The gender pay gap at labour market entrance: Evidence from Germany

• Author: Andreas BEHR,Katja THEUNE
• Date: 01 March 2018
• Published date: 01 March 2018
• DOI: http://doi.org/10.1111/ilr.12037

International Labour Review, Vol. 157 (2018), No. 1

Copyright © The authors 2018

Journal compilation © International Labour Organization 2018

* University of Duisburg-Essen, Germany, emails: andreas.behr@vwl.uni-due.de; katja.

theune@vwl.uni-due.de.

Responsibility for opinions expressed in signed articles rests solely with their authors, and

publication does not constitute an endorsement by the ILO.

The gender pay gap

at labour market entrance:

Evidence from Germany

Andreas BEHR* and Katja THEUNE*

Abstract. The authors investigate wage differentials between German male and

female graduates at labour market entrance at all percentiles of the wage distribu-

tion. The wage differentials are decomposed using the single-index approach sug-

gested by DiNardo, Fortinand Lemieux(199 6) and Fortinand Lemieux(199 8),

and by endowment, price and returns-to-skill function effects. The results reveal

higher starting salaries for men at all percentiles of the income distribution, with

the pay gap varying in magnitude. The endowment and price effects are favour-

able for men throughout. The effect of the difference in the returns-to-skill function

works in favour of female graduates.

In this article, we provide evidence of the pay gap at labour market

entrance in Germany. While numerous studies analyse gender pay differ-

entials in general, little has been written on gender pay gaps among gradu-

ates at career entry.

There are several reasons why it is important to analyse entry wages. Ger-

hart(1990) claims that current salary differentials are largely due to starting

salary differentials. Moreover, pay rises and other forms of payment are often

based on current salaries (see Graham, Hotchkissand Gerhart,2000). Hence,

starting wage differences between men and women can be assumed to per-

sist over time. Entering the labour market under specic conditions may also

have an effect on starting wages and the long-term wage gap (see, for ex-

ample, Oreopoulos, von Wachterand Heisz, 2 012). While at labour market

entry, experience and human capital endowments are similar for men and

women, women tend to acquire less human capital on average than men

because, for instance, they subsequently bear children and devote more time

to parenting. By analysing entry wages, the focus is on the effect of study

performance, the choice of eld of study and other individual characteristics.

International Labour Review84

The empirical economic literature explaining income differences between

groups is rooted in the original works of Oaxaca(1973) and Blinder(1973).

Their approach has been considerably extended in the past 30 years by analys-

ing complete income distributions and differences between group incomes at

all percentiles.

The approach used in this analysis was suggested by DiNardo, Fortin and

Lemieux (1996) and Fortinand Lemieux(1998) and can be classied as a single-

skill index model. It uses an ordered probit model to estimate a exible returns-

to-skill function that transforms skills monotonically into wages. By comparing

two wage distributions, single-index approaches estimate the isolated effects

of differences in characteristics, in prices of characteristics, in skills and in the

wage structure.

Suen (1997) has convincingly argued that decomposition approaches

should make it possible to clearly distinguish between changes in skills and

changes in the wage structure (see also Yun,2009). Separating these effects is the

main analytical gain of the applied non-parametric decomposition techniques

suggested by Fortinand Lemieux(1998).

1

In addition, decomposition approaches

result naturally in estimates of the isolated components of wage differences at

all percentiles of the wage distribution.

Single-index models can be seen as a generalization of the standard human

capital competitive markets model of wage determination. The basic assump-

tions are that, in the labour market, relevant skills are priced by means of a re-

turns-to-skill function and can be measured by a skill index. The skill index itself

can be thought of as a weighted aggregation of individual labour market-relevant

characteristics. By introducing the concept of a skill index, changes in the rela-

tive importance of observed characteristics can be disentangled from changes in

the wage structure. Changes in the wage structure are dened as general shifts

in the returns-to-skill function that leave the ordering of individuals constant.

A comprehensive theoretical discussion of decomposition methods in eco-

nomics is provided by Fortin, Lemieuxand Firpo (2011). A detailed comparison

and empirical application of four different approaches is given by Behr(2013)

for the two-country case (United States and Germany).

The remainder of the article is organized as follows: the rst section pro-

vides a brief overview of the relevant literature; the second section discusses the

data set and the variable construction, and provides a rst descriptive analysis;

the third section presents the Oaxaca-Blinder decomposition; the single-skill

index model and the empirical results are presented in the fourth section; the

conclusion is set out in fth section.

Literature review

This section focuses on the literature analysing graduates’ wages at labour

market entry. While it is generally agreed that a gender pay gap exists in

graduates’ starting salaries, the empirical ndings on the extent of the gap

1 An alternative single-index approach has been suggested by Donald, Green and Paarsch

(2000).