Can We Still Learn Something From The Relationship Between Fertility and MothersEmployment Evidence From Developing Countries

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 25

Can We Still Learn Something From the Relationship Between Fertility and Mother's

Employment? Evidence From Developing Countries


Author(s): Julio Cceres-Delpiano
Source: Demography, Vol. 49, No. 1 (February 2012), pp. 151-174
Published by: Springer on behalf of the Population Association of America
Stable URL: http://www.jstor.org/stable/41408223
Accessed: 01-09-2016 13:17 UTC

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted
digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about
JSTOR, please contact support@jstor.org.

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
http://about.jstor.org/terms

Springer, Population Association of America are collaborating with JSTOR to digitize, preserve and
extend access to Demography

This content downloaded from 202.92.128.169 on Thu, 01 Sep 2016 13:17:07 UTC
All use subject to http://about.jstor.org/terms
Demography (2012) 49:151-174
DOI 10.1 007/sl 3524-0 1 1 -0076-6

Can We Still Learn Something From the Relationship


Between Fertility and Mother's Employment? Evidence
From Developing Countries

Julio Cceres-Delpiano

Published online: 19 January 2012


Population Association of America 2012

Abstract In this work, I study the impact of fertility on mothers' employment for a
sample of developing countries. Using the event of multiple births as an instrumental
variable (IV) for fertility, I find that having children has a negative impact on female
employment. In addition, three types of heterogeneity are found. First, the magnitude
of the impact depends on the birth at which the increase in fertility takes place.
Second, the types of jobs affected by a fertility shock (multiple births) are jobs
identified with a higher degree of informality, such as self-employment or unpaid
jobs. Finally, the heterogeneity analysis reveals that an unexpected change in fertility
is stronger at a higher education level of the mother and in urban areas.

Keywords Fertility Female labor force participation Developing countries

Introduction

The relationship between female labor force attachment and fertility has long
been of interest to scholars. Cristia (2008) pointed out three reasons for this
interest. First, the increase in female employment in the United States, among other
countries, after World War II can be explained by delayed childbearing and
reduced fertility (Goldin 1990). Second, evidence supports the fact that the
interruption of work due to childbearing is partially responsible for the female-
male wage gap (Korenman and Neumark 1992). Third, in the context of a
household production model, a reduction in labor force attachment after birth can
be seen as a substitution of market-intensive forms of child investment by time-
intensive alternatives of child care; thus, knowing the effect of childbearing on
mothers' employment provides information about the type of inputs invested in a
child (Blau and Grossberg 1992; Cceres-Delpiano 2006).

J. Cceres-Delpiano (El)
Department of Economics, Universidad Carlos III de Madrid, Calle Madrid 126, Getafe 28903
Madrid, Spain
e-mail: jcaceres@eco.uc3m.es

Springer

This content downloaded from 202.92.128.169 on Thu, 01 Sep 2016 13:17:07 UTC
All use subject to http://about.jstor.org/terms
152 J. Cceres-Delpiano

Even when limited to stud


considerable empirical evid
fertility on female labor pa
child, Rosenzweig and Wo
al. (1999) used the fact tha
family size. Angrist and E
siblings to estimate the eff
miscarriage in a woman's f
delayed childbearing on an
Cristia (2008), instead of l
women for which the end
women faced with fertili
study the impact of a first
However, less evidence exi
(Browning 1992; Schultz 2
the beginning of this sectio
in the labor market of dev
family activities (Mammen
underrepresentation in t
arrangements in underdev
in labor outcomes: women
market (Blunch et al. 200 1
- such as in developing c
mother's time-intensive ty
in child well-being. Ccere
Surveys (DHS) data for de
fertility changes the like
changes the likelihood of v
therefore an increase in h
the household: that is, a re
Cruces and Galiani (2007
studies, to my knowledge,
and provided evidence for
Angrist and Evans's (1998)
negative impact of numb
Argentina. Agero and Ma
from the DHS data and the
size, did not find a signific
mothers' employment, me
In this article, using micr
advance the literature in

1Although the authors did not fin


sector, they observed some trend
to be engaged in nonwage employ
women are in wage employmen
distribution. Both elements are l
sector.

Springer

This content downloaded from 202.92.128.169 on Thu, 01 Sep 2016 13:17:07 UTC
All use subject to http://about.jstor.org/terms
Fertility and Mother's Employment in Developing Countries 153

relationship between fertility and mother's labor force attachment. Second, un


Agero and Marks, by using multiple births as a source of variation in numbe
siblings, I am able to investigate a shock in the number of children in higher par
such as multiple births in a second, third, and fourth birth,2 which is the tar
population that we expect family planning programs to have in mind. Thi
analyze the impact of fertility on different measures of mothers' employment. T
essential for several reasons. In developing countries, labor markets are characte
by higher levels of informality, a considerable share of the employment in ru
regions, and heterogeneous payment alternatives (paid versus unpaid, for exam
These dimensions are important when individuals perceive jobs as alternatives
different quality or as providing a different menu of services. With this in min
study not only the overall impact on the labor supply provided by these mother
also the type of employment affected by fertility. Fourth, by studying a shif
fertility across different parities and for different subsamples, I provide evidenc
the heterogeneous impact of an increasing number of children on these differ
definitions of female employment and individuals.
The results of this article reveal that as with previous studies for developed countr
a shock in family size has a negative impact on female employment. Nevertheless
types of heterogeneity are found. First, the size and sign of the impact depend o
birth at which the increase in family size is studied: a negative impact of shift in fer
is observed at the first birth or third and higher birth, but a shift in the middle o
distribution of family size (second birth) can even be positive for some samples
definitions of mother's employment). Second, the types of jobs affected by a chang
fertility differ, depending on the birth at which the shift in number of children t
place. At lower births (parities), jobs of higher degrees of informality - unpaid jo
jobs that are harder to combine with childbearing, such as working away from hom
seasonal jobs with unpredictable schedules - are affected; at higher parities, all ty
jobs are affected by the shift in fertility.
The article is structured as follows. First, I describe the empirical specificat
and identification strategy. Next, I specify the data used in the analysis, the cr
applied to the construction of samples, and outcomes measuring mother's labor
attachment. Finally, I present the results are presented and offer conclusions.

Identification

The following equation corresponds to the relationship of interest in the current


analysis:

2 Agero and Marks, by the nature of the source of variation they used (infertility), captured the impact of
fertility at lower parities. In a context of heterogeneity in the impact of family size and individuals
behaving as a function of this heterogeneity, the parameters estimated by instruments can be interpreted as
local average treatment effects (LATE) (Angrist and Imbens 1994). In that sense, the external validity of
the estimates is compromised in a context where the population of interest for policy makers or
development institutions promoting family planning programs are those families in the upper tail of the
fertility distribution.

