The impact evaluation of rural credit on accessibility to education, health care and clean water in rural vietnam

UNIVERSITY OF ECONOMICS INSTITUTE OF SOCIAL STUDIES HO CHI MINH CITY THE HAGUE VIETNAM THE NETHERLANDS VIETNAM - NETHERLANDS PROGRAMME FOR M.A IN DEVELOPMENT ECONOMICS THE IMPACT EVALUATION OF RURAL CREDIT ON ACCESSIBILITY TO EDUCATION, HEALTH CARE AND CLEAN WATER IN RURAL VIETNAM BY VO VAN TAI MASTER OF ARTS IN DEVELOPMENT ECONOMICS HO CHI MINH CITY, NOVEMBER 2013 UNIVERSITY OF ECONOMICS INSTITUTE OF SOCIAL STUDIES HO CHI MINH CITY THE HAGUE VIETNAM THE NETHERLANDS VIETNAM - NETHERLANDS PROGRAMME FOR M.A IN DEVELOPMENT ECONOMICS THE IMPACT EVALUATION OF RURAL CREDIT ON ACCESSIBILITY TO EDUCATION, HEALTH CARE AND CLEAN WATER IN RURAL VIETNAM A thesis submitted in partial fulfilment of the requirements for the degree of MASTER OF ARTS IN DEVELOPMENT ECONOMICS By VO VAN TAI Academic Supervisor: Dr PHAM KHANH NAM HO CHI MINH CITY, NOVEMBER 2013 CERTIFICATION I hereby declare that the substance of this thesis is my own work and knowledge This dissertation has not been submitted for any other degree or diploma of the university or higher degree I certify that its contain has not been published or written by another person VO VAN TAI November 28, 2013 ACKNOWLEDGEMENTS During the time of studying in Vietnam - Netherlands programme for M.A in Development Economics, I have learned so much useful knowledge Therefore, I want to thank to the programme and all the teachers that have taught me I would like to express my deepest gratitude to my academic supervisor - Dr Pham Khanh Nam for his guidance and valuable comments in writing and finishing my M.A thesis His enthusiasm and encourage has supported me during the process of this thesis to finish it Besides my supervisor, I want to acknowledge the tremendous support that I received from Prof Dr Nguyen Trong Hoai – Dean of the programme for his assistance and great encouragement I also would like to thank Ms Vo Thi Phi - Director of TAN VUONG Food and Fisheries Import and Export Company Limited, where I am working, for her encourage and financial support, as well as the work supports of my colleagues, for me to study and finish the programme My grateful thanks to my classmates, Pham Tien Thanh and Cao Thi Tuyet Mai who helped me overcome the difficulties to finish my thesis Last but not least, I am truly grateful to my family for their love and spiritual support in my life Thanks for all VO VAN TAI November 28, 2013 ABSTRACT The research is conducted in order to evaluate the impact of rural credit program on living standard of the rural households The estimation is based on the secondary data, namely the Vietnam household living standard survey in 2010 (VHLSS2010) The research applied propensity score matching (PSM) method with various techniques in order to estimate the impact of rural credit program on living standard The results found that participation in rural credit only increase the expenditure on education while there is no evidence to conclude the relationship between rural credit program and such living standard indicators as accessibility to health care and clean water Moreover, the research also applied PROBIT model to investigate the factors that affect the probability of accessing to rural credit program The results showed that the probability of participating in rural credit program of the rural households are affected by such factors as age of household head, leadership status of household head, household size, house value, total land owned and managed by the household, household being poor or not, and geographic location The research also found that the rural credit program may not serve the poor because among the participants in the credit program, the number of the poor households is less than that of the non-poor households Finally, the research suggested policies and solution to improve the effectiveness of rural credit program in order to support the poor households i TABLE OF CONTENTS ABSTRACT i TABLE OF CONTENTS ii LIST OF FIGURES iv LIST OF TABLES iv CHAPTER 1: INTRODUCTION .1 1.1 Problem statement 1.2 Objectives