every required reading unit for the selected scope. Read its complete designated section, including necessary assumptions, tables and footnotes. A working URL or an editorial access date is not a runtime reading receipt. Record the actual version, location, scope and what it supports. If unavailable, use a previously verified equivalent source; if the required unit remains unavailable, identify that gap rather than teach it from memory. Start runtime_reading_log empty. Once reading is complete, use a substantive diagnostic or follow the reader's request for direct explanation. Advance one complete reasoning task at a time; skip mastered basics. Distinguish original facts, supplied teaching assumptions and inference. ## Required readings and runtime protocol ```json { "entry_id": "zh-ei17", "node_id": "EI-17", "content_version": "2026-09-21-review-v3", "export_mode": "public", "selected_branch": "all", "audience": "有微积分、线性代数和基本概率基础的高年级本科生至研究生", "learning_task": "重建容量约束的作用与Figure 7/9的不同运算，不伪造作者后验时间序列。", "required_readings": [ { "source_id": "EIDEF-S20", "title": "Supply Chain Constraints and Inflation", "authors": [ "Diego Comin", "Robert C. Johnson", "Callum Jones" ], "version": "IMES DP2025-E-15, October2025", "access": { "kind": "pdf_full_text", "uri": "https://www.imes.boj.or.jp/research/papers/english/25-E-15.pdf", "verified_access_at": "2026-09-21" }, "required_unit": { "locator": "§2.2.2 printed12–13；§2.6.1 printed17–18；§3 printed21–26；§4.1–4.3 printed27–33；Fig7 p29/PDF32；Fig9 p32/PDF35", "scope": "所列具名单元全文，包含公式、表头、表注及本篇采用结论所需上下文；不要求无关章节。", "purpose": "Figure 7保持同一组smoothed shocks、只把约束改为始终松弛；Figure 7c的2021–2022总体通胀高约1–2pp，与Figure 7a的2020Q2–2021Q2商品通胀加速约一半是不同面板/期间/分母。Figure 9联合冲击减容量单独路径另行识别交互。" }, "supports": "Figure 7保持同一组smoothed shocks、只把约束改为始终松弛；Figure 7c的2021–2022总体通胀高约1–2pp，与Figure 7a的2020Q2–2021Q2商品通胀加速约一半是不同面板/期间/分母。Figure 9联合冲击减容量单独路径另行识别交互。", "id": "EI-17-READ-1", "retrieved_at": "2026-09-21" } ], "optional_readings": [], "runtime_reading_log": [], "supplied_inputs": { "frozen_source_version": "2026-09-21-ei-defgh-preflight-final", "node": { "experiment_ids": [ "exp-ei17-static-capacity-complementarity", "exp-ei17-fig7-relax-constraints", "exp-ei17-fig9-interaction" ], "research_version": "IMES Discussion Paper 2025-E-15, October 2025", "static_capacity_toy": { "demand_inverse": "P = a - Q", "marginal_cost_inverse_supply": "P = Q + mu", "constraints": [ "Q <= K", "mu >= 0", "mu*(Q-K)=0" ], "slack_case": { "K": 80, "a_before": 100, "a_after": 120, "before": { "P": 50, "Q": 50, "mu": 0 }, "after": { "P": 60, "Q": 60, "mu": 0 } }, "binding_case": { "K": 50, "a_before": 120, "a_after": 130, "before": { "P": 70, "Q": 50, "mu": 20 }, "after": { "P": 80, "Q": 50, "mu": 30 } }, "lesson": "P↑,Q↑ for a positive demand shift requires a non-binding interior in this toy; with binding capacity, P can rise while Q remains fixed." }, "Fig7_contract": { "filter_shocks_under_model_with_potentially_binding_constraints": true, "counterfactual": "feed the same smoothed shocks through the model with constraints slack in all periods", "inflation_identity": "quarterly inflation expressed at annualized rate and de-meaned; counterfactual simulations retain measurement-error treatment", "uncertainty": "posterior simulations / 5th-95th percentile bands", "capacity_prior": "positive mass on binding durations