Springer

This content downloaded from 202.92.128.169 on Thu, 01 Sep 2016 13:17:07 UTC
All use subject to http://about.jstor.org/terms
1 54 J. Cceres-Delpiano

where ict represents a meas


country c, at year t, for a
other covariates for the sam
fixed effects, respectively
impact of other covariate
America, Africa, and Asia.
The impact of family si
by y5. As documented in t
this equation may be subject to an omitted variable bias (Schultz 2007).
Therefore, statistical inference based on OLS will provide an inconsistent
estimate of
To address this omitted variable bias, I apply an empirical identification strategy
similar to the one used by Black et al. (2005), Cceres-Delpiano (2006), and Angrist
et al. (2010). Specifically, the event of multiple births is used as the exogenous shift
in fertility. Unlike Black et al. (2005), Cceres-Delpiano (2006), and Angrist et al.
(2010), whose observations are children in a family, the observations in this analysis
correspond to mothers, and therefore there is only one observation per family. Let
mbsi denote the binary instrument, multiple birth, which takes a value equal to 1 for
a family (mother) i with a multiple birth in the s birth, and 0 otherwise. Specifically,
in the analysis, four subsamples are defined according to the value of s. The first
subsample consists of mothers with one or more births (1+) whose instrument is
mb' ' and the second one, families with two or more children (2+) whose instrument
is mb2; and so on.
Whether the occurrence of multiple births is an appropriate instrument
depends on the legitimacy of two well-known assumptions. First, the
correlation between multiple births and family size is different from zero. This
assumption implies that there should be enough correlation between multiple
births and family size so that an average difference in family size exists and
can be measured properly. Women who experience a multiple birth have some
ability to adjust their subsequent fertility. For example, a mother who would
like four children may simply stop having children if she delivers twins on her
third birth. This is particularly problematic when working with developing
countries, where desired fertility is higher. Nevertheless, heterogeneity in the
ideal number of children ensures that at least for some individuals, multiple
births produce a shift in family size. In the following sections, I show that
multiple births, in fact, shift the mother's number of children upward for
different family sizes.
The second assumption, which is nontestable, is noncorrelation between the
instrument and the error term in the regression. This assumption implies that the
impact that is observed over the variable of interest should be attributed to a change
in family size and not to other factors. Twins are the most common of the multiple
pregnancies. Unlike identical twins, the occurrence of fraternal twins varies, and

3 Ideally, as the sample size grows, I should allow that c* = : that is, the impact of all the variables in the
model could differ among countries. This is the equivalent of running a regression separately for each
country. However, the relatively small sample size for each country and the data-intensive requirements of
the identification strategy permit a point between a model fully flexible (with * = c) and a model that
assumes the impact that is constant across countries.

Springer

This content downloaded from 202.92.128.169 on Thu, 01 Sep 2016 13:17:07 UTC
All use subject to http://about.jstor.org/terms
Fertility and Mother's Employment in Developing Countries 155

several risk factors may contribute to an increase in their incidence.4 In the lit
two concerns are related to using multiple births as an instrument for fertili
multiple births have a higher incidence among mothers undergoing f
treatments and among women who come from families with a history of fr
twins. Nevertheless, given the sample under analysis (developing countries
costs associated with fertility treatments, the use of fertility drugs does not
be a concern in this analysis. Also, there is no a priori information that wom
differently based on this hereditary information or that hereditary facto
associated with a particular group of the population. A second concern rais
Rosenzweig and Zhang (2009) when studying the impact of fertility o
investment refers to the possibility that parents might allocate resou
compensate (reinforce) an endowment shock. In fact, among twins and
order multiple-birth children - for example, triplets and quadruplets - rates
birth weight and infant mortality are 4 to 33 times higher compared with
births. Moreover, twins and other higher-order multiple births are more
suffer lifelong disabilities if they survive (Martin and Park 1999). Thus, m
(parents) might react by allocating fewer hours to the labor market in order
more time with their children, or they could potentially increase their labor
provide funds to compensate for the negative endowment shock. Either hyp
case would invalidate the exclusion restriction due to an impact of multipl
beyond the channel of fertility. Ideally, one should address this prob
controlling for a measure of child endowment, as Rosenzweig and Zhang (2
However, DHS data do not have an appropriate endowment measure.5 Ther
despite the fact that the second assumption is nontestable, the random na
multiple births, the use of a sample of developing countries, the choic
observational unit under analysis, the inclusion of other variables that are c
with the incidence of multiple births (e.g., mother's age and education), an
analysis of the impact of twinning in a specific birth ( s ) make it more likely
assumption holds.
The impact of family size on mother's labor attachment, as presented in E
constant across observations, although this assumption may be unrealistic g
obvious heterogeneity in household's preference, production technolo

4 For the United States, the American Society for Reproductive Medicine (Martin and Park 1999
such factors. First, the incidence is higher among African Americans. Second, women who are
fraternal twins give birth to twins at a rate of 1 set per 60 births, which is higher than the nation
set per 90 births. Third, women aged 35^0 with four or more children are three times more lik
twins than women younger than 20 without children. Finally, multiple births are more comm
women who use fertility medication (Martin and Park 1999).
Rosenzweig and Zhang (2009) used child birth weight as proxy of child endowment. For some
year samples, DHS data have information on birth weight for some of the children in the hou
limitation in the sample and the high data requirements (sample size) when using multiple birth
of variation make this alternative not viable. Although I cannot directly address this concern
analysis, as a second best alternative and using auxiliary data constructed from census data
developing countries, I check the robustness of the impact of fertility on mother's labor employ
addition of a measure of child endowment. Specifically, I use disability status as a proxy
endowment, and I check the sensitivity of the estimates to the inclusion of this variable. The result
analysis support the robustness of the results, and they can be found in a working paper
downloaded from the author's webpage (http://sites.google.com/site/caceresjulio/home/research

Springer

This content downloaded from 202.92.128.169 on Thu, 01 Sep 2016 13:17:07 UTC
All use subject to http://about.jstor.org/terms
156 J. Cceres-Delpiano

constraints.6 Extensive liter


of addressing this heterog
Imbens (1994) have show
treatment effects (LATE) in
whose actions take this h
impact of an increase in fam
of multiple births had mo
Imbens and Angrist point
used. Nevertheless, when m
compliance,8 LATE can be in
(Angrist et al. 2010). That
mothers who wanted a spe
nevertheless, were pushed t
"nontreated"). Therefore,
families who sought an 5 p
quadruplets, and so on) addi
(those with larger family
benefits of family program

Data and Variables

The primary data source in the analysis is the Demographic and Health Surveys
(DHS; ICF Macro 2009). These surveys are nationally representative househ
surveys that provide data for a wide range of monitoring and impact evaluatio
indicators in the areas of population, health, and nutrition. The sample in e
country-year is typically a stratified random sample of all noninstitutional ho
holds.9 The universe of the survey is mothers who were 15 to 49 years old at t
time of the interview. The analysis is restricted to 40 developing countries for wh
there is an average of two sample years (Table 6 in the appendix). The criterion f
selecting these countries and years is based on obtaining a sufficient sample size
ensuring that the key information to construct the sample and variables is availa
and consistently measured.
The sample is restricted to women who had one child or more at the time of t
interview, independent of the fact that a child has survived or still lives with the mot
By doing so, I avoid restricting the sample of mothers to a specific family arrangem
to a specific number of surviving children, or to a specific number of children still li

6 The use of the concept of heterogeneity as synonymous with selection is often observed in the emp
literature. Here, I make reference to the fact that the parameter y5 cannot be summarized by scalar r
with a complete distribution (Heckman et al. 2006).
When Y is homogenous, multiple births being a valid instrument (as well as any other valid instrume
will allow us to identify all the relevant parameters such as the average treatment effect (ATE), aver
treatment on the treated (ATT), or average treatment on the untreated (ATUT) because they all are t
same (Heckman et al. 2006).
An extended discussion can be found in the author's website (http://sites.google.com/site/caceresju
home/research).
Giving this sample design, all regressions in the present analysis incorporate sample weights. Also,
standard errors have been adjusted to allow for cluster correlation.

