The Impact of Womens’ Financial

Empowerment on Intimate Partner

Violence

Submitted by

Ananya Garg, Manjari Goel, Rianna Basu

S4 Project Group E

ECO-3500-4

Development Economics

Prof. Hemanshu Kumar

Contents

I Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
II Literature Review . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
II.1 Household Bargaining Theory . . . . . . . . . . . . . . . . . . . . . . . . 2
II.2 Male Backlash Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
III Data and Empirical Strategy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
III.1 Outcome Variable . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
III.2 Independant Variable . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
III.3 Controls . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
III.4 Methodology and Identification Strategy . . . . . . . . . . . . . . . . . . 6
IV Descriptive Statistics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
V Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
V.1 Propensity Score Matching Results . . . . . . . . . . . . . . . . . . . . . 9
V.2 Relevance of PSM in our study . . . . . . . . . . . . . . . . . . . . . . . 10
V.3 Treatment Assignment . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
V.4 Balance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
V.5 Final Results: Without Controls . . . . . . . . . . . . . . . . . . . . . . . 12
V.6 Final Results: With Controls . . . . . . . . . . . . . . . . . . . . . . . . 12
VI Limitations and Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
VII References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
VIII Appendix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
Abstract

This paper seeks to establish a causal impact between Womens’ Financial Empowerment and
Intimate Partner Violence. Intimate partner violence (IPV) is a prevalent and dangerous prob-
lem that women worldwide face, which can not only impact their physical and mental health,
but also their childrens’ health. Being financially empowered increases the household bargaining
power of women, thus reducing the prevalence of violence by their husbands/partners. On the
other hand, financial empowerment triggers male backlash, which might increase the likelihood
of IPV. Using the National Family Health Survey - Round 5 (NFHS-5) data in India, our anal-
ysis reveals that women involved in making financial decisions, whether related to their own
income or their partners’, have a lower likelihood of ever encountering any form of Intimate
Partner Violence (IPV).
I Introduction

Violence against women is much prominent across the globe. More often than not, women are
subjected to the same by their own partners or spouses. This makes intimate partner violence
a significant threat to a woman’s overall well-being. While there are a myriad of factors that
lead to intimate partner violence or provide protection against it, the aim of our paper is to
analyse it through the factors associated with women’s financial empowerment.

Our motivation behind this topic draws from the people we surround ourselves with in our
own homes. It is common to hire female house help in India, and very often they are the bread-
winners of their households. However, we still see them being subject to domestic violence.
To that effect, we have tried to study the measures that determine financial empowerment of
women, and whether or not they decrease the prevalence of IPV. We have assessed if there are
other factors that contribute to the prevalence of IPV, through controls that are available in
the NFHS-5 dataset. Our question of interest is as follows: Does financial empowerment of
women lead to changes in the prevalence of IPV?

The main dataset that we have made use of for our study is the National Family and Health
Survey in India (NFHS-5) conducted between 2019 and 2021. Furthermore, we have used
propensity score matching to establish a causal link and derive results that are statistically sig-
nificant. Our findings suggest that financial empowerment among women leads to a significant
reduction in the likelihood of experiencing IPV.

We begin with an overview of the existing literature on this topic and its findings. Through
our literature review, we have explored the link between financial empowerment and IPV via
two main channels: the household bargaining model and the male backlash theory, in order
to see which theory our results fit best. This is followed by a discussion on the data and
methodology that we have employed along with the results of our study. Lastly, we have talked
about the limitations of our study in the Conclusion.

1
II Literature Review

According to the World Health Organisation, harmful behaviour against women resulting in
physical, psychological or sexual harm by an intimate partner is categorized as Intimate Partner
Violence. (World Health Organization: WHO). Data shows that almost one in three women
in a relationship report being victims of the same all over the world. Some of the risk fac-
tors associated with IPV include problems around paid employment opportunities for women,
community norms ascertaining different societal statuses for men and women, and control over
their partners by the male in a relationship, among others. In India, around 30 per cent of
ever-married women face physical and sexual violence, as per NFHS-5. Thus, it is increasingly
important to study the effect that women empowerment determined by economic factors can
have on intimate partner violence.