only starting 2020Q2" }, "Fig9b_contract": { "operation": "simulate named shock type together with filtered capacity shocks, then subtract the capacity-shocks-alone path", "nonlinear_interaction": true, "not_additive_100pct_decomposition": true }, "Figure7_scale_reading": { "aggregate_consumer_price_inflation": { "period": "2021-2022", "actual_minus_all_slack_counterfactual_pp_approx": "1-2", "panel": "Figure 7c / aggregate consumer price inflation" }, "goods_price_inflation_acceleration": { "period": "2020Q2-2021Q2", "constraint_share_approx": "about one half", "panel": "Figure 7a / goods price inflation" }, "identity": "different panel, period, and denominator; do not infer a permanent aggregate 50% share by dividing an aggregate pp gap by another number" } }, "experiments": [ { "id": "exp-ei17-static-capacity-complementarity", "node_id": "EI-17", "title": "产能互补松弛", "anchor": "ei17-kkt", "method": "capacity", "controls": [ { "key": "a", "label": "逆需求截距a", "default": 120, "min": 0, "max": 200, "step": 5, "kind": "number" }, { "key": "K", "label": "产能K", "default": 50, "min": 1, "max": 150, "step": 5, "kind": "number" } ], "inputs": { "a": 120, "K": 50 }, "algorithm": "最大化aQ−Q²,0≤Q≤K。a≥0,K>0；Q=min(a/2,K),P=a−Q,μ=a−2Q。显示slack与μ(Q−K)=0。a=0采用Q=P=μ=0。模型为静态教学经济，不是IMES复现。", "static_equivalent": "K80/a100→120：P,Q均50→60；K50/a120→130：Q50，P70→80，μ20→30；a100,K50：触边μ0。", "source_ids": [ "EIDEF-S20" ], "identity": "teaching_assumption", "description": "产能互补松弛", "outputs": { "a": 120, "K": 50, "q": 50, "p": 70, "mu": 20, "slack": 0, "complementarity": 0, "regime": "绑定且乘子为正", "identity": "one-period competitive teaching model; abstract price/output, not the IMES model" } }, { "id": "exp-ei17-fig7-relax-constraints", "node_id": "EI-17", "title": "同一输入路径，放松约束", "anchor": "ei17-fig7", "method": "relax", "controls": [ { "key": "capacityScale", "label": "教学容量路径倍数", "default": 1, "min": 0.5, "max": 2, "step": 0.1, "kind": "number" } ], "inputs": { "capacityScale": 1 }, "algorithm": "新增教学路径a=[100,120,130,110],K=[80,50,50,80]×scale；受限逐期用静态模型，松弛逐期P=Q=a/2。不是论文通胀，不生成后验置信带。作者实验流程另列同θ、同smoothed shocks、all-slack反事实及measurement error。", "static_equivalent": "scale1时受限价格[50,70,80,55]，松弛[50,60,65,55]，价格差[0,10,15,0]。均为抽象价格单位。", "source_ids": [ "EIDEF-S20" ], "identity": "teaching_assumption", "description": "同一输入路径，放松约束", "outputs": { "rows": [ { "period": 1, "a": 100, "K": 80, "constrained_q": 50, "constrained_p": 50, "slack_q": 50, "slack_p": 50, "difference": 0, "mu": 0 }, { "period": 2, "a": 120, "K": 50, "constrained_q": 50, "constrained_p": 70, "slack_q": 60, "slack_p": 60, "difference": 10, "mu": 20 }, { "period": 3, "a": 130, "K": 50, "constrained_q": 50, "constrained_p": 80, "slack_q": 65, "slack_p": 65, "difference": 15, "mu": 30 }, { "period": 4, "a": 110, "K": 80, "constrained_q": 55, "constrained_p": 55, "slack_q": 55, "slack_p": 55, "difference": 0, "mu": 0 } ], "identity": "same demand sequence, constraint switched off; additional teaching data, not Fig.7 posterior simulation" } }, { "id": "exp-ei17-fig9-interaction", "node_id": "EI-17", "title": "条件效应与非线性交互", "anchor": "ei17-fig9", "method": "interaction", "controls": [ { "key": "demandShock", "label": "教学需求截距增量u", "default": 30, "min": 0, "max": 80, "step": 5, "kind": "number" }, { "key": "capacityReduction", "label": "教学容量减少v", "default": 30, "min": 0, "max": 60, "step": 5, "kind": "number" } ], "inputs": { "demandShock": 30, "capacityReduction": 30 }, "algorithm": "基准a100,K80；分别计算P(0,0),P(u,0),P(0,v),P(u,v)。