本文採用新的隨機共同邊界方法，將其擴充至Fourier富伸縮成本函數，針對西歐地區十個國家的銀行業進行成本效率之分析，資料期間涵蓋1996年至2010年。不同於Battese et al. (2004), O’Donnell et al. (2008), and Huang et al. (2011a) 等人利用線性規劃法，本文應用隨機共同邊界法來估計技術缺口比率，進而做跨國間的效率比較，此法的特點在於技術缺口比率可以設為一些反映國家環境差異的外生變數之函數，而線性規劃法則無法做此設定。實證結果顯示採用線性規劃方法所估計出的技術缺口比率與共同成本效率會有低估的現象，技術缺口比率以及共同成本效率在1996年至2000年間逐步上升，此結果支持金融市場的整合可以增進效率，然而，到2000年之後則反轉向下，特別是在2007年至2010年次級房貸風暴時期明顯惡化。此外，進一步的分群進行分析的結果顯示，小規模、高獲利、或是較保守的銀行相對來說較具有效率。 / This paper aims to gain further insights into cost efficiency using the newly developed metafrontier approach under the framework of the Fourier flexible cost frontier for banking industries across 10 Western European nations during the period 1996-2010. Unlike Battese et al. (2004), O’Donnell et al. (2008), and Huang et al. (2011a), who suggest using programming techniques, the stochastic metafrontier is formulated and applied to obtain the technology gap ratio (TGR) for efficiency comparisons among countries. One salient feature of our method is that the TGR can be specified as a function of some exogenous variables that reflect group-specific environmental differences, while the mathematical programming is not allowed to do so. Empirical results show that both TGR and metafrontier cost efficiency (MCE) are underestimated by programming techniques. The TGR and MCE exhibit a gradual upward trend during 1996-2000 and then followed by a downward trend, especially after the subprime crisis of 2007-2010. This suggests that a more integrated financial market is able to improve banking efficiency. Smaller banks tend to be more cost efficient than larger ones. Higher profitable banks and more conservative banks are related to greater efficiency.

金控銀行和獨立銀行在傳統在做績效評估時，可能都只考慮相同的技術水準，如此可能無法正確來衡量不同群體的的差異，而忽略其潛在的效率改善指標。因此，本文將先採用隨機邊界模型(SFA)，估出兩體系的隨機邊界模型，接著，再運用Rao (2006)所提出的共同邊界模型(metafrontier)，來進行兩個體系的銀行效率評估。 本研究運用了民國91年到民國97年期間，13家金控銀行與24家獨立銀行的資料為樣本，去分析此兩個群體的效率比較，可得以下結論： 1.在金控銀行與獨立銀行的個別隨機邊界中，兩個群體在於技術效率的表現上差異不大。 2.金控銀行的技術效率變動有越來越小的趨勢；而獨立銀行術效率沒有明顯的縮小的趨勢。 3.獨立銀行的技術缺口比TGR有顯著的大於金控銀行的TGR。 4.在共同邊界的技術效率中，獨立銀行的技術效率顯著的大於金控銀行 / Most of traditional banking performance evaluation analyses assume both financial holding banks and independent banks share the same level of technology, thus it may not able to identify the managerial efficiency difference of different groups correctly . In this research, a SFA model (Battese and Coelli, 1995) is used first to estimate the technologies and inefficiency factors for both systems. Then a deterministic linear programming metafrontier (Rao, 2006) is adopted to evaluate the technology gap ratio between two bank systems. We collect data of 13 financial holding banks and 24 independentbanks from 2001~2008. After analyzing two systems and making comparison, the main conclusions are as follows: 1.Two bank systems have a minor difference in SFA. 2.The variance on technical efficiency becomes smaller with time in financial holding banks, but not in independent banks. 3.Independentbanks’ TGR is significantly larger than financial holding banks’. 4.In metafrontier, technical efficiency of independent banks is significantly larger than financial holding Bbnks.