近年來有關台灣銀行業生產力的實證文獻，皆利用Färe, Grosskopf, Norries, and Zhang（1994）以產出或投入距離函數為基礎的生產面Malmquist 生產力指數，將生產力指數分解成與生產結構有關的技術效率變動指數與生產技術變動指數，冀掌握銀行產業技術的動態特質。但隨著時間經過，生產技術可能呈現非中性技術變動，使得各期生產邊界存在相交現象，導致以Färe et al.作法，來判斷受評估銀行生產力進步或退步，可能產生誤判。同時，一些有關生產力變動來源與型態等技術相關訊息也因而被忽略。本文承襲Färe, Grifell-Tatjé, Grosskopf, and Lovell（1997）與Nin,Arndt, and Preckel（2003）作法，討論台灣銀行業生產力、產出偏誤技術變動與投入偏誤技術變動。主要實證結果顯示，近十餘年間台灣銀行業生產邊界呈現非中性技術變動，產出偏誤、投入偏誤兩項偏誤技術變動，全期間以1.44%速度帶動生產力改善；惟就生產面而言，銀行業投入與產出組合調整方向並不利於生產力之提升。
未來，銀行業提高資本使用、資金使用，取代勞動投入，而提高淨投資/放款淨額比，將有助於進一步改善生產力。

There are recently substantial papers that use the DEA-Malmquist productivity index developed by Färe, Grosskopf, Norries, and Zhang (1994) and identify the sources of productivity change in the Taiwanese banking industry. However, DEA-Malmquist productivity index may give misleading information when technical change is not neutral.
This paper, following Färe, Grifell-Tatjé, Grosskopf, and Lovell (1997) and Nin, Arndt,& Preckel (2003), tends to decompose further the technical change component of DEA-Malmquist productivity index, which we express as the product of a magnitude index, the output-biased index, and the input-biased index. The approach is illustrated with a set of 36 banks in Taiwan from1995 to 2005. Empirical results show that Taiwanese banking industry experienced a non-neutral technical change. Combined
output-biased and input-biased technical change index value, these two biased technical change components make together positive contribution to productivity change at a rate of 1.44 per year during periods from 1995 to 2005. During the same time periods, the direction of adjustment of production point between two periods is less favorable situation for whole DMUs. For the purpose of productivity improvement, Taiwanese banking industry may increase further the investment/loan ratio associated with more use of physical capital and funds.