Publication thesis

Global Journals Inc. (A Delaware USA Incorporation with “Good Standing”; Reg. Number: 0423089) Sponsors: Open Association of Research Society Open Scientific Standards Publisher’s Headquarters office Global Journals ® Headquarters 945th Concord Streets, Framingham Massachusetts Pin: 01701, United States of America USA Toll Free: +001-888-839-7392 USA Toll Free Fax: +001-888-839-7392 Offset Typesetting Global Journals Incorporated 2nd, Lansdowne, Lansdowne Rd., Croydon-Surrey, Pin: CR9 2ER, United Kingdom Packaging & Continental Dispatching Global Journals Pvt. Ltd. E-3130 Sudama Nagar, Near Gopur Square, Indore, M.P., Pin: 452009, India Find a correspondence nodal officer near you To find nodal officer of your country, please email us at [email protected] eContacts Press Inquiries: [email protected] Investor Inquiries: [email protected] Technical Support: [email protected] Media & Releases: [email protected] Impact of Microfinance on Women’s Economic Empowerment: A Case Study on Omo Microfinance in Gimbo Woreda, Ethiopia Lecturer at Jimma University, Ethiopia. Tesfaye Melaku, Jemal Abafita(PhD) CONTENTS Page No. Chapter – 1 Introduction 1-5 1.1. Background of the Study 1.2. Statem ent of the Problem 1.3. Objective of the Study 1.3.1. General Objective 1.3.2. Specific objectives 1.4. Significance of the Study 1.5. Scope and Lim itation of the Study 1.6. Organization Chapter – 2 Literature Review 5-10 2.1. Theoretical Literature Review 2.1.1. Definition and Scope of Microfinance 2.1.2. Conceptualizing Wom en’s Em powerm ent 2.1.3. Microfinance and Wom en Econom ic Em powerm ent 2.1.4. Overview of Microfinance Sector in Ethiopia 2.1.5. Overview of Om o Microfinance Institution (OMFI) 2.2. Em pirical Literature Chapter – 3 Research Methodology 11-17 3.1. Brief Description of the Study Area 3.2. Data Type and Source 3.3. Study Population 3.4. Sam pling Design and Technique 3.4.1. Sam ple Size Determ ination 3.5. Method of Analysis 3.5.1. Propensity Score Matching (PSM) 3.5.2. Estim ation of the Propensity Scores 3.6. Description of Variables Chapter – 4 Result and Discussion 4.1. General Characteristics of the Respondents 4.2. Effect on Asset Ownership 4.3. Effect on Incom e 4.4. Effect on Saving 4.5. Effect on Decision Making 18-34 4.6. Estim ation Econom etric Model 4.6.1. Determ inants of Wom en’s Involvem ent in Decision Making 4.6.2. Propensity Scores 4.6.3 Choosing Matching Algorithm 4.6.4 Testing the Balance of Propensity Score and Covariates 4.6.5. Estim ating Average Treatm ent Effect on Treated (ATT) 4.6.6. Sensitivity Analysis Chapter – 5 Conclusion and Recommendation 5.1. Sum m ary 5.2. Recomm endations 35-37 Chapter - 1 I. INTRODUCTION 1.1. Background of the Study Gender discrim ination, m arginalization, unequal treatm ent and unequal access to resources between wom en and m en hinder wom en’s progress. Due to these inequalities wom en are still vulnerable to poverty and they are m arginalized from different econom ic, social and political activities (Batliwal, 1994). Gender equality is a m eans to prom ote growth, reduce poverty and particularly to em power wom en (World Bank, 20 0 1). In subsistence agriculture and low incom e developing countries, m icrofinance provision to rural areas is taken as a m echanism to handle poverty and to em power wom en econom ically (Gebrat, 20 13). The term em powerm ent stands for a broad range of concepts and has different m eaning in different contexts (Baden, 1997; Malhotra et al., 20 0 2). Different authors define the word em powerm ent according to the need of their work in different ways. Em powerm ent is the enhancem ent of assets and capabilities of diverse individuals and groups to engage, influence and hold accountable to the institutions which affect them (Bennett, 20 0 2). Wom en’s em powerm ent is a m eans to prom ote growth, reduce poverty and prom ote better governance (World Bank, 20 0 1). Em powerm ent is not a top-down strategy rather it is a bottom -up process. In m ost cases, the m eaning of em powerm ent has three dim ensions that are econom ic em powerm ent, social em powerm ent and political em powerm ent but som etim es it includes cultural em powerm ent. Since the focus of this study is on econom ic aspect, let us define what econom ic em powerm ent m eans(Mayoux. 20 0 5). Particularly, econom ic em powerm ent is defined as wom en’s access to savings and credit which gives them a greater econom ic role in decision m aking through their decision about savings and credit (Mayoux. 20 0 5). Econom ic em powerm ent looks for guarantee of skills, capabilities, resources and access to secure and sustainable incom es and livelihoods as well as access to assets and resources (Luttrell, 20 0 9). Econom ic em powerm ent of wom en with m icrofinance program m es, focusing and claim ing to em power wom en, have becom e popular am ong donors and NGOs in recent years. The shift in developm ent policies from the em phasis on wom en’s active role in production as a m eans to m ore well-organized developm ent, to the approach of wom en’s em powerm ent through wom en organizing for greater selfsufficiency, has also m eant a change in policies for the im provem ent of wom en’s econom ic role (Baden, 1997). The birth of m icrofinance was in Bangladesh in 1970 ’s. Muham m ad Yunus shocked by the appalling poverty and hum an suffering in Bangladesh and began to think about m icrocredit to im prove the situation (Batem an, 20 10 ). Microfinance is a financial service which is provided to those who are excluded from the form al financial sector. Microfinance can be categorized as form al, sem i-form al, and inform al (Elisabeth, 20 13). Microfinance is becom ing an instrum ent for the em powerm ent of the rural poor households particularly wom en in a way that is self-sustaining (Gebrat, 20 13). 1 © 2017 Global Journals Inc. (US) According to Batem an (20 10 ), the im portance of m icrofinance to the poor is for the following reasons. Microfinance serves as social solidarity in poor com m unities, gives everyone the opportunity to escape from poverty if they want, it saves the poor from traditional m oney lenders, and it is im portant in helping wom en em powerm ent and helps the poor in term s of consum ption sm oothing. Today, in m any developing countries m icrofinance plays crucial role in alleviating poverty. It is a real developm ent instrum ent for the im provem ent of the econom ic life of the poor, particularly wom en . Ethiopia is one of the poorest countries in the world and to address this challenge, the governm ent is im plem enting different developm ental program s like licensing the form al credit sector to reach the rural poor at the grass root level. Providing financial services to the poor particularly for wom en is central to econom ic em powerm ent. The inform al financial system is one of the m ost im portant sources of finance for poor households in rural parts of Ethiopia due to lim ited access to financial institutions. Under the inform al financial schem e the m ain sources of loans are friends, relatives and m oneylenders (Al-Bagdadi and Bruntrup, 20 0 2). The rapid growing m icrofinance industry in Ethiopia is a recent phenom enon (Ebisa et al., 20 0 2). The establishm ent of form al credit sector in Ethiopia dates back to 1995. Particularly, the licensing and supervision of m icrofinance institution proclam ation of the governm ent encouraged the spread of m icrofinance institutions in rural areas (Getaneh, 20 0 5). The m icrofinance industry in Ethiopia has shown a significant qualitative and quantitative growth since its establishm ent. The form al base has been laid by the issuance of proclam ation No. 40 / 96 which established the licensing and supervision of MFI as share com panies in accordance with the com m ercial code of Ethiopia (Al-Bagdadi and Bruntrup, 20 0 2). Microfinance contributes to poverty reduction through increasing incom e, accum ulations of capitals, and diversification of incom e sources for investm ent (Tesfay, 20 0 3). In Ethiopia studies show that, m icrofinance program s benefited the poor in term s of increased incom e, em ploym ent creation, changing the saving habits of households and their expenditure pattern drastically increased on different goods and services (Haym anot, 220 0 7; Balam urugan, 20 12; Gebrat, 20 13). As client of m icrofinance institutions, wom en’s incom e was increased proportionally as m en incom e increased (Gebru and Paul, 20 11). Om o MFI is one of the MFI established in Ethiopia in the South Nations Nationalities and Peoples Region (SNNPR) following the proclam ation No. 40 / 1996 in 1997, which is intended to fill the shortage of form al financial institutions by m eeting the needs of the poor households and sm all scale borrowers in incom e generating schem es. OMFI is operating in all zones and woredas of the region. According to Balam urugan (20 12), OMFI serves m ore than 8 72,0 0 0 loan and saving clients. Agriculture in the study area is characterized by rain-fed and subsistence nature which serve as m ain source livelihood. Though, OMFI is the only financial institution in the study area, wom en clients of OMFI in Gim bo woreda are too few in num ber com pared with the total population of the woreda because of its lim ited outreach in all kebeles. Therefore, this study is intended to identify and analyze the im pact of m icrofinance on wom en’s econom ic em powerm ent in Gim bo Woreda of South Nations Nationalities and Peoples Region. © 2017 Global Journals Inc. (US) 2 1.2. Statement of the Problem In the world as well as in Ethiopia wom en constitute m ore than fifty percent of the total population. But these parts of the population have been discrim inated and m arginalized from different activities specifically from econom ic aspects in the past with no appropriate earnings. This action m akes wom en poor and dependent on their husbands or parents in the fam ily. About 1.3 billion people who live below absolute poverty line and, seventy percent (70 %) of them are wom en (UNDP, 1995 as cited in Balam urugan, 20 12). Em powering wom en are vital, both to recognize the right of wom en and to attain developm ental objectives like econom ic growth, poverty reduction, health, education, welfare and the like (Golla et al., 20 11). Wom en em powerm ent takes three dim ensions. These are econom ic, social, and political em powerm ent but it encom passes beyond this dim ension like cultural em powerm ent (Luttrell, 20 0 9). Wom en’s work in m ost parts of the world, m ainly in developing countries, but not resulted in the sam e level of econom ic em powerm ent as that of m en (CIDA, n.d). Following the collapse of derge regim e, the EPRDF governm ent adopted different policies to address the poor in general and to em power wom en in particular through affirm ative action. Organizations like NGOs and CSOs on the other hand tried to play their role in solving these problem s of discrim ination and m arginalization of wom en. Microfinance is one tool to em power wom en, though it is not given em phasis in the past. However, currently governm ents and various organizations have began to recognize m icrofinance as an im portant intervention in em powering the poor particularly wom en. In developing countries, traditional beliefs and other obstructions such as discrim ination and unequal treatm ent, heavy dom estic workload, high rate of illiteracy am ong wom en, have restricted their roles in the household decision m aking and lim ited wom en’s involvem ent in the econom y and their access to resources. Unless wom en are em powered econom ically, they would be unable to play significant role in econom ic developm ent. Providing credit which is easily accessible to them is one m eans of em powering wom en to run their business (Ablorh, 20 11). The im pact of m icrofinance on wom en em powerm ent is still debatable. Though m icrofinance plays a great role, there is no agreem ent that m icrofinance program s have positive effects on econom ic status of wom en (Aghion and Morduch, 20 0 5). Som e em pirical findings show that m icrofinance has positive im pact on wom en’s econom ic em powerm ent while others argue that m icrofinance has negative im pact on wom en’s em powerm ent. Optim ist advocators of m icrofinance argue that m icrofinance has positive im pact in em powering wom en through an increase in household consum ption expenditure, ability to m ake sm all and large purchase, control over assets, involvem ent in fam ily decision m aking, m obility and freedom from fam ily dom ination are listed as channels through which wom en could be em powered (Hashem i et al., 1996; Schuler et al., 1996; Pitt and Khandker, 1997; Kato and Kratzer, 20 13; Awojobi 20 14). Sim ilarly, studies in Ethiopia depict that m icrofinance has significant im pact on wom en’s em powerm ent (Tesfay, 20 0 3; Haym anot, 20 0 7; Balam urugan, 20 12; Ahm ed, 20 13). On the contrary, other studies on m icrofinance show that m icrofinance has insignificant effect on wom en em powerm ent. They argue that wom en have little or no control over their loan and the loan is controlled by m ale relatives, a num ber of borrowers were to lose their property for repaying the loan. Thus opponents of m icrofinance argue that m icrofinance has negative im pact on wom en em powerm ent (Vengroff and Creevey, 1994; Goetz and Gupta, 1996; ILO, 1998; Kulkam i, 20 11). 