Female Empowerment in Northern India: Effects of the Political Reservation System on Gender Bias
May 10th, 2008
Sze-chuan Suen [email protected]
314-853-1784
Under the direction of Dr. Anjini Kochar
The government of India has attempted to address the low status of women
in society though a constitutional amendment that mandates a woman must be elected as village leader every third election cycle. Several papers have shown that female reservations have significant effects on local policy decisions, but it is unknown whether the amendment is effective in its primary goal to erode discrimination against women. In my thesis I use uniquely-tailored household and village data to investigate whether political gender reservations can decrease sex bias by evaluating changes in female investments of those living in villages with female village leaders. Using OLS regressions, I find that only when a female village leader has financial resources are there changes in sex ratios, immunization rates, and school enrollment that are consistent with greater female bargaining power. However, I also find that maternal bargaining power is correlated to low sex ratios. These results show that political reservations for females must be supplemented with sound financial resources in order to increase women’s status and bargaining power, but doing so without changing maternal son-preference will not be effective in fixing the skewed sex ratio.
Key Words: India, female sarpanch, son preference, female empowerment The author would like to express her heartfelt appreciation for Dr. Anjini Kochar for investing so much time and energy in this project. This would not have been possible without her. She would also like to thank SCID, the Rai Foundation, and the wonderful translators from Delhi Business School, and all the other mentors who contributed to the project.
2
Section 1: Introduction
Male-skewed gender ratios in East and Southeast Asia have been extensively
documented, and 22 to 37 million females are estimated to be “missing” from India alone
(Mohan, 2005), with 930 females for every 1000 males (Borooah 2004); evidence from other
countries has shown that the gender imbalance can reasonably be assumed to be outside the
range of natural biological fluctuation (Sen 1990). Son-preference also extends beyond the
skewed sex ratio and into other socio-economic factors; there is a significant gender disparity
in health and education measures as well. The causes of son-preference in India are complex
and involve both economic and cultural factors. The method of son-selection is also hard to
accurately document, as a variety of methods, from abortion to childhood neglect, could be at
play.
While the sociological motivations for sex preference are unclear, the phenomenon’s
existence shows that in general, the social perceived value of women is smaller than that of
men. The elevation of the status of women in society, which would increase female
bargaining power and raise female worth, could therefore be a lynchpin for solving this
problem, especially since tackling the problem by limiting the means of selection and
propaganda campaigns have shown limited success (Dale 2006). One of the most
comprehensive and enforced policies to this end was the 73rd amendment in the Indian
constitution, which mandates that one third of all rural elected governmental positions must
be reserved for women randomly on a rotating basis. Through this act, theoretically, women
would gain a political voice, be empowered to address their concerns about their villages, and
occupy a highly-visible position in the political ladder that would erode the perception that
3
women should not be involved in “men’s roles,” like politics, outside the home. This could
result in the empowerment of women in society, where female labor might become more
desirable, and in the home, where they might be able to counter extreme son preference by
having a larger say in household resource allocations. Therefore the gender of the village
leader could affect son preference through a variety of methods.
However, this long chain of cause and effect has been incompletely documented.
There have been studies showing that female sarpanches (village leaders) make decisions that
better address the needs of women, and that the village perception of the effectiveness of a
female as sarpanch changes with the number of female sarpanches the village has been
exposed to, but whether the policy has been effective in empowering women in non-political
contexts has not been explored. In this paper, I ask whether the 73rd amendment has been
able to achieve its ultimate goal of reducing gender discrimination, and more generally,
whether a policy that seeks to empower women through political power can succeed. To do
this, I look at the revealed gender preferences of parents.
I measure changes in gender discrimination through changes in the sex ratio, health
and educational investment differences across sexes and examine whether these investment
patterns change after a female sarpanch takes office through the 73rd amendment. The study
is based in Haryana, which the 2001 Indian census documented as one of the most gender-
skewed states in India.
In section two I review literature which discusses gender preference, health and
education disparities between male and female children and provide background on
government action, the 73rd amendment, and past findings concerning the effects of the
policy. In section three I explain the theoretical framework motivating the question. Section
4
four provides the methodology used to answer the question at hand, then in section five
presents the results of the analysis and in section six I conclude. The appendix, section seven,
provides definitions, tables, and graphs referred to throughout the previous sections.
Section 2: Literature Review
Gender Preference and Sex Ratios
Son-preference could be an age-old attitude in India, but the advent of amniocentesis,
ultrasound, and safe abortion practices in the 1980’s provided a reliable method to realize
these preferences (Guilmoto 2007). Ten million girls have been estimated to be “missing”
since 1985 (Dale 2006). Gender ratios have become especially skewed in the north during
the last decade, where Haryana has reached a sex-ratio of 820 females to 1000 males in the 0-
6 age range, and Punjab 793 females to 1000 males (Bhat, 2006). These numbers are
unnaturally low – female-to-male ratios in countries such as Europe or the Americas usually
hovers around 1.05 females to one male, where the slight rise above parity is an artifact of
the larger mortality rate of men at every age. India’s sex ratio figures show strong evidence
of being humanly-generated rather than some natural variation in male health or birth rates;
the likelihood of having a son is much higher if the parents already have one or two
daughters and no sons (Jha, 2006).
Health and Education Disparities between Genders
Son preference has also affected other measures of child investment, namely health
and education. Sex selection through neglect may take the form of decreased post-natal care,
5
breastfeeding, parental surveillance, food allocation, spending less on healthcare, or
inadequate prevention from disease, as in not getting all the available vaccinations (Guilmoto
2007). Arokiasamy et al., 2004, finds that there are higher immunization coverage rates for
males than females in all states besides Goa and Kerala, and north-central Indian states,
where the gender ratios are the most skewed, show the largest bias. In cases where disease
has already been contracted, girls are likely to be taken to a health care center later than boys
(Sachar 1990), which decreases their likelihood of effective treatment – there is significantly
higher mortality rates for hospitalized girls than boys. This disparity extends to outpatient
care, where girls are less likely to receive routine outpatient and preventive care (Sachar
1990).
Similarly, girls are less likely to attend school or go to private schools (Wu 2007),
which are perceived to be of higher quality but have higher costs. In Rajasthan, 71% of the
students in secondary and senior-secondary schools in 2003 were male (Wu 2005), and the
literacy gap between genders can rise as high as over 32.5% in north western India (Office of
the Registrar General and Census Commissioner 2001). Thus both health and education
show heavy sex discrimination.
Background on Gender Bias: Motivations and Results
Potential causes of son-preference range from the economic to social. While
practices vary between different ethnic, religious, and geographic groups, Indian tradition
generally requires bride’s family pay a dowry to the groom upon marriage. The bride
becomes a part of the husband’s family thereafter, and any productivity she generates
contributes to her husband’s family income – women are quite economically dependent on
6
men, even to the extent of not usually inheriting a husband’s property should he die
(Dasgupta 1995). The economic reasons against having a daughter are exacerbated by the
lack of a pension system in much of rural India, since aging parents can only depend on their
sons for financial support. Social costs include any children who are born to the married
couple – they are part of the husband’s caste and family lineage (gothra) and not the wife’s.
Thus parents sometimes pay exorbitant amounts in land or jewelry for a daughter to get
married, and they also lose her labor and productivity, as well as any children she might have.
There are therefore limited incentives for her parents to invest in her education or health, or
even to have female children at all.
Sex selection methods make son-preference hard to identify on a household level, as
amniocentesis and ultrasound technologies have made it easy to discreetly abort unwanted
female children without attracting the censure of neighbors or village members. While such
methods are not available to many living in the rural countryside where families may lack the
facilities or the resources for such a procedure, sex selection is hypothesized to occur through
more subtle channels such as childhood neglect. However, this may happen with less
frequency, as female infanticide may lead to greater emotional costs for the parents.
Whatever the reason behind the phenomenon, the abundance of males leads to many
social problems as the cohort ages; projected conservative estimates show about 25 million
missing wives by 2030. This may lead to bride importation and trafficking, which is already
happening – the 1981 census revealed that 30% of Indians live somewhere besides their
birthplace, and 80% of these were women who had moved for marriage (Dasgupta, 1995).
Whether the move is voluntarily or not, marrying a long way from the maiden home can
decrease the bargaining power of the wife and further disempowering women. The female
7
deficit may also lead to increased demand for traditional women’s roles in the home, and
increase the pressure for women to not pursue a career or acquire education, which in turn
may lead to weakening of the bargaining power of women in the home and skew sex ratios
further. The lack of women would therefore exacerbate the further marginalization of
women in public life and limit the ability for them to make independent decisions.
All of the manifestations of son-preference mentioned above are generated in part by
the low level of female bargaining power within the home and society. With greater social
status would come greater freedoms in economic and cultural practices for women, thereby
lessening the costs and increasing the returns for potential parents of girls – with a greater
ability to determine household resource allocations, wives might choose to send income or
resources back to her parents or visit them more often to contribute to their household
through labor. The husband may also require a smaller dowry when they marry since she is
bringing a larger set of social assets to his house, and also thus reducing the cost of having a
daughter. Wives may also directly decrease the manifestation of son-preference with greater
bargaining power by choosing to keep more girls; several studies have shown that mothers
show less bias towards sons than fathers (Thomas 1990, 1994, Liu 2008, Songa 2006, Glick
2000).
The geographical distribution of son-preference severity can support the female
bargaining power framework as well as the hypothesis that low female investments are due to
low female economic returns. The two causes of low female investment positively reinforce
each other; with less bargaining power, women cannot generate high economic returns, and
the lack of financial contribution to the home reduces potential female bargaining power.
Thus it is reasonable to see that son-preference is roughly inversely correlated with female
8
labor participation rates, which is higher in southern Indian states (Raju 2005), and also
inversely correlated with female education rates, which reach up to a 30% gap in northern
states (Pal 2003). The driving force of these negative correlations does not seem to be overall
family wealth, which one might be expected to be positively related to female labor
participation and female education. Son preference manifests strongly in rich states --
Haryana, with the third highest per capita income in India, holds within it four of the ten
districts with the most skewed child sex ratios in India (Indian Census 2001). This could be
due to the relatively high price of the medical procedures needed to identify and remove
unwanted female children so only richer states can afford to implement sex selection so
pervasively.
Government Action
The Indian government is aware of the sex preference issue and has enacted a variety
of programs that attempt to reduce sex bias through both economic and bargaining power
motivations of son preference. Some are incentive programs like Ladli, a Haryana
government scheme where a savings account is started at a girl’s birth and Rs. 100,000 is
given to the girl at age 18 (Central Chronicle, 2008). Others attempt to increase the social
cost of son-preference through as propaganda, as in the popular slogan “save the girl child.”
A variety of NGO and governmentally supported women’s support groups have also sprung
up to run village-based programs which support mothers and children, like the childcare
centers known as Anganwadi centers that appear throughout India. These groups may
provide nucleation points where women’s communities can form so women can find power
9
in numbers or simply allow women to become more productive, and therefore independent,
by offering better health, sanitation, or educational opportunities.
Changing female bargaining power would be more difficult, and the government
recognized that the problem extended beyond the facilities and resources women had access
to; gender discrimination is also sustained by the lack of representation of female needs and
ideas in both village and household level decision-making. A cultural paradigm shift was
needed to overturn attitudes about female status and power. In an attempt to spark this
mental shift the government implemented sweeping political reforms in the form of the 73rd
amendment to the constitution, the most visible women’s empowerment initiative, where
every third term the democratically elected village leader position is to be reserved for a
woman. If this plan was successful in changing attitudes that devalued women in the
workplace and in the home, not only would females be empowered within the political
sphere, but it women’s bargaining power at home would be affected as well.
The 73rd Amendment and Political Reservation System
In 1992 the Indian government passed the 73rd amendment to the constitution, which
mandated a three tiered, devolved level of government in rural areas. Although this situation
had nominally been the case, non-federal government levels were not in full operation due to
the lack of regular elections, prolonged supersessions, an overly powerful central government,
and lack of financial resources (Indian Constitution 73rd Amendment, 1992). The primary
impact of the 73rd amendment was not simply to devolve more power to rural government,
however. It also mandated that the three sub-central levels of government -- the district (zilla
parishad), block (panchayat samiti), and village level (gram panchayat) -- were all to have a
10
quota for the scheduled castes (SC) and scheduled tribes (ST) in accordance with their
relative population in that voting area, of which one-third were to be reserved for women.
Furthermore, one-third of the total elected seats were to be reserved for women, including SC
and ST women (Singh 1994). The particular seats reserved for women were to be randomly
selected and rotated, and on average, every third term the panchayat president (the sarpanch,
at the gram panchayat level) would be a woman. SC, ST, and women were groups
traditionally believed to have low bargaining power and social status; this sea change was an
effort to introduce a mechanism by which these under represented demographics might be
able to thwart discrimination and achieve higher social and political status.