Springer

This content downloaded from 202.92.128.169 on Thu, 01 Sep 2016 13:17:07 UTC
All use subject to http://about.jstor.org/terms
Fertility and Mother's Employment in Developing Countries 157

with the mother because these variables could be outcomes of fertility. Furt
consider women aged 1 8 to 40 who had their first birth between age 1 5 and 35
so, I focus on women in the middle of their childbearing years and who star
reproductive life neither too early (earlier than age 15) nor extremely lat
35 years). To avoid the decision of household formation, I also restrict the
mothers whose oldest child is younger than 14 years.10
The mother's measure of fertility, nu is defined as the reported number of
ever born.11 This definition of fertility differs as well from total number of
living with their parents. Nevertheless, I do not restrict number of childre
living at home because the decision to live at home or the decision of par
place their children with relatives (or any other third party) can be s
outcome of the fertility decision. Cceres-Delpiano (forthcoming), using m
births as source of variation of family size, found a positive impact of fertil
probability that a child is not living with his/her parents (mother).
To characterize mothers' labor force attachments, two groups of variab
defined. The first group is composed of variables that capture exten
intensive margins in the mother's labor force attachment. For all the
mothers are asked for their current working status. Using that information, I
dummy variable called "working," which takes a value of 1 if a mother is c
working, and 0 otherwise. Also, for all country-year samples, I know the
status of a mother during the past 12 months. Using this information, I
"worked last year" as a dummy variable that takes a value of 1 if the mothe
during the previous 12 months, and 0 otherwise. The third variable in this
the usual number of days per week worked. This information is rep
individuals who worked during the previous year. For individuals who did n
during the previous year, a value of 0 is inputted.
The second group of variables aims to characterize the mother's emp
across four dimensions: location, compensation type, employer, and f
In each dimension, the omitted category is "not working." For the dimen
location and for part of the country-year sample, I define two var
"working at home" and "working away from home," dummy variables th
a value of 1 if the mother's job is at home or outside the home, and 0
otherwise. For the aspect of compensation, the variables "unpaid" and "paid in
cash" are dummy variables that take the value of 1 when a mother holds an
unpaid job or if she is paid in cash for her work, and 0 otherwise.12 For the
facet of dependence, I define two variables, "salaried job" and "self-employment,"
which are dummy variables that take the value of 1 when a mother is an employee
in a business or is self-employed, and 0 otherwise. Finally, in terms of frequency,
two dummy variables are defined: "full year" and "seasonal or occasional" take the

10 These three restrictions reduced the sample size, respectively, by 1.9%, 1.2%, and 38.6% for the sample
1+; by 1.1%, 1.1%, and 43.2% for the sample 2+; by 0.8%, 1.1%, and 52.9% for the sample 3+; and by
0.6%, 1.0%, and 60% for the sample 4+.
The same qualitative results are obtained when using number of surviving children as a measure of a
woman's fertility. I also used a measure of fertility adjusted by child spacing, and in general, the estimates
for all regressions tend to be bigger. I opted for the most conservative measure.
A paid job is not necessarily a job for which a mother is paid in cash. Many jobs at a subsistence level
are characterized by payment in kind or services.

Springer

This content downloaded from 202.92.128.169 on Thu, 01 Sep 2016 13:17:07 UTC
All use subject to http://about.jstor.org/terms
158 J. Cceres-Delpiano

value of 1 if a mother had


year, and 0 otherwise.
Tables 7, 8, and 9 in the ap
characterizing mother's lab
Table 7 presents the sample
sample according to urba
statistics are presented for t
(1+), two or more (2+), thr
Following Bronars and Gro
multiple births by exploitin
for each child a woman had
household have the same ag
to multiple births. Because
sufficient statistical power;
sections. Using the algorith
multiple births, of which
Finally, I estimate Eq. (1), u
variables by country of resi
education, and mother's lite
evidence that the number
(Cceres-Delpiano 2006, for
father's or spouse's chara
father's education).

Results

First Stage: Multiple Births and Number of Children Ever Born

As mentioned earlier, one of the two conditions necessary to ensure a valid


instrument is a sufficiently high correlation between multiple births and family size
so that an average difference in family size exists and can be measured properly.
Table 1 presents the impact of multiple births on the number of children ever born
for samples 1+, 2+, 3+, and 4+. Each coefficient in the table corresponds to the OLS
estimate of the impact of multiple births dummy variable in a regression with a
dependent variable of the number of children ever born, and the specific functional
form the same as the one represented by Eq. (1) but replacing child ever born for the
instrumental variable.

13 Two educational levels are considered: mothers with no formal education (approximately 40% of all
mothers), and mothers with some years of education.
To be consistent with the explanation of the estimated parameter, the sample must be defined such that sample 1+
contains families aiming for a first "birth," some of whom go on to have just one child and others of whom have
multiple births. In general, for sample s+, I keep families aiming for the s birth. Then, I restrict the sample to those
with no multiple births in pregnancies prior to the pregnancy for which I study the unexpected change in family
size. Specifically, I restrict sample 2+ to families that did not have multiple births in the first birth, but I restrict
sample 3+ to families that did not have multiple births in either the first or the second birth. The results are not
sensitive to this restriction because families who have two or more events of multiple births are extremely rare.

Springer

This content downloaded from 202.92.128.169 on Thu, 01 Sep 2016 13:17:07 UTC
All use subject to http://about.jstor.org/terms
Fertility and Mother's Employment in Developing Countries 159

Table 1 Impact of multiple births on number of children (first stage): OLS estimates

All Urban Status Education

Unconditional Conditional Urban Rural None Some

Sample (1) (2) (3) (4) (5) (6)


1+ 0.8962** 0.7964** 0.7586** 0.8176** 0.8318** 0.7929**

(0.0343) (0.0302) (0.0403) (0.0418) (0.0590) (0.0333)


Observations 433,674 433,674 175,746 257,928 140,976 292,698
2+ 0.9037** 0.8365** 0.8099** 0.8507** 0.8449** 0.8371**

(0.0286) (0.0250) (0.0348) (0.0330) (0.0436) (0.0289)


Observations 320,482 320,482 119,492 200,990 115,909 204,573
3+ 0.9418** 0.8834** 0.9464** 0.8555** 0.8548** 0.9218**

(0.0282) (0.0260) (0.0474) (0.0312) (0.0393) (0.0345)


Observations 193,316 193,316 62,165 131,151 82,820 110,496
4+ 0.9224** 0.8973** 0.9416** 0.8833** 0.9261** 0.8713**

(0.0313) (0.0298) (0.0539) (0.0352) (0.0412) (0.0423)


Observations 105,428 105,428 28,619 76,809 52,263 53,165

Notes: Robust standard errors are in parentheses. The specification is the one
exceptions that the dependent variable is number of children ever born,
covariates and that the coefficient reported is the impact of multiple births.
error corresponds to a different regression. Covariates in the model (columns
by country of residence, year, urban status, mother's age, mother's year
literacy status. 1+, 2+, 3+, and 4+ represent the samples of families with one,
children, respectively.
**/?<. 01

The first two columns in the Table 1 present the impact of multiple births without
(unconditional) and with (conditional) other covariates in the model, respectively.
Columns 3-6 present the conditional impact of multiple births for the subsamples
defined by urban status (columns 3 and 4) and by mother's education level (columns
5 and 6). The first two columns show that the impact of multiple births is robust to
the inclusion of other covariates in the model.15 This finding is important because it
reveals that at least based on these observed variables, multiple births are not
strongly correlated with other covariates, and the positive impact observed on the
number of surviving children is not driven by the correlation with the other
covariates. Second, across all subsamples, there is a positive and statistically
significant impact of multiple births (at 1% significance level), which reduces the
concern about the bias associated to weak instruments. Third, for both the full
sample and the subsamples (by urban status or mother's education), the impact of
multiple births is greater at higher births. This finding is consistent with the idea that the
event of multiple births is more likely to shift family size over the desired fertility for a
higher percentage of the population (compilers) at higher births.