3 © 2017 Global Journals Inc. (US) According to Tesfay (20 0 3), in Ethiopia m icrofinance services have lim ited im pact on entrepreneurial developm ent, m icroenterprise sustaining and profitability. A study conducted by Yim er (20 11) on rural m icrofinance and wom en em powerm ent indicates that one third of the respondents included in the study did not perceive m eaning full changes in their life and the im pact of m icrofinance is not sam e and alike to all m atured wom en clients. Thus, there is no uniform ity am ong scholars and researchers on the im pact of m icrofinance on wom en econom ic em powerm ent. Therefore, this study is conducted to fill the existing literature gaps where there are inconclusive findings by including additional variables on previous studies. On the other hand, the researcher couldn’t find any research undertaken on the research question at hand in the study area. So, the finding of this study will help to visualize the im pact of m icrofinance on wom en econom ic em powerm ent in the study area. 1.3. Objective of the Study 1.3.1. General Objective The general objective of this study is to analyze the econom ic im pact of Om o m icrofinance institution in em powering wom en. 1.3.2. Specific objectives The specific objectives of this study are: ❖ To assess the im pact of m icrofinance on wom en’s access to resources and their control over assets. ❖ To investigate the contribution of m icrofinance on wom en’s participation in household decision m aking. ❖ To exam ine the effects of m icrofinance on wom en’s incom e. ❖ To exam ine the im pact of m icrofinance on the saving habits of wom en. 1.4. Significance of the Study Wom en as an essential part of the society and have im m ense potential, their participation and decision m aking on socio-econom ic issues in the past was very low due to different reasons. Econom ically em powering wom en is crucial. In the study area as well as in the western parts of the region, the researcher couldn’t find any research undertaken regarding the im pact of m icrofinance in em powering wom en. So this study m ay serve as a reference for further researchers who want to investigate in this regard. Furtherm ore, this study serves for concerned bodies as an input for further policy issues in the area. Finally, this paper com es with findings in the study area that depict the im pact of m icrofinance on wom en econom ic em powerm ent. 1.5. Scope and Limitation of the Study Regarding geographical scope, the study was conducted in SNNPR, Gim bo Woreda sub-branch OMFI. Whereas the subject m atter of the scope is lim ited to the im pact of Om o m icrofinance on wom en’s econom ic em powerm ent aspect. The study focuses only on the im pact of Om o m icrofinance on rural wom en’s econom ic em powerm ent aspect. Therefore, the study does not assess the im pact on urban wom en’s clients and does not include other dim ensions of wom en em powerm ent like social, political and cultural em powerm ent. On the other hand, few responsive rates of respondents due to different reasons lim ited this study. © 2017 Global Journals Inc. (US) 4 1.6. Organization This study was organized as follows. The first chapter describes about background of the study followed by a review of concepts and literature of m icrofinance and wom en’s em powerm ent. The third chapter describes the data and m ethodology part. The last two chapters that are, fourth and fifth chapters are devoted to the analysis of the data and conclusion and recom m endations respectively. Chapter - 2 II. LITERATURE REVIEW 2.1. Theoretical Literature Review 2.1.1. Definition and Scope of Microfinance According to Asian Developm ent Bank (ADB), m icrofinance is the provision of a broad range of finance such as deposits, loan paym ent services, m oney transfers, and insurance to poor and low-incom e households and their m icroenterprises (Binh, 20 0 0 ; ADB, 20 0 0 ). Microfinance is the provision of financial services to low-incom e clients, including self-em ployed. They give broad base function and term ed as developm ental tool (Ledgerwood, 1999). Microfinance is being recognized as one of the developm ent strategy for the poor (Otero, 1999). “Microfinance refers to the provision of form al services to poor and low-incom e peoples, as well as for others system atically excluded from the financial system . It em braces not only a range of credit products but also savings, m oney transfers and insurance (WB, 20 12). Microfinance is not sim ply banking, rather it is a m ulti-dim ensional developm ent instrum ent in which different activities usually engage such as sm all loans, inform al appraisal of borrowers and investm ents, collateral substitutes, such as group guarantees or com pulsory savings, access to repeat and larger loans based on repaym ent perform ance, stream lined loan disbursem ent and m onitoring, and secure saving products (Ledgerwood, 1999). Cornford differentiates the m eaning between m icrofinance and m icrocredit as m icrofinance is the provision of a broad range financial service to lowincom e m icroenterprises and households. The financial service range includes savings and loans while other products include insurance, leasing, and m oney transfers. On the other hand, m icrocredit em phasizes the provision of credit services to low-incom e clients, usually in the form of sm all loans for m icroenterprise and incom e generating activities (Cornford, 20 0 2). Microfinance covers m any varieties of institutional arrangem ents and approaches. They range from sm all self-helping groups with a handful of m em bers to huge organizations that have nationwide coverage and m illions of clients (McGuire and Conroy, 20 10 ). MFIs can be non-governm ental organizations (NGOs), savings and loan cooperatives, credit unions, governm ent banks, com m ercial banks, or non- bank financial institutions and clients of m icrofinance are characteristically selfem ployed, low-incom e entrepreneurs in both urban and rural areas. Traders, street vendors, sm all farm ers, service providers, and artisans and sm all producers, are those norm al clients of MFIs. These clients are poor but they are not poorest of the poor (Cornford, 20 0 2; McGuire and Conroy, 20 10 ). 5 © 2017 Global Journals Inc. (US) 2.1.2. Conceptualizing Women’s Empowerment Em powering wom en’s are vital, both to recognize the right of wom en and to attain developm ental objectives like econom ic growth, poverty reduction, health, education and welfare and the like (Golla et al., 20 11). The term em powerm ent has m any definitions in different socio-cultural perspectives, and does not interpret easily into all languages (Narayan, 20 0 2). ‘A w om an is econom ically em pow ered w hen she has both the ability to succeed and advance econom ically and the pow er to m ake and act on econom ic decisions’. (Golla et al, p: 4). There is no consensus how em powerm ent is viewed as outcom e or process, how power operates, strategies for inclusion, its im plication, approaches and definition. In m ost literature em powerm ent takes three dim ensions that are econom ic, social and political em powerm ent but it encom pass beyond this dim ension like cultural em powerm ent. But for the sake of this study, it focuses on econom ic em powerm ent aspect only. According to Luttrell (20 0 9) econom ic em powerm ent is defined asan econom ic em powerm ent that looks for guarantee of different skills, capabilities, resources, and access to assets and resources. Regarding types of power relation, it can be classified as power over (ability to influence), power to (organize and change existing hierarchies), power with (increased power from collective action), and power from within (increased individual consciousness). Em powerm ent is associated with the gender and developm ent approach and challenging the way in which the inclusion of wom en in the developm ent process can increase their work burden (Luttrell, 20 0 9). On the other hand, m eaning of em powerm ent refers to generally to the expansion of freedom of choice and action. This m eans one can decide on what she wants to do freely and act on it. The need of em powerm ent is to achieve quality of life and hum an dignity, good governance, pro-poor growth, project efficiency and enhanced service delivery. Em powerm ent, particularly, em powerm ent of poor people (wom en), rem ains in alm ost all developing countries is an ideal rather than a realism in which poor people’s experiences are pervaded by a com m on sense of powerlessness and voice lessens (Narayan, 20 0 2). Various dim ension indicators of em powerm ent includes econom ic, education, governance, health and m edia em powerm ent. Am ong others econom ic em powerm ent indicators includes m arket participation (labor force participation or com position in the m arket) m easured by accuracy of productivity and equity (ownership of land and other assets) (Chung et al., 20 13). According to UNDP (20 11), gender em powerm ent is m easured through political representation indicated by fem ale and m ale shares of parliam entary seats, representation in senior positions in the econom y depicted by fem ale and m ale shares in office and m anagerial position and power over econom ic resources indicated by professional and technical position. Poverty is one of the m ain indicators of wom en's disem powerm ent. As their poverty reduces wom en em powerm ent increases in various decisions m aking. Malhotra et al., (20 0 2) differentiates ordinary/ com m on dim ensions of em powerm ent and its operationalization into three areas. At household level econom ic em powerm ent is m easured through wom en’s control over resources and their role to fam ily support and at com m unity level asset and land ownership, access to credit, access to m arkets and representation in local trade associations are indicators/ m easurem ents of wom en’s econom ic em powerm ent. In the broader sense wom en econom ic em powerm ent is m easured through their representation in high paying works, wom en CEO’s and representation of their econom ic interest in m acroeconom ic policies and strategies of local and federal budgets (Malhotra et al., 20 0 2). © 2017 Global Journals Inc. (US) 6 2.1.3. Microfinance and Women Economic Empowerment In the past wom en having huge talent and potential have been discrim inated and m arginalized from different activities, particularly from econom ic aspects which is responsible for social exclusion. Em powering wom en has m any roles and seeks to m eet strategic gender needs through participation on resources and developm ent issues that concern the life of wom en. Most wom en particularly, those who are poor, are m ore vulnerable because of uneven distribution of resources and are unable to m eet the basic requirem ents that worsen the unm et gender needs. Thus, lack of entitlem ent arising from inadequate assets and capabilities, m akes wom en so poor and unable for the fulfillm ent of basic livelihood needs. The consequence is social and econom ic exclusion of a certain class and category of people and their consequent is disem powerm ent. That is why the idea of em powerm ent has influenced developm ent practitioners, developm ent agencies (both governm ents and NGOs), theoreticians and governm ents and donor agencies in the last decade (Padm a and Getachew, 20 0 4). Microfinance and wom en em powerm ent interface at both intrinsic and extrinsic levels in which, the extrinsic level of em powerm ent refers to gaining greater access to and control over financial and physical assets, while the intrinsic level involves changes within, such as the rise in self-reliance, confidence, m otivation and positive hope for the future (Yim er, 20 11). Microfinance enables to m obilize and organize the poor and wom en at grassroots level and offer hitherto denied access for critical assets to them (Am in et al., 1998). To achieve gender justice in the m icrofinance sector there is still a long way to go. In spite of increased access to sm all loans and savings, wom en’s access to m ore advanced products is still unequal in m any countries particularly in the LDCs. Evidences on m em bership of clients depict very little about the quality of the services accessed by wom en com pared with m en. Generally, wom en’s loan is lower than their spouse or m en, which constrains them to run good business that requires large loan. Most of their loan does not buy assets like land, house, m achinery and equipm ent. The fact is that, wom en are the m ajority of savers, but m en receive the m ajority of loans (Mayoux, 20 10 ). Availability of credit m ay also reduce willingness of custom ers and relatives to give interest-free loans and/ or access to m ore charitable form s of credit from traders. Moreover, a recent study has found that wom en’s access decreases com pared to that of m en as NGOs change to form al institutions, becom e m ore profitable and m ature (Mayoux, 20 10 ). The logic of m icrofinance’s potential for em powerm ent is sim ilar to the econom ic m odel of em powerm ent; m icrofinance m akes wom en econom ically independent by putting capital and financial resources in their hands. Econom ic independence results in higher bargaining power for wom en in their households and com m unities, and subsequently results in higher prestige and self-esteem . Here the functions are synchronous with its potential to em power (Kulkam i, 20 11). In som e cases, m icrofinance m ay even disem power wom en. The extent, to which wom en are able to benefit from sim ple financial services which do not take gender clearly into consideration, depends largely on context and individuals condition. Prim arily, there is the subject of financial indicators of access: wom en’s program m em bership, num bers and size of loans and repaym ent data cannot be used as indicators of real access or proxy indicators of their em powerm ent. Registration for loans in wom en’s nam es does not necessarily indicate even participation in decisions about loan application, as m en m ay sim ply negotiate loans with m ale program experts as an easier m eans of getting access to credit. 7 © 2017 Global Journals Inc. (US) Secondly, the contribution of financial services to increasing incom es varies widely. Experiences from South Asia and Latin Am erica show that, m ost wom en use their loans for their husband’s activities, either as a rational investm ent choice where their own econom ic activities are lim ited or because their husbands claim the m oney as their right due to gender inequality. Finally, wom en’s better contribution to household incom e does not ensure that wom en necessarily benefit or that there is any challenge to gender inequalities within the household though wom en seek to increase their power within joint decision-m aking process rather than seek independent control over incom e, neither of these outcom es can be assum ed to occur (Mayoux, 20 10 ). 