The Impact of Micro Finance on Household Welfare: Case Study of a Savings Group in Lao PDR

The Impact of Microfinance on Household Welfare: Case Study of a Savings Group in Lao PDR

by Kongpasa Sengsourivong

Master Thesis Department of Regional Cooperation Policy Studies Graduate School of International Cooperation Studies Kobe University

Filed July 2006

The Impact of Microfinance on Household Welfare: Case Study of a Savings Group in Lao PDR∗

Kongpasa Sengsourivong July 2006

∗

Copyright © 2006, Master Thesis submitted to the Department of Regional Cooperation Policy Studies, Graduate School of International Cooperation Studies, Kobe University, Japan .The views and interpretations in this paper are those of the author and do not necessarily reflect the position of Kobe University. E-mail: [email protected].

ACKNOWEDGEMENTS I am indeed indebted to my academic adviser, Associate Professor Koji KAWABATA for his insightful comments and support. I also would like to sincerely thank my two former academic advisers, Associate Professor Fumiharu MIENO who provided me with valued guidance, feedback and financial support for the follow-up survey in Laos, and Professor Hiroshi UENO who provided me of tremendous support during my first semester at the Graduate School of International Cooperation Studies. I am very grateful to the critical questions and invaluable comments from both Professor Seiichi FUKUI and Professor CHEN Kuang-hui during the time of being examination committee. Contribution from Professor TAKAHASHI Motoki is gratefully acknowledged. I also wish to thank Dr. Shalini MATHUR for her assistance in editing and structuring my thesis. In addition, I would like to express my gratitude to the Lao Women’s Union, especially Ms. Sysay LEUDEDMOUNSONE (The President of Lao Women’s Union), Ms. Boualone VONGDALASENE (the President of Lao Women’s Union at the Vientiane Capital), Ms. Saikham SENGKA (the President of Lao Women’s Union at Naxaithong city), and Ms. Khampane NAOVALARD (the Vice President of Lao Women’s Union at Naxaithong city) for their advice and support throughout the surveys. Thanks also to the Women and Community’s Empowering Project, particularly Mr. Khanthone PHAMUANG (the Manager of the Women and Community’s Empowering Project), and the other staff of the project for their support and cooperation during the survey and for providing me with data on savings groups.

ii

Furthermore, I am obliged to my former boss, Mr. Huw LESTER, Senior Manager of PricewaterhouseCoopers (Laos) Ltd, and Mrs. Rae DUNSTAN who edited and proof read my paper. Special thanks to my beloved sister and brother, Ms. Fongchinda SENGSOURIVONG and Mr. Apisid SENGSOURIVONG, for their invaluable support and comments; and to my friends, Ms. Chanmany VONGLOKHAM, Mr. Paliphonepheth BOUARAVONG, Ms. Phoummaly SIRIPHOLDEJ, Ms. Visany PHOMSOMBATH, and the students of the faculty of economics and business management (FEBM) from the National University of Laos, for kindly helping me to conduct surveys. I would also like to thank my friend, Mr. Bounthone SOUKAVONG, Economic Lecturer at FEBM, for kindly providing the students to help me conduct my survey. My gratitude to my friends, Mr. Santisouk PHOUNESAVATH and Mr. Buavanh VILAVONG for their invaluable comments; to Ms. Chansathith CHALEUNSHINH, Research Officer of National Economic Research Institute, for providing me relevant data and arranging of the surveys and to Ms. Daravanh VONGVIGIT, for kindly helping me on data processing. Many thanks to my Vietnamese friend, Ms. Nguyen Thi Thuy Vinh, for her helpful support on data processing and technical aspects of the econometric software. In working through this research, I also have leant to appreciate positive externalities through discussions with my colleagues. I am thankful to them. Moreover, I would like to thank Japan International Cooperation Agency (JICA) and Japan International Cooperation Center (JICE) for the financial support during study in Japan. Finally, I would like to truly thank my beloved parents, my dear wife and my cute daughter for always supporting me. SENGSOURIVONG, Kongpasa. Kobe, Japan, July 2006.

iii

EXECUTIVE SUMMARY

Many countries recognize that microfinance can play an important role in economic development as one of the tools for poverty reduction. Access to financial services is a major issue for both rural and urban areas of Laos. Consequently, the government of Laos recognizes that access to rural finance and microfinance could be one of the major tools for poverty alleviation and places microfinance activities as one of the priority programs for the agriculture and forestry sector in order to promote sustainable growth and poverty eradication under the National Growth and Poverty Eradication Strategy. Many studies have examined the relationship between microfinance and economic development, but until now, only two key studies were undertaken in Laos. While these two studies made an important contribution to this subject, they had some shortcomings as they did not correct for problems of the self-selection and endogenous program placement. This paper surmounts these issues by adopting the methods used by Coleman (1999) to estimate the effects on household welfare or outcomes by the participation in the savings group. The survey was conducted in six villages in 2005 - 2006, in a semiurban area of Laos. All these villages had savings groups which were in operation for various lengths of time. Of these, three villages had savings groups operating for more than a year, and are called “old” savings groups. The remaining three villages also started operating savings groups but these were in operation for less than a year. These are called “new” savings groups. In the six villages, villagers were allowed to self-select to be

iv

savings group members or nonmembers. The survey sample included members and nonmembers in the six villages. Members who experienced benefits from joining the savings group by either obtaining a credit or receiving a dividend are called the “treatment” group, and those who have not benefited from the groups are called the “control” group. All members of the “control” group were relatively new members with an average membership of 2.2 months. Hence, the effects on savings group members in the treatment group can be compared with the savings group members in the control group. In addition, differences in the length of time that savings group program has been available to members in both treatment and control groups is taken into account to obtain more precise impact estimates. Inclusion of nonmembers in all six villages allowed for the use of village fixed effect estimation to control the possibility that the order in which these six villages had savings group program placement is endogenous. With this kind of survey design, the impact of savings group program on household outcomes can be straightforwardly estimated. These positive outcomes include increase in household house value, household livestock production income, household agriculture production income, household rental expenses, and household education expenses. The results illustrate that the savings group participation has large positive and significant effects on all of these outcomes, except household yearly income from agriculture. However, this can largely be explained by issues relating to the robustness of the data for this indicator. In short, the participation of savings group can increase household asset, household income from self-employment activities and support the education of children.

v

Consequently, this paper’s findings have several important implications. Firstly, the large positive impact savings group has on household asset suggest that microfinance programs may improve household status in terms of wealth. Secondly, the positive significant effects of the savings group on productivity, particularly livestock and agriculture in terms of rental on rice fields, suggest that the savings group program may be a viable strategy for the poverty eradication. This is consistent with the National Growth and Poverty Eradication Strategy (NGPES) (2004: 65) of the Government of Lao PDR which recognizes the importance of microfinance and has placed it as the one of the high priority projects for the agriculture and forestry development plan. Thirdly, the great positive influence of the savings group program on household education expenses suggests that microfinance program may be one viable strategy to reach the millennium development goals in terms of education.

vi

TABLE OF CONTENTS CHAPTER TITLE

1

2

3

4

5

PAGE

Title Page

i

Acknowledgements

ii

Executive Summary

iv

Table of Contents

vii

List of Tables

ix

INTRODUCTION

1

1.1 Issues

1

1.2 Research Topic and Objective

5

1.3 Hypothesis

5

1.4 Methodology

6

1.5 Structure of the Paper

7

MICROFINANCE IN LAOS

8

2.1 Country Brief

8

2.2 Microfinance in Laos

9

LITERATURE REVIEW

19

3.1 Impact Studies of Microfinance on Different Economic and Social Indicators

19

3.2 Methodology

26

ANALYTICAL FRAMEWORK

31

4.1 Theoretical Framework

31

4.2 Model Specification and Methodology

36

SURVEY DESIGN

46

vii

6

7

5.1 Survey Design

47

5.2 Data Description

51

EMPIRICAL RESULTS

52

6.1 Impact on Household House Asset

55

6.2 Impact on Self-Employment Activities

59

6.3 Impact on Education

64

CONCLUSION

67

REFERENCES

71

APPENDIX A

80

Table 1: Descriptive Statistics for Variables of Whole Sample Size

80

Table 2: Descriptive Statistics for Variables by Treatment Group

84

Table 3: Descriptive Statistics for Variables by Control Group

88

Table 4: Descriptive Statistics for Variables by Nonmember

92

Table 5: Impact of Savings Group on Household House Value - GLS

96

Table 6: Impact of Savings Group on Yearly Self-Employment Income from Livestock – GLS

100

Table 7: Impact of Savings Group on Household Yearly SelfEmployment Income from Agriculture – GLS

103

Table 8: Impact of Savings Group on Household Monthly Rental Expenditure – GLS

106

Table 9: Impact of Savings Group on Household Monthly Educational Expenditure – GLS

109

APPENDIX B

112

Case study: Savings Groups in Naxaithong City

112

viii

LIST OF TABLES TABLE

TITLE

2.1 5.1

Current Practices in Laos Sample Size

PAGE 16 49

ix

CHAPTER 1 INTRODUCTION

1.1

Issues Microfinance is significant source of finance for poor, lower income people in

developing countries. It provides the funding for these people to run their micro-business and to smooth their household’s consumption. Poor, lower income people have difficulties in obtaining finance from formal financial institutions such as commercial banks, due to barriers such as high collateral requirements and complicated application procedures (Yunus, 2001; and Hulme &Mosley, 1996). However, there is strong demand for small-scale commercial financial services (for both credit and savings) among the economically active poor in developing countries. These and other financial services help low-income people improve household and enterprise management, increase productivity, smooth income flows, enlarge and diversify their microenterprises, and increase their incomes (Robinson, 2001). These effects were evident from a number of impact studies of microfinance. Based on the recent studies on this subject, microfinance has significant impact on income, expenditure, assets, educational status, health as well as gender empowerment. This positive impact of microfinance on income was confirmed in studies undertaken by Hulme and Mosley (1996); Mckernan (2002); Khandker et al. (1998); Copestake et al. (2001); Sichanthongthip (2004); Shaw (2000); Mosley (2001); and Copestake (2002).

1

Research by Pitt and Khander (1996 and 1998); and Khandker (2003) found that microfinance could increase household expenditure. Morduch (1998), however, argued that the eligible households that participated in the microfinance programs have strikingly less consumption levels than the eligible households living in villages without the programs. The positive impact of microfinance on household assets was confirmed in studies by Montgomery et al. (1996); Pitt and Khandker (19996 and 1998); Mosley (2001); and Coleman (1999 and 2002). However, Mckernan (2002) found an inverse relationship between participation in program and household assets. Mckernan also found that households with fewest assets benefit most from participating in a program. Research by Chowdhury and Bhuiya (2004); Holvoet (2004); and Pitt and Khandker (1996 and d1998) found that microfinance has a positive effect on education. Similarly, Chowdhury and Bhuiya (2004); Pitt and Khandker (1996); and Pitt et al. (1999) revealed that microfinance has a positive impact on health. Furthermore, Hashemi et al. (1996); and Pitt and Khandker (1998) noted that microfinance has positive effect on women empowerment. However, there exists a counter argument that microcredit programs inflicted extreme pressure on women by forcing them down to meet difficult loan repayment schedules (Goetz and Gupta, 1996). Beside the microfinance impact on the indicators mentioned above, Kyophilavong and Chaleunsinh, (2005), found that the behavior of the village savings group members was changed as a result of participating in a program. While previously savings were kept in the form of gold, livestock, jewelry, deposits in the bank and savings at home, members now saved in the savings group.

2

The Lao People’s Democratic Republic (Lao PDR) is one of the poorest countries in East Asia in terms of an estimated per capita income of US$ 390 in 2004 (World Bank Vientiane Office, 2006). Laos is classified by the United Nation as a Least Developed Country (LDC). According to the World Bank Vientiane Office (2006), 71 per cent of Lao population lived on less than US$2 a day, and 23 per cent on less than US$1 a day in 2004. In the same year, 34 per cent of the population lived under the national poverty line; infant mortality was 82 per 1,000 live births; and life expectancy was approximately 55 years. Consequently, the government of Laos recognizes that access to rural finance and microfinance could be one of the major tools for poverty alleviation and places microfinance activities as one of the priority programs for the agriculture and forestry sector in order to promote sustainable growth and poverty eradication under the National Growth and Poverty Eradication Strategy (NGPES) (2004). Since 1987, the broad approach of microfinance, including in kind and in cash revolving role fund, has been implemented by numerous development projects which includes those of government. However, the microfinance sector is still relatively new in Laos. Although donors have made a significant investment in the last few years in microfinance programs, the sector is developing very slowly. About one million economically active people potentially require access to formal or semi-formal microfinance services. However, almost three quarters can not reach them. Approximately 300,000 people recently accessed loan and savings services. Only 21 per cent have access to microcredit from the formal sector. 33 per cent are dependant on the semi-formal sector and project initiatives and the rest 46

3

per cent are obtaining financial service from the informal sector (Microfinance Capacity Building and Research Programme, 2005). Very few empirical studies have conducted to examine the effect of microfinance on individual, household, community, or institutional levels in Laos and to test whether or not microfinance is one of tools for poverty reduction. One empirical study of Lao microfinance on Saithani case1 by Sichanthongthip (2004) showed a positive impact of microcredit on income level of individual borrower. The study involved evaluating the impact of a microfinance program (village savings group) in a semi-urban area of Laos using a questionnaire to collect primary household data from members of the microfinance program at two points of time (before and after borrowing)

2

.

Sichanthongthip reported the results of the impact on income by applying econometric analysis. However, he did not control for selection bias in the sample. Another study which was not aware with such bias is the study by Kyophilavong and Chaleunsinh (2005) who estimated the impact of a village savings group in a semi-urban area of Laos by conducting a survey for both members and nonmembers of the village savings group. They presented only the means comparison of many impact indicators for both members and nonmembers. Without the correction of the selection bias problem, the results may overestimate the impact of the program. This study will attempt to evaluate the impact of microfinance programs in Laos by correcting for the bias described above. This study also selected savings groups in Naxaithong city as a case study. The city is located in a semi-urban area of Vientiane, the

1

Saithani case is derived from that the Saithani Small and Rural Development Project (Saithani Project) which was established in 1996. The Project is a cooperative management between Lao Women’s Union and the Foundation for Integrated Agricultural Management (FIAM) (Sichanthongthip, 2004:21). 2 Data before the borrowing was collected by respondent recall.

4

Capital of Laos, and the program established as part of the Women and Community’s Empowering Project. The location of the study was selected because most of the savings groups are located in and around the capital. Results of a recent survey found that most of the 357 savings groups operating in Laos were located in and around Vientiane (Microfinance Capacity Building and Research Project (2003) cited in Chaleunsinh, 2004:7).

1.2

Research topic and objective Many studies examining the impact of microfinance at the household level,

enterprise level and macroeconomic level have shown a positive impact of microfinance on two sets of indicators – economic and social indicators – in both developing and developed countries. However, such studies have not been widely conducted in Lao PDR. Therefore, this paper will address this gap by examining the impact of microfinance to household welfare or outcomes in Lao PDR. More specifically, it will investigate the impact of savings groups at the household level in a semi-urban area of Laos. The primary objective of a savings group in Laos is to improve the living status of borrowers and their families to bring them out of poverty. The main purpose of this paper will be to evaluate the success of savings group against their primary objective.

1.3

Hypothesis Most of poor people and lower income people join microfinance program in Laos

because they can access credit with specified interest rate which is lower than that obtained from the informal money lender. They can, thus, save money. Therefore, it is

5

hypothesized that members with a long-term participation on the savings group may have better quality of life in terms of wealth, income and expenses.

1.4

Methodology To achieve the research objective, the author adopted the survey design and

research methodology of Coleman (1999) taking into account bias for self-selection and endogenous program placement which was not corrected in the previous studies relating to Laos. The author conducted a survey of 251 households in six villages in a semi-urban area of Laos – Naxaithong district which is located 16 kilometers from Vientiane, the Capital of Laos. Of these, three villages had savings groups operating for more than a year, and are called “old” savings groups. The remaining three villages also started operating savings groups but these were in operation for less than a year. These are called “new” savings groups. In the six villages, villagers were allowed to self-select to be savings group members or nonmembers. The survey sample included members and nonmembers in the six villages. Members who experienced benefits from joining the savings group by either obtaining a credit or receiving a dividend are called the “treatment” group, and those who have not benefited from the groups are called the “control” group. All members of the “control” group were relatively new members with an average membership of 2.2 months. Hence, the effects on savings group members in the treatment group can be compared with the savings group members in the control group. In addition, differences in the length of time that savings group program has been available to members in both treatment and control groups is taken into account to obtain more precise impact estimates. Inclusion of nonmembers in all six villages allowed for

6

the use of village fixed effect estimation to control the possibility that the order in which these six villages had savings group program placement is endogenous.

1.5

Structure of the paper This paper is organized as follows: •

Chapter 2 presents an overview of microfinance in Lao PDR.

•

Chapter 3 presents a literature review of previous studies on the impact of microfinance on economic and social indicators.

•

Chapter 4 discusses the theoretical framework and identifies the empirical modeling and estimation methods to be used for this study.

•

Chapter 5 describes the survey design and data.

•

Chapter 6 outlines the results of empirical analysis.

• Chapter 7 summarizes the results and draws policy implications.

7

CHAPTER 2 MICROFINANCE IN LAOS

Many countries realize that microfinance can play an important role in economic development as one of the tools for poverty reduction. The government of Laos also recognizes that role and has placed microfinance activities as one of the priority programs for the agriculture and forestry sector in order to promote sustainable growth and poverty eradication under the National Growth and Poverty Eradication Strategy (NGPES). This chapter will outline the general background on the development of microfinance in Laos, provide an overview of the current state of play in Laos in this area.

2.1

Country Brief The Lao People’s Democratic Republic (Lao PDR) is one of the poorest countries

in East Asia with an estimated per capita income of US$ 390 in 2004 (World Bank Vientiane Office, 2006). Laos is classified by the United Nation as a Least Developed Country (LDC). According to the World Bank Vientiane Office (2006), in 2004, 71 per cent of the Lao population lived on less than US$2 a day, and 23 per cent on less than US$1 a day. In the same year, 34 per cent of the population lived under the national poverty line; infant mortality was 82 per 1,000 live births; and life expectancy was approximately 55 years. In 2004, Laos had a population of around 5.8 million and a land area of 236,800 square kilometers. The country is characterized by a high degree of geographic, cultural

8

and linguistic diversity with very poor infrastructure. Moreover, villages, particularly those populated by ethnic minority groups, tend to be extremely isolated and practice subsistence agriculture. Laos is also a landlocked country which is located in the center of the Mekong region, bordered by Thailand, Vietnam, Southern China, Cambodia and Myanmar, of which, the first three are experiencing rapid economic growth. Nevertheless, Laos has significant economic potential because of its rich natural resources (such as forestry, minerals and hydro-electric power) and its proximity to major Asian economies. Agriculture is the major economic sector contributing 51 per cent of GDP and employing 80 percent of the labor force. The industry sector accounts for 23 per cent and services for 26 per cent (World Bank Vientiane Office, 2006).

2.2

Microfinance in Laos Since 1987, the broad approach of microfinance in Laos (including in kind and in

cash revolving role fund) has been implemented by numerous development projects which include those initiated by the government. However, the microfinance sector is still relatively new in Laos. Although donors have made a significant investment in the last few years in microfinance programs, the sector is developing very slowly. About one million economically active people potentially require access to formal or semi-formal microfinance services. However, almost three quarters cannot reach them. Approximately 300,000 people recently accessed loan and savings services. Only 21 per cent have access to microcredit from the formal sector, 33 per cent are dependant on the semi-formal sector and project initiatives and the rest 46 per cent are obtaining financial services from the informal sector (Microfinance Capacity Building and Research Programme, 2005).

9

This section will briefly describe the development of microfinance in terms of microfinance providers3 which consist of the formal, semiformal and informal sectors. The information is sourced from Enterplan (2003).

2.2.1

Formal sector

The formal banking system in Laos consists of: •

The Bank of Laos (the central bank);

•

Three state-owned banks (the Banque pour le Commerce Exterieur Lao (BCEL), Lao Development Bank (LDB)4, and the Agricultural Promotion Bank (APB));

•

Three joint venture banks (Joint Development Bank, Lao-Viet Bank and Vientiane Commercial Bank);

•

Six foreign commercial banks with branch offices and one foreign commercial bank with a representative office (FCBs)5.

The headquarters of each bank are located in Vientiane. APB is the largest bank in term of branch network in Lao PDR. It has one head office in Vientiane, 18 branches in the 17 provincial capitals and Vientiane Capital City, and 55 sub-service units at the district level. The head office of LDB is also located in Vientiane and it has 17 branches (one in each province). BCEL has three branches. Lao-Viet Bank has one branch in

3

This section borrows extensively from Enterplan (2003). It was merged from two state-owned banks: Lao May Bank and Lane Xang Bank. 5 Public Bank (Malaysia), Bangkok Bank, Bank of Ayudhya, Krung Thai Bank, Siam Commercial Bank, Thai Military Bank (all Thai Bank) all of which have branch offices in Vientiane. Standard Chartered Bank has a representative office and only provides offshore guarantee facilities for overseas clients doing business in Laos. 4

10

Champasack province. The rest (JVBs and FCBs) have no branches and no provincial outreach, and only operate in the area around Vientiane municipality. Practically, only one formal financial institution, APB6, has a large outreach and a true track record in rural finance. APB was established by Decree in 1992 (Decree 92/PM 1992) to assist the development of the agricultural sector in Laos. Since March 2000, APB has been largely under the provision of the Decree on Commercial Banks (Decree02/PR 2000) after operating under that separate mandate. In April 2002, APB had about 130,000 borrowers. Of these, about 123,000 were in groups (group lending). The activities of APB have focused on the provision of subsidized loans which use funds from the government and donors.