The relationship between the financial empowerment of women and IPV is well analysed
via two broad channels or theories:- the household bargaining model and the male backlash
theory.

II.1 Household Bargaining Theory

The household bargaining model suggests that an increase in a woman’s economic indepen-
dence in the form of higher education, increased wages, better employment opportunities, etc.
creates a balance in the power dynamic between the partners (Dildar 4-6). This further leads
to an increase in the woman’s bargaining power within the household or the relationship, thus,
reducing the risk of IPV (Aizer; Guarnieri and Rainer; ALGÜL and YARBAŞI; Heath). An
improved financial status of a woman gives her enough resources and backing to leave her re-
lationship owing to abuse. This ‘power’ is reduced in case the woman is unemployed, has zero
financial assets, or is financially dependent on her partner (ALGÜL and YARBAŞI 201).

For instance, a study analysing whether economic empowerment acts as a protective factor
against IPV in Turkey exploits the National Survey on Domestic Violence and runs a log re-
gression to capture female exposure to violence (Dildar 9). It uses independent variables that
record a woman’s employment status, wealth ownership and relative income. Results prove

2
that women with higher income relative to their partner face a reduced probability of IPV (15).
While there is some evidence supporting male backlash theory, after controlling for endogeneity
by using average women’s employment status in a province, the results again support the idea
that economic employment of women reduces the likelihood of IPV (10, 16).

In addition to that, Heath’s study in Bangladesh on the impact of women’s access to em-
ployment opportunities and their autonomy within the household on domestic violence yields
similar results as the aforementioned. Bargaining power here is proxied by education and age
at marriage (16). Women who have higher autonomy before they enter the labour market tend
to have more bargaining power and higher chances of fleeing their partner in case of abuse (24).
Among women with paid employment opportunities, an additional year of education leads to
a decrease in domestic violence (16).

Within India, the longitudinal study conducted by Raj et al. on married women in Maha-
rashtra found that financial empowerment in the form of women owning bank accounts tends to
reduce the risk of IPV (197). Additionally, joint control over the husband’s income signalling fi-
nancial autonomy within the household further reduces the potential for violence against women
(202). Similar findings by Bhattacharya and Bhattacharya, based on NFHS-III data, suggest
that spousal violence decreases for women with increased formal education, family wealth and
financial independence (244).

II.2 Male Backlash Theory

The male backlash theory, on the contrary, proposes that men are likely to use violence as a
means to assert their dominance over a woman with economic independence (Guarnieri and
Rainer 5). This is a method for the partner to “offset the increased bargaining power that her
income would otherwise bring” (Heath 4). Therefore, the risk of IPV is higher for financially
empowered women – even more so if they earn more than their partner.

In the previously mentioned study by Heath, there is evidence that women with less ed-
ucation and a younger age at marriage, who get paid for work are at a higher risk of facing

3
domestic violence (24). The bargaining power of women can thus, be countered by men by way
of domestic violence. This provides reassurance to the male and also cements their dominance
in the household against a financially empowered woman. On the differential results in urban
and rural areas in sub-Saharan Africa, Stöckl et al. find that household wealth has a greater
impact in rural areas (4). However, this impact is only visible for women between ages 25 to
34, and 35 to 49 (5). Women working outside the household increased their exposure to IPV in
both rural and urban areas; the result is also consistent with women earning more than their
partners. Furthermore, this exposure was exacerbated in the rural setting amongst women with
higher educational attainment.

Overall, the results are mixed. When discussing female autonomy within the household,
results from a study by Tsai state that joint management of household finances reduced the
“severity” of IPV, as opposed to women who tracked them independently in the Philippines
(12). It contributes not only to some facets of the male backlash theory but also posits the
idea that the economic empowerment of women by way of their controlling over finances is
multifaceted (15).

Within India as well, Sohini Paul observes that there is a positive relation between women’s
labour force participation and physical and emotional violence faced by women (224). Increased
economic empowerment of women that are employed leads to a higher decision-making power
which undermines the conventional power that men possess (225). Therefore, they often resort
to violence, yielding results in favour of the male backlash theory.

