市內網路服務經營效率之評估與預測－DEA與GM模型之整合應用

本文旨在整合資料包絡分析法(DEA)與灰色模型(GM)，據以評估中華電信C地區9個營運處市內網路服務，在1998年至2002年的經營效率，並預測2003年之營運績效。本文首先以GM (1,N)篩選投入變數；其次，利用DEA CCR及BCC模式，評估經營效率、投入產出調整途徑，以GM (1,1)殘差修正模型，預測2003年投入與產出數據，並重新帶入DEA模式，預測2003年可能之營運情況。研究發現：(1)透過GM (1,N)模型，發現六種投入項與產出項（營收額）之灰關聯係數，依序為交換機建設數(4.806)、員工數(3.139)、成端電纜數(2.968)、配線電纜數(1.295)、營業區域面積(4.806)、員工數(3.139)、成端電纜數(2.968)、配線電纜數(1.295)、營業區域面積(0.976)、營業中心數(0.374)，故本文選擇前四項作為投入項。(2)1998年至2002年之平均技術效率為0.864，顯示存在約14%的效率改善空間，其中源於純技術無效率及規模無效率各約7%。1998年至2002年平均不足產出為315百萬元，投入過多分別為員工數147人、交換機門號數為60,983門、成端電纜125,378對、配分電纜119,044對。(3)整合GM與DEA模型，預測2003年之平均技術效率為0.855、純技術效率為0.931及規模效率為0.918；預測2003年平均不足產出為344百萬元，投入過多分別為員工數75人、交換機門號數為95,897門、成端電纜114,881對、配線電纜136,710對。

關鍵字

資料包絡分析法；灰色模型；市內網路服務；經營效率

並列摘要

This study integrates the Data Envelopment Analysis (DEA) and the Grey Model (GM) to evaluate the operating efficiencies of the nine branches in District C of the Chunghwa Telecom Co. Ltd. from 1998 to 2002, and to forecast these branches' operating efficiencies in the year of 2003. We first use GM (1,N) model to screen the six input variables. We then utilize the CCR and BCC models of the DEA to estimate operating efficiencies and calculate the input and output radical movement and slack moviment, we apply the GM (1,1) residual-correction model to predict the input and output variables in 2003, and further use these numbers to evaluate the operating efficiencies through DEA analysis. We find that: (1) Based on the grey relationship coefficients from the GM (1,N) model, we choose the following four DEA input variables from the six initial input variables: the available phone numbers provided by local switches; the number of employees; the number of pairs of backbone cables; and the number of pairs of distribution cables. (2) The average technical efficiency from 1998 to 2002 is 0.864. That is, there exists about 14% inefficiency with 7% from pure technical inefficiency and the other 7% from scale inefficiency. (3) The predicted results from the integrated GM and DEA models show that, in 2003, the average technical efficiency is 0.855, and the pure technical efficiency and scale efficiency are 0.931 and 0.918, respectively. The predicted deficient amounts of operating incomes (the output variable) in 2003 is NT$344 million, and the predicted amounts of excess inputs in the above four input variables are 75 employees, 95,897 phone numbers, 114,881 pairs of backbone cables, and 136,710 pairs of distribution cables, respectively.