Impact of Microcredit Program on Women's Empowerment in Rural Bangladesh

Gias Uddin Ahmed Choudhury

(T number 661112-T010)

Impact of Microcredit Program on Women’s Empowerment in Rural

Bangladesh

Business Administration Master’s Thesis

30 Credits

Term: Spring 2012

Supervisor: Per Skålen

i

Acknowledgement

Finalizing a thesis is a hard driving process. First of all, I am thankful to Almighty Allah without His blessings, the thesis would not have been finished. I could not have finished this process without active and sincere support and cooperation of some individuals and institutions. I am grateful to all of my well-wishers and academic mentors for making this thesis successful.

I would like to express my sincere gratitude to my honorable supervisor Professor Per Skålen, Karlstad University, Sweden, for his inspiring suggestions and helpful assistance in the time of writing this thesis. His patience, motivation, enthusiasm, and immense knowledge always insisted me to reach this stage of my study. His guidance helped me in all the time of research writing of this thesis. I am indebted to him not only for his tremendous academic support, but also for giving me the opportunities to complete the study.

And last of all to my loving, supportive, encouraging, and patient wife whose faithful support during the stages of this study, is so appreciated. He has supported me in countless ways, giving me good advice and helping me all the time. I would like to thank her from the bottom of my heart. I also wish to thank my beloved children for doing their best to understand a mother who extended her support with time and devotion during my study.

August 03, 2020

Gias Uddin Ahmed Choudhury

ii

Abstract

Background – This study is an attempt to explore the relationship between microcredit and the socio-economic empowerment of women in rural Bangladesh. Microcredit is simply the extension of a small amount of collateral-free institutional loans to jointly liable poor group members to generate employment and income enhancing activities. As it is too difficult for poor members to get loan from the formal credit institutions, Grameen Bank (GB) or other Non-Government Organizations (NGOs) provide small loans to vulnerable groups of the society by which they are expected to empower over his counterparts.

Research questions – RQ1: How does micro-credit affect different indicators of women empowerment in the rural areas of Bangladesh?

RQ2– Is the impact different from the male counterparts in the sample households?

Purpose – This study is an effort to find the impact of microcredit on a number of indicators of women’s empowerment in the rural areas in Bangladesh.

Methodology – Quantitative Regression Techniques such as Ordinary Least Square (OLS) and Instrumental Variable (IV) method have been applied to get the relationship between microcredit and women empowerment.

Conclusion – Applying nationally representative cross-section survey data, Bangladesh Integrated Household Survey (BIHS) 2015, this thesis is intended to find the causal linkage between microcredit and women empowerment’s with different dimensions of women’s decisions are taken as empowerment indicators: production, resources, income, leadership, savings and time. The analysis has been conducted at the household level. The study assumes that women empowerment is endogenous. After controlling for endogeneity in the estimation by using an instrumental variable (IV) ‘distance to the market’ this study finds a significant relationship between microcredit and different dimensions of women’s empowerment. Participation in the microcredit program is found to be significant in explaining some of the outcome indicators of empowerment for the sampled households.

Keywords: Microcredit, Women Empowerment, Ordinary Least Squares, Probit Model, Instrumental Variable Regression.

Paper type: Master’s thesis.

iii

Abbreviations

OLS: Ordinary Least Squares HH: Households

BIHS: Bangladesh Integrated Household Survey GB: The Grameen Bank

MC: Microcredit

WE: Women Empowerment

WEAI: Women’s Empowerment in Agricultural Index NGOs: Non-government Organizations

GOs: Government Organizations

5DWE: Five Dimensions of Women Empowerment 5DME: Five Dimensions of Men Empowerment IV: Instrumental Variable

FCS: Food Consumption Score HHS: Household Hunger Scale WB: The World Bank

ME: Marginal Effects

IFPRI: International Food Policy Research Initiative

iv

List of figures

Figure 1: Conceptual framework of women’s empowerment through microcredit ... 18

Figure 2: Indicators of Empowerment and percent of women not empowered ... 19

List of tables Table 1: Women’s Empowerment in Agricultural Index (WEAI) ... 23

Table 2: Definition and measurement of the key variables ... 24

Table 3: Descriptive statistics ... 28

Table 4: Probit Estimates of microcredit on women empowerment (Production, Resources, and Income Indicators of Women Empowerment) ... 30

Table 5: Probit Estimates of microcredit on women empowerment (Leadership and Time Indicators of Women Empowerment) ... 31

Table 6: Probit Estimates of microcredit on women empowerment (Production, Resources, Income, Speaking in Public, and Leisure Indicators of Women Empowerment) ... 32

Table 7: Marginal Effects ... 35

Table 8: Robustness Check ... 356

Table A8: Determinants (Except Access to Credit) of Women Empowerment (Method of Ordinary Least Squares) ... 49

Table A9: District Fixed Effect ... 50

Table A10: Empowerment by age group ... 51

Table A11: Robustness Check for Empowerment by age group ... 52

Table A12: Comparison of Women Empowerment by Borrower and Nonborrower Groups without microcredit: OLS ... 53

Table A13: STATA programming codes used in regression estimation ... 54

v

Table of Contents

Acknowledgement ... i

Abstract ... ii

Abbreviations ... iii

List of figures ... iv

List of tables ... iv

1. Introduction ... 1

1.1 Background of the research ... 1

1.2 Problem statement... 2

1.3 Importance of the Study ... 3

1.4 Research Gap ... 4

1.5 Study objectives ... 5

1.6 Research hypothesis ... 5

1.7 Structure of the thesis ... 7

2. Concept of empowerment and context of microfinance with reference to Bangladesh ... 8

2.1 Empowerment ... 8

2.2 Economic empowerment of poor women ... 8

2.3 Social empowerment of poor women ... 8

2.4 Microcredit... 9

2.5 Grameen Bank (GB) ... 9

2.6 Grameen Bank (GB) model for disbursing loan ... 10

3. Review of literature... 12

3.1 Previous works ... 12

3.2 Existing research gaps and contribution ... 16

4. Theoretical framework ... 17

4.1 Household utility function ... 17

4.2 Conceptual channel from microcredit to women empowerment ... 18

5. Methodology ... 20

5.1. Research approach ... 20

5.2 Research method ... 20

5.3 Sources of data ... 20

5.4 Population and sampling ... 21

5.5 Sample size ... 21

5.6 Women Empowerment in Agricultural Index (WEAI) ... 22

5.7 Definitions of Key Variables ... 24

vi

5.8 Empirical model and estimation method ... 26

5.9 Validity and reliability ... 27

6. Empirical findings and results ... 28

6.1 Descriptive Statistics... 28

6.2 Estimates of microcredit on women empowerment: Probit Model ... 29

6.3 Estimates of microcredit on women empowerment: Instrumental Variable (IV) Probit Model ... 32

6.4. Marginal Effects ... 34

6.5. Robustness Check ... 36

7. Conclusion, Policy Implications, and Limitations ... 38

7.1 Conclusion ... 38

7.2 Policy implications ... 39

7.3 Limitations and directions of future research ... 39

References ... 40

Appendix A: Some Case Studies of Crossing Poverty Lines after Having

Grameen Bank Microcredit ... 46

1

1. Introduction

In order to inform the inquisitive minded people about the brief idea of the research, this chapter will explain the background, problem statement, research objectives, and research hypothesis. Finally, the structure of the thesis has been presented.

