資料包絡分析法(Data Envelopment Analysis, DEA)為一評估企業經營績效之重要方法之一。此方法概念源自1927年義大利經濟學家Pareto所提出之柏拉圖最適境界(Pareto Optimality),後續分別由Charnes, Cooper and Rhodes (1978)及Banker, Charnes, Cooper(1984)提出CCR模式與BCC模式。該等模型之限制為所有變數均為非零非負。DEA法的研究背景是以美國金融市場為主,但是否適用於我國目前金融市場之規模與型態,為一值得討論之議題。其中包含變數之選取等問題皆尚無定論。本研究以中華民國2006年至2009年四年間,我國十三家上市上櫃金融控股公司旗下子銀行為樣本,利用投入導向DEA法探討樣本銀行於研究期間績效表現之差異性,並推論衡量銀行產業績效之最佳投入產出變數組合。本研究實證發現:(1) 採資產總額、員工人數、利息費用、存款和薪資費用為投入變數,利息收入為產出變數來衡量樣本銀行績效表現,結果較符合現實情況。(2) 樣本銀行於研究期間內經營績效有逐年下降之趨勢,此結果主要導因於純技術效率的退步。
Data Envelopment Analysis is an important operating performance evaluation method by Charnes, Cooper and Rhoades (CCR) model of Charnes et al.(1978) and Banker, Charnes and Cooper (BCC) model of Banker et al. (1984). The limitation of these models is all parameters need to be positive. These researches were based on the information gathered from the American banking industry many years ago. Whether it is applicable in the present Taiwan banking industry is a crucial subject, which includes the choice of parameters. The present study is conducted to evaluate the operating efficiency of 13 subsidiary banks of Financial Holding Companies in Taiwan over a five-year period, i.e., 2006-2009, through output-oriented DEA approach. Besides, this study also explores the optimal portfolio to evaluate operating performance of banks. The result indicates that (1) it is more realistic to evaluate operating performance of banking by using total assets, number of employees, interest expenses, deposits and salaries as inputs and interest income as output. (2) the performance of subsidiary banks has been declining year by year, it resulted from the recession of pure technical efficiency.