Impact evaluation of microcredit on welfare of the vietnam rural household

UNIVERSITY OF ECONOMICS HO CHI MINH CITY VIETNAM INSTITUTE OF SOCIAL STUDIES THE HAGUE THE NETHERLANDS VIETNAM - NETHERLANDS PROGRAMME FOR M.A IN DEVELOPMENT ECONOMICS IMPACT EVALUATION OF MICROCREDIT ON WELFARE OF THE VIETNAM RURAL HOUSEHOLD A thesis submitted in partial fulfilment of the requirements for the degree of MASTER OF ARTS IN DEVELOPMENT ECONOMIC S By PHAM TIEN THANH Academic Supervisor: DR PHAM BAO DUONG HO CHI MINH CITY, OCTOBER 2012 DECLARATION I certify that the contents of thesis have been carried out and written by me to the best of my knowledge and with the support in preparing this paper from many different sources I certify that this thesis has not been submitted to any other programs or journals HCMC, October 15", 2012 PHAM TIEN THANH ACKOWLEDGEMENTS This thesis is impossible to be achieved without the support and assistance of the following people: Firstly, I would like to express my greatest gratitude to Dr Pham Bao Duong, my academic supervisor, who advised and instructed and supported me during the process of this thesis His expertise and his suggestions have provided a good basis for the improvement of my research His enthusiasm and encouraging is also a motivation for me to achieve me thesis I would like to give my special thaaks to Prof.Dr Nguyen Trong Hoai, Dean of Vietnam—The Netherlands Programme and Dr Pham Khanh Nam, Academic Director of Vietnam —The Netherlands Programme Their knowledge and enthusiasm has supported me a lot during my thesis writing process This is also a good opportunity to express my appreciation to all the lecturers who equipped me with valuable knowledge during my study at Vietnam —The Netherlands Programme I would also like to appreciate Mr Nguyen Khanh Duy, Lecturer at the Faculty of Development Economics, University of Economics, Ho Chi Minh City His support with data as well as using econometrics software is a great contribution to the completion of my thesis Lastly, I am grateful to my beloved parents who gave moral support and encouraged me to finish my thesis during writing process ABSTRACT This research evaluates the impact of microcredit on the welfare of households living in the Vietnam rural areas, especially the poor The research is analyzed based on a data of the Vietnam household living standard survey (VHLSS) in the year 2008 The difference of the research in comparison with the previous studied about the relationship between microcredit and welfare is the employment of propensity score matching (PSM) method, thus it reflects the impact of microcredit on rural households’ living standard better and more precisely The result shows that microcredit will result in better welfare of rural households via a greater increase in the income and consumption per’ capita per‘ month of the participating households However, the result about the poor mral households showed that microcredit does not result in a higher increase in income of the participants than that of the nonparticipants, but contributes to a greater rise in the consumption The research also showed the determinants on the accessibility to microcredit programs of the households living in rural regions The results found out that the probability of accessing the microcredit sources of the rural households in Vietnam is still low Moreover, the proportionate of accessibility to microcredit of the poor household is even less that of the non- poor households, which means microcredit programs mistarget the poor households From those results, the research gives policy recommendations to improve microcredit programs in rural