2.1.4. Overview of Microfinance Sector in Ethiopia Most of the population in Ethiopia resides in rural area under widespread poverty. A program like m icrocredit m akes the governm ent to reach these m arginalized populations at the grassroots level. The history of m icrofinance in Ethiopia goes to the m id of 1990 ’s. As a recent phenom enon, m icrofinance service in Ethiopia was introduced for the first tim e in 1994 as an experim ent when the relief society of Tigray (REST) attem pted to rehabilitate drought and war affected people through rural credit schem e in Tigray region (Yohannes, 20 0 6). Regarding ownership, financial foreign investm ent in Ethiopia is prohibited by law. Particularly, proclam ation No. 84/ 1994 precludes a foreign national from undertaking banking business in Ethiopia, and no person is perm itted to own m ore than twenty percent (20 %) of a banking com pany’s shares. On the basis of this proclam ation, MFIs in Ethiopia should be established as share com panies, the capital thereof owned fully by Ethiopian nationals and/ or organizations wholly owned and registered under the laws with a head office in Ethiopia (Getaneh, 20 0 5). The m icrofinance sector in Ethiopia is characterized by its rapid escalation, wide geographic coverage and increasing num bers of clients, large shares of governm ent, focus on rural households, prom oting credit and saving products (Ebisa et al., et al., 20 12). Recently m icrofinance institutions in Ethiopia are increasing rapidly to provide credit for the poor with various technical assistances. The num ber of m icrofinance in Ethiopia reached 33 with total capital and asset 3.8 billion and 13.3 billion birr respectively. Of the existing m icrofinance institution seventy five percent of the total capital in the industry is occupied by Am hara credit and saving institution (ACSI), Dedebit credit and saving institution (DCSI), Orom iya and Om o credit and saving institutions. In Ethiopia, out of the existing m icrofinance institutions fifty percent of the MFI are operating in Addis Ababa (Gashaw, 20 14). 2.1.5. Overview of Omo Microfinance Institution (OMFI) OMO m icrofinance institution (OMFI) was established in the South Nations Nationalities and Peoples Region (SNNPR), which is intended to fill the shortage of form al institutions by m eeting the needs of the poor households and sm all scale borrowers in incom e generation schem es. The prim e m ission of OMFI is im proving the econom ic status of low incom e, productive poor people particularly poor wom en in the region through increased access to credit and saving services. OMFI exerts its m axim um effort to bring about accelerated and sustainable econom ic developm ent in the region by the provision of efficient, effective and sustainable financial services to econom ically active poor people through effective partnership with GO and NGOs. © 2017 Global Journals Inc. (US) 8 OMFI is established in 1997 and at the sam e tim e registered as a share com pany as per the requirem ent of proclam ation num ber 40 / 1996 which states the provision of licensing and supervision of the business of m icrofinance institutions. It has been established with five shareholders, these are, SNNP regional governm ent, southern Ethiopia peoples developm ent association (SEDA), Wondo Tradin g Com pany and two natural persons 1. OMFI operates for the achievem ent of governm ent policies and strategies. Currently it is operating in 11 branches and 69 sub-branches which constitute exactly m ore than half of the region’s geographical outreach (Dilayehu, 20 14). The m ain services provided by MOFI to its clients are credit, savings, pension fund adm inistration, and m icro lease. The target groups of clients includes from the agricultural sector, m icro and sm all scale enterprises or business, petty traders, handcraft, and services sectors. 2.2. Empirical Literature Various studies within the country or across countries m ay find different results on the im pact of m icrofinance in em powering wom en. Som e studies argue that m icrofinance has significant role in em powering wom en while other argues that m icrofinance has no role in em powering wom en. Even if m icrofinance plays a great role, there is no agreem ent that m icrofinance program s have positive effects on econom ic status of wom en (Aghion and Morduch, 20 0 5). Microfinance credit provision by Gram een Bank and Bangladesh Rural Advancem ent Com m ittee (BRAC) argues that m icrofinance has significant effect in em powering wom en through increased m obility, econom ic security, involvem ent in m ajor decision m aking, ability to m ake large purchase, freedom from fam ily dom ination, political and legal awareness, participation in public protests and political cam paigning (Hashem i, et al., 1996). Im pact participation by gender three group based credit program s, Gram een Bank, Bangladesh Rural Advancem ent Com m ittee (BRAC) and Bangladesh Rural Developm ent Board’s (BRDB) on wom en and m en in general find that significant effect on the wellbeing of the poor household but the effect is greater if wom en are the program participant (wom en annual household expenditure increased m ore than m en) (Pitt and Khandker, 1997). A study by Roxin et al (20 10 ), on im pact of m icrofinance in Sierra Leone revealed that MF had im proved clients’ business expansion, increased their incom e and expenditure. Their study depict that m icrocredit has considerable im pact on econom ic em powerm ent but it has only initial im pact on social em powerm ent. At the sam e tim e their finding reveals no im pact on wom en political em powerm ent. Different scholars argue that, m icrofinance has no role in em powering wom en since wom en have little or no control over their loan and the loan is controlled by m ale relatives, a num ber of borrowers were to lose their property for repaying the loan (Vengroff and Creevey, 1994; Goetz and Gupta, 1996). The study in Tigray region on the im pact of m icrofinance on poor wom en shows that directly or indirectly, m icrofinance services provided by Dedebit credit and saving institution (DECSI) are contributing to the sustenance and im provem ent of the life of the poor wom en and their households. This study was conducted using m ulti-stage sam pling with descriptive m ethod of analysis. The evidence from this study depict that m icrofinance has positive im pact in increasing incom e, diversifying sources and reducing variability of incom e. It also show that increased consum ption, 1 Natural persons are those individual persons who are the shareholders of OMFI 9 © 2017 Global Journals Inc. (US) im proved living condition in term s of house repairs and expansions, m edical services and capital accum ulation in the form of increased saving. This study critically depicts that wom en em powerm ent in term s of im proved attitude and respect of their husband, increased self-confidence and self-im age (Tesfay, 20 0 3). A research conducted at ACSI indicates that m icrofinance participant wom en are m uch better than non-participants in term s of household asset holding, yearly average off farm incom e, and involvem en t in decision m aking process in the household. The study was conducted using sim ple random sam pling m ethod with log it econom etric m ethod of analysis. Sim ilarly, the estim ation result of the logit m odel indicates out of 23 explanatory variables used 15 of them are significant (Gebrat, 20 13). The study by Haym anot (20 0 7), using descriptive statistics and binom ial logit regression m ethod of analysis reveals that m icrofinance has a positive im pact on wom en econom ic em powerm ent in term s of increased participation of wom en in the household decision m aking, and im proved living standard condition of its clients. Matured clients of ACSI have im proved their household incom es, asset possession level, and saving habit; thereby positively affecting their ability to fully participate in household decision m aking. A study by Yim er (20 11), at ACSI using explanatory research m ethod, show that in m ajority cases m atured wom en clients have gone substantial change in m any dim ensions, to m ention som e indicators, like change in term s of skills essential for m aking and m anaging businesses, level of confidence and self-esteem and worth, personal cash assets, level of financial independence, incom e and diversifying incom e sources. But one third of the respondents included in the study did not perceive m eaning full changes in their life. Therefore, the im pact of m icrofinance is not sam e and alike to all m atured wom en clients. A study in Harari m icrofinance institution (HMFI) using descriptive statistics and logistic regression depict that, HMFI participants have been better in their incom e, im proved their saving habits and control over resources than non-participants. The study used 123 sam ple respondents of which 15 of them are non-participants used as a control group. At the sam e tim e, the study used PSM m ethod of analysis to identify the im pact of HMFI on its clients. But the validity of the control group is questionable to apply PSM m ethod of im pact evaluation because of low num ber of non-participants which m ay result biased estim ates of ATT. Moreover the study doesn’t carry out sensitivity analysis to see the effect of unobservable covariates between participants and non-participants (Ahem ed, 20 13). The study in SNNPR of OMFI by Balam urugan at Wondogent indicates that m icrofinance has significant effect on wom en em powerm ent. The descriptive statistics and regression analysis of the study was conducted using before and after m ethod of analysis. OMFI contributes to social and econom ic em powerm ent of wom en in the study area. Wom en’s hope and self-confidence im proved through active participation in OMFI. Wondogent OMFI gives to wom en significant changes in term s of em ploym ent creation and incom e generation, saving habits and decision m aking. Therefore the study concludes that Wondogent OMFI affects wom en in term s of social and econom ic em powerm ent (Balam urugan, 20 12). Most of the studies conducted on wom en’s em powerm ent particularly econom ic em powerm ent were carried out using either descriptive statistics or logistic regression. Few studies tried to use PSM m ethod of im pact evaluation particularly in Ethiopia is rare and even those conducted on sim ilar topic fails to carry out full PSM procedures. Therefore, in the existing literature there is no uniform ity am ong scholars and researchers on the im pact of m icrofinance on wom en econom ic em powerm ent and their findings are yet inconclusive. © 2017 Global Journals Inc. (US) 10 Chapter - 3 III. RESEARCH METHODOLOGY 3.1. Brief Description of the Study Area Kaffa Zone is one of the 14 adm inistrative zones found in the SNNPR state located in the south western part of Ethiopia at a distance of 449 Km from Addis Ababa and 729 Km from the capital of SNNPR state, Hawassa. It is bordered to the north and east by Orom ia region, to west and south west by Benchi Maji Zone, to North West by Shaka Zone and to east by Konta special woreda. The Zone is located between 6 0 24 1 and 8 0 13 1 north latitude and 35 0 48 1 and 36 0 78 1 east longitudes. The Zone constitutes ten rural woredas (Gim bo, Gewata, Gesha, Sayilem , Bita, Chena, Decha, Telo, Cheta, and Addiyo) and one town adm inistration, Bonga, which is the capital of Kaffa Zone. Gim bo woreda is located between 70 23 1 and 70 49 1 north latitude and 36 0 0 0 1 and 36 0 471 east longitudes. It is bordered to the north by Orom ia region, to west Cenna and Gewata woreda, to south by Decha woreda and to east by Addiyo woreda. The woreda contains 33 rural kebeles and three urban kebeles 2 . The total land area of Gim bo woreda is 832.5 square kilom eter. It has three agro-ecological zones i.e. high lands (0 .3%), m id-lands (74.4%) and low lands (15.3%). According to Kaffa Zone finance and econom ic developm ent office population projection in 20 14 (based on the population and housing census of 20 0 7), the total population of the Gim bo woreda is 114113 of which 5730 9 are fem ales. Of the total population,9740 1 (85.4%) of the population resides in rural areas depending on subsistence agriculture and 16712 (14.6%) are urban dwellers. Regarding ethnic com position, Kafficho which is indigenous native and other ethnics Am hara, Orom o, Tigray, Guragie lives inside the woreda. The m ajority of the inhabitants are followers of the Orthodox, Protestant and Muslim religion respectively. Agriculture is the m ajor econom ic activity practiced in the area followed by trade. More than 85% of the population of the woreda is engaged in m ixed agricultural farm ing (crop and livestock) and the rem aining population rem ains in com m ercial/ trading activity, governm ent sector em ploym ent, and wage laborer. OMFI is the only institution in the woreda that provides financial services to the poor households particularly to wom en (KZFaED statistical abstract bulletin, 20 14). 2 Kebeles are sm all sub-divided local adm inistrative units in the com m unity 11 © 2017 Global Journals Inc. (US) Fig. 3.1: Map of the study area Source: KZFaED departm ent, 20 14 3.2. Data Type and Source Prim ary data was collected by m eans of a structured questionnaire responded by OMFI m atured clients (being clients 3 to 5 years), incom ing clients (clients for 1 to 2 years), and non-clients (loan applicants but yet not given) in Gim bo woreda. At the sam e tim e, sem i-structured interviews were held with clients to get additional inform ation of respondents’ opinions, perceptions and attitudes to verify inform ation given by clients. An interview of different officials and experts was conducted at different levels. This study was also used secondary data obtained from various sources like reports, m anuals, abstracts etc. Mainly quantitative data was used. 3.3. Study Population The study population consists of all poor wom en in the study area. These include all wom en clients of OMFIs and wom en non-clients (loan applicants as control groups). 