Village councils, or gram panchayats, vary in size with village population, and do not
have jurisdiction over urban areas. Each gram panchayat is composed of a number of
panches, or council members, and one sarpanch, or village president, all of whom are elected
by popular vote. All members of the gram panchayat are residents within the village. The
gram panchayat identifies the needs of the village, plays a limited role in allocating public
resources, implements development programs, and identifies below poverty line (BPL)
residents, who then qualify for certain public programs. This includes the construction of
local infrastructure, such as roads and drainage systems.
The 73rd Amendment gave the panchayats the power of taxation, and additional
funding for operations can come from the federal government, allocated through four general
programs. In addition, there is a drinking water funding scheme, welfare funding, and a
general gram panchayat operations fund. The gram panchayat also holds public land, or
shamlat land, in the village, and can rent it off or otherwise use it to generate income.
11
The changes made in the 73rd amendment were implemented starting in 1993, and
villages began obeying the women and SC and ST quotas in their 1995 cycle election. While
all villages held elections roughly every 5 years, the election year vary both inter- and intra-
village, with some villages holding elections in 1994 or 1996 and again in 2000.
With this constitutional amendment, India joined a growing number of countries that
legislated women’s participation in government. Worldwide, the number of laws mandating
women’s inclusion in the political process has increased drastically; as of 2003, there were
twenty-two countries with laws requiring all political parties to nominate a minimum
percentage (ranging from 20% to 50%) of women as candidates for national legislative office
(Baldex 2003). These laws have led to an eight percentage point increase in the number of
women elected to national governing bodies (Htun and Jones 2002). Therefore the laws are a
success to the extent that women are in government; however, are these women successful in
implementing real change? Much early anecdotal evidence points to women sarpanches
merely operating as proxies for their fathers or husbands, or independent female sarpanches
becoming victims of violence or ousted using manipulated votes of no-confidence or other
political maneuvers (McSweeney 2008).
Fortunately, there exists a body of past work in evaluating the success of the 73rd
amendment in regards to women’s empowerment. Chattopadhyay and Duflo (2004) have
found that contrary to the popular stories, substantial differences in expenditure patterns can
be attributed to the gender of the sarpanch. Women leaders invest more in public goods
more closely aligned with women’s concerns, which varied across region, and less in men’s
concerns (Chattopadhyay 2004). In southern India, less than 20% of women were found to
have been persuaded to run for the sarpanch position by their husbands, and were generally
12
wealthier, more politically experienced, and more politically knowledgeable than the average
woman (Rao 2008). Additionally, villages reserved for women leaders were found to have
less corruption (Duflo 2005).
In terms of attitude changes, there is even some evidence that mandated female
leadership might reduce prejudice against women in political positions. Beaman et. al (2008)
finds evidence that first-time women leaders receive worse evaluations from their
constituents than their male counterparts while second time females are rated equivalently.
However, residents of villages with female sarpanches report being less satisfied with public
goods than villages with male sarpanches, although they may have more of them and some of
this dissatisfaction stems from public goods outside the jurisdiction of the panchayat (Duflo
2005). Therefore it is yet unclear whether reservations for female sarpanches are able
successfully urge villagers to accept that politics can be a place for women, although not for
lack of studies on the subject.
To this date, however, there has been no exploration of whether greater political
representation of women are able to effectively change the status of women in society, the
primary goal of the 73rd amendment. While female leaders may be capable and can
effectively address women’s concerns in the village, they may still not be an effective tool in
empowering all women in non-political realms and changing gender discrimination. I
therefore ask directly in this study whether women sarpanches are effective at increasing the
value of women in general by looking at changes in bargaining power of women in the
household.
13
Section 3: Theoretical Framework
Women’s Bargaining Power Changes Resource Allocations to Children by Sex
The non-unitary model of bargaining power for intra-household resource allocation
(Thomas 1990) assumes that when the preferences of the husband and the wife differ on how
household resources will be allocated, the outcome will depend on their respective
bargaining powers, which can be determined by contributions to the household. These in
turn can be determined by a variety of factors, such as the magnitude of financial
contributions, labor (such as childcare or housekeeping), or some form of ability (like having
a political connection, social status, or knowledge). The past literature shows that there is
evidence to suggest that maternal resource allocation patterns differ from paternal ones
regarding childcare and investment; if so, bargaining power changes between parents,
induced by policies aimed at empowering women, the result would be reflected in resource
shifts for children by gender.
The non-unitary model of bargaining power has been empirically tested, and there is
much evidence in favor of it over the unitary model for inter-household resource allocation.
Thomas (1990) finds in Brazil that maternal unearned income (pensions, etc) has almost a
twenty times larger effect on child survival than paternal unearned income, and mother’s
education has a bigger impact on daughter’s height and weight for height measures while
father’s education has a larger affect on son’s (Thomas 1990, 1994).
Since then there have been a plethora of studies on the effects of parental bargaining
power on child health and education. A maternal head of household is likely to have taller
children than in a household with a male head (Liu, 2008). In West Africa, mother’s
14
education raises daughter’s schooling while father’s education raises school for both sexes
(Glick, 2000), while in China maternal education has a weaker effect on daughter’s
secondary school enrollment than paternal education does (Songa 2006) but the opposite is
true for primary school enrollment. These studies all show that bargaining power between
parents can and does have an effect on health and education investments of children by
gender, and therefore can change the degree of child sex bias in resource allocation in the
household. In general, these studies agree that mothers seem to show less son-preference, so
in increase in female bargaining power should result in a greater sex ratio and larger female
health and education investments.
Mechanisms by Which Female Leaders Empower Women
Greater political representation can directly empower women through voicing their
political concerns and helping them acquire village resources for their needs. Although these
direct changes are likely to be the main affect of the policy, female sarpanches can also
change female returns and bargaining power in a variety of indirect ways.
One is to directly do so by increasing the productivity of women in the village. If
female sarpanches efficiently address the needs of the village, it would publicly demonstrate
their ability to perform tasks in a public domain, and perhaps erode the social barriers for
women to serve the community in other non-traditional ways. The increased scope of
allowable “women’s work” would yield potentially better matches between the workers and
their work, thus generating higher returns for the whole community. Women would then
have a higher probability of serving in a visible position while receiving a wage, they may be
valued for their social and economic contribution to their family; the social return can extend
15
to both her husband’s family and her maiden home, if her maiden family is in the village.
This would justify increased investment in her from both her parents and her husband’s
family. Additionally, if female sarpanches enact programs that address women’s needs, then
the women of the village may be more productive by having better resources with which to
perform their work, be that through better infrastructure, health, or childcare, again
generating benefits for both parental and spouse families. The outside wage for women
would then increase with the rise in women’s productivity. In addition to increasing the
economic returns for women, this allows a higher threat point when bargaining in intra-
household resource allocation decisions.
This leads to the second mechanism female sarpanches might be able to change
female bargaining power -- through women’s social status and communal resources. In
“Bargaining and Gender Relations,” Bina Agarwal argues that rights in communal resources,
social support systems, and support from the state and nongovernmental organizations affect
the bargaining power of women within the household (Agarwal 1997). The introduction of
female sarpanches has the potential to change all three of these factors, as the sarpanch
obviously has the right to divide communal resources, and theoretically does so through the
support of the entire village. She may be able to introduce governmental programs or NGOs
into the village that support women’s rights in general. Bina Agarwal continues, saying that
notions about the division of labor, resources, and perceptions concerning the relative
contribution and needs of men and women also play a role in the range of bargaining power
for women. These, too, have the potential of being changed, as women are introduced into a
new field and would therefore have new contributions and needs. However, this may be
mitigated by what the Amartya Sen calls the “perceived contribution response,” where the
16
community perceives the contribution of the women to be lower than in reality, and so
continues to be undervalued.
Finally, the bargaining power of women in the village may increase due to the feeling
of self-worth that comes with successfully contributing to one’s community. Sen argues that
there may be psychological factors governing successful bargaining, and “the underdog
comes to accept the legitimacy of the unequal order and becomes and implicit accomplice”
(Sen 1990). However, if traditional gender roles are disrupted through the inclusion of
women into politics or increased productivity in women, women may come to expect more
from their communities and bargain more efficiently for decision-making power.
As seen in the previous section, a rise female bargaining power should decrease son-
preference, and female inclusion in local governance could theoretically achieve higher
bargaining power for women. However, there are some doubts as to whether female
sarpanches are able to achieve these goals. I now turn to examining that question in northern
India using parental investment in female children as an indicator of the perceived value of
females.
Section 4: Methodology
I ask whether the reservations for females in leadership positions are able to change
the perceived value of females in society through increasing female bargaining power. To do
this, I draw from three sources for my data. The first is a small subset of the Study for Fiscal
Decentralisation done collaboratively by the Centre for Research in Rural and Industrial
Development (CRRID) and the Stanford Center for International Development (SCID) in
17
2005-2006 in Pubjab, the second is village and household surveys of nineteen villages in
Haryana, India, and the third is the Indian Census -- town and village census data on the
villages of interest from the 1991 census, and state data from the 2001 census.
I first confirm that reservations for female sarpanches can successfully affect political
decision-making by observing that villages reserved for female sarpanches have different
panchayat expenditure patterns than those without. If female sarpanches are unable to
influence panchayat decisions, there is little possibility that they are able to affect child
investment measures. Female sarpanch efficacy has been shown before in Duflo’s work in
West Bengal and Rajasthan, and from the analysis presented below, I find that it holds true
for Punjab as well. Both Rajasthan and Punjab border on Haryana, and Punjab is Haryana’s
most demographically similar neighbor, so this offers substantial evidence that this finding
holds true in Haryana as well.
I then analyze whether female sarpanches are able to influence the status of females
by measuring parental investment in girls. The Haryana data, supplemented with the census
data, is used to identify whether the presence of female sarpanches are correlated with
behavioral changes in gender discrimination at the household level. I use the sex ratio and
gender differences in immunizations/prenatal checkups as a way to measure changes in
household gender preferences towards infants, and I use the gender differences in educational
quantity and quality to measure changes towards school-age children.
Econometric Methods
1. Effect of women sarpanches on panchayat expenditures
18
In order to find whether female sarpanch reservations affect panchayat expenditures, I
use the regression:
Y j = α + β1(d_reserved for female)j + β2(controls) j + ε j (1)
where j denotes village. Y can be whether funds were spent, or how much was spent, on
thirteen different expenditure categories (see appendix for complete variable definitions).
Control variables include measures that reflect the “neediness” of the village – for example, a
large number of landless poor might encourage spending on community buildings. Variables
reflecting the popularity of the sarpanch (such as the percentage of votes received by the
winner) are also held constant, as I am attempting to isolate the effect of reserving a sarpanch
position for a woman, not the effectiveness of any particular sarpanch.
Note that the dummy indicator here is whether the sarpanch seat is reserved for a
woman, not whether the sarpanch was a woman. The estimated coefficient will therefore
capture the effect of the policy in isolation, not of the affect of the sex of the sarpanch.
Female sarpanch reservations are randomly assigned to one third of all villages, and this
assignment is rotated every term, so a village will be reserved to have a female sarpanch
every third term. This exogenous variation in female sarpanch assignment ensures that there
is no selection bias.
One important consideration is also the panchayat income. I control for variables that
are sources of potential income (village resources such as public land that can generate rent,
or fishing ponds), and also estimate the regression above where Y is the total panchayat
income to see whether female-reserved villages systematically receive different income than
those which are not. There does not seem to be any significant differences in female
panchayat incomes between male and female sarpanch terms.
19
2. Female Sarpanches and sex ratios, health, and education
After seeing whether female sarpanches can influence panchayat decisions, I wish to
see whether this sort of highly-visible female exposure in a traditionally male-dominated
realm can affect household behavior towards female children. The variables of interest
include decisions made before children are born (though sex ratios), after birth (by looking at
immunization rates), and during childhood (through educational quality changes).
I first check that bargaining power is a significant factor when considering sex ratios
using the regression:
Sex ratiojc = α + β1(average maternal education)jc + β2(average paternal education)jc
+ β3(controls) jc + ε jc (2)
where j denotes village and c denotes cohort was born (cohorts are composed of all
individuals born in a given year). So (average maternal education)jc designates the average
education of all mothers in village j who had a child in cohort c, and similarly for paternal
education. Controls is a vector of controls which includes the log village earnings and
village characteristics, such as village population by sex and number of village households.
Here, maternal education is being used as a proxy for bargaining power, to see if it influences
the sex ratio.
I then repeat this analysis for my other variables of interest – the vaccination sex ratio
and primary school enrollment sex ratio. (The vaccination sex ratio is the percentage of
females vaccinated in a cohort to the percentage of all individuals vaccinated in a cohort; the
primary school enrollment sex ratio is the percentage of females who attend at least one year
of school in the cohort to the percentage of all individuals who did so). The cohorts here are
20
created in a similar manner as before – for the vaccinations, a cohort is composed of all
individuals who could have received a vaccination in a given year, and for education, a
cohort is composed of all individuals who could have been enrolled in primary school for the
first time (all children aged 6) in a given year. Numerous other studies have shown that
maternal bargaining power has an affect for such measures of child health as height and
height-for-weight (Thomas, 1990, 1994; Liu 2008) so it is reasonable to expect a statistically
non-zero coefficient on the maternal education variable.
