(Luận văn) the effect of financial development on economic growth, evidence from asian countries

UNIVERSITY OF ECONOMICS INSTITUTE OF SOCIAL STUDIES HO CHI MINH CITY THE HAGUE t to VIETNAM THE NETHERLANDS ng hi ep VIETNAM - NETHERLANDS w n PROGRAMME FOR M.A IN DEVELOPMENT ECONOMICS lo ad ju y th yi pl THE EFFECT OF FINANCIAL DEVELOPMENT al n ua ON ECONOMIC GROWTH: EVIDENCE FROM va n ASIAN COUNTRIES ll fu oi m at nh z BY z k jm ht vb TRẦN THANH GIANG l.c gm MASTER OF ARTS IN DEVELOPMENT ECONOMICS om an Lu n va ey t re HO CHI MINH CITY, JULY 2014 UNIVERSITY OF ECONOMICS INSTITUTE OF SOCIAL STUDIES HO CHI MINH CITY THE HAGUE t to VIETNAM THE NETHERLANDS ng hi ep VIETNAM - NETHERLANDS w n PROGRAMME FOR M.A IN DEVELOPMENT ECONOMICS lo ad ju y th yi pl n ua al THE EFFECT OF FINANCIAL DEVELOPMENT n va ON ECONOMIC GROWTH: EVIDENCE FROM fu ll ASIAN COUNTRIES oi m at nh z A thesis submitted in partial fulfilment of the requirements for the degree of z By k jm ht vb MASTER OF ARTS IN DEVELOPMENT ECONOMICS om l.c gm TRẦN THANH GIANG ASSOC PROF DR NGUYỄN VĂN NGÃI an Lu Academic Supervisor: n va ey t re HO CHI MINH CITY, JULY 2014 ii t to DECLARATION ng hi ep This is to certify that this thesis entitled “The effect of financial development w n on economic growth: evidence from Asian countries”, which is submitted by me in lo ad fulfillment of the requirements for the degree of Master of Art in Development ju y th Economic to the Vietnam – The Netherlands Programme The thesis constitutes only my original work and due supervision and acknowledgement have been made yi pl in the text to all materials used al n ua Trần Thanh Giang n va ll fu oi m at nh z z k jm ht vb om l.c gm an Lu n va ey t re iii ACKNOWLEGEMENT t to ng hi I would not be able to write and finish my dissertation without the help and support ep of people surrounding me w Above all, I would like to express my greatest appreciation to my supervisor, Assoc n lo Prof Nguyễn Văn Ngãi, for his invaluable comments and advices, patient guidance, ad encouragement in during the time of doing this thesis I have been strikingly lucky y th ju to have supervisor who cared so much my thesis and answered to all my questions yi Without his guidance, my thesis would not have been possible pl al I would also like to offer my special thanks to Dr Trương Đăng Thụy and Dr Phạm n va develop this thesis n ua Khánh Nam for the econometric guidance and valuable suggestions that help to ll fu Besides my mentors, special thanks also to all the lecturers at the Vietnam – at the program oi m Netherlands Program for their knowledge of all the course, during the time I studied nh at In addition, I would like to thank my friends and people who are always beside me z z and support for my thesis but are not above mentioned vb Last, but not least, I am very deeply grateful to my family Without their warm ht k jm encouragement and attention, I would not be possible to complete this dissertation om l.c gm an Lu n va ey t re iv ABBREVIATIONS t to WB World Bank ng hi OECD Organization for Economic Cooperation and Development ep MENA Countries in the Middle East and North Africa w GLS Generalized Least Squares n lo OLS Ordinary Least Squares ad FEM Fix Effects Model y th ju REM Random Effects Model yi GMM The Generalized Method of Moments Estimation pl n ua al n va ll fu oi m at nh z z k jm ht vb om l.c gm an Lu n va ey t re v ABSTRACT t to ng hi This study estimates the effect of financial development on economic growth in ep Asian countries in the period from 2000 to 2011 Based on unbalanced panel data, w this effect is examined by Fixed effects model (FEM) and the first difference n lo Generalizes Methods of Moments approach (GMM) The findings indicate that ad financial development has significant impacts on economic growth on both y th estimation techniques However, these impacts depend significantly on estimation ju yi