3.4. Sampling Design and Technique The survey was used cross-sectional data. Under this study, m ulti-stage and purposive sam pling m ethods were used. At the first stage Gim bo woreda is selected am ong ten woredas and one town adm inistration in Kaffa Zone because it is one of the two sub-branches of OMFI opened their office early (Bonga and Gim bo, beginners) and com pared with Bonga town, Gim bo woreda encom passes © 2017 Global Journals Inc. (US) 12 large clients. Gim bo woreda has a total of thirty five Kebeles. For the purpose of facilitating its service delivery, Gim bo woreda sub-branch OMFI has five clusters or ketenes (Gim bo, Wush Wush, Kuti, Kayikelo and Gojeb). In the second stage, three clusters were selected purposively according to the distance from the woreda town, one from the rem ote, one from the m iddle and one from the nearest. In the third stage, two kebeles were selected from each cluster random ly. Finally, from these six kebeles, respondents were selected using sim ple random sam pling m ethod from the list file of the clients in the institution. Sim ilarly, non-clients (control groups) are those applicants to take loan from Gim bo OMFI sub-branch office in the near future after they fulfill the institution selection criteria. To assess the im pact of m icrofinance, it is necessary to com pare the outcom es of clients with control groups that have sim ilar characteristics. The control groups are future clients that are very sim ilar to clients in their overall characteristics. The justification for the use of purposive sam pling is intended to include wom en clients only. 3.4.1. Sample Size Determination To determ ine the sam ple size, the researcher tried to consider inform ation from prior studies in the sam e topic, the available budget at hand for the study and tim e fram e to accom plish this study within the calendar were considered. Prior studies like Haym anot (20 0 7), Balam urgan (20 12) and Ahm ed (20 13) used their sam ple size 171, 120 , and 123 respectively. In addition by taking into account m y budget, tim e and its feasibility, for this study data was collected from 20 0 wom en. Of the total sam ples 115 of them are non-clients which are used as a control group for the study. Regarding the distribution, 15 clients were selected from each 5 kebeles and 20 clients from 1 kebele since it has large num ber of client com pared with the rest 5 kebeles. Non-clients were selected by 1.35 ratio scale to 1 client (1.35:1). 3.5. Method of Analysis The em pirical analysis of this research was em ployed both descriptive statistics and regression analysis. The descriptive statistics was used m easure of dispersions (m ean, SD, variance), percentages, tables and m aps. The regression analysis was em ployed logit to estim ate propensity score m atching using STATA software. 3.5.1. Propensity Score Matching (PSM) Propensity score m atching m ethod of im pact analysis is a m ethod of com paring m icrofinance clients and non-clients in the program m e areas, where both groups experiences sim ilar com m unication facilities, socio-econom ic characteristics, topography, developm ent infrastructure program s and others, to exam ine whether there is econom ic variation between program participants and nonparticipants. The assum ption behind this study is that at m ost m icrofinance benefits the poorest of the poor at the grass root level. The justification for choice of PSM m ethod over other im pact analysis m ethod like DID m ethod was that, PSM m ethod uses only cross-sectional data collected at point of tim e while DID m ethod needs baseline data. Another justification for the use of PSM m ethod was that, self selection bias can be best controlled or m inim ized by using PSM and PSM reduces dim ensionality. In non-experim ental data, PSM com pares treatm ent effects across participant and m atched non-participant units. PSM assum es selection bias is based on only observed characteristics (not account for unobserved factors). Because of the above reasons, PSM m ethod is chosen for this study. 13 © 2017 Global Journals Inc. (US) There are a range of assum ptions to hold PSM m ethod of analysis. To hold PSM, first participants and non-participants have sim ilar distribution of observed characteristics and have sim ilar distribution of unobserved characteristics (if not it causes problem s of selection bias), the sam e set of questionnaire is distributed for both groups with the sam e econom ic environm ent and assum ption of unit hom ogeneity (there is no unobserved heterogeneity).Finally the assum ption of conditional independence (there is no reverse causality) m ust be hold. The PSM is defined as the conditional probability of receiving treatm ent (participant) given pretreatm ent characteristics (Rosenbaum and Rubin, 1983). () = ( = 1 /) ={/} (1) Where D= (1, 0 ) is the binary variable indicating whether a wom an has em powered (=1) or not (=0 ) and X is a m ultidim ensional vector of pre-treatm ent characteristics (observable characteristics) and p(X) is the propensity score. Let Wi1 and W i0 represents the outcom e when wom en are participant in m icrofinance and the outcom e when not participate respectively. So, the difference between the treated and control group is given as, (2) Where, W i1is the outcom e if treated and W i0 is the outcom e of untreated. Let equation (2) is expressed as Bio to express the causal effect, the treatm ent variable takes 1 if the individual I receives treatm ent and 0 otherwise. Then, ATT of an individual I can be expressed as: ATT=E(W i1/ B=1-E(W i0 / B=1)) (3) The E (Wi0 / B=1) from equation (3)is unobservable outcom e known as counterfactual. In other words E(Wi0 / B=1) is the average outcom e of treated individuals had they not received the treatm ent). E[W i1/ B=1]-E[W io/ B=0 ]= ATT+E[W i0 / B=1]-E[Wi0 / B=0 ] (4) Selection bias is shown by the difference between left hand side of equation (4) and ATT. Since the m ain param eter interest is ATT, it can be defined as: ATT=E[W i0 / B=1]E[W i0 / B=0 ]=0 (5) In estim ating propensity scores, all variables that affect participation in m icrofinance are included. Therefore, the average treatm ent effect on those treated conditional on propensity score p(x) is given as: ATT=Ep(x)/ B=1{E [W i1/ B=1, p(x)]-E [W i0 / B=0 , p(x)] (6) ATT is the difference between expected outcom e values with and without treatm ent for those who actually participate in treatm ent. In equation (6), the PSM estim ator is the m ean difference in outcom es over the com m on support region, appropriately weighted by propensity score distribution of participants (Caliendo and Kopeinig, 20 0 5). ATT is average treatm ent effect on treated (i.e the effect of treatm ent) if the wom an participate in m icrofinance (B=1) and otherwise (B=0 ). © 2017 Global Journals Inc. (US) 14 According to Becker and Ichino (20 0 2), the assum ption of com m on support region falls between 0 and 1 (i.e. 0 p(x) 1). This im plies that the test of balancing propensity is perform ed only on the observations whose propensity score belongs to the com m on support region of the propensity of treated and control groups. Those individuals that lay outside the com m on support region would be excluded in treatm ent estim ation and this im proves the quality of m atching to estim ate ATT. In order to estim ate the m issing counterfactual outcom e for each treated observation different m atching estim ators are used: Nearest Neighbor m atching, Kernel m atching, caliper m atching and Radius Matching. The selection of m atching algorithm is tested using lower value in pseudo R 2 , balancing test (num ber of insignificant explanatory variables after m atching) and better num ber of m atched observation. By com paring the result of all m atching estim ators, kernel with bandwidth 0 .5 is selected for this study with different criteria. In this regard, kernel m atching algorithm m atches several nonparticipants with a participant. 3.5.2. Estimation of the Propensity Scores The probability of wom en clients to be em powered (wom en’s involvem ent in m ajor decision m aking), P i is given as; P i=E(Y=1/ Xi)= 1 (7) 1+−(1+2 ) The logistic representation of wom en’s involvem ent in m ajor decision m aking is; P i= 1 = 1+− (8) The probability of wom en’s does not involve in m ajor decision m aking is given as; 1-Pi= 1− 1− = 1+ 1+− = 1 1+ ……… (9) (10 ) = the odds ratio in favor of wom en’s involvem ent in m ajor decision m aking, i.e. ratio of the probabilities that wom en participate in m ajor decision to the probabilities that not participate in decision m aking.Taking the natural logarithm ; Li=ln( ) =Zi = β0 +β1β.1+β2 X2 +β3 X3 +…. +βn Xn-1− (11) By taking the error term into consideration, the log odds ratio m odel becom es Zi= β0 +β1X1+β2 Xi 2 +β3 X3 +...+βn Xn +u (12) Where ✓ ✓ ✓ ✓ ✓ P i is the probability of participating in a program m e Zi is a function of explanatory variables (Xi) Xi is the explanatory variables βo is an intercept β1,β2 …βn ’s are slopes of the equation in the m odel 15 © 2017 Global Journals Inc. (US) ✓ Li is log of the odds ratio which is linear in Xi’s and B’s ✓ U i is the disturbance/ error term Here Z, takes two possible values i.e. z=1 wom en are econom ically em powered m eans participates in m ajor decision and z=0 if not. 3.6. Description of Variables For the purpose of this study different variables were selected based on econom ic theory and previous em pirical findings from the existing literatures on sim ilar studies. In im pact evaluation study, variables choice m ust be those variables which affect both participants and non-participants (i.e. both treated and non-treated groups share characteristics of X covariates). Heckm an et al., (1998), argues that, only variables that affects both program participation and outcom es should be included in the estim ation propensity scores. Thus, in this study variable which affect clients OMFI and non-clients are selected depending on observable characteristics of respondents in the study area. Outcom e variables (im pact indicators): In this study three outcom e variables nam ely average yearly household incom e (ayhi), personal cash saving (pcs), and asset ownership and possession are used as an indicator of the im pact of OMFI on wom en econom ic em powerm ent. Average yearly household incom e is a continuous variable which determ ines wom en's status in signifying their em powerm ent. It has positive expected sign. Existence of personal cash savings is also expected positively related with wom en em powerm ent. Household assets ownership (ownast) like farm land, house, jewelry and livestock’s etc are assum ed that positively related with wom en em powerm ent. The dependent variable is wom en’s involvem ent in m ajor decision m aking in the fam ily which includes decisions on sales/ purchase/ rent/ repair of house, farm land, livestock’s like ox and cows, grains and the like. As a result, wom en’s involvem ent in m ajor decision m aking, as a proxy variable is used to indicate the econom ic em powerm ent of wom en. This variable was used in Hashem et al., (1996) as independent variable to explain wom en’s em powerm ent. Independent variables: Age, respondents (wom en) level of education, wom en’s spouse level of education, m arital status, head of the household, being m em ber of other MFI, num ber of household size, ecology and initial wealth. Age of respondent (age): It is a continuous variable. Older wom en have no independence and em powerm ent than younger wom en as they are housewife and not participate on decision m aking. So, it is hypothesized that age is positively related. Marital status is categorical variable it m ay relate positively or negatively. Respondents (wom en) level of education (wle): It is a continuous categorical variable. If wom en have m ore education their confidence will increase. Thus, it is expected that education has positively related. Spouse level of education (sle): it is a continuous categorical variable. It is expected that education is positively or negatively correlated. Head of the household (hhd): dum m y variable = 1 if head of household is wom en otherwise 0 . If wom en are household head it is expected positive if not negative. Num ber of household Size: it is a continuous variable. It m ay relate positively or negatively. © 2017 Global Journals Inc. (US) 16 Being m em ber of other MFI (bm ofi): it is dum m y variable (if wom en are clients of other MFI =1 otherwise=0 ). It m ay relate positively or negatively Ecology (eco): dum m y if the place has good ecology to engage in productive work (=1) and =0 if not. It is expected positive or negative relationship. If the environm ent has good ecology, wom en’s have encouraged to work hard and they need credit for the purchase of agricultural input and if not they discourage. Initial w ealth: dum m y if a wom an has initial wealth (=1) and = 0 if a wom an has not initial wealth. The expected sign is positive, if there is initial wealth otherwise negative. Table 3.1: Sum m ary variables description Its expected sign + + + + +/ + +/ + +/ +/ + +/ - Definition Variable nam e Ou tco m e variable Average yearly income Ayhi Personal cash saving Pcs Ownership of asset Ownast D e p e n d e n t variable Wom en’s involvement in m ajor decision m aking In d e p e n d e n t/ e xp lan ato ry variable s Age of respondents (women’s) Age Marital status Mrsta Wom en’s level of education Wle Spouse level of education Sle Head of the household (dum m y if wom an is head=1) Hhd Num ber of household size Being m ember of other MFI (dum m y if wom en is client of other MFI=1) initial wealth Ecology Nhsize Bm ofi Am inw Eco Source: Ow n com putation, 20 16 17 © 2017 Global Journals Inc. (US) Chapter - 4 VI. RESULT AND DISCUSSION This section presents both the descriptive and econom etric result and findings of the study. The study exam ined the im pact of OMFI on wom en’s econom ic em powerm ent based on prim ary data collected from wom en clients and non-clients in the study area. The questionnaire was designed in line with the pre-determ ined objectives of the study and distributed to the sam pled respondents. The inform ation given in the questionnaire was checked with sem i-structured interview from random ly selected sam pled respondents. 