Efficiency Predictions by Fuzzy Piecewise Auto-regression

Abstract

轉換為輸入與輸出關係來預測生產力與效率已廣範應用於許多生產與服務作業管理領域，因此，作業研究與計量經濟等領域普遍認為效率分析與預測為其主要的研究議題。在過去幾年，當資料包絡分析已經成功形塑資訊經濟議題時，效率預測卻顯少有學者進行研究。效率預測在當今許多規畫應用中扮演著關鍵角色。一個好的預測方法可以有效輔助決策者做效率預測。因此，不同於過去傳統效率預測方法，本研究採用模糊分段自我迴歸方法與效率追趕指標去處理效率預測問題。藉由可能性與必然性回歸模式應用於台灣22家商業銀行效率預測，透過2002到2005年三期資料去驗證2005年到2006年效率值，並且以此為基礎去預測2007年各銀行的效率。基於過去各商業銀行的績效表現，透過這個分析架構，我們可以提供給各銀行未來效率值可能區間。經由指示性研究發現與比較，透過模糊分段自我迴歸與效率追趕指標處理，可以有效預測相對效率議題。 Predicting productivity and efficiency during the transfer of input to output is a key issue in many manufacturing and service operation applications. Operation research and econometrics acknowledge that efficiency analysis is a major research issue. Data envelopment analysis (DEA) has substantially reshaped the result of information economics in previous years. However, the extent to which asymmetric information is relevant for efficiency prediction has rarely been sought empirically. Efficiency prediction plays a crucial role in many state-of-the-art applications and planning. Forecasting methodologies that can accurately predict efficiency scores can help in strategic decision-making. This study focuses on fuzzy piecewise auto-regression and the catching-up efficiency index (CIE), which supports efficiency prediction. In this study, two regression models were formulated by utilizing data from commercial banks in Taiwan from 2002 to 2005. These models were used to validate banking efficiency scores for 2005 and 2006, and to predict the banking efficiency scores for 2007. The results of a thorough computational analysis provide a range indicating the prediction value for each bank based on fuzzy regression characteristics.