2.2.2

Semiformal sector

The semiformal financial sector in Laos is poorly developed. In this sector, activities may be classified in two categories: microfinance initiatives and illustrative International Non Government Organization (INGO) initiatives.

A.

Microfinance Initiatives

Many microfinance initiatives provide microfinance in urban and rural areas of Laos through program such as the Microfinance Project, Cooperative de Credit de Soutien aux Producteurs (CCSP), Project de Developpement Rural du District de Phongsaly (PDDP), the Rural Development Cooperative (RDC), and Credit Union Pilot Project. 6

Lane Xang Bank before merging with Lao May Bank also provided micro lending with a focus on microtraders (Bank of the Lao PDR, 2002: 31).

11

(1)

The Microfinance project

The United Nations Capital Development Fund (UNCDF) and the United Nations Development Programme (UNDP) with cooperation of the Government of Laos launched the microfinance project in late 1997. This project ran microfinance activities in Oudomxay and Sayaboury Provinces using the group lending methodology. It also includes the Sihom Project Savings and Credit Scheme (SIPSACRES) – an urban credit and savings cooperative founded under UNDP project in 1995. However, the UNDP and UNCDF did not provide funding for this project after 2002. Now, the Lao government manages the project7. While data for 2002 is unavailable, estimated data for 2000 showed that the microfinance project serves about 3,525 clients in total in October 2000 (United Nations Capital Development Fund, 2001). (2)

Cooperative de Credit de Soutien aux Producteurs (CCSP)

CCSP was established in 1996 on the initiative of a group of Lao entrepreneurs. It consists of nine member-owned cooperatives. Microcredit provided to their clients in the vicinity of Vientiane is for the purposed of agriculture, handicraft, small industry and services. CCSP also accepts savings from member and issues loan to members secured by group guarantee and compulsory savings. In September 2002, there were about 1,000 active savers and about 650 active borrowers in this initiative. (3)

Project de Developpement Rural du District de Phongsaly (PDDP)

PDDP was formed in 1997. It runs village banks for farmers in 57 villages in Phongsaly city with the support from Agence Francais Developpement (AFD). PDDP also requires compulsory savings from their members and leverages these with AFD 7

For the ones operating in Sayaboury and Oudomxay provinces are recently under the provincial government authorities. SIPSACRES is currently under the supervision of Finance Department, Vientiane Capital.

12

funds. In March 2003, about 2,350 borrowers received a loan from PDDP. In this initiative, loans are provided for cash crops, livestock, handicraft and petty trade and it applies a joint liability system for loan guarantee. (4)

Rural Development Cooperative(RDC)

RDC began running its activities in August 2001 when it received the funds from the Vientiane Municipal Development Fund. Its operation is implemented in thirty villages in Vientiane Capital City. Its services include both credit and savings. Loans are issued for different kinds of economic activities such as agriculture, handicraft and trade. These loans require physical collateral and a 20 per cent savings deposit for guarantee. RDC had between 400 and 500 borrowers in 2003. (5)

Credit Union Pilot Project

The Credit Union Pilot Project is one of the subprojects of Asian Development Bank TA cluster 3413 – LAO: Rural Finance Development. In March 2003, the savings and credit unions (SCUs) were established in Vientiane, Savanakhet and Luang Prabang Provinces. The three pilot credit unions have licensing as a condition of the ADB Banking Sector Reform Programme Loan (BSRPL).

B.

Illustrative INGO initiatives

In addition to the microfinance initiatives, there are about 1,600 village revolving funds (VRFs) that have been supported by donors and NGOs (Bank of the Lao PDR, 2002: 31). International Non Government Organizations (INGOs) use village revolving funds (VRFs) to widely provide financial services in the rural areas. These organizations

13

tend to focus on the operation in the areas which are not covered by banks including the APB. In these areas, villagers can only access to credit from friends, family members, and moneylenders. Around 35 INGOs and a few bilateral and multilateral agencies are supporting rural development projects in Laos. These projects have set up cash or in-kind funds for villagers as part of their integrated programmes. Assistance is provided for agricultural equipment, gravity water and irrigation supplies, seeds, rice banks for food security, animal banks, and medicinal drug fund. The projects depend on subsidized credit and when this credit finishes when the project ends. There is rarely any local resource mobilization. As a result of the above, a substantial number of projects are not sustainable. The Lao Women’s Union (LWU) is a key intermediary for INGOs rural financial projects. LWU was established in 1955 and is a mass organization which has been implemented at three levels – district, provincial and national. LWU operates in every village through the country. It is active in a significant share of INGOs projects as an intermediary, an organizer, and as the Lao Government partner. Due to its reach into the villages, LWU has frequently been the implementing agent or partner for INGOsupported projects. In addition to LWU, there are three more mass organizations 8 and five government offices 9 at district and provincial level to run microfinance programs throughout the country (Microfinance Capacity Building and Research Programme, 2005).

8

Lao Youth Union, Lao Front for National Construction Office and Federation of Trade Union. Agriculture Office, Planning and Investment Office, Health Office, Labour and Social Welfare Office, and Finance Office. 9

14

In addition, there are many NGOs running microfinance initiatives in Laos including CARE, CONSORTIUM, CUSO, Mennonite Central Committee (MCC), Save the Children Australia (SCA).

2.2.3

Informal sector

This sector comprises moneylenders, rotating savings and credit schemes which locally known as Houai (daily, weekly, and monthly), traders and rich farmers. According to Bagchi et al. (2002), inter-household loans are important among rural households and most of these loans are made in-kind. Nevertheless, rotating savings and credit (Houai) is a significant source of loan money for investment or emergency needs in the urban or semi-urban areas. Moneylender According to the Bank of the Lao PDR (2002), the role of moneylenders in both urban and rural areas is estimated to be quite important – approximately 50 per cent of rural villages appear to have access to money lender services. Most professional moneylenders operate close to the markets and lend for quick turnover activities. According to professional observation, rates have been stable for the last few years ranging from 10 and 30 per cent per month with lower interest rates in urban areas due to competition. Supplier credit for agricultural inputs is rare. Some foreign suppliers across the Mekong River accept precious metals as collateral for merchandise purchases. As described above, there are many different approaches to microfinance in Laos. Bagchi et al. (2002) analyzed each microfinance practice in Laos and provide commentary on each type of initiative. These comments are summarized in Table 2.1.

15

Table 2.1: Current practices in Laos10 No.

Approach/ Models

1

Institutional methodology (solidarity methodology focused on institutional sustainability) Credit union

2

3

Direct bank operation

4

Co-operative

5

Self help group

10

Implementing agencies GoL/Ministry/ Bank INGOs Bi and or Provincial Multilateral Authority/ agency Mass organization √ √

√

√

√

√ √

√

√

Limitations/ Challenges

• Not targeting poor people • Not flexible enough • Limited outreach • Rural households do not have access to financial services • Lack of confidence of people as a depositor • Limited outreach • Rural households do not have access to financial services • Lack of skilled human resources • Accessibility is difficult because of lots of paper work and formality • Lack of confidence of people as a depositor • Limited outreach • Rural households do not have access to financial services • Lack of skilled human resources • Does not disclose its financial information or audit results to the public • Limited outreach • Rural households do not have access to financial services • Not properly designed • Ownership is big challenge • Most of the time village elite people have easy access to funds • Often poorer people excluded from the benefits • Highly subsidized

Table 2.1 is sourced from the market research study done by Bagchi et al. (2002).

16

No.

Approach/ Models

Implementing agencies GoL/Ministry/ Bank INGOs Bi and or Provincial Multilateral Authority/ agency Mass organization √ √ √ √

6

Village or community bank

7

Informal kinds (cereal and animal bank) initiatives

√

√

√

8

Village Revolving Fund

√

√

√

Limitations/ Challenges

• Not properly designed • Ownership is big challenge • Most of the time village elite people have easy access to the fund • Often poorer people excluded from the benefits • Emphasis on credit rather than savings • Sustainability is questioned • Mainly introduced for food security • No cash transactions made • Not financial services • Pushing back towards the non cash economy • Highly subsidized • Very informally organized • Very limited option for local resources mobilization • Ownership is a question • Often poorly designed • Most of the time village elite people have easy access to funds • Often poorer people excluded from benefits • Emphasis on credit rather than savings • Highly subsidized • Leadership is a crucial factor • Sustainability is questioned

Source: Annex 5 of Bagchi et al. (2002).

Although the practice of microfinance in Laos is still new and at an early stage of development, this study aims to examine the influence of microfinance on household’s outcomes. Therefore, the rest of the paper will conduct an in-depth investigation of the

17

effects of savings groups on household welfare or outcomes to determine whether or not the household’s status improves with access to microfinance.

18

CHAPTER 3 LITERATURE REVIEW

Most existing studies on the impact of microfinance examine two sets of indicators11 – economic and social indicators – at different levels12. Despite the variation in the methods used and the results of studies conducted in various countries, the main impact of microfinance impact is on change in income, expenditure, assets, educational status, health as well as gender empowerment. The studies that have examined the impact of microfinance on these indicators are discussed below.

3.1

Impact studies of microfinance on different economic and social indicators 3.1.1

Impact studies of microfinance on income

The effect on income has been analyzed at the individual, household and enterprise levels. Hulme and Mosley (1996), conducted various studies on different microfinance programs in numerous countries, and found strong evidence of the positive relationship between access to a credit and the borrower’s level of income. The authors indicated that the middle and upper poor received more benefits from income-generating credit initiatives than the poorest. McKernan (2002), moreover, evaluated three significant microcredit programs in Bangladesh and discovered that the profit for selfemployed activities of households can be increased by program participation. These 11

Economic indicators are normally measurements for microfinance impact as income, level and patterns of expenditure, consumption and assets. Social indicators to measure the impact of microfinance became popular in the early 1980s as educational status, access to health services, nutritional levels, anthropometric measures and contraceptive use, for example (Hulme, 2000). 12 Hulme (2000) identified levels of assessment in different units as individual, enterprise, household, community, institutional impacts and household economic portfolio such as households, enterprise, individual and community.

19

programs were also examined at the village-level impacts in the study of Khandker et al.(1998) which showed that they have positive impact on average households’ annual income, especially in the rural non-farm sector. Copestake et al. (2001), estimated the effect of an urban credit programme – a group-based microcredit programme – in Zambia, and found that microcredit has a significant impact on the growth in enterprise profit and household income in case of the borrowers who have received a second loan. Sichanthongthip’s study (2004) also pointed to a positive impact of microcredit on the income level of individual borrowers. This can be seen from the higher monthly income earned after the member accessed credit, in the empirical study of Lao microfinance on Saithani case. Shaw (2000) studied two microfinance institutions (MFIs) in Southeastern Sri Lanka and showed that the less poor clients’ microbusiness that accessed loans from microfinance programs could earn more income than those of the poor do. Mosley (2001) evaluated the impact of loans provided by two urban and two rural MFIs on poverty in Bolivia. He found that the net impact of microfinance from all institutions, at the average level, was positive in relation to borrowers’ income, even though that net impact for poorer borrowers might be less than the net impact on richer borrowers. Copestake (2002) conducted the case study of the Zambian Copperbelt, applying the village bank model to investigate the effect on income distribution at the household and enterprise levels. The study showed that the impact on income distribution depends on who obtains the loan, who move on to larger loans and who exits the program: group dynamics was also an important factor. As he discovered, “Some initial levelling up of business incomes was found, but the more marked overall effect among borrowers was of income polarization.”(Copestake, 2002: 743).

20

3.1.2

Impact studies of microfinance on expenditure

Expenditure is another indicator to measure the impact of microfinance. Pitt and Khandker (1996 and 1998) estimated the effect of microcredit obtained by both males and females for the Grameen Bank and two other group-based microcredit programs in Bangladesh on various indicators. They showed that the clients of the programs could gain from participating microfinance programs in many ways. It can be seen that income per capita consumption could be increased by accessing a loan from a microcredit program such as the Grameen Bank. Khandker (2003) also conducted research on the long-run impacts of microfinance on household consumption and poverty in Bangladesh by identifying types of impact in six household’s outcomes as outlined bellow: •

Per capita total expenditure;

•

Per capita food expenditure;

•

Per capita non-food expenditure;

•

The incidence of moderate and extreme poverty;

•

Household non-land assets

The author found that the microfinance effects of male borrowing were much weaker than the impact of female borrowing and there was decrease in return to borrowing all the time. Moreover, he noted that the impact on food expenditure was less pronounced than the one on non-food expenditure. Besides, he showed that the poorest gained benefits from microfinance and microfinance had a sustainable impact in terms of poverty reduction among program participants. In addition, the author discovered that there was spillover effect of microfinance to reduce poverty at the village level. In contrast, the impact was less noticeable in reducing moderate rather than extreme poverty. Morduch

21

(1998), however, argued that the eligible households that participated in these three microfinance programs have strikingly less consumption levels than the eligible households living in villages without the programs.

3.1.3

Impact studies of microfinance on wealth

A further indicator of the impact microfinance is wealth. Montgomery et al. (1996) examined the performance and impact of two microfinance programs in Bangladesh. They found that there were positive impacts of a microcredit program, such as the Bangladesh Rural Advancement Committee’s (BRAC’s) Rural Development Program (RDP), on both enterprise and household assets. Clearly, even though total value of household assets had a slight increase after the borrowers obtained last loans, there had significant increase in the value of productive assets. Pitt and Khandker (1996 and 1998) also noted that the microcredit had a positive impact on women’s non-land assets. Mosley (2001) also pointed out that there was positive impact of microfinance on asset levels. He further stated that accumulation of asset and income status was generally highly correlated, which led to extreme correlation between income poverty and asset poverty. Coleman (1999) investigated the impact of a village bank on borrower welfare in Northeast Thailand. He found that there was a slight impact of program loans on clients’ welfare. However, he discovered that the village bank had a positive and significant impact on the accumulation of women’s wealth, particularly landed wealth but this result included bias from measured impact (discussed in methodology below). In contrary to the positive results, Mckernan (2002) found an inverse relationship between participation in

22

program and household assets. Mckernan also found that households with fewest assets benefit most from participating in a program.

3.1.4

Impact studies of microfinance on educational status

Many impact studies of microfinance have focused on educational status. Chowdhury and Bhuiya (2004), studied the impact of a microfinance program, BRAC poverty alleviation program, in Bangladesh, and found that both member and nonmember groups of BRAC had improved in educational performance. However, the BRAC member households benefited much more than poor non-member households. Furthermore, girls gained more than boys. Holvoet (2004) investigated the effects of microfinance on childhood education by examining two microfinance programs in South India – one with direct bank-borrower credit and another one with group mediated credit. The author showed that loans to women, through women’s groups, had a significant positive impact on schooling and literacy for girls, whereas it remained mainly unchangeable in the case of boys. However, in case of direct individual bank-borrower lending, there was no improvement in educational inputs and outputs for children. Pitt and Khandker (1996) found that a credit to the participants provided by a microfinance institution like the Grameen Bank, could grow school enrolment for children. They found, for example, that in case of the Grameen Bank and Bangladesh Rural Development Board’s (BRDB) Rural Development RD-12 program, credit lending to women had a significantly positive impact on schooling for boys (Pitt and Khandker, 1998).

23

3.1.5

Impact studies of microfinance on health

Indicators related health issues are also applied as proxies to examine the impact of microfinance. Chowdhury and Bhuiya (2004) found that microfinance program, led to a good improvement in child survival and nutritional status. Pitt and Khandker (1996) also noted that there was a rise in contraceptive use and decrease in fertility in case of the participants obtaining a credit provided by the Grameen Bank. However, there was no evidence to prove that an increase in contraceptive use or a decrease in fertility resulted from the participation of women in group-based credit programs. But fertility reduction was observed and contraceptive use slightly increased in case of men’s participation (Pitt et al., 1999).

3.1.6

Impact studies of microfinance on empowerment

Microfinance also leads to the empowerment of women. Hashemi et al. (1996) studied two main microfinance programs in Bangladesh, the Grameen Bank and the Bangladesh Rural Advancement Committee (BRAC). They noted that the participation of the programs had important positive impacts on eight different dimensions of women’s empowerment: •

Mobility,

•

Economic security,

•

Ability to make small purchases,

•

Ability to make larger purchases,

•

Involvement in major household decisions,

24

•

Relative freedom from domination by the family (especially, women’s ownership of productive assets),

•

Political and legal awareness,

•

Participation in public protest and political campaigning.

In another study, Pitt and Khandker (1998) found that the behavior of poor households was significantly changed in case of women’s participation in the program credit, in Bangladesh. It, for example, could be seen that every 100 additional taka credit provided to women by the microcredit programs, namely the Grameen Bank, BRAC and BRDB, increased yearly expenditure for household consumption by 18 taka, whereas that provided to men from the same programs grew yearly household consumption expenditure by 11 taka. However, there exists a counter argument that microcredit programs inflicted extreme pressure on women by forcing them down to meet difficult loan repayment schedules (Goetz and Gupta, 1996). Besides the microfinance impact on the indicators mentioned above, one study tried to examine how the savings group in Laos affects the behavior of member of a village savings group. It showed that the behavior of the village savings group members was changed as a result of participating in a program. While previously savings were kept in the form of gold, livestock, jewelry, deposits in the bank, and savings at home, members now saved in the savings group (Kyophilavong and Chaleunsinh, 2005).

Different methodologies have been adopted to analyze the impact of microfinance programs. These are discussed in the next section.

25

3.2

Methodology Empirical studies on the impact of microfinance can be distinguished into two

main groups: those that were not concerned about selection bias problem13 and those that were. A large number of impact studies of microfinance programs did not take into account selection bias. According to Chen’ s review of 11 impact studies of the Grameen Bank in Bangladesh, no study corrected the selection bias (Chen (1992) cited in Coleman 1999, p.109). Shaw (2004) also studied two microfinance programs in Sri Lanka, and used a questionnaire and conducted interviews in one semi-urban and two rural groups. The author presented only median comparisons of client incomes among four household income groups (extreme poor, poor, near-poor and nonpoor), at time of their first loan (June 1994) and at the time the research was conducted (June1999). However, he did not take into account selection bias. Sichanthongthip (2004) evaluated the impact of a microfinance program of the village savings group in a semi-urban area of Laos and used a questionnaire to collect primary household data from members of the microfinance program at two points of time (before and after borrowing). He reported the results of the impact on income by applying econometric analysis. On the other hand, he also did not control for selection bias. Another study which also did not take into account such bias is the study by Kyophilavong and Chaleunsinh (2005) who estimated the impact of a village savings group in a semi-urban area of Laos by conducting a survey for both member and nonmember of village savings group. They presented only a comparison of the mean values of many impact indicators for both members and nonmembers.

13

As Holvoet (2004: 32) noted, “Selection bias may occur because of nonrandom program placement, through selection by program staff, or because of self-selection by program participants”. For further details, please read more in Baker (2000), and Greene (2000). For numerous discussions for methods to correct that bias, please refer to Heckman and Robb (1985), Moffitt (1991), and Ravallion (2005).

26

A number of researchers, however, have attempted to correct the selection bias. Hashemi et al. (1996) evaluated the effect of rural credit programs in Bangladesh by undertaking ethnographic research in six villages during the period 1991-94 and conducting a survey in late 1992. The authors classified their sample into four groups consisting of: •

Grameen Bank members,

•

BRAC members,

•

Nonmembers living in the Grameen Bank villages (who would have been eligible to join either BRAC or Grameen Bank),

•

A comparison group living in villages without the Grameen Bank or BRAC programs but who would have qualified to join the credit programs.

They also tried to address the possibility of selection bias by including nonparticipants and participants in Grameen Bank villages and comparing them with women living in villages without microcredit programs. In contrast, Hashemi et al. (1996) did not control for possibility of endogenous program placement, even though, the authors presented the effects of credit program on eight dimensions of empowerment by applying logistic regression models as mentioned in section 3.1.6 above. Hulme and Mosley (1996) also tried to solve for selection bias by studying different credit programs in a number of countries. In their study, they included eight microfinance institutions which provide group lending. Of these eight institutions, two were used a control group in the case a loan had been approved for participants but they had not yet received any amount of the loan. However, only the means of different

27

outcome variables for both treatment and control groups were introduced. There were no statistical analyses of the differences between the two groups. In addition, the possibility of endogenous program placement could not be controlled with their available data (Coleman, 1999: 109). Recently, many papers on the evaluation of microfinance programs have adopted an econometric approach and taken account of both selection bias and nonrandom program placement (Pitt and Khandker 1996&1998; Pitt et al. 1999; Coleman 1999 & 2002; Khandker 2003; Khandker et al.1998; McKernan 2002; Morduch 1998). The methodology was applied by Pitt and Khandker (1996&1998) to attempt to correct both selection bias and nonrandom program placement in group lending, is described as the below. The authors used survey data of the Grameen Bank and two other group lending programs in Bangladesh (the Bangladesh Institute of Development Studies (BIDS) and the World Bank). They conducted a quasi-experimental household survey of 87 villages in 29 thanas14. They sampled randomly for both members and nonmembers from villages that had a microfinance program. They also randomly chose households from villages without a program. In this case, credit program availability was applied as an identifying variable. The authors, however, identified that systematic variation will occur between the two kinds of villages because of the possibility of endogenous program placement. Thus, village fixed effects estimation was applied to control for unobserved variation between villages. Nevertheless, households living in program villages, which are exogenously excluded from the program by program rules which restrict participation for household with more than 0.5 acres of land were principally excluded from membership 14

A thana is the administrative center for numerous villages.