conditional=P(u,v)−P(0,v)；interaction=conditional−[P(u,0)−P(0,0)]。非线性不强制贡献加总100%。", "static_equivalent": "默认价格[50,65,50,80]；单独需求15、容量单独0、容量背景下需求30、交互15。教学价格，不是作者Figure9数值。", "source_ids": [ "EIDEF-S20" ], "identity": "teaching_assumption", "description": "条件效应与非线性交互", "outputs": { "baseline": { "a": 100, "K": 80, "q": 50, "p": 50, "mu": 0, "slack": 30, "complementarity": 0, "regime": "松弛", "identity": "one-period competitive teaching model; abstract price/output, not the IMES model" }, "demand_only": { "a": 130, "K": 80, "q": 65, "p": 65, "mu": 0, "slack": 15, "complementarity": 0, "regime": "松弛", "identity": "one-period competitive teaching model; abstract price/output, not the IMES model" }, "capacity_only": { "a": 100, "K": 50, "q": 50, "p": 50, "mu": 0, "slack": 0, "complementarity": 0, "regime": "恰好触边，乘子为零", "identity": "one-period competitive teaching model; abstract price/output, not the IMES model" }, "joint": { "a": 130, "K": 50, "q": 50, "p": 80, "mu": 30, "slack": 0, "complementarity": 0, "regime": "绑定且乘子为正", "identity": "one-period competitive teaching model; abstract price/output, not the IMES model" }, "demand_effect_without_capacity": 15, "capacity_alone_effect": 0, "conditional_demand_effect": 30, "interaction": 15, "total_effect": 30, "identity": "price-unit four-state teaching example; difference operation mirrors Fig.9b, not estimated inflation" } } ] }, "static_equivalents": [ { "experiment_id": "exp-ei17-static-capacity-complementarity", "text": "K80/a100→120：P,Q均50→60；K50/a120→130：Q50，P70→80，μ20→30；a100,K50：触边μ0。", "outputs": { "a": 120, "K": 50, "q": 50, "p": 70, "mu": 20, "slack": 0, "complementarity": 0, "regime": "绑定且乘子为正", "identity": "one-period competitive teaching model; abstract price/output, not the IMES model" } }, { "experiment_id": "exp-ei17-fig7-relax-constraints", "text": "scale1时受限价格[50,70,80,55]，松弛[50,60,65,55]，价格差[0,10,15,0]。均为抽象价格单位。", "outputs": { "rows": [ { "period": 1, "a": 100, "K": 80, "constrained_q": 50, "constrained_p": 50, "slack_q": 50, "slack_p": 50, "difference": 0, "mu": 0 }, { "period": 2, "a": 120, "K": 50, "constrained_q": 50, "constrained_p": 70, "slack_q": 60, "slack_p": 60, "difference": 10, "mu": 20 }, { "period": 3, "a": 130, "K": 50, "constrained_q": 50, "constrained_p": 80, "slack_q": 65, "slack_p": 65, "difference": 15, "mu": 30 }, { "period": 4, "a": 110, "K": 80, "constrained_q": 55, "constrained_p": 55, "slack_q": 55, "slack_p": 55, "difference": 0, "mu": 0 } ], "identity": "same demand sequence, constraint switched off; additional teaching data, not Fig.7 posterior simulation" } }, { "experiment_id": "exp-ei17-fig9-interaction", "text": "默认价格[50,65,50,80]；单独需求15、容量单独0、容量背景下需求30、交互15。