← BACK TO HOME
CM216E | WEEK 09 | PROF. Sonic Wu

區域整合
決策樹 Regional Integration
Decision Tree

Rules of Origin (ROO) / RVC 與供應鏈遷移 (Remapping) Rules of Origin (ROO), RVC & Supply-Chain Remapping

今日核心任務：把自由貿易協定（FTA）的「原產地規則 (Rules of Origin)」與「區域價值含量 (RVC)」翻成可執行的決策樹。你要引導 AI 幫你搭出「Tariff Calculator」的邏輯，而不是直接問答案；並用影片的「潤學」成本比較模型，做出供應鏈遷移/供應商替換的決策。課後：匯出對話紀錄 Log 3、閱讀 Hill Ch.9，並記錄你的認知負荷變化。 Core mission today: translate FTA rules—Rules of Origin (ROO) and Regional Value Content (RVC)—into an executable decision tree. Use AI to build the logic of a “Tariff Calculator” (do not ask for the final answer), and apply the video’s cost-comparison (“run”) mindset to supply-chain remapping decisions. After class: export Chat Log 3, read Hill Ch.9, and record your cognitive-load change.

研究視角：認知負荷 Research Lens: Cognitive Load

本週把「區域整合」從抽象名詞拉回到實務：你會發現 ROO/RVC 本質上是一套法規版演算法。我們要觀察：AI 是否能有效降低你在處理複雜法規時的認知負荷？ This week turns “regional integration” into operational reality: ROO/RVC are basically regulation-as-algorithm. We track whether AI meaningfully reduces your cognitive load when handling complex rules.

  • 課後作業：匯出對話紀錄 (Log 3)；閱讀 Hill Ch.9。 After class: export Chat Log 3; read Hill Ch.9.
  • 本頁輸入會自動儲存（存在你的瀏覽器 localStorage）。 Your inputs autosave (stored in your browser localStorage).

關鍵轉折：自由貿易 ≠ 低門檻 Critical Shift: “Free Trade” ≠ Low Friction

自由貿易協定會降關稅，但也會透過原產地規則設下「入場門檻」。結果可能不是效率提升，而是貿易轉移 (Trade Diversion)：企業為了合規而「潤」，不一定是因為技術更好。 FTAs lower tariffs, but they also create “gates” via rules of origin. The outcome may not be efficiency—it can be trade diversion: firms “run” for compliance, not necessarily because capabilities improved.

Pre-test 1：為什麼墨西哥汽車零件出口變多，不必然代表技術突飛猛進？（提示：USMCA、RVC、貿易轉移） Pre-test 1: Why can a surge in Mexico’s auto-parts exports after USMCA be “trade diversion,” not “technology progress”?
Pre-test 2：為什麼 RCEP 像「鬆散社團」，而 USMCA 更像「排他堡壘」？（提示：門檻 40% vs 75% + 工資條款） Pre-test 2: Why is RCEP a “loose club,” while USMCA behaves like an “exclusive fortress” (40% vs 75% + wage rules)?

THEORY: REGIONAL INTEGRATION

區域整合不只是「簽了協定、關稅變低」。真正的難點是：協定為了防止非成員「搭便車」，會設計一整套原產地規則 (Rules of Origin, ROO) 與區域價值含量 (Regional Value Content, RVC)。這套規則通常不是人腦直覺能完成的，你需要把它翻譯成決策樹——而這正是本週的 Algorithmic Thinking。 Regional integration is not just “tariffs go down after an agreement.” To prevent non-members from free-riding, FTAs impose Rules of Origin (ROO) and Regional Value Content (RVC). These checks rarely fit human intuition—you must translate them into a decision tree. That translation is today’s Algorithmic Thinking.

輸入層 INPUTS
規則層 ROO LOGIC
決策層 RVC DECISION
> 點擊上方任一層，載入「Tariff Calculator」邏輯。 > CLICK A LAYER ABOVE TO LOAD THE “TARIFF CALCULATOR” LOGIC.

Club vs Fortress

同樣叫 FTA，但「門檻設計」差很多： Both are FTAs, but the “gate design” differs sharply:

  • RCEP：像鬆散社團，門檻較低（常見約 40% RVC）且允許累積規則 (cumulation)，有利於亞洲供應鏈分工。 RCEP: a loose club—lower thresholds (often ~40% RVC) and cumulation, encouraging fragmented supply chains across Asia.
  • USMCA：像排他堡壘，門檻高（汽車常見 75% RVC）甚至加上工資條款，目標是把供應鏈鎖在北美，降低亞洲比重。 USMCA: an exclusive fortress—high thresholds (auto often 75% RVC) plus wage rules, designed to lock supply chains inside North America and reduce Asia exposure.
  • 副作用：可能造成貿易轉移：企業為免稅而改找區域內供應商，未必是效率最優。 Side effect: trade diversion: firms switch to inside-bloc suppliers for duty-free status, not necessarily because it is the efficiency optimum.