of the research 1.3 Research questions 1.4 Research structures CHAPTER 2: LITERATURE REVIEW 2.1 Theory of impact evaluation methods 2.2 Empirical studies on impact of rural credit on living standard of the rural households 2.3 Empirical studies on determinants of the participation in rural credit programs 2.3.1 Characteristics at household head level 2.3.2 Characteristics at household level 2.3.3 Characteristics at commune level CHAPTER 3: METHODOLOGY 12 3.1 Analytical framework 12 3.2 Models and estimation strategies 12 3.2.1 Determinants of demand for rural credit 13 3.2.2 Impact evaluation by PSM .15 3.3 Data description 18 ii CHAPTER 4: OVERVIEWS OF RURAL CREDIT AND ACCESS TO EDUCATION, HEALTH CARE, CLEAN WATER IN VIETNAM 19 4.1 Rural credit market in Vietnam 19 4.1.1 The formal credit sector 19 4.1.2 The semi-formal credit sector 20 4.1.3 Informal credit sectors 20 4.2 Accessibility to education, health treatment and clean water in Vietnam 21 4.2.1 Accessibility to education in Vietnam 21 4.2.2 Accessibility to health treatment in Vietnam 23 4.2.3 Accessibility to clean water in Vietnam 24 CHAPTER 5: EMPIRICAL RESULTS 26 5.1 Non-parametric analysis .26 5.1.1 Descriptive Statistics .26 5.1.2 Participation in the rural credit program of the poor 28 5.1.3 Impact evaluation using two-sample t-test methods 29 5.2 Results of determinants of the participation of the rural household in rural credit programs .31 5.3 Impact evaluation using PSM methods 36 5.3.1 Balancing Test .36 5.3.2 Impact Evaluation via PSM 39 CHAPTER 6: CONCLUSION 43 6.1 Conclusion 43 6.2 Policy Implication 45 REFERENCES 47 APPENDIX .52 iii LIST OF FIGURES Figure 2.1: Evaluation using a with and without comparison Figure 3.1: Analytical framework on how rural credit affects the accessibility to education, health care and clean water 12 Figure 5.1: The comparison about educost, health and waterexp between participants and non-participants 30 LIST OF TABLES Table 2.1: Empirical studies about determinants of the participation in rural credit programs 10 Table 3.1: Determinants on demand for rural credit .13 Table 3.2: Indicators reflecting living standard in rural area 17 Table 4.1: Sources of rural credit 21 Table 4.2: Monthly consumption for education per capita .22 Table 4.3: Monthly consumption for health care per capita 23 Table 4.4: Percentage of households by main source of drinking water 25 Table 5.1: Descriptive statistics of all variables .26 Table 5.2: Participation in the rural credit program of the poor .28 Table 5.3: Impact of rural credit on living standard of rural households using Independent Sample T-Test Method 29 Table 5.4: Correlation Matrix of the continuous independent variables in PROBIT model 31 Table 5.5: Determinants of participating in rural credit program 32 Table 5.5a: Models of determinants of participating in rural credit program 33 Table 5.6: Impact of rural credit using NN technique 39 Table 5.7: Impact of rural credit using Stratification technique .40 Table 5.8: Impact of rural credit using Kernel Matching technique .41 Table 5.9: Summary of Impact of rural credit using PSM techniques 42 iv CHAPTER INTRODUCTION 1.1 Problem statement Vietnam has remarkable achievements in poverty reduction, specifically a report by World Bank and Vietnam General Statistics Office (GSO) in 2013 stated the poverty rate in Vietnam has decreased from 60 percent in 1990s to 14.2 percent in 2010 (using official poverty lines of Vietnam Ministry of Labour Invalids and Social Affairs (MOLISA) at VND500,000/person/month for the urban area and VND400,000/person/month for the rural areas), and nearly 30 million people have escaped poverty However, a lot of people in rural Vietnam are still living in poverty with very low living standard, specifically, World Bank and GSO in 2013 reported that the poor, especially those living in rural area, have less opportunities to access to education (as the Vietnam Population and Housing census by GSO in 2009, the illiteracy rate in rural area is percent, which is percent higher than that in urban area), have limited access to formal health care (a report by GSO in 2010 showed that 