4.1. General Characteristics of the Respondents From the total sam pled respondents, the data was collected from 196 respondents. Of the total respondents 84 of them are clients of OMFI while rem ain 112 of them are non-clients. Non-clients are those respondents that cam e to the organization for loan after they fulfill the requirem ents but not yet given loan. Regarding the response rate of the questionnaire, 98.8%of client respondent returned the questionnaires while 97.3% of the control groups were returned the questionnaire. Of the total respondents 2% of wom en were not willing to give inform ation because som e of them were on work and som e others were not available at the tim e. The religion statistics of the respondents show that 73.98%,16.84%, and 9.18% of them are followers of Orthodox, Protestant and Muslim religions respectively. The total sam ple result of age distribution depict that 27.55% respondents were between the age 20 to 30 while 48.47% of the respondents are between the age 31 to 40 and 14.29% of respondents are between the age of 41 to 50 . From the total sam ples 3.57% of respondents are above the age of 50 years and 6.12% of the respondents don’t know their age. The m inim um and m axim um age of the respondents is 20 and 60 respectively and their m ean age is 37. The age distributions of the respondents indicate that m ost of the respondents are in the working or productive age group. Table 4.1: Age distribution of the respondents 3 Max Min Std. Dev. Mean Obs 60 20 7.579832 37.0 1523 196 Variable Age Source: ow n com putation, 20 16 To exam ine separately, the m inim um and m axim um age of client respondents’ is 20 and 57 where as non-client respondents’ age is 20 and 60 . Regarding m arital status, the sam ple result show that 70 .41% of the respondents were m arried whereas out of the total respondents 15.31%, 8.67% and 5.61% of them are widows, single and divorced respectively. The descriptive statistics depict 53.57% of the respondents are headed by their husbands while 41.84 % of the respondents are head of them selves. Out of the total respondents only 4.59 %of respondents responded that they are headed by their fam ily. The m ean household head difference 3 Obs indicates num ber of observations © 2017 Global Journals Inc. (US) 18 between client and non-clients is -0 .1686684 with p-value 0 .0 40 1 which is significant at 5% level of significance. The result indicates that m uch of the clients are heads of their fam ily (see Table3 in the appendix). Table 4.2: Distribution of m arital status and head of the fam ily Head of the household Total Others Husband Myself 17 (8.67) 138 (70 .41%) 30 (15.31%) 11(5.61%) 1 6 2 0 4 91 6 4 12 41 22 7 196 (10 0 %) 2(4.59%) 140 (53.57%) 54(41.84%) Marital status (m rsta) Single Married Widowed Divorced Total Source: Ow n com putation, 20 16 Education status of respondents show that 46.94 % of them are illiterate, 27.0 4 % of them can read and write, 15.31 % of them are learned from grade 1 to 4, 6.63 % of them are learned from 4 to 8, 3.57 % them are learned from grade 8 to 12 and 0 .51 % of them are above grade 12. The m ean education level difference between client and non-non-clients is -0 .78 2237 and the p-value 0 .0 0 0 (highly significant at 1% of significance level) which leads to reject the null hypothesis that is there is no difference between the groups in wom en’s education level (i.e. clients are better educated than non-clients) (see table 1 in the appendix). Regarding their (wom en) spouse education level the sam ple result depict that 14.29% of their husband are illiterate, 29.0 8 % of them can read and write, 28.57 % of them are learned from grade 1 to 4, 21.94% of them are learned from 4 to 8 and 6.12% of them are learned from 8 to 12. The m ean spouse education level difference between client and non-clients’ husband is -0 .264881 and the p-value 0 .10 47 (which is insignificant) that leads to accept the null hypothesis that there is no difference between the groups in spouses’ education level (see Table 2 in the appendix). Table 4.3: Distribution of respondents and their spouse level of education Spouse level of education Percent 14.29 29.0 8 28.57 21.94 6.12 - Frequency 28 57 56 43 12 - 10 0 .0 0 144 Respondents level of education Illiterate Read and write Grade 1-4 Grade 4-8 Grade 8-12 Above grade 12 Percent 46.94 27.0 4 15.31 6.63 3.57 0 .51 Frequency 92 53 30 13 7 1 Illiterate Read and write Grade 1-4 Grade 4-8 Grade 8-12 Above grade 12 Total 10 0 .0 0 196 Total Source: ow n com putation, 20 16 19 © 2017 Global Journals Inc. (US) The respondents’ fam ily size ranges from 1-10 . The m inim um and m axim um fam ily sizes of clients are 1 and 10 while for non-clients it is 1 and 9. On average, both client and non-client respondents have 5 fam ily sizes. The m ean household fam ily size difference between client and non-clients is -0 .6325327 and the p-value 0 .0 262 (which is significant) that leads to reject the null hypothesis that there is no difference between the groups in fam ily size (i.e. clients have less fam ily size than non-clients) (see Table 4 in the appendix). Table 4.4: Distribution of fam ily size of respondents separately Non-clients Clients Percent 12.50 8.93 Frequency 14 10 No. fam Size 1 2 Percent 4.76 7.14 Frequency 4 6 No. fam Size 1 2 16.96 19 3 14.29 12 3 14.29 24.11 16 27 4 5 16.67 23.81 14 20 4 5 9.82 11 6 8.33 7 6 9.82 1.79 11 2 7 8 1.79 2 9 21.43 1.19 2.38 18 1 2 7 8 10 10 0 .0 0 112 Total 10 0 .0 0 84 Total Source: Ow n com putation, 20 16 Concerning the m ain occupation, the sam ple result depicts that m ore than three-fourth of the respondents (79.7%) were engaged in the agriculture sector. Agriculture, being the m ain occupation in the study area, it is followed by trade activities which accounts about 14.72% of occupation activities in the area. The rem ain 4.57 % of respondents reply that they are engaged in governm ent sectors like agricultural and health extension workers, teachers and 1.0 2% were engaged in different works. Separately, 77.11% of client respondents’ m ain occupation is agriculture while for non-client respondent it accounts 81.42%. Trade accounts 19.28% and 11.5% for both client and non-clients respectively. Three clients responded that their m ain occupation is governm ent em ployers. Six nonclients occupation is governm ent em ployee while two of them are engaged in different activities. Table 4.5: Distribution of respondents’ occupation Non-clients Clients Percent Freq. Percent Freq. Percent Freq. 81.42 11.50 5.31 1.77 91 13 6 2 77.11 19.28 3.61 65 16 3 79.70 14.72 4.57 1.0 2 157 29 9 2 10 0 .0 0 112 10 0 .0 0 84 10 0 .0 0 196 Source: Ow n com putation, 20 16 © 2017 Global Journals Inc. (US) 20 J ob Agriculture Trade Gov’t em ployer Other Total Total Table 4.6: Sum m ary of descriptive statistics of selected variables Std. Dev. Mean Obs Variable .6433454 11.4740 1 .6592537 1.13766 1.3520 41 33.88265 2.178571 .9438776 196 196 196 196 Religion Age Marital status Wom en’s level of education 1.130 473 1.76530 6 144 Spouse level of education .60 23458 1.9850 5 1.25 4.443878 196 196 J ob/ occupation Num ber of fam ily size .5720 464 1.627551 196 Head of the fam ily Source: Ow n com putation, 20 16 Am ong the total sam ple, one observation reported that she has been client of OMFI since its establishm ent which is the m axim um period for 15 years. Of the total sam ples 30 % of the respondents were clients since 20 0 0 E.c, while m ore than 72% of the respondents were been clients since 20 0 4 E.c. The m ajority of clients’ loan size lie between 30 0 0 - 50 0 0 . The m axim um and m inim um loan size is 90 0 0 and 150 0 during the last five years from recent to back loan cycles. The m axim um loan size is determ ined by OMFI while the m inim um size is determ ined by client request or dem and. The descriptive statistics depict that, when the loan cycles increases, the average am ount of loan is also increase. Of the sam ple taken, 34% of the clients reply that, they take the loan for the purpose of buying ox for their farm activities. On the other hand, 22.78%, 20 .25% and 16.46% of clients take their loan for the purpose of buying agricultural input, for fattening (sheep and goat) and to sm all trade respectively. The rem aining 12.66% of the clients use the loan for different purpose like to build/ repair their house and for household consum ption. Table 4.7: Distribution of loan purpose Percent Freq. Loan purpose 34.18 22.78 20 .25 16.46 12.66 27 18 17 13 9 To buy ox To buy agricultural inputs Fatting Trade Other 10 0 .0 0 84 Total Source: Ow n com putation, 20 16 Regarding loan repaym ent, 77.79% of clients reported that they didn’t face loan repaym ent problem . But 22.21% of the client's reply that during the last five loan periods either in one or two loan period they face loan repaym ent due to die and stolen of ox, sheep, use of their loan for household expenses 21 © 2017 Global Journals Inc. (US) and illness. They repaid their loan by selling household assets and by borrowing from relatives and neighbors. Regarding group form ation, the m inim um size of the group is five clients. 4.2. Effect on Asset Ownership Am ong sam pled clients 78% of them have at least one ox and 90 % of them have m ore than two sheep with five m axim um sheep. Regarding household utensils, 83% of clients reported that they have cooking utensils and radio while 17% of clients answered that they have full household m aterials and Radios, Television and DVD player. Regarding hom e ownership 9.64% client respondents said that they didn’t have house, they are bein g living in rent-house. These clients are those health extension workers, agricultural extension workers and teachers. 44.58% of clients have ‘’Sar bet’’ while 45.78 % of them have ‘’Korkoro bet’’. The client respondents explained that, they owned their house m ostly after they were being OMFI client. Regarding house im provem ents, 19.51%clients’ house was not im proved while 43.37% are able to build additional houses and 24.39% of clients were able to decorate their house. 12.73% of clients reported that they don’t know whether there is im provem ent or not. Accordingly, 10 .84% of the respondents explained that OMFI has ‘’very-high’’ im pact on their access and control over assets while 39.76 % of clients reported that the im pact is ‘high’. 46.99% of clients explained that the im pact of OMFI on their access and control over assets is ‘m edium ’ and 2.41% answered it has ‘low’ im pact. Out of 112 non-clients, 7.96% of them have no house and 59.29%, 28.32% and 4.42% of future clients have owned ‘Sar bet’, ‘Korkoro bet’ and both ‘Sar bet’ and ‘Korkoro bet’ respectively. Of these nonclients 45.28% have no im provem ent in their house, 14.15% and 6.6% of them have been able to built better house and built additional room s. 22.64% of respondents were able to decorate their house while 11.32% of them reported that they don’t know whether there is im provem ent or not. Of 112 future clients 39.82% are very-high interested and 45.13% highly interested to participate in the loan program of OMFI. 14.16% of non-clients have m edium level of interest to participate in OMFI loan program while one observation confirm s that low level of interest to participate in the program , only for the sake of her neighbors she is going to participate. The t-test result indicates that, the m ean difference on asset ownership between client and non-clients is 0 .0 8 0 3571 and the p-value is 0 .1233 which is insignificant at 10 % significance level. This leads us to accept the null hypothesis that there is no difference between clients and non-clients in asset ownership (see Table 7 in the appendix). In general, the above descriptive and t-test analysis depict that OMF has lim ited im pact on access and control over asset between program participants and non-participants. This is because of their unwise use of loan for consum ption expenditures and their lim ited entrepreneurship on the use of loan. 4.3. Effect on Income In order to see the im pact on incom e, respondents’ average yearly incom e was asked. Accordingly, client respondents reported that 8 .34 % of them estim ate their average yearly incom e lie between 10 0 0 1 to 20 0 0 0 and 1.19 % of the clients earn estim ated average yearly incom e in between 20 0 0 1 to 30 0 0 0 . The rem ain 23.8 %, 3.57 % clients incom e falls between 50 0 1 to 10 0 0 0 and 30 0 0 1 to 40 0 0 0 respectively while 60 .71% of clients reported that their incom e falls under 50 0 0 . Their m inim um and m axim um incom e is 10 0 0 and 80 0 0 0 with the m ean incom e of 6131.311. © 2017 Global Journals Inc. (US) 22 Sim ilarly, 10 .71 %, 11.6% and 2.68 % of non-clients explained that their estim ated average yearly incom e falls between 10 0 0 1 to 20 0 0 0 , 50 0 1 to 10 0 0 0 and 20 0 0 1 to 30 0 0 0 respectively.75 % of nonclients reported that their average incom e was less than 50 0 0 . The m inim um and m axim um incom e of non-clients is 90 0 and 30 0 0 0 with the m ean incom e of 7957.619. Table 4.8: Average yearly incom e of respondents > 50 0 0 1 40 0 0 1 to 50 0 0 0 30 0 0 1 to 40 0 0 0 20 0 0 1 to 30 0 0 0 10 0 0 1 to 20 0 0 0 50 0 1 to 10 0 0 0 Up to 50 0 0 2 3 1 7 20 51 2.39% 3.57% 8.34% - 1.19% 3 12 23.8% 13 60 .71% 84 - 2.68% 10 .71% 11.6% 75% Average yearly incom e Clients (84) Percent (%) Non-clients (112) Percent (%) Source: Ow n com putation, 20 16 92.96% of clients reported that their incom e was increased because of adequate m arket for their business, good agricultural season and profitability of their business and 2.77% of clients explained that their incom e was greatly increased while 4.33% of clients explain that, their incom e has no change because of illness of the fam ily and dead of livestock like ox and sheep. Accordingly, 50 % of the respondents explains that OMFI has high im pact in increasing their incom e while only 7.32% of clients report that the im pact is very high in increasing their average yearly incom e. 39.0 2% of clients explained that the im pact of OMFI on increasing their incom e is m edium and 2.44% answered it has low im pact. A single observation states that OMFI has very low im pact in affecting her incom e. Com pared with clients, non-clients incom e was not increased. Only 32.14% of non-clients reported that their incom e has shown im provem ent due to good agricultural season and adequate m arket. The t-test result for average yearly household incom e depict that, the m ean difference between clients and non-clients average yearly incom e is -590 94.64 and the p-value is 0 .0 0 0 0 which is highly significant at 1% significance level (see Table 5 in the appendix). This leads to reject the null hypothesis that there is no difference between clients and non-clients average yearly household incom e. The im plication is that clients of OMFI earn better average yearly household incom e than non-clients. The result is consistent with the findings of (Tesfay, 20 0 3; Haym anot, 20 0 7; Roxin et al, 20 10 ; Gebru an d Paul, 20 11; Balam urugan, 20 12; Gebrat, 20 13and Ahm ed, 20 13; Kato and Kratzer, 20 13). 