1.1 Background of the research

Elimination of Poverty is a great challenge specifically for the developing and poor economies. Recent study by the World Bank (2019) finds that one in every four persons is living below the poverty line in Bangladesh. This implies that poverty is still a crucial issue for the policy makers in Bangladesh. A society cannot move forward without including the vast number of poor people. Including poor people in income generating process can gradually lead to empowerment. Given the family structure and unequal allocation of intra- household resources, empowerment of women is crucial to ensure welfare within a household. Empowerment of women not only improves financial capacity of the poor people but also changes power arrangement that sustained their marginalization and subordination.

Women are an important economic actor in the society. They take part in joint production processes to provide food for the family. They are simultaneously the primary caretakers of children and the elderly and the sick in the family. There is conclusive evidence from a number of studies conducted during the 1980s. The main findings of their studies are that mothers typically spend their income on food and health care for children, while men spend a higher proportion of their income for personal requirements.

Quisumbing et al. (1995) found that survival of a child would 20 times more likely in urban Brazil when income is controlled by women than men.

Yet women face significant restraints in maximizing utility in income or

expenditures. Very often they do not have equal access to productive inputs or to markets

for their goods and they own only 15 percent of the land across globe (USDS, 2014). They

also usually work longer hours than men both in house and in farms. Even then they earn

lower wages than men. Hence, it’s important to enhance bargaining power in decision

making process in the households. Microcredit could at least increase empowerment in the

families so that they can have assess and command over some goods and services.

2

1.2 Problem statement

In many poor economies including Bangladesh, non-governmental organizations (NGOs) pioneered micro-credit programs for the poor to uplift their economic conditions.

Following NGOs, government organizations have also started such programs mainly to monitor NGOs activities and have introduced such programs on a limited scale. Many of these programs mainly target women because more often than not financial markets are restricted for them and they have fewer bargaining powers over wage in the labor market.

Their role in the decision-making process is also very restricted specifically in the rural areas. The Grameen Bank (GB) of Bangladesh is perhaps one of best examples to provide collateral free loans to the poor people for small scale production operations. Over 90%

of GB’s clients are women. The GB believes that every human being has some potentials to contribute to the economy and its assumptions became quite correct when GB see that loan recovery rate for the women is quite satisfactory than their male counterparts.

Prior works on microcredit (Pitt & Khandker 1998; Pitt et al. 1999; Pitt et al. 2003;

Pitt 2000) find that the effects of credit program participation vary significantly across the genders in the programs. Pitt and Khandker (1998) find that consumption expenditure increases 18 Tk for every 100 Tk borrowed by women, but only 11 Tk for every 100 Tk borrowed by men. Pitt et al. (2003) find that provision of credit to women significantly improves the conditions of health and nutrition for both boys and girls, while provision of credit to men has no significant effect. These empirical observations imply that women empowerment may have increased bargaining power within the household in many areas such as production, use of income, enjoying leisure time.

Women empowerment is a multi-faceted concept while microcredit is one of the important indicators to enhance female empowerment (Malhotra & Mather, 1997; Johnson 2005; Kabeer 1999). Earlier empirical investigations find that microfinance can play a substantial role to empower women both economically and socially (Malhotra & Schuler, 2005; Pitt & Khandker, 1998; Pitt, Khandker, & Cartwright, 2006; Armendáriz & Morduch, 2010; Anderson & Eswaran, 2009; Bardhan & Klasen, 1999). A few studies find no impact (Banerjee et al. 2013) or adverse impact of microcredit on women’s decision-making power, specifically where loans are controlled by the husbands (Goetz & Gupta 1996;

Rahman 1999; Leach & Sitaram 2002). A good number of studies find that the economic impact of microfinance empowers women mainly through increase in income which also raises bargaining power within a household. Others argue the social impact of microfinance on women’s empowerment with greater autonomy in decision-making, and political and social inclusion.

Both academia and policy makers have identified microcredit is one of the useful

strategies to alleviate poverty. It’s a structured program under which small amount of loans

3

are granted to poor persons without collateral. This loan is provided to the jointly liable poor group members, especially to the poor women, for self-employment and income generating activities. Smith and Eric (2007) state that Muhammed Yunus established Grameen Bank in Bangladesh in 1976 which has been shifted into a formal bank in 1983 through a special law. Now it has become one of the largest microfinance organizations in Bangladesh.

Collateral is essential to get credit facilities in formal banking system in Bangladesh.

However, rural poor especially women have little to offer as collateral. When Grameen Bank model of microcredit has been introduced in Bangladesh, only then poor people were allowed to get loan without collateral. The inherent capacity of the poor people has been recognized and they are quite capable to earn money. Through the model of microfinance, a huge number of underprivileged poor especially women are able to take self-employment, creating self-confidence, and increasing awareness. Following Grameen Bank model microcredit, the others organizations such as in Bangladesh Rural Advancement Committee (BRAC), Association for Social Advancement (ASA) etc. have been established aimed for reducing poverty in Bangladesh.

Population density of Bangladesh is one of the highest in the world. About 168 million people live here and almost half of them are women (World Population 2014).

United Nations Development Program (UNDP 2014) states that the number of female- headed-households living below the poverty lines is significantly higher than that of male- headed households in Bangladesh. Even though, women constitute almost 50% of the total population, they enjoy unequal status in every aspect in life compared to men. They are mainly confined in domestic work, while men are exclusively working in the public domain like politics. Sebstad and Cohen (2000) mentions that women have limited control over household resources due to heavy domestic workloads and low level of education that cause them to vulnerable situation.

