永續投資提案如何承受不利情境：從假設到可重算的決策底稿

Trellis專家作者Daniel Aronson在11月5日提出一個永續投資提案的實務問題：如果價格、採用率或執行時程不如預期，原本的財務理由還成立嗎？本篇以其新方法文章為起點，對照英國財政部正式投資評估指引與獨立學術研究，說明壓力測試如何讓決策假設更可核對，而不是保證投資報酬。

In a November 5 expert contribution to Trellis, Daniel Aronson asks whether a sustainability investment case still works when prices, adoption or delivery timing disappoint. This educational report compares that fresh methodology article with formal UK Treasury guidance and independent academic research. Stress testing can make assumptions more inspectable; it does not guarantee an investment return.

節能、廢料回收與供應鏈轉型提案常同時牽涉採購、廠務及財務部門。原物料售價、回收率、設備稼動與導入成本都可能改變結果。專家作者的顧問案例提供方法示意，但未公開的客戶數據與成果不應被延伸成一般企業可複製的績效承諾。

Energy efficiency, materials recovery and supply-chain transition proposals often connect procurement, plant operations and finance. Commodity prices, recovery yields, utilisation and implementation costs can change the result. The contributor’s consulting example illustrates a method, but undisclosed client data and outcomes are not adopted as a generally reproducible performance promise.

Wharton的研究解讀指出，對機率不明的情況，人們可能具有模糊厭惡。研究對象是荷蘭投資人及非投資人，並非企業投資委員會；將此結果用來理解提案溝通，屬於有界的方法推論，不能宣稱它已證明永續專案的報酬或董事會偏好。

Wharton’s research reporting describes ambiguity aversion when probabilities are unclear. Its empirical setting involved investors and non-investors in the Netherlands, not corporate investment committees. Applying that finding to proposal communication is a bounded methodological interpretation; it does not establish a sustainability project’s return or predict board preferences.

英國財政部2026年更新的正式指引依Five Case Model組織評估，並要求處理不確定性、樂觀偏誤及敏感度分析。對企業而言，可先建立基準情境，逐一改變重要假設，再評估多個不利條件同時發生時的結果。這是可借鏡的評估方法，不是直接移植英國公共支出規範。

HM Treasury’s updated 2026 guidance organises appraisal through the Five Case Model and addresses uncertainty, optimism bias and sensitivity analysis. A corporate team can establish a base case, change important assumptions individually, and then combine several adverse conditions. This is a method to consider rather than a direct transplantation of UK public-spending requirements.

方法的產物應是可重算的結果分布與決策條件，而不是只有一個漂亮的報酬率。比較時需保存價格來源、回收率試算、資本支出及人力假設，並記錄哪些變數足以使方案不再成立。模擬次數多不等於資料品質高，機率分布若缺乏根據，結果仍有侷限。

A useful output is a recalculable range of results and decision conditions rather than one attractive return figure. Price sources, recovery-yield trials, capital expenditure and staffing assumptions should remain traceable. The analysis should identify variables that make a proposal cease to work. More simulations cannot compensate for unsupported probability distributions or poor inputs.

永續效益與財務效益可共同評估，但兩者的衡量單位及證據來源應清楚區分。壓力測試可以揭示資本需求、現金流及執行風險的關聯；它不取代正式審計、投資決策或管理階層對資料完整性的責任，也不使未經驗證的環境成果取得保證。

Environmental and financial benefits can be considered together while retaining distinct measurement units and evidence sources. Stress testing can show connections among capital needs, cash flow and execution risk. It does not replace formal audit, investment decisions or management responsibility for data, and cannot provide assurance of unverified environmental outcomes.

以下為SNN編輯分析與建議，並非來源確認的台灣企業事件。外部客戶的減碳投資要求與跨境原物料價格變動，可能透過報價、設備採購及回收收入假設，傳導至台灣製造業廠務評估團隊及台灣供應鏈採購財務團隊，影響設備投資核准及廢料回收方案評估；本篇方法不代表台灣已新增投資法定義務。揭露前證據基礎設施應連結基準現金流模型、金屬價格來源紀錄、回收率試驗紀錄、設備報價版本、導入時程假設、不利情境參數、敏感度分析表及投資核准條件。第一，台灣製造業廠務評估團隊應在設備採購送審前，將設備報價版本與導入時程假設對照基準現金流模型，保留修訂原因。第二，台灣供應鏈採購財務團隊應在客戶減碳要求或金屬價格變動時，依金屬價格來源紀錄與回收率試驗紀錄重算不利情境參數，記錄方案失效的條件。第三，資本預算審查委員會應在資金核准前，比較敏感度分析表與投資核准條件，要求負責人及後續驗證日期。這些建議只適用於確有相關交易與資料的企業，不預設所有台灣專案具有同樣報酬或風險。

SNN editorial analysis and recommendations: external customer decarbonisation investment requests and cross-border commodity-price changes may flow through quotations, equipment procurement and recovery-revenue assumptions to Taiwan manufacturing plant-appraisal teams and Taiwan supply-chain procurement-finance teams, affecting equipment-investment approval and waste-recovery proposal evaluation. This method does not establish new Taiwan statutory investment duties. Pre-Disclosure Evidence Infrastructure should connect base-case cash-flow models, metal-price source records, recovery-yield trial records, equipment-quotation versions, implementation-timing assumptions, adverse-scenario parameters, sensitivity-analysis tables and investment-approval conditions. First, Taiwan manufacturing plant-appraisal teams should compare equipment-quotation versions and implementation-timing assumptions with base-case cash-flow models before submitting equipment procurement for review, retaining reasons for revisions. Second, Taiwan supply-chain procurement-finance teams must recalculate adverse-scenario parameters using metal-price source records and recovery-yield trial records when customer decarbonisation requirements or metal prices change, recording proposal failure conditions. Third, capital-budget review committees should compare sensitivity-analysis tables with investment-approval conditions before authorising funding, assigning owners and subsequent verification dates. These recommendations depend on actual transactions and available data; they do not assume identical returns or risks for all Taiwan projects.

下一步不是增加更多無依據的模擬，而是將後續實際價格、回收率與導入進度回填，檢驗最初假設。若結果偏離預估，應更新模型及決策條件；良好的版本紀錄讓修正可追溯，但仍不能保證專案一定成功。

The next step is to compare actual prices, recovery yields and delivery progress with the original assumptions, rather than simply increasing the number of unsupported simulations. Divergence should trigger model and decision-condition updates. Version records make those revisions traceable without guaranteeing project success.