28

consideration by any of the three programs, that were sampled for this survey. This might have resulted from the possibility of collinearity between the village-specific dummy variables, identifying fixed effect, and the availability of the program. Many impact studies then applied a similar methodology. As can be seen from the studies by Pitt et al. (1999); Khandker (2003); Khandker et al.(1998); McKernan (2002); and Morduch (1998). But, Khandker (2003) did some further interesting work on the data set. He used the same data set as Pitt and Khandker (1996) and, then did a follow up survey of the same households in 1998/99 to come with panel data. However, according to Coleman claimed to the eligibility criteria for membership consideration:

Most group lending programs…do not impose such eligibility criteria. Rather, they attempt to attract the relatively poor and dissuade the relatively rich from participating by the small size of loans, the high frequency of meetings, and the stigma of belonging to a poor person’s credit program. Hence, the method of Pitt and Khandker could not be implemented in most group lending programs. Moreover, even in the context of the three Bangladesh programs they studied, their survey found that some 18-34% of program participants in fact had wealth that should have excluded them from participating. Hence, the use of this eligibility criterion as a key exclusion restriction may not be appropriate (Coleman, 1999:110).15 The methodology applied by Coleman (1999) did not require the existence and enforcement of exogenously imposed membership criteria to identify program impact. As part of his study, a unique survey which allows for the use of relatively straightforward estimation techniques was applied for data collection. Then, the survey was done four

15

Morduch’s study in 1998 (cited in Coleman 1999: 110) found that in program village, Pitt and Khandker marked “eligible” for households’ participation of the microcredit program as any households in program village, consisting households that should have been excluded principally. On the other hand, they follow exactly the rule on the eligibility criteria for marking household as eligible or not eligible in non-program village. Thus, both “treatment” group and “control” group was not conformed each other, and it led to overestimated results on program impact.

29

times over the course of a year, during 1995-1996, for both members and nonmembers of the village bank in 14 villages in Northeast Thailand. Six of those villages were identified as “control” villages that were recognized to receive NGO support for village bank within one year after the identification. It means that there was self-selection for villagers in six control villages as participants had already decided whether or not they want to be membership of the village bank. The rest of eight villages were “treatment” villages of which seven villages had a village bank for two to four years and one village started its village bank suddenly after the first survey. The comparison between the “old” village bank members in the eight treatment villages and the “new” village bank members in the six “control” villages could be undertaken. In addition, the author identified precise impact estimator as a variation of the length of time for the program availability in the treatment villages. When nonmembers in all villages were included in the sample, this allows for the use of village fixed effect estimation to control the possibility that the order in which these 14 villages received program support is endogenous. Based on empirical evidence, most studies showed positive impact of microfinance on different dimensions of outcomes at different levels, even though they applied various methodologies. Until now, no study on impact of microfinance in Laos has corrected the selection bias and endogenous program placement. Therefore, this study will try to take into account that bias and the author will apply method used by Coleman (1999) to evaluate the impact of a village savings group in a semi-urban area of Laos.

30

CHAPTER 4 ANALYTICAL FRAMEWORK

This paper will to examine the impact of microfinance program by applying two sets of indicators – economic and social indicators. To test the hypothesis that members with a long-term participation on the savings group may have better quality of life in terms of wealth, income and expenses, the theoretical framework, model specification and methodology will be discussed in this chapter.

4.1

Theoretical Framework In economic perspectives, consumer theory is used to identify the consequences

of consumer behavior by maximizing consumer utility. Therefore, on the basis of the framework of consumer theory, the impacts of microfinance program can be illustrated by using the concept of utility maximization. To evaluate the effects of group-based credit program participation on household behavior and intra-household resource allocation, it will consider a simple model that generates an efficiency argument for targeted credit for the rural poor. This paper will follow the framework that was developed and used by Pitt and Khandker (1996). Assume that households of size n comprise two working age adults (the male head and his wife) plus n-2 dependents. The households maximize a lifetime utility function which contains time-specific utility functions of the form:

31

U t = U (n, Qi , H i , li )

(1)

Where: Qi: a set of market goods consumed by household member i, Hi: a set of non-market household-produced goods allocated to member i, li : leisure time consumed by household member i. To generalize (1), each of the two adult household members, denoted by f and m, needs to maximizes his (if m) or her (if f) own utility uit,

u it = u i (n, Qi , H i , l i ) , i = f,m

(2)

Where household social welfare is some function of the individual utility functions U it = U (u ft , u mt ) , it can be represented as

U it = λu ft + (1 − λ )u mt

,0 ≤ λ ≤ 1

(3)

In this, λ is the weight given to women’s preferences in the social welfare function of household. The parameter λ can be thought to represent the bargaining power of female household members relative to males in determining the intra-household allocation of resources. When λ =0, female preferences are given no weight and the household’s social welfare function is identical to that of the males. The household-produced goods H include “household care” activities such as preparation of food, childcare, and the gathering of fuel. 32

H = H ( Lmh , L fh , G, F )

(4)

where, Lmh : time devoted to the production of H by males,

L fh : time devoted to the production of H by females, G : a vector of market goods used as inputs in the production of H F : a vector of technology parameters that affect efficiency in H good production. Due to socio-cultural factors, relatively few poor women work in the wage labor market. The reservation wage for market work is, therefore, relatively high. In addition to this preference effect on female wage employment, workers typically must commit to a full day’s employment even in the spot labor market. If men’s time (or that of other household members) is a poor substitute for women’s time, and if important H-good outputs, such as child care and food preparation, must be “produced” daily (cannot be stored), then working a full day may entail foregoing the production and consumption of highly valued H goods. Thus, the non-storability and time-intensity of production of household goods H, the indivisibility of time allocation in the wage labor market, and high reservation wages due to cultural impediments to wage employment outside the home all result in most women being engaged in the production of household goods H in every period to the exclusion of employment in market activities. These effects are magnified if λ is small and male preferences tend to favor certain kinds of H-goods produced on women's time. However, there are also economic activities that produce goods to sell in the market that is not culturally frowned upon. These activities produce what we refer to as Z-goods. These activities permit part-day labor and do not require that production occur

33

away from the home. For many Z-goods a minimum level of capital is necessary, even though many of these production activities can be operated at low levels of capital intensity. This minimum is often the result of the indivisibility of capital items. For example, dairy farming requires no less than one cow, and hand-powered looms have a minimum size. For other activities, such as paddy husking, where the indivisibility of physical capital is not an issue, transaction costs (or the high costs of information) take the place of the minimal level of operations. In many societies these indivisibilities may be inconsequential, but among the rural poor of many developing countries, including Laos, household income and wealth is so low that the costs of initiating production at minimal economic levels are quite high. Poverty alleviation programs, such as the Rural Works Programs, which target households by drawing them into (in-kind) wage labor, have a comparatively small direct effect on the time allocation and productivity of women. In addition, transportation and other transaction costs in labor markets may be so high as to make part-day labor un-remunerative. Formally, we represent the production function for the Z-goods as:

Z = Z ( K , Lmz , L fz , A, J )

Where Lmz : labor time of head devoted to the production of Z, L fz : labor time of wife devoted to the production of Z, K : capital in Z production, A : a vector of variable inputs

34

(5)

J: a vector of technology parameters that affect efficiency in Z-good production (information). Positive production requires a minimal level of capital K, Kmin. The production function (5) can be operated at a non-zero level when Lmz or L fz are zero, but not when both are zero. For example, although at least one cow is required in the production of milk, any person’s labor can be used to obtain the milk. In other examples, Kmin may stand for the minimum information which is required to produce and market home production. Households maximize lifetime utility subject to a budget constraint that requires that the present discounted value of expenditure on goods and leisure equal the present value of all wealth, defined as assets plus the discounted present value of the time endowments, and the two production function equations (4) and (5). Household ability to borrow has significant influence on the time path of household consumption. Households having very low levels of initial assets as collateral may not be able to borrow to achieve the minimum capital requirements necessary to operate the Z-good activity. At very low levels of income and consumption, reducing current consumption to accumulate assets for this purpose may not be optimal because it may seriously threaten health, production efficiency, and life expectancy, as shown in Gersovitz (1983). Therefore, for many households, the Z-good activity is never carried out (and Lfz = 0) and women who do not work in the wage labor market devote all their time to production of the non-market good H and to leisure.

35

4.2

Model Specification and Methodology 4.2.1

Model Specification

From the model mentioned above, the reduced-form determinants of credit program participation include: •

The prices of market time,

•

The price of the purchased market good Q,

•

The prices of the market inputs into H-good production including the cost of averting a birth and other determinants of fertility,

•

The prices of variable inputs into Z-good production,

•

The price of the capital good,

•

Age and education levels of the borrower and spouse,

•

Access to transfers from non-resident relatives and,

•

Village-level characteristics (V).

Whether or not poor households, especially the women, are credit-constrained is a complex issue. Rashid and Townsend (1994) concluded a detailed study of this issue in the context of targeted group-based lending. They recommend that inefficient consumption and production outcomes may be a result of risk, private information, communications and enforcement difficulties. Significant evidence shows that collateral and education, the health risks and intermittency of employment associated with childbirth, and cultural barriers result in the limitation of women participation in the formal credit market. Rashid and Townsend note that the evidence does not in itself imply that outcomes are inefficient if, for example, women have access to other sources

36

of finance such as transfers, or if male household members obtain funds for female household members. A primary focus of this paper is to estimate the impact of credit programs on various household outcomes such as household consumption, household income, and household asset. The author proposes to estimate the conditional demand equation for each outcome to be investigated, conditioned on the household’s program participation as measured by the quantity of credit borrowed. As noted earlier, the methodology adopted is that developed by Pitt and Khandker (1996). Consider the reduced form equation (6) for the level of participation in one of the credit programs by household i in village j ( C ij ), where level of participation will be taken to be the value of program credit

C ij = X ij β c + V j γ c + Z ij π + ε ijc

(6)

where Xij : a vector of household characteristics (e.g., age and education of household head), Vj: a vector of village characteristics (e.g. prices and community infrastructure), Zij : a set of household or village characteristics distinct from the X’s and V’s in that they

affect Cij but not other household behaviors conditional on Cij (see below),

β c , γ c , and π are unknown parameters, ε ijc : a random error having three components

ε ijc = μ j + η ij + eijc

(7)

37

where μ j : an unobserved village-specific effect,

η ij : an unobserved household-specific effect, eijc : a non-systematic error uncorrelated with the other error components or the regressors.

The conditional demand for household outcome Yij conditional on the level of program participation Cij is Yij = X ij β y + V j γ y + C ij δ + ε ijy

(8)

where β y , γ y and δ are unknown parameters and ε ijy is comprised of

ε ijy = (αμ j + μ jy ) + (θη ij + η ijy ) + eijy

(9)

where α and θ are parameters (corresponding to correlation coefficients), μ jy and η ijy are additional village and household-specific errors uncorrelated with μ j and η ij , respectively, and eijy is a non-systematic error uncorrelated with other error components or with the regressors. If α ≠0 or θ ≠0 the errors ε ijy and ε ijc are correlated. Econometric estimation that does not take this correlation into account will yield biased estimates of the parameters of equation (8) due to the endogeneity of credit program participation Cij.

4.2.2

Methodology

Econometric estimation of this equation (8) will yield biased parameter estimates if ε ijy and ε ijc are correlated and this correlation is not taken into account. According to

38

Coleman (1999), two different sources which lead to such correlation are (1) selfselection into the savings group and (2) nonrandom program placement. He illustrated sources of such correlation that for the first source of correlation, by considering a sample of households drawn only from with villages with a savings group16, some households will have selected to be savings group members, and others will have not selected to be members. ε ijy and ε ijc will almost definitely be correlated in this sample. For example, if there are many entrepreneurial households joining savings group, then unmeasured “entrepreneurship” would affect both the decision to become a member and impact measures such as income and assets. In contrast, if ε ijy and ε ijc were negatively correlated, estimation of savings group impact would be biased downward. For example, the relatively poor may join the savings group more than the rich who might feel stigmatized in a group for poor people. Coleman also demonstrated the second source of correlation as follows:

Consider another commonly used sample, which includes households of village bank members from some villages and randomly selected households from villages without a village bank…Now it is possible for ε ij and μ ij to be correlated across villages if village bank placement is not random. For example, if some village are viewed as more entrepreneurial or better organized, have more dynamic leaders and such leadership spills over to effect others’ behavior in the village, or are simply poorer (e.g., living in flood-prone or drought-prone areas),

16

In original paper of Coleman (1999) used word of “village bank” but hence use word “savings group” to conform to this study. Actually, village bank in Coleman’s case and savings group in this study are quite similar concept because of they both are applied from village bank model of the Foundation for International Community Assistance (FINCA) but the village bank is much more dependent on external fund than internal fund which contrast to the savings group. Please see Appendix B section 1.2.1 for point of view of savings group and section 2.2.1 for its source of fund.

39

and if NGOs use such criteria to determine village bank placement, then ε ij and

μ ij will be correlated (Coleman, 1999: 113)17. To deal with the case of correlation between ε ijy and ε ijc 18, therefore, the author will follow the method of Coleman (1999)19 to measure the impact of savings group on the household outcomes. As mentioned in section 3.2 of chapter 3, Coleman (1999) conducted a unique survey which permits the estimation of an alternative, simpler specification allowing for the use of relatively straightforward econometric techniques to measures program impact. By applying Coleman’s method, the author conducted a survey on 251 households in six villages in semi-urban area of Laos (Naxaithong district). Of these, three villages had savings groups operating for more than a year, and are called “old” savings groups. The remaining three villages had also started operating savings groups that were in existence for less than a year. These are called “new” savings groups (see table 5.1 of the next chapter). Both old and new savings groups cannot be identified as “treatment” villages and “control” villages as the case of the village banks in Northeast Thailand done by Coleman (1999). This is a result of slight difference in the way funds are managed. Unlike the village banks in Coleman’s case, savings groups in this case study are much more dependent on internal funds than external funds (NGOs’support) (see more

17

The words of “village bank” and the symbols of “ ε ij ” and “ μ ij ” used in the paper of Coleman (1999)

are equivalent to the words of “savings group” and symbols of “ ε ij ” and “ ε ij ” used in this study c

y

respectively. 18 To cope with such correlation, Moffitt (1991) suggests three standard procedures: using instrumental variables, using panel data, and assuming an error distribution of the outcome variable without treatment. For more explanation was mentioned in the paper of Moffitt (1991: 295-305). 19 Aghion and Morduch (2005) also suggested that the method of Coleman (1999) is one of the approaches for impact evaluation to address selection bias.

40

explanation in Appendix B under section 2.2.1). It means that some savings group members can obtain a credit in short time (two or three days) after the establishment of savings group by depending on internal fund (savings from members). Or some new savings group members can access a loan quickly after self-selecting to become a member20. However, there were many villagers who had already self-selected to become members of savings groups, but had not obtained any credit from savings groups yet. This was the case in both old and new savings groups, particularly among old members21 in the old savings groups who were with the savings group for more than a year. This may be result of their lack of demand for credit because they may have enough funds to run their activities or to smooth their consumption. Their motivation for participating in the savings group may have been to benefit from the yearly dividends they could expect from their monthly deposit with the savings groups. The classification above allows the author to set a “treatment” group in which savings group members have gained benefit from joining the savings group by either obtaining a credit or receiving a dividend. Similarly, a “control” group in which savings group members have not benefited from the savings groups can be identified. The control group of savings group households would presumably, on average, share the same unobservable characteristics (such as entrepreneurship, gender attitude, etc.) as the treatment group of savings group members (Coleman, 1999 and Mosley, 1997). In both

20

It is normally one or two day(s) to get a loan after becoming a member of savings group if the savings group has enough funds and the member meet the requirements for borrowing. 21 13.98 percent of member sample (n=93) in old savings groups has never borrowed from the savings groups since they have been a savings group member. Their membership ages varied from 11.5 months to 35.5 months.

41

old and new villages, both members and nonmembers were surveyed. With this survey design, equation (8) can be replaced by a single impact equation22 as follows:

Yij = X ij α + V j β + M ij γ + Tij δ + υ ij

(10)

Where Yij ,Vj , and Xij are defined as before Mij : a membership dummy variable equal to 1 if household ij self-selects into the savings group, and 0 otherwise. Tij : a dummy variable equal to 1 if a self-selected member has already had gained benefit from savings groups23, and 0 otherwise.

δ : measures the average impact of the savings group on Yij. The membership dummy variable Mij can be thought of as a proxy for the unobservable characteristics that lead households to self-select into the savings group. For example, it captures the unobserved variables that led to the correlation across households between ε ijy and ε ijc . The variable Tij which measures availability of the savings group to members who have self-selected is exogenous to the household. This is unlike the amount borrowed (which may not be exogenous with respect to the village, as discussed below). According to Coleman’s (1999) specification, the variable Mij which captures unobservable household characteristics, can eliminate the correlation between Tij and ε ijy

22

This equation was adopted from the study of Coleman (1999). This is slightly different from Coleman’s case (1999) which he used access to program credit as distinguish between control and treatment village but this study, the use of a gain from participation of savings group by either obtaining a credit or receiving a dividend is to be the proxy for distinguishing between control and treatment group.

23

42

due to self-selection at the household level. In addition, if the order in which villages have savings group program placement is random with respect to unobserved village characteristics, then efficient and unbiased estimates can be obtained with Vj as a vector of specific village characteristic affecting Yij. However, if the order is not random with respect to unobservable village characteristics, then using specific village characteristics as regressors will lead to a biased estimate of impact. This is a difference on the second source of bias discussed above. While in this study does not control villages and treatment villages, the problem of non-random program placement is resolved by the author dividing the six villages which have own savings groups in terms of different times of establishment (that is, three old and three new savings groups). However, the order in which villages had program support to establish a savings group may not be random. For example, if the most dynamic villages established a savings group with the program support before less dynamic villages, then Tij and ε ijy will be positively correlated and δ will be biased. One method to eliminate this bias is through village fixed effects estimation (Coleman, 1999). If the order of savings group placement is random 24 , however, then village fixed effects estimation is still unbiased, but less efficient than using specific village characteristics as regressors. The empirical model in (10) can be improved upon by recognizing that some treatment members have received benefit from joining savings groups longer than others. In the six villages, the age of the savings group differed from one month to three years.

24

Unofficial information obtained from the Project’s staff and the District Lao Women’s Union at Naxaithong city noted that the villages which have three old savings groups were target villages. Before the establishment of savings groups in those villages, the project proponents came to the villages to discuss the possibility of savings group establishment with local authorities of those villages. By contrast, the local authorities in the villages with the “new” savings groups, themselves had approached the project proponents and requested them to establish a microfinance program in their villages.

43

Of these six villages, there were members who classified to the treatment group and have their membership ages varied as the age of the savings group. Giving that the cumulative amount that a member can borrow grows over the life of the savings group, and that a member can receive a dividend every twelve months, one would expect to see greater impact in villages with older savings group. Hence the empirical model estimated, which is based on the one used by the Coleman (1999), is as follows:

Yij = X ij α + V j β + M ij γ + MAMTij δ + μ ij

(11)

Where μij is error representing unmeasured household and village characteristics that determine outcomes; the treatment dummy variable Tij is replaced by MAMTij, the numbers of months the savings group has been operating in the village. In other word, MAMTij can be thought of as the number of months participants have gained benefits from participation of savings groups. MAMTij is zero for members in control group and for nonmembers in the six villages. Therefore, MAMTij is a more precise measure of program availability than Tij. Now, δ measures the impact per month of program availability. Like the equation (10), if the order of program placement is random with respect to unobservable village characteristics, then efficient and unbiased estimates can be obtained with Vj as a vector of specific village characteristics. If program placement, however, is not random with respect to unobservable village characteristics, then MAMTij and ε ijy can be eliminated with village fixed effects. This specification (11) is considerably easier to estimate (if Yij is uncensored, then OLS is appropriate), according to Coleman (1999).

44

The regression results of the equation (11) by employing Coleman’s (1999) methodology, to evaluate the impact of savings group on household outcomes will be presented in chapter 6. Before moving to that, survey method and data will be described in the next chapter.

45

CHAPTER 5 SURVEY DESIGN

Many microfinance programs provide financial services to the poor and lower income people in urban and semi-urban areas of Laos. The Women and Community’s Empowering Project (WCEP) is one of the many programs which launched microfinance programs in semi-urban areas of Laos. The project provides microfinance through a “savings group” in three districts in Vientiane, the capital of Laos. One of these three cities was selected for this case study (see Appendix B for the overview of the project and this case study). The reason this case study was conducted in the Vientiane area is due to the fact that most of savings groups operate in and around the capital. It can be seen that of the 357 savings groups throughout the country that are being monitored by seven agencies25, a significant number of savings groups are running in Vientiane Capital City. Most of the groups in Vientiane are operating with the technical support of two main bilateral projects – “Small and Rural Development Project for Women” and “Capacity Building Project for Women and Community,” – between the Central Lao Women’s Union, the Foundation for Integrated Agricultural Management (FIAM) and Community Organizations Development Institute: Thailand (CODI) (the postal survey conducted by the Microfinance Capacity Building and Research Project (2003), cited in Chaleunsinh,

25

Seven agencies consist of District Lao Women’s Union, District Lao Youth’s Union, District Planning Office, District Social Welfare Office, District Finance Office, District Agriculture and Forestry Office and branches of Agriculture Promotion Bank (Chaleunsinh, 2004).

46

2004:7). This section discusses how the survey conducted and provides a description of the data used for this study.

5.1

Survey design26 In September 2005, a survey was conducted of 251 households in six villages in

the Naxaithong district which have their own savings group. At the time of survey, three villages had just established “new” savings groups. Two of these were three months old, one was existent for just one month (see table 5.1). Other three villages had “old” savings groups ranging in existence from just over one year to almost three years. These were selected from the list of savings groups provided by the project administers of projects. Chose to the three “new” savings groups selected for this study under distance condition. It means that the three old savings groups were not far from the three new savings groups – under 15 Kilometer. Both members and nonmembers of the village savings group were encouraged to interview. Households which had a least one person as a member of a savings group were selected for interview and households which has no one joining the savings group were also chosen for interview as shown in table 5.1. A random sampling method was not employed because a quasi-census survey of member and nonmember households was conducted. Both old and new savings groups included two kinds of members:

26

The method for survey design was the one used by Coleman (1999) for conducting survey in 14 village banks in Northeast Thailand as mentioned in section 3.2 of Chapter 3. However, the case of savings groups Laos are mostly depended on internal fund which mobilized money from member deposit to lend out to their members (see more explanation in Appendix B under section 2.2.1). Therefore, hence could not identified the new village savings groups to be as “Control Villages” and the old village savings groups as “Treatment Villages”; however, there are “Control” group and “Treatment” group for both old and new village savings groups as identified in this section, which are quite different from the Coleman’s case.