areas as well as to support more poor households to access to microcredit sources TABLE OF CONTENT DECLARA ACKNOWLE DMEN ABSTRACT TABLE OF CONTEN LIST OF ABBRIVIATION LIST OF TABLES AND FIGURES CHAPTE INTRODU 1.1 Problem Statement I - 1.2 Objectives of the study 1.3 Research questions 1.4 Organization of the research CHAPTER II ' LITERARTURE REVIEW Overview of Poverty 1 Definition 1.2 Method of defining poverty 2.2 Overview of Microcredit 2.2 Some definitions 2.2.2 Characteristics of Microcredit 2.2.3 Overview of rural credit market in Vietnam 11 2.2.4 Overview of microcredit program in Vietnam 12 2.3 Empirical Study 17 2.3 Impact of micro credit on welfare/ living standard of the rural households 17 2.3.2 Determlnants Of the accessibility to microcredit programs 24 CHAPTER RESEARCH METHODOLOGY AND DATA DESCRIPTION Model of determinants of access to credit 30 3.2 Im pact Evaluation techniques .34 3.2 Some Definition ., 34 3.2.2 Impact evaluation using PSM technique 34 3.2.3 Impact evaluation using DID technique .38 3.3 Data Description .41 3.3.1 Survey area 41 3.3.2 Data sources 41 3.3.3 Sample selection 41 CHAPTER IV RESU Descriptive Statistics .44 4.2 Determinants on microcredit participation 46 4.3 Impact of microcredit on welfare of rural households using PSM 51 4.4 Impact of microcredit on welfare of the rural poor using PSM .52 4.5 Impact of microcredit on welfare of rural households using DID with fixed effect .55 4.6 Comparison between the results of PSM and DID method .56 4.7 Comparison with previous studies 57 CHAPTER V CONCLUSION, POLICY RECOMMENDATION AND LIMITATION Conclusion .59 5.2 Policy Recommendation 62 5.3 Limitation 64 REFERENCE 65 LIST OF ABBREVIATIONS LIST OF TABLES AND FIGURES List of Tables Table : Poverty Rate in Vietnam Table 2: Characteristics of Microcredit Programs in Vietnam from 2005 to 20 1 12 Table 2.3 : Main Characteristics of the MFIs in 2011 14 Table 2.4: Characteristics of Microcredit Programs by VB SP from 2005 to 201 16 Table 5: Table Table 2: Table 3.3: Table 1: Table 4.2: Table 4.3 : Table 4.4: Table 5: Table 6: Table Table 4.8: Table 9: Table 10: Table 11: Table 12 Table 13 Table 14: Table 15: List of Figures Figure : Gross Loan Portfolio of microcredit in Vietnam from 2005 to 2011 13 Figure 2 : Characteristics of Microcredit Programs by VBSP from 2005 to 2011 15 Figure 2.3 : Determinants on Accessibility to Microcredit and Welfare Indicators .29 Figure : Illustration of Impact Evaluation Using DID Method .39 4.1 12 2.2 Impact on log of average consumption per capita 2.2 I Stratification method Analytical standard errors n treat n CG:)tr 290 254 634 Note: the numbers of treated and controls rcfer to actual nearest neighbour matches Bootstrapped standard errors n treat contr n 290 ATT Std Err 237 254 Note: the numbers of treated and controls refer to actual nearest neighbour matches 2.2 Stratification method Analytical standard errors n treat contr n ATT Std Err 17 604 Note: the numbers cf treated and controls refer to actual matches wlthin radius Bootstrapped standard errors n treat contr n ATT Std Err 362 I¶ote: the numbers of treated and contrcls refer to actual matches within radius 2.2 3 Stratification method ATT estimation with the Stratification method Analytical standard errors n treat n contr ATT 290 1994 124 Std Err 5.054 Bootstrapped standard errors n treat n contr ATT Std Err 290 1994 124 020 8l 6.299 2.2 Impact on consumption per capita 2.2 1.Nearest Neighbor Matching method Analytical standard errors n treat n ccntr 290 73 284 254 029 Note: the numbers of treated and controls refer to actual nearest neighbour matches Bootstrapped standard errors n treat n contr ATT 254 Std 7? 