15 The same robustness is observed for the subsamples defined by countiy-region, urban status, and
mother's education level. To save space, I did not include them in the table.

Springer

This content downloaded from 202.92.128.169 on Thu, 01 Sep 2016 13:17:07 UTC
All use subject to http://about.jstor.org/terms
160 J. Cceres-Delpiano

Finally, it is also worth no


line with those reporte
identification (multiple bi
although the analyses are c

Second Stage: Fertility an

Table 2 reports, for the f


(odd columns) and the IV
mother's employment. The
children ever born on the
extensive and intensive
definitions of mother's la
employer, and frequency.
seasonal or occasional em
estimates confirm the com
mother's employment, wh
occasional jobs, self-emplo
impact of an increase in th
When focusing on the tw
power associated with inst
errors and fewer statist
important differences can
Using multiple births a
children, and focusing f
confirm a negative impa
of mothers with two or
that a mother is curre
approximately three per
more children; this likelih
percentage points. In rel

16 Angrist et al. (2010), for Isra


depending on the sample conside
0.67 to 0.817. These point estimat
explained by the differences in t
completed fertility among less-
appendix), with 1972 as the media
al. (2010), the average birth coho
number of children is estimate
mothers born between 1948 and
impact of multiple births for indiv
a smaller group of compilers, wh
This positive sign for some of t
and the omitted variable bias asso
omitted variables, usually a con
independent of the factors respon
appendix) show that those indiv
capital (education) but are also lar
unpaid jobs, self-employment, or

Springer

This content downloaded from 202.92.128.169 on Thu, 01 Sep 2016 13:17:07 UTC
All use subject to http://about.jstor.org/terms
Fertility and Mother's Employment in Developing Countries 161
*

m < - -- cNNr-m
in m in <-) "3" m m ~- -
ff) Tj- Tt -H (N 5 fN
OOpOCNfN popp
^ , , , , ,

* * * * *
^ /-~s ^ ^ ^ ^ ^ ^ ^ /( ^ ^
h h in OO ON V IO in r- r-
OO -^ r^- ' ON ON ^ H rt (S TJ- -- ' <4 <N
OOOOO-H ^
Xf opop^p popp oppp
+ ^ O
^ + iwlwlw lwlw w I

*
)_- ^ ^ 4- ^ ^ ^-v * r-S
on - ' cN m oo on on < ocsr-o
<N 00 <N < t - ^ ro m ON t 0' t
i- i *- i i- i CO C4) (N - H in (N
pppOCN^ OOOO OOOO
*
g I w I w I w |W|W I w I w

* * * * *
* "V * ^ ^ *^ ^-N * ^ * ^T4
oo fN (N r- m m in on ^- on
ao ^ ^ ^ j4 t m - 1 tr m --
^'- oooo 2 2 n? 2
^ opop popp
+ hJ O o
CO O I w I w I w I w I W w , w

m r- no in r^. (Ni on - on
VO <- i OO >- 00 OO po in T}- o -H ON
h *- *- i -h i- i <4 <4 -
O O O O ^ ' - 1 oppp
^ OO ,

!
lEL
*
it /-s
*
& /-s
OON^J-ONONt^ Tj- ON in - OO ON rj-
OOt^-OONO
*
* ^ -

(N ^
*
^ ^1

(N ^
*
^ s ^

^
I en
^OOO^-H'-H
^
OOOO
popp

popp
<- i

5 4- ^ O
<3
I w I w I w I w I w W ( w
-


<-
*
0
t^- ? in oo ^ ON ~~- ^ co on -<
ON <N ON <N O NO ^^^ - *
1

H (S) H 1 ^ *- * y- I Tt (N CNOON
OOOON^ OOOO OOOO
D J> , ,

1
*3
I * * *

00 NO OO OO Tt M NO (N (N
* * *

TJ-ONO^HOO ON ^ *-
-0^00 OOOO O O (N
- <
Z +
Z + O
^ O
I w I w I w I w I
w w I w
w
1<+4
<+4
0
M
<u
c
g <u
.1 * s
<D

a
>

t b SS
2 88I ^
<4

<8
3 2 2 t
1
to
0 ^ ( JH
1
>% p >%
- 1 !
- "i
ce o p !
N - !s 13 e *2
.2*
3 -
i 2
i 2 -
I - 1
-i o 3oi 3
1 2
2 1
1 *2 ce
ce
&
^ - P o ou 3 o OD &H ce

Springer

This content downloaded from 202.92.128.169 on Thu, 01 Sep 2016 13:17:07 UTC
All use subject to http://about.jstor.org/terms
162 J. Cceres-Delpiano

> -g
r-c^ToS'^l-4 (N ^ " ^ ^ <
m m | 1 <
2
-H(NN^H^"(N ^ ^ (N
<N

> ~ ~
S f 8 f f
'S
f 3
> T3

59
* * * d +
* ^ ^ O ^
t4^ in -n?
OO ^ ^ ^ ^ _ ^ Cg _|_
OOOOOO - < O O O 0)<N
r OOOOOO ppop "
+ bJ U +
3 + l^|W|W Iw W U , ,-H
SS
s
S
ct ^
*3 ^ >%
^ *
1" h Om^HfN 2
) 1- 1- -H (N - ' <N t CD
^ - 1 <N - ' <N - < g -
OOOOOO oppp P-=s
> ^
, i- I- I-
e -
2
-S s
*
* ^
*^ *
^
$ la
^
^ ^ m lo
^^-^- rt - - - 1 ft4
-<00000 -< g
OOOOOO , +3
+ ^ , w w cg
+ l^lwlw I w w .2 3
,a 'o

nn 00
O

ONVOO'sOVOOO Tf 1 00 ^ 'S S
oo - >,
0-H-H-H(N-H <- <- >- 1 >,
poop ^ ,5
>> 4 <3
>> W I W I W I W I W
^i
<

* * * 43 Cd
* ^ - ^ ^ ^ ^ <D
m ^ - i ^ t4 ^ ^ 5 > ^
I^OOnO hw
- < - < 2 >
OOOOOO < 'S +3
+ jj w w w w OH
(N O I w w w I w -.Cd,
2 -
-
0 v-

ta <u
^ ^ ^ ^ .S2rt2
n 00 <n m r^mONO ca =3 'S
OiO'-Hr-HOO'O t - (N OO i-i O&O
1 - - < ^ ' >1
000 pop popp te a g
>> , *p
>> , w w w , w | w ^ *p ' g 0 t-,
' t-,
u

M
<D ^^ H
3
* * * * * <lT tS
^ ^ _.
^ <N - - ON - / _. *
OO O -^J- O (N o Os 00 ^ S
- i -
/5 OOOOOO ,;
4- ^ -5 <L>
- iw w w iw w 0 D D o
+3

