Accounting for Selection Effects in the Analysis of Wage Inequality in Germany

本项目的目的是分析过去几十年来德国工资不平等的加剧。为了评估所观察到的工资分配变化和劳动力市场群体之间工资差异的驱动因素，重要的是要考虑到选择有偿工作可能会随着时间的推移而变化，并且可能会因群体而异。鉴于就业波动、长期就业增长、女性劳动力参与率上升以及转向非全日制工作，构成变化对于分析工资不平等非常重要。该项目量化了在观察到的和未观察到的工人特征的差异下，不同群体之间以及随着时间的推移工资分配的差异。到目前为止，本项目使用DiNardo等（1996）的权重调整法和Firpo等（2014）的RIF回归法，分析了可观测特征的构成和工资结构效应。在分析性别工资差距时，采用Albrecht等人（2009年）的方法，对不可观测的选择进行了控制。关于如何在结果分布的分析中控制对不可观测的选择，只有一小部分但最近非常活跃的方法学文献，参见Huber和Melly（2015），Arellano和Bon-homme（2016）和D 'Hautefoeuille等人。第二个资助期的项目建议书由五个部分组成：第2A部分扩展了项目第一阶段的工作，并详细探讨了估计工资数据分位数回归的方法问题。这一部分的目的是在文献的最新发展的基础上作出方法上的贡献，并作为实证分析的方法基础。第2B部分估计了带有选择校正的分位数回归，以详细分析工资随时间的演变。我们还分析了为什么德国工资不平等的增加在2010年已经结束，从而扩展了以前的工作。这一部分将侧重于不同数据集的适用性和不同工具的适用性，以控制基于不可观察的有偿工作的选择。第2C部分分析了性别工资差距，同时还考虑到广泛的方法论方法，对不可观测的选择进行了校正。第2D部分讨论了教育的大幅扩张如何影响教育学位的质量和回报的问题。我们将研究这样一个假设，即高等教育招生人数的增加降低了毕业生的平均质量，从而降低了高等教育的回报。第2 E部分分析了个人劳动力市场转型的变化与工资演变之间的关系。特别是，我们将重点关注临时非全时工作发生率增加和男性非全时工作显著增加的影响。

The objective of this project is to analyze the rise of wage inequality in Germany over the last decades. For an assessment of the drivers for the observed changes in the wage distribution and the wage dif-ferences between labor market groups, it is important to take into account that the selection into paid work may change over time and may differ across groups. In light of employment fluctuations, long-term employment growth, rising female labor force participation, and a shift towards part-time work, composition changes are very important for the analysis of wage inequality. This project quantifies the differences in wage distributions between groups and over time that are free of differences in ob-served and unobserved worker characteristics. So far, this project has analyzed the composition and wage structure effects of observable characteristics, using the reweighting approach by DiNardo et al. (1996) and the RIF regression approach of Firpo et al. (2014). Selection on unobservables has been controlled for using the method of Albrecht et al. (2009) in the analysis of the gender wage gap. There exists only a small but recently very active methodological literature on how to control for selection on unobservables in the analysis of outcome distributions, see Huber and Melly (2015), Arellano and Bon-homme (2016), and D'Hautefoeuille et al. (2017). The project proposal for the second funding period consists of five parts: Part 2A extends work from the first project phase and explores in detail the methodological issues involved in estimating quantile regression for wage data with selection correction. This part aims to make a methodological contribution building on recent developments in the literature and serves as a methodological basis for the empirical analysis. Part 2B estimates quantile regressions with selection correction for a detailed analysis of the evolution of wages over time. We extend the previous work by also analyzing as to why the increase in wage inequality in Germany has come to an end in 2010. This part will focus on the suitability of different datasets and the suitability of different instruments to control for selection into paid work based upon unobservables. Part 2C analyzes the gender wage gap, while also correcting for selection on unobservables considering a wide range of methodological approaches. Part 2D addresses the question of how the considerable educational expansion has influenced the quality and the returns to educational degrees. We shall study the hypothesis that increasing enrollment in higher education has deteriorated the average quality of graduates and thus reduced the returns to higher education. Part 2E analyzes the relationship between changes in individual labor market transitions and the evolution of wages. In particular, we will focus on the effects of increasing incidence of temporary part time work and the notable increase in male part-time work.