Content uploaded by Tooraj Jamasb
Author content
All content in this area was uploaded by Tooraj Jamasb on Apr 23, 2020
Content may be subject to copyright.
WORKING PAPER
Copenhagen School of Energy Infrastructure | CSEI
Ashish Kumar Sedai
Rabindra Nepal
Tooraj Jamasb
Electrification and Socio-Economic Empowerment
of Women in India
CSEI Working Paper 2020-
10
CBS Department of Economics 10-
2020
Electrification and Socio-Economic Empowerment
of Women in India
Ashish Kumar Sedai∗Rabindra Nepal †Tooraj Jamasb ‡
April 16, 2020
Abstract
This study examines the effect of quality of electrification on empowerment of women
in terms of economic autonomy, agency, mobility, decision-making abilities, and time
allocation in fuel collection in India. It moves beyond the consensus of counting elec-
trified households as a measure of progress in gender parity, and analyzes how the
quality of electrification affects women’s intra-household bargaining power, labor sup-
ply and fuel collection time. We develop a set of indices using principal component
analysis from a large cross-section of gender-disaggregated survey. We use two stage
least squares instrumental variables regression to assess the causal effect of access and
hours of electricity on women’s empowerment using geographic instrumental variables
along with district and caste fixed effects. The results show that quality of electrifi-
cation has significant positive effects on all empowerment indices. However, the effect
differs at the margin of deficiency, location, living standards and education. The study
recommends revisiting the paradigm of access to electrification and women empower-
ment by focusing on the quality of not only extensive but also intensive electrification
to enhance life and economic opportunities for women and their households.
Keywords: Women Empowerment, Quality of Electrification, Instrumental Variables
JEL Codes: D13, D63, H42, Q43
∗Department of Economics, Colorado State University. Email: ashish.sedai@colostate.edu
†Department of Acconting Economics and Finance, University of Wollongong. Email: rnepal@uow.edu.au
‡Department Economics, Copenhagen Business School. Email: tj.eco@cbs.dk
1 Introduction
Gender equality is desirable in itself and is necessary for holistic economic development
[Duflo, 2012]. According to the United Nations “achieving gender equality and empowering
all women and girls” emphasizes that an economy cannot achieve its potential without equal
gender participation [Fukuda-Parr, 2016]. Though there have been several policy initiatives
to foster this egalitarian view, a substantial gap still persists [Duflo, 2012, Hendriks, 2019].
In order to promote gender equality, these initiatives have focused on improving women’s
economic engagement through financial inclusion [World Bank, 2013], political participation
as well as equal land and property rights [Cherchi et al., 2019].
Access to electricity and other sources of energy have been widely discussed as an effec-
tive pathway to empower women [Samad and Zhang, 2019, Khandker et al., 2014, Gould
and Urpelainen, 2018], but the discussion of empowering women through reliable and ade-
quate electricity at the intensive margin has remained elusive [Kennedy et al., 2019]. Thus,
empirical evidence on the effect of quality of electricity on women’s intra-household resource
allocation, bargaining, time allocation and safety has been lacking.
Women in developing countries spend more time in the household than men hence, the
quality of household electrification disproportionately affects them [Dinkelman, 2011, O’Dell
et al., 2014]. However, the focus in the literature has been whether access to electricity
benefits the households, especially women, through increases in labor supply, schooling of
children, household income and expenditure as well as women’s economic decision making
ability and mobility [Samad and Zhang, 2019, Khandker et al., 2014, Rao, 2013, O’Dell et al.,
2014].
According to the Government of India reports, there is no scope for improvement as India
has already achieved 100% electrification of all villages (Saubhagya Report 2019). However,
intensive margin requires policy attention in India because of the official definition of elec-
trification laid down by the government, “a village is electrified if the basic infrastructure is
in place; power is being supplied to schools, health centers and other public places; and at
1
least 10% of households are receiving electricity” [Agrawal et al., 2020].
Despite the micro evidences of the benefits women derive through electrification, a macro
examination of the impact of the quality of electrification and its potential for empowering
women has remained elusive. This study analyzes quintiles of electricity deficiency within
the household and examines the causal effect of falling into those quintiles on women’s socio-
economic freedom, agency, mobility and decision making ability. We compare the impact
of electrification on poor and non-poor, rural and urban women using the novel framework
of women empowerment laid down by Kabeer [1999] in defining a woman’s autonomy: (i)
economic freedom, referring to women’s ownership, control and decision making ability over
economic resources, allocation and labor supply, (ii) agency in household decision making
ability and reproductive freedom and, (iii) mobility, in terms of freedom to travel alone.
We operationalizes women’s empowerment as a step towards gender equality, access to and
control over resources and power to influence matters that concern or affect them [Kabeer,
1999].
We use the India Human Development Survey (IHDS), 2012, which includes hours of
electricity in a day, women’s engagement in household and personal care activities as well as
day to day decision making. We analyze the effect of electrification on women empowerment
variables related to economic freedom, reproductive freedom, mobility and decision-making
ability. We use (i) Ordinary Least Squares (OLS) with district and caste fixed effects (FE)
and (ii) Two Stage Least Squares instrumental variable (2SLS-IV-FE) regressions with a
geographic instrumental variable Fang [2003][Khandker et al., 2014]Rao [2013][Chakravorty
et al., 2014], average hours of electricity at the village/PSU1level as an instrument along
with district and caste fixed effects. We use principal component analyses (PCA) of nineteen
empowerment variables to create five indices and regress them on outages and quintiles of
power outages.
The study finds strong positive causal effect of power outages on all empowerment indices
1The data used includes observations from rural and urban areas. We denote the regional aggregation of
participation through villages in rural areas and through PSUs in urban areas, Desai and Vanneman [2018].
2
at the national level. Outages have the strongest effect on poor women’s economic free-
dom, agency and household decision making. They also have significant effects on women’s
agency and mobility, but the magnitude of the effect is smaller than the effects of outages
on economic and household decision making. The instrument: Mean village/PSU level elec-
trification satisfies the first stage weak, over and under identification tests, and has a strong
effect on outages at the household level. The second stage instrumental variable regres-
sions support the hypothesis of differential impacts of energy deficit on women depending on
household income, location and women’s education. Education matters the most, followed
by location and income status for women in conjunction with the quality of electrification
at the household level.
Section 2 reviews the literature on the significance of electrification for women with focus
on the Indian context. Section 3 presents the theoretical and econometric model, and the
data used in the study. Section 4 presents and contrasts the results from the OLS Fixed
effects and 2SLS-IV regressions. Section 5 discusses the results. Section 6 concludes.
2 Literature Review
The discussion on the effect of quality o electrification is divided into two sub-sections. The
first section reviews the literature on the effects of electrification and its impact on women’s
social and economic outcomes worldwide. The second section looks at the literature in terms
of deficiency of electrification in India and the programs undertaken by the government to
address the problem of electrification.
2.1 Background & Literature
Previous research has shown that access to electrification empowers women in myriad ways,
first and foremost through improved lighting in rural areas which increases rural household’s
income, business income and wages [Chakravorty et al., 2014]. A study by Kanagawa and
3
Nakata [2008] in Assam, India, shows that women in households without electricity hardly
undertook any reading irrespective of their level of education. Since availability of lighting
extends the effective workday, it allows women to leave certain household chores for the night
enabling them to participate in more formal economic activity during the day [Kanagawa and
Nakata, 2008]. However, these outcomes depend heavily on the reliability of the electricity
service. In case of erratic power supply, these benefits may fail to materialise. For instance,
electrified households receiving no electricity at night may not be significantly different from
households without grid connection. This has a direct bearing on the economic condition
of women. Deficiency of electricity through demand or supply channels can both lead to
sub-optimal time allocation to home production by women which reduces their labor supply
and increases unpaid care work [Dinkelman, 2011].
Lack of electricity dissatisfies women as children have less flexible study time at home
[Kennedy et al., 2019]. From a non-economic standpoint, street and household lighting al-
lows women to commute after dark [Standal and Winther, 2016]. Unlike kerosene lighting,
electricity provides good quality light required for reading which has potentially long-term
productivity impacts [Chakravorty et al., 2014]. Along with electric lighting, electrical ap-
pliances reduce time and effort for household chores leaving room for other productive activ-
ities [Standal and Winther, 2016]. Time conservation reduces the engagement in intensive
household activities allowing leisure and improvement in living standards while providing op-
portunity to participate in the labour market [Dinkelman, 2011] [Samad and Zhang, 2019].
The study by Standal and Winther [2016] in West Bengal, Uttar Pradesh and Jharkhand
finds that electricity affects everyday life in terms of providing important resources and en-
hancing women’s opportunities to perform their role as care workers more efficiently and in
a qualitatively better way.