methods and proxies for financial development The results of FEM and first pl difference GMM imply that financial depth and domestic credit to private sector al n ua have negative impact on growth, but there is no relationship between stock market va development and economic growth On the other hand, while a positive relationship n between the ratio of commercial – central bank assets and growth rate of real GDP fu ll per capita is shown by FEM, this indicator is not related to growth rates in GMM oi m results nh at Key words: Financial development, Economic growth, relationship, effect, z endogeneity, fixed effects, random effects, Asian countries z k jm ht vb om l.c gm an Lu n va ey t re vi TABLE OF CONTENTS LIST OF TABLES ix t to LIST OF FIGURES x ng hi CHAPTER 1: INTRODUCTION ep Problem statements 1.1 w n 1.2 Research objectives lo ad 1.3 Research scope and data y th 1.4 Research structure ju yi CHAPTER 2: LTERATURE REVIEW pl ua al 2.1 Theoretical literature n 2.1.1 Endogenous growth theory va n 2.1.2 Theories of financial development fu ll 2.2 Empirical studies 16 oi m CHAPTER 3: RESEARCH METHODOLOGY 29 nh at 3.1 Model Specification 29 z z 3.2 Measurements of Variables 31 vb jm ht 3.2.1 Measurements of financial development 31 k 3.2.2 The determinants of economic growth 34 gm l.c 3.3 Data collection 39 om 3.4 Research methodology 39 an Lu 3.4.1The common constant method (Pooled OLS) 40 3.4.2 The random effects method (REM) 41 ey vii t re 3.4.5 The generalized method of moments estimation (GMM) 45 n 3.4.4 Choice of panel regression model 42 va 3.4.3 The Fixed effects method (FEM) 41 CHAPTER 4: RESEARCH RESULTS 48 4.1 Overview the economic growth and the financial development in the regions of t to Asia 48 ng hi 4.1.1 Overview the economic growth in the regions of Asia in the period 2000 - 2011: ep 48 w 4.1.2 Overview the financial development in the regions of Asia in 2000 - 2011 50 n lo ad 4.2 The descriptive statistic of the sample 57 ju y th 4.3 Empirical results 62 yi 4.3.1 Results of tests for panel regression model 62 pl ua al 4.3.2 Discussions on the research results 65 n 4.3.3 Discussions on the results of first difference GMM 69 va CHAPTER 5: CONCLUSION AND POLICY IMPLICATION 75 n fu ll 5.1 Conclusions 75 m oi 5.2 Policy implications 77 nh at 5.3 Research limitations 78 z z 5.4 Suggestions for further research 79 vb jm ht REFERENCES 81 APPENDIX A 85 k gm APPENDIX B 91 l.c APPENDIX C: DESCRIPTIVE STATISTIC OF VARIABLE 94 om an Lu APPENDIX D: PANEL REGRESSION MODEL 96 APPENDIX E: RESULTS OF BREUSCH – PAGAN LM TEST 102 viii ey GMM 105 t re APPENDIX G: THE REGRESSION MODEL RESULTS OF FIRST – DIFFERENCE n va APPENDIX F: RESULTS OF HAUSMAN TEST 103 t to LIST OF TABLES ng hi ep w Table 3.1: The expected sign of variables in model 38 n lo Table 3.2: Tests for choosing a panel regression model 44 ad Table 4.1 Descriptive statistics of the sample observation 58 y th ju Table 4.2: The corrrelation on the sample observations 61 yi Table 4.3: The results of F test and Breusch – Pagan test 63 pl ua al Table 4.4: The results of Hausman test 63 Table 4.5: The results of FEM regression model 64 n n va Table 4.6: The results of first difference GMM 70 ll fu oi m at nh z z k jm ht vb om l.c gm an Lu n va ey t re ix LIST OF FIGURES t to Figure 2.1: The role of financial development in economic growth 27 ng hi Figure 3.1: Analytical framework 37 ep Figure 4.1: The average growth rate of real GDP per capita in Asia regions in 2000 w – 2011 50 n lo Figure 4.2: Financial development in Central Asia 51 ad Figure 4.3: Financial development in South - East Asia 51 y th ju Figure 4.4: Financial development in South Asia 52 yi Figure 4.5: Financial development in Eastern Asia 53 pl ua al Figure 4.6: Financial development in Western Asia 54 Figure 4.7: The ratio of liquid liabilities to GDP across Asia regions 55 n n va Figure 4.8: The ratio of