Essays on the Econometric Analysis of Distributional Effects with Applications to Labor Market Institutions in Germany

Abstract

The goal of this dissertation is to contribute to a better understanding of the impact of two prominent labor market institutions in Germany: the recently introduced federal minimum wage and the influence of unions on the earnings structure. This is achieved by examining how changes in these institutions affect the distribution of wages and earnings as well as wage and earnings inequality (chapters 2 and 4). Additionally, chapter 3 discusses more generally the limitations and prospects of certain econometric tools for specific distributional analyses in the context of policy evaluation studies. In chapter 2, we employ a distribution regression difference-in-differences approach to assess the causal effects of Germany's national minimum wage on the distribution of hourly wages, monthly earnings, and hours worked. We also explore the use of different bite measures for our difference-in-differences analysis. Utilizing two large-scale databases that account for confounding pre-trends, the analysis for hourly wages reveals that the introduction of the minimum wage shifted distributional mass from below to its threshold and led to substantial spillover effects. Hours changed only to a minor extent. Regarding the effect on the distribution of monthly earnings, we find impacts at the lower middle and middle parts of the distribution. We conclude that the minimum wage explains most of the recent fall in wage and earnings inequality. The third chapter addresses methodological challenges in combining unconditional quantile regressions in the form of RIF regressions with difference-in-differences approaches. A formal assessment of the pitfalls involved when employing a “pooled RIF-DiD” approach is provided, which is shown to lead to misleading estimates for the quantile treatment effect on the treated. Based on this, two alternative methodologies are proposed to overcome these shortcomings: the “DDiD-RIF” and “QDiD-RIF” approaches, which are linked to the underlying identifying assumptions. A Monte Carlo simulation and a practical application further demonstrate that the proposed methods overcome the identified shortcomings of the pooled RIF-DiD approach. In the final fourth chapter of this dissertation, the distributional impacts of the decline in union coverage in Germany from the mid-1990s to 2014 are explored. Employing a grouped quantile regression approach tailored to the German sector-region level collective bargaining context, the study finds significant increases in earnings inequality both between and within worker groups. De-unionization particularly affected the earnings differential between genders and educational levels. A simulation-based approach further provides evidence regarding the implied effect on the unconditional distribution of earnings, with effects being most pronounced at the lower to lower-middle parts of the distribution.