Maternal bargaining power is more commonly measured through more direct
measures of female contributions to the household, such as female income or pensions, etc.
However, 95% of the females in the dataset are unemployed and generate no income. It is
likely that their main contribution to the household is through non-income labor such as
household chores, child care, and work in their own fields. Female education may therefore
be an indirect measure of female bargaining power, since education may make these efforts
more efficient and therefore vary with the female contribution to the household. However,
maternal education may not accurately capture the full female contribution to the household
since it is not a direct measure of female household contribution, so while the results
presented later from this analysis are suggestive, alternative interpretations of the results are
valid.
After determining whether bargaining power has an effect on sex ratio, I turn to
finding the effect of a female sarpanch on sex ratio. Theoretically, as outlined previously, a
female sarpanch has the ability to affect female bargaining power in the village; through this
avenue she might change the sex ratio. As in the expenditure analysis above, the regression
21
here uses the exogenous discontinuity between having and not having a female leader to
examine sex-ratio differences by exploiting the randomness of female sarpanch assignment
across villages over time. I compare cohorts in villages which have had a female sarpanch to
cohorts in villages which have not, and selection bias is not a concern because village
sarpanch seats are reserved for women at random times. As in the bargaining analysis, the
sex ratio cohorts are made up of all individuals born in the same year, and the sex ratio of the
cohort is the subset of all females born in the cohort over the total number of individuals born
in the cohort. This sex ratio comparison can be repeated for all years/cohorts starting from
when the first village had a female sarpanch in order to reveal whether there is a difference
between the reserved and non-reserved villages. Thus the comparison is at the cohort level
across villages, and not across time, so it is unnecessary to be concerned about time trends,
and the inclusion of village fixed effects will not confound the analysis.
The exogenous nature of assignment of female sarpanch reservations is therefore
crucial to the analysis. The method of implementation of the policy ensures that the timing
of the reservation is random and it is therefore highly unlikely that there is any systematically
biased village characteristic for all the villages assigned to have a female sarpanch in a
particular election, but the sample size used here is small so it may contain villages that
suffer from such systematic bias simply by chance. I therefore confirm that village
characteristics are similar between the villages that had female sarpanches in 1995, 2000, and
2005 (see Table 2).
In order to find the effect of female sarpanch exposure on the variables of interest, I
use the regression:
yjc = α + β1(d_femaleLeaderVariable)jc + β2(shamlat) jc + β3(controls) ijc + ε ijc (3)
22
where i denotes individual in village j, cohort c. See appendix for complete variable
definitions. In the base regression, yjt is the sex ratio for village j for cohort c, which is
regressed on a dummy for having a female sarpanch in the village to find whether exposure
to female sarpanches can affect the sex ratio.
In equation (3), shamlat denotes the amount of public land a village owns. The rental
of shamlat land is a major source of income to the village panchayat, who have complete
discretion over the spending of these funds -- unlike much of its federally-allotted budget. It
therefore provides a measure of the economic power of the sarpanch.
Here, ‘controls’ is a vector of control variables that include village characteristics,
since the effect is expected to be different for different village populations, resources
available, and village average incomes. These factors may limit the “social effectiveness” of
a female sarpanch – for instance, if a village is five or ten times the size of a typical village,
as is the case for certain outliers in the sample, fewer village members may be aware that the
sarpanch is a woman, diluting the effect of having a woman in power. These controls are not
theoretically necessary if the sample size is large, since the female sarpanches are assigned
randomly, but including them here may significantly help overcome the issue of the small
sample size.
Son preference is also examined through the health and education investment
differences between sexes, because these measures may provide additional information
distinct from sex ratio patterns. Vaccination sex ratios and primary school sex ratios (see
definitions in appendix and above) are regressed against whether that village has been
exposed to a female sarpanch or not in order to determine whether female sarpanches can
affect these measures of investment.
23
The affect of a female sarpanch on son preference indicators may change over time
and across villages of different characteristics (such as village income or average parental
education), and therefore I include a sensitivity analysis section exploring these relationships.
To see whether the affect of a female sarpanch changes over time, I estimate the same
regressions while replacing the female leader dummy with a variable measuring the number
of years since the village had a female leader (starting at the beginning of her term). The
coefficient would then estimate the effect of exposure to female leader over time. I also run a
separate regression that includes the square of this variable to explore the possibility that the
effect of the female sarpanch on sex ratio might be increasingly time sensitive – i.e., the
effects of having a female sarpanch may be “forgotten” (in terms of the gender ratio or
education ratio, etc) more rapidly as time goes on. This might occur if there is a social effect
to the influence of having a female sarpanch – originally many households may decrease
their preference for males, but as more and more households revert back to their previous
beliefs, the societal pressure to have more girls will drop, causing a parabolic effect on the
sex ratio. Conversely, this could happen in the opposite direction, if societal pressures for
accepting women achieved critical mass in the village and more and more people were
pressured into, and eventually accepted, that females had greater value than they originally
believed.
In the second part of the sensitivity analysis I examine whether there is heterogeneity
within the results. There are two sources of heterogeneity – variation by the resources the
sarpanch has at her disposal, and variation by the characteristics of the village. Each of the
three sex bias indicators – sex ratio, vaccination sex ratio, and education sex ratio – is
regressed on interactions between the female sarpanch variable with indicators of sarpanch
24
resources (like available shamlat land) and village characteristics like average log village
income, average village maternal education, and average village paternal education. The
regression is:
yjc = α + β1(village characteristic * d_has ever had a female sarpanch)jc
+ β2(shamlat) jc + β3(village characteristic)jc + β4(controls) ijc + ε ijc (4)
where i denotes individual, j village, and c cohort, and village characteristic can be shamlat
land, earnings, average village maternal education, or average village paternal education and
controls are as specified in equation (3). I am therefore able to determine whether the female
sarpanches with greater access to resources (shamlat land) are more able to influence sex
ratios than those who don’t and whether female sarpanches with wealthier or more educated
villages are more influential.
Background on the Datasets
In order to ask the question of whether female sarpanches can change the perceived
value of women in households, the reservation system must be effective in bringing a woman
into office. While compliance with the 73rd amendment is complete within the sample, there
is ample anecdotal evidence that female sarpanches do little to influence panchayat decisions.
Duflo has already shown that female sarpanches are effective in initiating policies that cater
to the needs of women in Rajasthan, which borders on Haryana on the west, but Rajasthan is
demographically dissimilar to Haryana. Punbaj, however, to the northwest of Haryana, is
quite similar in size, population, literacy rates, and other demographic measures (see Figure 1
and Table 1). I briefly compare the difference between various demographic measures
between Haryana and its neighbors (standardized by national standard errors) to see that
25
Punjab is Haryana’s the most similar neighbor by far, and is actually very similar to Haryana
in absolute terms in many demographic categories as well.
Therefore Punjabi data from a subset of the CRRID-SCID Study of Fiscal
Decentralisation, which includes panel data on 300 villages at the village level, is used to
examine the relationship between female sarpanch reservations and panchayat expenditures.
If the female sarpanches elected through the reservation system are effective in representing
female concerns, then their villages would be expected to have different expenditure patterns
than non-reserved villages. The survey data was collected at the village level, and includes
whether the sarpanch position was reserved for a female, whether the panchayat spent funds
on public goods (which are broken down into 13 categories) in 2004-2005 and if money was
spent, how much. Sarpanch elections were held in 1993, 1998, and 2003; legally, one third
of the villages were reserved for female sarpanches. The 2003 reservations are analyzed here,
and in compliance with the law, 32% of villages were reserved for a female sarpanch that
year. The dataset also includes measures of village resources and sources of funding as well
as the proportion of poor (as measured by land ownership, income) and needy in the village
(i.e., pension drawing status).
After finding whether panchayat spending was influenced by female reservations, I
turn to the question of whether female sarpanches are able to directly influence the perceived
value of women in society. For this question, data was obtained from household and village
surveys of nineteen villages in Haryana state (in Faridabad, Rohtak, and Sonipat districts),
outside Delhi, India. Surveys were verbally conducted and translated from Hindi.
Nineteen villages were randomly chosen from village lists obtained from block
development offices. Village surveys were verbally conducted with the panchayat or with
26
the sarpanch. Survey information included village-level characteristics, such as panchayat
member characteristics and amount of various village resources. Since the 73rd amendment
was implemented, all villages have held elections once in 1994-1996 (exact year differs
between villages), 2000, and 2005.
Household surveys were conducted with 10-15 households in every village, except
one village which only had 4 households. At least two below poverty line (BPL) households
were included from every village to insure variation in income, and social economic group
variation was ensured by covering at least one household from every caste in the village.
Besides these variations, households were chosen randomly from the village resident list.
Household representatives were asked to give ages, relationships, educational characteristics,
and children of all household members along with other individual characteristics as well as
provide household level information (such as household income, landholdings, etc).
This dataset is uniquely tailored to answer the question of how gender ratios change;
all family members (including away or dead children) are documented by age, gender,
whether they were vaccinated, and educational level and status. The village and household
sex ratios over time can therefore be reconstructed using the ages of all household members,
even if they are currently away or dead. All individuals are assigned a cohort by the year
they were born and the village they were born in; the sex ratio (females born to total born) is
then calculated by village and cohort. In this way one can “look back in time” in the sex
ratio data to examine the sex ratios of cohorts born in villages before the sarpanch position
had been reserved for a woman and after.
Immunization rates are reconstructed in a similar manner, since immunization takes
place in the year after birth. The relationship between primary school enrollment (whether
27
one currently attending or completed one year of primary school) and female sarpanch
exposure can also be determined if one makes the assumption that the decision to attend
primary school for all children is made when they are six years of age. I designate all
individuals who have had a female sarpanch by reservation in their village prior to age six as
having parents who have been exposed to a female sarpanch, and I can see whether there is a
difference between school sex ratios (percentage of 6-year-old females who will complete
primary school to the percentage of all 6-year-olds) of those cohorts which have parents who
have been exposed to a female sarpanch and those which do not have such parents.
The dataset therefore provides unique information needed in variable construction –
may datasets do not include immunization or prenatal checkup history for all individuals, for
example. It is for this reason that it is chosen over other public datasets which have larger
sample sizes. This dataset includes information on over 1600 individuals in 18 villages, but
the sample size shrinks as those without various control variables must be excluded, resulting
in a minimum sample size of 213 individuals for certain regressions. The sample size also
varies by dependant variable; while all individuals surveyed provided information on age and
education, vaccination data could only be obtained for individuals with a living parent in the
survey household, cutting the maximum sample size to just over 564 (may be less due to
omission of control variables used when regressions were run).
The source of the advantages for this dataset -- that the questions were tailored to
answer questions about sex ratios -- could have potentially generated biased answers since
there are legal and social repercussions to gender selection. Therefore all questions
concerning births and health of children were framed in the context of gathering general
household information instead of asking expressly about gender preference.
28
Compliance with the 73rd amendment was close to complete: all but one of the
villages had a female sarpanch once in the last three terms, with only two exceptions to the
expected female sarpanch pattern. (The exceptions were one village which reported an non-
mandated female sarpanch in 1996 and mandated one in 2005, and another village which
reported a mandated female sarpanch in 1990, before the policy was implemented.) One
third of all panchayat members were female in all villages. Thus all villages are standardized
in that each had only one female sarpanch through the reservation system.
The third source of data used is the census of India. The census is taken every ten
years by the Registrar General and Census Commissioner of India, and includes state, district,
and village level information on population and household size and number. 1991 census
data is chosen over other censuses for use in the controls since female sarpanches are only
assigned after the passage of the 73rd amendment in 1992, and changes in the villages
reflected in the 2001 census could be the direct result of sarpanch actions. The 2001 census
data is used to show that comparisons between Haryana and Punjab are valid; the two states
are demographically similar and effects of the reservation system should be comparable.
Section 5: Results
A graphical representation of the total number of children born by sex over time for
all villages is shown in Graph 1. There are significantly fewer girls born than boys after
around the 1980’s, and the absolute sex ratio over this period is lower than parity. Sex ratios
for all villages, by years normalized to the first year that a female sarpanch was in office, are
shown in Graph 2. A simple visual analysis of the graph may lead one to believe that if there
29
changes due to female sarpanches the changes are not immediate or do not occur in all
villages, which motivates the sensitivity analysis.
Female Sarpanch Reservations and Expenditure patterns
The analysis of whether reservations for female sarpanches are able to influence
panchayat decisions confirms that reservations indeed make a difference on village
expenditures (See Table 3a, 3b, and 3c). I look at thirteen categories in which panchayats
can invest, and for each I examine whether the reserved or non-reserved sarpanches of
funded it and the amount they funded it by.
The regressions indicate that female sarpanches selected through reservations tend to
spend more on irrigation canals and tanks, drinking water and water works (significant at the
15% level), streetlights (significant at the 10% level), and salaries for panchayat members.