Often, women spend the majority of their time in cooking and collecting firewood pri-
marily because biomass fuels are perceived to be cost effective, however, they are highly
polluting and adversely affect the health of women [Bansal et al., 2013]. Their lower calorific
4
value necessitates prolonged cooking hours thereby enhancing the exposure to hazardous
emissions. Although the use of Liquified Petroleum Gas (LPG) has been widely discussed as
the alternative to clean cooking, the principal constraint to widespread adoption is the fuel
cost [Gould and Urpelainen, 2018] coupled with weak bargaining power of women in rural
households for having LPG connections [Bansal et al., 2013]. Household electrification holds
tremendous potential for improving the status of women and enhancing their quality of life.
Standal and Winther [2016] show that across all levels of education, women in electrified
households were much more likely to read during a day compared to unelectrified households.
Virtually no reading by women took place in households without access to electricity which
has clear implications for women continuing their education [Kanagawa and Nakata, 2008].
Further, as households add new electrical appliances, the impact on women’s life is even
greater. They allow women to perform household chores with greater efficiency, for instance,
the amount of time required to process food or spices can be considerably reduced through the
use of a simple grinder. Aside, electrification is often associated with information by exposure
to media such as radio and television, which plays a crucial role in raising awareness and
educating women [Samad and Zhang, 2019]. Exposure to radio and television spurs fertility
decline via increased use of contraception [Stephenson et al., 2006]. It also enables women to
lessen the grip of traditional and cultural norms and participate more actively in the society
[Standal and Winther, 2016].
Empowerment of women through electrification underscores the need for adequate access
to energy as a means towards inclusive development. The principal question is the discovery
of an optimal price and quantity given the constraints. Burgess et al. [2020] show how treating
electricity as a freebie creates economic inefficiency because it develops a social norms that
everyone deserves power independent of payment, subsidies, theft, and nonpayment. This
creates a sinking circle whereby electricity distribution companies lose money, government-
owned distribution companies restrict access and hours of supply is no longer governed by
market forces [Burgess et al., 2020]. The link between payment and supply is severed, thus
5
reducing customers’ incentives to pay, hence the equilibrium outcome is uneven, and sporadic
access undermines growth [Burgess et al., 2020].
Policy makers face a dilemma wherein electricity supply is quintessential for household
welfare, especially women’s empowerment, but providing electricity for free does not seem
to be a feasible solution. In this context, this study contributes to understanding how ad-
ditional hours of electricity affects women’s empowerment depending on the income levels
in rural and urban areas. Better knowledge of the marginal benefits and costs to women
of hours of electricity could provide policy makers with better understanding of appropriate
pricing mechanisms depending on the margin of deficiency and the objective of gender par-
ity. Faults in the existing structures of the public-private distribution grid cannot sidestep
the significance of electrification for women and understates the potential impact of pro-
viding energy services to women who would not sit idly. Although designing appropriate
pricing mechanisms is important and energy supply to women through targeted measures is
a justified concern, it is beyond the scope of this study.
2.2 The Context in India
Domestic energy concerns loom large for women in a country where 65% of the population
resides in rural areas with uneven and often unreliable access to electricity supply [Samad
and Zhang, 2019], also shown by figure 1. Despite making great strides to improve the
access to electricity supply since 2005, reliability of the supply has largely been neglected
by the government which can be the potential actor to empower women by enabling them
to be more efficient in household activities, enhance their educational and awareness levels,
enter workforce and start businesses 2. Often uneven and irregular access acts as a serious
impediment to welfare outcomes that electrification holds for women.
The early focus on electrification in India had been on building up industrial capacity
2Authors elaboration from the average hours of electricity in a day at the household level from IHDS,
2005-2012. In 2005, the average household electricity hours in a day was 16 hours in 2005 which stagnated,
even reduced to 15.68 hours a day in 2012.
6
and facilitating the use of electric pumps for irrigation purposes. It was only in the 1980s
that the central planning system started considering electricity as a basic input to household
production [Palit et al., 2014]. A major impetus was observed in 2005 following the forward-
looking Electricity Act of 2003 (EA), when the government launched Rajiv Gandhi Grameen
Vidyutikaran Yojana (RGGVY) which led to a large increase in the village electrification
rate from 59% in 2000 to 74% by 2010, albeit at the expense of massive losses to distribution
companies due to subsidies and thefts [Pargal and Ghosh Banerjee, 2014]. Though the
reforms succeeded in increasing the rate of electrification, it failed at ensuring the reliability
of the service as evidenced through the IHDS survey 2005-2012 and also shown through high
frequency satellite images by [Min et al., 2017]. In fact, Min et al. [2017] found that many
villages that were officially deemed as ‘electrified’ under RGGVY remained in dark for years
after completion of electrification projects.
In 2015, RGGVY was subsumed in Deen Dayal Upadhyay Gram Jyoti Yojana (DDUGJY)
which aimed to provide continuous power supply to the households [Jain, 2018]. In 2019,
India achieved 100% village electrification as per the Saubhagya, government of India re-
port [Mehra and Bhattacharya, 2019]. A village is deemed ‘electrified’ if at least 10% of
the households along with public spaces like school, post office, and basic infrastructure
have access to electricity [Nouni et al., 2008]. However, according to Smart Power India
report (2017), approximately 237 million Indians lacked access to reliable electricity. In the
government’s efforts to electrify communities, the emphasis has been on providing grid con-
nections to increase the count of electrified households while the reliability concerns have
largely escaped the attention of policy makers. Not only does it mask the problem of poor
household’s access, it also does not consider the quality of electricity service provided which
is essential for improving the quality of life for women. As a result, the impetus has been
on counting the electrified households, neglecting the reliability of the service. However, to
reap the socio-economic benefits of electrification, it is important to move beyond the binary
of classifying households as either having access to electricity or not and argue for a richer
7
definition which also incorporates the quality of the provided service.
In India, fuel wood collection and cooking forms a large part of household work whose
burden primarily falls upon women. According to Sharma et al. [2019], women in rural
areas in Jharkhand spend an average of 2hrs 45 mins on cooking activities and 2 hrs 30
mins on an average to collect firewood on a typical day. Not only is it arduous, the usage
of traditional fuel is extremely hazardous to health, especially for inhabitants who stay
inside the house for longer duration, and additionally, environmentally degrading. In India,
the ambient pollutant (PM10) concentration was drastically high around 20000µg/m3, even
higher near the cooking locations Sharma et al. [2019]. This is higher than the benchmark set
by Environmental Protection Act, 1986 around 150µg/m3. Further, health tests to measure
smoke levels in the lungs found that women had an average carbon monoxide (CO) reading of
7.77ppm, while children had CO levels similar to those from smoking about seven cigarettes
per day Parikh [2011]. These findings show grave impacts of using biomass fuels on women’s
health and wellbeing, in light of the fact that 80% of energy needs are met by biomass fuels.
Electrification is often found to spur the switch to cleaner fuels such as Kerosene, Liquified
Petroleum Gas and electrical cooking which saves women from the collection and usage of
biomass, and allows them to use their time in more productive activities.
Exposure to electronic media, especially television has made women more assertive,
thereby, increasing their autonomy in the household. Jensen and Oster [2009] find that
the access to cable television results in lower acceptance of spousal abuse, lower preference
for sons and greater likelihood of sending girls to school in rural India. Television viewing
may also improve women’s domestic productivity and welfare through greater knowledge.
Home-based technology such as electricity can reduce household dependence on girls’ labor
and reduces the opportunity cost of sending girls to school. K¨ohlin et al. [2011]. However,
for energy access to translate into empowerment, it needs to be reliable and affordable. Ac-
cording to K¨ohlin et al. [2011], benefits emerging from the improved access to energy are
substantial for women but the size of benefits is constrained by frequent power cuts. There-
8
fore, providing reliable electricity supply, instead of merely counting electrified households,
should be an important policy priority.
3 Methodology
The model is based on the intra-household resource allocation and uses a Nash bargaining
solution [Cohen and Glazer, 2017] under the assumption that an additional hour of elec-
tricity at the margin of electricity deficiency has significant effects on the socio-economic
outcomes of women in the household where the modelled household consists of a wife and
husband. Following gender disparity in relative bargaining power and the unequal time spent
by women in the house [Duflo, 2012, Kabeer, 1999], we assume that household electrifica-
tion disproportionately benefits women, and that the path to increased household welfare is
largely through the wife [Duflo, 2012, Mayoux and Anand, 1995]. Thus, the women in the
household have a higher preference for more electricity hours compared to their husband.
