過去使用隨機邊界模型探討生產效率的論文，大多依循Aigner et al. (1977) 與 Meeusen and Van Den Broeck (1977)，假設組合誤差項彼此間統計獨立，似乎過於強烈且缺乏學理支持。Bandyopadhyay et al. (2006) 放寬組合誤差項間相互獨立的限制，率先假設無效率項與隨機干擾項服從二變量聯合半常態分配，重新推導出橫斷面資料的組合誤差項機率密度函數。與傳統不相關模型相較，須多估計一個相關係數。因為縱橫資料漸趨普及，本研究打算將Bandyopadhyay et al. (2006) 生產函數模型擴展至縱橫資料成本函數，以便充分萃取資料中有用的訊息。依循Greene (2005) 的固定效果隨機邊界模型，在有相關性組合誤差與縱橫資料架構之下，容許固定效果與無效率項同時並存，然後發展可避免incidental parameters problem 的計量模型。1 鑒於東歐諸國在1991 年蘇聯解體前，實施社會主義制度，各行業包括銀行業皆受到嚴格管制，生產要素價格易遭到扭曲，可以利用影子價格(shadow price) 技巧處理，進而分析生產要素配置無效率議題。Kumbhakar andWang (2006) 證明成本函數模型若不同時考慮配置效率，導致有偏誤的係數估計值，影響後續技術效率和規模經濟等估計結果的正確性。如將前段提到固定效果隨機邊界成本函數模型應用於東歐諸國銀行業，同時探討技術與配置效率，兩者合稱經濟效率或X 效率，極具重要政策與經濟意義。 In a standard stochastic frontier model, dated back to Aigner et al. (1977) and Meeusen and Van Den Broeck (1977), assumes that the composed error terms are mutually independent. This arises from the fact that the statistical noise is out of control by a firm, which in essence must be independent of the controllable inefficiency term. Pal and Sengupta (1999) and Smith (2004) have recently considered the possibility of correlated noise-inefficiency in the context of stochastic frontier models. They claim that the controllable errors may be indirectly affected by random disturbances. The model of Schmidt and Sickles (1984) fixed effects formulation is known as being suffered from the problem of identification, because the fixed effects play two roles. They represent the heterogeneity and inefficiency. Greene (2005) proposes a novel model that is able to distinguish both heterogeneity and inefficiency, while maintains the assumption of independent composed errors. This research project attempts to derive the likelihood function of the correlated error components for a stochastic cost frontier model, under the framework of panel data, which deals with both heterogeneity and inefficiency. The issue of incidental parameters will be discussed and formally treated. In addition, it will derive the analytical expressions for the firm level technical inefficiency score. The new model will be applied to examine the technical and allocative efficiency of the transition countries, i.e., East European countries, since the input prices there may be seriously distorted by the government’s regulation. Hence, the input prices tend to be adjusted slowly in response to the market conditions. A shadow prices technique is introduced to represent the degree of allocative inefficiency.