本文以資料包絡分析法 (Data Envelopment Analysis,簡稱DEA),探討台灣中小企業銀行轄下分行之經營效率,因一家銀行經營績效的表現,是該銀行各分行經營績效的總合,故只要能提升分行效率,總行表現自然隨之提升。在要素的選取準則方面,投入要素包括員工人數、利息支出、業務費用等三項,而產出要素則有利息收入及稅前淨利等二項。研究結果發現:整體有效率之分行占17.65%,因已達相對效率,故無須做進一步的調整,以現有營運方式及規模繼續營運即可。分行經營無效率來自於純技術無效率僅占2.52%,欲改善無效率情況,應從管理面下手,調整資源配置方式,以最適當的投入產出組合來創造最大的產出。分行經營無效率來自於規模無效率占12.60%,其中處於規模報酬遞增階段者,應增加規模來改善效率,其餘處於規模報酬遞減階段者,故應縮減生產規模來改善效率。分行經營無效率來自於純技術無效率和規模無效率比率高達67.23%,若欲改善無效率情況,建議經營管理者可參考本研究建議之標竿企業作為學習目標,依差額變數分析表對分行投入量縮減的建議,進行降低投入規劃,以提升技術效率。
The Data Envelopment Analysis, DEA was adopted in this study to explore the operational efficiency of the Taiwan Business Bank branches. As the operational performance of a bank is the total operational performance of the bank branches, as long as the efficiency of the branches can be enhanced, the head office’s performance will also be improved. In terms of factor selection criteria, the three input factors include: the number of employees, interest expenditure, and operating expenses. On the other hand, there are two output factors, including interest income and net income before tax. The results show that the branches with overall efficiency accounted for 17.65%. As the relative efficiency was reached, there was no need to make further adjustments, but to continue with the existing operating mode and scale. The operational inefficiency of the branches that came from pure technical inefficiency accounted for only 2.52%. In order to improve the inefficiency condition, one should start from the management aspect to adjust the resource allocation method and maximize output through the most appropriate input and output portfolios. The branches’ operational inefficiency that came from scale inefficiency accounted for 12.6%. In particular, those in the increasing returns to scale stage should enhance the scale to improve efficiency, while those in the decreasing returns to scale stage should therefore reduce the production scale to improve efficiency. The ratio of the branches’ operational efficiency that came from pure technical inefficiency and scale inefficiency reached as high as 67.23%. In order to improve the inefficiency condition, it is suggested that managers refer to the recommended benchmark companies in this study as the learning objects. Further, based on the slack variable analysis table, it is suggested that the branches’ input volumes be reduced to plan lowered inputs, thereby enhancing the technical efficiency.