.s*t

S1:3
1:3 s
e -
'S

'S "g - I s
43

gl'I
|2 |2"2
CL "2 -
> -
22 1 * 11l|i
* > - *
3


8 Sa
c
iS
!
?
vh
22
; 1% 1 * g 1% <2
SS 0 '/'I

N
R
o<L)
_b <u
o<L) o<L)
^ >>~ f
H*
H -t- <u
o --^a0
^ *7^
CDuce- I I ^ it >>~
-
O M ^ C

M 0 3 * Il I
-3

f2

Springer

This content downloaded from 202.92.128.169 on Thu, 01 Sep 2016 13:17:07 UTC
All use subject to http://about.jstor.org/terms
Fertility and Mother's Employment in Developing Countries 163

in the appendix, this change is approximately a 5% to 7% reductio


employment.
From the second group of variables, compared with general outcomes
employment, a negative impact of family size is less clear.18 Neverthe
elements are worth mentioning. First, as for general measures of female
family size does not have an effect on the different measures of female
for the sample facing a shift in family size in the second birth. Second, t
an unexpected increase in family size on female employment at a first birth
of mothers) is driven mainly by a reduction in unpaid jobs, jobs located
home, and seasonal or occasional jobs. These job types are the one
associated with more informality (unpaid jobs, and seasonal or occasion
coupled with factors that are not complementary with childbearing (w
from home or an unclear schedule in seasonal or occasional jobs). Third
parities - specifically, the 3+ sample of mothers - a negative impact of sh
size is found even among definitions of mother's employment that are as
more formality, such as jobs that are paid in cash or self-employment.
Comparing OLS and IV estimates, for all statistically significant IV coeffi
underestimates (in absolute value) the impact of an increase in the number
The sign of this bias is informative when compared against the ones reporte
studies. Specifically, Angrist and Evans (1998), among other U.S. studies, f
relation to the IV ones, OLS estimates overestimated (in absolute value) t
fertility on mother's labor attachment. The explanation that sometimes is
literature relies on the assumption that the variables omitted from Eq. (1) a
some abstract concept of ability}9 which increases the opportunity cost
home (i.e., increases the likelihood of employment); jointly, those
decided to have a bigger family, on average, belong to groups with low
ability. Thus, OLS estimates provide a joint estimate of the impact of fe
families who aim for a bigger family have, on average, lower ability, and
lower opportunity cost of staying at home. All the same, different assu
required to make the same conjecture about the omitted variable (abilit
covariance with fertility decisions when I try to rationalize the findings. In
introducing home production in the model, with ability being an input
production, would allow an increase in ability to decrease the opportun
home production at higher levels of ability.20 Therefore, the OLS estim
underestimate the impact of a fertility change, since the impact of

One must be careful of reading these findings as evidence against a negative impact of c
mother's labor engagement. In addition to the loss of power associated with the use of
variables, by defining narrow outcomes for female employment, I define a smaller group
making the estimates noisier.
Agero and Marks (2008) argued, for example, that women with high career-based unob
as "ambition or talent," might be the ones choosing to have smaller families and that thes
overrepresented in the labor market.
Formally, the plim(yo/j) = ys + p co v {ability, n) IV {n) with p as the impact of a
propensity to engage in labor market activities; and with a valid instrument, the plim(y IV)
while IV provides us with a consistent estimate of the negative impact of fertility, y, the O
asymptotic bias equal to p because cov(ability,n) < 0 and p ^ 0. Therefore, wi
production in the model, (p < 0), |plim(yo/J| < |plim(y/F)| = |y|.

Springer

This content downloaded from 202.92.128.169 on Thu, 01 Sep 2016 13:17:07 UTC
All use subject to http://about.jstor.org/terms
164 J. Cceres-Delpiano

discounted because families


face a lower cost of takin
explanation based on ability
differences between OLS an
a higher proportion of in
occasional-seasonal jobs.
Two types of heterogeneit
change in family size seem
births (1+ sample) or at high
of the distribution of fam
observed in relation to the k
births, families (mothers) w
informality or those that a
in family size at higher pa
better-quality jobs in terms
When employment is di
sions), a natural concern ar
To address this concern, I
employment. Although th
proper measure of formal
formal and informal jo
employment are used: "cu
employed, unpaid, or at-h
dummy variable for hold
informal," a dummy variab
or working for a relative d
formal," a dummy variab
and salaried job during th
First, the sample means f
formality and informali
the sample to families wit
hold a job that, accordin
Third, and most importan
female employment are co
estimated magnitude for
definitions of mother's em

1 This lack of agreement is related


an increasing proportion of unpro
Conference of Labour Statistician
employment (Jtting and Laigles
related to characteristics of the j
This proportion goes from app
Jtting and Laiglesia (2009) rep
America (1995-1999), and 70% f
marginally smaller than the ones I
I have not excluded jobs associ
countries are overrepresented in

Springer

This content downloaded from 202.92.128.169 on Thu, 01 Sep 2016 13:17:07 UTC
All use subject to http://about.jstor.org/terms
Fertility and Mother's Employment in Developing Countries 165

Table 3 IV estimates of the impact of number of children on measures of mother's forma


employment

Job Type Mean 1+ Mean 2+ Mean 3+ Mean 4+

Currently Informal Job 0.346 -0.0224 0.358 -0.0278 0.386 -0.0388* 0.417 -0.0352
(0.0194) (0.0193) (0.0178) (0.0268)
Last Year Informal Job 0.470 -0.0285* 0.489 -0.0268* 0.524 -0.02051 0.559 -0.0467**
(0.0121) (0.0115) (0.0122) (0.0156)
Current Formal Job 0.153 -0.0084 0.142 0.0056 0.129 -0.0019 0.112 0.0030

(0.0144) (0.0143) (0.0135) (0.0175)


Last Year Formal Job 0.159 -0.0007 0.147 0.0002 0.132 -0.0034 0.111 0.0267

(0.0162) (0.0148) (0.0134) (0.0207)

Notes: Robust standard errors are in parentheses. The specificati


coefficient reported comes from a different regression. Covariates
country of residence, year, urban/rural status, mother's age, moth
literacy status. 1+, 2+, 3+, and 4+ represent the samples of families w
children, respectively.
V - *10; */? <.05; **p<. 01

impact is significant only for the models in which t


outcome of informal employment. Specifically, I
increase in family size reduces female informal emp
percentage points for sample 1+, and by up to 4.7 pe
which corresponds to an impact that varies from
baseline means, respectively.
These types of heterogeneity are helpful when read
recent findings of the non-impact of different defini
"paid" employment. Given the source of variation use
the impact of fertility at lower parities. At lower p
sacrifice unpaid jobs rather than paid jobs (as
employment), which given the evidence provided in
leave only at higher-parity births (larger family size)
The usual question faced when pooling 40 coun
informative these estimates are based on such a hete
Even when including fixed effects by country, the
average over "compliers" in different countries. Idea
children should be estimated separately for each
subsequent cost in power associated with smaller sam
divide the 40 countries into three more homogen
countries, African countries, and India. Table 4 presen
of number of children on the outcomes "working
expected, I obtain larger standard errors and fewer si
less, almost all the estimates are still negative, and for
estimates are still statistically significant. While for L
a significant negative impact for the sample of famili