教学价格，不是作者Figure9数值。", "outputs": { "baseline": { "a": 100, "K": 80, "q": 50, "p": 50, "mu": 0, "slack": 30, "complementarity": 0, "regime": "松弛", "identity": "one-period competitive teaching model; abstract price/output, not the IMES model" }, "demand_only": { "a": 130, "K": 80, "q": 65, "p": 65, "mu": 0, "slack": 15, "complementarity": 0, "regime": "松弛", "identity": "one-period competitive teaching model; abstract price/output, not the IMES model" }, "capacity_only": { "a": 100, "K": 50, "q": 50, "p": 50, "mu": 0, "slack": 0, "complementarity": 0, "regime": "恰好触边，乘子为零", "identity": "one-period competitive teaching model; abstract price/output, not the IMES model" }, "joint": { "a": 130, "K": 50, "q": 50, "p": 80, "mu": 30, "slack": 0, "complementarity": 0, "regime": "绑定且乘子为正", "identity": "one-period competitive teaching model; abstract price/output, not the IMES model" }, "demand_effect_without_capacity": 15, "capacity_alone_effect": 0, "conditional_demand_effect": 30, "interaction": 15, "total_effect": 30, "identity": "price-unit four-state teaching example; difference operation mirrors Fig.9b, not estimated inflation" } } ], "source_id_aliases": { "EIDEF-S10": "BFDE-JEP", "EIDEF-S14": "EIBC-S07M", "EIDEF-S16A": "EIBC-S05A", "EIDEF-S16B": "EIBC-S05B", "EIDEF-S21": "EI-S06-M3-202605" } } ``` ## Supplied entry 本篇任务：重建一篇研究中“约束怎样进入价格、同一组冲击怎样做反事实、不同冲击为什么不能线性分账”的完整关系。主要读物是Comin、Johnson和Jones的 *Supply Chain Constraints and Inflation*，固定为IMES 2025-E-15、2025年10月版。正文与三个小实验约20分钟；它们不代替整篇宏观模型及附录的复现。[^paper] ## 一、论文真正要区分的，是冲击与约束如何共同作用 价格上涨可以来自需求增加，也可以来自生产或进口受限。只画一条价格曲线，很难分清这些解释。论文把国内外生产、投入联系、价格调整和可能生效的产能上限放进同一个动态模型，再用价格与数量数据共同识别。研究样本是美国1990Q1–2023Q4的季度资料，不是2026年的通胀预测。它衡量产能约束历史影响的核心反事实，是先在允许约束生效的模型中滤出冲击，再保持同一组冲击，只把约束改成始终松弛来比较路径。[^data][^counter] 本篇沿三层推进：第一层是优化条件，说明产能影子价格为何改变定价；第二层是Figure 7的**保持同一组冲击、移除约束**实验；第三层是Figure 9的**某类冲击与容量冲击共同作用**实验。它们分别回答机制、反事实和交互，不应该混成一个“供应链贡献占比”。 ## 二、约束不是一条标签：它有可行性和乘子条件 在论文§2.2.2中，企业价格决策要面对产量上限。简写某期某企业的产量为 $Y$、上限为 $\bar Y$、乘子为 $\mu$，必须同时满足 $$ Y\le\bar Y,\qquad \mu\ge0,\qquad \mu(Y-\bar Y)=0. $$ 最后一条是互补松弛：有闲置容量时，乘子必须为0；乘子严格为正时，产量必在上限。反过来，仅有 $Y=\bar Y$ 并不强迫 $\mu>0$，因为需求恰好把产量推到边界时，增加一点容量也可能没有一阶价值。[^model] 论文的价格一阶条件中，边际成本与产能乘子一起影响最优价格；§2.6.1线性化的国内和进口价格方程额外出现由乘子带来的项。其机制不是“所有企业突然改变了市场力量”，而是在同一生产与定价结构下，产能上限改变继续扩大供给的影子成本。论文同时有投入替代、国内外来源及粘性价格，不能把它缩成下面的静态例子。[^model] 为了手算这三个条件，我们另设一个**静态教学经济**：逆需求 $P=a-Q$，边际生产成本为 $Q$，总产出不得超过 $K>0$。其竞争配置可由最大化总剩余重建： $$ \begin{aligned} &\max_{0\le Q\le K}\left[ \int_0^Q(a-z)\,dz-\frac{Q^2}{2} \right]\\ &\qquad=\max_{0\le Q\le K}(aQ-Q^2). \end{aligned} $$ 在 $a>0,K>0$ 下，正产量一阶条件为 $a-2Q-\mu=0$，配合上界互补松弛，得到 $$ \begin{aligned} Q&=\min(a/2,K),\\ P&=a-Q,\\ \mu&=P-Q=a-2Q. \end{aligned} $$ 这是单期竞争教学例，不是论文的垄断竞争动态模型。它保留了同样重要的对象区别：成本、容量和影子价格。 | 教学情景 | a | K | Q | P | μ | |---|---:|---:|---:|---:|---:| | 需求较低，容量松弛 | 100 | 80 | 50 | 50 | 0 | | 同一容量，需求增加 | 120 | 80 | 60 | 60 | 0 | | 需求不变，容量较紧 | 120 | 50 | 50 | 70 | 20 | | 已受限后需求继续增加 | 130 | 50 | 50 | 80 | 30 | | 恰好触边 | 100 | 50 | 50 | 50 | 0 | 在松弛内点，需求增加使价格和数量都上升；约束已经生效时，需求继续增加可以只推价格、不推数量。这是给“价格↑、数量↑是需求冲击”直觉加上必要条件，而不是取消供需分析。
## 三、从不可直接观察的容量走到模型估计 论文§3使用消费品与服务价格、消费支出、工业生产、劳动生产率以及进口价格和支出等观测，将不同系列映射到模型变量。原始系列有自己的统计定义；例如工业生产在这里是商品部门产出的代理，不是精确相同的对象。增长和通胀序列按作者处理去均值。[^data] 容量约束是否生效不是直接从一张“全国容量”表读取。作者用约束持续时间和结构参数刻画分段状态方程，给定候选参数与持续时间后滤出平滑冲击，再检查产生的模型路径是否与候选约束状态相容。基准先验只允许2020Q2起约束生效，因此更早时期的零乘子并不是由数据自由发现的普遍事实。[^data] 这一步告诉我们如何评价研究结果：必须同时看观测、模型映射、先验与近似解。论文还说明，数值上的分段线性近似可能产生小的负乘子残差；理论KKT要求非负，不应把数值近似误差解释成“真实产能约束具有负影子成本”。[^counter] 我们不在本篇重估后验或重新运行Dynare；但可以准确重建每次实验固定什么、改变什么、最后比较什么。这已经比只复述“产能影响通胀”多走了最重要的一步。 ## 四、Figure 7：同一组冲击，拿掉约束 论文§4.2的迭代可用一张流程表重建。对每次后验抽样 $b$，先得到参数 $\theta^{(b)}$ 和约束持续时间，再在**允许约束生效的模型**中从数据滤出冲击路径 $\hat\varepsilon^{(b)}$。随后保持这些冲击，改用始终松弛的约束运行模型。[^counter] $$ \begin{aligned} X^{(b)}_{\mathrm{constrained}} &=F(\theta^{(b)},\hat\varepsilon^{(b)};\text{允许绑定}),\\ X^{(b)}_{\mathrm{slack}} &=F(\theta^{(b)},\hat\varepsilon^{(b)};\text{始终松弛}). \end{aligned} $$ 这里的 $F$ 是本篇为解释作者求解流程所用的记号。