Simulation and Semiparametric Estimation of MicroeconometricModels

本项目由最大模拟似然估计、半参数最小距离估计和具有验证数据信息的非线性变量误差计量模型的估计三个相互独立但又相互关联的研究项目组成。这些项目研究许多重要的微观计量经济学模型的估计和统计检验的模拟和半参数方法。这些方法将有助于估计和检验离散面板数据模型、因变量有限的联立方程模型、多市场非均衡模型、具有非负性和/或数量约束的消费者需求和生产系统、以及具有变量误差的非线性模型。更具体地说，该项目的第一部分涉及推广最大模拟似然估计方法的偏差校正程序，这些方法是根据先前的NSF对离散选择模型的拨款而开发的。这些估算和推断程序被应用于使用印度尼西亚的数据对消费者需求和/或生产系统进行经验估算。第二部分介绍了具有定性和有限因变量的联立方程模型、多市场非均衡模型和具有条件期望规范的样本选择模型的经典最小距离估计方法的半参数估计过程。这类模型的半参数估计方法在目前的计量经济学文献中是有限的。第三部分发展了考虑验证数据和调查面板数据的信息以及所有变量的测量误差的估计方法。建议的方法将被应用于使用收入动态小组研究的数据来研究劳动力供应模型。

This project consists of three separate but related research projects in maximum simulated likelihood estimation, semiparametric minimum distance estimation and estimation of nonlinear error-in-variables econometric models with validation data information. These projects investigate simulation and semiparametric methods of estimation and statistical tests for many important microeconometric models. The methodologies will be useful for the estimation and testing of discrete panel data models, simultaneous equation models with limited dependent variables, multimarket disequilibrium models, consumer demand and production systems with nonnegativity and/or quantity constraints, and nonlinear models with error-in-variables. More specifically, the first part of the project involves the generalization of bias correction procedures for maximum simulated likelihood estimation methods developed under the previous NSF grant for discrete choice models. These estimation and inference procedures are applied to empirical estimation of consumer demand and/or production systems with data from Indonesia. The second part introduces semiparametric estimation procedures for the classical minimum distance estimation method for simultaneous equation models with qualitative and limited dependent variables, multimarket disequilibrium models, and sample selection models with conditional expectation specifications. Semiparametric estimation methods of such models are limited in the current econometric literature. The third part develops estimation methods which take into account the information of validation data and survey panel data and the measurement errors in all of the variables. The proposed methodologies will be applied to the study of a labor supply model using data from the Panel Study of Income Dynamics.