47

•

Members who had gained some benefit from the savings groups by either obtaining a credit or receiving a dividend from the savings group since they had been members. These members are classified as the Treatment group.

•

Members who were identified as Control group, are an inversion of the treatment group.

Questionnaires were done in three sets. One was for households in the six villages, which were interviewed for both members and nonmembers. The questionnaire conducted for this study is a replication of Hulme and Mosley (1996). A similar study was conducted for Saithani case of Lao microfinance by Sichathongthip (2004) and it was also reproduced by Mosley (2001) in the study of microfinance and poverty in Bolivia. However, the questionnaire was slightly modified in some parts to be more suitable and relevant for the purposes of this study. The household questionnaire contained data on household characteristics, assets, income, expenditure, deposits and borrowing. Some data on household assets, income, expenditure were asked for the current period (at survey date of September 2005) as well as recall period (5 years before conducting the survey). In each household which have a member of savings group, an adult or head of household, who knows almost everything about the household finances, was invited for interview27. The same method was done for households which had no members of the savings group.

27

Most interviewees were women (wives) because, in Lao custom, most wives know most about income and expenses of their household. As Sheck-Sandbergen and Choulamany-Khampoui (1995: 91) noted about women in Laos, “Women are generally good at financial management and accounting because of their social and economic experience in managing the household finances and the local economy: they are the sellers, buyers, traders, middle-women and entrepreneurs” (quoted by Kunkel and Seibel 1997:116).

48

Table 5.1: Sample size

No. Savings groups

I. 1 2 3 II. 4 5 6

Old savings group in: Nakountay village Huannamyene village Dongluang village Sub-total New savings group in: Phonekeo village Phonesavanh village Sisavard village Sub-total Grand total

Established Number Number of date of Number of nonmember savings of household household group household in savings of savings in village group group

Sample size by: Member Member in in control treatment group group Nonmember

Total of sample size

01-Oct-02 12-Jun-03 01-Apr-04

215 353 184 752

186 217 75a 478

29 136 109 274

39 34 16 89

3 1 0 4

19 13 8 40

61 48 24 133

15-Jun-05 01-Jun-05 10-Aug-05

95 123 59 277 1,029

80 56 53 189 667

15 67 6 88 362

8 19 15 42 131

19 9 20 48 52

6 17 5 28 68

33 45 40 118 251

Note: a) this figure is assumed equal to the number of families in the savings group.

Source: Author’s survey data, September 2005 and March 2006.

49

The second questionnaire was for collecting village information including characteristics of the villages such as the presence of schools and prices of goods in each village. This interview was conducted with the head of the village and members of the group committee at the village level. The last questionnaire was for general information on savings group. It included questions regarding the number of the group members, sources of funds, group deposit balances, deposits and credit methods, and methodology for solving bad debts (in case). This interview was conducted in depth for the group committee at the village level. In September 2005, the household surveys were administered by third year students of the faculty of economics and business management from the National University of Laos, together with the author. These students were trained on how to conduct a survey before interviewing the villagers. During survey period, the author was the leader of the team and supervised all the students to ensure that correct information was obtained. In addition, the author conducted the village surveys as well as in-depth interviews with the group committees. Finally, the author conducted a follow up survey in March 2006. This survey was targeted at collecting more data on village characteristics and additional data on deposits, credits and other areas which the previous surveys had not covered. Information for the last survey was mostly obtained from the group committee and gathered by in-depth interviews with the chief of Lao Women Union at Naxaithong district. In addition, secondary data from project documents (summary report, progress report, savings group manual and other) were obtained from the project, Lao Women Union at Municipality level and group committee.

50

5.2

Data description The data used in this analysis is taken from the author’s survey in September 2005

and March 2006 as described above. All variables used for this study were drawn from the studies of Coleman (1999 and 2002); Hulme and Mosley (1996); Pitt and Khandker (1996 and 1998); and from Annex 1 of Gaile and Foster (1996). The main dependent variables which represent proxies for the household’s outcomes are classified in three main parts: household monthly expenditure, household yearly income and household assets. Household monthly expenditure consists of two main categories: household monthly food expenditure and household monthly non-food expenditure. Household yearly income is classified to two main parts: household yearly self-employment income and household yearly non self-employment income. Household assets comprise household owned land and house assets, and household owned non-land and house assets. Independent variables are mainly characteristics of households and villages but the variable of most interest for this study is the number of months a person has been a member of a savings group. Household characteristic variables are age, gender, education level, household size, for example. Village characteristic variables consist of distance to market, price of goods, having a school and being near a river, for example. More details of data description are shown in the Appendix A: table 1-4.

51

CHAPTER 6 EMPIRICAL RESULTS

This chapter presents and interprets the results of estimating a single impact equation (11) for a wide variety of household outcomes. These outcomes include household house asset value28, household annual self employment income from livestock, household annual self employment income from agriculture, household monthly rental expenditure, and household monthly educational expenditure29. Before moving to those results, some particular innovations and specifications for estimation of the equation (11) will be briefly described below. As mentioned in chapter 4, this study applies the analytical method used by Coleman (1999). Four innovations borrowed from his study have been incorporated in the equation (11) for measuring the impact of the savings group: 1) It allows for use of village fixed effects30; 2) It identifies a control group of members who have not benefited from the savings groups and then surveys “treatment” members, “control” members and nonmembers, thus allowing for the use of a member dummy variable

28

This value stands for value of house with land (not empty land or land for rice field and crop plant). Barnes (1996:4) noted that house asset is one of the physical assets representing for the wealth of household. 29 For other proxies representing for household outcomes as shown in Appendix A-table 1-4 are not reported in this paper. According to the econometric exercise, those outcomes do not show significant correlation with the regressor of interest (months of savings group membership). 30 This was used by Pitt and Khandker (1996&1998); Pitt et al. (1999); Khandker et al. (1998); Mckernan (2002); Coleman (1999&2002).

52

to be a proxy for unobservable differences between members and nonmembers; 3) It uses the value of land owned by household five years before this survey was conducted to be a proxy for initial household wealth31; 4) It uses the length of time that the savings group has been in a village to obtain more precise impact measures. In addition, this study uses four specifications from Coleman (1999) to illustrate the importance of the four innovations to correct the bias resulting from self-selection and endogenous program placement. The four specifications are described below: 1) First is the “correct” specification which use village fixed effects and include a member dummy variable and variable for land value owned by the household five year before this survey. As mentioned in chapter 4, fixed effects estimates will be consistent (and unbiased if the dependent variable is uncensored) but possibly inefficient. 2) Second is the specification which is identical to the first but this includes a vector of specific village characteristics 32 that one might expect to influence the dependent variables. If the order of program placement is random, and the author was fortunate enough to choose all the relevant 31

Proportion of land value is 91.38 per cent of household wealth in this sample and land value is an excellent proxy for wealth which is similar to Coleman’s (1999) case. This land value comes from the value of empty land or land for agriculture or any land which is not for house building. According to Coleman (1999), collecting data on land owned five years before the surveys is relatively easy. Because land transactions tend to be large and important for a household, yet relatively infrequent, households can easily recall land transactions made in the previous five years, and land owned five year earlier (and its value) can be deduced. Of course, collecting similar data on other assets is not feasible without surveying over several years. 32 These village characteristics include dummy variable for that village has either a pig pond which has water throughout the year or be near river; dummy variable for that village has paved road or near main road (Km 13 road); dummy variable for that village has primary school grade5; a distance from village to main market; a price of traditional chicken (Gailard) per Kg; and daily wage for construction.

53

observable village characteristics, then the estimation of nonfixed effects (“village characteristics” or VC) model will be efficient and consistent (and unbiased if the dependent variable is uncensored). If the order of program placement is not random, or the author chose the wrong village characteristics, then the estimation of this model will be biased and inconsistent. 3) Third is “naive” estimates which ignore the first two innovations. For example, the naïve estimates use specific village characteristics rather than fixed effects, and they leave out the member dummy variable. 4) Fourth is “super-naive” estimates which are identical to the third but also ignore the third innovation by leaving out the variables for land value owned five years before the survey33. In this study, assuming that dependent variables (household outcomes) are uncensored, the method of ordinary least squares (OLS) is applied for estimating the equation (11). In addition, the White test is applied to test for heteroskedasticity which lead to unbiased estimators of OLS. Then, the generalized least squares estimation (GLS) is used to correct heteroskedasticity, called weighted least squares (WLS) estimations (Wooldridge, 2003: 268-276). The regression result of the equation (11) which covered all above methods will be discussed bellow under categories of the impact on household house assets, selfemployment activities and education.

33

According to Coleman (1999), the naïve models correspond to the models most commonly used to measure impact, which ignore selection bias, endogenous program placement, and prior wealth.

54

6.1

Impact on household house asset To measure the impact of savings group participation on household house assets,

all four specifications, mentioned above, were applied to the equation (11). Four sets of regression results for impact of savings group on household house asset value, corresponding to four specifications, are presented in Table 5 (Appendix A). Jointly significant test of explanatory variables shows strongly significant evidence at the 1 per cent level for all four methods regressions. All four specification regressions show that the savings group has positive and significant impact on household asset value. It can be seen that both fixed effect estimation and nonfixed effect estimation demonstrate reliable explanatory of the monthly savings group membership with large and significance at the 5 per cent level (292,097; p = 0.0493) and at the 10 per cent level (292,097; p = 0.0522) respectively. In contrast, the regression results of both naïve and super-naïve models which ignore the first two and the first three innovations respectively, also show significant impact with large proportion of coefficients of the month of savings group member at the 1 per cent level as (335,021; p = 0.0066) and (358,260; p = 0.0041) respectively. These results are in contrast with those of the study of Coleman (1999 & 2002) that showed insignificant effects of village bank on house value. Table 5 (Appendix A) shows that the coefficient of the member dummy variable in the household house asset value is insignificant (1,481,225; p = 0.6816), indicating that unobservable differences between members and nonmembers (such as entrepreneurship, preferences, etc.) are of little consequence. This is consistent with the result of the coefficient on member dummy variable and the house value in the fixed effect model found by the Coleman (1999 & 2002). Therefore, in the fixed effect model, there is no

55

correlation between the member dummy variable and the household house asset value. Hence, there is no correlation of selection bias from unobservables. In addition, the coefficient on the value of household owned land 5 years ago is positive but it is statistically insignificant at least at the 10 per cent level against a twoside alternative by employing the first three specifications. However, if we look at Pvalue level, the fixed effects and nonfixed effects estimations show those coefficients as (0.068769; p = 0.1683) and (0.068769; p = 0.1710) respectively, implying that those coefficients are acceptable at the 16.83 per cent and 17.1 per cent level. It means that the initial wealth of household could slightly increase the value of household house asset. Nevertheless, the coefficient of female-owned land value 5 years ago is positive and statistically significant at the 1 per cent level to women’s wealth in the study of Coleman (1999). One might expect the women who are the head of a household to participate more actively in the savings group than women who are not the head of household. However, the coefficient on this variable and the household house asset value in all four specifications regressions is insignificant at least 10 per cent level. This may be explained by the fact that 52 per cent of women who are the household head are either single, widowed, divorced. This implies that there was no male adult in the households to help with any household activities, particularly economic activities. According to Kunkel and Seibel (1997:106), in Lao Loum group 34 , there is a division of labor, but not fixed,

34

As Kunkel and Seibel (1997:6) note, “The ethnography of Laos comprises the lowland Lao Loum who grow paddy in the river valleys along the Mekong and its tributaries but also include the upland Tai Dam, who are non-Buddhist; the upland Lao Theung many of which practice slash-and-burn agriculture in the hills above the valleys and on the mountain slopes; and the mountain-top Lao Soung who have long resisted government efforts at resettlement and the substitution of new for old cash crops.” In this case study, all sample come from Lao Loum group.

56

between the two sexes in households for doing household work, agriculture work, livestock, collection of forest products and handicrafts. Many of roles of women and men are overlap. However, women who do or run family activities, especially economic activities, (or are entrepreneur) have positive significant increase of household house asset value. It can be seen that the coefficient of gender variable is positive and significant in relation to household house asset value in all four estimation models. For example, there is positive correlation between gender variable and household house asset value (14,021,407; p = 0.093), corresponding to the fixed effect estimation model (see Appendix A: Table 5). Education level is highly significant in all four estimation models at the 5 per cent level. For example, in case of fixed effect estimation model, the coefficient on maximum education by individuals is large and significant (854,694; p = 0.0442). If education’s years can be considered as a proxy for human capital, then this result likely stands for the complementarities of human capital and physical capital in production (Coleman, 1999:120). Coleman (1999) also discovered the positively significant correlation between education level for both females and males and women’s wealth but they were shown in case of the “super-naive” estimation model only. Age of the individuals is totally positive significance at the 5 per cent level, corresponding to the four specification models. In case of the fixed effect estimation model, for example, the coefficient on the age of individuals is largely significant (312,418; p = 0.023). It could be interpreted that the older of the individuals in the households who participate in the savings group, the more significant is the increase in their household house asset value. This result was also found by Coleman (1999). He

57

revealed that the coefficient on age-sex categories with women age 40 to 59 and women age 60 and over were positive and significant at the 5 per cent and 1 per cent levels respectively to women’s wealth. However, his result was shown by the super-naïve estimation model only. People, who have run their business or family activities for a long time, have been increasing their household wealth. More experience in running the business or family activities, may lead to effectiveness and efficiency in production by the accumulation of knowledge and experience in that field. It can be seen that the coefficient on number of months doing business are significant at the 1 per cent level to household house asset value, corresponding to the four estimation models (see Appendix A: Table 5). It is not surprising that the influence of the number of relatives in the village on household house asset value is positive and significant at the 1 per cent level for the four estimation models. For example, the coefficient on the number of relatives in the village to the household house asset value is large and significant (1,552,376; p = 0.0001), in case of the fixed effects estimation model. The number of relatives represents village relationships that would help each other when facing serious situations including selfemployment activities, or to share happiness among them. Having a chief of a savings group committee or member of the group committee in household has a large and statistically significant impact on household house asset value at the 1 per cent level for the four estimation models. It is statistically significant (12,703,787; p = 0.0026) in case of the fixed effect estimation model, for example. Similarly, Coleman (1999) found large positive correlation between the independent

58

variable of having a village chief or assistant chief in the household and women’s wealth for the first three specifications. One might expect that living in the village that has paved roads or been a near main road may improve household status more than living in the village does not have these advantages. The nonfixed effects, naïve and super-naïve models were negative effect of a variable for village characteristic – village has paved road or been near main road – on household house asset value with statistically significant level at 5 per cent as shown in Table 5 (Appendix A). This result is somewhat anomalous and difficult to explain. However, overall, the empirical results show that participation of savings group has been increasing household asset value by employing all four estimation models.

6.2

Impact on self-employment activities The effects of savings group on self-employment activities are explained by three

impact categories: household annual self employment income from livestock, household annual self employment income from agriculture, and household monthly rental expenditure35.

6.2.1

Impacts on household annual self employment income from livestock

By regressing the equation (11) with the four estimation models, the results of regressions for the effects of savings group on household yearly self employment income from livestock are shown in Table 6 (Appendix A). The fixed effect, nonfixed effects, naïve and super-naive estimates are all positive for the coefficient on the month of

35

Most of rental expenditure comes from rental of self-employment activities such as rental of rice field for planting rice, rental on shop for trading and rental on garage for vehicles fixing service.

59

savings group membership, implying that savings group participation by a member of the household increases the yearly income of household livestock production. Only two estimation models, fixed effects and nonfixed effects, produce statistically significant for that coefficient at the 10 per cent level, (19,962.47; p = 0.0606) and (19,962.47; p = 0.0640) respectively. In contrast, those results were not significant in the study by Coleman (1999) of the village bank in Northeast Thailand. The size of household also has a positive and significant effect on household annual self employment income from livestock. Table 6 (Appendix A) showed all positive relationship of the coefficient of household size to the yearly income of household from livestock production by employing all four estimation models, and these coefficients are statistically significant at the 1 per cent level. It implies that more member in the household could increase the annual income of household from livestock production. For example, in case of the fixed effects estimation model, increase in household size by one person could push up the household yearly income from livestock production by 318,013 Kip. Household which had a chief or member of the savings group committee had a large increase in the household yearly income from livestock production. It can be seen that all four estimation models result in the positive coefficients of the variable for household which has chief or member of savings group committee, which are statistically significant at the 1 per cent level as shown in Table 6 (Appendix A). In addition, Table 6 (Appendix A) shows that the fixed effects estimates produce positive coefficients on the number of civil servants in the household which is significant at the 10 per cent level (413,843.2; p = 0.0979).

60

6.2.2

Impacts on household annual self employment income from

agriculture The estimation of savings group impact on household yearly income from agriculture production is also done employing equation (11) with the four specifications. Table 7 (Appendix A) shows that all four estimation models result in significant negative correlation between the month of savings group membership and the yearly income of household from agriculture production. The coefficients of the month of savings group membership in the four models are significant at the 1 per cent level. However, the study of Coleman (1999) did not find any significant relationship between the month of village bank membership and household agriculture production, corresponding to the four models. Although there is negative relation between the month of savings group membership and the household income from agricultural self-employment activities, this result is still ambiguous. It is difficult to conclude that the participation of savings group decreases the household income from the agricultural self-employment activities, because such a negative relationship of the coefficient may have other reasons behind it. One possibility accounting for this negative relationship is prices of agricultural products which fluctuate throughout the year. Another possibility is related to underestimated real household income from agricultural self-employment activities. This can be explained as data of the yearly household collecting only from a part of agricultural products sold in the markets that are excludes the value of agriculture products for the owner’s household consumption and those reserved for reproducing (cultivating). The annual household income from the agricultural self-employment activities shown in the regressions,

61

therefore, is lower than the real value of the agriculture products. Some sample households had a zero value of household income from the agriculture products because of the non-sale of agricultural products (only used for household consumption), even though they produced those products. This was particularly the case for households who rented the rice fields for plantation36. Member dummy variable is positive and significant in relation to the yearly income from household agricultural production. Its coefficient is large and significant (286,830.8; p = 0.0808) for the fixed effects estimation model and (286,830.8; p = 0.0847) for the nonfixed effects estimation model. This implies that unobservable differences between member and nonmembers have large consequences. In other word, being member of savings group can increase the yearly income of household agricultural production. In contrast, in the study by Coleman (1999), there is an insignificant relation between member dummy variable and household agricultural production. Having a female household head has a negative and significant relationship to the yearly income of household agricultural production, corresponding to the four estimation models. Its coefficient is statistically significant at the 1 per cent level for all cases of the estimations as shown in Table 7 (Appendix A). This result shows that household head who is not female crucially increase the annual income of household agricultural production. This may be true because, in rice plantation, heavy tasks like ploughing or harrowing are mainly men’s work, while it is the women who select the seed, uproot seedlings, transplant and weed (Kunkel and Seibel, 1997:106).

36

Most of them did not sell their rice, only for own consumption, after payment of the rental fee to landlord by some amount of rice, regarding to the agreement between landowner and leasee.

62

The size of the household and the number of months doing business has a positive and significant relationship to the yearly income of household agricultural production as can be seen from the results of all four estimation models. Their coefficients are statistically significant at the 1 per cent level as shown in Table 7 (Appendix A). The number of generations that the family has been in the village has a positive impact on the yearly income of household agricultural production. Its coefficient is significant at the 5 per cent level. It means that more number of generations’ family in village can help agriculture work to push more products, and then household income may increase. It is not surprising that the value of household owned land 5 years ago or initial wealth has positive effect on the yearly income of household agricultural production because land is essential capital for agricultural plantation. Focus on the variable of village characteristic, the distance from the village to main market has a negative effect on the annual income of household agricultural production. This is true in reality because if goods, especially agricultural products, are located far away from the market (city market), their price are lower than the ones sold in the main market.

6.2.3

Impacts on household monthly rental expenditure

Table 8 (Appendix A) show the regression results of the equation (11) on household monthly rental expenditure corresponding to the fixed effects, nonfixed effects, naïve and super-naïve models. All four models indicate the positive correlation between the month of savings group membership and the household monthly rental expenditure.

63

Its coefficient is significant at the 5 per cent level for the first three specifications and at the 1 per cent level for the super-naïve model. This result could be interpreted as showing that the participation in a savings group leads to increase in household monthly rental expenditure. In other words, the savings group participation supports the persons who have no physical assets to run household self-employment activities such as rice plantation and opening a shop for trading. Education level of individuals in the household also has a positive and significant impact on household monthly rental expenditure. Its coefficient is statistically significant at the 5 per cent level for all four estimation models. The result could be explained by the fact that people who achieve high education level could be more confident to do their business in new ways walkout investing their own capital (especially physical asset).They conduct their business by renting or leasing assets.

