• Tidak ada hasil yang ditemukan

There is now plenty of evidences of the impact of microfinance on the lives of the poor borrowers. This study also found such evidence of impact. Different analysis techniques were used to investigate the level of impact where the value of asset holding of the participants was taken up as the major outcome variable. Since the study is based on a panel data set and includes all the households including the dropout clients, it is less prone to the bias of comparing the incoming clients to the older ones.

Looking into the pattern of dropout, we found that poverty status of the clients and their group, availability of alternative services, their loyalty to BRAC along with their satisfaction with the service have significant association with the possibility of dropout. Flexibility in loan

repayment was observed as a major area to work on to improve client retention. However, finding appropriate flexibility that does not impede the financial health and ‘good practices’ of microfinance is a major challenge.

REFERENCES

Alarakhia S, Barua P. Sectors scan of TUP enterprises: Identifying the determinants of sustainability, Research and Evaluation Division, BRAC: 2005 (unpublished).

Barua P, Sulaiman M. Is BDP ultra poor programme working? Survey of some key issues. Dhaka and Ottawa: BRAC and Aga Khan Foundation Canada, 2006:vi, 20p (CFPR Working Paper Serious No. 16).

Carter M R, Barrett CB. The Economics of Poverty Traps and Persistent Poverty: An Asset-based Approach”. Social Science Research Network: 2005.

Dufhues T, Heidhues F, Buchenrieder G. Participatory product design by using conjoint analysis in the rural financial market of Northern Vietnam. Asian Economic Journal vol. 18(1): 81-114. 2004.

Hayashi F. Econometrics. Princeton: Princeton University Press, 2000.

Karlan, D. Microfinance Impact Assessments:The Perils of Using New Members as a Control Group.

Journal of Microfinance, 2001: 3(2): 75 – 85.

Morduch J Income Smoothing and Consumption Smoothing, Harvard Institute of Economic Research Working Papers 1727, Harvard - Institute of Economic Research. 1995.

Nath SR, Khan MK. Schooling and literacy. In: Shareef H (Editor). Towards a profile of the ultra poor in Bangladesh: Findings from CFPR/TUP baseline survey. Dhaka and Ottawa: BRAC and Aga Khan Foundation Canada, 2004: 133-49.

Perdana AA. Risk management for the poor and vulnerable. CSIS working paper series no. 93.

Department of economics, CSIS, Jakarta: 2005:2:19 (http://www.csis.or.id/papers/wpe093). Accessed on 15 May 2007.

Palli Karma-Sahayak Foundation (PKSF). “Maps of microcredit coverage in upazilas of Bangladesh”, Dhaka: 2004: pp. 13-23.

Pawlak K, Szubert D. Counting on your prospective clients: Guiding principals in measuring microfinance client satisfaction and loyalty: MFC Spotlight Note No. 8. Microfinance Centre.

Ravallion M. The mystery of the vanishing benefits: Ms Speedy analyst’s introduction to evaluation”, World Bank, Washington DC, USA. 1999.

Schreiner M. A simple poverty scorecard for Bangladesh. Microfinance risk management”, L.L.C. Saint Louis, USA. 2006: pp. 2-31 (unpublished).

Sulaiman M. Do relationships matter? an empirical study of social capital in rural Bangladesh: p.26-28.

Social capital and economic well-being. Dhaka and Ottawa: BRAC and Aga Khan Foundation Canada, 2006. vi, 32p. (CFPR Working Paper Series No. 15).

Staiger D, Stock JH. Instrumental variables regression with weak instruments. Econometrica 1997;65(3):557-86.

Tudenschi G A, Karlan D. Microfinance Impact: Bias from Dropouts. Policy Paper, Yale University, USA.

Zaman H. Assessing the poverty and vulnerability impact of micro-credit in Bangladesh: A case study of BRAC. The World Bank. 2000.

Zimmerman F, Carter MR. Asset Smoothing, Consumption Smoothing and the Reproduction for Inequality under Risk and Subsistence Constraints. Staff Paper No. 402. Agriculture and Applied Economics. 1999.

30 Annexure

Annex-1. Mark Schreiner's poverty score-card derived from BBS HIES

Indicators Attributes Points

1 What type of latrine do you have? Open field pit pucca

sanitary or water seal pacca

Assigned Score 0 7 12

2 How many household members are 11 years old or younger 4 or more 3 2 1 0

Assigned Score 0 7 12 17 26

3 Does any household member work for a daily wage? Yes No

Assigned Score 0 7

4

How many living rooms does the house have (excluding ones used

for business and kitchen)? 1 2 or 3 4 or more

Assigned Score 0 3 9

5 Do all children ages 6 to 17 attend school in your household? No

No children age

6 to 17 Yes

Assigned Score 0 4 6

6 Does the household Own a TV set? No Yes

Assigned Score 0 11

7 How many hectares of cultivable land does the household own? Less than 0.34 0.34 to 0.99 1 to 1.99 2 or more

Assigned Score 0 3 4 9

8 What is the main construction material of the walls of the house? Hemp/hay/bamboo or mud brick C.I. sheet/wood Brick/cement

Assigned Score 0 5 7

9 Does the household own any cattle? No Yes

Assigned Score 0 9

10 Does the house have a separate kitchen? No Yes

Assigned Score 0 4

Total

Annex-2. Steps for instrumental variables (IV) analysis OLS model : Yi01LSi2HHi+ui

i i 2 i 1 0

i NL HH u

Y =β +β +β +

Condition of IV method: The conditions of instrumental variables to be used are relevance i.e. there must be relationship between IV and endogenous variables to be instrumented (LS in our case), and exogeneity i.e. correlation between IV and error term must be zero.