7.07 percent of people in rural area cannot access to health treatment, and the rate of the poor is higher than that of the non-poor), and have less opportunities to have a good job Therefore, improving living standard of the poor in rural area is considered as a top concern of the Vietnam Government With the objective of improving living standard of the rural poor, the Government has applied many programs such as free health care, food assistance, house assistance, education assistance, credit programs, etc Among these programs, rural credit program is considered one of the most effective programs to improve living standard of the poor (World Bank, 2012) The Government has applied a great number of the credit programs that provide loans to support the rural households Rural credit program has been applied in many countries and researched by many authors Waheed (2009) stated that credit may increase the living standard of the borrowers via improving their incomes, and then better education as well as health care To confirm the role of rural credit, Pitt and Khandker (1996) found that credit program has a positive significant impact on such factors as education, selfesteem, organizational and management skills, etc Pitt et al (2003) also concluded that participation in credit program has effect on the health status of children In addition, Coleman (2006) found that rural loan has positive effect on households’ living standards including clean water accessibility As a report by CARE program (2009), with a microfinance loan, the poor can run their own business and production, then they can generate income to pay for education of their children, pay for health treatment as well as access more to clean water The main objective of this research is to investigate whether rural credit programs have positive impact on the living standard of the households living in the rural areas in Vietnam via increasing their accessibility to education, health treatment and clean water Morduch and Haley (2002) found that when the poor are provided with credit, they can improve their living standard or at least smooth their expenditure However, due to the budget constraint, the people in the rural areas, especially the poor, have difficulty in accessing to formal and semi-formal credit sources Therefore, the government as well as financial institutions also makes efforts to support the poor to access to formal and semi-formal credit sources via providing programs that target the households in need of borrowing, especially the poor In order to target the poor effectively, many researches on the demand for credit of the rural poor have been conducted The credit providers need the information on characteristics of the households who are more likely to participate in credit programs This research also investigates the determinants of participation in rural credit programs of rural households in Vietnam The research aims at evaluating the impact of rural credit program on accessibility to education, health care and clean water of rural household, especially the poor, as well as investigating the factors that affect the participation in credit program of the rural households In order to achieve these objectives, the research probit credit age2 gender edu mar ostatus hsize depend hvalue land poorhh location distance commune135 road1 post Iteration Iteration Iteration Iteration Iteration Iteration 0: 1: 2: 3: 4: 5: log log log log log log likelihood likelihood likelihood likelihood likelihood likelihood = = = = = = -1894.2116 -1705.3079 -1702.5076 -1700.3157 -1700.2827 -1700.2827 Probit regression Number of obs LR chi2(15) Prob > chi2 Pseudo R2 Log likelihood = -1700.2827 = = = = 3615 387.86 0.0000 0.1024 -credit | Coef Std Err z P>|z| [95% Conf Interval] -+ -age2 | -.0001041 0000197 -5.28 0.000 -.0001427 -.0000655 gender | 0166138 0743933 0.22 0.823 -.1291944 162422 edu | 036363 024123 1.51 0.132 -.0109172 0836432 mar | 2720754 2254208 1.21 0.227 -.1697413 7138921 ostatus | 3455981 1217036 2.84 0.005 1070634 5841327 hsize | 0726222 0171044 4.25 0.000 0390983 1061462 depend | -.1203115 0398621 -3.02 0.003 -.1984397 -.0421832 hvalue | -9.00e-08 1.12e-07 -0.80 0.421 -3.09e-07 1.29e-07 land | 4.95e-06 1.86e-06 2.66 0.008 1.30e-06 8.59e-06 poorhh | 8058172 0602259 13.38 0.000 6877767 9238578 location | 153051 0729618 2.10 0.036 0100486 2960534 distance | 0005112 0017361 0.29 0.768 -.0028914 0039138 commune135 | 0709772 0744973 0.95 0.341 -.0750348 2169893 road1 | 1715693 075999 2.26 0.024 022614 3205247 post | -.1410493 0788622 -1.79 0.074 -.2956163 0135176 _cons | -1.44638 2523856 -5.73 0.000 -1.941047 -.9517136 -Note: failures and successes completely determined estat class Probit model for credit True -Classified | D ~D | Total -+ + + | 134 127 | 261 | 653 2701 | 3354 -+ + Total | 787 2828 | 3615 Classified + if predicted Pr(D) >= True D defined as credit != -Sensitivity Pr( +| D) 17.03% Specificity Pr( -|~D) 95.51% Positive predictive value Pr( D| +) 51.34% Negative predictive value Pr(~D| -) 80.53% -False + rate for true ~D Pr( +|~D) 4.49% False - rate for true D Pr( -| D) 82.97% False + rate for classified + Pr(~D| +) 48.66% False - rate for classified Pr( D| -) 19.47% -Correctly classified 78.42% 58 reg credit age age2 gender edu mar ostatus hsize depend hvalue land poorhh location distance commune135 road1 post Source | SS df MS -+ -Model | 68.6356463 16 4.28972789 Residual | 547.031297 3598 152037603 -+ -Total | 615.666943 3614 170356099 Number of obs F( 16, 3598) Prob > F R-squared Adj R-squared Root MSE = = = = = = 3615 28.21 0.0000 0.1115 0.1075 38992 -credit | Coef Std Err t P>|t| [95% Conf Interval] -+ -age | 0041349 0028602 1.45 0.148 -.0014728 0097427 age2 | -.0000624 000027 -2.31 0.021 -.0001154 -9.33e-06 gender | 0128653 019048 0.68 0.499 -.0244807 0502113 edu | 0072664 0065251 1.11 0.266 -.0055269 0200597 mar | 0655079 0541777 1.21 0.227 -.0407141 17173 ostatus | 0979731 0345291 2.84 0.005 0302746 1656716 hsize | 0191772 0047159 4.07 0.000 0099311 0284232 depend | -.027265 0107043 -2.55 0.011 -.0482521 -.0062779 hvalue | -1.88e-08 2.29e-08 -0.82 0.412 -6.36e-08 2.61e-08 land | 2.54e-07 1.04e-07 2.43 0.015 4.88e-08 4.58e-07 poorhh | 2591004 017334 14.95 0.000 225115 2930859 location | 0541649 0195195 2.77 0.006 0158944 0924353 distance | 0003989 0004994 0.80 0.424 -.0005801 001378 commune135 | 0257609 0204702 1.26 0.208 -.0143734 0658952 road1 | 0442473 0198265 2.23 0.026 0053751 0831196 post | -.0415691 0215907 -1.93 0.054 -.0839003 0007621 _cons | -.0568966 094436 -0.60 0.547 -.24205 1282569 vif Variable | VIF 1/VIF -+ -age | 40.05 0.024971 age2 | 39.73 0.025172 location | 2.02 0.496110 commune135 | 1.98 0.505998 hsize | 1.34 0.745146 distance | 1.33 0.751157 edu | 1.27 0.789041 gender | 1.24 0.804510 depend | 1.22 0.820016 poorhh | 1.15 0.867660 ostatus | 1.08 0.922302 mar | 1.08 0.923289 hvalue | 1.08 0.929680 road1 | 1.07 0.935771 land | 1.03 0.973080 post | 1.02 0.981999 -+ -Mean VIF | 6.10 59 5.3 Impact evaluation using PSM methods 5.3.1 Balancing Test (p.36-38) pscore credit age age2 gender edu mar ostatus hsize depend hvalue land poorhh location distance commune135 road1 post [pw= wt9], pscore(model1) blockid(blockf1) comsup level (0.005) numblo(5) (0 real changes made) **************************************************** Algorithm to estimate the propensity score **************************************************** The treatment is credit 1= Credit | Participati | on; 0=NOT | Freq Percent Cum + | 2,828 78.23 78.23 | 787 21.77 100.00 + Total | 3,615 100.00 Estimation of the propensity score (sum of wgt is Iteration 0: Iteration 1: Iteration 2: Iteration 3: Iteration 4: 8.0301e+06) log pseudolikelihood log pseudolikelihood log pseudolikelihood log pseudolikelihood log pseudolikelihood = = = = = -1821.7256 -1641.6488 -1637.5558 -1636.115 -1636.1146 Probit regression Number of obs Wald chi2(16) Prob > chi2 Pseudo R2 Log pseudolikelihood = -1636.1146 = = = = 3615 303.25 0.0000 0.1019 -| Robust credit | Coef Std Err z P>|z| [95% Conf Interval] -+ -age | 0291931 0156673 1.86 0.062 -.0015143 0599005 age2 | -.0003785 000161 -2.35 0.019 -.0006941 -.0000629 gender | 0404641 0799139 0.51 0.613 -.1161643 1970926 edu | 038901 0253568 1.53 0.125 -.0107974 0885995 mar | 2926973 2141651 1.37 