6.3

Impact on education The impact of savings group on education can be seen from household monthly

educational expenditure. Table 9 (Appendix A) shows the impact of savings group on household monthly educational expenditure by regressing the equation (11) with respect to the four specifications. The month of savings group membership has positive relation to the household monthly educational expenditure. Its coefficient is significant at the 5 per cent level for the four estimation models. This relationship indicates that the participation in a savings group can increase the monthly educational expenditure of household. It can be seen that being member of savings group for one more month could raise the household monthly education expenditure by 5,670 Kip, in case of the fixed

64

effects model, for example. One can say that the people joining savings group can support for their children’s schooling. However, this result could not be found in the study by Coleman (1999). Turning to village characteristic variables, the village which has a primary school until grade 5 has a positively significant effect on the household monthly education expenses. Table 9 (Appendix A) shows that the coefficient of this variable with statistically significant level at the 10 per cent, corresponding to the nonfixed effects, naïve and super naïve models. This result implies that a village which has a primary school until grade 5 can support more children to continue their schooling until higher education level. To conclude, there are many positive impacts of savings group participation at household level. This is seen by applying the fixed effects, nonfixed effects, naïve and super-naïve estimation models. These household impacts are shown in three categories, namely household assets, household self-employment activities and education. However, some theoretically fundamental relationship results might present the adverse effects from such as the negative relationship between the months of savings group membership and the household income from agricultural self-employment activities. This is more likely related to data collection issues. However, this paper has some limitations. First, even though this applied the generalized least squares estimation (GLS) to correct heteroskedasticity, called weighted least squares (WLS) estimations, there may be correlation among independent variables. This may lead to biased estimations of the impact on household outcomes. Second, there may be endogenous issue in variables because this study did not cover a test for the

65

endogeneity. Third, this study did not take into account the spillover effects to nonmembers in the six villages. Fourth, the values of data on household assets, incomes and expenditures are nominal. Fifth, the use of cross-section data to evaluate the impact of the savings group, may show only the short term effect and may not predict the long term impact of savings group which may be possible to ascertain by using panel data. This is a suggested area for further research.

66

CHAPTER 7 CONCLUSION

Microfinance affects household welfare or outcome through a number of channels. Research studies have showed that microfinance has significant impact on assets, income, expenditure, educational status, health as well as gender empowerment. In this paper, the effects on household outcomes by the participation in the savings group in a semi-urban area of Laos were estimated by application of the survey design and research methodology of Coleman (1999) taking into account bias for self-selection and endogenous program placement which was not corrected in the previous studies relating to Laos. The survey sample included members and nonmembers in six villages that have own savings groups. These savings group were in operation for various lengths of time. Members who experienced benefits from joining the savings group by either obtaining a credit or receiving a dividend are called the “treatment” group, and those who have not benefited from the groups are called the “control” group. All members of the “control” group were relatively new members with an average membership of 2.2 months. In order to demonstrate the importance of selection bias correction, the paper compared the results of the “correct” empirical specification to those of the three other specifications which do not take account for some or all of the bias. This paper provides estimates of the influence of participation in the savings group on household house value, household livestock production income, household agriculture production income, household rental expenses, and household education

67

expenses by applying four empirical specifications. The empirical results illustrate that the participation in the savings groups has large positive and significant effects on all of these outcomes, except household agriculture production income. This may be explained by the fluctuation of seasonal agricultural prices, and under reporting of the real value of household agricultural income. This is because data collected pertain only to income from agriculture products which were sold in the markets, not those used for selfconsumption. The results of this study differ from those of Coleman (1999). He did not find significant impact of village bank loan on household outcomes once corrections were made for self-selection and endogenous program placement. Only the more “naive” specifications (i.e. without these corrections) significantly overestimate program impact. This difference in the results between this study and Coleman’s (1999) study may come from the different contexts of these two cases, even though these two cases follow the same village bank model (see Appendix B: section 2.2.1). Regarding to Coleman’s results, he provided some explanations as:

Thailand…is a relatively wealthy developing country, with annual GDP growth at close to 10% for the past two decades, and many villagers already have access to low-interest credit from financial institutions such as the BAAC37. In fact, the average wealth of survey households was 529,586 baht, and average household low-interest debt, excluding village bank debt, was 31,330 baht, of which 9,342 baht was held by women. In such an environment, it should not be surprising that loans of 1,500 to 7,500 baht would have a negligible impact. Indeed, a common complaint of women surveyed was that the size of village bank loans was far too small for them to be productive. The village bank model of group lending, as practiced throughout the world, sometimes takes a “one size fits all” approach, based on a first loan of US$50 and a maximum of US$300. These limits are in fact commonly used in much poorer countries in Africa, where they represent a 37

BAAC stands for the semi-statal Bank for Agriculture and Agricultural Cooperatives (Coleman, 1999: 111).

68

lot of money to a village household. They are also the limits used in Northeast Thai villages, and they are arguably too low. On the other hand, the significant number of members who had worked themselves into debt by borrowing without having identified a productive activity to invest in points to the need for project field staff to redouble efforts to stress the need to invest the loans productively (Coleman, 1999:133). In contrast to Thailand, Laos is one of the poorest countries in East Asia with an estimated per capita income of US$ 390 in 2004. Real GDP grew by an annual average rate of 6.3 per cent in the 1990s and was 7 per cent in 2005 (World Bank Vientiane Office, 2006). In terms of access to financial services, about one million economically active people potentially require access to formal or semi-formal microfinance services. However, almost three quarters cannot reach them. Approximately 300,000 people recently accessed loan and savings services. Only 21 per cent have access to microcredit from the formal sector, 33 per cent are dependant on the semi-formal sector and project initiatives and the rest 46 per cent are obtaining financial service from the informal sector. In this survey sample, only 6 per cent and 4 per cent of those sample surveyed access could credit at low interest rates and make deposits with APB respectively (Appendix B: Table 5-6). These may be factors that lead to the villagers’ need for a credit for productive purpose from the savings group. This may, therefore, lead to have a significant impact on their household outcomes. This paper also has some limitations. First, even though this applied the generalized least squares estimation (GLS) to correct heteroskedasticity, called weighted least squares (WLS) estimations, there may be correlation among independent variables. This may lead to biased estimations of the impact on household outcomes. Second, there may be endogenous issue in variables because this study did not cover a test for the

69

endogeneity. Third, this study did not take into account the spillover effects to nonmembers in the six villages. Fourth, the values of data on household assets, incomes and expenditures are nominal. Fifth, the use of cross-section data to evaluate the impact of savings group, may show only the short term effect and may not predict the long term impact of savings group which may be possible to ascertain by using panel data. The use of panel data for conducting a follow up survey for an impact analysis is recommended area for further study. This will help in dealing with problems such spillover effects and endogeneity of explanatory variables. However, this paper’s findings have several important implications. Firstly, the large positive impact savings group has on household asset suggest that microfinance programs may improve household status in term of wealth. Secondly, the positive significant effects of the savings group on productivity, particularly livestock and agriculture in terms of rental on rice fields, suggest that the savings group program may be a viable strategy for the poverty eradication. This is consistent with the National Growth and Poverty Eradication Strategy (NGPES) (2004: 65) of the Government of Lao PDR which recognizes the importance of microfinance and has placed it as the one of the high priority projects for the agriculture and forestry development plan. Thirdly, the great positive influence of the savings group program on household education expenses suggests that microfinance program may be one viable strategy to reach the millennium development goals in term of education.

70

References Materials in English Aghion, Beatriz Armendariz and Jonathan Morduch, 2005, The Economics of Microfinance, London, England: The MIT Press. Asian Coalition for Housing Rights, 2005, Laos Grassroots Savings Network: Part 2, Update:

7th

December

2005,

Retrieved

June

9,

2006,

from

http://www.achr.net/Countries/Laos/LaoSavingss2.htm. Bagchi, Palash K., Sarah Wood, and Phetdavone Chaleunsouk, 2002, “Market Research Study: A Sector Analysis Update for Microfinance in Lao PDR,” Lao PDR: Concern Worldwide, July, unpublished. Baker, Judy L. 2000, Evaluating the Impact of Development Projects on Poverty: A Handbook for Practitioners, Washington DC: The World Bank. Bank of Thailand, 2006, Table 89: Rates of Exchange of Commercial Banks in Bangkok Metropolis, Retrieved June 9, 2006, from http://www.bot.or.th/bothomepage /databank/EconData/EconFinance/Download/Tab89.xls. Bank of the Lao PDR, 2002, The Banking and Financial Sector of Lao PDR: Financial Sector Note, Vientiane, Laos: the Bank of Lao PDR, 15 July. Barnes, Carolyn, 1996, Assets and The Impact of Microenterprise Finance Programs, Washington, DC: Management System International. Chanleunsinh, Chansathith, 2004, “Village Savings Groups in Lao PDR: Factors affecting their Growth: Case study of FIAM/CODI Savings Group projects in Vientiane Capital/Municipality”. National Economic Research Institute, Vientiane, Laos, 10th December, unpublished.

71

Chowdhury, A.M.R., and Bhuiya, A. 2004, “The Wider Impacts of BRAC Poverty Alleviation Programme in Bangladesh,” Journal of International Development, 16: 369-386. Coleman, Brett E., 2002, “Microfinance in Northeast Thailand: Who Benefits and How Much?” Economics and Research Department Working Paper no.9, Asian Development Bank, April. Coleman, Brett E., 1999, “The impact of group lending in Northeast Thailand,” Journal of Development Economics 60: 105-141. Copestake, J., 2002, “Inequality and The Polarizing Impact of Microcredit: Evidence from Zambia’s Copperbelt,” Journal of International Development 14: 743-755. Copestake, J., Bhalotra, S., and Johnson, S., 2001, “Assessing the Impact of Microcredit: A Zambian Case Study” The Journal of Development Studies 37 (4): 81-100. Enterplan, 2003, “Towards s Strategy to Promote Rural and Microfinance,” FIRST Initiative, Lao PDR: Development of a Rural and Microfinance Strategy and Legal and Regulatory Framework, June, unpublished. Gaile, L.Gary and Jennifer Foster, 1996, Review of Methodological Approach to the Study of the Impact of Microenterprise Credit Programs, Washington, DC: Management System International. Gersovitz, Mark, 1983, “Savings and Nutrition at Low Incomes,” Journal of Political Economy 91(5): 841-855. Goetz, A.M., & Gupta, R.S. 1996, “Who Takes the Credit? Gender, Power, and Control Over Loan Use in Rural Credit Programs in Bangladesh,” World Development 24 (1): 45-63.

72

Greene, William H. 2000, Econometric Analysis, Upper Saddle River, New Jersey: Prentice-Hall Press. Hashemi, S.M., Schuler, S.R., Riley, A.P., 1996, “Rural Credit Programs and Women’s Empowerment in Bangladesh,” World Development 24 (4): 635-653. Heckman, James J. and Richard Robb, 1985, “Alternative Methods for Evaluating the Impact of Interventions: An Overview,” Journal of Econometrics 30: 239-386. Holvoet, Nathalie, 2004, “Impact of Microfinance Program on Children’s Education: Do the Gender of the Borrower and the Delivery Model Matter?” Journal of Microfinance 6(2): 27-49. Hulme David and Mosley Paul, 1996, Finance Against Poverty, 1rst and 2nd Vols. London: Routledge. Hulme, D., 2000, “Impact Assessment Methodologies for Microfinance: Theory, Experience and Better Practice,” World Development 28 (1): 79-98. Khandker, S.R., Samad, H.A., and Khan, Z.H. 1998, “Income and Employment Effects of Microfinance-credit Programmes: Village-level Evidence from Bangladesh,” The Journal of Development Studies 35 (2): 96-124. Khandker, Shahidur R, 2003, “Micro-Finance and Poverty: Evidence Using Panel Data from Bangladesh,” Policy Research Working Paper 2945, World Bank, Washington DC. Kunkel, C.R. and H.D. Seibel, 1997, Microfinance in Laos, Bangkok, Thailand: Asia Pacific Rural and Agricultural Credit Association.

73

Kyophilavong, Phouphet and Chansathith Chaleunsinh, 2005, “The Impact of the Savings Group and Its Service Issues in Laos,” Lao Journal of Economics and Business Management 1(3): 43-66. Ledgerwood, Joanna, 2002, Microfinance Handbook: An Institutional and Financial Perspective, Washington, D.C.: World Bank, Sustainable Banking with the Poor project. McKernan, Signe-Mary, 2002, “The Impact of Microcredit Programs on SelfEmployment Profits: Do Noncredit Program Aspects Matter?” The Review of Economics and Statistics 84(1): 93-115. Microfinance Capacity Building & Research Programme, 2005, Rural & Microfinance Statistics in Lao PDR 2004, Vientiane, Lao PDR: National Economic Research Institute, and Concern Worldwide Lao PDR. Moffitt, Robert, 1991, “Program Evaluation with Nonexperimental Data,” Evaluation Review, 15(3): 291-314. Montgomery, R., Debapriya Bhattacharya, and David Hulme, 1996, “Credit for The Poor in Bangladesh: The BRAC Rural Development Programme and The Government Thana Resource Development and Employment Programme,” chapter 12 in Finance Against Poverty. David Hulme and Paul Mosley, eds. London: Routledge. Morduch, J. 1998, “Does Microfinance Really Help the Poor? New Evidence from Flagship Programs in Bangladesh,” First complete draft. Retrieved June 3, 2006, from

http://www.wws.princeton.edu/~rpds/downloads/morduch_microfinance

_poor.pdf.

74

Mosley, Paul, 1997, “The Use of Control Groups in Impact Assessments for Microfinance,” Social Finance Unit Working Paper No. 19, The International Labour Office, Geneva. Mosley, Paul, 2001, “Microfinance and Poverty in Bolivia,” The Journal of Development Studies 37(4): 101-132. National Economic Research Institute, 2004, Rural & Microfinance Statistic in Laos 2003, Vientiane, Laos: National Economic Research Institute. National Growth and Poverty Eradication Strategy, 2004, Vientiane, Lao PDR., June. Pitt, M.M., and Khandker, S. R., 1996, “Household and Intrahousehold Impact of the Grameen Bank and Similar Targeted Credit Programs in Bangladesh,” World Bank Discusssion Papers 320, The World Bank, Washington DC. Pitt, M.M., Khandker, S.R., Mckernan, S.M., & Latif, M.A, 1999, “Credit Programs for The Poor and Reproductive Behavior in Low-Income Countries: Are the Reported Causal Relationships the Result of Heterogeneity Bias?” Demography 36 (1): 121. Pitt, Mark M., and Shahidur R. Khandker, 1998, “The Impact of Group-Based Credit Programs on Poor Households in Bangladesh: Does the Gender of Participants Matter?” Journal of Political Economy, 106 (5): 958-996. Rashid, Mansoora and Robert M. Townsend, 1994, “Targeting Credit and Insurance: Efficiency, Mechanism Design and Program Evaluation,” ESP Discussion Paper No. 47, The World Bank, Washington, DC. Ravallion, Martin, 2005, “Evaluating Anti-Poverty Programs,” Policy Research Working Paper 3625, World Bank, Washington DC.

75

Robinson, Marguerite S, 2001, The Microfinance Revolution, Sustainable Finance for the Poor, Washington, D.C.: World Bank. Schenk-Sandbergen, Loes and Outhaki Choulamany-Khamphoui, 1995, Women in Rice Field and Offices: Irrigation in Laos. Gender Specific Case Studies in Four Villages, Heiloo (The Netherlands): Empowerment. Shaw, J., 2004, “Microenterprise Occupation and Poverty Reduction in Microfinance Programs: Evidence from Sri Lanka,” World Development 32 (7): 1247-1264. Sichanthongthip, Chanhsana, 2004, Loan Repayment and Impact on Income in Lao Microfinance: Saithani Case, (Master thesis, Sophia University, 2004). Japan: Sophia University. Slover, Curtis, Jon Wynne Williams, Jose Garson, and Eric Duflos, 1997, Microfinance in Rural Lao PDR: A National Profile, United Nations Development Programme and United Nations Capital Development Fund, June. United Nations Capital Development Fund, 2001, Feasibility Study Mission Report: Microfinance & Sustainable Livelihood, UNDP/UNCDF LAO/96/CO1 – LAO/96/020, 25th January, Retrieve July 7, 2006, from United Nations Capital Development

Fund

Web

site:

http://www.uncdf.org/english/countries/laos/

microfinance/technical_review_reports/FEA.pdf. Wooldridge, Jeffrey M. 2003, Introductory Econometrics: A Modern Approach, 2nd ed., South-Western. World Bank Vientiane Office, 2006, “Lao PDR Economic Monitor,” Vientiane, Laos, The World Bank, April.

76

Yunus Muhammad, 2001, Banker to The Poor: The Autobiography of Muhamad Yunus, founder of the Grammeen Bank, Dhaka, Bangladesh: The University Press Limited.

77

Materials in Lao £¤¡¾−¦‰¤À¦ó´£¸¾´À¢˜´Á¢¤¦¿ìñ®Á´È¨ò¤Á콧÷´§ö−, £øÈ´õ¡È¼¸¡ñ®¡÷È´êɺ−À¤ò−, (²ò´£˜¤êó 01, −½£º−͸¤¸¼¤¥ñ−, ®Ò´ó¸ñ−êó), £¤¡¾−»È¸´´õì½¹¸È¾¤¦ø−¡¾−¦½¹½²ñ−Á´È¨ò¤ Áì½

¦½«¾®ñ−²ñ©ê½−¾§÷´§ö−

(¯½Àê©Äê).

[Women

and

Community’s

Empowering Project, Savings Group Manual, 1st ed., Vientiane, n.d, Cooperative Project

between

Lao

Women’s

Union

and

Community

Organizations

Development Institute (Thailand)]

£¤¡¾−¦‰¤À¦ó´£¸¾´À¢˜´Á¢¤¦¿ìñ®Á´È¨ò¤Á콧÷´§ö−,

쾨¤¾−¡¾−¥ñ©ª˜¤¯½ªò®ñ©Â£¤

¡¾−¦‰¤À¦ó´£¸¾´À¢˜´Á¢¤¦¿ìñ®Á´È¨ò¤Á콧÷´§ö−: ª÷ì¾ 2002 À«ò¤¡ñ−¨¾ 2005, (−½£º−͸¤¸¼¤¥ñ−, 2005), £¤¡¾−»È¸´´õì½¹¸È¾¤¦ø−¡¾−¦½¹½²ñ−Á´È¨ò¤ Áì½ ¦½«¾®ñ−²ñ©ê½−¾§÷´§ö− (¯½Àê©Äê). [Women and Community’s Empowering Project, Report on the Implementation of Women and Community’s Empowering Project: October 2002 - September 2005, Vientiane, 2005, Cooperative Project between Lao Women’s Union and Community Organizations Development Institute (Thailand)]

¦ø−¡¾¤¦½¹½²ñ−Á´È¨ò¤ì¾¸ ¦.¯.¯ 쾸 Áì½ ¦½«¾®ñ−²ñ©ê½−¾§÷´§ö− (¯½Àê©Äê), ¡÷È´êɺ−À¤ò− ¡ñ®¡¾−Á¡ÉÄ¢¯ñ−¹¾£¸¾´ê÷¡¨¾¡: 쾨¤¾−¡¾−¦‰¤À¦ó´ Áì½¢½¹¨¾¨ ¡÷ú´êɺ−À¤ò− Ã− ¦.¯.¯ 쾸, (−½£º−͸¤¸¼¤¥ñ−, ®Ò´ó¸ñ−êó), ¦ø−¡¾−¦½¹½²ñ−

78

Á´È¨ò¤ ¦.¯.¯ 쾸 Áì½ ¦½«¾®ñ−²ñ©ê½−¾§÷´§ö− (¯½Àê©Äê). [Lao Women’s Union: Lao PDR and Community Organizations Development Institute: Thailand, Savings Group and Poverty Reduction: Study on Promotion and Expansion of Savings Group in Lao PDR, Vientiane, n.d, Lao Women’s Union: Lao PDR and Community Organizations Development Institute: Thailand, (unpublished)]

79

Appendix A

Table 1: Descriptive statistics for variables of whole sample size

Variable descriptions Dependent variables (in Kip unless stated otherwise): Household monthly food expenditure Households monthly rental expenditure Household monthly transportation fee expenditure Household monthly educational expenditure Household monthly clothing expenditure Household monthly medical expenditure Household monthly expenditure on household utensils Household monthly other major expenditure Household monthly total non-food expenditure Household monthly total expenditure Value of household owned house Value of household owned land Value of household owned land and house Value of household owned agriculture asset Value of household owned livestock asset Value of household owned other enterprise asset Value of household owned savings at house Value of household owned other asset Value of household owned non-land and house asset Value of household owned total asset

Mean

Median

417,481 10,267 210,553

300,000 0.00 45,000

2,100,000 1,200,000 3,600,000

20,000 0 0

329,570 84,549 432,975

251 251 251

167,807 136,084 105,106 95,754

55,000 58,333 30,000 25,000

3,000,000 1,200,000 2,000,000 2,400,000

0 0 0 0

333,357 184,019 234,065 222,659

251 251 251 251

91,748 817,319

25,000 480,000

3,500,000 6,080,000

0 12,000

316,805 998,698

251 251

1,234,800 885,000 33,964,779 20,000,000 498,000,000 20,000,000 532,000,000 46,000,000 2,524,714 0 5,549,922 400,000 459,721 0

7,000,000 287,000,000 111,000,000,000 111,000,000,000 70,000,000 220,000,000 20,000,000

67,000 1,152,860 0 45,449,824 0 7,010,000,000 0 7,010,000,000 0 7,606,691 0 19,242,134 0 2,228,895

251 251 251 251 251 251 251

1,242,315 3,458,696 13,235,368

Maximum Minimum

200,000 0 4,100,000

107,000,000 190,000,000 338,000,000

545,000,000 55,270,000

111,000,000,000

80

0 0 0

Std. Dev. Observations

7,490,676 14,566,528 33,072,800

251 251 251

300,000 7,020,000,000

251

Appendix A

Table 1: Descriptive statistics for variables of whole sample size (continued)

