Little women,
stunted children:
Intergenerational
consequences
women’s status
in India
Diane Coffey
Princeton University
photo credit: Kyle Merrit Ludowitz
October 22, 2014
13 per
1000
neonatal mortality
& GDP per capita,
2005
causes of neonatal death, India 2005
low birth weight/prematurity
infection
birth asphyxia/trauma
congenital
tetanus
other causes
source: Million Deaths Study
two papers
1. what are the basic facts about maternal
nutrition in India?
just main
findings
• paper 1: pre-pregnancy body mass & weight gain
2. how does mom’s social status affect child
health?
in-depth • paper 2: women’s status & child height
(with Reetika Khera & Dean Spears)
how are papers 1 & 2 related?
1. maternal nutrition in India is a surprisingly
big problem
•
•
estimate pre-pregnancy body mass, weight gain during
pregnancy
make comparisons with sub-Saharan Africa
2. causes & consequences: social status seems to
lead to poor nutrition and poor child outcomes
•
•
show an effect of mother’s social status on child height
present evidence that poor maternal nutrition is an
important mechanism linking the two
paper 1:
pre-pregnancy body mass &
weight gain during pregnancy
in India & sub-Saharan Africa
research questions: what fraction of Indian
women are underweight before pregnancy,
how much weight do they gain during
pregnancy, and how does this compare to
poorer sub-Saharan Africa?
data: NFHS, 2005 (India)
29 DHS, 2000-2010 (sub-Saharan Africa)
why would we want to know the prevalence of
pre-pregnancy underweight?
can we just use the fraction underweight
among non-pregnant women as a proxy for
pre-pregnancy underweight?
this will be biased if women who get
pregnant are different from those
who don’t in ways that correlate with
nutrition...
age, fertility &
underweight
41.4%
fraction of pre-pregnant women
who are underweight (BMI<18.5)
(means and 95% CIs shown)
15.7%
weight gain during pregnancy
is remarkably low
(means and 95% CIs shown)
cumulative distribution functions of weight
for women at 3, 6, and 9+ months gestation
in India and sub-Saharan Africa
could differences in the status of women be
driving these differences in nutrition?
sex gaps in India are larger than in
sub-Saharan Africa.
education
work outside the home
sex ratios at birth
child mortality
child height
pre-pregnant
Indian women (41%)
age gradient among
Indian women 15-49:
15 percentage points
Indian
men,
35-50
(25%)
pre-pregnant
African women
(16%)
paper 2:
women’s status and children’s
height in India: evidence from
joint rural households
research questions:
no
can a mother’s social status
influence her children’s health?
if so, how?
why is this a hard question to answer?
measurement
• social status is often
measured in
“subjective” ways
• other measures suffer
from systematic
reporting bias
• education/education
gaps are not really
about status
omitted variables
• high social status
women live in different
households
• high social status
women have different
human capital
joint households in rural India
household heads
older brother
8% of rural
children under
five lived in joint
households with
two daughters in
law in 2005
younger brother
the children we study
joint households in rural India
• joint HH are characterized by patriarchy and
age hierarchy (Mandelbaum, 1948)
– older brothers have higher status than younger
brothers, wives inherit from husbands (Seymour, 1993)
• daughters in law are at the bottom of the
household hierarchy
– they do lots of physical work and their behavior is
highly regulated (Jeffrey, Jeffrey & Lyon, 1984)
how do daughters in law enact social rank?
• veiling
• lowering her gaze
• sitting on the floor
• remaining quiet in the
presence of senior men
and women
• eating last
www.nationalgeographic.com
photo credit: dinodia.com
our strategy
measurement
• use a woman’s rank in a joint
household, an observable
demographic fact based on
the age of her husband
omitted variables
• use joint household fixed
effects, so we’re looking at
differences between cousins
• large literature on rank in
joint households
• control for properties of
individual children and
nuclear families within joint
HH
• verify that differences in
status emerge from intrahousehold rank
• verify that results are not
driven by parents sorting into
household rank
three analyses
o one: do women married to the younger brother
really have lower status?
o literature review & empirical confirmation
o two: are children of lower ranked daughters in law
really shorter than their cousins ?
o rule out explanations other than social rank
o three: how does mother’s rank affect child height?
o it’s not health care
o maternal nutrition: differences in neonatal mortality,
body mass index, and birth weight (using new primary data)
one:
do women married to younger
brothers really have lower
status than women married to
older brothers?