The objective is to maximize the joint utility of the household, the husband and wife
who live for two periods. The husband’s preferences for the indivisible good are positive but
lower than the wife’s. We test the comparative static effect of additional electricity hours
on the wife’s bargaining power in intra-household resource allocations. The model shows
that if the wife’s preference δ≥0.5 for household electricity is more than the husband’s,
then a marginal increase in expenditure on the purchase of an additional hour of electricity
increases the wife’s bargaining power. This model has been extensively used in the field of
micro-finance and women empowerment but has been relatively under explored in the field
of energy and women empowerment.
3.1 Theoretical Model
In the first period, the wife expresses demand for an electricity Dfor which her preference
δ(0,1) is higher than the preference of the husband (1 −δ). To acquire more hours of
9
electrification the household must save S(we alternatively use savings to reflect expenditure
on the purchase of electricity). For simplicity, we assume that the household earns the same
income in both periods, although the assumption can be relaxed to account for inflation
adjusted income. Both the husband and the wife are risk averse and have utility functions
that exhibit Constant Relative Risk Aversion (CRRA).
Wife’s utility function: UW=U(c1) + U(c2) + δD. Husband’s utility function: UH=
U(c1) + U(c2) + (1 −δ)D. CRRA utility function implies: U(c1) = c(1−θ)
1/(1 −θ) and
U(c2) = c(1−θ)
2/(1 −θ). Household utility function is derived from the Nash bargaining
model:
UHH = (UW)γ(UH)(1−γ)(1)
where γis the relative bargaining power of the wife and (1−γ) is the relative bargaining
power of the husband in intra-household resource allocations. Given that the preference for
the indivisible good Dis higher for the wife, we have δ > (1 −δ). The household faces the
following constraints: (i) S≥0; savings in the first period has to be positive to acquire the
indivisible good D, (ii) Y≥C1+S; income in the first period should be equal to consumption
in period 1 and savings in period 1, (iii) Y+S≥C2+D; income and savings from the
first period must be more or equal to consumption in period 2 and the expenditure on the
purchase of higher hours of electricity D.
Identifying consumption as a function of income and savings, the household maximization
problem with choice of savings Sand purchase the reliable electricity Dis given by the
household utility function (1) with the constraints:
Log(UH H ) = (1−γ)[U(Y−S)+ U(Y+S−D)+ (1 −δ)D]+ γ[U(Y−S)+U(Y+S−D) +δD]
subject to
S≥0
Y≥C1+S
10
Y+S≥C2+D
(2)
First order conditions of the maximization exercise ∂UH H /∂S in equation 3.1 gives the
neccessary condition for the household to purchase the available electricity (C1)−θ= (C2)−θ.
The savings neccessary to buy electricity is given by S=D/2. The marginal effect of
purchasing electricity on the joint utility shows that the optimal savings rate is given by
S=Y−(2δγ + 1 −δ−γ)−1/θ (3)
Here θis the parameter of relative risk aversion. Marginal effect of bargaining power on
the savings function ∂2Log(UHH )/∂S∂γ in equation 3 shows that with an increase in the
bargaining power of women in the household, the savings for electrification in period 1 also
increases.
∂S/∂γ = 1/θ[2δγ + 1 −δ−γ](θ−1)/θ (2δ−1) (4)
Equation 4 is a positive for any value of γ(0,1) if δ > 0.5. Partial derivative of wife’s
bargaining power with changes in the savings
∂γ/∂S = [θ(Y−S)−θ−1]/2δ−1 (5)
Equation 5 shows that an increase in savings for electrification increases women’s bargaining
power in intra-household resource allocations.
3.2 Empirical Model
We are interested in estimating the causal effect quality of electricity on women empower-
ment. We consider that the outcomes are conditional daily hours of electricity and power
11
outages, the baseline estimate is as follows:
Yi=αi+δQEi+βXi+γdi+θci+i(6)
Where, Yirepresents the outcome of interest, empowerment for women iof caste ciin district
di.Xij is a vector of individual and household observable socioeconomic and demographic
characteristics: real income, household adult education, women education, age, number of
children, household size, age of the household head, sex of the household head, rural/urban
and caste. QEiis the hours of electricity outage in the house of the i-th women of c-th caste
in the d-th district. We control for geographic and cultural characteristics at the district
and caste levels with district and caste dummy variables. The error term iis assumed to
be randomly distributed. α,β, and δare the unknown parameters to be estimated. The
interest in the analysis is to estimate the effect of hours of electrification, measured by the
coefficients δ.
If villages/PSUs were randomly selected, and electricity distribution happened randomly
then the baseline estimation in equation 6 would have provided unbiased estimates of the
impact of electrification. However, in practice individuals and households are not randomly
assigned to hours of access to electricity. There is an element of self-selection and sorting
involved. The decision to acquire electrification is based on both observed (income, household
size, age, sex, caste and locality) and unobserved characteristics at the village/PSU level.
There are unobserved characteristics such as household preference for electricity, the relative
bargaining power of the women in the household, productive potential of the household
and the ability to perceive benefits from electrification. These issues lead to endogeneity,
self-selection and sorting in acquiring reliable electrification. Households that are likely
to be excluded and left to be subsidized with irregular electricity, such as poor and the
less educated, are more inclined to acquire higher hours of electricity indicating a positive
selection bias, especially in households where women have strong bargaining power Donoso
12
et al. [2011].
Endogeneity can manifest in various ways. For example, it could be due to time varying
omitted variable bias motivated by unobserved factors at an individual and household level,
or that individual’s perception about potential benefits of electrification leads to a positive
self-selection bias. There could be reverse causality with positive or negative self selection.
Household could stop bargaining for electrification once they reach a certain threshold of
income and be able to afford alternate sources of electrification causing a negative selection
bias. Hence, a proposition could be made that households who are at the extreme ends of
electricity deficiency may have low marginal benefits from an additional hour of electrifica-
tion. Endogeneity may also arise from simultaneity if outcomes such as household income and
electrification are jointly determined. Thus equation 6 would yield biased impact estimates.
To this effect, we expect the baseline estimate to underestimate the coefficients.
To address this problem of endogeneity, we instrument household hours of electricity with
average hours of electricity at the village/PSU level. The same instrument has been used
previously on the effect of electrification on household and non-farm income [Khandker et al.,
2014, Chakravorty et al., 2014]. These studies have used the average access of electricity at
the village level. In our study we substitute average access by average hours of electricity at
the village/PSU level. A technical view of access vs. hours of electricity is potential power
(capacity) vs. energy (power*hours). Some systems are capacity constrained and some are
energy constrained. In this case, the system is mostly energy constraint. For example, in
the case of water with only two hours of access one can store some water but with electricity
this is not possible.
We use average hours of electricity at the village/PSU as this would affect a household’s
choice of electricity hours through peer-effects and demonstration effects. Higher hours of
electricity at the village level could induce a household which is below the average level
to acquire more hours of electricity. In India, where caste and religion form the basis of
social structure, this social structure influences how a particular village/PSU is populated
13
[Newman and Thorat, 2010]. The structure of social organization influences a household’s
choice of electricity hours given the local hours of electricity [Khandker et al., 2014, Rao,
2013]. The first stage estimate of instrumental variables (IVs) regression is obtained by
estimating the following equation
QEi=αi+λIij +βXi+γdi+θci+i(7)
Where Iij is the vector of instruments of mean hours of electricity in village/PSU j3.
λis the vector of coefficient of the effect of average hours of electricity at the village/PSU
level on household’s hour of electricity. If neighbors acquire more hours of electricity and
realize economic and social gain of better quality of electrification, then the status of fewer
electricity hours may signals lower socioeconomic standing and a case for social depravity,
which we expect households would want to avoid.
We expect that the higher the hours of electricity in a village/PSUs, the higher will be the
likelihood of a household in that village/PSUs to acquire more hours of electricity, provided
the household can afford the cost. The exogeneity condition for the instrument also holds
because mean village/PSUs level electricity does not directly affect a woman’s economic
and social agency in household decision making processes. Women’s financial autonomy,
mobility, agency and socio-economic decision making ability depends on their education,
economic and social condition at home, and the relative intra-household bargaining power
[Kabeer, 1999]. We expect a negative self selection bias as empowered women already have
higher household income and education would have an alternate source of electricity, the
outcome of which is that they do not need to acquire state/agency provided electricity.
We follow a two stage least squares linear probability instrumental variable (2SLS-IV)
equation Semykina and Wooldridge [2010], with cross-section data as it permits for unspec-
3There are 1503 villages in the survey with 42152 household level observations, hence there is sufficient
variation of approx 28 observations at the village/PSU level, see Khandker et al. [2014] for geographic
instrumental variables using IHDS, (2012).