The common limitation in all these studies, however, is that of a lack of a significant causal
relationship between factors associated with economic empowerment and IPV. While correlation
is easy to establish, the causality is compromised due to endogeneity issues. By utilizing the
approach of propensity score matching, we have tried to generate causal implications of financial
empowerment on IPV. Owing to the lack of utilization of this method in studies conducted
within India, our paper is relevant and contributes to the existing literature.

4
III Data and Empirical Strategy

This paper uses the National Family Health Survey - Round 5 (NFHS 5) conducted through
2019-2021 as its primary dataset. We make use of the individual survey questionnaire, known
as the Individual’s Recode of the NFHS-5 which consists of responses by 7,24,115 women.

III.1 Outcome Variable

The outcome variable in this study is Intimate Partner Violence (IPV), which is measured
through 13 different variables in the survey. Some of these measures include whether hus-
band/partner has slapped the respondent, whether respondent has been subject to sexual vi-
olence by husband/partner and whether respondent has bruises because of husband/partner.
All 13 measures have been used to create one variable called IPV which takes the value 1 if a
respondent has faced even a single one of the 13 forms.

On the other hand, IPV takes the value 0 only if the respondent has faced none of the
forms of intimate partner violence recorded in the survey questionnaire. We find that 63,851
women answered this question, out of which 27.8% have been been subject to violence by their
husband/partner .

III.2 Independant Variable

The key independant variable in this paper if financial empowerment of women (finemp), which
is measured by the decision making power a respondent has with respect to her income as well
as her husband/partner’s income. The variable takes the value 0 if the respondent has no say
in how to spend her income as well as her husband’s income. However, if the respondent has a
say in decision making regarding even one of the two incomes, the variable takes the value 1.
As it would be expected, there are no observations where the respondent has a say in how to
spend her husband’s income but has no say in how to spend her own.

Interestingly, we observe that out of the 76,257 who have answered this question, 98.6% are
financially empowered. It is important to note that this is because even if a respondent and

5
her husband jointly make decisions regarding any of the two incomes, the woman is considered
to be financially empowered (Akilova et al. 68).

III.3 Controls

For the final regressions, the paper uses a host of controls from the NFHS 5 survey that could
potentially affect the prevalence of IPV as well as financial empowerment of women. These con-
trols include number of sons and daughters the respondent has lost, whether husband/partner
consumes alcohol, the education level of the respondent’s husband/partner, religion of the re-
spondent, whether the respondent is currently pregnant, the respondent consumes alcohol, age
of the respondent et cetera. These controls are used in order to account any inaccuracies that
our results could portray sue to the problem of Omitted Variable Bias.

III.4 Methodology and Identification Strategy

In order to establish a causal link between womens’ financial empowerment and prevalence of
IPV, we use Propensity Score Matching (PSM) to replicate a Randomized Control Trial (RCT)
framework, where the treatment and control groups differ solely because of the treatment itself
(Akilova et al. 68). With respect to our study, women who are financially empowered form
the treatment group, and those who are not form the control group. In this paper, Propensity
score matching is done through the following steps:

Probit Regression

As the first stage of PSM, we ran a probit regression of the key independant variable (finemp)
on various baseline observable covariates. The purpose of this step was to obtain propensity
scores, which are essentially the probability that an observation will be in the treatment group.
Through this exercise, we received a propensity score for all 76,257 women in the sample.

Propensity Score Matching

In the second stage, each respondent in the treatment group is matched with a respondent in
the control group with the same propensity score through a one to one matching procedure.
This ensures that now we have a sub sample of 6,129 women created which are divided into

6
treatment and control groups, with each woman in the treatment group having a match in the
control group. In simpler words, PSM helped eliminate those observations in the treatment
and control groups that differ greatly on the basis of the observable covariates.