0.172 -.1270586 7124533 ostatus | 3015596 126513 2.38 0.017 0535986 5495205 hsize | 0596352 0195285 3.05 0.002 0213601 0979103 depend | -.0681167 0433297 -1.57 0.116 -.1530413 0168079 hvalue | -2.04e-07 1.13e-07 -1.81 0.070 -4.25e-07 1.70e-08 land | 5.67e-06 2.09e-06 2.71 0.007 1.57e-06 9.77e-06 poorhh | 8269005 0659447 12.54 0.000 6976513 9561497 location | 148289 0807213 1.84 0.066 -.0099219 3064999 distance | 0019216 0017431 1.10 0.270 -.0014948 005338 commune135 | 0663608 0807308 0.82 0.411 -.0918687 2245902 road1 | 1325476 0828113 1.60 0.109 -.0297595 2948547 post | -.1165129 0828194 -1.41 0.159 -.2788359 0458102 _cons | -2.196268 4354092 -5.04 0.000 -3.049654 -1.342881 -Note: failures and successes completely determined 60 Note: the common support option has been selected The region of common support is [.01227176, 1] Description of the estimated propensity score in region of common support Estimated propensity score Percentiles Smallest 1% 0266434 0122718 5% 058233 012369 10% 0901394 0124644 Obs 3608 25% 1295794 012883 Sum of Wgt 3608 50% 75% 90% 95% 99% 1639496 2469692 4561517 522389 6307957 Largest 909662 9882353 1 Mean Std Dev .2143825 141199 Variance Skewness Kurtosis 0199372 1.478109 4.933195 ****************************************************** Step 1: 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 blocks ********************************************************** Step 2: Test of balancing property of the propensity score Use option detail if you want more detailed output ********************************************************** Variable ostatus is not balanced in block Variable commune135 is not balanced in block The balancing property is not satisfied Try a different specification of the propensity score Inferior | 1= Credit of block | Participation; 0=NOT of pscore | | Total -+ + -0 | 422 30 | 452 | 1,608 270 | 1,878 | 547 221 | 768 | 222 229 | 451 | 21 33 | 54 | | -+ + -Total | 2,821 787 | 3,608 Note: the common support option has been selected ******************************************* End of the algorithm to estimate the pscore ******************************************* 61 pscore credit age age2 gender edu mar hsize depend hvalue land poorhh location distance road1 post [pw= wt9], pscore(model2) blockid(blockf2) comsup level (0.005) numblo(5) (0 real changes made) **************************************************** Algorithm to estimate the propensity score **************************************************** The treatment is credit 1= Credit | Participati | on; 0=NOT | Freq Percent Cum + | 2,828 78.23 78.23 | 787 21.77 100.00 + Total | 3,615 100.00 Estimation of the propensity score (sum of wgt is Iteration 0: Iteration 1: Iteration 2: Iteration 3: Iteration 4: 8.0301e+06) log pseudolikelihood log pseudolikelihood log pseudolikelihood log pseudolikelihood log pseudolikelihood = = = = = -1821.7256 -1644.8662 -1640.7612 -1639.2788 -1639.2784 Probit regression Number of obs Wald chi2(14) Prob > chi2 Pseudo R2 Log pseudolikelihood = -1639.2784 = = = = 3615 297.74 0.0000 0.1002 -| Robust credit | Coef Std Err z P>|z| [95% Conf Interval] -+ -age | 0292295 0155035 1.89 0.059 -.0011567 0596157 age2 | -.0003761 0001593 -2.36 0.018 -.0006884 -.0000639 gender | 0356974 0796165 0.45 0.654 -.1203479 1917428 edu | 0540313 0244462 2.21 0.027 0061176 101945 mar | 2946141 2143532 1.37 0.169 -.1255104 7147386 hsize | 0648543 0192746 3.36 0.001 0270767 1026319 depend | -.0732395 0432512 -1.69 0.090 -.1580102 0115313 hvalue | -2.00e-07 1.14e-07 -1.76 0.079 -4.23e-07 2.32e-08 land | 5.80e-06 2.10e-06 2.77 0.006 1.70e-06 9.91e-06 poorhh | 829667 0654989 12.67 0.000 7012915 9580424 location | 1919181 0667914 2.87 0.004 0610094 3228267 distance | 0021811 001743 1.25 0.211 -.0012351 0055974 road1 | 1283144 0829011 1.55 0.122 -.0341687 2907975 post | -.1212529 0828228 -1.46 0.143 -.2835827 0410769 _cons | -2.21936 4332246 -5.12 0.000 -3.068465 -1.370255 -Note: failures and successes completely determined 62 Note: the common support option has been selected The region of common support is [.01221087, 1] Description of the estimated propensity score in region of common support Estimated