Variable descriptions Dependent variables (in Kip unless stated otherwise): Household yearly self employment income from agriculture Household yearly self employment income from livestock Household yearly self employment income from handicraft & textile Household yearly self employment income from trading Household yearly self employment income from repairing& fixing service Household yearly self employment income from rice mill & construction Household yearly self employment income from vehicle service Household total yearly self employment income Household yearly wage & salary income Household yearly income from remittance Household yearly rental income Household yearly monetary items income Household yearly other income Household total yearly non-self employment income Household yearly total income

Mean

Median

Maximum Minimum

Std. Dev. Observations

2,419,044

1,000,000

81,500,000

0

5,979,001

251

1,578,175

300,000

30,000,000

0

3,486,485

251

1,888,122

100,000

25,400,000

0

3,243,976

251

3,376,932

0

194,000,000

0

18,386,628

251

379,482

0

21,900,000

0

2,459,819

251

1,648,562

0

190,000,000

0

15,460,301

251

301,793

0

40,000,000

0

2,944,454

251

11,592,111

5,475,000

196,000,000

0

25,294,651

251

2,205,797 407,822 79,084 887,139 21,514 3,601,356

0 0 0 0 0 500,000

24,000,000 19,068,600 10,800,000 213,000,000 3,600,000 213,000,000

0 0 0 0 0 0

3,799,600 1,704,379 738,221 13,461,641 253,644 13,979,828

251 251 251 251 251 251

15,193,467

8,200,000

213,000,000

380,000

28,634,965

251

81

Appendix A

Table 1: Descriptive statistics for variables of whole sample size (continued)

Variable descriptions Independent variables: Months as VSG member Does household has a VSG member?(0/1) Value of household owned land 5 years ago Sex of household head (female=1) Gender of respondents (entrepreneur) (female=1) Maximum education of respondents (entrepreneur) (years) Household size Age of respondents (entrepreneur) (years) Number of months doing business Number of Generations of family in village Number of relatives in village Are you member or head of the group committee? (0/1) Number of civil servant in household Number of wage employment in HH Number of school age in HH Number of dependent on your income in HH Village is near river (0/1) Village has pig pond which has water throughout the year (0/1) Village has either pig pond which has water throughout the year or be near river (0/1) Village is located in district capital (0/1)

Mean

Median

Maximum Minimum

Std. Dev. Observations

8.08

1

36

0

12

251

0.73 1 29,042,629 10,000,000 0.0996 0.00 0.888 1

1 911,000,000 1 1

0 0 0 0

0.45 82,223,525 0.30 0.32

251 251 251 251

4.90

5

19

0

3

251

5.51 41 161 0.502 2.42 0.064

5 41 120 0 1 0

12 75 813 6 40 1

1 18 1 0 0 0

2 12 146 1 4 0.24

251 251 251 251 251 251

0.267 1.08 1.57 2.64 0.434 0.375

0 1 2 2 0 0

4 5 6 10 1 1

0 0 0 0 0 0

0.56 1.19 1.26 1.67 0.50 0.48

251 251 251 251 251 251

0.566

1

1

0

0.50

251

0.096

0

1

0

0.29

251

82

Appendix A

Table 1: Descriptive statistics for variables of whole sample size (continued)

Variable descriptions Dependent variables (in Kip unless stated otherwise): Does village have paved road or near main road (Km 13 road)? (0/1) Village has irrigation (0/1) Does village have school until secondary level?(0/1) Does village have primary school up to grade5? (0/1) Distance from village to main markets Price of one cattle Price of one buffalo Price of pig per Kg Price of duck per Kg Price of traditional chicken (Gailard) per Kg Daily wage for harvesting rice Daily wage for planting rice Daily wage for construction

Mean

Median

0.287

0

1

0

0.45

251

0.191 0.096

0 0

1 1

0 0

0.39 0.29

251 251

0.661

1

1

0

0.47

251

22 1,610,359 3,260,956 14,596 14,572 18,857 18,307 27,875 24,243

20 1,500,000 3,250,000 11,667 15,000 19,000 20,000 35,000 25,000

33 2,000,000 4,000,000 25,000 15,000 20,000 30,000 50,000 27,500

15 1,300,000 2,500,000 10,000 13,500 18,000 0 0 20,000

6.84 233,007 524,051 5,097 617 896 9,339 18,081 2,043

251 251 251 251 251 251 251 251 251

Source: Author’s survey data, September 2005 and March 2006.

83

Maximum Minimum

Std. Dev. Observations

Appendix A

Table 2: Descriptive statistics for variables by treatment group

Variable descriptions Dependent variables (in Kip unless stated otherwise): Household monthly food expenditure Households monthly rental expenditure Household monthly transportation fee expenditure Household monthly educational expenditure Household monthly clothing expenditure Household monthly medical expenditure Household monthly expenditure on household utensils Household monthly other major expenditure Household monthly total non-food expenditure Household monthly total expenditure Value of household owned house Value of household owned land Value of household owned land and house Value of household owned agriculture asset Value of household owned livestock asset Value of household owned other enterprise asset Value of household owned savings at house Value of household owned other asset Value of household owned non-land and house asset Value of household owned total asset

Mean

Median

Maximum

Minimum

429,160 2,061 209,551

300,000 0 66,000

2,100,000 210,000 2,790,000

20,000 0 0

356,530 18,594 376,340

131 131 131

201,508 161,304 119,523 98,393

70,000 100,000 30,000 25,000

2,700,000 1,000,000 2,000,000 2,400,000

0 0 0 0

348,466 194,430 246,709 241,647

131 131 131 131

119,380 911,720

25,000 615,000

3,500,000 5,900,000

0 44,500

421,754 1,036,465

131 131

1,340,880 36,757,225 43,986,808 80,744,034 2,702,595 4,906,851 412,977

1,000,000 25,000,000 25,000,000 56,303,000 0 400,000 0

7,000,000 267,000,000 500,000,000 700,000,000 60,000,000 150,000,000 18,000,000

67,000 500,000 0 500,000 0 0 0

1,197,825 43,616,512 66,102,737 94,659,232 7,597,104 16,750,015 2,081,643

131 131 131 131 131 131 131

1,078,321 4,311,510 13,412,254

200,000 0 3,820,000

50,000,000 190,000,000 338,000,000

0 0 0

4,602,353 19,478,105 36,561,896

131 131 131

94,156,287

62,100,000

720,000,000

1,080,000

111,000,000

131

84

Std. Dev. Observations

Appendix A

Table 2: Descriptive statistics for variables by treatment group (continued)

Variable descriptions Dependent variables (in Kip unless stated otherwise): Household yearly self employment income from agriculture Household yearly self employment income from livestock Household yearly self employment income from handicraft & textile Household yearly self employment income from trading Household yearly self employment income from repairing& fixing service Household yearly self employment income from rice mill & construction Household yearly self employment income from vehicle service Household total yearly self employment income Household yearly wage & salary income Household yearly income on remittance Household yearly rental income Household yearly monetary items income Household yearly other income Household total yearly non-self employment income Household yearly total income

Mean

Median

Maximum

Minimum

2,537,023

700,000

81,500,000

0

7,571,050

131

1,466,794

500,000

15,000,000

0

2,435,613

131

2,134,983

500,000

25,400,000

0

3,686,305

131

2,664,695

0

194,000,000

0

17,334,078

131

552,672

0

21,900,000

0

2,944,155

131

1,703,733

0

150,000,000

0

13,617,852

131

427,481

0

40,000,000

0

3,754,035

131

11,487,380

6,160,000

194,000,000

0

23,500,748

131

2,327,977 448,352 102,290 45,802 27,481 2,951,902

0 0 0 0 0 400,000

19,720,000 19,068,600 10,800,000 6,000,000 3,600,000 26,068,600

0 0 0 0 0 0

3,842,135 1,945,576 954,863 524,222 314,534 4,678,700

131 131 131 131 131 131

14,439,282

8,520,000

194,000,000

1,000,000

23,505,613

131

85

Std. Dev. Observations

Appendix A

Table 2: Descriptive statistics for variables by treatment group (continued)

Variable descriptions Independent variables: Months as VSG member Does household has a VSG member?(0/1) Value of household owned land 5 years ago Sex of household head (female=1) Gender of respondents (entrepreneur) (female=1) Maximum education of respondents (entrepreneur) (years) Household size Age of respondents (entrepreneur) (years) Number of months doing business Number of Generations of family in village Number of relatives in village Are you member or head of the group committee? (0/1) Number of civil servant in household Number of wage employment in HH Number of school age in HH Number of dependent on your income in HH Village is near river (0/1) Village has pig pond which has water throughout the year(0/1) Village has either pig pond which has water throughout the year or be near river (0/1) Village is located in district capital (0/1)

Mean

Median

Maximum

Minimum

15 1 25,345,857 0.09 0.92

15 1 12,000,000 0.00 1.00

36 1 300,000,000 1.00 1.00

1 1 0 0.00 0.00

12.51 0.00 38,953,728 0.29 0.27

131 131 131 131 131

4.69

5.00

14.00

0.00

3.41

131

5.65 42.29 155.19 0.50 2.27 0.08

5.00 43.00 120.00 0.00 2.00 0.00

11.00 68.00 531.00 6.00 12.00 1.00

2.00 21.00 1.00 0.00 0.00 0.00

1.98 10.74 137.00 1.01 2.49 0.28

131 131 131 131 131 131

0.28 1.18 1.67 2.66 0.56 0.36

0.00 1.00 2.00 3.00 1.00 0.00

2.00 5.00 5.00 8.00 1.00 1.00

0.00 0.00 0.00 0.00 0.00 0.00

0.54 1.29 1.17 1.60 0.50 0.48

131 131 131 131 131 131

0.62

1.00

1.00

0.00

0.49

131

0.12

0.00

1.00

0.00

0.33

131

86

Std. Dev. Observations

Appendix A

Table 2: Descriptive statistics for variables by treatment group (continued)

Variable descriptions Dependent variables (in Kip unless stated otherwise): Does village have paved road or near main road (Km 13 road)? (0/1) Village has irrigation (0/1) Does village have school until secondary level? (0/1) Does village have primary school up to grade5? (0/1) Distance from village to main markets Price of one cattle Price of one buffalo Price of pig per Kg Price of duck per Kg Price of traditional chicken (Gailard) per Kg Daily wage for harvesting rice Daily wage for planting rice Daily wage for construction

Mean

Median

Maximum

Minimum

0.38

0.00

1.00

0.00

0.49

131

0.26 0.12

0.00 0.00

1.00 1.00

0.00 0.00

0.44 0.33

131 131

0.74

1.00

1.00

0.00

0.44

131

22.75 1,573,664 3,270,992 15,485 14,668 18,985 18,168 32,664 23,798

20.00 1,500,000 3,250,000 11,667 15,000 19,000 20,000 40,000 25,000

33.00 2,000,000 4,000,000 25,000 15,000 20,000 30,000 50,000 27,500

15.00 1,300,000 2,500,000 10,000 13,500 18,000 0 0 20,000

6.63 231,660 579,322 5,334 577 928 8,091 17,081 1,970

131 131 131 131 131 131 131 131 131

Source: Author’s survey data, September 2005 and March 2006.

87

Std. Dev. Observations

Appendix A

Table 3: Descriptive statistics for variables by control group

Variable descriptions Dependent variables (in Kip unless stated otherwise): Household monthly food expenditure Households monthly rental expenditure Household monthly transportation fee expenditure Household monthly educational expenditure Household monthly clothing expenditure Household monthly medical expenditure Household monthly expenditure on household utensils Household monthly other major expenditure Household monthly total non-food expenditure Household monthly total expenditure Value of household owned house Value of household owned land Value of household owned land and house Value of household owned agriculture asset Value of household owned livestock asset Value of household owned other enterprise asset Value of household owned savings at house Value of household owned other asset Value of household owned non-land and house asset Value of household owned total asset

Mean

Median

Maximum Minimum

435,058 39,396 311,404

300,000 0 50,000

1,640,000 1,200,000 3,600,000

20,000 0 0

354,763 180,419 671,649

52 52 52

129,619 95,556 88,875 104,131

73,334 45,834 30,000 25,000

870,000 510,000 800,000 1,000,000

0 0 0 0

182,782 110,758 166,766 205,913

52 52 52 52

79,058 848,039

30,000 477,223

700,000 5,050,000

0 44,167

143,501 1,004,497

52 52

225,000 1,238,047 100,000 52,921,605 0 15,400,000,000 100,000 15,400,000,000 0 10,603,436 0 30,604,352 0 2,342,297

52 52 52 52 52 52 52

1,283,097 890,000 6,010,000 33,009,317 15,000,000 287,000,000 2,210,000,000 15,000,000 111,000,000,000 2,240,000,000 40,750,000 111,000,000,000 3,156,410 0 70,000,000 8,011,365 500,000 220,000,000 726,731 0 15,000,000 574,519 1,887,392 14,356,418

195,000 0 4,950,000

5,000,000 23,944,385 244,000,000

2,260,000,000 52,800,000 111,000,000,000

88

0 0 0

Std. Dev. Observations

967,112 5,072,488 35,608,003

52 52 52

348,000 15,400,000,000

52

Appendix A

Table 3: Descriptive statistics for variables by control group (continued)

Variable descriptions Dependent variables (in Kip unless stated otherwise): Household yearly self employment income from agriculture Household yearly self employment income from livestock Household yearly self employment income from handicraft & textile Household yearly self employment income from trading Household yearly self employment income from repairing& fixing service Household yearly self employment income from rice mill & construction Household yearly self employment income from vehicle service Household total yearly self employment income Household yearly wage & salary income Household yearly income from remittance Household yearly rental income Household yearly monetary items income Household yearly other income Household total yearly non-self employment income Household yearly total income

Mean

Median

Maximum Minimum

Std. Dev. Observations

2,370,000

570,000

15,000,000

0

3,645,969

52

1,580,385

250,000

24,000,000

0

3,852,912

52

1,703,500

0

10,000,000

0

2,879,942

52

3,952,885

0

94,100,000

0

14,868,408

52

0

0

0

0

0

52

19,231

0

1,000,000

0

138,675

52

0

0

0

0

0

52

9,626,000

5,050,000

113,000,000

0

17,829,858

52

1,870,000 249,816 113,462 38,462 34,615 2,306,355

0 0 0 0 0 900,000

14,400,000 7,995,450 4,000,000 2,000,000 1,800,000 14,400,000

0 0 0 0 0 0

3,213,138 1,182,835 578,378 277,350 249,615 3,238,564

52 52 52 52 52 52

11,932,355

6,600,000

121,000,000

1,000,000

19,371,116

52

89

Appendix A

Table 3: Descriptive statistics for variables by control group (continued)

Variable descriptions Independent variables: Months as VSG member Does household has a VSG member?(0/1) Value of household owned land 5 years ago Sex of household head (female=1) Gender of respondents (entrepreneur) (female=1) Maximum education of respondents (entrepreneur) (years) Household size Age of respondents (entrepreneur) (years) Number of months doing business Number of Generations of family in village Number of relatives in village Are you member or head of the group committee? (0/1) Number of civil servant in household Number of wage employment in HH Number of school age in HH Number of dependent on your income in HH Village is near river (0/1) Village has pig pond which has water throughout the year (0/1) Village has either pig pond which has water throughout the year or be near river (0/1) Village is located in district capital (0/1)

Mean

Median

0 1 36,661,946 0.13 0.88

0 1 9,000,000 0.00 1.00

0 1 770,000,000 1.00 1.00

0 1 0 0.00 0.00

0 0 112,000,000 0.34 0.32

52 52 52 52 52

5.27

5.00

13.00

0.00

3.06

52

5.54 38.63 161.06 0.50 2.21 0.10

5.00 35.50 110.50 0.00 1.00 0.00

11.00 68.00 813.00 4.00 12.00 1.00

2.00 18.00 2.00 0.00 0.00 0.00

2.16 13.53 162.29 0.98 2.82 0.30

52 52 52 52 52 52

0.37 1.10 1.83 2.65 0.08 0.42

0.00 1.00 2.00 2.00 0.00 0.00

4.00 4.00 6.00 8.00 1.00 1.00

0.00 0.00 0.00 0.00 0.00 0.00

0.71 1.16 1.37 1.63 0.27 0.50

52 52 52 52 52 52

0.44

0.00

1.00

0.00

0.50

52

0.00

0.00

0.00

0.00

0.00

52

90

Maximum Minimum

Std. Dev. Observations

Appendix A

Table 3: Descriptive statistics for variables by control group (continued)

Variable descriptions Independent variables: Does village have paved road or near main road (Km 13 road)? Village has irrigation (0/1) Does village have school until secondary level? (0/1) Does village have primary school up to grade5? (0/1) Distance from village to main markets Price of one cattle Price of one buffalo Price of pig per Kg Price of duck per Kg Price of traditional chicken (Gailard) per Kg Daily wage for harvesting rice Daily wage for planting rice Daily wage for construction

Mean

Median

0.02

0.00

1.00

0.00

0.14

52

0.02 0.00

0.00 0.00

1.00 0.00

0.00 0.00

0.14 0.00

52 52

0.44

0.00

1.00

0.00

0.50

52

24 1,678,846 3,134,615 11,484 14,356 18,288 15,769 11,913 25,769

18 1,750,000 3,000,000 10,833 14,000 18,000 25,000 15,000 25,000

33 2,000,000 4,000,000 20,000 15,000 20,000 30,000 50,000 27,500

15 1,300,000 2,500,000 10,000 13,500 18,000 0 0 22,500

7.81 191,830 298,942 2,303 605 572 12,887 12,764 1,447

52 52 52 52 52 52 52 52 52

Source: Author’s survey data, September 2005 and March 2006.

91

Maximum Minimum

Std. Dev. Observations

Appendix A

Table 4: Descriptive statistics for variables by nonmember

Variable descriptions Dependent variables (in Kip unless stated otherwise): Household monthly food expenditure Households monthly rental expenditure Household monthly transportation fee expenditure Household monthly educational expenditure Household monthly clothing expenditure Household monthly medical expenditure Household monthly expenditure on household utensils Household monthly other major expenditure Household monthly total non-food expenditure Household monthly total expenditure Value of household owned house Value of household owned land Value of household owned land and house Value of household owned agriculture asset Value of household owned livestock asset Value of household owned other enterprise asset Value of household owned savings at house Value of household owned other asset Value of household owned non-land and house asset Value of household owned total asset Household yearly self employment income from agriculture

Mean

Median

Maximum Minimum

Std. Dev. Observations

381,539 3,799 135,363

300,000 0 7,500

1,140,000 150,000 1,500,000

30,000 0 0

247,036 19,816 260,151

68 68 68

132,086 118,490 89,745 84,265

32,500 50,000 24,584 25,000

3,000,000 1,200,000 2,000,000 1,200,000

0 0 0 0

385,963 202,077 253,594 198,019

68 68 68 68

48,221 611,968 993,507 29,315,860 66,232,290 95,548,151 1,698,971 4,906,500 345,588 2,068,912 3,017,360 12,037,331

20,000 600,000 358,833 6,080,000 702,500 6,230,000 19,000,000 267,000,000 10,000,000 2,420,000,000 30,000,000 2,450,000,000 100,000 30,000,000 200,000 65,500,000 0 20,000,000 150,000 107,000,000 0 23,656,050 4,825,000 131,000,000

0 100,743 12,000 899,020 129,000 962,150 0 42,974,388 0 294,000,000 0 299,000,000 0 4,155,598 0 10,876,800 0 2,424,959 0 12,902,080 0 5,697,910 0 22,803,727

68 68 68 68 68 68 68 68 68 68 68 68

108,000,000 2,229,265

41,215,000 2,460,000,000 1,500,000 19,000,000

300,000 303,000,000 0 3,462,084

68 68

92

Appendix A

Table 4: Descriptive statistics for variables by nonmember (continued)

Variable descriptions Dependent variables (in Kip unless stated otherwise): Household yearly self employment income from livestock Household yearly self employment income from handicraft & textile Household yearly self employment income from trading Household yearly self employment income from repairing& fixing service Household yearly self employment income from rice mill & construction Household yearly self employment income from vehicle service Household total yearly self employment income Household yearly wage & salary income Household yearly income from remittance Household yearly rental income Household yearly monetary items income Household yearly other income Household total yearly non-self employment income Household yearly total income Independent variables: Months as VSG member Does household has a VSG member?(0/1)

Mean

Median

Maximum Minimum

Std. Dev. Observations

1,791,059

175,000

30,000,000

0

4,740,913

68

1,553,735

0

11,000,000

0

2,515,442

68

4,308,603

0

182,000,000

0

22,539,720

68

336,029

0

19,200,000

0

2,363,542

68

2,788,235

0

190,000,000

0

22,992,377

68

290,441

0

18,250,000

0

2,214,526

68

13,297,368 2,227,206 450,569 8,088 3,156,941 0 5,842,804

5,000,000 0 0 0 0 0 450,000

196,000,000 24,000,000 10,000,000 400,000 213,000,000 0 213,000,000

0 0 0 0 0 0 0

32,677,643 4,153,388 1,550,881 51,551 25,853,716 0.00 25,914,373

68 68 68 68 68 68 68

19,140,172

8,100,000

213,000,000

380,000

40,947,644

68

0 0

0 0

0 0

0 0

0.00 0.00

68 68

93

Appendix A

Table 4: Descriptive statistics for variables by nonmember (continued)

Variable descriptions Independent variables Value of household owned land 5 years ago Sex of household head (female=1) Gender of respondents (entrepreneur) (female=1) Maximum education of respondents (entrepreneur) (years) Household size Age of respondents (entrepreneur) (years) Number of months doing business Number of Generations of family in village Number of relatives in village Are you member or head of the group committee? (0/1) Number of civil servant in household Number of wage employment in HH Number of school age in HH Number of dependent on your income in HH Village is near river (0/1) Village has pig pond which has water throughout the year (0/1) Village has either pig pond which has water throughout the year or be near river (0/1) Village is located in district capital (0/1) Does village have paved road or near main road (Km 13 road)? (0/1) Village has irrigation (0/1) Does village have school until secondary level? (0/1)

Mean

Median

Maximum Minimum

30,337,816 0.09 0.82 5.01

5,000,000 0.00 1.00 5.00

911,000,000 1.00 1.00 19.00

5.24 41 172 0.51 2.87 0.00

5.00 39 121 0.00 1.50 0.00

0.16 0.88 1.19 2.59 0.47 0.37

Std. Dev. Observations

0 113,000,000 0.00 0.29 0.00 0.38 0.00 3.92

68 68 68 68

12.00 75 624 6.00 40.00 0.00

1.00 19 1 0.00 0.00 0.00

2.25 13.69 152.41 1.13 5.27 0.00

68 68 68 68 68 68

0.00 1.00 1.00 2.00 0.00 0.00

2.00 4.00 6.00 10.00 1.00 1.00

0.00 0.00 0.00 0.00 0.00 0.00

0.44 0.99 1.28 1.85 0.50 0.49

68 68 68 68 68 68

0.56

1.00

1.00

0.00

0.50

68

0.12 0.31

0.00 0.00

1.00 1.00

0.00 0.00

0.32 0.47

68 68

0.19 0.12

0.00 0.00

1.00 1.00

0.00 0.00

0.40 0.32

68 68

94

Appendix A

Table 4: Descriptive statistics for variables by nonmember (continued)

Variable descriptions Independent variables Does village have primary school up to grade5? (0/1) Distance from village to main markets Price of one cattle Price of one buffalo Price of pig per Kg Price of duck per Kg Price of traditional chicken (Gailard) per Kg Daily wage for harvesting rice Daily wage for planting rice Daily wage for construction

Mean

Median

0.68

1.00

1.00

0.00

0.47

68

20.97 1,628,676 3,338,235 15,262 14,551 19,044 20,515 30,853 23,934

20.00 1,500,000 3,500,000 11,667 15,000 19,000 20,000 35,000 25,000

33.00 2,000,000 4,000,000 25,000 15,000 20,000 30,000 50,000 27,500

15.00 1,300,000 2,500,000 10,000 13,500 18,000 0 0 20,000

6.30 252,645 535,607 5,310 664 871 7,877 16,678 2,040

68 68 68 68 68 68 68 68 68

Source: Author’s survey data, September 2005 and March 2006.