4.4. Effect on Saving Sam pled respondents explained their saving experience before they were client of OMFI as follows. Am ong 84 clients 92.77 % of respondents explained that they didn’t have saving account at an y institution before they join OMFI. Only 7.23% of clients reported that they have saving account in CBE with their m ale partner. 8 0 .52% of clients explained that they haven’t knowhow or awareness about saving while 14.29% and 5.19% of the clients indicated that lack of m oney and distance of financial institutions affect them not to save. On the other hand, 92.8 6% of non-clients explained that they haven’t saving account at any institution. They said that we opened saving account after the selection or recruit of OMFI agent in the near past 23 © 2017 Global Journals Inc. (US) for the purpose of loan. Only 7.14 % of non-clients have saving in the form of equb 4 and account at CBE. 70 .48 % of non-clients explained that they haven’t knowhow or awareness about saving while 12.38% and 17.14% of the clients indicated that lack of m oney and distance of financial institutions affect them not to save. Out of 84 sam pled respondents, 36.62% and 25.35% clients explained that they are saving to cover em ergency cases and for their loan repaym ent purpose respectively. On the other hand 28.17% of client respondents explained their saving is used for both em ergency cases and loan repaym ent while rem ain 9.86% respondents explained that they are saving for safety of their cash and to cover household expenses. Of 112 non-clients 64.86% and 35.14% of respondents explained that they are saving for em ergency cases and for the sake of loan (i.e. indicates after recruit saving). Respondents were asked to explain the im pact OMFI on their saving habits and they elucidate as follows. Out of 84 clients 16.87% and 39.76% of respondents m arked that OMFI has very-high and high im pact in im proving their habits respectively. 43.37% of clients explained that OMFI has m edium level im pact in im proving their saving habits. Non-clients were asked their level of interest to participate in OMFI loan program and they responded as 39.8 2%, 45.13% 15.0 4% of them have very-high, high and m edium interest to join the program . From the t-statistic test, the m ean difference on personal cash saving between clients and non-clients is – 0 .698799 and the p-value is 0 .0 0 0 0 which is highly significant at 1% significance level. This leads us to reject the null hypothesis that there is no difference between clients and non-clients in personal cash savings (see Table 6 in the appendix). This im plies that clients of OMFI have better cash savings than non-clients. In general, OMFI has im proved clients saving habits than non-clients and this finding is consistent with (Haym anot, 20 0 7 and Ahm ed, 20 13; Kato and Kratzer, 20 13). 4.5. Effect on Decision Making Wom en were asked to indicate their level of involvem ent in m ajor decision m aking in the household as a proxy to depict the im pact of OMFI on their econom ic em powerm ent. They are asked to indicate their involvem ent on deciding to purchase or sale land, house, ox (dom estic anim als in general) and on the use of their loan, saving, buy clothing, household foods etc. If wom en are able to decide activities independently or jointly with their spouse, it is considered as they em powered since they involved and able to affect the decision m aking process. Accordingly, 25.3% of clients and 5.31% non-clients explained that decisions on selling or buying of land and house were m ade independently by them selves. 62.65% of clients and 41.59% of non-clients m ade their decision jointly with their spouse while 12.0 5% of clients and 53.1% of non-clients decisions m ade by their spouse on selling or buying of land and house. The ram m ing 3.54% of non-clients decisions was m ade by other relatives. Regarding decision m aking to buy or sell ox, cow etc (dom estic anim als), 25.3% of clients and 5.31% of non-clients able to m ake decisions independently and 59.65% of clients and 39.8 2% of non-clients can decide jointly with their spouse. For instance, the t-test result on decision m aking to sell/ buy land indicates that, the m ean difference on 4 Equb is a traditional way association in which peoples save their m oney for pre-determ ined tim e, usually a week or a m onth and receives their savings in a rotation. © 2017 Global Journals Inc. (US) 24 decision m aking to sell/ buy land between client and non-clients is 0 .4244567 and the p-value is 0 .0 0 0 1 which is significant at 1% significance level. This leads us to reject the null hypothesis that there is no difference between clients and non-clients in decision m aking (clients are better decision m akers than non-clients)(see the table 7.1 in the appendix). Sim ilarly, the t-test result on decision m aking to sell/ buy ox (dom estic anim als) indicates that, the m ean difference on decision m aking to sell/ buy ox between client and non-clients is 0 .8922952 and the p-value is 0 .0 0 93 which is significant at 1% significance level. This leads us to reject the null hypothesis that there is no difference between clients and non-clients in decision m aking to sell/ buy ox (dom estic anim als) (i.e. clients are better decision m akers) (see the table 7.1 in the appendix). From the above descriptive statistics it can be concluded that, clients are better decision m akers than non-clients. Thus, their im proved or better decision m aking ability of clients was the result of the loan program which increases their incom e, saving habits and their overall confidence and status at all. Table 4.9: Wom en’s decision m aking at household level Non-clients Clients Description Other* J ointly Herself Others* J ointly Herself - - - 3(1.22%) 52(63.41%) 29(35.37%) 5(5.31%) 84(74.34%) 23(20 .35%) - 55(66.27%) 28(33.37%) 7(7.0 8%) 77 (68.14%) 28 (24.78%) - 47(56.63%) 36(43.37%) 17(15.93%) 71 (62.83%) 24 (21.24%) 4 (2.41%) 52(62.65%) 29(34.94%) 6(6.19%) 64 (56.64%) 42 (37.17%) - 38(45.78%) 45(54.22%) Decision to buy cooking m aterials 25(22.32%) 66 (58.93%) 21 (18.75%) 4 (3.61%) 54(65.0 6%) 26(31.33%) Decision to pay for health expenses 34(30 .98%) 61(53.98%) 17 (15.0 4%) 11(12.0 5%) 52(62.65%) 21(25.3%) 61(54.87%) 45 (39.82%) 6 (5.31%) 14(15.66%) 49(59.65%) 21(25.3%) 59(53.1%) 47 (41.59%) 6 (5.31%) 11(12.0 5%) 52(62.65%) 21(25.3%) 30 (26.79%) 70 (62.5%) 12(10 .71%) 7(7.23%) 56(67.47%) 21(25.3%) 45(33.63%) 69(61.0 6%) 6(5.31%) 12(12.2%) 52(63.41%) 20 (24.39%) Decision on the use of loan Decision on the use of saving Decision to buy household food Decision to buy cloth Decision to pay edir, m ahiber...etc fees Decision to sell/ buy ox Decision to sell/ buy land, house Decision to repair house Decision to rent farm land Source: ow n com putation, 20 16 Others* indicate that decision is m ade m ostly by their husbands or fam ily 25 © 2017 Global Journals Inc. (US) Respondents asked to explain the im pact of OMFI as very-high, high, m edium and low. Accordingly 10 .8 4% and 45.78% of the clients explained OMFI has very-high and high im pact in im proving their saving habits respectively. While 42.17% of clients elucidate that OMFI has m edium im pact in im proving their level of saving. Only two observations reported OMFI has low im pact on their decision m aking. The result is sim ilar with the findings of(Tesfay, 20 0 3; Haym anot, 20 0 7; Gebru and Paul, 20 11; Balam urugan, 20 12; Gebrat, 20 13 and Moham ed, 20 13;Kato and Kratzer, 20 13). In general, the descriptive statistics indicates that, OMFI increased wom en’s average incom e, im proved their saving habits and im proved their participation in m ajor decision m aking process in the fam ily than those non-clients in the study area. Regarding asset ownership and capital form ation, OMFI has lim ited im pact on its clients com pared with non-participants. The overall descriptive result depict that OMFI has positively affected its clients in increasing their involvem ent in m ajor decision m aking process, im proved their level of confidence, self esteem and reduce their poverty than non-program participants. The result of the above descriptive statistics was strengthened from the interview explanation held. According to sub-branch m anger currently OMFI has given special attention to wom en and efforts are m ade to benefit wom en’s from our services. As a result, wom en were benefited from the loan program during the last years and they im proved their overall living status. Loan officer explained that,<<I believe that,wom en were benefited from our loan program and as m uch as we can we are supporting and encouraging them and m any of our m em ber had im proved their living condition. Their incom e is increased. If you see their saving account, for sure, you will understand about their awareness on their saving habits im provem ent. Generally, I can say that, they are benefited from the loan program in m any ways>>. 4.6. Estimation Econometric Model Under this sub-section, the logistic regression m odel and propensity scores for m atching of clients and non-clients were presented. To estim ate the effect of propensity scores, logit m odel is em ployed because there is no difference on result between logit and probit m odel (Caliendo and Kopeinig, 20 0 5). Before looking the econom etric regression result, it is better to check the fitness of the m odel usually the problem of heteroscedascity and m ulticollinearity. Accordingly, the problem of heteroscedascity which is com m on in cross-sectional data was checked and solved by robustness of standard error before the estim ation of the m odel. The problem m ulticollinearity which is the relationship between continuous explanatory variables and coefficient of contingency (the association between discrete variable) was checked by different tests. To detect m ulticollinearity problem , variance inflation factor (VIF) was calculated and the result depict that the data had no problem s of m ulticollinearity (see Table 9 in the appendix). Likewise the contingency coefficients were com puted to check the association am ong discrete variables. The value of contingency coefficients lies between 0 and 1 in which 0 indicating no association between the variables and values close to 1 indicates high degree of association. Since contingency coefficient is not greater than 0 .75 all discrete variables can be used in the regression analysis (see Table 8 in the appendix). The Pseudo R 2 indicates the overall significance of the m odel. The m odel was estim ated by STATA 13.0 software using propensity score m atching m ethod of analysis. © 2017 Global Journals Inc. (US) 26 4.6.1. Determinants of Women’s Involvement in Decision Making The logistic regression m odel used nine explanatory variables such as age, m arital status, head of the household, wom en’s level of education, spouse level of education, num ber of household size, being m em ber of other m icrofinance institutions, ecology and am ount of initial wealth. The dependent variable is a binary variable taking a value of 1 if a wom an is client and 0 if not. The logistic regression estim ate was m ade to identify factors that affect wom en’s involvem ent in m ajor decision m aking in the fam ily. Accordingly the logit regression estim ate depict that wom en ’s involvem ent in m ajor decision m aking is significantly affected by age, wom en’s spouse level of education, num ber of fam ily size, head of the household, being m em ber of other MFI and am ount of initial wealth. But variables like wom en’s level of education, m arital status and ecology were insignificant in affecting wom en’s econom ic em powerm ent. The result of wom en’s level of education is consistent with Haym anot, (20 0 7)and m arital status is consistent with the finding of Gebrat, (20 13) in affecting wom en’s involvem ent in m ajor decision m aking process. The result of the m odel indicates that age of respondents is significantly affects wom en em powerm ent at 1% significant level. This is m ay be because of aged wom en’s relatively can’t decide on the household issues and dom inate by their husband. The positive relationship between age of respondents and wom en em powerm ent is consistent with the finding of Ahm ed (20 13). If wom en are head of their fam ily they are better chance to involve and decide on their asset and other fam ily issues. Therefore, the variable head of the household affects wom en involvem ent in m ajor decision m aking significantly at 1% significance level. Respondent’s spouse level of education significantly affects wom en’s involvem ent in m ajor decision m aking at 1% significant level. As the spouse’s level of education increases their awareness an d attitudes towards their wife changes and husbands start to consult their wife on m ajor decision issues in the household. Num ber of household size of respondents affects wom en’s em powerm ent significantly at 10 % significant level. The justification is that as fam ily size increases their incom e will increase by engaging in various incom e generating activities. The variable being m em ber of other m icrofinance institution (credit experience) affects wom en’s em powerm ent significantly at 1% significance level. The justification is that, wom en who have m em ber of other m icrofinance institution have better knowledge how to use the loan and invest and it is positively related with wom en em powerm ent. Am ount of initial wealth also significantly affects wom en’s em powerm ent at 5% significant level. Wom en who have initial wealth have the opportunity to start their business earlier and when com bined with their loan, they m ay have better capital to engage in a better business (see Table 10 in the appendix). 