As a result, we obtain treatment and control groups that are very similar with respect to the
observable characteristics, but only differ in the fact that one receives the treatment and one
does not. This helps establish a causal link between the key independant variable(finemp) and
the dependant variable(IPV). It is important to note that we are limited by the covariates used
in the study to obtain balanced treatment and controls groups. The possibility of the existence
of unobservable characteristics that might define treatment assignment cannot be denied.

However, for the purpose of this study, we use 7 reasonable covariates such as ethnicity of
the respondent, education level of the respondent, whether the respondent is literate or not,
whether the respondent is currently employed, husband’s desire for children et cetera. The
paper aims to find robust results within the scope of these observed covariates.

Regressions on the Matched Sample

Finally, we use this matched sample of 6,129 women to run regressions of IPV on financial
empowerment. The regressions are run with and without the controls that are mentioned in
the above subsections. The two regression equations we use are:
1. Without Controls

Yi = β0 + β1 X1i + µ (1)

IP Vi = β0 + β1 finempi + µ (2)

2. With Controls

Yi = β0 + β1 X1i + Ci + ε (3)

IP Vi = β0 + β1 finempi + Ci + ε (4)

7
IV Descriptive Statistics

Before comprehending the results of the study, it is important to understand the different
variables in the questionnaire in order to establish a foundation for any reasoning and inferences
that one wishes to make.

As it is evident from the Table 1, 27.8% of the women selected for the domestic violence
module have faced some kind of abuse from their husband/partner. What is even more inter-
esting is that 36.7% women in the whole sample justify intimate partner violence if the wife is
disrespectful towards the husband, and 26.4% justify it if the wife is unfaithful. These statistics
are important because they do affect the prevalence of IPV in a household. If a woman justifies
IPV, she herself and the other women in her household are more likely to experience it. As
for emotional violence, which this study does not include as a measure of Intimate Partner
Violence, is experienced by 12.5% of the respondents.

The average age at first marriage is 21 for the women in the sample. This is unusual due to
two factors, the high prevalence of child marriage in India as well as the minimum legal marriage
age for women. As for financial empowerment of women, strikingly, 98.6% of the women who
have answered this question are financially empowered. It is imperative to understand that this
result is because of the way the study is conducted, which assumes a woman to be financially

8
empowered even if both her and her husband jointly decide how to spend the income that
comes into the household. Finally, the mean level of education of the respondent and their
husbands/partners is very similar.

V Results

Amidst conflicting literature on the subject, the results from our regression analysis suggest
that financially empowering women reduces the risk of them facing intimate partner violence in
India. This argument is in line with the Household Bargaining Theory, which suggests that the
participation of women in financial decision-making (about their income and/or their husband’s
income) increases their bargaining power and autonomy in the household, thereby reducing the
risk of IPV.

V.1 Propensity Score Matching Results

The results from the probit regression of women’s financial empowerment on the observed
covariates are summarized in the top half of Table 2. The pscore is statistically significant at
the 1% level with a p-value of 0.00 and a positive coefficient, indicating that women with a
higher pscore were more likely to be assigned treatment status i.e., more likely to be financially
empowered.

The bottom half of Table 2 shows the overall difference between treatment and control

9
groups for the matched versus the unmatched sample. The T-stat in the first row is statistically
significant, implying that there are significant differences between the treatment group and the
control group in the unmatched sample. The effect in the case of the matched sample is
statistically insignificant, implying that there is balance between the treatment and control
group after PSM.

V.2 Relevance of PSM in our study

Table 3 shows us the results of a IPV on pscore, which is the variable for the propensity
scores for each respondent who has answered the question regarding decision making with re-
spect to income. The interpretation of the negative coefficient on pscore is that a higher pscore,
meaning, the higher the probability of receiving the treatment (being financially empowered),
the lower the likelihood that the respondent would experience IPV. The fact that this coefficient
is significant at the 1% level proves the relevance of PSM in our study, because it showcases how
being in the treatment group does have a significant negative impact on the outcome variable.

