Research on modeling principle and application of non mean model非均值模型建模原理及其应用研究.pdfVIP

摘 要 构建合适的模型是计量经济研究的第一步，常用的经典均值模型存在一定局限 性，一方面反映的特征有限，关注的只是变量的条件均值分布，许多细部特征无法 描绘；另一方面模型假定过于严格，如随机项需满足同方差、服从正态分布等条件， 现实事件很难完全满足这些条件，故许多社会经济问题的分析无法运用均值模型解 决。随着现代计量经济学的发展和研究的需要，许多弥补均值模型缺陷的模型应运 而生。 本文按照模型反映的不同分布特征，将计量经济模型分为均值模型和非均值模 型两大类。非均值模型主要产生于二十世纪七八十年代，是在均值模型的基础上发 展而来，弥补了均值模型的一些缺陷成为现代计量经济学的重要组成部分。 在现代计量经济学的框架下，本文将非均值模型分为四类：条件分位数回归模 型；方差模型；概率模型（包括离散选择模型、持续时间模型）；分布函数模型。 文中简单回顾了非均值模型的理论和应用研究，概括了建模原理及建模特点，并与 均值模型进行了比较分析。此外，针对山东省人力资源社会保障情况的调查数据， 将几种非均值模型汇总在一起，对调查数据进行分析，说明非均值模型在实例中的 应用，最后，总结了非均值模型的发展和意义，并点出文章存在或需要进一步解决 的问题。 关键词：非均值模型 建模原理 比较分析 应用实例 i Abstract The first step of the econometric study is to build a proper model. Generally, the classical econometric model is considered to be the traditional mean regression model. The mean regression model has some limitations, on one hand, which can only reflect limited features and is focused on conditional mean; on the other hand, it has to meet some rigorous hypotheses, such as error must follow a normal distribution with zero mean valued and homo-skedastcity, so it is not applicable in some practical situation. So the analysis of many social and economic problems cant be solved by the mean regression model. Due to the development of econometrics and the needs of the research, many models making up for the defect of the mean regression model emerge at a historic moment. According to different distribution characteristics of models，econometric model is divided into two categories: the mean regression model and the non-mean regression model. The non-mean regression model mainly appeared in 1970s and 1980s, the development of which is based t