14
ified correlation between Iiand αi, but requires Iito be uncorrelated with the error ui.
yi=αi+δQEi+βX1i+γdi+θci+i(8)
Where, yiis the dependent variable. For our study, these are the 5 empowerment indices
created by classifying 19 empowerment variables of women iin village/PSU jbased on the
IHDS, 2012 survey using a principal component analysis (see Table 2). QEiis the hours
of power outage and allowed to be correlated with the i, all other notations are similar to
equation 7. The model does not reject the hypothesis of no selection bias and allows for
arbitrary correlation between the unobserved effect and the explanatory variables. We use
caste and district fixed effects in the instrumental variables to derive a more robust casual
effect by controlling for the observed time invariant characteristics as these have significant
impact on women’s empowerment in India [Chandrasekhar and Ghosh, 2018]. The model
allows for any type of correlation between unobserved effects, explanatory and instrumental
variables, and does not require any specification of the reduced form equations for endogenous
variables. It makes no assumptions about errors distribution.
4 Data
The data used for this analysis is mainly from the third wave of the Indian Human Devel-
opment Survey (2012) [Desai and Vanneman, 2018]. IHDS are nation-wide random sample
surveys covering all 28 states of India except Meghalaya and Union Territories of Andaman
and Nicobar Islands and Lakshadweep, depending on population density 4. IHDS have rep-
resentative individual, household and gender disaggregated survey and wide-ranging topics
at household and individual level on demographic and socio-economic characteristics. The
survey also covers key features of the households at aggregated level including caste, religion
4The surveys are jointly carried out by researchers from the University of Maryland and the National
Council of Applied Economic Research (NCAER) in New Delhi [Desai and Vanneman, 2018].
15
and village level infrastructure. These features help control for community and district level
characteristics in the analyses as they can directly affect the outcomes of interest. The sur-
vey covers 13,706 households from urban areas and 28,446 households from rural areas. We
use only the third wave of the survey by combining the Individual, Household and Eligible
women’s questionnaire (IHDS, 2012), since the IHDS 2005 does not include all the variables
to create the indices of empowerment. We select 19 variables relating to gender relations, all
of which are only available for the third wave of the survey with 1,503 villages and 42,152
households.
The survey covers key socio-economic aspects of gender relations, agency, mobility and
decision making processes as shown in table 1.
Insert Table 1 here
It shows the descriptive statistics for the sub groups classified by quintiles of power outage
with difference in means tests across the quinitiles to compare these samples. There are signif-
icant differences among women in economic decision making, mobility, agency and household
decision making ability between the quintiles of deficiency. A pronounced difference can be
seen between women in the first and the second quinitile of deficiency, highlighting that
the having less that 5 hours of outage of electricity in a day does not substantially affect
women’s empowerment, but the second quinitile 5-14 hours and third quintile 15-24 hours of
deficiency have a strong bearing on the empowerment variables. We explore the correlation
between the empowerment variables and use these variations to create empowerment indices
as shown in table 2 using Principal Component Analysis (PCA).
We choose the component which has an eigen value greater than one as is the common
practice with PCA [Fang, 2003]. Component loading for economic empowerment includes
three variables: cash in hand for household expenditure, women’s ownership of property, and
women’s employment. PCA analysis in Table 2 shows that cash in hand for expenditure has
significantly higher weight than property ownership and employment in terms of determining
the economic freedom for women.
16
Insert Table 2 here
The analysis assigns ranks and factor loads (weight) to each variable within each empow-
erment category. We have five standardized empowerment indices with a mean of zero and
a standard deviation of one namely: economic freedom, economic decision making ability,
agency, mobility and household decision making ability. For simplicity, we use empowerment
indices instead of individual variables with the district and caste fixed effects and treat the
OLS model in equation 6 as the baseline model following the fixed-effects estimation in nor-
mal linear models [McCaffrey et al., 2012]. Coefficients in all regression tables are deviations
from zero as the indices are standardized with mean zero and standard deviation of one.
5 Results
For expositional purposes, we start the analysis at the extensive margin with an instrumental
variable regression of the access to electricity on women’s empowerment indices. We use
mean access to electricity at the Village/PSU/District/State level as an instrument for the
analysis at the extensive margin following the save IV used by Khandker et al. [2014], Rao
[2013], Chakravorty et al. [2014]. Table 3 shows that reliable access to electricity has strong
positive effects on women’s economic freedom, economic decision making, agency, mobility
and household decision making.
Insert Table 3 here
Access to electricity increases women’s economic freedom by 0.34 standard deviations, women’s
economic decision making by 0.25 standard deviations, women’s agency by 0.58 standard de-
viations, women’s mobility by 0.38 standard deviations and household decision making ability
by 0.44 standard deviations. Samad and Zhang [2019] have carried out a similar analysis
and have shown the effects of access to electricity on women’s socio-economic agency at the
extensive margin only. We start the analysis at the intensive margin using an OLS fixed
effects regression with district and caste fixed effects as shown in equation 6. Specification
17
1 in Table 4 shows the index of women’s economic freedom created by the PCA of cash in
hand for expenditure, current employment, decision making ability about work and property
ownership.
Insert Table 4 here
An additional hour of power outage in a day reduces women’s economic freedom by -0.02 SD.
Once we add additional controls in specification 2, the effect of power outage on women’s
economic freedom is still significant at 1% but the magnitude of effect decreases to -0.01
SD. Comparing specification 3, 4 and 5 shows that power outage at the second quinitile
(5-14 hours of outage a day) has the strongest negative effect on women’s economic freedom.
Although any power outage has a negative effect on women’s economic freedom, it is the
second quintile of outage that has the strongest bearing on economic freedom. Household
income has a positive effect on women’s economic freedom. Further, women’s education has
a positive effect on economic freedom. Women’s age has a positive effect on their economic
freedom. Given that the poorest population have the lowest access to electricity [Samad
and Zhang, 2019], the number of children for these women have a positive effect on their
economic freedom. Women headed households have a positive effect on women’s economic
freedom. Household size has a negative effect and except for the Brahmin caste which is the
base of the analysis for caste, belonging to any other caste does not individually have any
bearing on women’s economic freedom in the sample.
Figure 2 shows the effect of an additional hour of power outage on women’s empower-
ment and fuel collection time of girls for the poor and non-poor. Figure 3 shows women’s
empowerment with interactions at income levels and education of women. The results of the
non-linear OLS fixed effect regression shows that poor women expectedly have lower levels
of empowerment with additional hours of power outage. Education is more important than
their economic status in terms of determining the effect of power outage on their economic
freedom, economic decision making ability, agency, mobility, household decision making abil-
ity and fuel collection time. Overall, the figures show that income and education are key in
18
determining empowerment with electrification. Empowerment indices are affected differently
by income and education interacted with electrification.
Table 5 compares the effect of margins of electricity deficiency on women’s economic
decision making, agency, mobility, household decision making and fuel collection time for
girls using the baseline OLS fixed effects model in equation 6.
Insert Table 5 here
An additional hour of outage has a strong negative impact on all the variables in the bivariate
and multivariate regression. As in economic freedom, households belonging to the second
quinitile of electricity deficiency bear the strongest negative effect on women’s empowerment:
-0.01 SD on economic decision making, -0.01 SD on women’s agency, -0.03 SD on mobility
and a significant but smaller effect on women’s household decision making ability. Quintile
1 of power outage has a stronger effect than quintiles 2 and 3 in terms of women’s agency.
An additional hour of outage increases the fuel collection time of young women under the
age of 15 by approximately 2 minutes. In comparison with other indices, women’s agency in
terms of use of contraceptives, membership in social organizations, decision in own sickness
and the illness of the child is most affected by power outages and is significant across all
quintiles of power outage.
Given the presence of endogeneity in the relationship between electricity and women
empowerment, as well as self-selection in electrification, we use instrumental variables re-
gression to examine the intensive margin of electricity deficiency. The instrumental variable
regression in Table 6 shows the causal effect of an additional hour of outage on women’s
empowerment indices and fuel collection time in the household of adult women and children
under the age of 15 in the household.
Insert Table 6 here
Results show that power outage affects all facets of women’s empowerment under consid-
eration. An additional hour of power outage in a day reduces women’s economic freedom,
19
agency, mobility and household decision making by 0.02 SD and the effect is significant.
Power outage reduces women’s economic decision making by 0.01 SD. Household income
has a negative effect on women’s economic and household decision making which confirms
to the de-feminization theory with higher incomes in India [Abraham, 2013]. An additional
hour of outage increases the woman’s fuel collection time by 1.06 minutes, the boy’s by 4.96
minutes and the girl’s by 2.56 minutes. Apart from income which shows a strong negative
selection bias in the IV regression, all other covariates exhibit similar signs as the baseline
model in equation 6, but the significances are different correcting for selection bias in the
baseline estimates.
Table 7 shows the effect of moving from one quinitile of electricity deficiency to the next
quintile on women’s economic empowerment for poor and non-poor households in rural and
urban areas.