关键不是重新把松弛模型估到现实数据，而是对同一次抽样使用同一组冲击：否则两条路径的差还会混入重新估出的冲击变化。 Figure 7的纵轴是**去均值的季度通胀年化值**，不是同比通胀。若季度对数变动为 $\pi_t$，年化尺度是 $4\pi_t$；以百分比显示时还要乘100。图比较原始数据与反事实中位数，区间来自后验模拟，并保留观测的测量误差处理。约1000次后验抽样形成的第5至95分位区间，不是本篇静态例子能够生成的置信带。[^counter] 实际读图要把面板、期间和分母分开。Figure 7c对应总体（aggregate/headline）消费价格通胀：在2021–2022年，实际路径比始终松弛约束的反事实高约1–2个百分点。另一方面，**Figure 7a及其正文说“约一半的加速”时，分母是2020Q2–2021Q2的商品价格通胀加速**；正文还另讨论2022年后半年商品通胀回落。这不是把总体通胀的“2个百分点”除以某个“4个百分点”所得的永久50%比例。图7的商品、服务和总体面板必须分别读，不能把更窄对象/期间的份额改写成2021–2022总体通胀的固定贡献率。[^counter] 我们用四期**教学变式**演示“同冲击、不同约束”的运算，不抄造论文曲线。输入需求截距 $a=(100,120,130,110)$，容量 $K=(80,50,50,80)$，逐期使用上一节静态规则；拿掉上限后，同一 $a$ 得到 $P=Q=a/2$。 | 教学期 | 相同需求a | 容量K | 受限价格 | 始终松弛价格 | 价格差 | |---|---:|---:|---:|---:|---:| | 1 | 100 | 80 | 50 | 50 | 0 | | 2 | 120 | 50 | 70 | 60 | 10 | | 3 | 130 | 50 | 80 | 65 | 15 | | 4 | 110 | 80 | 55 | 55 | 0 | 这里是抽象价格单位，不是通胀百分点；逐期独立的静态重算也没有学习预期或跨期价格调整。它只让我们亲手确认反事实比较固定的是输入，而不是把一条曲线任意下移。
## 五、Figure 9：容量改变其他冲击的效果 Figure 9a让各类冲击分别进入模型；Figure 9b则把某类冲击与容量冲击一起放入，再减去容量冲击单独作用的路径。作者用这一步观察同一种冲击在容量路径不同的背景下怎样表现。[^interaction] 令 $G(u,v)$ 表示给定其他模型条件时，具名冲击 $u$ 与容量冲击 $v$ 产生的某一结果。Figure 9b对应的比较结构是 $$ G(u,v)-G(0,v), $$ 而不是单纯的 $G(u,0)-G(0,0)$。它们的差是交互项： $$ J=G(u,v)-G(0,v)-G(u,0)+G(0,0). $$ 用同一个静态教学模型，基准 $a=100,K=80$；需求冲击把 $a$ 增加30，容量冲击把 $K$ 减少30： | 输入组合 | a | K | 价格P | |---|---:|---:|---:| | 无冲击 | 100 | 80 | 50 | | 只有需求冲击 | 130 | 80 | 65 | | 只有容量冲击 | 100 | 50 | 50 | | 两者一起 | 130 | 50 | 80 | 单独需求效应是15；在较紧容量下，条件需求效应是 $80-50=30$，交互项为15。容量冲击单独出现时只是恰好触边，价格变化为0，却不能据此说它与联合结果无关。这正是非线性的重要之处。 因此把“需求单独贡献15、容量单独贡献0”相加，不能解释联合变化30。分配交互项需要另外定义归因规则，论文的Figure 9b不是自动加总为100%的分账表。Figure 7改变约束制度而固定全套冲击；Figure 9改变冲击组合并允许约束响应。两个实验回答不同问题。
## 六、重建与迁移 任务一：$a=100,K=50$时为什么产量在上限，但乘子为0？把a再提高到110会怎样？ 解析：无约束最优产量正好是50，增加容量没有边际收益，所以触边不必有正乘子。a升到110后，产量仍50，价格60，$\mu=10$；此时放宽容量才能增加可满足需求。 任务二：有人为“无约束反事实”重新估计了一套更小的需求冲击，然后比较通胀。它还是本篇Figure 7的实验吗？ 解析：不是。比较中同时改变了约束和冲击来源，差距不能只归于放松约束。作者流程先在允许约束的模型中滤出冲击，再保持同组冲击进入无约束模型；测量误差及后验抽样也要沿其原约定处理。 任务三：用上一节四格价格重算Figure 9b式差分、无容量冲击的需求效应和交互项。 解析：依次为 $80-50=30$、$65-50=15$、$30-15=15$。它们都是教学价格差，不是研究论文的通胀贡献。若把三者都作为独立贡献相加，会再次重复计数。 迁移任务：一家制造商的产品价格上涨，但实物交付持平。这能否直接识别其产能约束？ 解析：可以提出容量绑定的候选机制：需求增强时只能涨价、不增产。但还可能有产品组合、计划检修、配给或价格调整时点等解释。需要可用产能、排产、实物订单和交期；论文中的宏观模型结果不能直接识别这家公司的瓶颈，更不能直接推出利润。可迁移的是“把约束和冲击共同建模”的方法，而非某一个宏观份额。 [^paper]: EIDEF-S20，Diego Comin、Robert C. Johnson、Callum Jones，*Supply Chain Constraints and Inflation*，IMES Discussion Paper 2025-E-15，2025年10月；固定72页公开版本。[全文](https://www.imes.boj.or.jp/research/papers/english/25-E-15.pdf)。 [^model]: 同源§2.2.2，印刷pp.12–13（PDF物理pp.15–16），生产、价格条件与互补松弛式(6)–(7)；§2.6.1印刷pp.17–18（PDF pp.20–21），定价关系。本文静态总剩余例是另设教学模型，不是作者企业问题的完整复制。 [^data]: 同源§3，印刷pp.21–26（PDF pp.24–29），数据、参数、持续时间与先验；季度美国资料1990Q1–2023Q4。未重新估计底层序列或后验。 [^counter]: 同源§4.1–4.2，印刷pp.27–30（PDF pp.30–33）；Figure 7位于印刷p.29／PDF p.32，脚注34–35含近似乘子、测量误差说明。本文保留其反事实流程和图读对象，不提供伪造的作者逐期模拟数据。 [^interaction]: 同源§4.3，印刷pp.31–33（PDF pp.34–36）；Figure 9位于印刷p.32／PDF p.35，图下正文明确“联合冲击路径减容量冲击单独路径”。 ## Additional teaching material ### 图表的静态等价与默认结果 K80/a100→120：P,Q均50→60；K50/a120→130：Q50，P70→80，μ20→30；a100,K50：触边μ0。 ```json { "a": 120, "K": 50, "q": 50, "p": 70, "mu": 20, "slack": 0, "complementarity": 0, "regime": "绑定且乘子为正", "identity": "one-period competitive teaching model; abstract price/output, not the IMES model" } ``` scale1时受限价格[50,70,80,55]，松弛[50,60,65,55]，价格差[0,10,15,0]。