Semi-Parametric Estimation of Sample Selection Models

本研究的目的是发展样本选择模型估计的半参数方法。这些方法将允许同时估计选择方程和结果方程。除了对两种方程样本选择模型的估计外，还将考虑更复杂的样本选择模型的估计。所有这些模型都有某些共同的特征，但它们也有自己的结构。这样的结构对模型施加了特定的识别限制。该项目将开发半参数估计方法，探索模型的系统结构。研究的系统估计方法中有半参数似然方法。将进行蒙特卡罗实验来研究所提出的估计器的小样本性能。这项研究很重要，因为样本选择模型有许多重要的应用，例如在求职行为的分析中，并且将要开发的方法将允许更好地估计这些模型和其他模型。

The purpose of this research is to develop semi-parametric methods for the estimation of sample selection models. These methods will allow for the simultaneous estimation of choice equations and outcome equations. In addition to the estimation of the two equations sample-selection models, the estimation of more complicated sample-selection models will also be considered. All of these models have certain features in common, but they also have structures of their own. Such structures impose specific identification restrictions on the models. This project will develop semi-parametric estimation methods which explore the systematic structures of the models. Some of the systematic estimation methods to be investigated are the semi-parametric likelihood methods. Monte Carlo experiments will be performed to investigate the small sample performance of the proposed estimators. This research is important because sample-selection models have many important applications, such as in the analysis of job-search behavior, and the methods to be developed will permit for better estimating these and other models.