literature on rank among daughters-in-law
• more people to whom a second daughter-inlaw must defer than a first daughter-in-law
(Mandelbaum, 2005)
• “senior wives tend to dominate young inmarrying wives” (Dyson & Moore, 1983)
• lower ranked daughters-in-law are most likely
to eat last from a common pot (Pariwala, 1994)
one: really lower ranking?
empirical confirmation of lower social rank
regressions in this presentation
• β is the coefficient of interest
• h is a joint household fixed effect
• Cih are controls, which change depending on
the specification
one: really lower ranking?
decision making authority
• does the woman have “final say” in decisions
related to:
•
•
•
•
•
her own health care?
large household purchases?
daily purchases?
visits to her relatives and friends?
what to do with the money her husband earns?
• subjective, but the household fixed effect helps
deal with some concerns
• regress an indicator for “say” in decisions on an
indicator for being the lower ranked woman using
household fixed effects
one: really lower ranking?
decision making authority
(NFHS 2005)
one: really lower ranking?
mobility: time spent outside
• prior literature has identified women’s mobility
outside the home as a measure of social status
(Rahman & Rao, 2004; Kabeer, 1999)
• we analyze data from the India Time Use Survey
• 1999 survey, all adults in 12,750 rural households in 6
states
• 1.2% of rural households interviewed (n=312) had two
daughters-in-law
• data on time use for the “typical” day before the survey
one: really lower ranking?
mobility: time spent outside
one: really lower ranking?
mobility: time spent outside
really lower ranking?
two:
no
are children of lower ranked
daughters in law shorter
than their cousins ?
is this really because of
mother’s social rank?
robustness checks
 direct effects of household size,
birth order, or other demographics
 pre-marriage sorting of mothers
into intra-household rank
 differences between older and
younger brothers (father differences)
two: really shorter due to mom’s rank?
in the same household, are
children of lower ranking
mothers shorter than children
of higher ranking mothers?
two: really shorter due to mom’s rank?
nonparametric comparison of children
of lower and higher ranking women
two: really shorter due to mom’s rank?
children’s height & mother’s rank
two: really shorter due to mom’s rank?
why control for child’s age?
two: really shorter due to mom’s rank?
children’s height & mother’s rank
two: really shorter due to mom’s rank?
robustness checks
 direct effects of household size,
birth order, or other demographics
 pre-marriage sorting of mothers
into intra-household rank
 differences between older and
younger brothers (father differences)
two: really shorter due to mom’s rank?
reasons to use demographic controls
maybe grandmothers prefer earlier born
grandchildren (or even the first born), regardless of
mothers’ status?
maybe having older cousins increases
children’s exposure to disease?
maybe low ranking daughters in law have
their children’s younger, on average?
two: really shorter due to mom’s rank?
demographic controls
•
•
•
•
•
first born to mother
single birth
mother’s age at birth
birth order in joint household
birth order in nuclear family (not shown)
two: really shorter due to mom’s rank?
children’s height & mother’s rank
two: really shorter due to mom’s rank?
robustness checks
 direct effects of household size,
birth order, or other demographics
 pre-marriage sorting of mothers
into intra-household rank
 differences between older and
younger brothers (father differences)
two: really shorter due to mom’s rank?
reasons to be concerned about
mother characteristics
maybe inferior women sort into being
lower ranked daughters-in-law?
if this is true, then they may be worse
mothers for reasons that do not have to
do with rank in a joint household.
two: really shorter due to mom’s rank?
do inferior women sort into being
lower ranked daughters-in-law?
maybe yes: rank among daughters-inlaw is common knowledge, parents
might take it into consideration
maybe no: marriage markets are thin,
lots of constraints to optimize
we can check!
two: really shorter due to mom’s rank?
mother characteristics
mother
comparisons
• height
• years of
education
• literacy
• age at marriage
regression
controls
• (mother’s)
height
• years of
education
• age at marriage
two: really shorter due to mom’s rank?
no differences on mothers’
pre-marriage characteristics (NFHS 2005)
two: really shorter due to mom’s rank?
children’s height & mother’s rank
two: really shorter due to mom’s rank?
height difference present for all
maternal heights
two: really shorter due to mom’s rank?
robustness checks

direct effects of household size, birth
order, or other demographics
 pre-marriage sorting of mothers into
intra-household rank
 differences between older and
younger brothers (father differences)
two: really shorter due to mom’s rank?
