多項績效指標之營運單位之兩階段評估程序

Abstract

在多指標的評比中，管理者經常面臨制定指標的權重的問題，為求計算各單位的綜合績效並呈現出組織內部績效差異跟排序。當評量各單位的績效時，由於各單位想最大自己的效率值而持有不同的意見。但是以此觀點去評估可能造成過於樂觀的情況。因此，管理者必須以最妥協的方法去評量各單位的績效，以減少過於樂觀的風險。我們建議一個方法由各單位自己為標準去評量以及綜合各個績效分數。在最妥協的情況下我們能知道各個單位間的差異以及排序。 本篇文章提出一個新的評估程序透過兩階段進行。第一階段我們利用一線性規劃模型，讓各單位以效率值為1為基準，求出所認為的最妥協權重來評量其他單位的差異。第二階段我們代入交叉效率的概念，以各單位所選擇的權重來評量自我的表現。利用交叉評比的方法求得的平均綜合分數來排序各單位。最後，以台灣某一縣消防隊其25個分隊的數據為例說明本研究所建議的評估程序。

Managers frequently want to assess units under their governance with multiple performance indices. Charnes, Cooper and Rhodes [CCR, 1978] introduced Data Envelopment Analysis to have the self-appraisal efficiency for each unit. Doyle and Green [1994] employed CCR and further obtained peer-appraisal cross efficiency to rank the units. Liu and Peng [2008] and Liu, Peng and Chang [2006] ranked the units with a common set of weights for the indices while the total virtual gap of the units is minimized. Here, we proposed a two-phase procedure to rank the units. In the first phase, we adopt the model of Liu, Peng and Chang [2006] to determine a set of weights as the object unit has efficiency equal to 1 while the total virtual gap of the units to the object unit is minimized. Each unit acts as the object unit alternatively. In the second phase, we proposed a procedure about the concept of cross efficiency. The average cross efficiency of every units are then used to rank units. Finally, we apply the proposed two-phase procedure to assess 25 fire departments of a county in Taiwan.