以資料包絡分析法評估供應鏈網路之效率

Abstract

由於近年來受到全球化以及快速變遷的市場環境的影響，企業的生存除了仰賴有效的運用資源之外，更必須成功的與供應鏈上下游夥伴配合以達到整體的最大利益。為了正確的衡量一個的運作績效，績效評估系統必須要能夠接受來自不同領域的影響因子來提供決策者更完整的資訊。然而，傳統的績效評估方法往往局限於少數固定的影響因子，或是只針對供應鏈上的單一組織做績效評估，因此無法針對不同的影響因子提供完整的分析結果，或是忽略供應鏈上角色之間的互動關係。 資料包絡分析法在最近已成為一個普遍的績效評估方法，並且在過去的研究中已經被成功的套用在不同領域的生產系統上計算其效率。而網路資料分析法改進了傳統的資料分析法，藉由增加的限制式來維持系統內部結構與內部流程之間的互動及影響。使用資料分析法最大的好處之一，在於可以同時接受許多量化的因子，讓使用兩者之間沒有明確關聯的因子計算效率的方法得以實現。 本研究提出一個修改的網路資料包絡分析法模型。此修改模型將可以(1)以供應鏈整體的觀點來評估效率；(2)同時處理供應鏈網路中買方賣方的結構以及不同時間點或不同期數的影響；(3)能夠處理惡性產出-惡性產出會降低生產效率，且無法直接以傳統的資料包絡分析法模型處理。修改過的模型將可以接受更多種不同的影響因子作為輸入及輸出，以提供決策者在不同狀況時的需求。本研究最後將以情境分析及實例測試來驗證此修改模型可以有效地分辨出效率差異。

Recently, due to globalization and the highly competitive and fast-changing environment, the survivability of an organization relies on not only the ability to use its resources effectively and efficiently, but the ability to cooperate with the its supply chain partners to optimize the overall profits. In order to assess the performance of an organization, the performance measurement system must be generic enough to provide an integrated view on important factors from different aspects. However, traditional means of performance measurement are either based on a small set of predefined factors, which prevent a complete view of the overall system, or limited to a single organization, which ignore the interaction between supply chain players. Data envelopment analysis (DEA) has become a popular method to evaluate the performance in terms of efficiency, and has been applied to production systems in various fields. Network DEA improves the traditional DEA models by adding constraints to maintain the internal structure and operations of the subject being evaluated. One of the main benefits of using DEA is the model accepts multiple positive quantitative data as inputs and outputs of the system, making evaluating performance using factors with unknown relationship possible. In this study, a modified network DEA model is proposed, which has the ability to (1) evaluate the efficiency of a supply chain as a whole; (2) consider the buyer-seller structure as well as multiple time periods when evaluating efficiency; (3) handle “undesirable outputs,” which cannot be used in traditional DEA models directly. The modified network DEA model may accept any quantitative factors as inputs and outputs depending on the decision makers’ needs. To prove the effectiveness of the model, a scenarios test and a real-world data test are conducted.