Insert Table 7 here
One quinitile increase power outage on an average reduces poor women’s economic freedom
by 0.14 SD and non-poor women’s economic freedom by 0.20 SD. In rural areas a quintile
increase in outage reduces economic freedom of women by 0.15 SD and in urban areas the
effect is negative but the coefficient is insignificant. In terms of economic decision making
of poor women, an increase in the quintile of power outage reduces poor women’s economic
decision making by 0.10 SD, and for the non-poor women by 0.12 SD. In rural areas the
economic decision making is reduced by 0.16 SD and in urban areas by 0.10 SD. Overall, an
additional quinitle of power outage reduces economic freedom and economic decision making
more for the non-poor women than for poor women. Outage negatively affects women’s
economic freedom and decision making in rural areas as compared to urban area. Results
show that non-poor and more educated women in rural households are likely to suffer more
from power outage as compared to poor women in urban areas with lower education levels.
Table 8 shows the effect of an additional quintile of power outage on women’s agency,
mobility and household decision making ability.
20
Insert Table 8 here
Outage does not have a significant effect on poor women’s agency. It has a negative and
significant effect on non-poor women’s agency by 0.15 SD. Women’s agency is reduced by
0.17 SD in rural areas and 0.27 SD in urban ares. In terms of women’s mobility, an additional
quintile of power outage reduces poor women’s mobility by 0.25 SD and non-poor women’s
mobility by 0.15 SD. Mobility for rural women is reduced by 0.18 SD, for urban women the
effect is negative but insignificant. Household decision making for poor women is reduced
by 0.22 SD. For non-poor women the household decision making ability reduces by 0.14 SD.
The household decision making ability for rural women reduces by 0.15 SD and in urban
areas it decreases by 0.20 SD. Overall, similar to the effect on economic empowerment, an
increase in quintile of outage affects rural non-poor women’s agency, mobility and decision
making more than poor and non-poor women in urban areas.
6 Conclusion
This study analyzes how empowerment of women is affected by the quality of electricity
outages at the household level. We investigate the causal link between access and quality
of electricity and women’s empowerment using a large gender-disaggregated survey from the
India Human Development Survey, 2012. We use principal component analysis to combine
19 variables that elicit information on women’s economic freedom, intra-household decision-
making and resource allocation into five indices of empowerment: economic freedom, eco-
nomics decision-making ability, agency, mobility and household decision making ability.
The study finds that an additional hour of power outage and the quintiles of outages
reduce women’s empowerment and bargaining power, but the effect is not homogeneous
across all women population. An hour of electricity is more beneficial for women in the
second quintile of electricity deficiency where household on average has no electricity between
5-14 hours a day. There is not much difference in the effect of electrification for those who
21
the second and third quintiles of outage where on average the household faces outages for
15-24 hours a day.
A number of factors in conjunction with electricity deficiency affect women’s empower-
ment in India. Educated women tend to lose the most from outages in rural and urban areas
conforming the loss of labor supply by educated women with energy deficiency [Chakravorty
et al., 2014, Dinkelman, 2011]. Household income has a strong impact on women’s empow-
erment associated with electricity deficiency, as also suggested by [Kanagawa and Nakata,
2008], khandker2014benefits. The analysis shows that having access to electricity is not suf-
ficient to empower women and the quality of electricity affects women’s position on all five
dimensions of empowerment.
Women’s labor force participation, education, health and exposure to electronic media
are key intermediary factors through which electrification enhances women’s empowerment.
We find that the quality of electricity is associated with positive improvements in all five
enabling factors for empowerment. We then investigate how the intermediate factors may
affect women’s empowerment depending on education and household characteristics such as
income and location. We find suggestive evidence that women’s labor force participation and
education are the most important determinants of their intra-household bargaining power.
Our results suggest that counting the electrified households is not sufficient and quality
is an important policy lever for empowering women. However, electrification alone is un-
likely to ensure significant progress in important dimensions of women’s empowerment, in
particular, for decision-making ability and economic freedom. Sustained efforts in improving
women’s earning opportunities, education, household income and urbanization are impor-
tant for improving their agency and empowerment. These enabling factors can be improved
in other ways besides electrification. One possible approach is the use of micro-finance in
provisioning of energy access to women. Policy measures targeting pervasive social norms
and gender stereotypes are also needed to reduce gender inequality.
22
References
V. Abraham. Missing labour or consistent” de-feminisation”? Economic and Political
Weekly, 2013.
A. Agrawal, A. Kumar, and T. J. Rao. 100% rural electrification in India: Myth or reality?
2020.
M. Bansal, R. Saini, and D. Khatod. Development of cooking sector in rural areas in india—a
review. Renewable and Sustainable Energy Reviews, 17:44–53, 2013.
R. Burgess, M. Greenstone, N. Ryan, and A. Sudarshan. The consequences of treating
electricity as a right. Journal of Economic Perspectives, 34(1):145–69, 2020.
U. Chakravorty, M. Pelli, and B. U. Marchand. Does the quality of electricity matter?
evidence from rural india. Journal of Economic Behavior & Organization, 107:228–247,
2014.
C. Chandrasekhar and J. Ghosh. The financialization of finance? demonetization and the
dubious push to cashlessness in India. Development and change, 49(2):420–436, 2018.
L. Cherchi, M. Goldstein, J. Habyarimana, J. Montalvao, M. O’Sullivan, C. Udry, A. Gruver,
et al. Empowering women through equal land rights. Technical report, The World Bank,
2019.
L. Cohen and A. Glazer. Bargaining within the family can generate a political gender gap.
Review of Economics of the Household, 15(4):1399–1413, 2017.
S. Desai and R. Vanneman. National Council of Applied Economic Research, New Delhi.
India Human Development Survey (IHDS), 2012. Ann Arbor, MI: Inter-university Con-
sortium for Political and Social Research, University of Michigan, 2018.
T. Dinkelman. The effects of rural electrification on employment: New evidence from South
Africa. American Economic Review, 101(7):3078–3108, 2011.
S. B. Donoso, Y. Altunba¸s, and A. Kara. The rationale behind informal finance: Evidence
from roscas in bolivia. The Journal of Developing Areas, pages 191–208, 2011.
E. Duflo. Women empowerment and economic development. Journal of Economic Literature,
50(4):1051–79, 2012.
W. Fang. Comparison and application of principal component analysis & factor analysis [j].
Statistical Education, 5(5):14–17, 2003.
S. Fukuda-Parr. From the millennium development goals to the sustainable development
goals: shifts in purpose, concept, and politics of global goal setting for development.
Gender & Development, 24(1):43–52, 2016.
C. F. Gould and J. Urpelainen. Lpg as a clean cooking fuel: Adoption, use, and impact in
rural India. Energy Policy, 122:395–408, 2018.
S. Hendriks. The role of financial inclusion in driving women’s economic empowerment.
Development in Practice, 29(8):1029–1038, Nov. 2019.
V. Jain. Powering India at household level: State effort, issues and concerns. 2018.
R. Jensen and E. Oster. The power of TV: Cable television and women’s status in India.
The Quarterly Journal of Economics, 124(3):1057–1094, 2009.
N. Kabeer. Resources, agency, achievements. Development and Change, 30:435–464, 1999.
ISSN 1467-7660.
M. Kanagawa and T. Nakata. Assessment of access to electricity and the socio-economic
impacts in rural areas of developing countries. Energy Policy, 36(6):2016–2029, June 2008.
23
URL https://ideas.repec.org/a/eee/enepol/v36y2008i6p2016-2029.html.
R. Kennedy, A. Mahajan, and J. Urpelainen. Quality of service predicts willingness to pay
for household electricity connections in rural India. Energy Policy, 129:319–326, 2019.
S. R. Khandker, H. A. Samad, R. Ali, and D. F. Barnes. Who benefits most from rural
electrification? evidence in India. The Energy Journal, 35(2), 2014.
G. K¨ohlin, E. O. Sills, S. K. Pattanayak, and C. Wilfong. Energy, gender and development:
what are the linkages? Where is the evidence? 2011.
L. Mayoux and S. Anand. Gender inequality, roscas and sectoral employment strategies:
questions from the South Indian silk industry. Money-go-rounds: The Importance of
Rotating Savings and Credit Associations for Women (Washington, DC: BERG, 1995),
pages 179–196, 1995.
D. F. McCaffrey, J. R. Lockwood, K. Mihaly, and T. R. Sass. A review of stata commands
for fixed-effects estimation in normal linear models. The Stata Journal, 12(3):406–432,
2012.
M. Mehra and G. Bhattacharya. Energy transitions in india implications for energy access,
greener energy, and energy security. 2019.