95

Maximum Minimum

Std. Dev. Observations

Appendix A

Table 5: Impact of savings group on household house value - GLS.

Independent variable

Fixed Effects model Coefficient

Months as VSG member Does household have a VSG member?(0/1) Sex of household head (female=1) Gender of individuals (entrepreneur) (female=1) Maximum education of individuals (entrepreneur) (years) Household size Age of individuals (entrepreneur) (years)

Nonfixed effects model

“Naïve” model

“Super-naive” model

Coefficient

Std. Error

Coefficient

Std. Error

Coefficient

Std. Error

292,097**

Std. Error 147,784

292,097*

149,683

335,021***

122,254

358,260***

123,456

1,481,225

3,606,033

1,481,225

3,652,365

-8,395,409

6,494,524

-8,395,409

6,577,969

-8,192,279

6,446,844

-8,324,504

6,502,591

14,021,407**

6,760,261

14,021,407**

6,847,120

14,277,586**

6,841,554

10,800,449*

6,236,708

854,694**

422,396

854,694**

427,823

875,203**

4,152,112

912,698**

405,066

1,037,618 312,418**

752,418 136,471

1,037,618 312,418**

762,086 138,225

1,031,433 317,848**

761,675 136,362

1,141,685 347,865***

753,676 137,941

96

Appendix A

Table 5: Impact of savings group on household house value - GLS. (continued)

Independent variable

Fixed Effects model Coefficient

Number of months doing business Number of Generations of family in village Number of relatives in village Are you member or head of the group committee? (0/1) Number of civil servant in household Value of household owned land 5 years ago

Nonfixed effects model

“Naïve” model

“Super-naive” model

Coefficient

Std. Error

Coefficient

Std. Error

Coefficient

Std. Error

29,502***

Std. Error 11,177

29,502***

11,321

29,811***

11,347

33,317***

11,108

843,836

1,191,177

843,836

1,206,481

796,147

1,209,852

542,343

1,190,404

1,552,376***

387,633

1,552,376***

392,613

1,548,088***

392,040

1,544,427***

381,940

12,703,787*** 4,169,217

12,703,787*** 4,222,785

12,597,429***

4,225,101

12,995,561***

4,189,169

5,058,192

3,566,208

5,058,192

3,612,029

5,216,180

3,483,950

5,351,551

3,463,556

0.0688

0.0498

0.0688

0.0504

0.0697

0.0502

97

Appendix A

Table 5: Impact of savings group on household house value - GLS. (continued)

Independent variable

Fixed Effects model Coefficient

Village has either pig pond which has water throughout the year or be near river (0/1) Does village have paved road or near main road (Km 13 road)? (0/1) Does village have primary school up to grade5? (0/1) Distance from village to main markets Price of traditional chicken (Gailard) per Kg

Std. Error

Nonfixed effects model

“Naïve” model

“Super-naive” model

Coefficient

Std. Error

Coefficient

Std. Error

Coefficient

Std. Error

-44,768,692

36,726,790

-45,892,240

36,782,065

-47,125,621

36,182,599

-17,908,249** 7,655,365

-18,358,229**

7,784,479

-18,674,677**

7,906,479

45,390,649

38,241,483

46,212,441

38,344,587

47,538,868

3,764,1126

583,219

424,715

599,591

416,257

585,607

424,001

-258

1,783

-429

1,846

-171

1,841

98

Appendix A

Table 5: Impact of savings group on household house value - GLS. (continued)

Independent variable

Fixed Effects model Coefficient

Std. Error

Daily wage for construction F-statistic= 2.858314 Prob(F-statistic)= 0.000149 R-squared=0.181513

Nonfixed effects model

“Naïve” model

“Super-naive” model

Coefficient

Std. Error

Coefficient

Std. Error

Coefficient

Std. Error

-1,018

1,542

-886

1,613

-977

1,613

F-statistic=2.858314 Prob(F-statistic)= 0.000149

F-statistic=3.044732 Prob(F-statistic)= 0.000080

F-statistic=2.911104 Prob(F-statistic)= 0.000217

R-squared=0.181513

R-squared=0.181768

R-squared=0.166006

Note: the superscripts ***, ** and * denote rejection at 1 per cent, 5 per cent and 10 per cent critical values. Source: Author’s survey data, September 2005 and March 2006.

99

Appendix A

Table 6: Impact of savings group on yearly self-employment income from livestock - GLS.

Independent variable

Fixed Effects model Coefficient

Months as VSG member Does household have a VSG member?(0/1) Sex of household head (female=1) Gender of individuals (entrepreneur) (female=1) Maximum education of individuals (entrepreneur) (years) Household size Age of individuals (entrepreneur) (years) Number of months doing business Number of Generations of family in village Number of relatives in village

Nonfixed effects model Coefficient

19,962*

Std. Error 10,589

19,962*

Std. Error 10,725

-557,276

358,369

-557,276

362,974

-300,623

207,914

-300,623

80,203

845,785

12,190

“Naïve” model

5,374

Std. Error 6,546

5,948

Std. Error 6,388

210,585

-368,213*

204,348

-371,531*

202,203

80,203

856,652

-6,467

855,017

-84,635

861,953

33,842

12,190

34,277

2,017

33,190

4,484

33,169

318,013*** 4,370

97,539 10,151

318,013*** 4,370

98,792 10,281

306,585*** 1,298

95,545 9,339

310,764*** 2,549

95,568 9,296

-809

818

-809

828

-761

806

-689

792

107,168

83,641

107,168

84,716

110,511

78,823

110,934

78,738

46,844

32,868

46,844

33,290

48,340

31,638

49,922

32,272

100

Coefficient

“Super-naive” model Coefficient

Appendix A

Table 6: Impact of savings group on yearly self-employment income from livestock – GLS (continued)

Independent variable

Fixed Effects model Coefficient

Are you member or head of the group committee? (0/1) Number of civil servant in household Value of household owned land 5 years ago Village has either pig pond which has water throughout the year or be near river (0/1) Does village have paved road or near main road (Km 13 road)? (0/1) Does village have primary school grade5? (0/1) Distance from village to main markets Price of traditional chicken (Gailard) per Kg

1,378,544***

Std. Error 483,964

413,843.2* 0.0023

Nonfixed effects model Coefficient

“Naïve” model Coefficient

1,378,544***

Std. Error 490,183

249,040.7

413,843.2

0.0029

“Super-naive” model Coefficient

1,413,605***

Std. Error 473,251

1,445,757***

Std. Error 479,883

252,240.5

389,296

246,407

395,150

244,608

0.0023

0.0029

0.0023

0.0027

-3,014,492

2,745,199

-2,626,708

2,695,507

-2,669,113

2,678,944

-654,752

684,148

-481,329

656,940

-493,312

659,751

2,453,660

2,861,017

2,144,790

2,823,562

2,186,486

2,806,690

-16,485

54,007

-21,839

53,026

-22,598

52,962

-99.92

117

-39

99

-31.38

99.5

101

Appendix A

Table 6: Impact of savings group on yearly self-employment income from livestock – GLS (continued)

Independent variable

Fixed Effects model Coefficient

Std. Error

Daily wage for construction

Nonfixed effects model Coefficient 83

F-statistic= 2.296443 Prob(F-statistic)= 0.002577 R-squared=0.151228

Std. Error 69

F-statistic=2.296443 Prob(F-statistic)= 0.002577 R-squared=0.151228

“Naïve” model Coefficient 39

Std. Error 56

F-statistic=2.281757 Prob(F-statistic)= 0.003352 R-squared=0.142720

Note: the superscripts ***, ** and * denote rejection at 1 per cent, 5 per cent and 10 per cent critical values. Source: Author’s survey data, September 2005 and March 2006.

102

“Super-naive” model Coefficient 35

Std. Error 56

F-statistic=2.388660 Prob(F-statistic)= 0.002513 R-squared=0.140397

Appendix A

Table 7: Impact of savings group on household yearly self-employment income from agriculture - GLS.

Independent variable

Fixed Effects model Coefficient

Months as VSG member Does household have a VSG member?(0/1) Sex of household head (female=1) Gender of individuals (entrepreneur) (female=1) Maximum education of individuals (entrepreneur) (years) Household size Age of individuals (entrepreneur) (years) Number of months doing business Number of Generations of family in village Number of relatives in village

Nonfixed effects model Coefficient

-28,206.57***

Std. Error 9,286.57

-28,206.57***

Std. Error 9,405.89

286,830.8*

163,582.4

286,830.8*

165,684.2

-541,902***

108,493

-541,902***

72,762

129,788

-41,017.54**

“Naïve” model

-19,530.14***

Std. Error 7,505.15

Coefficient Std. Error *** -20,080 7,381.74

109,887

-512,509***

107,429.6

-486,574*** 107,728

72,762

131,455.9

103,554.6

128,713

74,531

127,055

19,608.96

-41,017.54**

19,860.9

-39,569.5**

19,779.25

-28,725.78

17,799

74,184.43*** 6,792.35

25,956.19 4,806.29

74,184.43*** 6,792.35

26,289.69 4,868.04

74,259*** 5,987.76

25,928 4,768.41

80,356*** 7,609*

25,993 4,585

2,310.43***

573.26

2,310.43***

580.63

2,347.08***

579.52

2,375.99***

572.196

137,675.9**

58,040.65

137,675.9**

58,786.38

138,985**

57,809

127,145.7** 56,916

11,847.96

21,677.13

11,847.96

21,955.64

11,319.47

22,153.05

11,529.7

103

Coefficient

“Super-naive” model

22,208.8

Appendix A

Table 7: Impact of savings group on household yearly self-employment income from agriculture – GLS (continued)

Independent variable

Fixed Effects model Coefficient

Are you member or head of the group committee? (0/1) Number of civil servant in household Value of household owned land 5 years ago Village has either pig pond which has water throughout the year or be near river (0/1) Does village have paved road or near main road (Km 13 road)? (0/1) Does village have primary school up to grade5? (0/1) Distance from village to main markets Price of traditional chicken (Gailard) per Kg

Nonfixed effects model Coefficient

536,173.8

Std. Error 563,871

82,783.2 0.001251***

“Naïve” model Coefficient

536,173.8

Std. Error 571,115.9

118,974.4

82,783

0.000365

“Super-naive” model

506,895

Std. Error 569,192.6

Coefficient Std. Error 507,812.4 567,530

120,503

107,797.9

115,761.5

158,546

105,499

0.001251***

0.00037

0.001144***

0.000367

-1,348,716

5,157,754

-1,576,873

5,166,868

-1,521,957

5,156,553

-1,464,704

3,444,240

-1,550,032

3,444,890

-1,536,444

3,437,276

4,407,984

8,395,579

4,576,076

8,394,287

4,510,155

8,377,608

-118,734**

59,639.55

-116,338**

59,163.12

-116,370**

59,044.5

-697.26

998.98

-724.56

999.12

-720.7

996.85

104

Appendix A

Table 7: Impact of savings group on household yearly self-employment income from agriculture – GLS (continued)

Independent variable

Fixed Effects model Coefficient

Std. Error

Daily wage for construction

Nonfixed effects model Coefficient 637.56

F-statistic= 1.492664 Prob(F-statistic)= 0.093412 R-squared= 0.103790

Std. Error 792.33

F-statistic= 1.492664 Prob(F-statistic)= 0.093412 R-squared= 0.103790

“Naïve” model Coefficient 663.61

Std. Error 793.01

F-statistic=1.537622 Prob(F-statistic)= 0.082866 R-squared=0.100871

Note: the superscripts ***, ** and * denote rejection at 1 per cent, 5 per cent and 10 per cent critical values. Source: Author’s survey data, September 2005 and March 2006.

105

“Super-naive” model Coefficient Std. Error 656.64 791.38 F-statistic=1.607735 Prob(F-statistic)= 0.067655 R-squared=0.099043

Appendix A

Table 8: Impact of savings group on household monthly rental expenditure - GLS.

Independent variable Months as VSG member Does household has a VSG member?(0/1) Sex of household head (female=1) Gender of individuals (entrepreneur) (female=1) Maximum education of individuals (entrepreneur) (years) Household size Age of individuals (entrepreneur) (years) Number of months doing business Number of Generations of family in village Number of relatives in village

Fixed Effects model

Nonfixed effects model

Coefficient

Std. Error

Coefficient

138.904**

60.75

138.904**

Std. Error 61.53

175.558

1,355

175.558

1,372

2,153

1,313

2,153

514

1,036

483.38**

“Naïve” model

“Super-naive” model

Coefficient

Std. Error

Coefficient

144.28**

45.69

144.421***

Std. Error 45.613

1,330

2,127.22

1,358.16

2,120.69

1,352.03

513.95

1,049.55

460.495

1,024.501

367.014

1,024.34

204.63

483.38**

207.25

488.83**

207.36

498.64**

208.93

-164 -342.71***

439.83 124.83

-164 -342.71***

445.48 126.43

-163.57 -345.779***

439.61 123.026

-155 -343.272***

439.1 121.85

-9.44**

4.09

-9.44**

4.14

-9.41**

4.008

-9.364**

3.985

8,346***

2,666

8,346***

2,700.43

8,417.66***

2,626.15

8,413.08***

2,616.92

-325**

134

-325**

135.42

-335.64***

131.47

-333.027***

130.446

106

Appendix A

Table 8: Impact of savings group on household monthly rental expenditure – GLS (continued)

Independent variable Are you member or head of the group committee? (0/1) Number of civil servant in household Value of household owned land 5 years ago Village has either pig pond which has water throughout the year or be near river (0/1) Does village have paved road or near main road (Km 13 road)? (0/1) Does village have primary school up to grade5? (0/1) Distance from village to main markets Price of traditional chicken (Gailard) per Kg

Fixed Effects model

Nonfixed effects model

Coefficient

Std. Error

Coefficient

-2,927**

1,212

-1,736** 0.0000027

“Naïve” model

“Super-naive” model

Coefficient

Std. Error

Coefficient

-2,927**

Std. Error 1,228.03

-2,959.98**

1,217.54

-2,912.395**

Std. Error 1,204

891

-1,736**

902.81

-1,734.42**

837.19

-1,684.74**

824.22

0.0000035

0.0000027

0.0000036

0.0000027

0.0000035

-46,764.79

83,436.6

-46,847.70

84,100.69

-46,844.96

83,916.4

-30,659.31

55,918.17

-30,707.13

56,154.46

-30,713.54

56,031.4

78,514.40

137,373.4

78,573.59

137,835.7

78,567.17

137,533.2

-898.635

942.369

-897.67

932.11

-898.15

930.08

-9.975

16.173

-9.984

16.258

-9.98

16.22

107

Appendix A

Table 8: Impact of savings group on household monthly rental expenditure – GLS (continued)

Independent variable

Fixed Effects model Coefficient

Std. Error

Daily wage for construction

Nonfixed effects model Coefficient 8.714

F-statistic= 2.851135 Prob(F-statistic)= 0.000155 R-squared=0.181139

Std. Error 12.837

F-statistic=2.851135 Prob(F-statistic)= 0.000155 R-squared=0.181139

“Naïve” model Coefficient

Std. Error

Coefficient

8.730

12.911

8.72

F-statistic=3.099860 Prob(F-statistic)= 0.000060 R-squared=0.184452

Note: the superscripts ***, ** and * denote rejection at 1 per cent, 5 per cent and 10 per cent critical values. Source: Author’s survey data, September 2005 and March 2006.

108

“Super-naive” model Std. Error 12.88

F-statistic=3.309186 Prob(F-statistic)= 0.000032 R-squared=0.184518

Appendix A

Table 9: Impact of savings group on household monthly educational expenditure - GLS.

Independent variable Months as VSG member Does household has a VSG member?(0/1) Sex of household head (female=1) Gender of individuals (entrepreneur) (female=1) Maximum education of individuals (entrepreneur) (years) Household size Age of individuals (entrepreneur) (years) Number of months doing business Number of Generations of family in village

Fixed Effects model

Nonfixed effects model

Coefficient

Std. Error

Coefficient

5,670.26**

2,768.37

5,670.26**

Std. Error 2,803.94

7,079.72

29,520.13

7,079.72

29,899.42

-73,346.61

49,733.75

-73,346.61

-21,040.16

29,964.27

6,287.33

“Naïve” model

“Super-naive” model

Coefficient

Std. Error

Coefficient

5,961.41**

2,328.64

5,975.62**

Std. Error 2,323.98

50,372.75

-71,354.07

49,884.70

-71,423.84

49,472.92

-21,040.16

30,349.26

-20,434.61

28,795.59

-16,808.54

28,283.16

3,930.84

6,287.33

3,981.35

6,318.587

3,875.395

6,305.82

3,879.46

20,529.96*** 1,592.31

5,012.84 1,142.47

20,529.96*** 1,592.31

5,077.25 1,157.15

20,845.82*** 1,550.24

4,978.39 1,129.29

20,763.81*** 1,464.46

4,937.83 1,108.08

-210.32**

97.97

-210.32**

99.23

-201.72**

100.08

-207.34**

97.26

-19,403.54**

8,778.61

-19,403.54**

8,891.41

-19,900.6**

8,926.37

-19,117.51**

8,493.06

109

Appendix A

Table 9: Impact of savings group on household monthly educational expenditure – GLS (continued)

Independent variable Number of relatives in village Are you member or head of the group committee? (0/1) Number of civil servant in household Value of household owned land 5 years ago Village has either pig pond which has water throughout the year or be near river (0/1) Does village have paved road or near main road (Km 13 road)? (0/1) Does village have primary school up to grade5? (0/1) Distance from village to main markets

Fixed Effects model

Nonfixed effects model

Coefficient

Std. Error

Coefficient

-135.896

1,292.505

-19,447.98

“Naïve” model

“Super-naive” model

Coefficient

Std. Error

Coefficient

-135.897

Std. Error 1,309.112

-186.27

1,232.45

-242.82

Std. Error 1,241.64

76,584.68

-19,447.98

77,568.68

-23,655.88

7,3921.90

-24,936.75

73,712.6

11,323.64

21,117.49

11,323.64

21,388.81

12,310.31

19,502.08

11,545.31

19,684.12

-0.0000758

0.0000979

-0.0000758

0.0000992

-0.0000769

0.0000914

-321,772.9*

166,552.9

-328,547.7**

155,380.5

-327,368.3**

154,914.8

-142,432.3

98,762.85

-144,617.5

93,385.55

-144,249.4

93,205.14

408,506*

223,862.9

413,289.4*

214,578.9

411,673.4*

214,079.6

1,380.33

1,975.93

1,476.9

2,101.4

1,454.18

2,092.53

110

Appendix A

Table 9: Impact of savings group on household monthly educational expenditure – GLS (continued)

Independent variable

Fixed Effects model Coefficient

Std. Error

Price of traditional chicken (Gailard) per Kg Daily wage for construction F-statistic= 1.670459 Prob(F-statistic)= 0.045622 R-squared=0.114734

Nonfixed effects model Coefficient

“Naïve” model

“Super-naive” model

Coefficient

Std. Error

Coefficient

-27.08

Std. Error 26.85

-28.12

25.11

-28.35

Std. Error 25.05

16.933

21.342

17.75

19.87

17.94

19.82

F-statistic=1.670459 Prob(F-statistic)= 0.045622 R-squared=0.114734

F-statistic= 1.835944 Prob(F-statistic)= 0.024821 R-squared=0.118129

Note: the superscripts ***, ** and * denote rejection at 1 per cent, 5 per cent and 10 per cent critical values. Source: Author’s survey data, September 2005 and March 2006.