propensity score Percentiles Smallest 1% 0271312 0122109 5% 0577847 0124028 10% 0905254 0126613 Obs 3608 25% 130523 0144201 Sum of Wgt 3608 50% 75% 90% 95% 99% 165646 2434986 4575583 5193872 6262064 Largest 9182045 98944 1 Mean Std Dev .2140685 1401231 Variance Skewness Kurtosis 0196345 1.509062 5.082356 ****************************************************** Step 1: 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 blocks ********************************************************** 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 of treated and the number of controls for each block Inferior | 1= Credit of block | Participation; 0=NOT of pscore | | Total -+ + -.0122109 | 410 32 | 442 | 1,615 270 | 1,885 | 557 225 | 782 | 220 221 | 441 | 18 35 | 53 | | -+ + -Total | 2,821 787 | 3,608 Note: the common support option has been selected ******************************************* End of the algorithm to estimate the pscore ******************************************* 63 Table 5.6: Impact of rural credit using NN technique (p.39) attnd educost credit, pscore(model2) comsup boot The program is searching the nearest neighbor of each treated unit This operation may take a while ATT estimation with Nearest Neighbor Matching method (random draw version) Analytical standard errors n treat n contr ATT Std Err t 787 588 625.058 200.750 3.114 Note: the numbers of treated and controls refer to actual nearest neighbour matches Bootstrapping of standard errors command: statistic: attnd educost credit , pscore(model2) comsup attnd = r(attnd) Bootstrap statistics Number of obs Replications = = 3615 50 -Variable | Reps Observed Bias Std Err [95% Conf Interval] -+ -attnd | 50 625.0585 -148.664 228.3745 166.1228 1083.994 (N) | 117.3355 970.7469 (P) | 336.3713 984.0334 (BC) -Note: N = normal P = percentile BC = bias-corrected ATT estimation with Nearest Neighbor Matching method (random draw version) Bootstrapped standard errors n treat n contr ATT Std Err t 787 588 625.058 246.931 2.531 Note: the numbers of treated and controls refer to actual nearest neighbour matches attnd health credit, pscore(model2) comsup boot The program is searching the nearest neighbor of each treated unit This operation may take a while ATT estimation with Nearest Neighbor Matching method (random draw version) Analytical standard errors n treat n contr ATT Std Err t 787 588 0.750 0.402 1.863 - 64 Note: the numbers of treated and controls refer to actual nearest neighbour matches Bootstrapping of standard errors command: statistic: attnd health credit , pscore(model2) comsup attnd = r(attnd) Bootstrap statistics Number of obs Replications = = 3615 50 -Variable | Reps Observed Bias Std Err [95% Conf Interval] -+ -attnd | 50 7496824 -.1771286 44805 -.1507078 1.650073 (N) | -.254522 1.442536 (P) | 0456274 1.612903 (BC) -Note: N = normal P = percentile BC = bias-corrected ATT estimation with Nearest Neighbor Matching method (random draw version) Bootstrapped standard errors n treat n contr ATT Std Err t 787 588 0.750 0.520 1.441 Note: the numbers of treated and controls refer to actual nearest neighbour matches attnd waterexp credit, pscore(model2) comsup boot The program is searching the nearest neighbor of each treated unit This operation may take a while ATT estimation with Nearest Neighbor Matching method (random draw version) Analytical standard errors n treat n contr ATT Std Err t 787 588 6.227 4.660 1.336 Note: the numbers of treated and controls refer to actual nearest neighbour matches Bootstrapping of standard errors command: statistic: attnd waterexp credit , pscore(model2) comsup attnd = r(attnd) Bootstrap statistics Number of obs Replications = = 3615 50 -Variable | Reps Observed Bias Std Err [95% Conf Interval] -+ -attnd | 50 6.227446 -2.320445 4.586969 -2.990412 15.4453 (N) | -4.228356 12.11467 (P) | 175853 12.23774 (BC) 65 Note: N P BC = normal = percentile = bias-corrected ATT estimation with Nearest Neighbor Matching method (random draw version) Bootstrapped standard errors n treat n contr ATT Std Err t 787 588 6.227 4.638 1.343 Note: the numbers of treated and controls refer to actual nearest neighbour matches