1.3 Importance of the Study

In underdeveloped and developing countries, women have inferior status due to gender-based discrimination. Positions of women are not the same as their counterpart, and occupational choices are usually narrower than men. On the other hand, women represent almost 50 percent of the population in these countries (Goswami 1998). Thus, a society or country cannot go ahead properly without involving this large number of women into the income generating process. Although, the microcredit program features differ from country to country, depending upon socio-economic conditions and microcredit is not a panacea to empower the poor women in reducing the poverty from the society.

Nevertheless, the idea of microcredit to empower the poor women in Bangladesh is still

an effective way to address poverty. Therefore, I think disbursing microcredit to poor

4

women would assist in income generating process mainly in the underdeveloped and developing countries for empowering them. Hence, this study has potential policy implications.

1.4 Research Gap

While for decades microcredit has focused on the development of women and the minimization of poverty, development specialists know comparatively little about how microcredit activities limit the scope of poverty reduction and empower women in different dimensions (Khandker, 2005). Despite its many success stories, it is important to understand the actual impact of microcredit program to justify support for microcredit over its alternatives. Given that women are excessively disadvantaged in Bangladesh and that they are the majority of microcredit beneficiaries, we must carefully consider the following policy questions. Do women actually benefit from the microcredit programs? If so, how much do they benefit the different indicators of women empowerment when they engaged in microcredit programs?

A handful of studies in the recent past have examined the relationship between microcredit programs and women empowerment in different contexts or dimensions. The main dimensions they have focused are: education, health, marriage, savings etc. However, almost all of the studies suffer from likely biased estimates due to potential due to endogeneity of decisions involved in credit programs and unobserved household characteristics. It is plausible that more empowered women are more likely to join in credit programs which would cause reverse causality in the equation of women empowerment.

Another point that can be mentioned here is that although many studies have focused on rural areas, these can be applied to sub-population in small surveys and also most of them are non-causal.

Considering the above facts and findings, this study is an attempt to address those two limitations and causally investigate the relationship between microcredit and women’s empowerment. Specifically, this study asks ‘does microcredit enhance borrowers’

participation in their household’s decision-making process’? There are many sectors

involved in household decision making process including education and health. However,

this study will emphasize how participation in microcredit would affect various dimensions

of women empowerment.

5

1.5 Study objectives

Considering the above problem statement, importance of the study, and research gaps, the main objective of this research is to apply nationally representative large -scale survey data, Bangladesh Integrated Household Survey 2015 (BIHS 2015), to find the causality between microcredit and women’s empowerment in Bangladesh. This study has two other general objectives:

1. To investigate the impact of microcredit on the empowerment of women for some selected indicators in the rural areas in Bangladesh.

2. To investigate whether the impact of microcredit substantially differ with the male’s counterpart.

1.6 Research hypothesis

I hypothesize that borrowers’ participation in microcredit activities operated by Non-Government Organizations (NGO) such as Grameen Bank (GB) leads to greater empowerment of credit members compared to non-credit members. And the credit membership is generally positively associated with women's empowerment. Conforming with the objectives of this study, the following hypothesis would be tested empirically:

Hypothesis 1: Null Hypothesis: Microcredit does not have any impact on the women’s control over household production.

This domain of production as an indicator of women empowerment has two major indicators. One is input in productive decisions and another is autonomy in production.

The former is basically decision making over food and cash-crop farming, livestock and fisheries. The latter is basically autonomy over buying inputs, what crops to grow and what livestock to raise and so and so forth. Considering data frequencies, representations, and domestic culture, I take up and include ‘input in productive decisions’ as a representative indicator of production.

Hypothesis 2: Null Hypothesis: Microcredit does not have any impact on the women’s control over household resources.

This domain of resources as an indicator of women empowerment has three major

indicators. The first one is ownership of assets which basically is sole or joint ownership

of major household assets. The second one is purchase, sale or transfer of assets which is

whether respondent participates in decisions to buy, sell, or transfer of assets. And the last

one is decisions on credit which is decision making concerning credit. Considering data

6

frequencies, representations, and domestic culture, I take up and include ‘ownership of assets’ as a representative indicator of production.

Hypothesis 3: Null Hypothesis: Microcredit does not have any impact on the women’s control over household income.

This domain of income as an indicator of women empowerment has only indicator control over use of income. It shows the sole or joint control over income and expenditures.

Hypothesis 4: Null Hypothesis: Microcredit does not have any impact on the women’s control over household leadership.

This domain of leadership as an indicator of women empowerment has two major indicators. One is ‘group member’ in productive decisions and the other one is ‘speaking in public’. The former is asking whether respondent is an active member in at least one economic and social group. The latter is asking whether the respondent is comfortable speaking in public concerning different issues such as family disputes, bargaining over wages for public works programs etc. Considering data frequencies, representations, and domestic culture, I take up and include ‘speaking in public’ as a representative indicator of leadership.

Hypothesis 5: Null Hypothesis: Microcredit does not have any impact on the women’s control over household time.

This domain of production as an indicator of women empowerment has two indicators: workload and leisure. Workload contains allocation of time to productive and domestic tasks and leisure contains satisfaction with the available time for leisure activities.

Considering data frequencies, representations, domestic culture, and domestic importance,

I take up and include ‘leisure’ as a representative indicator of time in final analysis.

7

1.7 Structure of the thesis

This thesis is organized as follows. This thesis is divided into seven chapters.

Chapter 1 describes background of the research, problem statement, importance of the study, research gap, study objectives, and research hypothesis.

Chapter 2 sets out concept of empowerment and the context microfinance with reference to Bangladesh the background and context for this study. It summarizes the evolution of microfinance sector in Bangladesh and the primary goals targeted by these institutions.

Chapter 3 describes review of literature to find out research gaps in the study of household level microcredit and women empowerment. The review of Literature is to grasp theoretical foundation to identify possible channels through which changes in intra- household bargaining occur in the distribution of small loans to poor women.

Chapter 4 explains theoretical framework which includes household utility function and conceptual framework from microcredit to women empowerment.

Chapter 5 lays down the Data and Methodology. The empirical methodology is outlined for testing this hypothesis. The section sets out the difficulties involved in controlling for biases, and explains the concern for endogeneity in this context. The section conceptualizes treatment measure as participation in a microcredit program, for at least one complete loan cycle. Using a comparable group of prospective clients who have accessed loans for the first time in the past month we are able to estimate the impact of treatment on indicators of women’s empowerment. Estimators for treatment effects are outlined. Further, matching and instrumenting is proposed to address problems stemming from endogeneity and simultaneity bias.