If the coefficient on pscore were not significant, it would render the entire research question,
as well as the methodology to be futile, since being in the treatment or the control group would
essentially be the same thing in that case, since the likelihood of receiving the treatment has
no significant effect on the outcome variable. Hence, the results in this table prove how being
in the treatment group does have an effect on the outcome variable

10
V.3 Treatment Assignment

As we can observe from Table 4, out of the 6,129 women in the matched sample, 5,106
are in the treatment group and 1,023 are in the control group.The difference in the number of
respondents in the treatment and control group can be attributed to the nature of the data
in the NFHS itself. As observed in the summary statistics, around 98% of the women in the
dataset are deemed to have financial autonomy. Thus, variation in the size of the untreated and
treated groups are a result of a high incidence of financially empowered women in the dataset.

V.4 Balance

Table 5 considers baseline characteristics such as ethnicity of the respondent, husband’s desire
to have children, husband’s alcohol consumption, education level of the respondent, literacy
level of the respondent, whether the respondent is working, and the relative income of the
respondent as compared to her partner.

The difference between treatment and control groups in terms of husband’s desire for chil-
dren and respondent’s employment status is statistically significant at the 10% and 5% level
respectively. However, these differences can be ignored, as the magnitude of the downward bias
is negligible (-2.9 and -4.3 respectively), and the mean value of each variable is almost the same
in the treatment and control groups.

Furthermore, when including a large set of covariates, small differences tend to become
statistically significant. However, they are of little practical importance to the results overall

11
since most of the other variables show balance, which can be observed in Table 5. Thus, we
can conclude that there is balance in the baseline observable characteristics in the treatment
and control groups.

V.5 Final Results: Without Controls

Table 6 in the Appendix reports the results from running a simple linear regression of IPV on
Women’s Financial Empowerment on the matched sample. As observed in Column 2 of the
table, women’s participation in financial decision making i.e., their ability to make decisions
regarding their own income and/or their husband’s income, reduces their probability of experi-
encing IPV by approx. 7.36%. This result is statistically significant at the 1% significance level.

V.6 Final Results: With Controls

As observed in Column 1 of Table 6 (in the Appendix), we have added a host of controls to
increase the precision and robustness of our causal analysis, and account for Omitted Variable
Bias. After adding these controls, we find that financially empowered women still have a re-
duced likelihood of facing IPV by roughly 10%, and that this result is statistically significant
at the 5% significance level. Controls like Husband’s Occupation, Respondent’s Religion, Re-
spondent’s Alcohol Consumption and Husband’s Education Level also have a significant effect
on IPV. However, they do not eliminate the significance of our key independent variable.

These results also indicate that not being financially empowered leads to an increased proba-
bility of experiencing IPV by 56.7% and 38% respectively, as observed by the positive constants

12
in both the columns, both being significant at the 1% level. Women that are financially em-
powered are on more equitable grounds with their partners and more likely to exert a lower
tolerance towards IPV. Thus, granting women financial autonomy such that they partake in
decision-making regarding their and/or husband’s income, reduces the risk of IPV.

VI Limitations and Conclusion

Our analysis is robust since adding controls does not eliminate the significance of our results.
Without controls, financial empowerment reduces the likelihood of ever experiencing IPV by
approx 7.36%, which is significant at the 1% level. After robustness checks, this changes to a
10% reduction in the likelihood, significant at the 10% level. Within the scope of our study,
we find that financial empowerment of women does decrease the prevalence of IPV, and it is
imperative for future research to explore more accurate ways of establishing causal links, and
then discussion policy recommendations that could be born out of such analyses.

Our treatment group came out to be larger than the control group, which could reduce the
statistical power of our analysis. However, it can be seen that there is balance between the
groups in terms of observable characteristics. The biases are minimal and most differences are
statistically insignificant. Additionally, we avoided 1 to 5 or 1 to 10 matching to prevent our
treatment and control groups from being imprecise and having too much variability.