27 © 2017 Global Journals Inc. (US) Table 4.10: Results of logistic regression Num ber of obs= 196 F(9, 186)= 132.46 Prob > F= 0 .0 0 0 0 R-squared= 0 .7196 Root MSE= .2690 1 Linear regression P>t T Robust Std. Err. 6 Coef 5. Treated (trtd) 0 .0 0 9* 0 .350 0 .137 0 .0 0 9* 0 .0 51*** 0 .0 0 5* -2.64 0 .94 1.49 -2.66 1.96 2.82 .0 0 1649 .0 34696 .0 258815 .0 166268 .0 112116 .0 3530 47 -.0 0 43554 .0 324927 .0 386928 -.0 44180 5 .0 220 113 .0 9940 24 0 .0 0 0 * 9.54 .0 750 943 .7165681 0 .649 -0 .46 .0 777547 -.0 35420 6 0 .0 13** 2.50 2.46e-0 6 6.16e-0 6 Age Marital status (m rsta) Wom en level of education (wle) Spouse level of education (sle) No. fam ily size (nhsize) Head of the household (hhd) Being m ember of other MF institution (Bm ofi) Ecology (Eco) Am ount of initial wealth (Am tinw) 0 .435 -0 .78 .1499552 -.1174421 _ cons Source: Ow n com putation, 20 16 *, ** and *** are indicating variables that are significant at 1%, 5% and 10 % significance level respectively 4.6.2. Propensity Scores From the total sam ple, propensity score m atching estim ation result discards three observations from clients but it doesn’t discard any observation from non clients. As indicated from table 11 below, 112 of non-clients (untreated) are on com m on support region and 81 of the clients (treated) are on com m on support region (see Table 13 in the appendix). 5 6 Coef. represents to m ean coefficients Std. err is to m ean Standard error © 2017 Global Journals Inc. (US) 28 Table 4.11: Distribution of com m on support psm atch2: support Com m on psm atch2: Treatm ent Assignm ent Total On suppor Off suppor 112 84 112 81 0 3 Untreated Treated 196 193 3 Total Source: Ow n com putation, 20 16 Fig. 4.1: Presentation of com m on support region before m atching Source: Ow n com putation, 20 16 The m inim a and m axim a criterion deletes all observations whose propensity score is sm aller than the m inim um and larger than the m axim um propensity scores (Caliendo and Kopeing, 20 0 5). 29 © 2017 Global Journals Inc. (US) Fig. 4.2: Presentation of com m on support region after m atching Source: Ow n com putation, 20 16 Accordingly, the result of estim ated propensity score varies in between 0 .0 0 0 to 0 .998 with the m ean of 0 .10 3 for untreated and from 0 .119 to 0 .999 with the m ean of 0 .85 for the treated. That is, clients whose estim ated propensity scores less than 0 .119 and larger than 0 .998 are not included in the m atching exercise. That is [0 .119, 0 .999] and [0 , 0 .998] are propensity scores for treated and untreated respectively. Therefore, by m inim a and m axim a criterion, taking the m inim um propensity score from the treated and the m axim um score from the untreated form s the com m on support region. Thus, the com m on support regions lie between [0 .119, 0 .998] which show none of observations was dropped from non-clients in the sam ple (see Table 11 in the appendix). Table 4.12: Distribution of estim ated propensity scores Maxim um Minim um Std. Deviation Mean Observations 0 .9981149 0 .9997642 .9997642 0 .0 0 0 0 20 7 0 .1190 692 0 .0 0 0 0 20 7 0 .223330 7 0 .1776374 0 .424432 0 .10 350 1 0 .8531177 0 .4247653 112 84 196 Source: Ow n com putation, 20 16 © 2017 Global Journals Inc. (US) 30 Untreated Treated Total Fig. 4.3: (a) Kernel density estim ate for non-clients (i.e. m ost of non-clients are found in the left m iddle left partly) Kernel density estimate 5 4 Density 3 2 1 0 0 .2 .4 .6 psmatch2: Propensity Score .8 1 Kernel = epanechnikov, bandwidth = 0.0414 Fig. 4.3: (b) Kernel density estim ate for clients (i.e. m ost of clients are found in the right hand side m iddle left partly) The above graphs depicts that there is wide area in which the propensity score of clients are sim ilar to those non-clients. 4.6.3 Choosing Matching Algorithm Different m atching estim ators can be used to m atch the treated and the untreated in the com m on support region. The question of choosing m atching algorithm depends on the pseudo-R2, balancing test and num ber of m atched observations (Dehejia and Wahba, 20 0 2). So that for this data kernel 0 .5 is chosen based on the above criteria i.e. low pseudo R2 (pseudo R2= 0 .18 4), the balancing test that balances all explanatory variables (6 insignificant explanatory variables) after m atching and the largest m atched num ber of observations (193) are considered. 31 © 2017 Global Journals Inc. (US) Table 4.13: Results of checking m atching algorithm s Num ber of m atched observation Pseudo R 2 Balancing test * 193 193 193 193 0 .344 0 .331 0 .263 0 .236 2 2 3 4 193 193 193 0 .730 0 .730 0 .730 5 5 5 193 193 193 0 .315 0 .243 0 .184 3 4 6 Matching algorithm Nearest neighbor (NN) NN (1) NN (2) NN (3) NN (4) Radius 0 .1 0 .25 0 .5 Kernel 0 .1 0 .25 0 .5 Source: Ow n com putation, 20 16 4.6.4 Testing the Balance of Propensity Score and Covariates After m atching, every covariates m ean between the two groups in the m atched sam pled has been reduced and pseudo-R 2 should be relatively low (Caliendo and Kopeinig, 20 0 5). The m ajor aim of propensity score estim ation is to balance the distributions of relevant variables in both groups. Below from table 13 before m atching age, wom en level of education (wle), num ber of household size (nhsize), being m em ber of other m icrofinance institutions (bm ofi), and am ount of initial wealth (am tinw) were significantly different for the two groups of respondents. But after m atching these significant variables were insignificant which indicates that the differences in covariates m ean between the treated and untreated groups was elim inated and now the covariates between the groups is balanced (see Table 14 in the appendix). Table 4.14: Propensity scores and covariates balancing Me an t-test Varia ble t p>t Control Treated Unm atched Matched 26.12 4.40 -2.82*** -3.73 0 .0 0 0 0 .0 0 0 0 .0 0 5 0 .0 0 0 .0 9922 .72158 35.848 37.751 .86771 .86288 31.262 31.494 Unm atched Matched nm atched Matched 0 .22 -0 .0 3 0 .827 0 .975 2.1696 2.20 11 2.190 5 2.1975 Unm atched Matched Marital status (Mrsta) 5.0 8*** 1.17 0 .0 0 0 0 .246 .60 714 1.1783 1.3929 1.3951 Unm atched Matched Wom en level of education (Wle) 1.63 -0 .77 0 .10 5 0 .442 1.6518 2.0 588 1.9167 1.9383 Unm atched Matched Spouse level of education (Sle) © 2017 Global Journals Inc. (US) 32 _ pscore Age 2.26** 0 .23 0 .0 25 0 .820 4.1696 4.6985 4.80 95 4.7778 nm atched Matched 2.11** 2.19 19.34*** -0 .66 -0 .14 0 .0 6 0 .0 36 0 .0 30 0 .0 0 0 0 .512 0 .887 0 .955 1.5536 1.5296 .15179 1.0 22 .910 71 .89853 1.7262 1.716 .9881 .98765 .90 476 .90 123 Unm atched Matched Unm atched Matched Unm atched Matched 5.59*** 0 .0 0 0 2.36 0 .0 20 8114.8 10 657 16145 150 15 Unm atched Matched No. household (Nhsize) size Head of the fam ily (Hhd) Being mem ber m icrofinance (Bm ofi) of Ecology (Eco) Am ount of initial wealth (Am tinw) Source: Ow n com putation, 20 16 *** and ** show level of significance at 1% and 5% respectively (before m atching ). The fairly low pseudo-R 2 and the insignificant likelihood ratio tests supports the hypothesis that both treated and non-treated groups have sim ilar distribution in covariates X after m atching (i.e. there is com plete balance in the characteristics in both groups). After this procedure, we can com pare observed outcom es for participants with control groups that lie in the com m on support region (see Table 14 in the appendix). Table 4.15: Ch-square test for the joint significance of variables LR chi2 Pseudo R 2 Sam ple 192.80 41.24 0 .720 0 .184 Unm atched Matched Source: Ow n com putation, 20 16 The result of all the above tests indicate that the m atching algorithm being chosen and used is com paratively best for this data and thus, now it is possible to estim ate ATT for clients of OMFI. 4.6.5. Estimating Average Treatment Effect on Treated (ATT) To m eet the objectives of this study, this part evaluates the program ’s im pact on the outcom e variable (i.e. average yearly incom e, personal cash saving and asset ownership) for their significant effect on wom en clients (participant), after pre-intervention differences were controlled (See table 12 in the appendix). 33 © 2017 Global Journals Inc. (US) Table 4.16: Average treatm ent effect on the treated T-stat 2.21** 2.80 ** 0 .90 S.E. Difference Controls Treated Sam ple Variable 1989.43316 4391.18762 3750 .0 4694 8141.23457 ATT Average incom e(ayhi) 90 5.841628 .120 760 812 2537.33352 .10 81840 2 1399.70 352 .780 70 4869 3937.0 370 4 .888888889 ATT ATT Personal cash saving (pcs) Asset ownership (Ownast) Source: Ow n com putation, 20 16 ** show the level of significance at 5% Table 4.16 depict the estim ation result of the outcom e variables in which out of the three outcom e variables two of them (i.e. average yearly incom e and personal cash saving) are statistically significant while one variable (asset ownership) is statistically insignificant but positive ATT. Thus, the program intervention has resulted in a positive and statistically significant m ean difference between the client and non-clients wom en in term s of increase in incom e and cash saving. From the above table, the result of ATT is positive indicating that average yearly incom e, personal cash saving and owning asset has been im proved because of m icrofinance program in the study area. Therefore, m icrofinance program in the study area has been im proved wom en’s econom ic em powerm ent as shown from table 4.15 and the m ean difference value of the outcom e variables between client and non-clients wom en was positive. 4.6.6. Sensitivity Analysis Since PSM cannot alleviate the potential problem of unobservable variables, sensitivity analysis m ust be carried out to check the robustness. The sensitivity analysis was carried out on the estim ated average treatm ent effect for the outcom e variables and the m atching estim ator result depict that there is significant effect on the program participants. The sensitivity analysis result (i.e. at e γ=1 up to 1.6) indicates that there was no unobserved variable that affect estim ates of ATT or program m e participants. Thus, it can be concluded that the im pact estim ates of ATT are insensitive to unobserved selection bias and clearly indicates that OMFI has positive im pact on its clients (see Table 15 in the appendix). © 2017 Global Journals Inc. (US) 34 Chapter - 5 V. CONCLUSION AND RECOMMENDATION 5.1. Summary Due to the widespread of inequalities, gender discrim ination and deprivation of rights, wom en were vulnerable to poverty and they were denied from various socio-econom ic activities particularly in LDCs. To end this ignorance, gender equality is a way to prom ote growth, reduce poverty, equal access to resources and involve in decision m aking process at different levels which em powers their econom ic capacity. Credit is one m eans of em powering wom en’s econom ic capacity. MF provision to poor wom en is taken as a m echanism to reduce poverty and em power wom en econom ically. Microfinance provides social cohesion in poor comm unities, gives opportunity to escape from poverty for the poor particularly wom en. The m ain aim of this study was to analyze the econom ic im pact of Om o m icrofinance institution in em powering wom en with a case study in Gim bo woreda, South nation nationalities and peoples region (SNNPR), Ethiopia. Using m ulti-stage sam pling technique, the input data was collected from 6 rural kebeles of which 84 m icrofinance clients and 112 non-clients with structured questionnaires. Sem iinterviews were used to get additional inform ation and cross check inform ation provided in the questionnaire. Control groups were those loan applicants to take loan after they m eet the selection criteria of the institution. The data was analyzed by using both descriptive statistics and econom etric m odels. The econom etric m odel was carried out using propensity score m atching m ethod of analysis. The result of descriptive statistics reveal that 9.64% of clients and 7.96% of non-clients didn’t have house, 44.58% of clients and 59.29% of non-clients have Sar bet while 45.78% of clients and 28 .32% of non-clients owned Korkoro bet. 10 .84% and 39.76 % of client respondents explains that OMFI has ‘’very-high’’ and “high” im pact on their access and control over assets. While alm ost half of clients (46.99%) and (2.41%) clients explained that the im pact of OMFI on their access and control over assets was “m edium ” and “low” respectively. It clearly shows Om o m icrofinance has lim ited im pact on access and control over their asset or slight differences between program participants and non-participants. OMFI has significant im pact in increasing their average yearly incom e. 92.96% of client’s incom e is increased duet their participation in the program .Sim ilarly saving has significant im pact on the savings of respondents showing difference on their saving habits. 