近年來台灣景氣持續低迷，台灣綜合商品零售業的成長也受到不小的衝擊，在市場漸趨成熟與飽和的情況下，產業內競爭也漸趨激烈。如何在外在經濟環境欠佳，內在市場飽和下，有效提升經營效率，加強產業競爭能力，是目前台灣綜合商品零售業者最重要的課題。 　　本研究以資料包絡分析法（Data Envelopment Analysis）為基礎，並結合Metafrontier分析法，針對台灣地區2000年到2006年25家綜合商品零售業者，以客觀的投入產出資料進行廠商的經營效率分析。本研究結果顯示： 　　1.整體而言，連鎖式便利商店的經營績效平均處於相對有效率狀態，購物中心的經營績效平均處於相對無效率狀態。 　　2.廠商技術無效率的原因主要為純粹技術效率的問題。廠商的規模報酬狀態與所屬業態關聯不大，而與廠商是否屬於連鎖集團較有關聯。 　　3.比較metafrontier模型與group frontiers模型所評估的效率值，顯示各業態廠商的平均技術水準有顯著差異，連鎖式便利商店的效率水準較接近metafrontier的效率邊界。 　　4.依據Norman and Stoker（1991）的效率值強度分類方法，發現在兩模型之下，各業態廠商屬於「邊緣無效率單位」與「確實無效率單位」的比例都較大，表示經營績效有很大的改善空間。 　　5.差額分析與敏感度分析方面，百貨公司業與超級市場業在兩模型之下的結論不一致，推論應是兩者評估的廠商樣本，其衡量的技術水準不同。各廠商依循group frontiers模型所得結論改善經營績效，將可以在業態內成為相對有效率的廠商，但如果想要進一步提高效率，則必須改變廠商本身所處的業態環境所造成的技術水準限制，便可依循metafrontier模型所得結論改善經營績效，所以選用何種結論，仍須由業者依據廠商未來發展自行判斷。 / Since the recession in Taiwan economy has arrived in recent years, it has a great effect on the growth of the Taiwan retail sale of general merchandise. Because the market has been saturated, the competition has become fierce in this industry. How to improve the management efficiency and the industrial competitiveness in the present economic environment is the most important issue of the Taiwan retail sale of general merchandise. 　　This study uses inputs and outputs data of 25 DMUs of the Taiwan retail sale of general merchandise from 2000 to 2006, applying Data Envelopment Analysis with Metafrontier model to evaluate the operational efficiency of those firms. The findings of this study are as below: 　　1. On average, the chain convenience store industry shows more technically efficient than the shopping mall industry. 　　2. The main source for the firms being inefficient is pure technical inefficiency. The returns of scale of the firms are not much related to the types of business operation, but to the firms being the chain store ones or not. 　　3. Comparing meta-frontier model with group frontiers model, we find that there are significant differences in the average technical efficiency levels among the various types of business operation, and the chain convenience store industry is most close to meta-frontier. 　　4. According to the classification of the efficiency by Norman and Stoker (1991), we find a major part of the firms is marginally inefficient and distinctly inefficient. It implies there is a great room for the firms to improve their managerial performance. 　　5. The slack analysis indicates there are differences in the resource wastes between the department store industry and the supermarket industry. The differences mainly come from the technological differences between the two industries. The information generated from the group frontier model is helpful to improve the firms’ efficiency within the group. However, if the firms want to further improve their efficiency, they should rely on the information from the meta-frontier model, which sheds light on the technology gap among sub-industries.