Chapter 6 discusses empirical findings of the associations between credit and women empowerment.

Finally, chapter 7 presents main findings and conclusions of this research and

briefly hints the potentially of extending the research into further arenas with limitations.

8

2. Concept of empowerment and context of microfinance with reference to Bangladesh

This section briefly explains concepts of microcredit, social and economic empowerment of women, Grameen Bank (GB) model’s functional features.

2.1 Empowerment

Eyben et al. (2008) states that empowerment is basically regarding the power within themselves. It facilitates people to accomplish courageous things they never thought themselves to be capable of doing and the power that comes from working along with others to claim what is correctly theirs. Empowerment broadens poor people freedom of choice and action, expanding their assets and abilities and enabling them to participate in, negotiate with, influence, control and hold accountable institutions that affect their lives.

On the other hand, Empowerment is a process, by which people learn to think critically about their own situation and possibilities, and see things in a different way. Empowerment implies that individuals and organized groups are able to assess their world differently and to understand that vision by changing the relations of power that are keeping them in doing tasks.

2.2 Economic empowerment of poor women

Eyben et al. (2008) explains that economic empowerment means capacity of poor women to participate in, contribute to and benefit from growth process and recognizes the value of their contributions. It also makes it possible for them to negotiate a fairer distribution of the benefits of growth. At the same time people thinking beyond immediate survival needs and thus able to recognize and exercise agency and choice.

Kumar et al. (2015) shows that the financial sustainability emphasizes women’s own income, generating activities. Women’s entrance to savings and credit gives them a greater economic role in decision making process. When women control decisions regarding credit and savings, they will optimize their own and the welfare of households. The investment in women’s economic activities will improves employment opportunities for women.

2.3 Social empowerment of poor women

Eyben et al. (2008) define social empowerment as taking steps to change society so

that one’s own place within it is respected and recognized on the terms on which the person

themselves want to live, not on terms dictated by others. Empowerment comes in as a

9

process whereby women develop a sense and capacity to improve the quality of their social relationships and to secure respect, dignity and freedom from violence.

After giving the specific attention to enhancement of women’s entrance in micro finance, women increased economic activity and control over income with improves their skills, mobility, and access to knowledge and support networks (Kumar et al. 2015).

2.4 Microcredit

Mandal (2009) defines microcredit as a system of credit delivery and savings mobilizing scheme which is designed to meet the unique financial necessities of the poor.

The financing scheme allows the receivers to improve their living standard by getting loan without collateral. The credit approach has evolved as an economic development tool intended to benefit low-income group people including the self-employment. Micro financing institute could be non-governmental organizations (NGOs), savings and loan cooperatives, credit unions, government bank, commercial banks, or non-bank financial institutions.

On the other hand, microcredit programs extend small loans to very poor for generating income through self-employment. Unlike the other formal banking systems, this loan program does not have any requirement of collateral security from the borrowers. The overall success of microcredit programs rests on an alleviation of poverty and long-term sustainability (Chowdhury, 2004).

Character, Capital and Capacity are the three C’s in microcredit program. Character means historical background of a borrower such as borrower’s honesty and capacity to refund the loan and how a borrower handles his previous debt liability. Capital means present assets of borrower has which would be helpful for him to repay the loan timely and Capacity means how much debt a borrower can maintain as per his income, and still be able to pay that debt off (Grameen Bank, 2018a).

2.5 Grameen Bank (GB)

After a famine caused severe starvation in 1974 in Bangladesh Professor

Muhammad Yunus, an economist experimented to find out a way to remove poverty from

Bangladesh. He thought from small amount of loan a poor person could start his own

business and it would be a sustainable solution to hunger poverty. He established Grameen

Bank project in the village of Jobra, Bangladesh in 1976. It was shifted into a formal bank

in 1983 through a special law and now it is a one of the largest microfinance organization

in the world. Owners of the Grameen Bank are poor borrower of this bank and most of

them are women. Grameen Bank is different financial organization in Bangladesh. It

founded absolutely for the poor people who has a few or no assets. Method of Grameen

10

Bank has appreciated all over the world than any other microfinance model (Smith & Eric 2007).

2.6 Grameen Bank (GB) model for disbursing loan

A bank branch is set up with a branch manager and a number of bank workers, covering an area of about 15 to 22 villages. The manager and workers start by visiting villages to familiarize themselves with the local milieu in which they will be operating and identify prospective clientele, as well as explain the purpose, functions, and mode of operation of the bank to the local population. Groups of five prospective borrowers are formed; in the first stage, only two of them are eligible for, and receive, a loan. The group is observed for a month to see if the members are conforming to rules of the bank. If the first two borrowers successfully repay their weekly instalments, then the next two members are become eligible to apply for their own loans. Similarly, when they make repayments successfully then the final member may apply for loan. Because of these restrictions, there is substantial group pressure to keep individual records clear. In this sense, collective responsibility of the group serves as collateral on the loan (Grameen bank, 2018 b).

Grameen Bank (GB) credit delivery system has the following features ¹ :

1. There is an exclusive focus on the poorest of the poor. Exclusivity is ensured by:

i) Establishing clearly the eligibility criteria for selection of targeted clientele and adopting practical measures to screen out those who do not meet them

ii) In delivering credit, priority has been increasingly assigned to women iii) The delivery system is geared to meet the diverse socio-economic development needs of the poor

2. Borrowers are organized into small homogeneous groups. A center is formed to organize activities of credit. The Centers are functionally linked to the Grameen Bank, whose field workers have to attend Centre meetings every week.

3. Special loan conditionality which are particularly suitable for the poor.

These include:

i) Loans given without any collateral

ii) Loans repayable in weekly installments spread over a year

iii) Eligibility for a subsequent loan depends upon repayment of first loan

iv) Individual, self-chosen, quick income generating activities which employ the skills that borrowers already posses

1

www.grameen.com

11

v) Close supervision of credit by the group as well as the bank staff vi) Stress on credit discipline and peer support solidarity.

vii) Special safe-guards through savings to minimize the risks that the poor confront

viii) Transparency in all bank transactions most of which take place at center

meetings.

12

3. Review of literature

This section highlights the relevant review of literatures linking microcredit to empowerment in the context of developing countries including Bangladesh. Through the related review of literatures, a research gap has been explored so that this thesis can address the issues in methodical ways.