These sarpanches were also less likely to spend on hospitals and health centers, but those that
did spent more than their male counterparts (both measures significant at 10% level). Less of
the reserved villages spent funds on public buildings and community centers. (All
differences between the two types of villages are significant at 5% level unless otherwise
stated.) Therefore out of the thirteen expenditure types, six categories show differences
between non-reserved and reserved sarpanch positions.
This shows that reserved panchayats act differently from non-reserved villages, and
since female reservations are randomly assigned, this difference must be due to the
reservation system. Previous literature shows that women sarpanches are better able to
address the needs of women in the village; while this analysis does not show that these
30
changes in Punjab cater to village women, they at least reflect the concerns of the female
sarpanch. These changes could potentially demonstrate to villagers that women have a role
in public office and decision-making positions in the home, thereby increasing the perceived
value of women. To see if they can, I must first determine whether female bargaining power
can influence the dependent variables I use in my analysis.
Sex Ratios and Bargaining Power
The first regression here determines whether maternal bargaining power can affect
sex ratios, where village average maternal and paternal education act as proxies for
bargaining power (see Table 4). The results show that both maternal and paternal education
are strong predictors of sex ratio at the 5% level, even controlling for village characteristics
such as village populations and households (and therefore household size). However, the
sign of the maternal education coefficient is negative, indicating that increases in maternal
bargaining power decrease the sex ratio (reduce the fraction of girls born). This is a contrary
result to that found in much of the literature, which shows positive correlations between
maternal bargaining power and child health. This anomaly could be due to attitudes specific
to Haryana, which is unique in having one of the lowest sex ratios in the country. However,
it could also be an effect of not having a direct measure of maternal bargaining power within
the home.
Since health and education sex ratios are also of interest, I also wish to find whether
female bargaining power is a force that can change girls’ health and education investment. In
the vaccination sex ratio regression, the maternal education coefficient is significant and
31
positive, indicating that mothers with more bargaining power give daughters more
vaccinations.
The next regression shows that maternal bargaining power does not change
daughters’ primary school enrollment rates, and I therefore should expect no change in
primary school sex ratios if the female sarpanch influences the perceived value of women
through maternal bargaining power. Examining this female investment indicator is still
valuable, since it can act as, in the language of medical testing, a “negative control” -- it will
help identify when changes in female investment patterns are due to an affect on maternal
bargaining power or due to some other change in the village.
If the above analysis is correct, then if the female sarpanch were able to increase
female bargaining power within the home it might depress the sex ratio still further, although
vaccination ratios should rise. I explore whether that is the case in the next section.
Female Sarpanches and Sex Ratios
I examine whether female sarpanches influence the perceived value of women in
society by measuring changes in the sex ratio, an indicator of the desirability -- and by proxy
the social value -- of women. The sex ratio can be followed over time in Graph 1. In the
base case, the sex ratio (fraction of female individuals in the cohort, where a cohort consists
of all individuals born in the same year) is regressed on whether that village has had a female
sarpanch at the time of a cohort’s birth. These results are shown in Table 5 (regressions 1, 2,
and 3). The estimates indicate that there seems to be no relationship between having ever
had a female sarpanch and the sex ratio.
32
Shamlat land does not significantly affect the sex ratio when village fixed effects are
taken into account; it would be reasonable to assume that shamlat land is not a fixed village
characteristic (as it may be sold or rented out in various quantities to provide income for the
panchayat) and therefore evaluating it with the inclusion of village fixed effects is correct.
One might be concerned that the accuracy of self-reported ages decreases with the
age of interviewee, causing measurement error and biased estimators; higher mortality rates
for older persons may also increase the variance of yearly sex ratio measurements even given
accurate reporting, which would influence the t-statistics. Therefore identical regressions
were run using only individuals born after year 1960. I find similar results regarding
coefficients and t-statistics, showing that the data from older people in the sample generally
reinforce the same trends in the data from younger people.
Over-fitting is not seen to be a concern even though the regression only uses 564 data
points for a considerable number of independent variables as the R value is actually quite low.
The regression suffers from some non-robustness, as the coefficients and significance levels
change dramatically as more variables are controlled for, but all the variables included in the
final regression are reasonable and theoretically justified. The amount of shamlat land
affects the degree of influence a sarpanch has by providing the ability to fund panchayat
projects. The village average income could reasonably be related to desirability of males
(necessity of old age support, etc). Individual characteristics such as savings and parental
education might be important for controlling for household-level characteristics. Village
population measures would also reflect the transparency and publicity of panchayat and
sarpanch actions (for instance, the smaller the village, the larger the fraction of constituents
the sarpanch may have had personal contact with) which would effect the degree to which
33
the sarpanch is able to influence public perception. This result, then, indicates that the
gender of the sarpanch is not able to elevate the perceived value of females enough to change
child sex preferences.
Sex Preference Indicators through Health Investment: Immunizations
A sufficient increase in the perceived value of females to overcome cultural and
social sex preferences might be difficult to achieve. Here, I examine changes in health
investment of female children that parents have already decided to have. This measure is
expected to vary independently of sex composition choices as parents may have more
autonomy over the level of investment in a child’s health than the sex of the child. Here, the
vaccination rate is an index of health investment
Vaccination rates have changed dramatically in rural villages. Out of the entire
sample, an average of 71% people are vaccinated against any sort of disease (the most
popular vaccinations include polio, diphtheria, BCG, and hepatitis B) while children under
the age of 14 had an average vaccination rate of 88%. Vaccinations are normally
administered at or soon after birth, with the child taking a series of booster shots that ends
before age two. Crucially, investment in vaccinations occur after the sex and health of the
child are known, so the sex of the child may well play a role in determining whether he or
she receives vaccinations or not. The bargaining power regressions show that mothers with
bargaining power favor giving vaccinations to their daughters, so there should be an increase
in the vaccination sex ratio if female sarpanches are able to increase female bargaining power.
34
Even so, the regressions show there is no significant change in the proportion of
female vaccinations as compared to total vaccinations (see Table 6 regressions 1, 2, and 3).
Ever having a female sarpanch does not change vaccination behavior for women.
Sex Preference Indicators through Educational Investment: Primary School Enrollment
While households may not respond to a female sarpanch by changing health
investments for girls, they might change other types of investments. The analysis from
Punjab showed that villages reserved for female sarpanches were less likely to spend
panchayat funds on hospitals and health clinics; if the female sarpanch was catering to female
desires, as suggested by previous literature, then perhaps health investment is not a priority
for women in the household. Then more bargaining power for females would not result in
greater health investment for either sex.
I therefore turn to education, a measure of investment that happens later in a child’s
life. The increase in the child’s age also changes the factors behind the parental investment
decision, since the parent has now already invested in the child. This gives another reason to
believe that analysis of educational investment patterns will generate an additional dimension
of information on sarpanch efficacy. I estimate whether having had a female sarpanch in the
village changes the primary school enrollment sex ratio, defined (in the appendix) as the
percentage of females in the cohort who attend at least one year of primary school to the
percentage of all children in the cohort who attend at least one year of primary school. The
analysis finds that those parents who live in a village which has had a female sarpanch by
reservation before their child is six years of age (when they must make the decision whether
35
to send their child to primary school) are not any more or less likely to enroll their girls in
primary schools (see Table 7 regression 1, 2, and 3).
While this is consistent with the findings concerning maternal bargaining power,
where the analysis found that daughter’s education investments do not vary with mother’s
bargaining power, the results from the previous two sections showed that female sarpanches
were not able to change sex ratios or health investment ratios, which maternal bargaining
power is important in predicting. Therefore this negative result is likely due to the lack of the
influence of the female sarpanch, not a lack of response due to the inability of female
bargaining power to change educational sex ratios.
Sensitivity Analysis: Allowing for a Nonlinear Effect of Female Sarpanch Exposure
1. Sensitivity of the Sex Ratio to Time from Female Sarpanch Exposure
In order to more fully understand whether the female sarpanch reservation has an
affect on the sex ratio, I explore whether there is a linear or parabolic relationship between
the time since the female sarpanch took office and the sex ratio (wee Table 5 regressions 4-9).
There does not seem to be any relationship.
2. Sex Ratio and Heterogeneous Exposure to Female Sarpanches
I also separate out various village characteristics which might influence how
receptive the village is to a female sarpanch. If the sarpanch has more disposable income, for
instance, she might be able to enact more effective or more visible programs, which could
change villagers’ perceptions about her efficacy and the desirability of a woman in politics
36
(see Table 8a). Shamlat land is a large source of a sarpanch’s non-earmarked funds, so I use
this as a measure of the amount of resources the sarpanch has available. I find that the
interaction term between having a female sarpanch and amount of shamlat land has a
statistically non-zero effect on sex ratios at the 10% level. The sign of the coefficient
indicates that the wealthier the female sarpanches are, the more depressed the sex ratio
becomes, which is consistent with the effects of greater maternal bargaining power as seen in
the bargaining power estimations. This would then imply that the efficacy of the female
sarpanch in elevating female bargaining power is dependent on the financial resources she
has at her disposal.
The resources of the residents of a sarpanch’s village may also affect how effective
she is. I test whether female sarpanch reservations can change sex ratios for villages that are
on average wealthier, have higher average maternal education, and have higher average
paternal education in three separate regressions. I find no significance for any of them,
indicating that whether female sarpanches change the perception of women is independent of
village average income and parental education.
3. Sensitivity of the Health and Education Sex Ratios to Time from Female Sarpanch
Exposure
As with the analysis of the sex ratio, I wish to check for changes over time as well as
for sarpanch and village characteristics that might make it more likely for a female sarpanch
to have an affect on health and education investment decisions. There are no linear or
parabolic relationships for the health (see Table 6 regressions 4-9) or education sex ratios
37
(see Table 7 regressions 4-9) over time since the female sarpanch was in office once village
fixed effects are controlled for.
4. Health and Education Sex Ratios and Heterogeneous Exposure to Female
Sarpanches
As in the sex ratio analysis, the amount of shamlat land a village has is an important
factor for female sarpanch efficacy (see Table 8b). The vaccination sex ratio increases with
shamlat land interacted with female sarpanch reservation while education investment does
not; this is consistent with the directions of change seen when maternal bargaining power is
increased. Along with the sex ratio results, this confirms that disposable funds are a crucial
factor influencing the ability for female sarpanches to change female bargaining power and
early-childhood investment decisions. This is result is reiterated with the education
regression, which shows that even female sarpanches with shamlat land do not change
education investment in girls (see Table 8c). The bargaining power analysis earlier showed
that maternal bargaining power does not influence girls’ educational attainment even while
being significant for the sex ratio and health regressions, so the fact that the sex ratio and
health investment patterns changed in villages with landed female sarpanches while
education did not is one clue that female sarpanches act through changes in maternal
bargaining power to affect these measures of female investment.
There are no changes in health or education investment patterns if the female
sarpanch village is wealthier or if fathers in the village are, on average, more educated.
However, if mothers are more educated in the female sarpanch village, there is a decline in
the vaccination sex ratio, indicating that fewer girls are vaccinated (see Table 8b). Therefore
38
the affect of the sarpanch does vary slightly with maternal education in the village, although
only by affecting vaccination sex ratios and no other measures of female investment. The
sign of the change is also interesting; it is contrary to the results that an increase in female
bargaining power would give. The mechanism for this change is unclear, and this finding is
unsupported by the data in the other two sex preference variables.
While the direction of the effect of female sarpanches with shamlat land is consistent
with increases in maternal bargaining power, the magnitude of the effect is slight. In the
maternal bargaining power analysis I found that an additional year of maternal education
results in a -0.00998 change in the sex ratio, 0.0475 change in the vaccination ratio, and no
change in the education sex ratio. A female sarpanch reservation in a village with one
additional acre of shamlat land results in a -0.000381 change in the sex ratio, 0.00102 change
in the vaccination ratio, and no change in the education sex ratio. A rough evaluation of the
magnitudes shows the effects are roughly 25 to 45 times stronger respectively for an
additional year of maternal education than the effect of the female sarpanch. This would
roughly mean a female sarpanch would need roughly 45 more acres of shamlat land to have
the same effect on female bargaining power as an additional year of maternal schooling
(assuming that non-linear effects are still negligible at this range). This is a large difference;
82% of the individuals surveyed live in villages with 80 acres or less of shamlat land, and
70% in villages with less than 40 acres. So while the effect of a female sarpanch with land is
significant and increases female bargaining power in the village, the magnitude of the change
is small.
39
Section 6: Conclusion
Several papers have shown that female reservations in the Indian panchayat
leadership system have significant effects on policy decisions and allow for greater
representation for women’s concerns. However, effective female governance may not
necessarily lead to the erosion of discrimination against women. In this paper I asked
whether political gender reservations can increase the perceived value of females by
evaluating changes in familial investment in girls living in villages with female village
leaders.
Looking at the discontinuity between villages before having a female sarpanch and
after having one, I confirmed that the female sarpanch reservation alters panchayat
expenditure patterns, showing that female sarpanches are able to effectively influence
panchayat decisions in the region of interest.
