本研究的目的在於探討區位因素是否影響廠商的生產技術效率，利用Aigner, Lovell and Schmidt(1977)所提出的隨機性生產邊界法，並進一步利用Reifschneider and Stevenson(1991)與Huang and Liu(1994)把可能造成技術無效率的解釋變數與隨機項加入隨機性生產邊界模型的一階段分析法，探討區位因素對生產技術效率的影響。 本文以民國九十年台灣電子業廠商為研究對象，利用FRONTIER 4.1電腦軟體進行運算。生產函數設定為Translog型式，以資本、勞動與原物料之三種要素投入及營業收入總額為產出估計生產技術效率，並同時進一步探討影響廠商生產技術效率之特性變數，包括廠商營運年數、年數平方項及廠商區位選擇之虛擬變數。本研究依特性變數設定五種不同研究模型，實證結果如下： 一、不同模型所估計之平均生產技術效率值介於0.904至0.927之 間，顯示我國電子產業生產技術效率具一定水準。 二、電子產業廠商設廠年數與生產技術效率之間存在非線性關係，即先遞減而後遞增之U字型關係，可說明電子業廠商是在累積相當經驗後，生產技術效率才會產生遞增效果。 三、區位選擇在工業園區之廠啇，其生產技術效率高於區位選擇不在工業園區之廠商。再進一步探討區位選擇在不同性質工業園區之技術效率，結果以區位選擇在科學工業園區之廠商生產技術效率最高，其次為區位選擇在加工出口區之廠商，第三為區位選擇在工業區之廠商，最差者為區位選擇不在任一種性質工業區之廠商。 四、透過概似比統計量檢定選擇最適模型，結果以設定廠商營運年數及廠商區位選擇之特性變數置入實證模型中為最適模型。 The purpose of this study is to examine how the location factor affects the production efficiency for Taiwan electronic firms. Based on the theory proposed by the Aigner, Lovell and Schmidt(1977), this research adopts the empirical method proposed by Reifschneider and Stevenson(1991) and Huang and Liu(1994) to estimate parameters of the stochastic frontier and the inefficiency model simultaneously. By using the computer program, FRONTIER 4.1., this research uses the electronic firms data collected from the Taiwan-Fukien District Census Survey in 2001. Our empirical models are based on the Translog production function which assumes capital, wage, and intermediate input as primary input variables and the sales as output. In addition, we assume that the technical inefficiency effects are affected by years of establishment, the squared term of years of establishment, and location dummy variables. The empirical results indicate that: (1) The estimated average efficiency values range from 0.904 to 0.927. It means that the production efficiency of Taiwan electronic firms shows a good path. (2) The U-shape of the quadratic relationship between the years of establishment and the inefficiency effects indicate that the accumulation of experience would improve the production efficiency for electronic firms. (3) Electronic firms located in industry park zone would produce higher production efficiency than those not located in. Furthermore, firms located in industry science park have the highest production efficiency. (4) Using the likelihood ratio test, we found that an appropriate empirical inefficiency model should include the variables of years of establishment and the location dummies.
