研究：碳會計方法差異可讓企業供應鏈排放呈現逾一倍落差

《Nature Sustainability》刊登的研究以 401 家歐洲企業與一組固定採購籃為例，指出供應鏈排放結果會因資料庫與計算方法而大幅改變。研究示例中，相同活動的估算從 170 萬噸到 350 萬噸二氧化碳當量不等；這不是實際營運改變，而是會計選擇造成的差異。

A study published in Nature Sustainability uses data from 401 European companies and a fixed purchasing basket to show how strongly supply-chain emissions can change with the database and accounting method selected. For the same activities, the study example ranged from 1.7 million to 3.5 million tonnes of CO2 equivalent; the difference came from accounting choices, not operational change.

企業淨零承諾通常高度依賴範疇三排放，但供應鏈活動資料必須透過排放係數、產業資料庫、邊界與基準年規則轉換為排放量。不同資料來源具有不同年份、地理範圍、產品分類與估算邏輯，因此同一批活動資料不必然產生可直接比較的結果。

Corporate net-zero claims often depend heavily on Scope 3 emissions, but supply-chain activity data must be converted through emission factors, industry databases, boundaries and base-year rules. Sources differ in year, geography, product classification and estimation logic, so identical activity data do not necessarily produce directly comparable results.

研究指出，401 家歐洲企業供應鏈排放的年度變化中，有 62.5% 無法由企業活動、產業變化或排放強度解釋；相較之下，直接排放的不可解釋變化略高於 5%。若公司只公布最終數字而未揭露資料庫、版本、邊界與重算規則，外部使用者難以判斷減量來自實際行動或方法替換。

The study reports that 62.5% of year-to-year variation in supply-chain emissions across 401 European companies could not be explained by company activity, industry change or emission intensity. The comparable unexplained variation for direct emissions was just over 5%. When a company discloses only a final number, users cannot tell whether a reduction came from real action or from changing a database, boundary or calculation rule.

企業應為每次盤查保留活動資料、排放係數來源、資料庫與版本、地理與時間適用範圍、組織及價值鏈邊界、基準年與重算政策。跨年度改用資料庫或方法時，應同步重算可比較基準並揭露影響；產業與標準制定者則可建立同業一致的最低計算規則及變更紀錄。

Each inventory should retain activity data, emission-factor sources, database and version, geographic and temporal applicability, organisational and value-chain boundaries, base year and recalculation policy. A change of database or method should trigger a comparable-baseline recalculation and an impact disclosure. Industry bodies and standard setters can also establish minimum common rules and controlled change logs.

研究的固定食品飲料採購籃在四個常用科學資料庫下得到 170 萬至 350 萬噸二氧化碳當量的結果，最大值超過最小值一倍。Eco-Business 報導連結至原始同儕審查文章；SNN 草稿同時保留 Eco-Business、Mongabay 與 Nature Sustainability 的可點擊來源，並將研究結果與編輯判讀分開。

The study’s fixed food-and-beverage purchasing basket produced estimates from 1.7 million to 3.5 million tonnes of CO2 equivalent across four commonly used scientific databases, making the highest result more than double the lowest. This draft preserves clickable links to Eco-Business, the underlying Mongabay report and the peer-reviewed Nature Sustainability paper, while separating reported findings from SNN analysis.

對台灣出口商與跨國供應鏈而言，客戶要求的碳數據若使用不同係數庫與邊界，可能造成供應商績效、採購比較與減碳認列偏差。企業除了交付排放數字，也需要交付足以重建計算的來源、版本、假設與變更歷史，才能支撐查證與決策。

For Taiwanese exporters and multinational supply chains, different factor libraries and boundaries can distort supplier performance, procurement comparisons and recognition of reductions. Delivering an emissions total is therefore insufficient; companies also need to deliver the sources, versions, assumptions and change history required to reconstruct the calculation.

SNN 編輯分析：碳盤查的可信度不能只依賴單一總量。可驗證的證據鏈應把活動資料、排放係數、方法版本、邊界、轉換步驟、重算決定與最終揭露綁定，讓第三方可以辨識數字變動究竟來自營運、資料修訂或方法選擇。

SNN editorial analysis: Carbon-inventory credibility cannot rest on a single aggregate. A verifiable evidence chain should bind activity data, emission factors, method versions, boundaries, transformations, recalculation decisions and the final disclosure so reviewers can distinguish operational change from data revision or methodological choice.

後續應追蹤研究所提產業一致規則是否被標準制定、保證與數位申報流程採納，也要觀察 GHG Protocol 等制度更新如何處理排放係數、方法變更、基準年重算與範疇三可比性。

Next steps include monitoring whether the proposed industry-consistent rules enter standards, assurance practice and digital reporting, and how updates by institutions such as GHG Protocol address factor selection, method changes, base-year recalculation and Scope 3 comparability.