部分参数识别模型的贝叶斯统计推断

在计量经济模型分析中，部分参数不可识别是一个常见现象。这类具有部分参数不可识别的计量经济学模型，通常被称为部分识别模型(partially identified models)。近些年来，对于部分识别模型，如何进行统计推断是计量经济学研究领域的一个热点研究问题。本项目针对该类模型，基于贝叶斯计量经济学方法，研究该类模型的统计推断问题，包括参数置信区间（集合）构建，模型设定性检验，以及模型选择等。最后，本项目将基于所发展的统计推断方法，研究宏观经济学、微观经济学以及金融学中常见的部分识别模型的统计推断问题，用以具体经济实践。

Econometric models often do not point identify parameters of interest. This class of models are referred to as partially identified models in the literature. In recent years, partial identification in econometrics has been a theoretically and empirically attractive area of research in econometrics, especially how to conduct credible statistical inference for these models. Based on Bayesian methods, this project will develop the statistical inference procedures for partially identified models, including confidence sets for identified sets, model specification tests, model selection and etc. Finally, our proposed procedures will be applied to some well-known partially identified models in macroeconomics, microeconomics and finance.