92.92% of non-clients didn’t have any saving account at any institution due to their lack of awareness. The finding of this study on the im pact OMFI in im proving wom en’s decision m aking was found significant. 12.0 5%of clients and 53.1%)of non-clients explained that they didn’t participate on decisions to sell or buy land and house. They indicated that the decision is m ade by their spouse. Thus, clients are better decision m akers than non-clients. The logistic regression result indicate that out of nine explanatory variables six variables were significantly affects wom en’s involvem ent in m ajor decision m aking. Age, wom en’s spouse level of 35 © 2017 Global Journals Inc. (US) education, num ber of fam ily size, head of the household, being m em ber of MFI and am ount of initial wealth were significant variables. Wom en’s level of education, m arital status and ecology were insignificant in affecting wom en’s econom ic em powerm ent (involvem ent in m ajor decision m aking). The estim ation of propensity score m atching result discards only three observations from clients and none from non-clients. The com m on support region lies between [0 .113, 0 .998] which show none of observations was dropped from non-clients in the sam ple. From the existing m atching algorithm kernel 0 .5 was chosen based on low pseudo-R 2 , m ore insignificant balancing test and by looking better num ber of observations in the com m on support region. The propensity score m atching estim ation result reveals that out of the three outcom e variables average yearly incom e and personal cash saving are statistically significant in affecting wom en’s econom ic em powerm ent, but access and control over asset (asset ownership) was statistically insignificant with positive ATT. Therefore, the program intervention has resulted in a positive and statistically significant m ean difference between the client and non-clients wom en in term s of increase in incom e and cash saving. ATT is positive indicating that average yearly incom e, personal cash saving and owning asset has been im proved. Sum m ing up, the findings of this study explicitly depict that, with its lim itation, OMFI had a positive im pact on wom en’s econom ic em powerm ent in the study area. 5.2. Recommendations As a policy indicator, the intervention of m icrofinance program is expected to im prove and em power the living standard of the poor’s particularly wom en at the grass root level and hence reduces poverty. As such the econom ic status of wom en and their level of participation in decision m aking will significantly im prove. Descriptive statistics of this study reveals that there is little difference between clients and non-clients in accessing, owning and control over resources. Sim ilarly the econom etric result depict ATT has statistically insignificant effect on wom en’s accesses to resources and control over asset. Thus, it can be concluded that, OMFI has lim ited im pact on wom en’s accesses to resources and control over asset. Therefore credit provision of OMFI should give priority in asset form ation, access to resources, acquire asset and able to control it. Taking these actions reduces their level of poverty and em powers wom en’s econom ic capacity. OMFI should take appropriate m easures to ensure its organizational m andates, objectives and com m it to benefit wom en from its services by providing training, advisory services and continuous follow-up to assist wom en’s econom ic em powerm ent. Linkages with other governm ental organizations like wom en and children offices and agricultural offices should be m ade to work cooperatively and address problem s. Though, the im pact of OMFI on wom en’s average yearly incom e is significant, efforts should continue to increase access to resources and accum ulation of assets that eventually help to wipeout or elim inate poverty and em power them . © 2017 Global Journals Inc. (US) 36 In conclusion, additional researches should be carried out to acquire m ore em pirical findings on the im pact of OMFIs on wom en’s access to resources, own asset and control over their resources. REFERENCES 1. Ablorh, W. (20 11). Microfinance and Socio-Econom ic Em powerm ent of wom en: Unpublished level thesis. Kwam e Nkrum ah University of science and Technology, Ghana. 2. ADB. (20 0 0 ). Microfinance Developm ent Strategy Finance for the Poor: Microfinance Developm ent Strategy , Asian Developm ent Bank. 3. Ahm ed, M. (20 13). Microfinance and Rural Wom en’s Em powerm ent: A cross-sectional level analysis. Unpublished thesis, Mekele University, Ethiopia. 4. Aghion, A. and Morduch, J . (20 0 5). The Econom ics of Microfinance: The MIT press Cam bridge, Massachusetts London, England. 5. Al-Bagdadi,H. and Bruntrup, M. (20 0 2). Microfinance Associations: The case of the Association of Ethiopian Microfinance Institutions, Eschborn. 6. Am in, R., Becker, S., and Bayes, A. (1998). NGO-Prom oted Micro-credit Program s and Wom en’s Em powerm ent in Rural Bangladesh: Quantitative and Qualitative Evidence. Journal of Developing areas, 221-236. 7. Awojobi, N., (20 14). Em powering Wom en through Microfinance: Evidence from Nigeria. Australian Journal of Business and Managem ent Research Vol.4 No.1 8. Baden, Z.O. (1997, Revised). Gender and Em powerm ent: Definitions, Approaches and Im plications for Policy. BRIDGE Institute of developm ent studies, Brighton, England. 9. Balam urgan, S. (20 12). A case study on Wom en Em powerm ent through Wondogent Om o Microfinance Institution: CLEAR IJRCM, Vol-0 2, Issue-0 3, pp. 43-63. 10 . Batem an, M. (20 10 ). W hy Doesn’t M icrofinance W ork? The destructive rise of local neoliberalism . Zed Books, London, England. 11. Batliwala, S. (1994). The Meaning of Wom en’s Em powerm ent: New Concepts from action. Pp. 127138 in Population policies Reconsidered: Health, Em pow erm ent and Rights. G.Sen, A. Germ ain, and L.C. Chen, eds. Cam bridge, MA: Harvard University Press. 12. Becker, S. and Ichino, A. (20 0 2). Estim ation of Average Treatm ent Effects Based on Propensity Score. The Stata journal, Vol-2 No. 4, pp. 358 -377. 13. Binh, T. (20 0 0 ). Special Evaluation Study on ADB’s Microfinance Developm ent Strategy. Mandaluy ong city , 1550 Metro Manilia, Philippines. 14. Caliendo, M. and Kopeinig, S.(20 0 5). Som e Practical Guide for the Im plem entation of Propensity Score Matching. Journal of econom ic survey s, Discussion Paper No. 158 8, University Of Cologne. 15. Chung, B., Kantachote, K., Polster, R. and Roets, K. (20 13). Indicators of Wom en’s Em powerm ent in Developing Nations: W orkshop in international public affairs, University of WisconsinMadison. 16. CIDA. (n.d). Wom en’s Econom ic Em powerm ent: A-CESO perspective. 17. Cochhran, S. (1997). Sam ple size Determ ination: J ohn Winley and sons, Inc., New York, USA. 18. Copas, J . (1998). Binary Regression Models for Contam inated Data: Journal of roy al statistical association, pp. 225-226. 19. Conford, R. (20 0 2). Microfinance Services: What do Pacific People Need? Asian Pacific Viewpoint, pp. 343-357. 37 © 2017 Global Journals Inc. (US) 20 . Dilayehu, D. (20 14). The Role of Microfinance Institutions in Accessing Credit and Poverty Reduction in Rural Ethiopia: In the case of OMFI in Dam ot Gale W oreda, Haw assa university , Ethiopia. 21. Ebisa et al., E., Nigussie, G., and Mitiku, F. (20 13). Filling the breach: Journal of business and Econom ic M anagem ent, Academ ia Publishing. 22. Elisabeth, H. (20 13). Microcredit and the Em powerm ent of wom en: An em pirical study on the effects of M icrocredit on W om en’s lives in Kibera, Nairobi. 23. Gashaw, T. (20 14). Microfinance Institutions in Ethiopia, Kenya and Uganda: Loan Outreach To The Poor and The Quest For Financial Viability. The Horn Econom ic and Social Policy Institute (HESPI), Policy Paper No. 0 2/ 14. 24. Gebrat, K. (20 13). Determ inants of Wom en’s Participation in Microfinance and their Decision Making in Household Incom e: Case of Am hara credit and Saving Institution, in Sekota W oreda. Mekele University, Ethiopia. 25. Gebrat, K. (20 14). Socio-Econom ic Effect of Microfinance Services on Wom en: International Journal of Academ ic Research in Econom ics and Managem ent Science, Vol, No.3, J om o Kenyatta University, Kenya. 26. Gebru, B. and Paul. I (20 11). Role of Microfinance in Alleviating Urban Poverty in Ethiopia: Journal of Sustainable Developm ent in Africa, Clarion University of Pennsylvania, Pennsylvania vol-13, No.6. 27. Getaneh, G. (220 5). Regulating Microfinance in Ethiopia: Making it More Effective. Essay s on regulation and Supervision. 28. Goetz, A. and SenGupta, R. (1996). Who Takes the Credit? Gender, Power, and Control over Loan Use in Loan Program s in Rural Bangladesh. W orld Developm ent, Vol-24, No.1: 45-63. 29. Golla, A., Malhotra, A., Nanda, P., and Mehra, R. (20 11). Understanding and Measuring Wom en’s Econom ic Em powerm ent: Definition, Fram ework and Indicators, International center for Research on w om en. 30 . Greene, W. (20 0 2). Econom etric Analy sis. 3 rd edition. New York, New York University, USA. 31. Gujarati, D. (20 0 4). Basic Econom etrics. 4 th edition. The McGraw-Hill Com pany, New York, USA. 32. Hashem i, S.M., Schuler, S.R. and Riley, A.P. (1996). Rural Credit and Wom en’s Em powerm ent in Bangladesh. W orld Developm ent, Vol-24. No.4 pp. 635-653. 33. Haym anot, E. (20 0 7). The Im pact of Microfinance on Wom en’s Econom ic Em powerm ent: A case study in Am hara Credit and Saving Institution. Unpublished level Thesis, Addis Ababa University, Ethiopia. 34. Heckm an, J ., Ichim ura, H. and Todd, P. (1997, eds). Matching As Econom etric Evaluation Estim ator: Evidence from Evaluating a J ob Training Program , Review of Econom ic studies, 261-294. 35. ILO, (1998). Sm all change, Big changes: W om en and M icrofinance. 36. Kaffa Zone Finance and Econom ic Developm ent Office (20 0 6). Statistical Abstract Bulletin. Bonga, Ethiopia. 37. Kato, M. and Kratzer, J . (20 13). Em powering Wom en through Microfinance: Evidence From Tanzania. ACRN Journal of Entrepreneurship perspectives, vol. 2, p 31. 59. 38. Kulkam i,S. V. (20 11). Wom en’s Em powerm ent and Microfinance: An Asian Perspective study. The 13 th in a series of discussion papers produced by the Asian and the Pacific Division, IFAD, Yale University. © 2017 Global Journals Inc. (US) 38 39. Ledgerwood, J . (1999). Sustainable Finance w ith the Poor: An Institutional and Financial Perspective. World Bank, Washington, D.C, USA. 40 . Luttrell, C., Quiroz, S., Scrutton, C., and Bird, K. (20 0 9, Revised). Understanding and Operatinalizing Em powerm ent: W orking paper 30 8, Overseas Developm ent Institute 111 Westm inster Bridge, London, England. 41. Maddala, G. (1992). Introduction to Econom etrics 2 nd edition. New York: Maxwell Macm illan Canada. 42. Malhotra, A., Schuler, S.R. and Boender, C. (20 0 2). Measuring wom en’s Em powerm ent as Variable in International Developm ent: Background Paper prepared for the W orld Bank W ork shop on poverty and Gender, New Perspective. 43. Mayoux, L. (20 0 5). Wom en’s Em powerm ent through Sustainable Microfinance: Rethinking Best practice. Discussion draft. 44. Mayoux, L. (20 10 ). Reaching and Em powering Wom en: Towards a Gender J ustice Protocol for a Diversified, Inclusive and Sustainable Financial Sector, International Consultant for Oxfam Novib’s W om en program . 45. Narayan, D. (20 10 ). Em pow erm ent and Poverty Reduction: A source book. World Bank, Washington, D.C, USA. 46. Otero, M. (1999). Bringing Developm ent Bank in to Microfinance: Journal of Microfinance 47. Padm a, M. and Getachew, A. (20 0 4). Wom en Econom ic Em powerm ent and Microfinance: A review on Experience of Awasa Wom en Clients. A paper presented in international Conference on Microfinance Developm ent in Ethiopia held at Aw asa, Ethiopia. 48. Pitt, M. and Khandker, S. (1997, Revised). The Im pact of Group Based Credit Program s on Poor Households in Bangladesh. J ournal of political Econom y , Brown University and World Bank. 49. Rajeev H. Dehejia & Sadek Wahba, (20 0 2). Propensity Score Matching Methods for NonExperim ental Causal Studies. The Review of Econom ics and Statistics, 8 4(1): pp. 151-161. 50 . Rosenbaun, P.R. and Rubin, D.B. (1983). The Central Role of the Propensity Score in Observational Studies for Causal Effects. Biom etrica, 70 (1), 41-55. 51. Roxin, H., Berkm uller, H., Koller, J ., Lawonn, J ., Pooya, N. and Schappert, J . (20 10 ). Econom ic Em powerm ent of Wom en through Microcredit: the case of the Microfinance investm ent and technical assistance facility (MITAF). Hum boldt University, Zu Berlin, Germ an. 52. Tesfay, A. (20 0 3). The Im pact of m icrofinance on W om en: A case study Dedebit Credit and Saving Institution. MA thesis of Addis Ababa University, Ethiopia. 53. UNDP (20 11). “Gender Inequality Index and Related Indicators.’’ Hum an Developm ent Report’’ 54. World Bank (20 0 1a). Engendering Developm ent: Through Gender Equality in Rights, Resources and Voice. W orld Bank Policy Research Report, Oxford University Press. 55. World Bank (20 12). A Guide to Regulation and supervision of Microfinance consensus guidelines CGAP: Washington D.C. 56. Vengroff, R. and Creevey, L (1994). Evaluation of Project Im pact ACEP Component of the Com m unity Enterprise Developm ent Project: The University of Connecticut. 57. Yim er, A. (20 11). Rural Microfinance and Wom en’s Em powerm ent: A case Study on Rural Wom en Clients in Dessie Zuria District, Ethiopia, International Journal of Rural Developm ent and Managem ent Studies, vol-5, No.1. 39 © 2017 Global Journals Inc. (US) Index A Affirmative · 3 Appalling · 1 Artisans · 5 C Cohesion · 35 Consensus · 6, 39 Covariates · 10, 16, 32, 33 E Elucidate · 24, 26 K Kafficho · 11 Kebeles · 2, 11, 13, 35 O Oromia · 11 Oromiya · 8 P Pervaded · 6 Precludes · 8 Propensity · 13, 14, 15, 16, 26, 27, 28, 29, 30, 31, 32, 35, 36 Protestant · 11, 18 R Ramming · 24 S Solidarity · 2 Global Journals www.globaljournals.org Reach out to us at [email protected]