111

F-statistic=1.952906 Prob(F-statistic)= 0.017032 R-squared=0.117802

Appendix B CASE STUDY SAVINGS GROUPS IN NAXAITHONG CITY

This section discusses the savings groups in Naxaithong city as a case study under the Women and Community’s Empowering Project (WCEP), one of the many programs which launched microfinance programs in the semi-urban areas of Laos. Before going to that point, we will briefly provide an overview of WCEP.

1.

Overview of the Women and Community’s Empowering Project38 1.1

Background, goal and activities

The Women and Community’s Empowering Project (WCEP), formed formally in October 2002, is cooperation between the Lao Women’s Union (LWU) – a Lao Government women's organization – and the Community Organizations Development Institute: Thailand (CODI). The project was implemented in three districts in Vientiane, the capital of Laos, namely Pak Ngum, Naxaithong and Sangthong cities which were selected by LWU because they were the poorest districts (Asian Coalition for Housing Rights, 2005), for the period from October 2002 to September 2005 with budget of 2,650,000 Baht39, supported by the Asian Coalition for Housing Rights (ACHR). The goal of the project is to improve quality of life for women and their families, and to support the activities for savings, credit and community fund to be a method for development. Its main activities have been: 38

This section borrows extensively from the Report on the Implementation of Women and Community’s Empowering Project: October 2002 - September 2005, produced by the Women and Community’s Empowering Project (2005). This report which was originally written in Lao language was translated by the author. 39 In 2002, average exchange rate was 42.96 Baht per one US dollar (Bank of Thailand, 2006).

112

Appendix B •

to establish the savings group and working capital fund;

•

to promote community activities with learning promotion, training and procession provision for community to be able to create activities for self development such as community development, community welfare provision, use of natural fertilizer, sharing of experience;

•

to improve the functioning of the savings group and its branches by developing accounting system, and developing administration and management system for savings group, branch, district fund and central fund;

•

to provide advice on self-employment opportunities as well as provide training to develop marketing and other related skills

1.2

Savings group 1.2.1 Points of view

Savings groups are still a new development in Laos. There are various definitions of a savings group. According to Chaleunsinh defined a savings group as:

A group of people who share the same voluntary spirit to help each other in solving problems on financial liquidity in order to improve solidarity and upgrade the living standard of people within the village. Basically, the savings group is an association which accumulates savings of the group’s members40 and loans them to members who need to borrow them and charging them a loan interest rate which must be higher than the savings’ rate of return. In addition, some strong savings groups also provide member support services, such as vocational training and establishing production group based on the potential of their members (Chaleunsinh, 2004: 4-5).

40

That must be equal to or higher than the minimum savings the group has established.

113

Appendix B In addition, according to the savings group manual of the Women and Community’s Empowering Project (n.d, 3), the savings group has to have its own organizational structure, which consists of group committee 41 , member, regulation, and savings group consultant, received promotion from government authority in each level and other related parties including benchmark village. More details on the organizational structure will be discussed in following sections. In the next three sections, most of the information is derived from the savings group manual of the Women and Community’s Empowering Project (n.d)42.

1.2.2 Objectives According to the savings group manual of the Women and Community’s Empowering Project (n.d), the main objectives of the savings group are listed down as the following: •

To reduce poverty and improve living status of women and their families;

•

To accumulate savings for village working capital fund;

•

To help each other and exchange knowledge and experiences of economic activities among members;

•

To strengthen women solidarity in a village;

•

To improve knowledge and capacity of women such as management, accounting, credit and banking;

41

Normally, the group committee at village level includes 5 people who are villagers (generally being membership of Lao women union at village level) in the village. They take caring every thing in the village savings group such as doing savings account, loan account, cash book, income and expenses account, and general ledger. 42 This savings group manual which was originally written in Lao language was translated by the author.

114

Appendix B •

To sustain living status of members and villagers and to have good welfare step by step;

•

To be one of the implementations of economic and social development plan which the Lao government issues in each period.

1.2.3 Establishment process of savings group The steps for establishing a savings group are described as bellows: Step 1: The project proponents meet local authorities at village level to discuss the possibility of savings group establishment in a village. Then, these authorities will consult with their villagers. Step 2: The local authorities at village level submit a list of the names who self select to be membership of a savings group43; to the promoter (hence may be the Lao Women Union at district level or the Project). Then, a meeting is held to provide information about policies of the group and to set up the savings group by: •

Election for group committee and group consultant;

•

Setting the roles for group committee, group consultant, and member;

•

Dividing the responsibilities for both group committee and group consultant;

•

Managing savings group with respect to the rules settled.

Step 3: Group committee and group consultant accept the first set of savings group members. The members have to pay their membership fee and deposit money with

43

The members included in the list should not be less than 20 people (Lao Women Union: Lao PDR and Community Organizations Development Institute: Thailand, n.d: 17).

115

Appendix B the savings group at the deposit day which has been set out. The group committee summarizes a financial statement after receiving the member savings. Step 4: Operations of the savings group’s activities, such as savings and credit activities, have to follow the rules. The group committee can receive a technical assistant from the promoter (the project) if the committee do not understand how to operate the activities, during implementation period. Step 5: Instead of savings activities, members can think about how to create activities, for example activities for earning extra income, welfare activity, environment and others, with respect to suitability and capacity of that local area. Step 6: The promoter of the project, the government authority at each level and other related organization have to follow up and promote and support a savings group to reach the development objective of poverty reduction.

1.2.4 Organizational structure of savings group The organizational structure of the savings group consists of group committee, savings group members, group consultant, the promoter of the project, the government authority and other related organizations. A.

Savings group member: A person who is selected and accepted to be a

member by the group committee, has attributes as the following: •

Self selection to be membership;

•

Harmonious, helpful, kindness;

•

Acceptable and agreeable to follow the savings group regulations;

•

Rational and acceptable to the majority of others;

116

Appendix B

B.

•

No gender and age discrimination;

•

Be a resident of the village.

Group committee: A person who received high scores from the members

by election or voting, has attributes like: •

Being a membership of savings group and self selection to be group committee;

C.

•

Trusted by others, knowledgeable, skillful and leadership experience;

•

Satisfied to do the job for others, honest, faithful, patient;

•

Healthy

•

Good attitude;

•

No discrimination on social class and ethic;

•

Accepted by majority of members.

Group consultant: A person who is a representative of the local authority

at the village level or a high respected person in the village or a teacher, is self selected and accepted by most of the members. D.

Savings group regulations: The creation of regulation with agreement of

members and group committee is based on particularity, suitability and capacity of a village or local. Therefore, there are some different and similar articles in regulations among different savings groups. However, the summary of savings group regulation should have different articles as: •

Membership criteria;

•

Minimum savings per day/per week/ per month.

•

Setting date, location and time of deposit;

117

Appendix B •

Setting membership fee;

•

Setting rules for members who save irregularly

•

Conditions for savings and dropping out from the group;

•

Setting loan interest rate for members such as loan for emergency sick; and normal loans (for doing rice field, crop plant, livestock, education and other);

•

Setting conditions for member borrowing such as condition and right of borrowing; and term of loan repayment;

•

Allocating benefit from total revenue at the year ended such as:

o Dividend payment to members 44 (based on calculation rate of member deposit or share ratio);

o Administration budget such as wage for group committee and group consultant; savings group expenditure which is savings group reserve;

o Village development fund; o Welfare fund; o Other fund which is beneficial to village, agreed by member and group committee; •

Setting aging or period of group committee and group consultant;

•

Regulation can be improved and changed with respect to agreement and suitability of most members.

44

Normally, dividend paid to the member is come from 70% of savings group profit and the rest of the profit (30%) is allocated to different categories of fund and reserve (for particular rate is depended on practice of each savings group), according to author’s survey data (September, 2005).

118

Appendix B 1.2.5 Results of the project implementation on savings group During the period of the project, savings groups have been established in number of villages in three target districts. It can be seen from table 1 that there were many villages demanding to voluntarily establish a savings group in own village. As a result, there were 24 more savings groups than the number planned. Table 2 shows that about 84% of the villages in the three pilot districts have their own savings groups. It means that savings groups have good outreach to provide financial services for villagers. Table 1: Number of savings groups established during the period from

October 2002 to September 2005 Areas

Number of Savings Group Planned number Actual number Variation Pak Ngum district 17 25 8 Naxaithong district 31 37 6 Sangthong district 17 27 10 Total 65 89 24 Source: Women and Community’s Empowering Project (2005: 42) Table 2: Number of savings group in three districts compared to number of

villages as the statistic of July 2005 Districts

Number of Villages

Savings Groups Number45

Number of member

Number of families

Pak Ngum district 53 55 5,201 3,963 Naxaithong district 56 38 6,128 3,877 Sangthong district 37 30 2,704 2,164 Total 146 123 14,033 10,004 Source: Women and Community’s Empowering Project (2005: 43)

45

Deposit balance in Kip 3,616,44,500 1,674,802,000 734,514,500 6,025,761,000

The number in this column is higher than the figures in third column of table 1 because it included the number of savings groups implemented before October 2002 as Pak Ngum district with 30 savings groups, Naxaythong district with one savings group, and Saengthong district with three savings groups (Women and Community’s Empowering Project, 2005).

119

Appendix B Savings groups in the three districts provide different kinds of loans to their members. The largest loan balance is loans for planting rice field in all three districts, followed by credit for trade and other services. Table 3: Loan balance for the savings groups in each district as April 2005 No.

1 2 3 4 5 6 7

Types of loan Crop plant Textile46 Trade and other services Emergency sick Sick Emergency Rice field planting

Total

Pak Ngum district No. of borrowers 230 387 273

Loan balance

Naxaithong district

Sangthong district

Loan balance 94,700,000 9,080,000 174,900,000

No. of borrowers 183 420 567

Loan balance

314,638,000 212,360,000 1,304,487,000

No. of borrowers 151 27 144

327

181,916,000

-

-

-

-

1,747

1,570,942,300

95 34 700

32,800,000 76,830,000 260,749,500

63 75 1,016

30,750,000 58,340,000 619,937,000

1,151 649,059,500

2,324

1,560,217,000

2,964 3,584,343,300 Note: “-” means no balance.

141,550,000 253,260,000 456,380,000

Source: Women and Community’s Empowering Project (2005: 45-46)

However, this paper will focus only on savings groups in Naxaithong district as the case study due to limitation of time and budget. The following section will discuss more details about the case study.

2.

Savings groups in Naxaithong district 2.1

Background

Naxaithong city, located in the area of Vientiane, the capital of Laos, is about 16 Km from the capital. A total of 56 villages in Naxaithong district are divided to six zones. 46

In Naxaythong district, this type of loan is provided borrowers, who do textile of Lao skirt, called Tam Hook.

120

Appendix B Its population is 57,129 people consisting of 28,761 females and there are 10,378 families (Women and Community’s Empowering Project, 2005: 40-41). The main economic activities in this district are planting rice fields, crop plants, livestock, textile and trade. According to the Women and Community’s Empowering Project (2005: 41), the living status of population in this district was poor before the establishment of the savings group. This could be seen from low income levels, lack of capital for production, and lack of finance for smoothing consumption and other activities. This led to the selling of Green Rice47 and borrowing money from the informal sector such as money lenders at high monthly interest rates of 10 percent to 30 percent. For example, in the 38 villages which have savings groups, 423 families sell green rice; 1,202 families obtained a loan from money lender; and 57 families were classified as being of a poor status before the establishment of the savings groups. In addition, in the absence of the savings group, the education level of the people was limited due to a lack of learning opportunities, especially for women and lack of knowledge and ability for management of different activities. In June 2001, a savings group was implemented in Naxaythong district. Now, there are 38 village savings groups available in four zones of this district as shown in the table 2.

2.2

Financial services of savings groups in Naxaithong District

Savings groups provided some basic financial services such as credit and savings to their members. The following section discusses the findings of the survey conducted in 47

As Chanleunsinh (2004:8) defined as, “Green rice selling or Khai Khao Khiew means farmers sell rice at a discounted price, which is evaluated by the middleman or the traders, before it is harvested. The traders/middle men then come to take the rice after it is harvested”.

121

Appendix B six of these villages. Six villages which the author did field survey in September 2005 and March 2006.

2.2.1 Source of fund The two main sources of funds for running savings group are internal funds and external funds48 . Internal funds come from deposit amounts from member of savings group (see table 4). At beginning of each month, normally at the first day of the month, the members have to deposit money to savings groups of at least the minimum required amount49 which is one of the membership rules. There is no interest paid on such deposits but members receive dividends every twelve months50. 70 per cent of the savings group profit is paid to each member as a dividend. The dividend payment to each member is based on share ratio or proportion of deposit of each member to total deposit balance of whole savings group. This implies that deposit amount per time from members who require more dividend tends to be increased every month. External funds are a soft loan from the central funds which are managed by the management committee of the project. This loan is lent to a savings group the funds of which are not sufficient fund for borrowers. The process of making a soft loan request

48

Besides the two main sources of fund, interest income from loan to members is one source of fund for savings group. 49 Minimum deposit amount is normally 5,000 Kip per month or equivalent to 0.46US dollar as exchange rate of 10,890 Kip per a dollar in September 2005, according to six savings groups of this case study. Until September 2005, there has been no limitation for maximum deposit amount in those six savings group. However, there is limitation on maximum deposit, up to one million kip per time, in case of savings group in Nakountay village at the time of follow up survey on 2nd March 2006. 50 (a)If members withdraw deposit before dividend payment date, they will not receive any dividend from their deposit accumulation. (b) If they withdraw at dividend payment date, they will still get dividend from their deposit, but if they need a sequent loan, they cannot withdraw their deposit. (c) They can withdraw only at the day of withdrawal, not the deposit day, (The deposit day is the day to make a plan to request for withdrawal).

122

Appendix B starts with members of savings group making borrowing plans to their savings groups. The plans are submitted to the committee at zone level, through the fund committee at district level to the central fund. The central fund committee issues the soft loan to the fund committee at district level. Then, the committee at district level will lend to the members of different groups who make borrowing plans with approval of the fund committee at district level (Women and Community’s Empowering Project, 2005: 63-64).

Table 4: Credit balance and source of fund in six savings groups as

September 2005

No. 1

Villages Nakountay

Date of the savings group established 01-Oct-02

Source of fund as September 2005

Population in village 1,089

Members in savings group 300

Loan balance as September 2005 Kip 167,000,000

Internal fund Kip 158,148,500

External fund Kip -

2

Huannamyene

12-Jun-03

1,710

338

136,200,000

109,031,000

-

3

Dongluang

01-Apr-04

1,048

153

74,700,000

56,499,000

6,000,000

4

Phonekeo

15-Jun-05

532

145

13,112,000

11,967,000

11,000,000

5 6

Phonesavanh Sisavard

01-Jun-05 10-Aug-05

680 393

74 121

650,000 1,200,000

650,000 1,120,000

-

Source: Author’s survey data, September 2005 and March 2006. In the survey sample, there were two savings groups which had external funds. One is a savings group in Dongluang village which received a soft loan from the fund committee of Naxaithong district amount of six millions Kip with one year term of loan and 2% interest rate per month, on 5th August 2005. However, the savings group could

123

Appendix B repay this loan by three months later. A savings group in Phonekeo village was another one which obtained soft loan from the fund committee of Naxaithong city. On the 1st of August, 2005, this savings group borrowed from the fund committee of Naxaithong city the amount of 11 million Kip with a one year term of the loan and 3% interest rate per month. At the time of follow up survey in this village, 3rd March 2006, this loan was yet not repaid. In addition, as table 4 shows four savings groups are dependent on internal funds (member deposit) in order to lend to their members. This implies that most savings groups in Laos are much based on savings mobilization for lending to their members rather than soft loans or external funds for the project. This is in contrast to the village bank system in Northeast Thailand of the study by Coleman (1999). Even though, this approach was promoted by Thai NGO, Community Organizations Development Institute: Thailand (CODI) which follow the “Village Bank” group lending methodology of the Foundation for International Community Assistance (FINCA)51 (Chaleunsinh, 2004: 7). This result is consistent with Chaleunshinh that: “Savings group” operation in Lao PDR, especially in Vientiane Capital, is quite different from the “Village Bank” system in Northeast Thailand, due to the fact that most of the village savings groups in Laos are currently based on money mobilized within their villages or internal funds rather than from soft loans from a project. However, this approach was promoted by the Thai NGOs, Foundation for Integrated Agricultural Management: FIAM, and Community Organizations Development Institute: Thailand or CODI, starting in 1997. Both organizations follow the “Village Bank” group lending methodology of the Foundation for International Community Assistance (Chaleunsinh, 2004:7).

51

Please refer to Ledgerwood (1999: 85) for more explanation of Village Bank.

124

Appendix B 2.2.2 Credit Savings groups lend money to their members on a monthly basis. Normally, the borrowing day is the day after deposit day. Members who require the loans fill the loan request form and submit it at the deposit day. The group committee collects all the request forms52 and summaries the amount of loan requested. They then check how much money the group has at that time53. The savings group provides different kinds of loan as shown in table 3. In the six savings groups of this sample, loan size is range from 100,000 Kip to 5,000,000 Kip with a 5% interest rate per month54 and 3 to 6 months term of loan55. Limitation of loan size is dependent on the practices of each savings groups. For example, in the case study of the six savings groups, the groups in Dongluang, Huannamyene and Phonesavanh villages have a minimum loan size of 100,000 Kip while other three savings groups have not settled for that. For maximum loan size, savings groups in Nakountay, Dongluang and Phonesavanh villages have settled at amount of 5 million Kip, one million Kip and 300,000Kip respectively, whereas the rest three savings groups have not settled for that but they consider the loan size depending on the borrowing request and how much the group have money for (deposit amount). 52

Generally, borrowing conditions are settled in different way among savings groups. In case of savings group in Nakountay village, some conditions for member borrowing are: (1) members have to have deposit amount with the group for guarantee; (2) If they borrow at 500,000kip, TV or freeze can be collateral; if they borrow amount more than 3,000,000kip, a land certificate can be collateral but it is also depended on how much volume of their deposit with the group; (3) if member has no any things to be guarantee, friends can be guarantor; (4) If a wife borrows, her husband has to be guarantor and vice versa. 53 Normally, savings group does not keep any money balance with the group after receiving deposit from members, then the day after, all money has to lend out to members depending on their request and borrowing conditions. In case of deposit balance of the group remaining at 1 million Kip, the group committee will deposit that balance to bank such as Agricultural Promotion Bank, an example of savings group in Nakountay village. 54 Sometimes, interest rate per month is 3%, an example of savings group at Nakountay village, it is depended on what kinds of loan. 55 Three new village savings groups have 3 months term of loan while three old village savings group practice on 6 months term of loan during the survey time. For example, savings group in Nakountay village, term of loan was 3 months for the period from year 2002 to 2004, but since 2 March 2005, term of loan has been 6 months.

125

Appendix B 3.

Other financial service providers in Naxaithong district According to the author’s survey in September 2005, some financial service

providers, apart from the savings group, provide basic financial services, mainly credit and deposits in the survey area. For credit access, 12% of sample survey could obtain a loan from financial providers. The Agricultural Promotion Bank (APB) was the first financial provider at 6%, followed by money lender at 4%, while other microfinance providers covered only by 1% of sample survey (see table 5). It is interesting that 17% of the control group had access to credit from other financial providers. Of that, APB was the most dominant financial provider at 12%. In terms of deposit service, about 9% of those surveyed in the sample deposited with other financial providers. APB and other microfinance providers covered by 4% per each while the 2% was covered by money rotating (see table 6). For control group sample, 8% of its sample could access to financial providers which APB and other microfinance providers shared that percent by half per each.

126

Appendix B Table 5: Loan balance at different sources of finance for sample survey in six

villages as September 2005

Mean by group:

Sources of credit:

APB 0.02 - 0.17 Interest rate per day (%): All sample (n=251): Number of borrowers 0.06 Loan amount (Kip) 40,035,438 Member (n=183): Number of borrowers 0.07 Loan amount (Kip) 54,851,885 Treatment group(n=131): Number of borrowers 0.05 Loan amount (Kip) 231,870 Control group (n=52): Number of borrowers 0.12 Loan amount (Kip) 192,452,308 Nonmember (n=68): Number of borrowers 0.03 Loan amount (Kip) 161,765

Other MFIs b) 0.167 - 0.2

Money lender 0.7 - 1

Relative & friend 0 - 10

0.01 41,833

0.04 39,982

0.02 21,978

0.12a) 40,139,231

0.02 57,377

0.04 51,833

0.03 30,145

0.15 54,991,240

0.02 80,153

0.04 10,687

0.03 41,729

0.14 364,439

-

0.04 155,489

-

0.03 8,088

Total

0.02 0.17 962 193,000,000 -

Note: a) One person could access to credit in both APB and relative &friends. b) This consists of savings group of Lao Women Union at Naxaithong city, Rural Development Cooperative Naxaithong and others. Source: Author’s survey data, September 2005.

127

0.06 169,853

Appendix B Table 6: Deposit balance at different institutions as September 2005

Mean by group:

Interest rate per year (%): All sample (n=251): Number of depositors Deposit balance (Kip) Member (n=183): Number of depositors Deposit balance (Kip) Treatment group (n=131): Number of depositors Deposit balance (Kip) Control group (n=52): Number of depositors Deposit balance (Kip) Nonmember (n=68): Number of depositors Deposit balance (Kip)

Deposit balance at different institutions: Other APB MFIs Money rotating

Total

8 - 24

0-3

0.04 32,207

0.04 12,530

0.02 1,801

0.09a) 46,538

0.05 30,514

0.05 17,186

0.02 2,377

0.12 50,077

0.05 40,840

0.06 22,634

0.02 3,321

0.14 66,794

0.04 4,500

0.04 3,462

0.01 36,765

Note: a) One person deposited with both APB and Money rotating Source: Author’s survey data, September 2005.

128

0.08 7,962 0.01 250

0.01 37,015