The average village maternal bargaining power is important in determining the sex
ratio and vaccination sex ratio, although not the primary school sex ratio. Even so,
reservations for female sarpanches alone are unable to affect village sex ratios or female
health and educational investments. Only in villages which have both a female sarpanch and
shamlat land are all female investment indicators consistent with an increase in maternal
bargaining power, and the size of the change varies linearly with the amount of shamlat land
the village has.
This shows that reserving sarpanch seats for females is not enough to change female
status and bargaining power in the village; the female sarpanch must also have disposable
income in order to be an effective leader and demonstrate that females belong outside the
40
home. Ironically, though, the policy would fail in achieving its goal of fixing the sex ratio
even if the government provided greater financial resources to female sarpanches -- greater
maternal bargaining power is correlated with lower sex ratios in Haryana. Political
reservations for females must be supplemented with sound financial resources in order to
increase women’s status and bargaining power, but doing so without changing maternal son-
preference will not be effective in fixing the skewed sex ratio.
While it is reasonable and expected that female sarpanches require resources in order
to be effective, it is somewhat surprising that female empowerment does not reduce son
preference. This is especially so in light of other studies, like Thomas’, which have shown
that female bargaining power increases investment in daughters. One possibility for maternal
son preference in Haryana is the unique level of dependence of mother’s on sons. As
reviewed earlier, females have very low status in northern India and have limited property
rights without male representation (Dasgupta 1995). This dependence is particularly striking
for widows who must rely on sons – unlike in the regions where Thomas and others based
their studies, widows in northern India often rely almost completely on sons for resources,
and therefore may show more son preference than fathers. It is therefore plausible that
increasing women’s bargaining power without addressing their financial dependence on sons
may increase gender bias.
This study focuses on villages in which there have only been one female sarpanch;
the literature shows that different results can be seen as a village is required to have a female
sarpanch more than once. Further studies need to explore the effect of repeated female
sarpanches on discriminatory attitudes. However, the present analysis shows that political
reservations alone are not an effective tool for changing discriminatory attitudes and should
41
be supplemented with financial resources if the goal is to allow female sarpanches to be
effective in increasing female bargaining power and social status. At the same time, this
result will be likely to exacerbate the sex bias problem, which is already of grave concern in
Haryana. I therefore see little hope that this policy will be able to address all measures of the
sex discrimination problem in India effectively.
42
Section 7: Appendix
Definitions:
Regression set 1:
Y j = α + β1(d_reserved for female)j + β2(controls) j + ε j
• Where j denotes village. • Where Y can equal total panchayat income, or whether there were expenditures
on (13 regressions), or how large the expenditures on (13 separate regressions): o electricity supply • irrigation canals and tanks • drinking water and water works • sewage and drainage • roads • schools • hospitals and health centers • street lighting • panchayat buildings and community centers • anganwadi center • sports stadium • salaries • other
• d_reserved for female is a dummy variable for whether the village was reserved
for a female sarpanch.
• Controls are a vector of control variables listed in the regression tables.
43
Definitions (cont.): Regression Set 2:
yjc = α + β1(d_femaleLeaderVariable) jc + β2(shamlat) jc + β3(controls) ijt + ε ijc
where i is the index for the individual belonging to j for village and cohort c. Each cohort is made up of all children born in a particular year.
• y can equal the sex ratio, prenatal checkup sex ratio, immunization sex ratio, school attendance sex ratio, or educational quality sex ratio which are defined as follows:
• sex ratio = (females born in village j in cohort c) / (total children born in
village j cohort c)
• prenatal checkup sex ratio = (females given prenatal checkup in village j cohort c) / (total children given prenatal checkup in village j cohort c)
• immunization sex ratios = (females immunized in village j cohort c) / (total
children immunized in village j cohort c)
• school attendance sex ratio = (percentage girls who are in or will complete 1 year of primary school in village j in cohort c) / (percentage of total individuals who will enroll in or complete 1 year of primary school in village j cohort c)
• D_femaleLeaderVariable is a dummy variable that denoting whether a female leader
has been in office in the village before or during the year of the births, the time between the birth year of the individual and the year the female sarpanch took office, and time between the birth year of the individual and the year the female sarpanch took office squared.
• Shamlat is the amount of shamlat land the village has
• Controls is a vector of control variables (see regression tables for complete list) which
include characteristics of individual i and village j. 1991 Village Characteristics, as seen on the regression tables, is part of the control vector and contains village characteristics from 1991. They are: village land area, population, female population, male child and female child population, no. of houses, and no. of households.
44
Graph 1: Total Males and Females Born by Calendar Year
Total No. born in Haryana sample (5 yr MA by calendar year)
0
5
10
15
20
25
30
1900 1920 1940 1960 1980 2000
Calendar year
# bo
rn MalesFemales
Graph 2: total number born, by sex and calendar year for all villages. Graph is smoothed using a 5-year moving average to reduce visual noise. The total number of females born after the 1990’s (when the 73rd amendment was ratified and implemented) are smaller than the total number of males born then. The trend seems to begin before the 1990’s, and may be part of the reason that the policy was enacted. I will be looking for discontinuities in the sex ratio of each village at the time the female sarpanch took office, which are 1995, 2000, 2005, depending on the village.
45
Graph 2: Sex Ratios Over Time
3 yr MA Sex Ratios
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
-70 -60 -50 -40 -30 -20 -10 0 10 20 30
Normalized Years
Sex
Rat
io (g
irls/
tota
l bor
n)
Graph 1: Shows the sex ratio over time, using a three-year moving average of sex ratios to reduce visual noise, with years normalized to the year that the female sarpanch took office. All sarpanches serve for a 5 year term, so villages had a female sarpanch for years 0 - 5.
46
Figure 1: Map of Indian States
Haryana's neighboring states. Previous literature shows that female sarpanches in Rajasthan (to the southwest of Haryana) are able to effectively influence panchayat expenditures. I do the same analysis for Punjab (to the northwest of Haryana) and see the same results. Map modified from Census of India 2001 accessed at http://www.censusindia.gov.in/maps/india2.jpg on May 1, 2009.
Table 1: Demographics of Haryana and its Neighbors
Total Population:Total 29443371.56 38029976.34 21144564 24358999 -0.085 5360926 -0.415 56507188 0.930 166197921 3.814Males 15908512.94 19485479.36 11363953 12985045 -0.083 6077900 -0.271 29420011 0.927 87565369 3.911Females 14188923.6 17947311.95 9780611 11373954 -0.089 3087940 -0.373 27087177 0.964 78632552 3.836
No. Households 5530855.829 6743173.138 3712319 4348580 -0.094 1221589 -0.369 9317675 0.831 25757640 3.269
Rural pop:Total 21214018.26 28352071.97 15029260 16096488 -0.038 5482319 -0.337 43292813 0.997 131658339 4.114Males 10902933.54 14731050.98 8052988 8516596 -0.031 2756073 -0.360 22426640 0.976 69157470 4.148Females 10311084.71 13627313.85 6976272 7579892 -0.044 2726246 -0.312 20866173 1.019 62500869 4.075
No. SC: Total 5207365.625 7353406.288 4091110 7028723 -0.399 1502170 -0.352 9694462 0.762 35148377 4.224No. SC: Male 2690273.75 3833753.766 2188585 3714350 -0.398 763333 -0.372 5067679 0.751 18502838 4.255No. SC: Female 2517091.875 3521369.987 1902525 3314373 -0.401 738837 -0.330 4626783 0.774 16645539 4.187
Male Births Last Year 298197.2286 399011.0159 238814 197007 0.105 59676 -0.449 810308 1.432 1873074 4.096Female Births Last Year 270004.3429 360983.7419 187658 155092 0.090 50402 -0.380 700126 1.420 1687395 4.155
Literacy Rate: Total 16019651.34 19473893.35 12093677 14756970 -0.137 4041621 -0.413 27702010 0.802 75719284 3.267Literacy Rate: Male 9615249.029 11875017.3 7480209 8442293 -0.081 2278386 -0.438 18047157 0.890 48901413 3.488Literacy Rate: Female 6404402.314 7693359.11 4613468 6314677 -0.221 1763235 -0.370 9654853 0.655 26817871 2.886
* the difference between Haryana and state means in national standard errorsData from Registrar General & Census Commissioner, India, Census of India 2001 accessed online at http://www.censusindia.gov.in/default.aspx on May 1, 2009
Himachal Pradesh Average
Diff. from Himachal (in national. s.e.)*
Rajasthan Average
Diff. from Rajasthan (in nat. s.e.)*
Mean of all states national s.e.
Haryana Average
Punjab Average
Diff. from Punjab (in national. s.e.)*
Uttar Pradesh Average
Diff. from Uttar Pradesh (in nat. s.e.)*
The demographics of India, Haryana, Punjab, Himachal Pradesh, Rajasthan, and Uttar Pradesh. We find the national standard error using data from all 28 Indian states, then use this as a standard by which we can use to compare the differences between Haryana and its neighbors. A visual inspection of the demographic distance between Haryana and its neighbors shows that Punjab is the most similar to Haryana -- most of Punjab's demographic measures are less than 0.1 standard errors away from Haryana's while the other states are much higher. We therefore use Punjabi expenditure data to confirm that reserved female sarpanches are able to affect panchyat expenditures and assume that this conclusion holds for sarpanches in Haryana.
Table 2: Average Village Statistics By Female Sarpanch Term In Haryana
AllNever 1995 2000 2005
Village Area (Sq km) in 1991 504.15 462.68 384.06 430.03 643.92Village Shamlat Land Amt 113.33 6.99 14.96 149 177.11Average VillIage Per Capita Income 1169.79 704.33 1257.87 1421.25 1094.71Ave. Rooms Per House 877.01 892.2 883.49 858.39 883.63Ave. TV Possession Rate 875.65 891.88 883.54 856.39 881.74Total Village Pop. in 1991 3175.7 1764.08 1834.5 2068.1 5575.41Percentage of Village are Female in 1991 0.47 0.46 0.48 0.46 0.46
1990Variable Obs Mean Std. Dev. Min Max
Village Area (Sq km) in 1991 233 462.6828 501.9881 171 1322.9Village Shamlat Land Amt 233 6.987124 1.743166 4 8Average VillIage Per Capita Income 233 704.3257 106.2533 642.5867 886.4035Ave. Rooms Per House 233 892.1974 308.7298 0 999Ave. TV Possession Rate 233 891.8755 309.6595 0 999
Total Village Pop. in 1991 233 1764.077 213.5378 1640 2130Percentage of Village are Female in 1991 233 817.0129 103.2826 757 994
1995Variable Obs Mean Std. Dev. Min Max
Village Area (Sq km) in 1991 132 384.06 253.9036 131.12 637Village Shamlat Land Amt 295 14.96271 18.90368 4 70Average VillIage Per Capita Income 295 1257.872 326.2619 716.6765 1703.704Ave. Rooms Per House 293 883.4881 319.3605 1 999Ave. TV Possession Rate 294 883.5442 319.8179 0 999Total Village Pop. in 1991 132 1834.5 391.9876 1444 2225Percentage of Village are Female in 1991 132 869.5 172.1533 698 1041
2000Variable Obs Mean Std. Dev. Min Max
Village Area (Sq km) in 1991 412 430.0331 158.7298 200.32 650.32Village Shamlat Land Amt 412 148.9951 263.9359 0 750Average VillIage Per Capita Income 511 1421.251 1216.043 287.6812 4282.412Ave. Rooms Per House 510 858.3873 347.1542 1 999Ave. TV Possession Rate 511 856.3933 349.6559 0 999
Total Village Pop. in 1991 412 2068.097 1138.377 465 3967Percentage of Village are Female in 1991 412 957.2184 523.0696 215 1827
2005Variable Obs Mean Std. Dev. Min Max
Village Area (Sq km) in 1991 401 643.9214 403.2707 187 1188Village Shamlat Land Amt 613 177.1062 221.2399 0 500Average VillIage Per Capita Income 613 1094.714 446.5239 452.6931 1668.317Ave. Rooms Per House 613 883.6346 319.0079 1 999Ave. TV Possession Rate 613 881.7406 321.6878 0 999
Total Village Pop. in 1991 401 5575.409 5581.774 821 13805Percentage of Village are Female in 1991 401 2577.823 2579.978 382 6385
Table 2 shows the village
demo- graphics of
Haryana villages for
those villages with female
sarpanches in 1995, 2000, and 2005 for comparison
across village types (top
box) and with more detail (sections
below). In general, the
three types of villages do not system- atically vary
by any particular
measure, and this justifies
our assumption that female sarpanch
assignment was indeed
random. Data from Registrar
General & Census
Commissioner, India,
Census of India 2001 accessed online at
http://www.censusindia.gov.in/default.aspx on May 1,
2009
In Villages with a Female Sarpanch in Year:
In Villages with a Female Sarpanch in Year:
In Villages with a Female Sarpanch in Year:
In Villages with a Female Sarpanch in Year:
In Villages with a Female Sarpanch in Year:
Table 3a: Female Sarpanch Reservations and Irrigation & Drinking Water Expenditures(1) (2) (1) (2)
VARIABLES
-0.0549 6525*** -0.0479 3864~(0.0387) (2351) (0.0482) (2376)0.000164 -11.17 0.000149 -6.054(0.000116) (7.051) (0.000144) (7.126)-1.66e-05 2.843 -0.000128 5.152(7.40e-05) (4.499) (9.22e-05) (4.547)2.95e-05 -3.747 0.000151 -6.430(8.46e-05) (5.144) (0.000105) (5.199)0.00794 -314.2 0.0306 -779.6(0.0504) (3067) (0.0628) (3100)-0.00943 308.4 -0.0300 848.0(0.0505) (3070) (0.0629) (3103)-0.00805 360.3 -0.0326 769.2(0.0504) (3067) (0.0628) (3099)-0.00890 379.5 -0.0302 809.5(0.0504) (3066) (0.0628) (3098)-0.000298 15.38 -8.20e-05 17.12(0.000369) (22.44) (0.000460) (22.68)0.00797 -952.3 0.00888 -729.1(0.0158) (961.9) (0.0197) (972.2)-0.0824 429.5 -0.129 -771.0(0.0966) (5874) (0.120) (5936)-0.0225 893.1 -0.0634*** 995.0(0.0194) (1180) (0.0242) (1192)0.0539 4090 0.0800 -14266**(0.103) (6282) (0.129) (6349)-0.0105 1001 0.0107 -1452*(0.0138) (838.7) (0.0172) (847.6)0.000682 -94.23 -0.000598 11.29(0.000957) (58.19) (0.00119) (58.81)0.00670 -559.9 0.0207 -1001(0.0274) (1664) (0.0341) (1682)-0.00100 -9.994 0.000226 -48.99(0.00126) (76.58) (0.00157) (77.39)
Constant 1.945*** 2976 1.864*** 39197**(0.306) (18611) (0.381) (18810)
Observations 291 291 291 291R-squared 0.073 0.092 0.063 0.065
In order to confirm that female sarpanches are able to effectively influence some measure of panchayat operations, we analyze whether villages reserved for female sarpanches allocate village expenditures differently. We here see that they do for irrigation/water tanks and drinking water. *** p<0.01, ** p<0.05, * p<0.1, ~ p<0.15. S.e. in parenthesis. Data from CRRID-SCID Study of Fiscal Decentralisation..