Springer

This content downloaded from 202.92.128.169 on Thu, 01 Sep 2016 13:17:07 UTC
All use subject to http://about.jstor.org/terms
166 J. Cceres-Delpiano

Table 4 IV estimates of the impac

1+ 2+ 3+ 4+

Latin America

Firststage 0.8271** 0.7855** 0.7894** 0.9017**


(0.0733) (0.0632) (0.0754) (0.0359)
Currently working -0.0951** -0.0427 -0.0415 -0.0313
(0.0319) (0.0358) (0.0426) (0.0516)
Worked last year -0.0796* -0.025 -0.0062 -0.0198
(0.0349) (0.0361) (0.0429) (0.0547)
Africa

Firststage 0.8162** 0.8458** 0.9161** 0.9017**


(0.0414) (0.0334) (0.0326) (0.0359)
Currently working -0.0058 -0.0114 -0.0263* -0.0471**
(0.0152) (0.0138) (0.0133) (0.0170)
Worked last year -0.0084 -0.0222 -0.0178 -0.0535**
(0.0148) (0.0138) (0.0132) (0.0175)
India

Firststage 0.6902** 0.6810** 0.8813** 0.9504**


(0.0944) (0.0908) (0.1324) (0.2950)
Currently working -0.1053 0.0176 -0.0231 -0.0806
(0.0691) (0.0829) (0.0859) (0.0984)
Worked last year -0.1118 0.0738 -0.0719 -0.0080
(0.0693) (0.0791) (0.0859) (0.0963)

Notes : Robust standard errors are in parentheses. The specificat


coefficient reported comes from a different regression. Covariat
country of residence, year, urban/rural status, mother's age, mo
literacy status. 1+, 2+, 3+, and 4+ represent the samples of familie
children, respectively.
*/?<.05; **p<. 01

for the sample of African countries, I find a signi


families with four or more births (4+).

Heterogeneity Analysis: Education, and Urban Statu

Table 5 presents the findings when the samples are d


and mother's education, respectively. The inter
heterogeneity has different motivations. First, the inf
of a shock of fertility on formal or informal employm
different than that in rural areas. In rural areas, with
a subsistence level and with a higher percentage of
activities, evidence of an impact of fertility on an inf
does not necessarily indicate a change in the amount
in urban areas. Second, in terms of education, theory

Springer

This content downloaded from 202.92.128.169 on Thu, 01 Sep 2016 13:17:07 UTC
All use subject to http://about.jstor.org/terms
Fertility and Mother's Employment in Developing Countries 167

I"1 " ^ ^
Or-H^OrncNNom ^
-
+ OOO S &
_ _ _ _ I i
hfNhO^^hn >
^^^- on
ON-Hr-HCS-H-H-H
w~
+ g +
X> +
_ D CN
-5 ^-(-nmOro^Tt^ _
_ 1-^<(^^ bp I
ooooooog
I
+ OOOo |
<L>
>

fe
-2
/- N ^ 1/3

Tt40rMiO-Hm^noo ; 43 >,
_o
od
^ OOrtooOr-.'ONOs ; 43 * >,
^ rONcsfS^HCSO^ cd cd
0
3
to poooooqp vh

D
03 +
t-
w g +5
1 lw w WIW,2'~H
tp
X C/3 Ve
* DD


OO^MOOOO^OIOO *3
^1^4040000'0

1

^j-'^-'-
- Jn
+ O Mg

* I w I w I w I w S
1 *^V ^ Po ^
f-

^400f0400NTj-Tt ^ -
0<Nir>Ttcnr-i040 0 od
. <O<no^O-
oooooooo i m -
0 g od
o o

s + f f ff
g

I ^ _ So
2
D
"S g OOOOOOOO rS^4
X <s + ?f f ef ^
sa
^ w
H * ^ 4_ ^ +- ^-s ^HO
Z d
d (Nm^OOfNTj-^OO
OOTj-rj-^040^t $

JO
I Z
.s g>
W

ce 4- OOOOOOOO Cd

D
i4 J -H I W I w I w I w jj ^
/3

s
Sfffgsg
8S?=>g8S?=>g
s . s- .f !
I 4 T s? s? s . 2

O
C/3
D


O
cONO-hOLIN^
cONO-hOLIN^ S i? t
OOOOO^O ?5ts S
sl||*3

+ oooo?? I 1 s
O

^ ^ ^ ^ 11
O
43
D ^ONONOONmOO -
o ^O^cor-oom/-}
(NCSrH^M^^rH fe cd
cdfe

cd sL fa
13
^ ^I w
+I w w w w
ce

<N I w I w p g tH
0 " . -
cd d
.* ^ ^ ^ ^ ^ ^
- - i i/~) ' - D .."
1 D '-Hfni^ONO'-iUO ^
M D /l(Nr4rHrt(SfflH -^
M -g qppqqqpo g ^
I i ?sfe?e?s &
<4-1

e .s s
f I 'i
5
d
SS g J 12
12
<+
S S 6|'5 S

D
s 1 1 1
II ni sir l
cd

J
^ sir GG<fH
^ <fH

D
g 3 3 V
D N -Oft 3 Cl
N A O D - *
d5 ^
J_ vh ^ Cd fe
o Cd & -
n D - i fe
-

S
-i g
p -i Ss ,5 ^S
,5 s ^ OOD^sS
as^ OOD^s 'S ,

*} Springer

This content downloaded from 202.92.128.169 on Thu, 01 Sep 2016 13:17:07 UTC
All use subject to http://about.jstor.org/terms
1 68 J. Cceres-Delpiano

education, individuals fa
Nevertheless, to explain th
to IV estimates in Table 2,
higher levels of education s
positively correlated with
with no (zero) years of edu
To simplify the exposition
focus on three outcomes
section: "worked last year,
2SLS estimates are present
The results in Table 5 con
fertility mainly affects info
the magnitude of the coef
outcome measuring inform
are statistically significant
mothers living in rural ar
likelihood that a mother he
Focusing on the impact of
heterogeneity analysis by ur
stronger among mothers li
informality in rural areas is
a shock in fertility, the degr
jobs that provide less protect
with my explanation about
impact is stronger among m
mothers with higher level
production, the arrival of c
especially among mothers w
market substitute for mothe
Voicu (2009), using data f
particularly stronger among

23 The same conclusions are obtain


formal." The results are available i
caceresjulio/home/research).
This result should be read jointly
during the past 12 months, and
Therefore, the results speak for a
for this sample of mothers.
25 Nevertheless, they found that mo
is, by moving from full- to part-ti
likely to stop working. Therefore
effects associated with an increase i
stronger preferences for time-inten
of child quality, and a higher margi
production of child quality.

Springer

This content downloaded from 202.92.128.169 on Thu, 01 Sep 2016 13:17:07 UTC
All use subject to http://about.jstor.org/terms
Fertility and Mother's Employment in Developing Countries 169

A natural question that arises is why mothers - and specifically, moth


higher education levels - would "need" a shift in family size to red
chances of being employed in these informal jobs that we think o
protected type of employment. One way of rationalizing these find
look at the payment of a job as a bundle of services: for example,
schedule flexibility, social status, security, and peer characteristics.
increase in family size would increase not only the cost of time in
activities but also the attractiveness of some jobs (scheme of payments) t
compatible with larger family sizes versus smaller families. Consistent w
hypothesis, Felfe (forthcoming) found that German mothers are willing t
a significant fraction of their wage to reduce hazardous working conditi
of their wage) and to enjoy a more flexible working schedule (44% to 56%
wages).

