Academic journal article Federal Reserve Bank of Atlanta, Working Paper Series

The Gap between the Conditional Wage Distributions of Incumbents and the Newly Hired Employees: Decomposition and Uniform Ordering

Academic journal article Federal Reserve Bank of Atlanta, Working Paper Series

The Gap between the Conditional Wage Distributions of Incumbents and the Newly Hired Employees: Decomposition and Uniform Ordering

Article excerpt

1 Introduction

Wage differentials among different types of workers, e.g. the gender earnings gap, wage differences between immigrants and natives, etc., have drawn much attention from labor economists and policy makers. There is an extensive literature on labor market outcomes, much of it focused on the analysis of the wage gap at the mean, median and other quantiles of the wage distribution. More recently, techniques have been provided for identifying entire distributions and general function of the distributions. These techniques provide the backdrop for the current paper's approach. One central object of inference in this paper is a summary measure of the "distance" between the entire distributions of interest. Our proposed summary measure makes clear that all other measures of the gap between two distributions are special, and all imply and are implied by well defined welfare functions. Seen in this light, comparison at the mean, median, or any particular quantile would appear to place too much weight on a part of the population, or too equal a weight everywhere. For example, Blau and Kahn (2006) documented the slowing convergence of the gender gap at the mean, median and 90th percentile levels. Albrecht et al. (2003) looked at wages differentials at different parts of the distribution to see whether the gender gap is larger in the upper tail than in the lower tail of the wage distribution due to a "glass ceiling" effect in Sweden. Kampkotter and Sliwka (2011) investigated average wage differences between newly hired and incumbent employees. While these focused examinations are informative and useful, recent papers have examined the wage differentials at the entire distribution level. For example, Massoumi and Wang (2012) employed a metric entropy measure proposed by Granger et al. (2004) to examine the gender wage gap based on the metric distance between two distributions. The measure is the metric member of the Generalized Entropy class of measures with very credible welfare theoretic foundations.

All measures of the gap provide strong ranking of outcome distributions since they are based on implicit "cardinal" welfare or weighting functions. They are inevitably subjective even though some are less extreme than others. In view of this, we explore weak uniform rankings based on the concept of stochastic dominance which allow assessments over entire classes of welfare functions. We do so by rigorous statistical tests for various orders of dominance.

A key issue of interest is about decomposition of observed gaps and rankings in order to identify the factors that underlie the overall wage differentials. Specifically, are those differentials associated with inequality or discrimination in the wage structure, or are they due to human capital composition effect. The classic decomposition method is due to Oaxaca (1973) and Blinder (1973). It is a regression-based method focusing on linear conditional mean decomposition. One major limitation of the Oaxaca-Blinder procedure discussed by Barsky et al. (2002) is that the decomposition provides consistent estimates of the structure and composition effect only under the assumption that the conditional expectation is linear. As suggested by Barsky et al. (2002), an alternative non-parametric decomposition may be based on propensity score reweighting methods, as advocated in DiNardo et al. (1996). The key advantage of this reweighting approach is that it identifies the counterfactual distribution under less restrictive assumptions and hence can easily be applied to more general distributional statistics, rather than the simple mean and quantiles.

Several recent papers, e.g. Firpo et al. (2007), Massoumi and Wang (2012), have applied this reweighting method for wage gap decompositions. Following the recent approach, this paper decomposes the wage gap between newly hired and incumbent employees across the entire distribution. The wage differences between newly hired and incumbent employees is a less studied topic in labor economics. …

Search by... Author
Show... All Results Primary Sources Peer-reviewed

Oops!

An unknown error has occurred. Please click the button below to reload the page. If the problem persists, please try again in a little while.