Amt. of funds for drinking water
No. votes received by 2003 sarpanch winner
No. of shops generating revanue% landless village householdsNo. candidates for 2003 sarpanch
Total No. of OBC BPL HouseholdsTotal No. of SC BPL Households
Funds for Drinking water?
Rental revanue from shops?
Amt of Shamlat LandShamlat land use in past 5 years
Any fishing ponds?
Area of fishing ponds
Total No. of all BPL householdsTotal No. of General Caste BPL
Reserved For Female SarpanchSGRY funds at District LevelTotal Village Population 2001Total Village Population 1991
Funds for irrigation?
Amt. of funds for irrigation
Table 3b: Female Sarpanch Reservations and Lighting & Building Expenditures(1) (2) (1) (2)
VARIABLES
0.0413* -1964 -0.120** 8909(0.0247) (1876) (0.0484) (6865)0.000139* -13.62** 1.52e-05 -7.305(7.42e-05) (5.628) (0.000145) (20.59)1.25e-05 -1.571 -0.000247*** 11.53(4.74e-05) (3.591) (9.26e-05) (13.14)-1.63e-05 1.440 0.000246** -5.591(5.42e-05) (4.106) (0.000106) (15.02)-0.00321 43.77 0.0348 -3542(0.0323) (2448) (0.0632) (8957)0.00271 -18.71 -0.0354 3689(0.0323) (2450) (0.0632) (8965)0.00398 -81.74 -0.0371 3469(0.0323) (2448) (0.0631) (8955)0.00311 -61.04 -0.0340 3409(0.0323) (2447) (0.0631) (8953)-0.000452* 45.74** 0.000574 -53.90(0.000236) (17.91) (0.000462) (65.52)0.00475 -68.39 0.0245 -2549(0.0101) (767.8) (0.0198) (2809)0.0265 2676 0.275** -34733**(0.0618) (4688) (0.121) (17153)-0.000859 416.6 0.0482** -6328*(0.0124) (941.5) (0.0243) (3445)0.0535 -6849 -0.247* 19440(0.0661) (5014) (0.129) (18345)0.0136 -1045 -0.0407** 1778(0.00883) (669.4) (0.0173) (2449)0.000121 0.347 -0.00108 340.7**(0.000613) (46.44) (0.00120) (169.9)-0.000862 -251.8 -0.0513 1827(0.0175) (1328) (0.0343) (4859)0.000480 -42.80 -0.00278* 132.9(0.000806) (61.12) (0.00158) (223.6)
Constant 1.703*** 20405 2.168*** 10601(0.196) (14855) (0.383) (54350)
Observations 291 291 291 291R-squared 0.063 0.064 0.111 0.059
No. votes received by 2003 sarpanch winner
No. candidates for 2003 sarpanch
In order to confirm that female sarpanches are able to effectively influence some measure of panchayat operations, we analyze whether villages reserved for female sarpanches allocate village expenditures differently. We here see that they do for the street lighting and panchayat building/community center catagories. *** p<0.01, ** p<0.05, * p<0.1, ~ p<0.15. S.e. in parenthesis. Data from CRRID-SCID Study of Fiscal Decentralisation.
% landless village households
Amt of Shamlat LandShamlat land use in past 5 years
Any fishing ponds?
Area of fishing ponds
Amt. of funds for lighting
Funds for buildings?
Rental revanue from shops?
Amt. of funds for buildings
Total Village Population 1991Total No. of all BPL households
Reserved For Female SarpanchSGRY funds at District LevelTotal Village Population 2001
Funds for lighting?
Total No. of General Caste BPL Total No. of OBC BPL HouseholdsTotal No. of SC BPL Households
No. of shops generating revanue
Table 3c: Female Sarpanch Reservations and Health & Salary Expenditures (1) (2) (1) (2)
VARIABLES
-0.0325* 9035* -0.0425 2220*(0.0166) (5424) (0.0593) (1242)4.75e-05 1.979 0.000372** -2.908(4.98e-05) (16.27) (0.000178) (3.725)3.15e-05 -2.290 -3.79e-05 -0.635(3.18e-05) (10.38) (0.000113) (2.377)-5.81e-05 0.205 3.48e-05 2.619(3.64e-05) (11.87) (0.000130) (2.718)0.00191 -268.8 -0.130* 361.6(0.0217) (7076) (0.0774) (1621)-0.000675 187.1 0.130* -390.7(0.0217) (7083) (0.0774) (1622)-0.00231 313.3 0.130* -304.5(0.0217) (7075) (0.0773) (1620)-0.00202 300.1 0.129* -361.2(0.0217) (7073) (0.0773) (1620)-6.65e-05 -10.27 -0.000968* 28.09**(0.000159) (51.77) (0.000566) (11.85)0.00305 -764.7 0.0326 -511.4(0.00680) (2219) (0.0243) (508.2)0.0237 4164 0.0502 4340(0.0415) (13552) (0.148) (3103)0.00979 785.6 0.0197 557.4(0.00834) (2721) (0.0298) (623.2)-0.0170 4219 0.0753 849.8(0.0444) (14494) (0.158) (3319)-0.00736 1185 0.00366 337.3(0.00593) (1935) (0.0212) (443.1)0.000394 -284.8** -0.000847 0.368(0.000411) (134.3) (0.00147) (30.74)0.00893 -692.6 0.0220 -258.5(0.0118) (3839) (0.0420) (879.1)0.000720 -129.1 0.000348 -30.78(0.000541) (176.7) (0.00193) (40.46)
Constant 1.901*** 7514 1.320*** -6060(0.132) (42940) (0.469) (9833)
Observations 291 291 291 291R-squared 0.090 0.030 0.084 0.120
Any fishing ponds?
Area of fishing pondsRental revanue from shops?
In order to confirm that female sarpanches are able to effectively influence some measure of panchayat operations, we analyze whether villages reserved for female sarpanches allocate village expenditures differently. We here see that they do for health clinics/hospitals and salaries. *** p<0.01, ** p<0.05, * p<0.1, ~ p<0.15. S.e. in parenthesis. Data from CRRID-SCID Study of Fiscal Decentralisation.
No. of shops generating revanue% landless village householdsNo. candidates for 2003 sarpanch No. votes received by 2003 sarpanch winner
Total No. of all BPL householdsTotal No. of General Caste BPL Total No. of OBC BPL HouseholdsTotal No. of SC BPL Households
Shamlat land use in past 5 years
Funds for health?
Amt. of funds for health
Amt of Shamlat Land
Reserved For Female SarpanchSGRY funds at District LevelTotal Village Population 2001Total Village Population 1991
Amt. of funds for salaries
Funds for salaries?
Table 4: Female Bargaining Power and Dependent Variables of Interest(1) (2) (3)
VARIABLES Sex Ratio
-0.00998*** 0.0475***(0.00361) (0.0114)0.0122*** -0.0261*(0.00387) (0.0137)
0.0124(0.00855)0.00705(0.00764)
0.000113 0.000851* -5.60e-05(0.000154) (0.000438) (0.000327)-0.0741* -0.219* -0.103(0.0430) (0.119) (0.0912)-0.0618 0.116 0.0420(0.0489) (0.145) (0.104)-2.33e-05 -0.000232 -4.92e-05(7.20e-05) (0.000213) (0.000153)0.000545 -0.00130 -3.93e-05(0.000538) (0.00150) (0.000869)-0.00105 0.00280 0.00101(0.000895) (0.00248) (0.00191)0.00162 -0.00692 0.00140(0.00215) (0.00637) (0.00458)-0.000601 0.00948 -0.00644(0.00326) (0.00961) (0.00700)-0.00179* -0.00901*** 0.00363*(0.00105) (0.00307) (0.00211)0.00123 0.00639*** -0.00350**(0.000852) (0.00245) (0.00163)
Village Fixed Effects? Yes Yes Yes
Constant 0.876*** 2.252*** 1.642***(0.265) (0.726) (0.559)
Observations 841 516 516R-squared 0.053 0.078 0.062
We find whether female bargaining power, proxied here with maternal education, has the ability to change our dependent variables of interest: the sex ratio, the vaccination sex ratio, and the education sex ratio. We find that it does influence sex ratios and vaccination rates but does not change female school enrollment. Increases in maternal bargaining power decreases the sex ratio and increases the vaccination sex ratio. *** p<0.01, ** p<0.05, * p<0.1. Standard errors in parentheses. Data from Haryana SCID Gunn Internship 2008 Surveys.
average education of mother with children born at time t in villageaverage education of fathers with children born at time t in village
Vaccination Sex Ratio
School Sex Ratio
log village average per capita income
amt. shamlat land in village
average education of fathers with children age 6 at time t in village
average education of mothers with children age 6 at time t in village
village female child population in 1991
village houses in 1991
village households in 1991
village has doctor
village land area in 1991
village population in 1991
village male population in 1991village male child population in 1991
Table 5: Female Sarpanch Reservations and the Sex Ratio(1) (2) (3) (4) (5) (6) (7) (8) (9)
VARIABLES Sex Ratio Sex Ratio Sex Ratio Sex Ratio Sex Ratio Sex Ratio Sex Ratio Sex Ratio Sex Ratio-0.0122 0.00119 -0.00303(0.0205) (0.0267) (0.0377)
2.36e-05 1.34e-05 3.40e-06 4.05e-05 -2.38e-05 0.00160(1.76e-05) (3.14e-05) (3.78e-05) (4.32e-05) (7.75e-05) (0.00110)
-9.01e-09 2.61e-08 -1.58e-06(2.11e-08) (4.96e-08) (1.09e-06)
2.11e-05 7.96e-05 0.000292** 3.03e-05 0.000129 0.000244* 2.72e-05 0.000130 0.00188(4.30e-05) (6.16e-05) (0.000133) (4.26e-05) (8.32e-05) (0.000134) (4.33e-05) (8.33e-05) (0.00120)
-0.0544** -0.149*** -0.0326 -0.0329 -0.0255 -0.167*(0.0228) (0.0364) (0.0396) (0.0476) (0.0418) (0.0962)7.77e-05 9.72e-05 9.29e-05 9.72e-05 9.15e-05 8.67e-05(4.84e-05) (8.05e-05) (7.93e-05) (8.05e-05) (7.94e-05) (8.08e-05)0.000143***0.000123 0.000137 0.000123 0.000136 0.000123(4.91e-05) (8.94e-05) (8.74e-05) (8.94e-05) (8.75e-05) (8.93e-05)0.000102** 9.08e-05 9.34e-05* 9.08e-05 9.01e-05 9.83e-05*(5.18e-05) (5.76e-05) (5.67e-05) (5.76e-05) (5.71e-05) (5.78e-05)0.000287 0.000330* 0.000324* 0.000330* 0.000320 0.000321(0.000192) (0.000195) (0.000194) (0.000195) (0.000194) (0.000195)0.00383 0.00778** 0.00749** 0.00778** 0.00770** 0.00800**(0.00303) (0.00369) (0.00361) (0.00369) (0.00363) (0.00369)0.0224 0.0497* 0.0482* 0.0497* 0.0481* 0.0463*(0.0211) (0.0277) (0.0275) (0.0277) (0.0275) (0.0278)-0.000801 -0.00340 -0.00319 -0.00339 -0.00354 -0.00303(0.00300) (0.00367) (0.00357) (0.00367) (0.00364) (0.00367)-0.0225 -0.0496* -0.0481* -0.0496* -0.0481* -0.0463*(0.0211) (0.0277) (0.0275) (0.0277) (0.0275) (0.0278)7.06e-05 -3.97e-05 -3.88e-05 -3.98e-05 -3.78e-05 -4.00e-05(0.000219) (0.000228) (0.000227) (0.000228) (0.000227) (0.000228)
village land area in 1991 No No Yes No No Yes No No YesVillage Fixed Effects? No No Yes No No Yes No No YesConstant 0.485*** 0.356 0.947*** 0.458*** 0.0501 0.0752 0.453*** 0.00181 3.215
(0.0119) (0.241) (0.300) (0.0201) (0.357) (0.388) (0.0235) (0.369) (2.122)Observations 1524 747 564 1524 564 564 1524 564 564R-squared 0.000 0.085 0.138 0.001 0.135 0.138 0.002 0.136 0.141
landholdings of individual i in villagefather's education of individual i in villagenumber of brothers of individual i in village
village has had a female sarpanch reservationtime since village had a female sarpanch
caste of individual i in village
time squared since village had a female sarpanch amt. shamlat land in villagelog village average per capita incomeeducation of individual i in village We find whether female
sarpanch reservations can affect the sex ratio by regressing it on whether the village has had a female sarpanch, the time since having a female sarpanch, and the that time squared. We do not find that female sarpanch reservations can influence the sex ratio. *** p<0.01, ** p<0.05, * p<0.1. Standard errors in parentheses. Data from Haryana SCID Gunn Internship 2008 Surveys.
