教師資料查詢 | 類別: 研究報告 | 教師: 黃河泉 HO-CHUAN HUANG (瀏覽個人網頁)

標題:平滑係數之隨機邊際模型的半母數貝氏分析
學年92
學期1
出版(發表)日期2004/01/01
作品名稱平滑係數之隨機邊際模型的半母數貝氏分析
作品名稱(其他語言)Semiparametric Bayesian Inference of the Smooth-Coefficient Stochastic Frontier Models
著者黃河泉
單位淡江大學財務金融學系
描述計畫編號NSC93-2415-H032-008;研究期間200408~200507;研究經費741,000
委託單位行政院國家科學委員會
摘要傳統隨機邊界模型通常或多或少假設所分析的廠商都採用相同的技術生產而不同的地方在於生產效率的部分。然而,實際上,廠商可能有不同的理由而採用不同的技術。因此,以往「相同技術」的假設似乎不切實際,而且可能導致效率衡量的錯誤。不同於以往,本篇文章提出一新的平滑係數之隨機邊際模型來衡量廠商無效率之程度,但同時又允許廠商使用不同之技術。我們的平滑係數之隨機邊際模型其實是Li, Huang, Li and Fu (2002)的平滑係數模型與Aigner, Lovell and Schmidt (1977)與 Meeusen and van den Broeck (1977)的隨機邊界模型之結合體。所有此模型之估計與推論都是仰賴貝氏模擬方法,特別是「吉卜斯-資料擴充」,來執行的。我們將運用我們的模型於一組實際資料來說明其實用性,並且打算與傳統隨機邊界模型比較廠商無效率之衡量有何差異。 Conventional stochastic frontier specifications often assume, implicitly or explicitly, that all firms under consideration share exactly the same technology and differs only with respect to their degree of inefficiencies. However, in practice, firms may adopt different technologies for a variety of reasons. As a result, this common-technology assumption appears to be inappropriate and may result in misleading, and even incorrect, measurement of inefficiencies. In contrast, this paper proposes a novel semiparametric smooth-coefficient stochastic frontier (SPSC-SF) model to measure firms' inefficiencies while allowing for different technologies adopted by individual firm. Our SPSC-SF model is a synthesis of the semiparametric smooth-coefficient model proposed by Li, Huang, Li and Fu (2002) and the stochastic frontier model pioneered by Aigner, Lovell and Schmidt (1977) and Meeusen and van den Broeck (1977). Estimation and inference are made possible by the Bayesian simulation algorithm, e.g., the Gibbs sampling with data augmentation. An real example will be used to illustrated the practical use of our model. In addition, we will compare the inefficiency measurements obtained by our SPSC-SF approach and the conventional SF model
關鍵字貝氏;半母數;平滑係數;隨機邊界;Bayesian;semiparametric;smooth-coefficient;stochastic frontier
語言
相關連結
Google+ 推薦功能,讓全世界都能看到您的推薦!