ARE THERE DIFFERENTIAL RETURNS TO SCHOOLING BY GENDER? THE CASE OF INDONESIAN LABOUR MARKETS†
| Author | Anil B. Deolalikar,Jere R. Behrman |
| DOI | http://doi.org/10.1111/j.1468-0084.1995.tb00029.x |
| Date | 01 February 1995 |
| Published date | 01 February 1995 |
OXFORD
BULLETIN
OF
ECONOMICS AND STATISTICS,
57.1
(1995)
0305-9049
ARE
THERE
DIFFERENTIAL
RETURNS
TO
SCHOOLING
BY
GENDER?
"HE
CASE
OF
INDONESIAN
LABOUR
MARKETSt
Jere
R.
Behnnan and
Anil
B.
Deolalikar
1.
INTRODUCTION
In
most developing countries, women have significantly less schooling than
men. For instance, primary school enrollment rates averaged
96
percent for
females and
114
percent for males in
1989
in all low-income countries, and
115
percent and
121
percent respectively in Indonesia. Secondary school
enrollment rates in the same year averaged
3
1
percent for females and
45
per
cent for males in all low-income countries, with parallel figures for Indonesia
of
43
percent and
51
percent.
As
one result
of
such gender differences in
schooling enrollments through the years, the average adult female literacy
rate for low-income countries was
48
percent in 1990, as compared
to
72
Percent for males.
For
Indonesia the reported 1990 adult female and male
literacy rates, respectively, were
68
percent and
86
percent
-
significantly
higher than in most other low-income countries, but
still
with a substantial
gender gap.'
Before jumping to the conclusion that such disparities reflect gender
discrimination
in
schooling investments,
it
is
important
to
know how the
rewards from schooling differ between males and females.
If
males have
higher private rates
of
return
to
schooling than females for a given level
Of
tThe authors are grateful
to
Manin Godfrey and anonymous referees for helpful comments
On,thiS
paper. The authors alone are responsible
for
all interpretations
in
this paper.
All
the statistics in this paragraph are based
on
UNESCO data that are presented in Wor!d
Bank
(
1992).
which is
also
the source
of
the low-income country grouping (including Zndonesra
and
42
other countries, mostly from Asia
and
Sub-Saharan Africa). The enrollment rates. are
s'-c&d
'gross
enrollment rates' that are based on the ratio
of
school enrollers
for
a glven
level relative to estimates
of
the population for the age ranges nomaliy assyla!ed
wth
*at
grade level. The enrollment rates can
(and
indeed often
do
for primary schoollng
In
many
countries
in
recent years) exceed
J
00
percent because there are older or younger
students
than
the
'nomd'
age range attending that school level. There are a number
of
Pfoblems with
the
comparability
of
aggregate schooling data across he and Space that are
dlscusd
in Behmm
md
Rozenzweig
(
1933, 1994).
but they still provide the best available
crude indicators
of
schooling investments across countries.
For
further comparisons
of
the
and recent changes of Indonesian gender gaps
in
schooling
relative
to
the
international
elxWrienCe, see Behrman and Schneider
(I
993).
97
'
BlXkWelI
Ltd.
1995.
Published
by
Blackwell
Publishes,
IOR
Cowley
Road.
oxford
0x4
1JF.
uK
238
Main
Street.
Cambridge.
MA
02
142,
USA.
98
BULLETIN
schooling, the greater schooling attainment among males may reflect an
efficient (although perhaps inequitable) household allocative response to
scarce resources. On the other hand,
if
the reverse is true, the greater school-
ing
of
males
relative to females may represent
a
serious misallocation
of
resources and
loss
of
efficiency.
Recently, a few studies have attempted to estimate labour market rates
of
returns to schooling for males and females in developing countrie~.~.~ These
studies suggest that, while there are significant differences in male-female
earnings even after controlling for schooling and experience, the labor
market rates of return to schooling do not differ significantly by gender. But
most
of
these studies do not control
for
other factors, such
as
gender differ-
ences in school repetition or the differential impact
of
unobserved household
and community endowments! Further,
by
focusing only on earnings func-
tions, most
of
the available studies confound gender differences
in
the impact
of
schooling on wage rates
with
those on hours worked.
In this paper
we
explore whether there are differential impacts
of
schooling
on labor market outcomes for men and women in Indonesia. We
go
beyond
most previous studies in this genre by estimating wage rate functions in addi-
tion to the standard earnings functions and controlling for schooling repeti-
tion and dropout rates and for unobserved household and community
heterogeneity. For the latter controls we explore alternatives
of
fixed and
random effects.
Section I1 summarizes the insights
of
simple human capital theory about
why there might be gender differences in the marginal impact of schooling on
labor market outcomes. Section
I11
describes the data. Section
1V
presents
the functional forms that we use and discusses some estimation issues.
Section
V
presents and discusses our estimates. Section
VI
concludes.
II.
ANALYTICAL lSSUES
Gender differences
in
wages may arise for several
reasons:
gender differences
in traits, such as manual dexterity, stamina or strength, that are valued by the
market; gender specialization
in
jobs and relative scarcity
of
one gender; and
These include Behrman and Wolfe
(1991)
for Nicaragua, and Birdsall and Behrman for
Brazil (1 991), Gannicott (1986) for Taiwan, Gindling
(1
988) for Brazil and Khandkar
(
1990)
for
Peru.
Also
see Schultz
(
1993b).
’
There
also
may be gender differences in the non-pecuniary returns
to
schooling, such as the
impact
on
health
and nutrition. The empirical evidence
on
these gender differences
in
develop
ing countries tends to indicate that the returns are greater,
if
they differ, for female schooling
than
for male schooling
(e.g.
see Behrman, 1990b; King and Hill, 1993; Schultz, 1993a.b).
In
this
study
we
consider
only
the returns in terms
of
the wage market, which are
an
important and
growing component of
the
total returns and the ones that are most emphasized in the literature.
Our
estimates
for
Indonesia
in
Behrman
and
Deolalikar
(
1991,
1993)
suggest that controls
for
school
repetition experience and unobserved heterogeneity significantly influence estimated
returns
to
schooling.
Also
see Behrman and Birdsall
(
1987) and Behrman
(
1987, 1990a.b) and
Schultz (1988. 1989). Khandkar (1990) is an exception with regard
to
considering wage rates
and controlling for unobserved household characteristics.
0
Basil
Blackwell
Lid.
1995.
To continue reading
Request your trial