B. Min, Z. O’Keeffe, and F. Zhang. Whose power gets cut? Using high-frequency satellite
images to measure power supply irregularity. The World Bank, 2017.
K. Newman and S. K. Thorat. Blocked by caste: Economic discrimination in modern India.
Oxford University Press, 2010.
M. Nouni, S. Mullick, and T. Kandpal. Providing electricity access to remote areas in
India: An approach towards identifying potential areas for decentralized electricity supply.
Renewable and Sustainable Energy Reviews, 12(5):1187–1220, 2008.
K. O’Dell, S. Peters, and K. Wharton. Women, energy, and economic empowerment: apply-
ing a gender lens to amplify the impact of energy access. 2014.
D. Palit, G. K. Sarangi, and P. Krithika. Energising rural india using distributed generation:
The case of solar mini-grids in Chhattisgarh state, India. 2014.
S. Pargal and S. Ghosh Banerjee. More power to India: The challenge of electricity distri-
bution. The World Bank, 2014.
J. Parikh. Hardships and health impacts on women due to traditional cooking fuels: A case
study of Himachal Pradesh, India. Energy Policy, 39(12):7587–7594, 2011.
N. D. Rao. Does (better) electricity supply increase household enterprise income in India?
Energy Policy, 57:532–541, 2013.
H. A. Samad and F. Zhang. Electrification and Women’s Empowerment: Evidence from
Rural India. The World Bank, 2019.
A. Semykina and J. M. Wooldridge. Estimating panel data models in the presence of endo-
geneity and selection. Journal of Econometrics, 157(2):375–380, 2010.
A. Sharma, J. Parikh, and C. Singh. Transition to LPG for cooking: A case study from two
states of India. Energy for Sustainable Development, 51:63–72, 2019.
K. Standal and T. Winther. Empowerment through energy? impact of electricity on care
work practices and gender relations. In Forum for Development Studies, volume 43. Taylor
& Francis, 2016.
R. Stephenson, M. A. Koenig, and S. Ahmed. Domestic violence and contraceptive adoption
in uttar pradesh, India. Studies in Family Planning, 37(2):75–86, 2006.
World Bank. Global Financial Development Report 2014: Financial Inclusion. World Bank
24
Publications, Herndon, United States, 2013. ISBN 978-0-8213-9990-3.
25
Table 1: Descriptive Statistics and Mean Difference at Three Quinitles of Power Outage,
India, IHDS, 2012
Variables Obs Mean sd Obs Mean sd Obs Mean sd Mean Diff. Mean Diff.
Q1 (Power Outage (0-4) hours) Q2(Power Outage (5-14) hours) Q3(Power Outage (15-24) hours) p (Q1-Q2) p (Q2-Q3)
Real Income (base 2005) 14830 92957.84 135116.50 11660 65648.32 105620.60 10100 66827.43 134304.60 ***
Women Education 12397 1.30 0.63 9841 1.16 0.48 8594 1.14 0.44 **
Women Age 12398 37.32 9.22 9842 36.46 9.57 8594 36.33 9.57
Number of Children 12395 2.36 1.32 9837 2.59 1.48 8593 2.72 1.62 * *
Household Head Education 14829 9.74 4.66 11660 8.35 4.94 10093 8.21 5.09 ***
Household Head Sex 14828 1.15 0.36 11656 1.14 0.34 10100 1.13 0.34
Household Head Age 14828 50.62 13.20 11656 49.66 13.49 10100 49.56 13.60
Household Size 14830 4.74 2.15 11660 4.89 2.36 10100 5.07 2.46 * *
Urban 14830 0.55 0.50 11660 0.32 0.47 10100 0.22 0.41 *** ***
Electricity Hours 14830 22.17 1.57 11660 14.23 3.01 10100 6.19 1.88 *** ***
Hours of Power Outage 14830 1.83 1.57 11660 9.77 3.01 10100 17.81 1.88 *** ***
Women Knowledge of Health 14683 2.60 0.63 11530 2.51 0.70 9964 2.34 0.76 * **
Fuel Collection Minutes
Adult Women 3031 153.54 109.23 3559 157.72 104.29 3148 155.14 107.52
Boys Under 15 176 129.48 82.89 323 133.21 96.96 406 111.45 82.96 * ***
Girls under 15 236 134.13 81.04 408 130.28 87.15 527 123.41 89.20 * **
Girl Harassment (0/1) 12241 0.25 0.43 9632 0.26 0.44 8356 0.34 0.47 ***
Empowerment for Women (0/1)
Cash in hand for expenditure 12377 0.93 0.26 9827 0.93 0.25 8577 0.90 0.30
Ownership of Prop erty 11996 0.19 0.39 9499 0.17 0.38 8168 0.16 0.37 *
Current Employment 12381 0.40 0.49 9828 0.45 0.50 8579 0.40 0.49
Permission to visit health center 12303 0.80 0.40 9729 0.69 0.46 8440 0.65 0.48 *** *
Permission to visit friends/relatives 12257 0.84 0.36 9695 0.76 0.43 8443 0.70 0.46 *** ***
Permission to visit grocery stores 11964 0.86 0.34 9287 0.80 0.40 8008 0.73 0.45 *** ***
Permission for short distance travel 12274 0.64 0.48 9706 0.49 0.50 8462 0.47 0.50 ***
Use of Contraception 11167 0.80 0.40 8791 0.79 0.41 7648 0.69 0.46 ***
Ownership of Joint Bank Account 9653 0.62 0.49 7119 0.54 0.50 5553 0.53 0.50 ***
Membership in Social Organizations 12383 0.08 0.27 9832 0.05 0.21 8581 0.05 0.22 ***
Decision on number of children 12044 0.94 0.24 9451 0.94 0.24 8247 0.90 0.30 *
Decision making ability about work 10579 0.50 0.50 8462 0.46 0.50 6706 0.40 0.49 * **
Decision-Purchase of Expensive items 12314 0.81 0.39 9792 0.80 0.40 8543 0.76 0.43 *
Decision-Purchase of Property/Land 12194 0.79 0.41 9652 0.77 0.42 8485 0.73 0.45 *
Decision-Wedding Expenditures 12352 0.84 0.36 9783 0.83 0.38 8522 0.78 0.41 **
Decision-Children Marriage 11779 0.92 0.27 9391 0.90 0.30 8147 0.84 0.36 **
Decision-Food items 12374 0.69 0.46 9827 0.60 0.49 8582 0.55 0.50 *** ***
Decision-Child Illness 11827 0.94 0.24 9422 0.92 0.27 8234 0.88 0.33 *
Decision-Own Sickness 12369 0.89 0.31 9821 0.88 0.32 8571 0.82 0.38 ***
∗∗∗p < 0.01, ∗ ∗ p < 0.05, ∗p < 0.1
26
Table 2: Descriptive Statistics: Principal Component Analysis (PCA) of empowerment vari-
ables, IHDS, 2012
Indices Rank Empowerment Variables (0/1) Obs. Mean sd Weight (PCA) Interval
Economic Freedom 4 Cash in hand for household expenditure 39460 0.91 0.29 0.31
3 Name on home ownership papers 38023 0.17 0.37 0.39
2 Currently Employed 39461 0.42 0.49 0.58
1 Decision about Work 39461 0.42 0.49 0.63 (-2.81 1.77)
Mobility 1 Can visit health center alone 39079 0.71 0.45 0.52
2 Can visit friends/relatives alone 38966 0.77 0.42 0.52
3 Can go to grocery shop alone 37358 0.8 0.4 0.48
4 Can go short distance travel alone 38980 0.53 0.5 0.47 (-2.20 0.85)
Agency 3 Currently use contraceptives 35101 0.74 0.44 0.11
4 Membership Social Organization 27769 0.55 0.5 0.09
2 Decision Child Illness 39481 0.06 0.23 0.69
1 Decision Own Sickness 38042 0.92 0.27 0.70 (-3.96 0.91)
Economic Decisions 3 Decide purchasing expensive item 39243 0.77 0.42 0.57
1 Decides whether to buy land/property 38867 0.75 0.44 0.59
2 Decide wedding expense 39294 0.8 0.4 0.57 (-2.13 0.56)
Household Decisions 1 Decision Child Marriage 37261 0.88 0.32 0.67
3 Decision Food Shopping 39465 0.58 0.49 0.30
2 Decision Number of Children 37474 0.91 0.29 0.65
4 Ownership of Joint Bank Account 39430 0.85 0.35 0.18 (-3.30 0.53)
Factor loads are the score of individual variables in the empowerment indices and all indices are
standardized with mean zero and standard deviation one
27
Table 3: Extensive Margin: Instrumental Variable Regression: The Effect of Access to
Electricity on Women’s Empowerment in India, IHDS, 2012
(1) (2) (3) (4) (5)
Variables Economic
Freedom
Economic
Decision Agency Mobility Household
Decision
Access to Electricity 0.34*** 0.25*** 0.58*** 0.38*** 0.44***
(0.04) (0.03) (0.04) (0.03) (0.04)
Log Real Income 0.06*** -0.01** 0.01 0.02*** 0.01
(0.01) (0.01) (0.01) (0.01) (0.01)
Women Education 0.10*** 0.04*** 0.01 0.14*** 0.06***
(0.02) (0.01) (0.01) (0.01) (0.01)
Women Age 0.02*** 0.02*** 0.01*** 0.02*** 0.01***
(0.00) (0.00) (0.00) (0.00) (0.00)
Number of Children 0.00 0.04*** 0.03*** 0.03*** 0.01**
(0.01) (0.00) (0.01) (0.01) (0.01)
Household Head Sex 0.51*** 0.13*** 0.13*** 0.25*** 0.14***
(0.02) (0.01) (0.02) (0.02) (0.02)
Household Head Age -0.00*** -0.01*** -0.01*** -0.01*** -0.01***
(0.00) (0.00) (0.00) (0.00) (0.00)
Household Size -0.03*** -0.06*** -0.04*** -0.02*** -0.04***
(0.01) (0.00) (0.00) (0.00) (0.00)
Forward Caste -0.00 -0.07*** -0.05** -0.02 -0.03*
(0.03) (0.02) (0.02) (0.02) (0.02)
Other Backward Caste -0.04 -0.07*** -0.07*** -0.14*** -0.07***
(0.02) (0.02) (0.02) (0.02) (0.02)
Scheduled Caste -0.09*** -0.12*** -0.10*** -0.00 -0.08***
(0.03) (0.02) (0.02) (0.02) (0.02)
Constant -1.84*** -0.13** -0.57*** -1.22*** -0.58***
(0.11) (0.06) (0.09) (0.07) (0.07)
District FE 370 370 370 370 370
Observations 13,622 33,381 29,324 31,992 22,024
R-squared 0.085 0.084 0.016 0.044 0.033
Robust standard errors in parentheses
∗∗∗p < 0.01, ∗ ∗ p < 0.05, ∗p < 0.1
Number of observations are reduced in Table 3 due to the variable women’s current employment which has
13,622 observations.