均为抽象价格单位。 ```json { "rows": [ { "period": 1, "a": 100, "K": 80, "constrained_q": 50, "constrained_p": 50, "slack_q": 50, "slack_p": 50, "difference": 0, "mu": 0 }, { "period": 2, "a": 120, "K": 50, "constrained_q": 50, "constrained_p": 70, "slack_q": 60, "slack_p": 60, "difference": 10, "mu": 20 }, { "period": 3, "a": 130, "K": 50, "constrained_q": 50, "constrained_p": 80, "slack_q": 65, "slack_p": 65, "difference": 15, "mu": 30 }, { "period": 4, "a": 110, "K": 80, "constrained_q": 55, "constrained_p": 55, "slack_q": 55, "slack_p": 55, "difference": 0, "mu": 0 } ], "identity": "same demand sequence, constraint switched off; additional teaching data, not Fig.7 posterior simulation" } ``` 默认价格[50,65,50,80]；单独需求15、容量单独0、容量背景下需求30、交互15。教学价格，不是作者Figure9数值。 ```json { "baseline": { "a": 100, "K": 80, "q": 50, "p": 50, "mu": 0, "slack": 30, "complementarity": 0, "regime": "松弛", "identity": "one-period competitive teaching model; abstract price/output, not the IMES model" }, "demand_only": { "a": 130, "K": 80, "q": 65, "p": 65, "mu": 0, "slack": 15, "complementarity": 0, "regime": "松弛", "identity": "one-period competitive teaching model; abstract price/output, not the IMES model" }, "capacity_only": { "a": 100, "K": 50, "q": 50, "p": 50, "mu": 0, "slack": 0, "complementarity": 0, "regime": "恰好触边，乘子为零", "identity": "one-period competitive teaching model; abstract price/output, not the IMES model" }, "joint": { "a": 130, "K": 50, "q": 50, "p": 80, "mu": 30, "slack": 0, "complementarity": 0, "regime": "绑定且乘子为正", "identity": "one-period competitive teaching model; abstract price/output, not the IMES model" }, "demand_effect_without_capacity": 15, "capacity_alone_effect": 0, "conditional_demand_effect": 30, "interaction": 15, "total_effect": 30, "identity": "price-unit four-state teaching example; difference operation mirrors Fig.9b, not estimated inflation" } ``` ## Experiment inputs and static equivalents ```json [ { "id": "exp-ei17-fig7-relax-constraints", "node_id": "EI-17", "title": "同一输入路径，放松约束", "anchor": "ei17-fig7", "method": "relax", "controls": [ { "key": "capacityScale", "label": "教学容量路径倍数", "default": 1, "min": 0.5, "max": 2, "step": 0.1, "kind": "number" } ], "inputs": { "capacityScale": 1 }, "algorithm": "新增教学路径a=[100,120,130,110],K=[80,50,50,80]×scale；受限逐期用静态模型，松弛逐期P=Q=a/2。不是论文通胀，不生成后验置信带。作者实验流程另列同θ、同smoothed shocks、all-slack反事实及measurement error。", "static_equivalent": "scale1时受限价格[50,70,80,55]，松弛[50,60,65,55]，价格差[0,10,15,0]。均为抽象价格单位。", "source_ids": [ "EIDEF-S20" ], "identity": "teaching_assumption", "description": "同一输入路径，放松约束", "outputs": { "rows": [ { "period": 1, "a": 100, "K": 80, "constrained_q": 50, "constrained_p": 50, "slack_q": 50, "slack_p": 50, "difference": 0, "mu": 0 }, { "period": 2, "a": 120, "K": 50, "constrained_q": 50, "constrained_p": 70, "slack_q": 60, "slack_p": 60, "difference": 10, "mu": 20 }, { "period": 3, "a": 130, "K": 50, "constrained_q": 50, "constrained_p": 80, "slack_q": 65, "slack_p": 65, "difference": 15, "mu": 30 }, { "period": 4, "a": 110, "K": 80, "constrained_q": 55, "constrained_p": 55, "slack_q": 55, "slack_p": 55, "difference": 0, "mu": 0 } ], "identity": "same demand sequence, constraint switched off; additional teaching data, not Fig.7 posterior