domestic credit to private sector to GDP across Asia regions fu 55 ll Figure 4.9: The ratio of commercial – central bank assets across Asia regions oi m 56 nh at Figure 4.10: The ratio of stock market capitalization to GDP across Asia regions z z 56 vb Figure 4.11: The scatter diagram among dependent variable and financial ht k jm development variables 59 gm Figure 4.12: The scatter diagram among dependent variable and control variables om l.c 60 an Lu n va ey t re x Figure B.5: Growth rate of real GDP per capita in Central Asia (%) t to Growth rate of real GDP per capita in Central Asia (%) ng hi 40 ep 30 20 w 10 n 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 lo -10 ad -20 Azerbaijan Georgia ju y th Armenia Kyrgyz Republic Tajikistan Kazakhstan yi pl Turkey n ua al va n Figure B.6: Financial development in Asian developing countries ll fu 100 at nh 90 oi m Financial development in developing countries 80 Ratio of Liquid liabilities to GDP z z 70 vb 60 Ratio of Commercial central bank assets jm ht 50 40 Ratio of domestic credit to private sector to GDP k 30 gm 20 l.c Ratio of stock market capatilization to GDP 10 om an Lu n va ey t re 93 Figure B.7: Financial development in Asian developed countries t to Financial development in developed countries 160 ng hi 140 ep Ratio of Liquid liabilities to GDP 120 100 w Ratio of Commercial central bank assets n 80 lo 60 ad 40 Ratio of domestic credit to private sector to GDP y th 20 ju Ratio of stock market capatilization to GDP yi pl n ua al va n APPENDIX C: DESCRIPTIVE STATISTIC OF VARIABLE fu ll Figure C.1: Data distribution figures of Financial development indicators 015 m oi Density 05 z 0 200 300 20 40 DEPTH 80 015 Density 005 01 an Lu 50 150 200 200 400 STOCK 94 600 ey 100 CREDIT t re n va Density 01 om 015 l.c 005 100 gm 02 60 BANK k 100 jm ht vb 005 z Density at 01 nh Density w 05 02 n 0 lo ad -20 -10 GROWTH_1 10 -20 20 ju 08 20 inf 40 60 015 y th yi 01 n 005 04 Density ua al Density 06 pl 02 va 15 GOV 20 25 0 30 100 200 ll 10 fu n 300 400 TO oi m at 025 nh 02 z 015 vb gm 005 k jm 01 ht Density z 20 40 60 EDU 80 l.c ep Density hi 04 ng 15 t to 06 Figure C.2: Data distribution figures of control variables 100 om an Lu n va ey t re 95 APPENDIX D: PANEL REGRESSION MODEL t to ng hi Estimation results of model ep Figure D.1 -1: The pooled OLS model w n lo Source ad df 575.892541 5041.8858 ju y th Model Residual SS Total 5617.77834 MS 328 115.178508 15.3716031 333 16.8702052 Number of obs F( 5, 328) Prob > F R-squared Adj R-squared Root MSE = = = = = = 334 7.49 0.0000 0.1025 0.0888 3.9207 yi GROWTH pl DEPTH INF GOV TO EDU _cons -.0096109 0314105 -.1962185 0057002 0095793 5.998403 Coef Std Err t n ua al 0047422 02727 0483809 0040521 0122061 8767924 P>|t| n va 0.044 0.250 0.000 0.160 0.433 0.000 -.0189399 -.0222357 -.2913945 -.0022713 -.0144329 4.273557 -.0002819 0850568 -.1010425 0136716 0335914 7.723249 ll fu -2.03 1.15 -4.06 1.41 0.78 6.84 [95% Conf Interval] oi m at nh Figure D.1-2: The FEM model z z vb Number of obs Number of groups = = 334 32 R-sq: Obs per group: = avg = max = 10.4 12 jm ht Fixed-effects (within) regression Group variable: id F(5,297) Prob > F = -0.8124 t P>|t| DEPTH INF GOV TO EDU _cons -.0459941 0181456 -.2912321 0668945 -.0011982 4.796717 0190295 0282139 1067523 0136771 0377244 3.269838 sigma_u sigma_e rho 4.1115071 3.3522938 60067769 (fraction of variance due to u_i) -2.42 0.64 -2.73 4.89 -0.03 1.47 0.016 0.521 0.007 0.000 0.975 0.143 [95% Conf Interval] -.0834439 -.0373788 -.5013189 0399782 -.0754391 -1.638271 -.0085444 0736701 -.0811452 0938109 0730428 11.23171 n va Coef an Lu GROWTH 8.57 0.0000 om Std Err = = l.c gm corr(u_i, Xb) k within = 0.1261 between = 0.0828 overall = 0.0559 4.89 96 ey F(31, 297) = t re F test that all u_i=0: Prob > F = 0.0000 Figure D.1-3: The REM MODEL t to ng hi Random-effects GLS regression Group variable: id Number of obs Number