28
Table 4: Intensive Margin: Baseline Result of OLS Fixed Effects Regression, Dependent
Variable: Women’s Economic Freedom
(1) (2) (3) (4) (5)
Variables Economic Freedom Economic Freedom Quintile-1 Qunitile-2 Quintile-3
Power Outage -0.02*** -0.01*** -0.00 -0.02*** -0.02
(0.00) (0.00) (0.01) (0.01) (0.01)
Log Real Income 0.05*** 0.06*** 0.05*** 0.02
(0.01) (0.02) (0.02) (0.02)
Women Education 0.09*** 0.05* 0.12*** 0.21***
(0.02) (0.03) (0.04) (0.06)
Women Age 0.02*** 0.02*** 0.02*** 0.02***
(0.00) (0.00) (0.00) (0.00)
Number of Children 0.01 -0.01 0.01 0.05***
(0.01) (0.01) (0.01) (0.01)
Household Head Sex 0.49*** 0.45*** 0.46*** 0.54***
(0.02) (0.04) (0.04) (0.05)
Household Head Age -0.01*** -0.01*** -0.01*** -0.01***
(0.00) (0.00) (0.00) (0.00)
Household Size -0.02*** -0.02** -0.01 -0.03***
(0.01) (0.01) (0.01) (0.01)
Urban 0.11** 0.05 -0.02
(0.05) (0.05) (0.09)
Forward Caste 0.01 0.01 0.00 -0.00
(0.03) (0.05) (0.05) (0.06)
Other Backward Caste -0.03 -0.01 -0.01 -0.07
(0.03) (0.05) (0.05) (0.06)
Scheduled Caste/Tribe -0.02 -0.07 0.04 -0.01
(0.04) (0.06) (0.06) (0.08)
Constant 0.18*** -1.32*** -1.46*** -1.25*** -1.21***
(0.02) (0.13) (0.21) (0.22) (0.32)
District FE 370 370 370 370 370
Observations 11,949 11,621 4,464 4,099 3,058
R-squared 0.198 0.262 0.272 0.355 0.304
Robust standard errors in parentheses
∗∗∗p < 0.01, ∗ ∗ p < 0.05, ∗p < 0.1
Note: For Table 4, the number of observations are listed from the baseline bivariate specification
29
Table 5: Intensive Margin: OLS Fixed Effect Regression: Effect of an additional hour of
power outage on Women’s Economic Decisions, Agency, Mobility, Household Decisions and
Fuel Collection time of Girls
(1) (2) (3) (4) (5)
Variables Economic Decision Agency Mobility Household Decision Fuel Collection Girls
OLS (Bivariate) -0.01*** -0.02*** -0.01*** -0.01*** 2.1***
(0.00) (0.00) (0.00) (0.00) (0.60)
OLS (Multivariate) -0.01*** -0.02* -0.01*** -0.01*** 2.45***
(0.00) (0.00) (0.00) (0.00) (0.69)
Quintile 1 -0.01 -0.03*** 0.00 -0.01***
(0.00) (0.01) (0.00) (0.07)
Quintile 2 -0.01* -0.01** -0.03*** -0.00**
(0.00) (0.00) (0.00) (0.00)
Quintile 3 -0.00 - 0.01** -0.00 -0.00
(0.01) (0.00) (0.00) (0.00)
Controls Y Y Y Y Y
District FE 370 370 370 370 370
Caste FE 4 4 4 4 4
Observation 30205 26571 29030 20637 1171
Robust standard errors in parentheses
∗∗∗p < 0.01, ∗ ∗ p < 0.05, ∗p < 0.1
30
Table 6: Intensive Margin: Instrumental Variable Regression: Causal effects of an additional
hour of power outage on Women Empowerment and Fuel Collection Time.
(1) (2) (3) (4) (5) (6) (7) (8)
Variables Economic Freedom Economic Decision Agency Mobility Household Decision Fuel Girls Fuel Boys Fuel Women
Power Outage -0.02*** -0.01*** -0.02*** -0.02*** -0.02*** 2.59*** 4.96*** 1.06**
(0.00) (0.00) (0.00) (0.00) (0.00) (0.99) (1.25) (0.48)
Log Real Income 0.04*** -0.02*** -0.00 0.01 -0.02** -1.52 3.72 -1.05
(0.01) (0.01) (0.01) (0.01) (0.01) (2.71) (2.87) (1.29)
Women Education 0.09*** 0.04*** 0.02 0.12*** 0.06*** 4.44 5.92 -7.60**
(0.02) (0.01) (0.01) (0.01) (0.01) (13.20) (12.60) (3.47)
Women Age 0.02*** 0.02*** 0.01*** 0.02*** 0.01*** 0.26 0.32 0.25*
(0.00) (0.00) (0.00) (0.00) (0.00) (0.40) (0.50) (0.15)
Number of Children 0.01 0.05*** 0.03*** 0.03*** 0.02*** 2.09 0.72 0.04
(0.01) (0.00) (0.01) (0.01) (0.01) (1.85) (2.25) (0.89)
Household Head Sex 0.49*** 0.12*** 0.09*** 0.21*** 0.12*** 7.45 11.04 -4.75*
(0.02) (0.01) (0.02) (0.02) (0.02) (7.40) (7.62) (2.82)
Household Head Age -0.01*** -0.01*** -0.01*** -0.01*** -0.00*** -0.08 0.05 0.01
(0.00) (0.00) (0.00) (0.00) (0.00) (0.25) (0.24) (0.10)
Household Size -0.02*** -0.05*** -0.02*** -0.02*** -0.02*** 1.55 0.61 1.34**
(0.01) (0.00) (0.00) (0.00) (0.00) (0.98) (1.10) (0.59)
Urban 0.05 0.01 -0.03 0.03 -0.01
(0.03) (0.02) (0.02) (0.02) (0.02)
Forward Caste 0.01 0.03* 0.04* -0.02 0.03 -3.23 19.78** 4.12
(0.03) (0.02) (0.02) (0.02) (0.02) (7.96) (8.94) (3.86)
Other Backward Caste -0.03 0.01 0.01 -0.06*** 0.00 -0.47 28.50*** -3.75
(0.03) (0.02) (0.02) (0.02) (0.02) (7.59) (9.39) (3.78)
Scheduled Caste/Tribe -0.02 0.03 0.04 -0.04** 0.02 -13.68 -0.01 -12.39***
(0.03) (0.02) (0.02) (0.02) (0.02) (8.93) (10.50) (4.18)
Constant -1.15*** 0.07 0.43*** 0.35*** 0.29*** -17.64 -46.00 182.72***
(0.33) (0.12) (0.12) (0.10) (0.11) (40.96) (55.28) (54.36)
District FE 370 370 370 370 370 370 370 370
Observations 11,621 29,343 25,819 28,203 20,054 1,043 788 8,193
R-squared 0.262 0.303 0.282 0.216 0.291 0.475 0.494 0.347
Robust standard errors in parentheses
∗∗∗p < 0.01, ∗ ∗ p < 0.05, ∗p < 0.1
31
Table 7: Intensive Margin: Instrumental Variable Regression: The Effect of Quintile of Power Outages on Women’s Economic
Freedom and Decision Making
(1) (2) (3) (4) (5) (6) (7) (8)
Variables Ec. Freedom Poor Eco Freedom Non-Poor Eco Freedom Rural Eco Freedom Urban Ec Decision Poor Eco Decision Non-Poor Eco Decision Rural Eco Decision Urban
Quinitle of Power Outage -0.14* -0.20*** -0.15*** -0.13 -0.10* -0.12*** -0.16*** -0.10**
(0.07) (0.03) (0.04) (0.10) (0.05) (0.02) (0.02) (0.04)
Log Real Income 0.01 0.03** 0.03** 0.01 -0.01 -0.03*** -0.03*** -0.04***