simulation" }, "input_uri": "https://ou-liu-red-sugar.github.io/notebook/labs/ei-defgh/inputs.json", "implementation": { "html": "/notebook/labs/ei-defgh/interactions.html?embed=1&experiment=exp-ei17-fig7-relax-constraints", "engine": "/notebook/labs/ei-defgh/engine.js", "static_available_without_js": true } }, { "id": "exp-ei17-fig9-interaction", "node_id": "EI-17", "title": "条件效应与非线性交互", "anchor": "ei17-fig9", "method": "interaction", "controls": [ { "key": "demandShock", "label": "教学需求截距增量u", "default": 30, "min": 0, "max": 80, "step": 5, "kind": "number" }, { "key": "capacityReduction", "label": "教学容量减少v", "default": 30, "min": 0, "max": 60, "step": 5, "kind": "number" } ], "inputs": { "demandShock": 30, "capacityReduction": 30 }, "algorithm": "基准a100,K80；分别计算P(0,0),P(u,0),P(0,v),P(u,v)。conditional=P(u,v)−P(0,v)；interaction=conditional−[P(u,0)−P(0,0)]。非线性不强制贡献加总100%。", "static_equivalent": "默认价格[50,65,50,80]；单独需求15、容量单独0、容量背景下需求30、交互15。教学价格，不是作者Figure9数值。", "source_ids": [ "EIDEF-S20" ], "identity": "teaching_assumption", "description": "条件效应与非线性交互", "outputs": { "baseline": { "a": 100, "K": 80, "q": 50, "p": 50, "mu": 0, "slack": 30, "complementarity": 0, "regime": "松弛", "identity": "one-period competitive teaching model; abstract price/output, not the IMES model" }, "demand_only": { "a": 130, "K": 80, "q": 65, "p": 65, "mu": 0, "slack": 15, "complementarity": 0, "regime": "松弛", "identity": "one-period competitive teaching model; abstract price/output, not the IMES model" }, "capacity_only": { "a": 100, "K": 50, "q": 50, "p": 50, "mu": 0, "slack": 0, "complementarity": 0, "regime": "恰好触边，乘子为零", "identity": "one-period competitive teaching model; abstract price/output, not the IMES model" }, "joint": { "a": 130, "K": 50, "q": 50, "p": 80, "mu": 30, "slack": 0, "complementarity": 0, "regime": "绑定且乘子为正", "identity": "one-period competitive teaching model; abstract price/output, not the IMES model" }, "demand_effect_without_capacity": 15, "capacity_alone_effect": 0, "conditional_demand_effect": 30, "interaction": 15, "total_effect": 30, "identity": "price-unit four-state teaching example; difference operation mirrors Fig.9b, not estimated inflation" }, "input_uri": "https://ou-liu-red-sugar.github.io/notebook/labs/ei-defgh/inputs.json", "implementation": { "html": "/notebook/labs/ei-defgh/interactions.html?embed=1&experiment=exp-ei17-fig9-interaction", "engine": "/notebook/labs/ei-defgh/engine.js", "static_available_without_js": true } }, { "id": "exp-ei17-static-capacity-complementarity", "node_id": "EI-17", "title": "产能互补松弛", "anchor": "ei17-kkt", "method": "capacity", "controls": [ { "key": "a", "label": "逆需求截距a", "default": 120, "min": 0, "max": 200, "step": 5, "kind": "number" }, { "key": "K", "label": "产能K", "default": 50, "min": 1, "max": 150, "step": 5, "kind": "number" } ], "inputs": { "a": 120, "K": 50 }, "algorithm": "最大化aQ−Q²,0≤Q≤K。a≥0,K>0；Q=min(a/2,K),P=a−Q,μ=a−2Q。显示slack与μ(Q−K)=0。a=0采用Q=P=μ=0。模型为静态教学经济，不是IMES复现。", "static_equivalent": "K80/a100→120：P,Q均50→60；K50/a120→130：Q50，P70→80，μ20→30；a100,K50：触边μ0。", "source_ids": [ "EIDEF-S20" ], "identity": "teaching_assumption", "description": "产能互补松弛", "outputs": { "a": 120, "K": 50, "q": 50, "p": 70, "mu": 20, "slack": 0, "complementarity": 0, "regime": "绑定且乘子为正", "identity": "one-period competitive teaching model; abstract price/output, not the IMES model" }, "input_uri": "https://ou-liu-red-sugar.github.io/notebook/labs/ei-defgh/inputs.json", "implementation": { "html": "/notebook/labs/ei-defgh/interactions.html?embed=1&experiment=exp-ei17-static-capacity-complementarity", "engine": "/notebook/labs/ei-defgh/engine.js", "static_available_without_js": true } } ] ``` ## Sources - [Supply Chain Constraints and Inflation](https://www.imes.boj.or.jp/research/papers/english/25-E-15.pdf): 同冲击、改约束的Figure 7反事实：Figure 7c在2021–2022总体消费价格通胀约高1–2pp；Figure 7a“约一半”对应2020Q2–2021Q2商品价格通胀加速，二者不是同一分母。