mother's education of individual i in villagenumber of sisters of individual i in villagebirthorder of individual i in village
if individual i in village has been ill in past year
Table 6: Female Sarpanch Reservations and Vaccination Sex Ratios(1) (2) (3) (4) (5) (6) (7) (8) (9)
VARIABLES
0.124** 0.102 0.163(0.0570) (0.100) (0.101)
-7.42e-05* -0.000207**-0.000164 -0.000283**2.93e-05 0.00203(4.40e-05) (8.29e-05) (0.000102) (0.000118) (0.000219) (0.00261)
1.12e-07* -1.68e-07 -2.18e-06(5.86e-08) (1.44e-07) (2.59e-06)
0.000138 0.000457* 0.000228 4.82e-05 0.000415* 0.000300 0.000119 0.000387 0.00314(0.000130) (0.000238) (0.000360) (0.000126) (0.000236) (0.000360) (0.000131) (0.000238) (0.00289)
0.00811 -0.130 -0.0778 -0.159 -0.132 -0.258(0.108) (0.129) (0.111) (0.130) (0.120) (0.239)-0.000275 -0.000209 -0.000244 -0.000210 -0.000236 -0.000229(0.000185) (0.000187) (0.000184) (0.000187) (0.000184) (0.000188)0.000293 0.000262 0.000277 0.000262 0.000289 0.000260(0.000208) (0.000210) (0.000206) (0.000210) (0.000206) (0.000210)3.76e-05 8.44e-05 3.29e-05 8.44e-05 6.13e-05 9.27e-05(0.000167) (0.000166) (0.000164) (0.000166) (0.000166) (0.000166)0.000866 0.000748 0.000728 0.000749 0.000741 0.000763(0.000695) (0.000691) (0.000691) (0.000691) (0.000691) (0.000692)-0.0184* -0.0192* -0.0195* -0.0192* -0.0208* -0.0189*(0.0107) (0.0108) (0.0106) (0.0108) (0.0107) (0.0108)0.108 0.0605 0.108 0.0602 0.109 0.0467(0.131) (0.133) (0.130) (0.133) (0.130) (0.134)0.0132 0.0112 0.00972 0.0112 0.0118 0.0119(0.00970) (0.00968) (0.00949) (0.00967) (0.00965) (0.00971)-0.108 -0.0610 -0.109 -0.0607 -0.110 -0.0472(0.132) (0.133) (0.130) (0.133) (0.130) (0.134)0.000785 0.000611 0.000804 0.000612 0.000767 0.000618(0.000554) (0.000553) (0.000549) (0.000553) (0.000549) (0.000554)
No Yes Yes No Yes Yes No Yes Yes
village Fixed Effects No No Yes No No Yes No No YesConstant 0.850*** -0.223 0.733 0.977*** 0.908 1.356 1.024*** 1.327 4.921
(0.0403) (1.017) (1.108) (0.0519) (1.099) (1.180) (0.0574) (1.156) (5.058)Observations 679 352 352 679 352 352 679 352 352R-squared 0.007 0.075 0.105 0.005 0.089 0.105 0.010 0.093 0.107
Vacc. Sex Ratio
Vacc. Sex Ratio
Vacc. Sex Ratio
Vacc. Sex Ratio
Vacc. Sex Ratio
Vacc. Sex Ratio
Vacc. Sex Ratio
Vacc. Sex Ratio
Vacc. Sex Ratio
We find whether female sarpanch reservations can affect the vaccination sex ratio by regressing it on whether the village has had a female sarpanch, the time since having a female sarpanch, and the that time squared. We do not find that female sarpanch reservations can influence the sex ratio after village fixed effects are taken into account. *** p<0.01, ** p<0.05, * p<0.1. Standard errors in parentheses. Data from Haryana SCID Gunn Internship 2008 Surveys.
number of sisters of individual i in villagebirthorder of individual i in village1991 Village characteristics?
landholdings of individual i in villagefather's education of individual i in villagenumber of brothers of individual i in villagemother's education of individual i in village
log village average per capita incomeeducation of individual i in village caste of individual i in village if individual i in village has been ill in past year
village has had a female sarpanch reservationtime since village had a female sarpanch reservationtime squared since village had a female sarpanch
amt. shamlat land in village
Table 7: Female Sarpanch Reservations and Education Sex Ratios(1) (2) (3) (4) (5) (6) (7) (8) (9)
VARIABLES0.145*** 0.0688 0.0591(0.0398) (0.0858) (0.0920)
-4.20e-05 -4.45e-05 -5.44e-05(3.34e-05) (7.24e-05) (9.23e-05)
4.43e-09 -1.33e-08 -4.31e-08(1.55e-08) (4.35e-08) (9.13e-08)
4.30e-05 -0.000151 -0.000337 -3.64e-05 -0.000153 -0.000313 -2.15e-05 -0.000153 -0.000300(8.84e-05) (0.000166) (0.000294) (8.73e-05) (0.000166) (0.000298) (8.83e-05) (0.000167) (0.000307)
-0.0388 -0.0186 -0.0618 -0.0285 -0.0588 -0.0348(0.0794) (0.100) (0.0810) (0.101) (0.0844) (0.104)-0.000289**-0.000310** -0.000291**-0.000311** -0.000298**-0.000314**(0.000142) (0.000147) (0.000142) (0.000147) (0.000142) (0.000147)3.63e-05 6.79e-05 3.68e-05 6.84e-05 3.86e-05 6.93e-05(0.000154) (0.000160) (0.000154) (0.000160) (0.000154) (0.000160)0.000175* 0.000170 0.000186* 0.000171 0.000190* 0.000172*(0.000103) (0.000104) (0.000101) (0.000104) (0.000101) (0.000104)0.00102** 0.000951* 0.00101** 0.000953* 0.00103** 0.000957*(0.000479) (0.000488) (0.000481) (0.000488) (0.000481) (0.000488)0.00779 0.00680 0.00719 0.00681 0.00739 0.00684(0.00691) (0.00724) (0.00697) (0.00724) (0.00701) (0.00724)0.0416 0.0377 0.0418 0.0376 0.0418 0.0374(0.0600) (0.0616) (0.0600) (0.0616) (0.0600) (0.0616)0.00357 0.00390 0.00418 0.00396 0.00442 0.00408(0.00704) (0.00720) (0.00698) (0.00719) (0.00697) (0.00719)-0.0408 -0.0369 -0.0410 -0.0368 -0.0410 -0.0366(0.0600) (0.0616) (0.0600) (0.0616) (0.0600) (0.0616)-0.000678* -0.000660* -0.000681* -0.000660* -0.000682* -0.000660*(0.000351) (0.000356) (0.000351) (0.000356) (0.000352) (0.000356)
1991 Village Characteristics? No Yes Yes No Yes Yes No Yes YesVillage Fixed Effects? No No Yes No No Yes No No YesConstant 0.777*** -0.0629 -0.107 0.870*** 0.198 0.103 0.820*** 0.121 0.187
(0.0241) (0.719) (0.816) (0.0400) (0.779) (0.869) (0.0296) (0.807) (0.986)Observations 881 295 295 881 295 295 881 295 295R-squared 0.015 0.154 0.158 0.002 0.153 0.158 0.000 0.153 0.157
Edu Sex Ratio
Edu Sex Ratio
Edu Sex Ratio
Edu Sex Ratio
amt. shamlat land in villagelog village average per capita income
Edu Sex Ratio
Edu Sex Ratio
Edu Sex Ratio
Edu Sex Ratio
village has ever had a female sarpanch reservationtime since village had a female sarpanch reservationtime squared since village had a female sarpanch reservation
Edu Sex Ratio
education of individual i in village
caste of individual i in village
mother's education of individual i in villagenumber of sisters of individual i in villagebirthorder of individual i in village
We find whether female sarpanch reservations can affect the school enrollment sex ratio by regressing it on whether the village has had a female sarpanch, the time since having a female sarpanch, and the that time squared. We do not find that female sarpanch reservations can influence the sex ratio after village fixed effects are taken into account. *** p<0.01, ** p<0.05, * p<0.1. Standard errors in parentheses. Data from Haryana SCID Gunn Internship 2008 Surveys.
if individual i in village has been ill in past yearlandholdings of individual i in villagefather's education of individual i in villagenumber of brothers of individual i in village
Table 8a: Female Sarpanch Reservation Interactions and Sex Ratio
(1) (2) (3) (4)VARIABLES
-0.000381**(0.000153)
0.0451(0.0548)
0.00774(0.00880)
-0.0130(0.00819)
everFsar 0.0569 -0.312 -0.0294 0.114(0.0446) (0.378) (0.0598) (0.0763)
momEduYearInt -0.0155* -0.0106(0.00819) (0.00713)
dadEduYearInt 0.0174** 0.0211***(0.00740) (0.00774)
0.000301** 0.000254* 0.000238* 0.000239*(0.000135) (0.000135) (0.000135) (0.000135)-0.0203 -0.0459 -0.0514 -0.0528(0.0473) (0.0496) (0.0479) (0.0478)0.000103 9.51e-05 0.000101 0.000101(8.02e-05) (8.06e-05) (8.22e-05) (8.20e-05)0.000122 0.000125 0.000115 0.000112(8.89e-05) (8.95e-05) (9.02e-05) (9.01e-05)9.10e-05 8.88e-05 8.64e-05 8.38e-05(5.74e-05) (5.77e-05) (5.77e-05) (5.76e-05)0.000312 0.000338* 0.000316 0.000306(0.000194) (0.000195) (0.000195) (0.000194)0.00678* 0.00780** -0.00143 -0.00129(0.00370) (0.00369) (0.00538) (0.00537)0.0490* 0.0494* 0.0552** 0.0572**(0.0276) (0.0277) (0.0278) (0.0277)-0.00387 -0.00347 0.00219 0.00236(0.00366) (0.00367) (0.00511) (0.00510)-0.0490* -0.0493* -0.0552** -0.0572**(0.0276) (0.0277) (0.0278) (0.0277)-2.40e-05 -3.00e-05 1.96e-05 3.09e-06(0.000227) (0.000228) (0.000228) (0.000228)
1991 Village Characteristics? Yes Yes Yes YesVillage Fixed Effects? Yes Yes Yes YesConstant -0.00553 0.176 0.179 0.133
(0.364) (0.379) (0.367) (0.365)Observations 564 564 564 564R-squared 0.148 0.139 0.149 0.152
Sex Ratio Sex Ratio Sex Ratio Sex Ratio
We find whether female sarpanch reservations interacted with shamlat land, average village earnings, maternal education, or paternal education have an affect on the sex ratio. We find that only the shamlat land interaction has significance, showing that female sarpanches must be properly funded to have any affect. *** p<0.01, ** p<0.05, * p<0.1. Standard errors in parentheses. Data from Haryana SCID Gunn Internship 2008 Surveys.
