本文是以汽車自製率由以往的50%以上,掉至不及一半的2003年與2005年,就台灣上市、櫃、及公開發行的汽車零組件廠商,分別採資料包絡分析法(DEA)與隨機性統計邊界法(SFA),進行品質變數對效率影響之比較。實証結果發現,在下述情況下兩法的結論是具顯著相關:一、若樣本未分類,則兩個年度的總效率(TE)排序,均具顯著相關;二、樣本不論分成大、中小規模,或有、無與國外技術合作,總效率(TE)、規模效率(SE)在兩個年度間均未呈改善現象;三、在品質變數中之TS-16949認證,僅在中小規模廠商上獲得對降低無效率具顯著的效果。
This paper tries to compare the quality variables' effects to inefficiency evaluated by Data Envelopment Analysis and Stochastic Statistical Frontier Approach in Taiwan's listed automobile parts companies during year 2003 and 2005. We chose the two years because the local content of automobile hadn't achieved a half in the period. The results found: (1) the rankings in total efficiency from the both methods all had the significant correlation in two years. (2) the both total efficiency and scale efficiency didn't show to improve in the period whether DEA or SFA. (3) the small and medium-sized companies with TS-16949 certification all had the significant relations to reduce the inefficiencies in both methods.