Conclusions

By using the Demographic and Health Survey (DHS) data for 40 developing
countries, I studied the impact of fertility on mother's employment. To solve
the problem of omitted variable bias, I used multiple births as a source of
variation in family size. As with previous evidence for developed countries, the
findings reveal that family size has a negative impact on female employment.
Nevertheless, two types of heterogeneity are exposed. First, the size of the
impact depends on the birth at which the increase in family size is studied.
Specifically, a negative impact of a shift in fertility is observed at lower or
higher parities. Second, the job types affected by a change of fertility are those
that are more informal, such as self-employment or unpaid jobs. Thus, an
unexpected change in family size affects jobs that probably are harder to
combine with childbearing (working away from home or in seasonal jobs with
an unclear schedule) and jobs that do not compensate a wealth shock coming
from an increase in family size (unpaid jobs). Finally, the heterogeneity analysis
reveals that an unexpected change in fertility is stronger at higher education
levels of the mother and also in urban areas.

Acknowledgments I would like to thank Cesar Alonso, Miguel Delgado, Eugenio Giolito,
Philippe Gagnepain, Ernesto Villanueva, and two anonymous referees for helpful comments and
discussion. Finally, I also want to thank the participants at the 23rd Annual Congress of the
European Society for Population Economics Society (June 11-13, 2009, Seville, Spain) for their
comments and advice. Financial support from the Spanish Ministry of Education (Grants BEC2006-
05710 and SEJ2005-8462) and from the European Commission (MRTN-CT-2003 -50496) are
gratefully acknowledged. A previous version of this article circulated under the title "Keeping the
Best for Last. Impact of Fertility on Mother's Employment. Evidence from Developing Countries."
The usual disclaimers apply.

Springer

This content downloaded from 202.92.128.169 on Thu, 01 Sep 2016 13:17:07 UTC
All use subject to http://about.jstor.org/terms
170 J. Cceres-Delpiano

Appendix

Table 6 Countries and years considered in the analysis

Country Year Country Year

1 Peru 1996, 2000 21 Niger 1998, 2006


2 Guatemala 1995, 1998 22 Nigeria 1999, 2003
3 Colombia 1995, 2000, 2005 23 Philippines 1998, 2003
4 Bolivia 1994, 1998, 2003 24 Rwanda 2000, 2005
5 Nicaragua 1998, 2001 25 Senegal 2005
6 Dominic Republic 1996, 1999, 2002 26 Togo 1998
7 Brazil 1996 27 Uganda 1995, 2001, 2006
8 Haiti 2000, 2005 28 Zambia 1996, 2002
9 Honduras 2005 29 Zimbabwe 1994, 1999
10 Bangladesh 1994,1997,2000 30 Burkina Faso 2003
11 Cameroon 2004 31 Benin 1996,2001,2006
12 CAR 1995 32 South Africa 1998

13 Cte d'Ivoire 1994 33 Chad 1997, 2004


14 Ghana 1998, 2003 34 Congo 2005
15 Indonesia 2003 35 Mozambique 1997, 2003
16 Kenya 1998, 2003 36 Cambodia 2000, 2005
17 Madagascar 1997, 2004 37 Ethiopia 2000, 2005
18 Malawi 2000, 2004 38 Guinea 2005
19 Mali 1996, 2001, 2006 39 Lesotho 2004
20 Namibia 2000 40 India 1999, 2006

Table 7 Descriptive statistics for the full sample

1+ 2+ 3+ 4+

General

Currently working 0.516 0.524 0.545 0.564


Worked last year 0.578 0.584 0.604 0.621
Usual number of days per week 3.137 3.122 3.146 3.146
(2.938) (2.944) (2.956) (2.963)
Location

Current job at home 0.134 0.138 0.142 0.147


Current job away from home 0.364 0.362 0.373 0.382
Type of Payment
Unpaid 0.065 0.069 0.075 0.081
Paid in cash 0.414 0.406 0.404 0.399

Employer
Current salaried job 0.184 0.172 0.160 0.141

Springer

This content downloaded from 202.92.128.169 on Thu, 01 Sep 2016 13:17:07 UTC
All use subject to http://about.jstor.org/terms
Fertility and Mother's Employment in Developing Countries 171

Table 7 (continued)

1+ 2+ 3+ 4+

Last year self-employed 0.302 0.318 0.350 0.383


Currently self-employed 0.239 0.248 0.269 0.296
Frequency
Full year 0.341 0.339 0.342 0.340
Seasonal or occasional 0.230 0.240 0.262 0.284

Living in Urban Area 0.382 0.348 0.295 0.244


Mother's Years of Education 4.921 4.380 3.501 2.720

(4.667) (4.468) (4.032) (3.526)


Share of Mothers Illiterate 0.34 0.373 0.436 0.506

Mother's Age 27.445 28.365 29.362 30.232


(5.151) (4.825) (4.323) (3.866)
Number of Children Ever Born 2.695 3.244 4.029 4.851

(1.539) (1.370) (1.208) (1.045)

Notes: Standard deviations are in parentheses. The standard devia

Table 8 Descriptive statistics by urban status

Urban Rural

1+ 2+ 3+ 4+ 1+ 2+ 3+ 4+

General

Currently working 0.489 0.486 0.499 0.512 0.533 0.545 0.564 0.580
Worked last year 0.553 0.545 0.555 0.563 0.594 0.605 0.625 0.640
Usual number of days per week 3.184 3.130 3.133 3.075 3.097 3.115 3.153 3.179
(2.965) (2.976) (3.001) (3.015) (2.914) (2.919) (2.928) (2.938)
Location

Current job at home 0.138 0.145 0.156 0.165 0.131 0.133 0.136 0.140
Current job away from home 0.364 0.347 0.344 0.344 0.365 0.370 0.386 0.396
Type of Payment
Unpaid 0.017 0.017 0.019 0.020 0.095 0.098 0.101 0.103
Paid in cash 0.499 0.486 0.487 0.489 0.358 0.360 0.365 0.366

Employer
Current salaried job 0.240 0.216 0.193 0.168 0.148 0.148 0.145 0.131
Last year self-employed 0.265 0.285 0.323 0.365 0.324 0.335 0.361 0.388
Currently self-employed 0.238 0.252 0.281 0.314 0.240 0.245 0.264 0.290
Frequency
Full year 0.387 0.382 0.386 0.386 0.314 0.317 0.323 0.325
Seasonal or occasional 0.141 0.143 0.157 0.173 0.283 0.291 0.305 0.320

Mother's Years of Education 7.507 6.939 5.856 4.660 3.321 3.016 2.514 2.094

(4.677) (4.659) (4.492) (4.163) (3.876) (3.699) (3.364) (3.041)

Springer

This content downloaded from 202.92.128.169 on Thu, 01 Sep 2016 13:17:07 UTC
All use subject to http://about.jstor.org/terms
172 J. Cceres-Delpiano

Table 8 (continued)

Urban Rural

1+ 2+ 3+ 4+ 1+ 2+ 3+ 4+

Share of Mothers Illiterate 0.160 0.184 0.240 0.321 0.451 0.474 0.518 0.566

Mother's Age 28.173 29.254 30.129 30.730 26.995 27.890 29.040 30.071
(5.236) (4.863) (4.330) (3.824) (5.045) (4.737) (4.279) (3.866)
Number of Children Ever Born 2.324 2.924 3.774 4.677 2.924 3.414 4.135 4.907

(1.323) (1.171) (1.057) (0.946) (1.617) (1.437) (1.251) (1.069)

Notes : Standard deviations are in parentheses. The standard deviation for proportion is not shown.