284 618 Err 18 936 Note: the numbers of treated and controls refer to actual nearest neighbour matches 2.2 Radius Matching method Analytical standard errors n treat n 170 - contr ATT Std Err 3.948 86.753 362 Note: the numbers of treated and contrcls refer to actual matches within radlus Bootstrapped standard errors n treat ATT n contr Std Err 902 170 Note: the numbers of treated and controls refer to actual matches within radlus 2.2 Stratification method Analytical standard errors n treat contr 290 n ATT 1994 77 Std Err 14 780 714 Bootstrapped errorsATT n treat n standard contr 290 1994 Std, Err 309 77.714 82 The impact of microcredit program on welfare of the ral poor using PSM 3.1 Estimation of propensity score Algorithm to estimate the propensity score The treatment is credit -0 ’ No ; | = - Yes | Freq Percent 241 70 6E 340 100.00 99 Total | 70.88 150 00 29 12 Estimation of the propensity score (sum of wgt is 3806e+05) Iteration 0: log pseudolikeiihood = -203 13416 Iteration 1: log pseudolikelihood = -167.80027 Iteration 2’ log pseudoiikelihood = -165.87674 Iteration 3: log pseudolikel ihood = -165.65165 Iteration 4: log pscudolikelihood = -165 85164 Probit regression Wald chi2(18) Log pseudolikelihood -165 85164 ’ Number of obs Prob > chi2 Pseudo R2 credit | | | | | | | | | | | 2546174 -.0114784 -.0000578 0738638 0461368 -.0677631 5703301 1593863 -.3689113 26e-06 5.46e- 06 0104119 gcofl | - l127]56 geof2 geo2 rbs ca2 po _cons = = = - 340 73 11 0.0000 1835 [95% Conf Interval] Coef hgender age age2 hedu hmar ost ethnic hhsize drate lpc hval distance ’ | -.3248203 | 4703378 | -.1457987 | 2659439 | 8480587 | -2.584804 Robust Std Err 3139148 0398684 0003734 0276461 3307367 4929231 2627605 0646162 1451254 7.45e- 06 97e- 06 0081855 2612711 233524b 2255054 204928 2631803 3086781 08646 0.81 -0.29 -0.15 67 0.14 -0.14 17 2.47 -2.54 98 2.77 27 417 0.773 877 008 889 0.891 030 014 0.011 329 0.006 203 -.3606443 -.089619 -.0007896 0196785 - 6020952 -1.033875 0553289 0327409 - 6533518 -7 34e-O 1.59c- 06 -.0056313 8698791 0b66622 000674 128049 6943687 8983484 1.085331 2860318 -.0844708 0000219 32e.0264551 -1 39 2.09 -0.71 1.01 2.75 -2 38 164 037 0.477 312 0.006 0.017 -.7825201 0283552 -.5474502 -.2498801 2430609 -4.714226 1328796 9123203 2558528 7817678 1.453057 -.4553815 -0 43 6b6 Note: the common support option has been selected The region of common support is [.01778183, 90920033] - 6248075 3993562 Description of the estimated propensity score in region of common support Estimated propensity score Percentiles 0200174 smallest 0177918 017?170 0361762 0540576 1257768 0200174 272886 908 75% 4406013 6000481 99% 824314 Obs 0194947 Sum of Wgt l4ean Std Dev Largest 824314 6413337 8747703 6742904 Variance Skewness 9092003 Kurtosis 3016466 2062667 0425459 5873015 565311 Step l: Identification of the optimal number of blocks Use option detail if you want more detailed output The final number of blocks is This number of blocks ensures that the mean propensity score is not different for treated and controls in each block Step 2: Test of balancing property of the propensity score Use option detail if you want more detailed output The balancing property is satisfied This table shows the inferior bound, the number cf treated and the number of controls fcr each Llock Inferior of block | of pscore | =0 No ; =1 ’ Yes 0177818 | 24 1666667 67 50 19 3333333 6666667 | 231 | Total 19 | 86 24 11 43 28 99 | 78 330 8333333 Total Note: the common support option has been selected End of the algorithm to estimate the pscore 84 3.2 Impact evaluation of microcredit on welfare using PSM ' 3.2 Impact on log of average income per