Relevance: correlation (IVi, LSi) ≠ 0 or, correlation (IVi, NLi) ≠ 0 Exogeneity: correlation (IVi, ei) = 0

Stages of IV: In 2SLS estimation process, amount of last loan is predicted by the instrumental variables and other exogenous variables in the first stage. The predicted amount of last loan instead of actual ones enters at the second stage with other exogenous variables.

i i 2 i 1 o

i IV HH v

LS =α +α +α + or, NLio1IVi2HHi+vi

i 2 i 1 0

i ˆ ˆ IV ˆ HH

L =α +α +α or, NLˆi=αˆ0+αˆ1IVi+αˆ2HHi

i

* i 2

* i

* 1 0

i LSˆ HH u

Y =β +β +β + or, Yi*01*NLˆi*2HHi+ui

Justification of IV: To estimate J statistics the number of instrumental variables has to be larger than the number of included endogenous variable what is the case here. In calculating J statistics, using the estimates of 2SLS, the value of asset is predicted3 to estimate the residuals.

* i 2

* i

* 1 0

i LS HH

Yˆ =β +β +β or, Yˆi*0*1NLi*2HHi

Y eˆi= −

i i 2 i 1 0

i IV HH

eˆ =π +π +π +ε

The residual is regressed against the instruments and other exogenous variables; and endogenous size of loan is excluded. Partial F-statistics of the instruments from the final regression multiplied by the number of instruments yields the J statistics. Thus, all four instruments satisfy the condition of exogeneity.

3 In calculating the predicted values, actual values of loan size is used instead of predicted ones from the first stage.

Annex-3. Determinants of participation using logit model

Relative economic status (1= better-off 0= otherwise) 0.447

(2.45)**

Female headed HH (1= yes 0= otherwise) 0.460

(0.97)

Highest education of HH member 0.045

(1.97)**

Violence index (0= no violence against women, ... 7= extreme violence) 0.355

(3.21)***

Knowledge index (0= no knowledge on legal issues, ….7= knowledge on all legal issues) -0.061

(1.64)

Improved creditworthiness (1= yes 0= otherwise) 0.105

(0.71)

Improved economic status (1 =yes 0= otherwise) -0.349

(1.68)*

Having mobility outside village (1= yes 0= otherwise) 0.095

(0.33)

Having food security (1= yes 0= otherwise) -0.393

(2.37)**

HH having sanitary latrine (1= yes 0= otherwise) 0.203

(1.45)

Source of cooking water (1= tube-well 0= otherwise) 0.802

(1.20)

HH Having electricity connection (1= yes 0= otherwise) 0.234

(1.24)

Number of HH member 0.112

(2.39)**

HH with at one female earner (1= yes 0= otherwise) -0.411

(2.54)**

Number of income source -0.137

(1.72)*

HH having at least one day labour (1= yes 0= otherwise) 0.250

(1.65)*

Amount of cultivable land (decimal) -0.001

(1.37)

Number of cow 0.072

(1.35)

Number of goat 0.058

-1.33

Number of crisis faced 0.461

(4.30)***

Constant -1.312

(1.76)*

Observations 1038

Pseudo R squire 0.061 Absolute value of z statistics in parentheses

Significant at 10%; ** significant at 5%; *** significant at 1%

Annex-4. Questionnaire for repeat survey

PART A

Sample Status: New Scale up New Comparison ID No

Member name --- --- --- Member No Joining Date

DD MM YY Name of VO and Code ---

Branch Office and Code ---

Village --- Union --- Upazila --- Zila --- Name of interviewer ---Date

DD MM YY Remark of cross checker 1= Satisfied 2= Not satisfied Signature after satisfaction

--- Remark of supervisor 1= Satisfied 2= Not satisfied Signature after satisfaction

--- 1. What is the status of household interview?

1= Household found and interviewed

2= Household found but could not be interviewed 3= Household was not found

2. If 2 and 3 why not found (mention reason)

1 = Respondent was not found by visiting 2-3 different days

2= Death of respondent 3 = merged with other household 4 = permanent migration (10 Km.

away from previous place) 5 = temporary migration

Others (……….………..…………..) 3. Member line No. according to baseline

4. Respondent’s line No. in present survey (see table 5):

Dokumen terkait