(0.03) (0.01) (0.01) (0.02) (0.02) (0.01) (0.01) (0.01)
Women Education 0.13 0.07*** 0.14*** 0.02 -0.07 0.02** -0.03 0.04***
(0.10) (0.02) (0.03) (0.03) (0.06) (0.01) (0.02) (0.01)
Women Age 0.02*** 0.02*** 0.01*** 0.02*** 0.01*** 0.01*** 0.01*** 0.01***
(0.00) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00)
Number of Children 0.02 0.02** 0.02*** 0.02 0.08*** 0.07*** 0.07*** 0.07***
(0.02) (0.01) (0.01) (0.02) (0.01) (0.01) (0.01) (0.01)
Household Size -0.03** -0.05*** -0.04*** -0.04*** -0.06*** -0.08*** -0.08*** -0.07***
(0.01) (0.01) (0.01) (0.01) (0.01) (0.00) (0.00) (0.00)
Forward Caste -0.01 -0.01 -0.01 -0.05 0.08* 0.03 0.06*** 0.00
(0.06) (0.04) (0.03) (0.07) (0.05) (0.02) (0.02) (0.03)
Other Backward Caste 0.01 -0.06* -0.06** -0.07 0.04 -0.01 0.01 -0.01
(0.06) (0.03) (0.03) (0.06) (0.04) (0.02) (0.02) (0.03)
Scheduled Caste/Tribe -0.05 -0.04 -0.07 -0.02 0.05 0.01 0.00 0.03
(0.09) (0.04) (0.04) (0.07) (0.06) (0.02) (0.03) (0.03)
Constant -1.08** -0.31 -0.55 -0.21 0.50* 0.31** 0.42*** 0.42*
(0.42) (0.34) (0.39) (0.64) (0.26) (0.13) (0.15) (0.22)
District FE 370 370 370 370 370 370 370 370
Observations 2,177 9,440 8,247 3,375 4,161 25,178 18,054 11,292
R-squared 0.365 0.226 0.255 0.250 0.348 0.294 0.284 0.312
Robust standard errors in parentheses
∗∗∗p < 0.01, ∗ ∗ p < 0.05, ∗p < 0.1
32
Table 8: Intensive Margin: Instrumental Variable Regression: Effect of Electricity Deficiency on Women’s agency, mobility and
household decision making ability for poor, non-poor, rural and urban areas, India, IHDS 2012
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12)
Variables Agency Poor Agency Non-Poor Agency Rural Agency Urban Mobility Poor Mobility Non-Poor Mobility Rural Mobility Urban HH Decision Poor HH Decision Non-Poor HH Decision Rural HH Decision Urban
Quinitle of Power Outage -0.09 -0.15*** -0.17*** -0.27*** -0.25*** -0.15*** -0.18*** -0.03 -0.22*** -0.14*** -0.15*** -0.20***
(0.06) (0.03) (0.03) (0.06) (0.06) (0.02) (0.03) (0.04) (0.05) (0.02) (0.03) (0.05)
Log real income 0.02 -0.01 -0.01 -0.00 -0.01 0.01 0.01 -0.02* -0.01 -0.02*** -0.03*** -0.02**
(0.03) (0.01) (0.01) (0.01) (0.02) (0.01) (0.01) (0.01) (0.02) (0.01) (0.01) (0.01)
Women Education -0.09 0.01 -0.01 0.02 0.11* 0.11*** 0.10*** 0.12*** 0.02 0.05*** 0.04** 0.06***
(0.08) (0.01) (0.02) (0.02) (0.07) (0.01) (0.02) (0.01) (0.05) (0.01) (0.02) (0.01)
Women Age 0.01*** 0.01*** 0.01*** 0.01*** 0.01*** 0.01*** 0.01*** 0.01*** 0.01*** 0.01*** 0.01*** 0.01***
(0.00) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00)
Number of Children 0.05*** 0.04*** 0.05*** 0.05*** 0.03*** 0.04*** 0.04*** 0.04*** 0.03*** 0.03*** 0.03*** 0.02***
(0.01) (0.01) (0.01) (0.01) (0.01) (0.01) (0.01) (0.01) (0.01) (0.01) (0.01) (0.01)
Household Size -0.02*** -0.04*** -0.04*** -0.03*** -0.03*** -0.03*** -0.03*** -0.03*** -0.04*** -0.04*** -0.04*** -0.03***
(0.01) (0.00) (0.00) (0.01) (0.01) (0.00) (0.00) (0.00) (0.01) (0.00) (0.00) (0.00)
Forward Caste 0.20*** 0.02 0.05* 0.05 0.12** -0.03 -0.01 -0.03 0.14** 0.01 0.06** -0.00
(0.06) (0.03) (0.03) (0.04) (0.05) (0.02) (0.03) (0.04) (0.06) (0.02) (0.02) (0.03)
Other Backward Caste 0.14** -0.01 0.03 -0.02 0.07 -0.08*** -0.06** -0.07** 0.10** -0.02 0.03 -0.04
(0.06) (0.02) (0.03) (0.04) (0.05) (0.02) (0.03) (0.03) (0.05) (0.02) (0.02) (0.03)
Scheduled Caste/Tribe 0.17** 0.01 0.06* -0.01 0.01 -0.06*** -0.07** -0.04 0.12* 0.01 0.04 -0.01
(0.07) (0.02) (0.03) (0.04) (0.06) (0.02) (0.03) (0.03) (0.07) (0.02) (0.02) (0.03)
Constant 0.13 0.73*** 0.75*** 1.05*** 0.52 0.73*** 0.81*** 0.52** 0.67** 0.55*** 0.52*** 0.95***
(0.31) (0.14) (0.16) (0.23) (0.37) (0.12) (0.14) (0.21) (0.28) (0.12) (0.14) (0.20)
Observations 3,747 22,068 15,920 9,901 4,048 24,150 17,272 10,933 2,340 17,714 11,928 8,129
District FE 370 370 370 370 370 370 370 370 370 370 370 370 R-squared
0.326 0.290 0.251 0.347 0.289 0.208 0.207 0.237 0.379 0.295 0.286 0.312
Robust standard errors in parentheses
∗∗∗p < 0.01, ∗ ∗ p < 0.05, ∗p < 0.1
33
Figures
Figure 1: Hours of electricity in India at the district level, 2012, Legend is average hours of
electricity at the district level in 2012. Source: India Human Development Survey 2012
34
Figure 2: Quintile of Power Outages and Women’s Empowerment for the Poor and Non-Poor,
result from OLS fixed effects regression. Quintile 1: 0-4 hours of power outage, Quintile 2:
5-14 hours of power outage, Quintile 3: 15-24 hours of power outage, IHDS, 2012
35
Figure 3: Quintile of Power Outages and Women’s Empowerment. Categories: Poor, Non
Poor, Urban, Rural, Education, Living Standards. Result from OLS fixed effects regression
with interactions. Quintile 1: 0-4 hours of power outage, Quintile 2: 5-14 hours of power
outage, Quintile 3: 15-24 hours of power outage. Education: 0-9th grade is grouped as
1, Matriculation (10th grade) to Higher secondary is grouped as 2, Bachelors and above
is grouped as 3. Poverty line is derived from Tendulkar’s (2012) cut-off line of poverty.
Living standard is classified into Poor, Middle Class and Comfortable following the survey
codebook, IHDS, 2012.
Note: The index is standardized with Mean=0 and Standard Deviation=1
36
37