Table 9 Descriptive statistics by mother's education

None Some

1+ 2+ 3+ 4+ 1+ 2+ 3+ 4+

General

Currently working 0.552 0.556 0.564 0.552 0.498 0.505 0.559 0.559
Worked last year 0.610 0.614 0.623 0.610 0.562 0.566 0.617 0.617
Usual number of days per week 3.226 3.199 3.207 3.226 3.107 3.092 3.128 3.128
(2.967) (2.975) (2.980) (2.967) (2.927) (2.931) (2.940) (2.940)
Location

Current job at home 0.117 0.120 0.123 0.117 0.143 0.149 0.166 0.166
Current job away from home 0.399 0.399 0.403 0.399 0.345 0.339 0.362 0.362
Type of Payment
Unpaid 0.101 0.103 0.104 0.101 0.045 0.048 0.057 0.057
Paid in cash 0.378 0.379 0.379 0.378 0.433 0.423 0.424 0.424

Employer
Current salaried job 0.155 0.156 0.151 0.155 0.200 0.183 0.147 0.147
Last year self-employed 0.345 0.348 0.359 0.345 0.279 0.299 0.389 0.389
Currently self-employed 0.244 0.243 0.255 0.244 0.237 0.251 0.320 0.320
Frequency
Full year 0.315 0.316 0.318 0.315 0.355 0.354 0.367 0.367
Seasonal or occasional 0.304 0.307 0.313 0.304 0.191 0.199 0.247 0.247

Living in Urban Area 0.164 0.159 0.152 0.164 0.494 0.462 0.348 0.348
(0.000) (0.000) (0.000) (0.000) (3.759) (3.656) (3.070) (3.070)
Share of Mothers Illiterate 0.815 0.814 0.817 0.815 0.095 0.106 0.160 0.160

Mother's Age 27.02 27.762 28.877 27.02 27.664 28.73 30.595 30.595
(5.019) (4.705) (4.247) (5.019) (5.204) (4.860) (3.834) (3.834)
Number of Children Ever Born 3.210 3.636 4.260 3.210 2.430 3.006 4.729 4.729

(1.686) (1.494) (1.287) (1.686) (1.385) (1.230) (0.983) (0.983)

Notes: Standard deviations are in parentheses. The standard deviation for proportion is not shown,

Springer

This content downloaded from 202.92.128.169 on Thu, 01 Sep 2016 13:17:07 UTC
All use subject to http://about.jstor.org/terms
Fertility and Mother's Employment in Developing Countries 173

Table 10 Frequency of multiple births for the complete sample of children

Type of Birth Frequency %

Singletons 1,135,219 97.97


Twins 23,180 2.00
Triplets 351 0.03
Quadruplets 4 0.00
Total 1,158,754 100

References

Agero, J. M., & Marks, M. (2008). Motherhood and female labor force participation: Evidence from
infertility shocks. American Economic Review: Papers and Proceedings, 98, 500-504.
Angrist, J., & Evans, W. (1998). Children and their parents' labor supply: Evidence from exogenous
variation in family size. American Economic Review, 88, 450-477.
Angrist, J., & Imbens, G. (1994). Identification and estimation of local average treatment effects.
Econometrica, 62, 467-476.
Angrist, J., Lavy, V., & Schlosser, A. (2010). Multiple experiments for the causal link between the
quantity and quality of children. Journal of Labor Economics, 28, 773-823.
Black, S., Devereux, P., & Salvanes, K. (2005). The more the merrier? The effect of family composition
on children's education. Quarterly Journal of Economics, 120, 669-700.
Blau, F., & Grossberg, A. (1992). Maternal labor supply and children's cognitive development. Review of
Economic and Statistics, 74, 474-481.
Blunch, N., Canagarajah, S., & Raju, D. (2001). The informal sector revisited : A synthesis across space
and time (Social Protection Discussion Paper Series, No. 0119). Washington, DC: World Bank.
Bronars, S., & Grogger, J. (1994). The economic consequences of unwed motherhood: Using twin births
as a natural experiment. American Economic Review, 84, 1141-1156.
Browning, M. (1992). Children and household economic behavior. Journal of Economic Literature, 30,
1434-1475.
Cceres-Delpiano, J. (2006). The impacts of family size on investment in child quality. Journal of Human
Resources, 41, 738-754.
Cceres-Delpiano, J. (Forthcoming). Impacts of family size on the family as a whole: Evidence from the
developing world. The B.E. Journal of Economic Analysis & Policy.
Cristia, J. (2008). The effect of a first child on female labor supply: Evidence from women seeking fertility
services. Journal of Human Resources, 43, 487-510.
Cruces, G., & Galiani, S. (2007). Fertility and female labor supply in Latin America: New causal
evidence. Labor Economics, 14, 565-573.
Felfe, A. C. (Forthcoming). The willingness to pay for job amenities: Evidence from mother's return to
work. Industrial and Labor Relations Review.
Goldin, . (1990). Understanding the gender gap. New York: Oxford University Press.
Heckman, J., Urzua, S., & Vytlail, E. (2006). Understanding instrumental variables in models with
essential heterogeneity. The Review of Economics and Statistics, 88, 389-432.
Hotz, V. J., McElroy, S., & Sanders, S. (2005). Teenage childbearing and its life cycle consequences:
Exploiting a natural experiment. Journal of Human Resources, 40, 683-715.
ICF Macro (2009). MEASURE DHS. Calverton, MD: ICF Macro. Retrieved from http://www.measuredhs.com
Jacobsen, J., Pearce, J. W., III, & Rosenbloom, J. (1999). The effects of childbearing on married women
labor supply and earnings: Using twin births as a natural experiment. Journal of Human Resources,
34 449-474.
Jtting, J., & de Laiglesia, J. R. (Eds.). (2009). Is informal normal? Towards more and better job
developing countries. Paris, France: OECD.
Korenman, S., & Neumark, D. (1992). Marriage, motherhood, and wages. Journal of Human Resou
27, 233-255.

Springer

This content downloaded from 202.92.128.169 on Thu, 01 Sep 2016 13:17:07 UTC
All use subject to http://about.jstor.org/terms
174 J. Cceres-Delpiano

Mammen, ., & Paxson, . (2000


Perspectives, 14, 141-164.
Martin, J. A., & Park, M. M. (1999
Reports Vol. 47 No. 24). Hyattsv
Rosenzweig, M. R., & Wolpin, .
household models. Journal of P
Rosenzweig, M. R., & Zhang, J
investment? Twins, birthweigh
1149-1174.

Schultz, T. P. (2007). Population policies, fertility, women's human capital, and child quality (IZA
Discussion Paper No. 2815). Bonn, Germany: Institute for the Study of Labor.
Troske, K. R., & Voicu, A. (2009). The effect of children on the level of labor market involvement of
married women: What is the role of education? (IZA Discussion Paper No. 4074). Bonn, Germany:
Institute for the Study of Labor.

Springer

This content downloaded from 202.92.128.169 on Thu, 01 Sep 2016 13:17:07 UTC
All use subject to http://about.jstor.org/terms

You might also like