半参数工具变量模型的统计方法、理论及其应用

项目名称： 半参数工具变量模型的统计方法、理论及其应用

项目编号： No.11301569

项目类型： 青年科学基金项目

立项/批准年度： 2014

项目学科： 数理科学和化学

项目作者： 杨宜平

作者单位： 重庆工商大学

项目金额： 23万元

中文摘要： 半参数工具变量模型在计量经济学中有着广泛的应用。该模型既含有参数分量, 又含有非参数分量,比单纯的参数回归模型或非参数回归模型有更大的适应性, 并具有较强的解释能力，而工具变量的引入能有效的解决内生解释变量所带来的问题。对该模型的统计方法和理论研究是当前重要而又富有挑战的课题之一。目前关于半参数工具变量模型的统计推断还未大量展开。本课题针对半参数工具变量模型拟提出有效的估计方法并讨论估计的渐近性质以及兴趣参数的区间估计；拟采用广义似然或经验似然等方法研究半参数工具变量模型的检验问题，考察所提出检验统计量的大样本性质；对半参数工具变量模型探讨有效的变量选择方法。此外，通过模拟研究和实证分析所提出方法的有限样本性质。本课题的研究期望进一步完善半参数工具变量模型的统计方法和理论，并在理论和应用上有所突破。

中文关键词： 工具变量；半参数回归；内生变量；经验似然；变量选择

英文摘要： Semiparametric instrumental variable models have a wide range of applications in econometrics. The models contain both the parametric components and the nonparametric components, which have more implements and stronger explanations than pure parametric or nonparametric regression models, and the introduction of instrumental variables can effectively solve the problem of the endogenous explanatory variables. It is becoming one of the important and challenging issues to study the statistical methods and theoretical properties of the models. Currently there is little research on the models. The topics will propose a series of the effective estimation methods, and discuss the asymptotic properties and confidence region construction of parameters of interest. We will study the test of semiparametric instrumental variable models by the generalized likelihood or empirical likelihood method, and investigate large sample properties of the proposed test statistics. A series of variable selection procedures will be proposed. Furthermore, the finite sample properties of the proposed methods will be assessed through simulation studies and some real examples. We hope to further improve statistical methods and theory of the semiparametric instrumental variable models and have a breakthrough in the theory and application.

英文关键词： instrumental variables；semiparametric regression；endogenous vairables；empirical likelihood；variable selection