RVC = The Math of Survival

RVC 最常見的形式之一（Net Cost Method）： A common RVC form (Net Cost Method):

RVC = (NC − VNM) / NC × 100%
  • NC：淨成本（例如出廠價扣除行銷等）。 NC: net cost (e.g., ex-works cost excluding marketing).
  • VNM：非原產材料價值（從區域外進口的零件總值）。 VNM: value of non-originating materials (inputs sourced outside the region).

微型例子：若 NC=$1,000、VNM=$300，則 RVC=70%。若門檻是 75%，不合格；你就得「Remapping」：把區域外零件換成區域內供應商，哪怕貴一點，也可能因免稅而總成本更低。 Micro example: if NC=$1,000 and VNM=$300, RVC=70%. If the threshold is 75%, you fail; you must “remap” inputs toward in-region suppliers—even if slightly more expensive—because duty savings can dominate.

提示：把法規「編碼」成決策樹 Tip: Encode Regulations into a Decision Tree

你會在後面的 Tariff Lab 用到一個簡化版流程：先確認 FTA 適用與累積，再算 RVC，最後做「合格/不合格」判斷與供應鏈調整。這就像現實世界的貿易合規軟體（例如 Thomson Reuters ONESOURCE）在做的事：人機分工，AI/系統處理重複算術，人類處理策略取捨。 In the Tariff Lab you will apply a simplified pipeline: verify FTA scope + cumulation, compute RVC, then decide pass/fail and propose supply-chain adjustments. This mirrors real-world trade compliance software (e.g., Thomson Reuters ONESOURCE): machines do repeatable math; humans do strategic trade-offs.

CASE STUDY: THE TARIFF LAB

本週的個案不是戲劇，而是「關稅與合規」的現場：當歐美對中國電動自行車課高關稅、同時 RCEP 提供新的免稅通道，企業的供應鏈決策會被 ROO/RVC 重新洗牌。 This week’s case is not a drama episode but a live “tariff + compliance” scene: when high tariffs hit China-made e-bikes while RCEP offers new duty-free pathways, ROO/RVC can reshuffle supply-chain decisions.

SUPPLY CHAIN REMAPPING THE TARIFF LAB

情境：Giant-X（台灣自行車製造商） Scenario: Giant‑X (Taiwan bicycle maker)

Giant‑X 的主要工廠在台中與昆山（中國）。面對歐美對中國電動自行車的高關稅，以及 RCEP 生效後的區域免稅機會，管理團隊必須做供應鏈重組。 Giant‑X runs major plants in Taichung and Kunshan (China). Facing high tariffs on China-made e-bikes and new regional duty-free opportunities under RCEP, the management team must remap the supply chain.

任務（分組） Team Task

  • 決定是否把車架焊接移到越南，或把組裝線移到波蘭（歐盟內）。 Decide whether to move frame welding to Vietnam, or move the assembly line to Poland (inside the EU).
  • 用 AI 先把 ROO/RVC 變成「可計算」的流程，再把結果轉成策略建議。 Use AI to convert ROO/RVC into a computable flow, then translate results into a strategic recommendation.

（課堂假設門檻：RCEP 40%、EU 45%。） (Class assumption: RCEP 40%, EU 45% thresholds.)

課堂用虛擬 BOM（可調） Class “Virtual BOM” (editable)

為避免同學不知道電池、馬達行情而導致 AI 幻覺，本頁提供一份標準化 BOM 當作起點。 To avoid price confusion and AI hallucination, we provide a standardized BOM as a starting point.

零件Component 來源地Origin 成本 (USD)Cost (USD)
Battery China 200
Motor Japan 150
Frame Vietnam 100
Other parts Taiwan 50
這不是「問 AI 答案」的遊戲，而是把 AI 當作計算引擎，逼你看懂規則的結構。 This is not “ask AI for the answer.” It is using AI as a computation engine so you can see the structure of the rule.

Survival Glossary

ROO (Rules of Origin)

原產地規則：判斷產品是否「算作」協定區域內原產，以決定能否享有優惠關稅。核心目的：防止區域外成員搭便車。 Rules of Origin: determines whether a product counts as originating within the FTA region and thus qualifies for preferential tariffs. Core purpose: prevent free-riding.