capita 3.2 I.Nearest Neighbor Matching method Analytical standard errors n treat contr n ATT 99 J82 Std Err 061 C 354 Note: the numbers of treated and controls refer tc act ualnearest neighbour matches Bootstrapped standard errors n treat ATT n contr Std Err Note: the numbers of treated and controls refer to actual nearest neighbour matches 3.2 Radius Matching method Analytical standard errors n treat ATT n contr Sto Err 76 113 064 Note: the numbers of treated and controls refer to acLual matches within radius " Bootstrapped standard errors n treat contr ATT n 76 Std Err 136 0.085 598 Note: the numbers of treated and controls refer to actual matches within radius 3.2 Stratification method ATT estimation with the Stratification method Analytical standard errors n treat contr n 99 ATT 231 Std Err 0.079 Bootstrapped standard errors n treat contr 99 231 n ATT 079 Std Err 557 85 394 3.2 Impact on average income per capita 3.2 1.Nearest Neighbor Matching method Analytical standard errors n treat contr n 9J ATT 62 Std Err 33.76:7 51 768 533 Notc: the numbers of treated and controls refer to actual nearest neighbour matches Bootstrapped standard errors n treat n contr ATT Std Err 51.718 656 Note: the numbers cf treated and controls refer to actual nearest neighbour matches 3.2 2.Near Radius Matching method Analytical standard errors n treat n contr ATT Std Err 76 3? 481 Note: the numbers of treated and controls refer to actual matches within radius Bootstrapped standard errors n treat contr n ATT Std Err 59 9317 453 Note: the numbers of treated snd controls rcfer to actual matches within radius 3.2 3.Near Radius Matching method ATT estimation with the Stratification method Analytical standard errors n trest n contr 99 ATT rtd Err 47.168 Bootstrapped standard errors n treat contr 95 n ATT 231 47.168 Std Err 31 438 86 3.2 Impact on log of average consumption per capita 3.2 Stratification method Analytical standard errors n treat n contr 99 62 ATT Srd Err Note: the numbers of Lreatcd and controls refer to actual nearest neighbour matches Bootstrapped standard errors n treat contr n ATT 62 Std 6rr 12% 13 0? 26 Note: the numbers of treated and controls refer to actual nearest neighbour matches 3.2 Radius Matching method Analytical standard errors n creat n contr ATT otd 76 Err 0.G6 949 Note: the numbers of trcated and controls refer to actualmatches within radius Bootstrapped standard errors , contr n treat n ATT Std Err 1º.9 0.07 519 Note: the numbers of treated and contrcls refer to actual matches within radius 3.2 Stratification method Analytical standard errors n treat contr 99 n ATT 23 Std Err Std Err 141 Bootstrapped standard errors n treat n contr ATT 141 0.051 2.754 87 º 3.2 Impact on consumption per capita 3.2 1.Nearest Neighbor Matching method Analytical standard errors n treat n contr 99 62 ATT Std Err 28 485 314 Note: the numbers of treated and controls refer to actual nearest neighbour matches Bootstrapped standard errors n treat n contr ATT Std Err 27 62 212 2.422 Note: the numbers of treated and controls refer to actual nearest neighbour matches 3.2.4.2 Radius Matching method Analytical standard errors n treat contr n ATT Std Err 2.964 Note: Lhe numbers of treated and controls refer to actual matches within radius , Bootstrapped standard errors n treat ATT n concr Std Err 98 O95 76 496 Note: the numbers of treated and controls refer to actual matches within radius 3.2 Stratification method ATT estimation with the Stratification method Analytical standard errors n treat contr n ATT Std Err 71 720 99 Bootstrapped standard errors n treat contr 99 n ATT 71.725 Std Err 20 512 496 Impact evaluation of microcredit