attnd watertap credit, pscore(model2) comsup boot The program is searching the nearest neighbor of each treated unit This operation may take a while ATT estimation with Nearest Neighbor Matching method (random draw version) Analytical standard errors n treat n contr ATT Std Err t 787 588 0.010 0.011 0.951 Note: the numbers of treated and controls refer to actual nearest neighbour matches Bootstrapping of standard errors command: statistic: attnd watertap credit , pscore(model2) comsup attnd = r(attnd) Bootstrap statistics Number of obs Replications = = 3615 50 -Variable | Reps Observed Bias Std Err [95% Conf Interval] -+ -attnd | 50 0101652 -.0010586 0131473 -.0162553 0365857 (N) | -.0166667 035443 (P) | -.0166667 035443 (BC) -Note: N = normal P = percentile BC = bias-corrected ATT estimation with Nearest Neighbor Matching method (random draw version) Bootstrapped standard errors n treat n contr ATT Std Err t 787 588 0.010 0.012 0.859 Note: the numbers of treated and controls refer to actual nearest neighbour matches 66 Table 5.7: Impact of rural credit using Stratification technique (p.40) atts educost credit, pscore(model2) blockid(blockf2) comsup boot ATT estimation with the Stratification method Analytical standard errors n treat n contr ATT Std Err t 785 2823 425.017 Bootstrapping of standard errors command: statistic: atts educost credit , pscore(model2) blockid(blockf2) comsup atts = r(atts) Bootstrap statistics Number of obs Replications = = 3615 50 -Variable | Reps Observed Bias Std Err [95% Conf Interval] -+ -atts | 50 425.017 3.520983 167.7136 87.98392 762.05 (N) | 130.9821 755.999 (P) | 130.9821 755.999 (BC) -Note: N = normal P = percentile BC = bias-corrected ATT estimation with the Stratification method Bootstrapped standard errors n treat n contr ATT Std Err t 785 2823 425.017 147.942 2.873 - atts health credit, pscore(model2) blockid(blockf2) comsup boot ATT estimation with the Stratification method Analytical standard errors n treat n contr ATT Std Err t 785 2823 0.457 - 67 Bootstrapping of standard errors command: statistic: atts health credit , pscore(model2) blockid(blockf2) comsup atts = r(atts) Bootstrap statistics Number of obs Replications = = 3615 50 -Variable | Reps Observed Bias Std Err [95% Conf Interval] -+ -atts | 50 4568533 -.0234709 3282342 -.2027579 1.116465 (N) | -.2150077 1.050182 (P) | -.1384337 1.168713 (BC) -Note: N = normal P = percentile BC = bias-corrected ATT estimation with the Stratification method Bootstrapped standard errors n treat n contr ATT Std Err t 785 2823 0.457 0.357 1.279 - atts waterexp credit, pscore(model2) blockid(blockf2) comsup boot ATT estimation with the Stratification method Analytical standard errors n treat n contr ATT Std Err t 785 2823 4.341 - Bootstrapping of standard errors command: statistic: atts waterexp credit , pscore(model2) blockid(blockf2) comsup atts = r(atts) Bootstrap statistics Number of obs Replications = = 3615 50 -Variable | Reps Observed Bias Std Err [95% Conf Interval] -+ -atts | 50 4.341479 -.6296841 3.208877 -2.107 10.78996 (N) | -1.569006 9.339189 (P) | -.356226 9.875525 (BC) -Note: N = normal P = percentile BC = bias-corrected 68 ATT estimation with the Stratification method Bootstrapped standard errors n treat n contr ATT Std Err t 785 2823 4.341 3.686 1.178 - atts watertap credit, pscore(model2) blockid(blockf2) comsup boot ATT estimation with the Stratification method Analytical standard errors n treat n contr ATT Std Err t 785 2823 0.009 Bootstrapping of standard errors command: statistic: atts watertap credit , pscore(model2) blockid(blockf2) comsup atts = r(atts) Bootstrap statistics Number of obs Replications = = 3615 50 -Variable | Reps Observed Bias Std Err [95% Conf Interval] -+ -atts | 50 0091433 9.12e-06 0078682 -.0066684 024955 (N) | -.0108599 021586 (P) | -.0111862 0194355 (BC) -Note: N = normal P = percentile BC = bias-corrected ATT estimation with the Stratification method Bootstrapped standard errors n treat n contr ATT Std Err t 785 2823 0.009 0.009 0.983 - 69 Table 5.8: Impact of rural credit using Kernel Matching technique (p.41) attk educost credit, pscore(model2) comsup bootstrap