RVC (Regional Value Content)

區域價值含量：用數學方式要求「區域內價值」達到門檻（例如 40% / 45% / 75%）。 Regional Value Content: a quantitative requirement that regional value exceeds a threshold (e.g., 40% / 45% / 75%).

Cumulation

累積規則：允許把多個成員國的投入視為「區域內」，降低供應鏈分工的合規摩擦。 Cumulation: allows inputs from multiple member countries to count as “in-region,” reducing friction for cross-country fragmentation.

Landed Cost

落地成本：出廠價 + 運輸/保險 + 關稅等。供應商比較不能只看單價。 Landed cost: ex-works price + freight/insurance + tariffs, etc. Supplier comparisons cannot rely on unit price alone.

快速檢查點 Quick Checkpoints

USMCA：比較「免稅但貴」vs「便宜但有稅」的落地成本，可能逼你改向墨西哥採購。 USMCA: landed-cost comparison (“duty-free but pricey” vs “cheap but taxed”) can force sourcing toward Mexico.

RCEP：累積規則讓中國布料可算入區域內原產，幫助越南成衣達到 40% 門檻。 RCEP: cumulation lets China-sourced fabric count as originating, helping Vietnam-made garments meet the 40% threshold.

教學設計提醒 Teaching Design Note

真實世界的合規工作通常由軟體 + 人工審核完成。課堂上請用「同一份虛擬 BOM / 物流費率」作為共同基準，避免因資料不一致造成錯誤比較。 In practice, compliance work is done by software + human review. In class, use the same “virtual BOM / logistics rates” baseline to avoid inconsistent data and invalid comparisons.

THE TARIFF CALCULATOR CHALLENGE

本週 AI 技能是 Algorithmic Thinking：你要引導 AI 把 ROO/RVC 的規則拆成「If–Then」決策樹，而不是直接索取答案。接著，套用影片的「潤學」成本比較模型，把計算結果翻成供應鏈遷移建議。 This week’s AI skill is Algorithmic Thinking: guide AI to decompose ROO/RVC into an If–Then decision tree—instead of asking for the final answer. Then apply the video’s cost-comparison (“run”) mindset to convert calculations into a supply-chain remapping recommendation.

User: Compliance_Analyst @ YZU_CM438_Lab

[CHECK-IN] 認知負荷自評 [CHECK-IN] Cognitive Load Self‑Rating (Before / After)

目的：回填你在「沒有 AI」與「用 AI 當腳手架」兩種狀態下，處理 ROO/RVC 的主觀負荷。先填 Before；完成兩個任務後再回來填 After。 Purpose: record your subjective load when handling ROO/RVC without AI vs with AI as scaffolding. Fill “Before” now; after completing Tasks 1–2, return and fill “After.”

[TASK 1] Tariff Calculator：決策樹 [TASK 1] Tariff Calculator: Decision Tree (The Logic)

任務：把 ROO/RVC 寫成「可執行的判斷流程」。至少要包含：FTA 適用/會員 → 累積規則 → RVC 計算方法與門檻 → 合格/不合格 → 不合格時的 Remapping（換供應商/換製程/換組裝地）。 Task: write ROO/RVC as an executable decision flow. Minimum nodes: FTA scope/membership → cumulation → RVC method + threshold → pass/fail → remapping actions if fail (change suppliers/process/assembly location).

[TASK 2] Tariff Lab：用 AI 做試算與建議 [TASK 2] Tariff Lab: Use AI to Compute & Recommend (AI as Calculator)

把 AI 當成合規助理：請它依照你在 Task 1 設計的流程，完成 RVC 試算與供應鏈 Remapping 建議。注意：你要引導它「先建流程、再算」，避免黑箱答案。 Treat AI as a compliance assistant: make it follow the flow you designed in Task 1 to compute RVC and propose remapping actions. Key: make it “build the flow first, then compute,” to avoid black-box answers.