on welfare using DID with Fixed Effect Impact on average income per capita Note: credit08 omitted because of collinearity (*) Fized — effects (within) ’Rsq: within - 0.0780 regression between - hhid 0.0070 Group variable: overall = 0.0321 Obs per group: = avg max = 179 F(2, 177) Prob › F corr(u_i, Xb) = 0.0U77 Coef avinc | year credit09 timecredit cons | | | | 96.39623 (omitted) 61 06953 475.7151 sigma_u | 296 205i4 sigma_e | 305 5032 rho | 48455091 r test that all u_i-0: 4.2 49 0.0008 std Err f 95% Conf Interval] 41 96409 30 023 13.58189 179 2106 65.71173 22 83438 93 20 83 0.354 0.000 -68.60977 430 6524 190.7488 520.7778 (fraction of variance due to u_i) Prob > F = 0.0000 F(178, 177) = Impact on log of average income per capita Note: credit08 omitted because of collinearity (*) Group variable: hhid Fixed-effects (within) regression Number of obs Number of groups R - sq: within = 0.3235 betwecn = 0.0005 Obs per group: = avg = corr(u_i, Xb) F(2, 177) overall = 0.1039 = -0.0044 lnavinc | Err year | credit08 | timecredit | cons | i a u a ho max Coef - 358 179 2 42.32 Prob › F Std [95% Conf Interval] 2900307 04673 21 500 197811 3822503 092364 5.977053 0731747 0254277 26 235.06 209 000 -.0520431 5.926872 2367712 027233 (omitted) 42 34 996 F test that all u_i-0: f a ct ion o r ian F(178, 177) = 09 89 to Prob › F = 0.0000 4.3 Impact on average consumption per capita Note: credit08 omitted because of collinearity Number Number groups (*) FlXed- effects Group variable: hhid Obs per group: = avg F(2 ,177) Prob > F max -’ = -0.0162 avexp | Coef Err year | credit08 | 121 2292 (omitted) 33 72849 timecredit | 031 115 1666 52 81557 sigma_u | sigma_e | rhO | 230 84967 245.54711 46917808 _COnS Đ.ả0i7 [J.0 ảQ [9S% Conf Interval] Std 54.6675 18 lh./5/LJ5 )0 j 187.791 10 19 9 346 0827 418.5206 (fraction of variance due to u_i) F test that all u i = F(178, 177) = 4.4 179 (within) regression R- sq: witkin - 208% LeLwecn = 0.0049 overall = 0.0980 corr(u_i, Xb) of obs o[ i.76 Prob > F = 0001 Impact on log of average consumption per capita Note: credit08 omitted because of collinearity (*) Fixed-effects (within) regression Croup variable: hhid R-sq: 5439 within Number of obs Number of groups = 358 - 179 Obs per group: = av g ma x = between = 0.0051 overall = 0.2015 corr(u_i, Xb) F(2 , 177) Rrob > F - rho | u_i) -0.0029 lnavexp | Coef year | credit08 | timecredit | cons 3814157 (omitted) 1451803 5.7923 sigma_u | sigma_e | 39940252 29068547 65372553 F test that all u_i= 0: Std Err [95% Conf Interval] 0399287 3026181 4602133 0217908 5.749423 2685698 5.835177 0625245 0217269 2.J 2t6 60 021 001) (fraction of variance due to F(178, 177) = 3.77 Prob › F = 0.0000 (*) Note: The credit08 variable (the participation in the program) is omitted because of its perfect correlattion with the other dummy variables (the dummy variable of households generated from using fixed effect model) Therefore, when using DID with fixed-effect regression, we should only consider the pre and post program impact over time (which means the coefficient of timecredit variable) 90 ... income of the poorest households than that of the medium-income households in Vietnam rural regions, then this leads to the positive impact on poverty alleviation In a research on the case of the Vietnam. .. accessing the microcredit sources of the rural households in Vietnam is still low Moreover, the proportionate of accessibility to microcredit of the poor household is even less that of the non- poor households,... 4.4 Impact of microcredit on welfare of the rural poor using PSM .52 4.5 Impact of microcredit on welfare of rural households using DID with fixed effect .55 4.6 Comparison