3.1 Previous works

Along with improving the income earning power, a usual assumption existed about micro-credit that it enables women especially poor women to have a strong empowerment within their family and community. Studies of Rahman (1986), Pitt and Khandker (1996), Hashemi et al.(1996) find that while women are contributing more income to the household then they possess more authority in the household decision making and also their self-esteem gets high in the perspective of the society which was quite rare while women were engaging themselves only in doing the household chores without contributing any income. There’s another way existed by which microcredit ameliorates women empowerment by creating the opportunities for them to interact with the outer world from the walls of their households. This has helped them to boast up their courage, self-esteem to exercise greater authority within their households and community.

Social entrepreneurship programs exhibit a lot of positivity in the empowerment of women in India by strengthening economic security, development of entrepreneurial behaviour and contributions to family (Datta & Gailey 2012).Also in India the Gram Mooligai Company limited (GMCL) which is the first female community enterprise providing micro-entrepreneurship facilities to rural women enhances women’s productive capabilities, leadership skills. Although the success of this program has not attained its full potential result due to the existence of caste system and patriarchal norms in rural India (Torri & Martinez, 2014).The study of Khan and Noreen (2012) finds the significance of microfinance in ameliorating the decision making capability of women within households by improving their participation in economic activities. In their study based on Bahawalpur district Pakistan showed that women empowerment depends on age, husband education, father inherited assets, marital status, a number of alive male children and amount of microfinance. Their analysis show that women using their loans by themselves have better effect of empowerment rather than used by others.

Unless the esurient of proper utilization of microfinance its impact wouldn’t be that

great as expected. Kato and Kratzer (2013) finds that in Tanzania women members of

microfinance institutions (MFI) are seemed to be more empowered than non-member

women. They conduct their survey by including data of 305 members of MFI’s and 149

non-members of MFI. The analysis of the above obtained data by Mann-Whitney U test

showed a significant difference between the members and non-members of MFI’s

13

regarding women empowerment. The result showed that women members of MFI’s seemed to possess more control over savings and income which enables them to have great self-efficacy, self-esteem and greater freedom of mobility outside the household.

Kabeer (2001) has identified four major types of reasons which affect women empowerment through micro credit. First of all, empowerment is a multi-dimensional concept, and some dimensions improve while others do not or even get worse. Secondly, Women are a heterogeneous category in based on class, religion, economic status, occupation, size of loan and also, they have individual differences. As a result, the empowering effect of microcredit will be different for different women. Thirdly, the time dimension is important as it can empowerment because it takes considerable time to take root in a patriarchal society. Longer time horizon may have effect on empowering whereas the shorter time horizon may not. Finally, methodological differences is very important because empowerment is a difficult concept to measure, and conflicting conclusions may emerge from differences in methodologies adopted to measure and evaluate. Ackerly (1995) gives a good idea about empowerment as an outcome. Borrower wisely invests money in a successful enterprise, her husband stops beating her, she sends her children to school, she improves the health and nutrition of the family, and she participates in major family decisions. She gains control over reproductive decisions.

Pitt et al. (2006) have examined the effect of mans and women’s participation in group-based microcredit programs on various indicators of women’s empowerment. They have found that credit programs led to women taking a greater role in household decision making processes. It ensures greater access to financial and economic resources, broader social networks, greater bargaining power with their husband and greater freedom of mobility. And also have increased spousal communication in general about family planning and parenting concerns. At the same time, the effects of male credit on women’s empowerment were generally negative. Male microcredit has a negative effect on overall measure of empowerment for eligible households and specifically on women’s control of resources, finance, freedom of movement, development of networks, fertility and parenting decision.

Amin et al. (1998) have found that women’s membership in NGO promoted credit

program has significantly positive relationship with women empowerment using

quantitative data. Qualitative evidence has found by conducted a series of focus group

sessions with the female recipients of NGO credits. Focus group sessions concentrated

on women’s status and rights. They are more confident, intelligent, and no more idle

because of their NGO membership. Women have greater courage now; they can stand on

their own two feet. That means that the loanees themselves support the positive

relationship between NGO credit membership and women empowerment.

14

Shami et al. (2018) examined the effects of microcredit on women empowerment using a cross-sectional survey to 474 old and new clients in Malaysia. They argued that microcredit empowers women only those who live in patriarchal family. It is because of the norms that empower man and give them authority of making household decision.

Microcredit empowers partially and potentially fosters gender equality. Participations in microcredit program increases women’s income, mobility out home, health expenditure decision, daily household expenditure, and loan order. The microcredit is an imperative economic resource which provides women a bargaining power, to improve their decisions and positions in the household and take control over their lives.

Steele et al. (1998) explore the impact of participation in women’s savings and credit groups that embodied by Save the Children USA on women’s empowerment, the use of contraceptive and fertility in a rural area of Bangladesh. Panel survey data are used in this study. The characteristics of women in savings groups are qualified by quasi-experimental design. The findings of this study showed that women who are attached in saving groups are more educated and socially independent than those women who are not attached to the program. Pre intervention measures of empowerment are taken into consideration to control selection bias and analysis the impact of savings groups on 1995, levels of empowerment and fertility behaviour. This study has found positive impacts of the credit groups owing to children’s education, age at marriage, and modern contraceptives use.

Afrin et al. (2008) examines that the micro credit programs help the rural women borrowers only to survive and those programs do not assistance them to amplify entrepreneurial capabilities. Primary data are collected from rural women borrowers of two leading organizations i.e., ASA in private sector and BRDB in public sector. Covariance and structural modelling program are verified in this study. By Confirmatory Factor Analytic (CFA) model, observed variables are connected to unobserved variables.

Multivariate Analysis technique is used to define the factors that responsible to development women entrepreneurship in the rural areas of Bangladesh with microcredit.

The results have found from this study that the financial management skills and the group of the women borrowers have important relationship in the development of rural women entrepreneurship in Bangladesh.

Rahman et al. (2009) demonstrate factors (age, education level, income etc.) that affected women empowerment on microcredit borrowers. Primary data are used from borrowers of two major microcredit institutions i.e., the Grameen Bank and BRAC, from three districts- Gazipur, Dinajpur and Chokoria, in Bangladesh. Quasi-experimental approach is used by comparing microcredit borrowers with non-borrowers as control group selected from non-programmed villages as to reduce heterogeneity bias in this study.