"你是資深貿易合規與原產地規則專家。請先用條列式寫出一個『Tariff Calculator』決策樹（FTA 適用→累積→RVC→判斷→Remapping）。 接著依照下列情境做試算，使用公式 RVC=(NC−VNM)/NC×100%。 - FTA: RCEP - 出口：越南製造後出口到澳洲 - 門檻：RVC 40% - 假設 NC（淨成本）= 零件成本總和 - BOM：電池(中國) $200、馬達(日本) $150、車架(越南) $100、其他零件(台灣) $50 請你： (1) 判斷哪些投入算『區域內原產』、哪些算 VNM (2) 算出 RVC，並給出是否合格 (3) 若不合格，提出 2 個 Remapping 方案（例如把某零件換成 RCEP 成員國供應），並重新估算 RVC（可自訂合理價格差異，務必說明假設）。" "You are a senior trade compliance and rules-of-origin expert. First, write a bullet-point ‘Tariff Calculator’ decision tree (FTA scope → cumulation → RVC → decision → remapping). Then compute the following case using RVC=(NC−VNM)/NC×100%. - FTA: RCEP - Export: made in Vietnam, exported to Australia - Threshold: RVC 40% - Assume NC (net cost) = total component cost - BOM: Battery (China) $200, Motor (Japan) $150, Frame (Vietnam) $100, Other parts (Taiwan) $50 Deliver: (1) Identify which inputs count as originating vs VNM (2) Compute RVC and state pass/fail (3) If fail, propose 2 remapping options (e.g., switch a component to an RCEP member supplier) and re-estimate RVC (you may assume reasonable price deltas, but state assumptions clearly)."
"你是採購與供應鏈策略顧問。請用『落地成本 (Landed Cost)』模型，示範如何用算式做出供應商選擇。 情境：特斯拉得州廠要買方向盤。 - 選項 A：中國供應商，單價 $50，運費 $5，美國對中關稅 25% - 選項 B：墨西哥供應商，單價 $60，運費 $2，符合 USMCA 原產地規則（免稅） 請你： (1) 計算 A 與 B 的 landed cost 並比較 (2) 用影片中的『潤學』視角，把它整理成一份 8 行以內的 Decision Memo：包含「訊號 vs 雜訊」、「風險是否不對稱」、「行動觸發點（threshold）」 (3) 最後給出一個可以泛化到 Giant‑X Tariff Lab 的決策框架（不用新數據）。" "You are a procurement and supply-chain strategy advisor. Use a landed-cost model to show how equations drive supplier choice. Scenario: Tesla Texas plant is sourcing steering wheels. - Option A: China supplier, unit price $50, freight $5, US tariff on China 25% - Option B: Mexico supplier, unit price $60, freight $2, qualifies under USMCA (duty-free) Deliver: (1) Compute and compare landed cost for A vs B (2) Using the video’s ‘run’ mindset, write an 8-line decision memo including ‘signal vs noise,’ ‘risk asymmetry,’ and an ‘action trigger threshold’ (3) Generalize into a decision framework that can be applied to the Giant‑X Tariff Lab without requiring new data."

學習檢核 (Ticket to Leave) Performance Check (Ticket to Leave)

在離開前，請確認你能回答以下問題： Before you leave, make sure you can answer:

  1. 貿易轉移：為什麼 USMCA 可能讓企業「潤」向墨西哥，但不代表全球效率更好？ Trade diversion: why can USMCA push firms to “run” toward Mexico without improving global efficiency?
  2. 累積規則：在 RCEP 情境下，中國布料為何可視為區域內原產？這如何影響越南成衣出口日本的合規難度？ Cumulation: under RCEP, why can China-sourced fabric count as originating, and how does that reduce compliance friction for Vietnam-to-Japan exports?
  3. 演算法化：請寫出 4～6 行 pseudo-code，描述你的 Tariff Calculator 如何判斷「免稅/不免稅」，以及不合格時的 remapping 選項。 Algorithmization: write 4–6 lines of pseudo-code showing how your Tariff Calculator decides duty-free vs not, and what remapping options you consider when failing.

SUBMISSION HUB｜成果輸出與上傳 SUBMISSION HUB｜Export & Upload

你在本頁的輸入（決策樹、認知負荷自評、反思、貼上的 AI 結果、以及附件檔）可以被打包成檔案下載，並且一鍵上傳到課程的 Google Sheets（上傳網址不會顯示在頁面上）。 Your inputs on this page (decision tree, cognitive-load ratings, reflection, pasted AI outputs, and an optional evidence file) can be packaged for download and uploaded to the course Google Sheet (the upload endpoint is not displayed on the page).

A. 你的身分資訊 A. Your profile

提示：以上資訊會寫入你的下載檔與 Google Sheets 記錄，方便後續學習分析。 Tip: this metadata is embedded in your downloaded file and the Google Sheets log for learning analytics.

B. 附件（可選） B. Optional evidence file

你可以上傳一個文字檔（例如：對話紀錄 Log 3、分組筆記）。系統會把內容一併打包，並嘗試寫入 Google Drive。 Upload a text-based file (e.g., Chat Log 3, group notes). The content will be packaged and sent to Google Drive.

C. 匯出 / 上傳 C. Export / Upload

把你剛剛在 AI 端得到的輸出貼回來，讓檔案與 Sheets 能完整保存你的思考歷程。 Paste the AI outputs back here so your file and the Sheets log can preserve your reasoning trail.