Figure 9联合冲击减容量单独路径是另一运算；近似乘子、先验、测量误差与后验区间保留。 ## Content relations ```json [ { "from": "zh-ei17", "relation": "part_of", "to": "industry-evidence", "reason": "主要 topic 归属" }, { "from": "zh-ei17", "relation": "uses_method", "to": "zh-ei03", "reason": "需求冲击" }, { "from": "zh-ei17", "relation": "uses_method", "to": "zh-ei04", "reason": "容量上限" }, { "from": "zh-ei17", "relation": "uses_method", "to": "zh-ei05", "reason": "投入网络" }, { "from": "zh-ei17", "relation": "uses_method", "to": "zh-ei09", "reason": "动态传导" }, { "from": "zh-ei17", "relation": "compares_with", "to": "zh-ei16", "reason": "原表综合阅读与结构模型研究" }, { "from": "ei17-question", "relation": "supported_by", "to": "EIDEF-S20", "reason": "容量互补松弛；同冲击松弛约束反事实与联合冲击减容量单独路径是不同运算。近似乘子、先验、测量误差与后验区间保留。", "locator": "§2.2.2 printed12–13；§2.6.1 printed17–18；§3 printed21–26；§4.1–4.3 printed27–33；Fig7 p29/PDF32；Fig9 p32/PDF35", "scope": "对应段采用的具名单元，不扩称完整论文复现", "at_section": "ei17-question" }, { "from": "ei17-kkt", "relation": "supported_by", "to": "EIDEF-S20", "reason": "容量互补松弛；同冲击松弛约束反事实与联合冲击减容量单独路径是不同运算。近似乘子、先验、测量误差与后验区间保留。", "locator": "§2.2.2 printed12–13；§2.6.1 printed17–18；§3 printed21–26；§4.1–4.3 printed27–33；Fig7 p29/PDF32；Fig9 p32/PDF35", "scope": "对应段采用的具名单元，不扩称完整论文复现", "at_section": "ei17-kkt" }, { "from": "ei17-estimation", "relation": "supported_by", "to": "EIDEF-S20", "reason": "容量互补松弛；同冲击松弛约束反事实与联合冲击减容量单独路径是不同运算。近似乘子、先验、测量误差与后验区间保留。", "locator": "§2.2.2 printed12–13；§2.6.1 printed17–18；§3 printed21–26；§4.1–4.3 printed27–33；Fig7 p29/PDF32；Fig9 p32/PDF35", "scope": "对应段采用的具名单元，不扩称完整论文复现", "at_section": "ei17-estimation" }, { "from": "ei17-fig7", "relation": "supported_by", "to": "EIDEF-S20", "reason": "容量互补松弛；同冲击松弛约束反事实与联合冲击减容量单独路径是不同运算。近似乘子、先验、测量误差与后验区间保留。", "locator": "§2.2.2 printed12–13；§2.6.1 printed17–18；§3 printed21–26；§4.1–4.3 printed27–33；Fig7 p29/PDF32；Fig9 p32/PDF35", "scope": "对应段采用的具名单元，不扩称完整论文复现", "at_section": "ei17-fig7" }, { "from": "ei17-fig9", "relation": "supported_by", "to": "EIDEF-S20", "reason": "容量互补松弛；同冲击松弛约束反事实与联合冲击减容量单独路径是不同运算。近似乘子、先验、测量误差与后验区间保留。", "locator": "§2.2.2 printed12–13；§2.6.1 printed17–18；§3 printed21–26；§4.1–4.3 printed27–33；Fig7 p29/PDF32；Fig9 p32/PDF35", "scope": "对应段采用的具名单元，不扩称完整论文复现", "at_section": "ei17-fig9" }, { "from": "ei17-kkt", "relation": "illustrated_by", "to": "exp-ei17-static-capacity-complementarity", "reason": "K80/a100→120：P,Q均50→60；K50/a120→130：Q50，P70→80，μ20→30；a100,K50：触边μ0。", "at_section": "ei17-kkt" }, { "from": "ei17-fig7", "relation": "illustrated_by", "to": "exp-ei17-fig7-relax-constraints", "reason": "scale1时受限价格[50,70,80,55]，松弛[50,60,65,55]，价格差[0,10,15,0]。均为抽象价格单位。", "at_section": "ei17-fig7" }, { "from": "ei17-fig9", "relation": "illustrated_by", "to": "exp-ei17-fig9-interaction", "reason": "默认价格[50,65,50,80]；单独需求15、容量单独0、容量背景下需求30、交互15。教学价格，不是作者Figure9数值。", "at_section": "ei17-fig9" } ] ``` ## Related entries