Distribution of the gender wage gap with endogenous human capital: evidence for Spain

Pages25-45
Publication Date07 April 2015
DOIhttps://doi.org/10.1108/EBHRM-05-2013-0012
AuthorLucía Navarro-Gómez,Mario F. Rueda-Narvaez
SubjectHR & organizational behaviour,Global HRM
Distribution of the gender wage
gap with endogenous human
capital: evidence for Spain
Lucía Navarro-Gómez and Mario F. Rueda-Narvaez
Departamento de Estadística y Econometría, Universidad de Málaga,
Málaga, Spain
Abstract
Purpose The purpose of this paper is to provide empirical evidence on gender wage discrimination
and how it is distributed among women in the Spanish labour market, where female participation has
been rising for decades. The empirical approach aims to assess to which extent discrimination is
evenly distributed or not among women, and how different subgroups of workers are affected by it.
Design/methodology/approach Using data from the Spanish section of the European Community
Household Panel (1994-2001) the authors estimate earnings equations for men and women using the
instrumental variable (IV) method proposed by Hausman and Taylor (1981). This aims to avoid biases
resulting from endogeneity of regressors. Building on these results, the authors follow the proposal of
Jenkins (1994) and estimate a bivariate wage distribution for women, containing individual expected
earnings with and without discrimination.
Findings The results show that discrimination is distributed unevenly across female workers and
that the degree to which women are discriminated against grows as they move upward in the wage
distribution. Also, when wage determinants are allowed to be endogenous, the results experience
drastic changes, both in average and distributional terms.
Research limitations/implications The results point to a glass ceilingoperating on female
earnings and also show that endogeneity of human capital should be taken into account when
analysing discrimination. Therefore, more empirical evidence in this line would be welcome.
Originality/value By using IV estimation of wages, the authors control for the existence of
endogeneity in earnings equations. Also, the authors provide unexplained wage differentials for
particular groups of female wage earners, specially according to education, experience and job tenure.
Keywords Promotion and compensation, Labour economics
Paper type Research paper
1. Introduction
The main theoretical challenge posed by research on wage discrimination is to explain
the sizeable observed differences between male and female average earnings (the gender
wage gap), whether caused by gender-based differences in characteristics and/or
preferences or by the existence of discrimination, which has to be modelled somehow.
The latter explanation involves answering the question quoted by Cain (1986): Under
what conditions is it possible for essentially equal goods to have different prices when
exchanged in competitive markets?”–that is,why and how can men and women differ in
what they receive for equally productive work? Theoretically, this situation should not
even arise. If any woman produces as much as a man, but receives lower wages, any
employer would take advantage of the situation in order to have a cheaper workforce,
thus obtaining extra benefits. One would expect many firms to do the same thing, so
that in the long run female wages would rise until they caught up with male wages. Evidence-based HRM: a Global
Forum for Empirical Scholarship
Vol. 3 No. 1, 2015
pp. 25-45
©Emerald Group Publis hing Limited
2049-3983
DOI 10.1108/EBHRM-05-2013-0012
Received 31 May 2013
Revised 30 October 2013
Accepted 8 November 2013
The current issue and full text archive of this journal is available on Emerald Insight at:
www.emeraldinsight.com/2049-3983.htm
The authors acknowledge and thank the financial assistance of the Andalusian Government
through Project No. P09-SEJ-4859.
25
Distribution of
the gender
wage gap
However, empirical research fails to explain the entire observed wage gap in terms of
differences in characteristics. This leaves the remainder of the gap consistent with the
presence of discrimination in the labour market. Much theoretical work has been
devoted to reconciling these two seemingly contradicting positions (for an overview,
see Altonji and Blank, 1999).
From the point of view of empirical research, the focus is usually on verifying and
quantifying wage discrimination. This leads methodologically to the application of
human capital theory (Becker, 1964/1993) and estimating some kind of Mincerian wage
equation (Mincer, 1974), separately for men and women. While the female parameters
are thought to represent what women receive when being discriminated against, the
male parameters are used to estimate the counterfactual wage a female worker would
receive if she were a man, but otherwise equal. From these results, the mean differenc e
in (log) wages can be decomposed into one part being explained by differences in
human capital and other characteristics (differences in productivity) and one part
arising because the labour market values wo rkerscharacteristics differ ently
depending on whether the worker is a man or a woman. This latter part (resid ual or
unexplained) is then taken as an estimate of the mean amount of discrimination in the
labour market (Oaxaca, 1973; Blinder, 1973).
More recently, the focus has shifted to the analysis of gender wage gaps in different
quantiles of the wage distribution using quantile regressions (QR) as the procedure to
estimate earnings equations. Most of these new distributional analyses rely on the
methodology proposed by Machado and Mata (2005). In essence, QR are performed for
male and female samples and then the difference in corresponding quantiles of the male
and female wage distributions are decomposed into a part explained by differences in
characteristics and another part due to differences in returns (coefficients) to those
characteristics. This has allowed researchers, among other issues, to test the existence
of a glass ceiling, that is, a discrimination profile that is increasing in the wages of
female workers. Studies in this line of work include Albrecht et al. (2003) for Sweden,
Arulampalam et al. (2007) for a number of European Union countries and, for the case
of Spain, Gardeazabal and Ugidos (2005), de la Rica et al. (2008) and del Río et al. (2011).
In a recent paper, Albrecht et al. (2009) provide an expansion of the Machado and Mata
(2005) procedure that takes into account sample selection (especially of women) in the
earnings equations. Beyond looking at gaps at wage quantiles, some distributional
analyses view discrimination as an individual trait that can be estimated. Originally,
Jenkins (1994) proposed using MCO estimates of wage equations to construct individual
measures of discrimination. Del Río et al. (2011) extend Jenkinss methodology to the use
of QR. Under this approach, individual estimates of discrimination allow for comparison of
different groups of women, not only according to their position in the wage distribution.
In Spain, empirical research on discrimination has been relatively frequent in recent
years due to the availability of new databases with suitable information on wages and
other attributes of workers. Interest has been devoted to two lines of work. On the one
hand, articles such as De la Rica and Ugidos (1995) use cross-sec tional data to estimate
the average amount of discrimination, applying to the female wage equation a variation
of Heckmans (1979) model for data with sample selection problems. On the other hand,
García et al. (2001) is the first in a series of papers to use QR to assess the degree of
discrimination at different points of the wage distribution. Evidence on discrimination
in the Spanish market tends to conclude that differences in characteristics have
a limited to no role in explaining observed (raw) wage differentials. This means that
most of the raw differential is due to differences in the coefficients of earnings
26
EBHRM
3,1

To continue reading

Request your trial

VLEX uses login cookies to provide you with a better browsing experience. If you click on 'Accept' or continue browsing this site we consider that you accept our cookie policy. ACCEPT