Furthermore, when it comes to strategies like PSM or RCTs, we cannot deny the possibility
of the existence of unobservable factors that might determine treatment assignment. As re-
searchers, we will always be limited by our tool sets to be able to account for such unobservable
factors. What we have attempted to do here is to establish as smooth as a causal link as we
could, given the confines of the NFHS dataset. In doing so, we recognize the complexity of our
task, as we try to see the interactions between various factors of our research. Through this
careful approach, we strive to contribute valuable insights to academic literature that are not
only methodologically rigorous but also cognizant of the unobservable factors within the realm
of our study.

13
VII References

Aizer, Anna. “The Gender Wage Gap and Domestic Violence.” American Economic Review,
vol. 100, no. 4, American Economic Association, Sept. 2010, pp. 1847–59.
doi.org/10.1257/aer.100.4.1847.

Akilova, Mashura, and Yamile M. Marti. “What Is the Effect of Women’s Financial Empow-
erment on Intimate Partner Violence in Jordan?” Global Social Welfare, vol. 1, no. 2, Springer
Science and Business Media LLC, Mar. 2014, pp. 65–74. Crossref, https://doi.org/10.1007/s40609-
014-0005-x.

ALGÜL, Yahya, and İkram Yusuf YARBAŞI. “EFFECTS OF WOMEN’S EMPLOYMENT

AND FINANCIAL EMPOWERMENT ON DOMESTIC VIOLENCE: THE CASE OF TURKEY.”
Erciyes University Faculty of Economics and Administrative Sciences Journal, no. 59, Erciyes
University, Aug. 2021, pp. 197–220. doi.org/10.18070/erciyesiibd.935635.

Bhattacharya, Shubhasheesh, and Sonali Bhattacharya. “Battered and Shattered: Will

They Get Justice? A Study of Domestic Violence Against Women in India Based on National
Family Health Survey, 2005.” The Journal of Adult Protection, vol. 16, no. 4, Emerald, Aug.
2014, pp. 244–58. doi.org/10.1108/jap-07-2013-0032.

Dildar, Yasemin. “Is Economic Empowerment a Protective Factor Against Intimate Part-
ner Violence? Evidence From Turkey.” The European Journal of Development Research,
vol. 33, no. 6, Springer Science and Business Media LLC, Sept. 2020, pp. 1695–728.
doi.org/10.1057/s41287-020-00311-x.

Guarnieri, Eleonora, and Helmut Rainer. “Female Empowerment and Male Backlash.”
SSRN Electronic Journal, Elsevier BV, 2018. doi.org/10.2139/ssrn.3198483.

Heath, Rachel. “Women’s Access to Labor Market Opportunities, Control of Household

Resources, and Domestic Violence: Evidence From Bangladesh.” World Development, vol. 57,

14
Elsevier BV, May 2014, pp. 32–46. doi.org/10.1016/j.worlddev.2013.10.028.

Paul, Sohini. “Women’s Labour Force Participation and Domestic Violence.” Journal of
South Asian Development, vol. 11, no. 2, SAGE Publications, June 2016, pp. 224–50.
doi.org/10.1177/0973174116649148.

Raj, Anita, et al. “Longitudinal Analysis of the Impact of Economic Empowerment on Risk
for Intimate Partner Violence Among Married Women in Rural Maharashtra, India.” Social Sci-
ence Medicine, vol. 196, Elsevier BV, Jan. 2018, pp. 197–203. doi.org/10.1016/j.socscimed.2017.11.042.

Stöckl, Heidi, et al. “Economic Empowerment and Intimate Partner Violence: A Secondary
Data Analysis of the Cross-sectional Demographic Health Surveys in Sub-Saharan Africa.”
BMC Women’s Health, vol. 21, no. 1, Springer Science and Business Media LLC, June 2021.
doi.org/10.1186/s12905-021-01363-9.

Tsai, Laura Cordisco. “Household Financial Management and Women’s Experiences of In-
timate Partner Violence in the Philippines.” Violence Against Women, vol. 23, no. 3, SAGE
Publications, July 2016, pp. 330–50. doi.org/10.1177/1077801216642869.

World Health Organization: WHO. Violence Against Women. 9 Mar. 2021, who.int/newsroom/fact-
sheets/detail/violence-against-women.

15
VIII Appendix

16