This study has showed that non-borrowers are equally empowered as like microcredit

borrowers and also have found that age and educational attainment of women are

15

significant factors in empowerment. Hashemi et al. (1996) found that credit had elevated women’s empowerment with involvement in decision-making within the household, mobility in the public domain, asset holding in their own names, and reduction in domestic violence. Rahman (1986) and Pitt and Khandker (1996) also find that the borrowing women have gained a greater say in decision-making processes within the household.

Kabeer (2001) distinguished improvement in empowerment outcomes for borrowing women in terms of several dimensions: greater sense of self-worth, greater degree of social inclusion, increased voice in household decision-making, reduced domestic violence, and greater propensity to educate daughters. Those studies gave the positive conclusion about the outcomes of empowerment. But in several other studies conducted by Goetz and Sen Gupta (1994), Ackerly (1995) find doubt about the effect of microcredit on women empowerment.

Ackerly (1995) took the borrower’s ‘accounting knowledge’ as the indicator of empowerment as she has knowledge of the input costs for loan-funded enterprise, its product yield and its profitability. Goetz and Sen Gupta (1994) conducted ‘managerial control’ as an index of indicator of empowerment. As the index is based on how the loan was used by the women whether they knew about it or they provided any labour input instead of loan. Montgomery et al. (1996) also considered managerial aspect as they took into account particularly the projects that were managed by the woman or jointly with her husband or some other male member of the household. Especially they used joint management as an indicator of lack of empowerment. Those three studies came up with negative conclusions about the empowering effect of credit as women did not have sufficient accounting knowledge or managerial control of the loan-funded enterprises.

The distinction between researchers about microcredit’s effect of empowering women existed on the basis of indicators. The process-based proxy indicators show that credit has no role in empowering women whereas the outcome based direct indicators which can directly measure what they require to assess find the significant effect of women empowerment. Kabeer (2001) finds that the choice of process indicators may leads to ambiguous result like showing divergence in the direction about outcome and process while they choose wrong indicators. Managerial control index used the following hypothesis that if women don’t contribute labour in the production process they wouldn’t be empowered (Goetz & Gupta, 1994) though there’s a lack of justification regarding this hypothesis.

Ackerly (1995) uses accounting knowledge as indicator but there is also less

justification about this indicator to be a concluding indicator. Montgomery et al. (1996)

considered sole management as an indicator of women empowerment and also considered

joint management as an indicator of lack of women empowerment. Still there’s exist

interpretation about joint management advocating women empowerment as women need

16

to overcome some gender related obstacles with the cooperation of males.

The problems of the above-mentioned studies are that most of the studies used poor proxy variable to women empowerment which has severe consequences on a statistical test of significance of coefficient of variables. The other issue is that applying those studies does not take much care of applying appropriate model in self-reported women empowerment.

3.2 Existing research gaps and contribution

Most of the prevailing studies related to microcredit and women empowerment are

shown association and the causal path is either not addressed or obscure. Another

deficiency in the existing studies are that they have used either case studies or qualitative

studies. Moreover, the most of the prevailing studies have applied to sub-populations. This

study is an endeavor to apply quantitative regression techniques to find the causality from

microcredit to female empowerment in the rural areas in Bangladesh. The concept of

women empowerment considered here is comprehensive which includes five dimensions

and ten indicators. The causal path has been addressed by applying instrumental variable

approach.

17

4. Theoretical framework

This chapter briefly discusses the household utility function and channels of microcredit to women.

4.1 Household utility function

If women get access to credit through microfinance institutions, their choice sets could be expanded by providing them with greater access to economic opportunities. This helps to economic empowerment. This translates into greater bargaining power in the intra-household decision making process. Pitt, Khandker, and Cartwright (2006) explains collective household model’ to motivate this change. In this simple collective household model, the two partners look for equilibrium. They maximize utility taking into account the utility of the other partner. Thus, the household simply optimizes a weighted utility function of the form:

𝑽 = 𝜽 𝒖 _𝟏 (𝒙) + (𝟏 − 𝜽)𝒖 _𝟐 (𝒙)

Where 𝜽 is between 0 and 1 inclusive and stands for the relative power of the female compared to male partner in the household decision making process; u 1 (x) captures the preferences of the female over a bundle of consumption goods ‘x’ and u 2 (x) are the male preferences over the same consumption bundle.

The weight 𝜽 can be regarded as the power distribution within the household.

When 𝜽 becomes zero, women do not get any preference (no power within the household) and utility function is specific to male. When 𝜽 becomes one, women do get all the preferences (all power within the household) and utility function is specific to female.

When 𝜽is between 0 and 1, the power has been distributed into male and female.

18

4.2 Conceptual channel from microcredit to women empowerment

After carefully reviewing the literature on women empowerment, through three channels microcredit can affect women empowerment. Figure 1 below shows the specific channels. First one is that empowerment can be achieved by enhancing income. The second channel is that empowerment can be brought through control over credit or savings. The third one is that women’s status can be increased through generating income.

Figure 1: Conceptual framework of women’s empowerment through microcredit Source: Salman (2009)

As poor people in the rural areas has lack or restricted access to the formal credit

market and they are not usually habituated to save for the hard times, women’s

access to credit is the first and foremost thing to consider. In a household there are

two potential clients who can borrow from the lenders: male and female. Empirical

observation confirms that when women take loan from GB or any other NGOs,

repayment rate of loan is very high compared to their male counterparts. Hence,

the lenders purposively select women as a potential client in formal labor market.

19

Once women get access to credit, this can be channelized three ways to empower women. Though a woman is initially a group borrower with the lender’s perspective, she is an individual borrower in the household.

Figure 2: Indicators of Empowerment and percent of women not empowered Source: Sraboni, Quisumbing, and Ahmed (2014)

Figure 2 gives us some ideas about the prevailing situations of women not empowered in the nationally representative Bangladesh Integrated Households Survey in 2011. The numerical figures are calculated by Sraboni, Quisumbing, and Ahmed (2014). These findings would guide us to get the improvement on the women’s empowerment scenarios in a recent survey of Bangladesh Integrated Household Survey 2015 (BIHS 2015).

Ten indicators constitute a Women Empowerment (WE) index. Of these indicators,

it is observed that more than 50% are not comfortable speaking in public and about

35% women do not have control or influence over ownerships of assets. Another

aspect of WE indicator is input in productive decisions in which about 37% women

are not empowered. Hence, this information’s would guide us to get a comparable

statistic in our current study.

20

5. Methodology

In this section, I will briefly detail out the sources of data and methodological aspects of this study. The two sub-sections below show data, empirical model and econometric method of data analysis.

