内生性随机前沿模型估计方法研究：无需工具变量的Copula方法

内生性是常见的计量问题，忽略内生性会导致估计量有偏且不一致。现有部分文献研究了内生性随机前沿模型的估计，但实现的前提是能够为内生性自变量寻找到合适的工具变量，而实际情况下合适的工具变量通常不容易获取。本文研究了在难以找到合适的工具变量的情况下内生性随机前沿模型的估计问题：结合Copula方法和极大模拟似然方法估计参数。此外，本文还构造了技术无效率的新的点估计，该点估计额外利用了内生自变量的信息，通常比JLMS法对应的点估计更有效。数值模拟表明，相比于已有研究，本文提出的方法估计精度更高。

The Estimation for Endogeneous Stochastic Frontier Models: Copula Approach without Instrumental Variables

Endogeneity is a common econometric problem. Ignoring Endogeneity will result in biased and inconsistent estimators. Some researches study the estimation methods for endogeneous stochastic frontier models, and these methods all need to find the instrumental variables for endogeneous independent variables. However the proper instrumental variables are usually hard to get. This paper aims at the situation when it’s hard to find the proper instrumental variables. Copula method and maximum simulated likelihood method are used to get the estimation of endogeneous stochastic frontier models and construct the new point estimation for technical inefficiencies. The new point estimation absorbs the information of endogeneous variables and is more efficient than the point estimation of JLMS. The numerical simulations show that the method in this paper gets higher accuracies compared with the existing methods.