birthorder of individual i in village
if individual i in village has been ill in past yearlandholdings of individual i in villagefather's education of individual i in villagenumber of brothers of individual i in village
amt shamlat land * had a female sarpanch reservationlog village average per capita income * had a female sarpanch ave. maternal education * had a female sarpanch reservationave. paternal education * had a female sarpanch reservation
amt. shamlat land in villagelog village average per capita income
education of individual i in village
caste of individual i in village
mother's education of individual i in villagenumber of sisters of individual i in village
Table 8b: Female Sarpanch Reservation Interactions and Vaccination Sex Ratio
(1) (2) (3) (4)VARIABLES
0.00102**(0.000445)
-0.199(0.171)
-0.0623**(0.0275)
0.00682(0.0261)
everFsar 0.00198 1.529 0.461*** 0.0881(0.123) (1.182) (0.177) (0.241)
momEduYearInt 0.0512** 0.0183(0.0239) (0.0194)
dadEduYearInt -0.0527** -0.0599**(0.0217) (0.0251)
0.000290 0.000113 0.000170 0.000213(0.000528) (0.000373) (0.000361) (0.000365)
-0.0823 -0.0683 -0.0755(0.135) (0.129) (0.130)
-0.000202 -0.000209 -0.000260 -0.000249(0.000186) (0.000187) (0.000189) (0.000190)0.000254 0.000249 0.000293 0.000307(0.000209) (0.000210) (0.000210) (0.000212)7.99e-05 0.000100 0.000130 0.000122(0.000165) (0.000167) (0.000165) (0.000166)0.000857 0.000580 0.000904 0.000879(0.000689) (0.000706) (0.000684) (0.000691)-0.0160 -0.0189* 0.00403 0.00183(0.0108) (0.0108) (0.0134) (0.0135)0.0678 0.0625 -0.0447 -0.00430(0.133) (0.133) (0.135) (0.135)0.0124 0.0115 0.00583 0.00631(0.00963) (0.00968) (0.0117) (0.0118)-0.0682 -0.0630 0.0443 0.00385(0.133) (0.133) (0.135) (0.135)0.000551 0.000562 0.000264 0.000416(0.000550) (0.000555) (0.000555) (0.000558)
1991 Village Characteristics? Yes Yes Yes YesVillage Fixed Effects? Yes Yes Yes YesConstant -0.248 0.521 0.350 0.644
(0.732) (1.122) (1.101) (1.102)Observations 352 352 352 352R-squared 0.119 0.108 0.136 0.123
Vacc. Sex Ratio
Vacc. Sex Ratio
We find whether female sarpanch reservations interacted with shamlat land, average village earnings, maternal education, or paternal education have an affect on the vaccination sex ratio. We find that only the shamlat land interaction has significance, showing that female sarpanches must be properly funded to have any affect. *** p<0.01, ** p<0.05, * p<0.1. Standard errors in parentheses. Data from Haryana SCID Gunn Internship 2008
Vacc. Sex RatioVacc. Sex Ratio
mother's education of individual i in village
education of individual i in village
caste of individual i in village
number of sisters of individual i in village
birthorder of individual i in village
if individual i in village has been ill in past yearlandholdings of individual i in villagefather's education of individual i in villagenumber of brothers of individual i in village
amt shamlat land * had a female sarpanch reservationlog village average per capita income * had a female sarpanch ave. maternal education * had a female sarpanch reservationave. paternal education * had a female sarpanch reservation
amt. shamlat land in villagelog village average per capita income
Table 8c: Female Sarpanch Reservation Interactions and Educational Sex Ratio
(1) (2) (3) (4)VARIABLES
-0.000153(0.000380)
0.259(0.177)
0.0332(0.0203)
0.0292(0.0206)
everFsar 0.0896 -1.724 -0.0585 -0.192(0.119) (1.219) (0.116) (0.197)
momEduYearInt -0.0180 -0.00347(0.0184) (0.0148)
dadEduYearInt 0.0124 0.00401(0.0146) (0.0154)
-0.000321 -0.000291 -0.000316 -0.000333(0.000297) (0.000295) (0.000295) (0.000295)-0.0177 -0.0692 -0.0403 -0.0369(0.101) (0.106) (0.102) (0.102)-0.000310** -0.000301** -0.000290* -0.000286*(0.000147) (0.000147) (0.000150) (0.000150)6.85e-05 5.33e-05 6.21e-05 5.20e-05(0.000160) (0.000160) (0.000162) (0.000162)0.000170 0.000172* 0.000162 0.000166(0.000104) (0.000104) (0.000105) (0.000105)0.000943* 0.00100** 0.000880* 0.000915*(0.000489) (0.000488) (0.000489) (0.000489)0.00658 0.00716 0.000894 0.00138(0.00727) (0.00723) (0.00978) (0.00978)0.0375 0.0329 0.0470 0.0418(0.0617) (0.0616) (0.0623) (0.0624)0.00365 0.00206 0.00335 0.00273(0.00723) (0.00729) (0.00933) (0.00936)-0.0367 -0.0321 -0.0463 -0.0410(0.0617) (0.0616) (0.0623) (0.0624)-0.000662* -0.000660* -0.000631* -0.000626*(0.000356) (0.000355) (0.000357) (0.000358)
1991 Village Characteristics? Yes Yes Yes YesVillage Fixed Effects? Yes Yes Yes YesConstant -0.123 0.253 0.0943 0.0941
(0.818) (0.850) (0.823) (0.826)Observations 295 295 295 295R-squared 0.158 0.164 0.168 0.166
Edu. Sex Ratio Edu. Sex Ratio
We find whether female sarpanch interaction variables can change the education sex ratio. We find no affect though the shamlat interaction variable showed significance for the other sex ratio measures. This supports our hypothesis that female sarpanch affects these sex ratio measures through increasing maternal bargaining power; since we found maternal bargaining power has no affect on school sex ratios. *** p<0.01, ** p<0.05, * p<0.1. Standard errors in parentheses. Data from Haryana SCID Gunn Internship 2008 Surveys.
Edu. Sex Ratio
Edu. Sex Ratio
mother's education of individual i in villagenumber of sisters of individual i in village
birthorder of individual i in village
if individual i in village has been ill in past yearlandholdings of individual i in villagefather's education of individual i in villagenumber of brothers of individual i in village
amt shamlat land * had a female sarpanch reservation
log village average per capita income
education of individual i in village
caste of individual i in village
log village average per capita income * had a female sarpanch ave. maternal education * had a female sarpanch reservationave. paternal education * had a female sarpanch reservation
amt. shamlat land in village
Reference List: Agarwal, Bina. 1997. “Bargaining and Gender Relations: Within and Beyond the Household,” Feminist Economics 3(1): 1-51. Arokiasamy, Perianayagam. 2004. "Regional Patterns of Sex Bias and Excess Female Child Mortality in India," Population Vol 59. ISSN 1634-2941. Accessed online at: http://www.cairn.info/article_p.php?ID_ARTICLE=POPE_406_0831 Ashish Bose, 2001. “Fighting Female Foeticide: Growing Greed and Shrinking Child Sex Ratio.” Economic and Political Weekly. Baldez, L, 2003. "Elected Bodies: The Gender Quota Law for Legislative Candidates in Mexico" Paper presented at the annual meeting of the American Political Science Association, Philadelphia Marriott Hotel, Philadelphia, PA. Beaman, Lori, Chattopadhyay, Raghabendra, Duflo, Esther, Pande, Rohini and Topalova, Petia B, 2008. “Powerful Women: Does Exposure Reduce Bias?” MIT Department of Economics Working Paper No. 08-14; HKS Working Paper No. RWP08-37. Available at SSRN: http://ssrn.com/abstract=1162358 Bhat R.L. and Namita Sharma, 2006. “Missing Girls: Evidence from Some North Indian States” Indian Journal of Gender Studies; 13; 351 Borooah, Vani, 2004. “Gender bias among children in India in their diet and immunisation against disease.” Social Science & Medicine 58. pp1719–1731 Central Chronicle (Bhopal). 2008. “Benefit of Ladli Laxmi Yojana extended to over 40 thousand girls,” India Environment Portal. Accessed online at: http://www.indiaenvironmentportal.org.in/node/246790 Chamarbagwala, Rubiana; Ranger, Martin. 2006. “India's Missing Women: Disentangling Cultural, Political and Economic Variables,” Center for Applied Economics and Policy Research, Economics Department, Indiana University Bloomington, Caepr Working Papers.pp.34. Chattopadhyay, Raghabendra, Ester Duflo, 2004. “Women as Policy Makers: Evidence from a Randomized Policy Experiment in India.” Econometrica, Vol. 72, No. 5. pp 1409–1443 Dale, Stephen. 2006. “India's Missing Daughters,” The International Development Research Centre. Accessed online at http://www.idrc.ca/en/ev-95719-201-1-DO_TOPIC.html on April 15th, 2009. Dasgupta, Partha. An Inquiry Into Well-being and Destitution. Oxford University Press, 1995. pp 323
Desai, Sonalde, Sonya Rastogi and Reeve Vanneman. 2005. ‘Gender Differences in Child Survival in India: What Do We Know? ’ Paper presented in the International Population Conference at the International Union for the Scientific Study of Population, France, 18–23 July, http://www.iussp.princeton.edu/download.aspx. Duflo, Esther, 2003. “Grandmothers and Granddaughters: Old Age Pensions and Intra-household Allocation in South Africa.” The World Bank Economic Review, Vol. 17. Duflo, Esther, Petia Topalova, 2005. “Unappreciated Service: Performance, Perceptions, and Women Leaders in India,” MIT Mimeo. Glick, Peter and David E. Sahna, 2000. “Schooling of girls and boys in a West African country: the effects of parental education, income, and household structure,” Economics of Education Review. Volume 19, Issue 1, Pages 63-87 Guilmoto, Christophe, 2007. “Characteristics of Sex-Ratio Imbalance in India, and future Scenarios,” 4th Asia Pacific Conference on Reproductive and Sexual Health and Rights: Hyderabad, India. LPED/IRD, Paris. United Nations Population Fund (UNFPA). Jha, Prabhat, Rajesh Kumar, Priya Vasa, Neeraj Dhingra, Deva Thiruchelvam, Rahim Moineddin, 2006. "Low male-to-female sex ratio of children born in India: national survey of 1·1 million households." The Lancet 367(9506), pp 211 – 218. Liu, Haoming., March 2008. “The impact of women's power on child quality in rural China.” China Economic Review 19(1). pp 101-115. Mishra, Vinod . T. K. Roy and Robert D. Retherford. "Sex Differentials in Childhood Feeding, Health Care, and Nutritional Status in India," Population and Development Review. 30(2), pp 269 - 295 McSweeney, Brenda Gael (ed.), Mieke Windecker (ed), Margaret Hartley (ed.) 2008.“Another Side of India: Gender, Culture and Development,” United Nations Educational, Scientific and Cultural Organization, Paris. Office of the Registrar General and Census Commissioner, India. “Census of India 2001.” Accessed online at: http://www.censusindia.gov.in/default.aspx Pal, Sarmistha, 2003. “How much of the Gender Differences in child school enrolment can be explained? Evidence from Rural India,” Working paper, Cardiff Business School. Raju, Saraswati, 2005. “Regional Patterns of Female Participation in the Labor Roce of Urban India.” The Professional Geographer 34(1). pp 42-49. Rao, Vijayendra, 2008. “Tokenism or Agency? The Impact of Women’s Reservations on Village Democracies in South India” Economic Development and Cultural Change. Vol. 56. pp501-530.
Sachar RK; Verma J; Dhawan S; Prakash V; Chopra A; Adlaka R. 1990. "Sex bias in health and medical care allocation.” Indian Journal of Maternal and Child Health. Apr-Jun 1(2). pp 63-5. Sen, Amartya, 1990. “Gender and cooperative conflicts’” in Irene Tinker, ed., Persistent inequalities: women and world development. New York, New York, Oxford University Press. pp123-149. Sen, Amartya, 1990. “More Than 100 Million Women Are Missing,” New York Review of Books 37(2). Singh, Hoshiar, 1994. “Constitutional Base for Panchayati Raj in India: The 73rd Amendment Act” Asian Survey, 34(9). pp. 818-827 Singh, Meeta and Vasu Mohan, 2005. “The Rise of Sex Selection in India,” Democracy at Large Vol. 2, No. 1. Songa, Lina., Simon Appletona and John Knightb, September 2006. “Why Do Girls in Rural China Have Lower School Enrollment?” World Development. 34(9). pp 1639-1653 The Government of India. “The Indian Constitution (Seventy-Third Amendment) Act, 1992” Accessed online at: http://indiacode.nic.in/coiweb/amend/amend73.htm Wu, Kin Bing., Pete Goldschmidt, Christy Kim Boscardin, and Mehtabul Azam. 2007. “Ch 5: Girls in India: Poverty, location, and social disparities,” in Maureen Lewis and Marlaine Lockheed, ed., Exclusion, gender and education: case studies from the developing world. Washington, D.C., Center for Global Development, 2007. p 119-143. Thomas, Thomas., 1990. “Intra-household resource allocation: An inferential approach.” The Journal of Human Resources 25(4):635{664. Thomas, Thomas., 1994. “Like Father, like Son; Like Mother, like Daughter: Parental Resources and Child Height.” The Journal of Human Resources Vol. 29, No. 4, Special Issue: The Family and Intergenerational Relations, pp. 950-988.