中文摘要：

本文首次对我国1980—2011年间36个工业大类行业CES生产函数中资本、能源和劳动力之间三种嵌套形式的参数做了非线性计量估计,并以资本/劳动力比率为门限变量,对我国CES生产函数的门限效应做了考察。本文研究表明：1对于三种CES嵌套形式,多数行业能拒绝替代弹性等于1的原假设,这表明目前很多研究在构建模型时选择使用Cobb-Douglas函数形式的做法有待商榷;2对于三种CES嵌套形式,仅有极少数行业不能拒绝规模报酬等于1的原假设,这表明在构建模型时假定规模报酬不变有较大风险;3总体上,多数行业能拒绝把三种要素放在一个核里的原假设,因此对CES函数进行适当的嵌套十分必要;4综合来看,资本与能源先聚合再与劳动力聚合的嵌套形式（KE）L比较符合我国工业实际情况;5多数行业门限效应显著,且门限值大多出现在20世纪90年代之后,表明不少行业随资本/劳动力比率增长到一定程度后,要素替代弹性、技术进步参数及规模报酬等的表现有显著变化。

英文摘要：

In this paper, we build two-level CES production functions with capital, labor and energy as inputs, arid is the first time to estimate the parameters using nonlinear measurement and compare all nesting structures, based on the data of 36 industries from 1980 -2011 in China. Also, the threshold effects of CES production function are tested by taking capital/labor ratio as threshold variable. The results show that： （ 1 ） for the three CES nesting structures, most industries can reject the elasticity of substitution equal to one, suggesting that the choose of C-D model in many researches is questionable. （2） for the three CES nesting structures, only few industries cannot reject the returns to scale equal to one, indicating that the hypothesis of constant returns to scale faces high risk in a model. （3） overall, most industries can reject that all three inputs can be put into one single nest. （4） on the whole, the nesting structure where capital and energy are combined first, fits the situation in China best. （5） for most industries, the threshold effects are significant, and the threshold mostly appeared after 1990s, which indicates the performances of elasticity of substitutions, technical progress parameters and returns to scale made a significant change as the capital/labor ratios rising.