The Translog Production Function and Variable Returns to Scale
提出一个允许可变规模报酬的超越对数生产函数估计框架，基于逆要素需求函数和非齐次生产技术，并用美国制造业数据证明齐次性和不变规模报酬不成立。
This paper examines existing methods of estimating the translog production function and provides a general framework that allows for variable returns to scale. The model is based on the inverse input demand function and embeds a nonhomothetic production technology. Previous estimation methods are valid only for homogeneous technologies with fixed scale effects. Estimation results for U.S. manufacturing show that neither homotheticity and homogeneity nor constant returns to scale is a proper characterization of the underlying structure of production, thereby vindicating the empirical relevance of the inverse demand framework that entails a nonhomothetic technology. Copyright 1992 by MIT Press.