5.1 Research approach

This study aims to test prevailing hypotheses that relates microcredit at the household level to the women empowerment within the household. This necessitates finding the relationships among variables in a particular specification. Thus, the deductive approach seems to be the most suitable approach. Usually, the process of deductive approach follows few steps. Firstly, the researcher has to select the theories that would be most suitable consistent with the aims and objectives of the study to test various hypothesis. Then, the study requires providing operational definitions of the variables involved in the study. Finally, the hypothesis has to be tested statistically in data to confirm or reject the hypothesis (Gray, 1992; Saunders et al. 2011).

5.2 Research method

This study employs quantitative research method to examine the relationship between the variables specified by an Equation. Consequently, this study aims to examine and explain the relationship between microcredit and women empowerment in the context of rural areas in Bangladesh. The Ordinary Least Square (OLS) has been applied where dependent variable is continuous, Bivariate Probit Model (PM) has been applied where both independent and dependent variable are dummy or indicator variables, and Instrumental Variable Probit Model (IVPROBIT) has been applied to address endogeneity issues. Lastly, robustness check has been applied.

5.3 Sources of data

In this thesis paper, I have used secondary data. In particular, this study uses Bangladesh Integrated Household Survey (BIHS) 2015 data which is very rich one of the nationally representative survey data in Bangladesh. ² BIHS collects detailed data on (a) agricultural production and practices, (b) dietary intake of individual household members, (c) anthropometric measurements of all household members, and (d) data to measure

2

The BIHS is an open access dataset available at http://www.ifpri.org/blog/ifpris-bangladesh-

integrated-household-survey-bihs-second-round-dataset-now-available. It is the initiative of

International Food Policy Research Institute (IFPRI) to intensively study the dimensions of

agriculture, food security, nutrition, and empowerment. Recently, IFPRI’s BIHS dataset has

emerged as a global public good to explore the standard of living in various dimensions in the rural

areas in Bangladesh.

21

women’s empowerment in agriculture index (WEAI). The BIHS 2015 covers about 6500 households in 325 Primary Sampling Units (PSU).

5.4 Population and sampling

Though secondary data are sometimes available at less effort, one should take into account the fact that these data may not fit researcher’s problem entirely. The data may have been gathered for different purposes in the previous studies (Ghauri & Grønhaug, 2010). This study has overcome the problems of secondary in quantitative study. One of the main objectives of collecting BIHS data is to find various dimensions of living conditions in the rural areas of Bangladesh with particular attention paid to women. The BIHS also contains separate module of women’s empowerment and detailed information in access to microcredit, I can potentially elucidate BIHS data to fit into my objective of studying the impact of microcredit on women empowerment. The model F of BIHS questionnaire contains information on loans or credit and module WE contain information on the indicators of Women Empowerment in Agriculture.

The BIHS questionnaires include several modules that provide an integrated data to answer a variety of research questions. There is also a separate questionnaire for self- identified primary male and female decision makers in sampled households. This study relied primarily on information concerning household demographics, educational attainment, occupation and employment, food and non-food consumption and expenditures, household-level agricultural production and livestock holding, household assets, and other modules.

5.5 Sample size

The BIHS 2015 sample consists of about 6500 households. This study analyses only farm households in the agricultural sector. The women empowerment in agriculture index (WEAI) relies on information collected from both primary male and female adults in the household. However, this study has only selected women in the final sample. This study drops the age if the age of household head is above 73 years and below 17 years to make consistent dataset. The final estimation sample consists of 6219 households. To examine women’s empowerment within the household, I divide the whole sample into two groups:

non borrower households and borrower households. The borrower groups are the one who is eligible to get loan (i.e. not over 0.5 acre of land), applied for the loan and finally has been approved. The nonborrower group has been created from the household members who have the maximum 0.5 acres of land but for some reasons they did not finally get the credit. As it is a group credit, all members have to abide by the group policies.

Hence, the final sample consists of about 4900 borrower households and about 1300

borrower households’ samples in final analysis.

22

5.6 Women Empowerment in Agricultural Index (WEAI)

Measurement of empowerment is a tricky one particularly women empowerment (WE) in rural setting of Bangladesh. Mahmud, Shah, and Becker (2012) have explained three issues of measurement. First, the process of empowerment is not directly observable assuming it is a latent variable. So, it can be approximated by proxy variables or indicators.

For instance, the initial resources that women can bring upon to the spouse may not translate into bargaining power in the rural sectors. For example, educated women may not translate her bargaining power on husband’s family if other factors are dominant over education in the family. Motivations and hard works can also be regarded as such types of factors. Some other indicators such as income and production decisions may have commonly identified as included observable actions. Second, concept of empowerment is a multi-dimensional process. Gender inequality exists across all spheres of socioeconomic dimensions. If we assume that increased bargaining power enhances women’s well-being by reducing gender inequality in socioeconomic dimensions, then the causal pathways of influence from WE to favourable outcomes also need to be identified. This implies indicators need to be specified and measured across various dimensions. Third, context is critical. The particular pathways of WE vary from context to context.

Even within the same context, all women may not experience empowerment within the same dimensions. Different ways household wealth and age can shape the empowerment process. To overcome all of these issues, Women Empowerment in Agricultural Index (WEAI) has been developed which is discussed below.

For the individual-level and household level analysis of women’s various dimensions of empowerment, Table 1 below shows the different indicators of women empowerment. The indicators in Table 1 show Women Empowerment in Agricultural Index (WEAI) in the context of rural areas in Bangladesh. WEAI is an innovative survey- based index developed by International Food Policy Research Initiative (IFPRI), the Oxford Poverty and Human Development Initiative (OPHI), and the US Agency for International Development (USAID) to measure and track women empowerment (WE).

It is a valuable tool for policy makers, academics, and development organizations to get

updated information on WE in the agricultural setting.

23 Table 1: Women’s Empowerment in Agricultural Index (WEAI)

Domain Indicator Definition of indicator

Production

Input in productive decisions

Sole or joint decision making over food and cash-crop farming, livestock, and fisheries

Autonomy in production

Autonomy in agricultural production (e.g., what inputs to buy, crops to grow, what livestock to

raise, etc.). Reflects the extent to which the respondent’s motivation for decision making reflects his/her values rather than a desire to

please others or avoid harm.