reasons to be concerned about
father characteristics
maybe younger brothers have worse
human capital in ways that affect their
ability to provide for nuclear families?
if this is true, then their children may be
shorter for reasons that do not have to do
with mom’s rank in the joint household.
two: really shorter due to mom’s rank?
father characteristics
brother comparisons
• height
• years of education
• literacy
• working at time of
survey
• working in a white
collar job
regression controls
• full sample
• years of education
• age at marriage
• select sample (not shown)
• (father’s) height
• employment
two: really shorter due to mom’s rank?
no differences between brothers
(NFHS 2005)
two: really shorter due to mom’s rank?
children’s height & mother’s rank
two: really shorter due to mom’s rank?
robustness checks
 direct effects of household size,
birth order, or other demographics
 pre-marriage sorting of mothers
into intra-household rank
 differences between older and
younger brothers (father differences)
two: really shorter due to mom’s rank?
three:
no
how does mother’s
intra-household rank
affect child height?
three: how does status affect height?
potential mechanisms
• health care inputs?
• maternal nutrition?
– maternal body mass
– neonatal mortality
– birth weight
• other things?
three: how does status affect height?
no differences in health care (NFHS 2005)
dependent variable:
(1)
prenatal
care
(2)
(3)
facility birth
health card
(4)
ever
vaccinated
Panel A: No controls except household FEs
lower rank
0.0588*
0.0697**
(0.0284)
(0.0250)
0.00366
h
(0.0248)
-0.0117
(0.0124)
Panel B: Household and year of birth FE
lower rank
0.0410
0.0665*
(0.0286)
(0.0261)
-0.0111
(0.0253)
-0.00384
(0.0126)
1076
last 3 births
1078
last 3 births
n (births since
2001)
sample:
788
last birth
1078
last 3 births
maternal nutrition: differences in maternal
BMI at time of survey (NFHS 2005)
h
three: how does status affect height?
maternal nutrition: differences in
early life mortality (NFHS 1993, 1999 & 2005)
• are children of lower-ranking daughters in law
more likely die in infancy?
IMR = NNM + PNM
(first year) = (first month) + (months 2-12)
• NNM is especially associated with birth weight
and maternal nutrition
three: how does status affect height?
the difference is due to neonatal mortality
lower rank
female child
year of birth FEs
household FEs
n
(1)
IMR
(2)
PNM
(3)
NNM
28.72+
(15.44)
-12.08
(12.72)


4.791
(10.51)
-5.561
(8.981)


26.43*
(11.03)
-10.04
(9.804)


3,227
3,095
3,703
three: how does status affect height?
birth
weights in a
district
hospital
low birth weight (<2.5 kgs)
Sitapur,
Uttar Pradesh
2013-14
(vaginal births)
birth weight by age
and mother’s rank
Sitapur, Uttar Pradesh
district hospital, 2013-14
birth weight by parity
and mother’s rank
summary of paper 2
• provided evidence that social status indeed
differs by a woman’s intra-household rank
• used a novel identification strategy to show
the children of lower ranking women are
shorter than their cousins
• provided evidence that one likely mechanism
leading from mother’s social rank to child
health is maternal nutrition
wrap up
• paper 2 sheds light on why paper 1 finds such poor
indicators of maternal nutrition
– daughters in law in joint households illuminate more
general social forces in Indian society
• worrisome implications for policy
– if poor maternal nutrition is importantly about low
social status, how can public actors promote the status
of young women?
– would it be better to focus on public goods?
• where my research is going next
– what are the impacts of poor sanitation on health?
– why does India have exceptionally poor sanitation?
thank you
photo credit: Gates Foundation
supplementary slides
photo credit: Gates Foundation
.25
where are the joint households?
.15
.2
UP
.1
BH
0
.05
RJ
MP
WB
MH
PJ KA
OR
HR
GJ
CHJH
KE
AP
AS
HP
JM
UC
TR TN
MN
GO
NA
DL
MZ
SK
AR
0
.05
.1
.15
.2
.25
percent of rural children under five in NFHS-3 data
background
comparison of households in our
sample with other rural households
background
comparison of children in our sample
with other rural children under five
background
not due to comparing children of
different birth orders in nuclear families
two: really shorter due to mom’s rank?
robustness check: larger sample
main results