reps(50) The program is searching for matches of each treated unit This operation may take a while ATT estimation with the Kernel Matching method n treat n contr ATT Std Err t 787 2821 374.196 Note: Analytical standard errors cannot be computed Use the bootstrap option to get bootstrapped standard errors Bootstrapping of standard errors command: statistic: attk educost credit , pscore(model2) comsup bwidth(.06) attk = r(attk) Bootstrap statistics Number of obs Replications = = 3615 50 -Variable | Reps Observed Bias Std Err [95% Conf Interval] -+ -attk | 50 374.1959 11.80717 172.5303 27.48328 720.9084 (N) | 121.8368 719.9103 (P) | 121.8368 719.9103 (BC) -Note: N = normal P = percentile BC = bias-corrected ATT estimation with the Kernel Matching method Bootstrapped standard errors n treat n contr ATT Std Err t 787 2821 374.196 171.210 2.186 - attk health credit, pscore(model2) comsup bootstrap reps(50) The program is searching for matches of each treated unit This operation may take a while ATT estimation with the Kernel Matching method n treat n contr ATT Std Err t 787 2821 0.460 Note: Analytical standard errors cannot be computed Use the bootstrap option to get bootstrapped standard errors Bootstrapping of standard errors 70 command: statistic: attk health credit , pscore(model2) comsup bwidth(.06) attk = r(attk) Bootstrap statistics Number of obs Replications = = 3615 50 -Variable | Reps Observed Bias Std Err [95% Conf Interval] -+ -attk | 50 4603996 0002455 3384209 -.2196827 1.140482 (N) | -.3022202 1.094709 (P) | -.3320419 1.094709 (BC) -Note: N = normal P = percentile BC = bias-corrected ATT estimation with the Kernel Matching method Bootstrapped standard errors n treat n contr ATT Std Err t 787 2821 0.460 0.326 1.414 - attk waterexp credit, pscore(model2) comsup bootstrap reps(50) The program is searching for matches of each treated unit This operation may take a while ATT estimation with the Kernel Matching method n treat n contr ATT Std Err t 787 2821 3.655 Note: Analytical standard errors cannot be computed Use the bootstrap option to get bootstrapped standard errors Bootstrapping of standard errors command: statistic: attk waterexp credit , pscore(model2) comsup bwidth(.06) attk = r(attk) Bootstrap statistics Number of obs Replications = = 3615 50 -Variable | Reps Observed Bias Std Err [95% Conf Interval] -+ -attk | 50 3.655444 1517648 3.917 -4.216062 11.52695 (N) | -3.914149 9.29926 (P) | -8.084994 9.29926 (BC) -Note: N = normal P = percentile BC = bias-corrected 71 ATT estimation with the Kernel Matching method Bootstrapped standard errors n treat n contr ATT Std Err t 787 2821 3.655 3.667 0.997 - attk watertap credit, pscore(model2) comsup bootstrap reps(50) The program is searching for matches of each treated unit This operation may take a while ATT estimation with the Kernel Matching method n treat n contr ATT Std Err t 787 2821 0.008 Note: Analytical standard errors cannot be computed Use the bootstrap option to get bootstrapped standard errors Bootstrapping of standard errors command: statistic: attk watertap credit , pscore(model2) comsup bwidth(.06) attk = r(attk) Bootstrap statistics Number of obs Replications = = 3615 50 -Variable | Reps Observed Bias Std Err [95% Conf Interval] -+ -attk | 50 0077894 0010174 0090333 -.0103637 0259424 (N) | -.0066754 0280448 (P) | -.0066754 0280448 (BC) -Note: N = normal P = percentile BC = bias-corrected ATT estimation with the Kernel Matching method Bootstrapped standard errors n treat n contr ATT Std Err t 787 2821 0.008 0.009 0.893 - *** 72 ... OF RURAL CREDIT AND ACCESS TO EDUCATION, HEALTH CARE, CLEAN WATER IN VIETNAM 4.1 Rural credit market in Vietnam 4.1.1 The formal credit sector ? ?In Vietnam, formal rural credit is defined as the. .. briefs the overviews of rural credit market in Vietnam and the current situation of living standard indicators such as education, health care and clean water Chapter V presents the results of the. .. in the researches on evaluating the impact of a program (rural credit) on the outcome (income/consumption) includes two main stages: (1) Investigate the determinants on the participation in credit