Semiparametric Analysis of the U.S. Labor Force: Quantile Regression with Sample Selection, Dynamic Choice of Education and Mobility

[320386 .布钦斯基最近的许多劳动经济学研究都致力于理解美国工人的工资结构。本研究采用一种新发展的分位数回归计量经济学技术，调查了男女劳动力的受教育回报和经验回报的变化。该技术描述了整个工资的有条件分配，也允许检查群体内部的不平等。分位数回归技术的使用引起了与样本选择偏差相关的严重理论问题。为了解决这个问题，本课题的第一部分致力于分位数回归估计量的渐近性质的理论推导。第二部分提出用非参数方法考察女性劳动力工资结构的变化。该应用程序使用1964年至1992年的人口普查数据。这一实证调查的结果将阐明与女性劳动力整体有条件分配变化有关的重要问题。项目的第三部分利用第二部分的结果来调查女性劳动力在不同时期的教育选择。该应用程序使用密歇根收入动态研究小组和国家纵向调查的数据来跟踪个人的长期情况。这两项实证分析的结果将有助于解释工资和收入的流动性。如果工资是固定的，那么工资不平等的变化将表明真正的福利影响，否则个人之间的差异将随着时间的推移而平均。* * *

9320386 Buchinsky Much recent research in labor economics is devoted to understanding the wage structure of American workers. This research investigates changes in the returns to schooling and experience for the male and female labor force, using a newly developed econometric technique for quantile regression. The technique describes the entire conditional distribution of wages and also allow an examination of within-group inequality. The use of the quantile regression technique raises serious theoretical concerns related to sample selection bias. In order to handle this, the first part of the proposed project is devoted to the theoretical derivation of the asymptotic properties of the quantile regression estimator. The second part proposes to use nonparametric methods to investigate changes in the female labor force wage structure. The application uses Census of Population Survey data from 1964 to 1992. The findings from this empirical enquiry would shed light on important issues that relate to changes in the entire conditional distribution of the female labor force. The third part of the project uses the results from the second part to investigate educational choices of the female labor force over time. The application uses data from the Michigan Panel Study of Income Dynamics and the National Longitudinal Survey to track individuals over time. The findings from the two empirical analyses would help explain mobility of wages and earnings. If wages are immobile, then changes in wage inequality would be indicative of real welfare implications, otherwise differences among individuals would average out over time. ***