Resources

Ownership of assets Sole or joint ownership of major household assets

Purchase, sale, or transfer of assets

Whether respondent participates in decision to buy, sell, or transfer his/her owned assets Income Control over use of

income and Savings

Sole or joint control over income and expenditures

Leadership

Group member

Whether respondent is an active member in at least one economic or social group (e.g., agricultural marketing, credit, water users’

groups)

Speaking in public

Whether the respondent is comfortable speaking in public concerning various issues such as intervening in a family dispute, ensure

proper payment of wages for public work programs, etc.

Time

Workload Allocation of time to productive and domestic tasks

Leisure Satisfaction with the available time for leisure activities

Source: Alkire et al. (2013)

The WEAI constitutes of two sub-indexes: five domains of Women’s

Empowerment (5DE) and the Gender Parity Index (GPI). The 5DE captures five

domains: decision making over agricultural production, decision making power over

productive resources, control over use of income, leadership rules in the community,

and time utilization. There are total ten indicators in five dimensions. In our study, five

indicators have been selected out of ten. These five indicators have been selected on

the assumption of cultural context in Bangladesh and the frequencies in the data so that

these indicators could be the representative of all ten indicators.

24

5.7 Definitions of Key Variables

Table 2 defines the key explanatory variable, other independent variables, and vector of dependent variables. The explanatory variables are segmented into three categories at the household level: demographics, socioeconomic, and food security.

Table 2: Definition and measurement of the key variables Types of Variables Name of the variables Metric Main variable of

interest Microcredit Indicator variable (1=access to microcredit; 0=otherwise) Other explanatory

variables

a) Demographics Household head marital status Indicator variable (1=household head married; 0=otherwise) Household head religion Indicator variable (1=household

religion is Muslim; 0=otherwise) Household head gender Indicator variable (1=household

head male; 0=otherwise) Household head age Age of household head in years Household head level of

education Categorical (0. Reads in class i; 1.

Completed class i; 2. Completed class 2; 3. Completed class 3; 4.

Completed class 4; 5. Completed class 5; 6. Completed class 6; 7.

Completed class 7; 8. Completed class 8; 9. Completed class 9; 10.

Completed SSC/Dakhil; 12.

Completed HSC/Alim; 14.

BA/BSC/Fazil; 15. BA/BSc Honors/Fazil; 16. MA/MSC and above/Kamil)

Household size Number of household members Total dependency ratio Number of members under 15 or

over 65 divided by the number of members between 15 and 65 b) Socioeconomic

Total land owned and operated

(ha) Hectares of land owned and

operated Number of rooms in the

household Number of rooms

Electricity connection Indicator variable (1=household has electricity connection;

0=otherwise)

Water source Categorical variable (1. Supply Water (piped), outside; 2. Supply Water (piped) inside house; 3.

Own tube well; 4. Community tube well; 5. Rain water; 6. Ring Well/ Indara; 7. Pond/River/

Canal; 9. Shallow tube well; 10.

25 Deep tube well; 11. Others; 12.

Other tube well Number of assets owned by the

household Number of different types of assets owned by the household Distance to the market Distance to market, km

Distance to public

transportation Distance to public transportation, km c) Food Security

Food consumption score Numeric score on 0 to 120 Household hunger scale Composite score ranging from 0

to 6 Dependent

variables (Indicators of Women

Empowerment)

Input in productive decisions Indicator variable (1=women has command over inputs in productive decisions;

0=otherwise)

Ownership of assets Indicator variable (1=women has command over ownership of assets; 0=otherwise)

Control over use of income Indicator variable (1=women has control over use of income;

0=otherwise)

Speaking in public Indicator variable (1=women can speak in public;

0=otherwise)

Leisure Indicator variable (1=women

has workload over leisure;

0=otherwise)

26

5.8 Empirical model and estimation method

I follow the empirical methods similar to Pitt and Khandker (1998) and Pitt, Khandker, and Cartwright (2003) to examine the effect of rural women’s credit participation on women empowerment. Specifically, this study estimates the conditional demands for women’s empowerment on the individual’s participation in the microcredit program. The reduced form system of equations that are to be estimated specified as follows:

𝑀𝐶 _𝑖 = 𝛼 _𝑚𝑐 𝑋 _𝑖 + 𝜋𝑍 _𝑖 + µ _𝑖 (1)

𝑊𝐸 _𝑖 = 𝛼 _𝑤𝑒 𝑋 _𝑖 + 𝛿 _𝑖 𝑀𝐶 _𝑖 + µ _𝑗 (2)

Where 𝑀𝐶 _𝑖 is a dummy variable measuring whether a women participate in the microcredit program, 𝑋 _𝑖 is a vector of household and borrowers characteristics such as age, education, number of children etc., 𝑊𝐸 _𝑖 is the vector of conditional demand for women’s empowerment outcomes, 𝑍 _𝑖 is the set of exogenous instruments that affects 𝑀𝐶 _𝑖 but not the measures of women’s empowerment, 𝑊𝐸 _𝑖 conditional on 𝑀𝐶 _𝑖 , 𝛼 _𝑚𝑐 , 𝜋, 𝛼 _𝑤𝑒 , and 𝛿 _𝑖 are the parameters to be estimated, 𝛿 _𝑖 is the main parameter of interest that measures the impact of participation of microcredit program on a number of indicators of women empowerment.

As both of the key independent variable (i.e. MC) and outcome variables are binary, the bivariate probit model has been applied estimate the impact of participation in the microcredit program impact. In the bivariate probit model endogeneity is addressed and treatment has been tried to identify by employing exogenous instrument ‘distance to the market from the home’, which is highly correlated with the timing of the decision to participate in the program, but is uncorrelated with indicators of empowerment. The reduced form endogenous model can be specified as:

𝑀𝐶 _𝑖 = 𝛼 _𝑚𝑐 𝑍 _𝑖 + 𝜀 _𝑖 (3)

𝑊𝐸 _𝑖 = 𝛼 _𝑤𝑒 𝑋 _𝑖 + 𝛿 _𝑤𝑒 𝑀𝐶 _𝑖 + 𝜀 _𝑗 (4)

Where, Z i is the vector of observables and the system is appropriately identified if at least one variable in Z i is not contained in X i .

To allow for the possibility that the unobserved determinants of program participation in the program are correlated with the unobserved determinants of women empowerment, it is assumed that ε i and ε j are distributed bivariate normal with E(ε i )= E(ε j )=

0, Var (ε i )= var(ε j )= 1, and Cov(ε i ,ε j )= ρ. For the case where both treatment and outcome

variables are dichotomous, there are four possible states WE i =1 or WE i =0, MC i =1 or