of groups = = 334 32 R-sq: Obs per group: = avg = max = 10.4 12 ep within = 0.0978 between = 0.1588 overall = 0.0913 Random effects u_i ~ Gaussian corr(u_i, X) = (assumed) w Wald chi2(5) Prob > chi2 = = 29.55 0.0000 n lo GROWTH ad -.0174571 0393201 -.2218614 0184767 0139279 5.25352 ju y th DEPTH INF GOV TO EDU _cons Coef yi z 0082673 0269725 0712008 0071266 019114 1.503053 -2.11 1.46 -3.12 2.59 0.73 3.50 P>|z| [95% Conf Interval] 0.035 0.145 0.002 0.010 0.466 0.000 -.0336606 -.0135451 -.3614125 0045087 -.0235349 2.307591 -.0012535 0921853 -.0823104 0324446 0513906 8.199448 pl 2.053621 3.3522938 2728758 (fraction of variance due to u_i) n ua al sigma_u sigma_e rho Std Err n va ll fu m oi Estimation results of model at nh Figure D.2-1: The pooled OLS model z z SS df ht vb Source MS 104.408579 15.0967025 Total 4975.57012 300 16.5852337 0107968 0285552 -.2480927 0080281 0155202 4.60912 0177247 0274753 0506151 0056802 0125624 1.706022 t 0.61 1.04 -4.90 1.41 1.24 2.70 P>|t| 0.543 0.300 0.000 0.159 0.218 0.007 [95% Conf Interval] -.0240862 -.0255171 -.347705 -.0031507 -.0092032 1.251604 0456798 0826276 -.1484803 0192069 0402436 7.966637 an Lu BANK INF GOV TO EDU _cons Std Err 301 6.92 0.0000 0.1049 0.0898 3.8854 om Coef = = = = = = l.c GROWTH gm 295 k 522.042893 4453.52723 jm Model Residual Number of obs F( 5, 295) Prob > F R-squared Adj R-squared Root MSE n va ey t re 97 Figure D.2-2: The FEM MODEL t to ng Fixed-effects (within) regression Group variable: id Number of obs Number of groups = = 301 30 R-sq: Obs per group: = avg = max = 10.0 12 hi within = 0.1656 between = 0.0842 overall = 0.0689 ep corr(u_i, Xb) F(5,266) Prob > F = -0.8498 = = 10.56 0.0000 w n Coef BANK INF GOV TO EDU _cons 0712947 -.022958 -.6250647 0760589 -.0589488 4.207273 lo GROWTH Std Err ad ju y th 0301948 0295659 1197558 0163898 0377705 3.383438 P>|t| 2.36 -0.78 -5.22 4.64 -1.56 1.24 0.019 0.438 0.000 0.000 0.120 0.215 [95% Conf Interval] 0118435 -.0811709 -.8608545 0437886 -.1333159 -2.454454 1307459 0352549 -.3892749 1083291 0154184 10.869 yi 4.7597486 3.2010708 68856496 pl (fraction of variance due to u_i) F test that all u_i=0: ua al sigma_u sigma_e rho t F(29, 266) = 5.81 Prob > F = 0.0000 n n va ll fu Figure D.2-3: The REM MODEL Number of obs Number of groups = = 301 30 Obs per group: = avg = max = 10.0 12 at nh R-sq: oi m Random-effects GLS regression Group variable: id z within = 0.1347 between = 0.1411 overall = 0.0915 z vb = = GROWTH Coef BANK INF GOV TO EDU _cons 0284564 0137372 -.3425755 0273702 0169124 2.700434 0231654 0281362 0765049 0098496 0197052 2.312929 sigma_u sigma_e rho 2.0920598 3.2010708 29929179 (fraction of variance due to u_i) 0738597 0688832 -.1926287 0466751 055534 7.233692 om -.0169469 -.0414087 -.4925223 0080652 -.0217091 -1.832824 l.c 0.219 0.625 0.000 0.005 0.391 0.243 [95% Conf Interval] gm 1.23 0.49 -4.48 2.78 0.86 1.17 P>|z| k z 32.54 0.0000 jm Std Err Wald chi2(5) Prob > chi2 ht Random effects u_i ~ Gaussian corr(u_i, X) = (assumed) an Lu n va ey t re 98 Estimation results of model Figure D.3-1: The pooled OLS model t to ng hi Source SS df MS ep 628.040221 4968.65208 321 125.608044 15.4786669 Total 5596.6923 326 17.1677678 w Model Residual Number of obs F( 5, 321) Prob > F R-squared Adj R-squared Root MSE = = = = = = 327 8.11 0.0000 0.1122 0.0984 3.9343 n lo Coef CREDIT INF GOV TO EDU _cons -.0133036 028056 -.2233368 0039021 0207128 5.753535 ad GROWTH Std Err t ju y th yi 0059683 0277754 0506562 0037007 0132289 8851951 P>|t| -2.23 1.01 -4.41 1.05 1.57 6.50 [95% Conf Interval] 0.027 0.313 0.000 0.292 0.118 0.000 -.0250456 -.0265888 -.3229969 -.0033786 -.0053135 4.012018 -.0015617 0827008 -.1236768 0111828 0467391 7.495052 pl n ua al n va ll fu Figure D.3-2: The FEM MODEL oi m at nh 327 32 Obs per group: = avg = max = 10.2 12 F(5,290) Prob > F = -0.8400 jm ht vb within = 0.1419 between = 0.0880 overall = 0.0572 corr(u_i, Xb) = = z R-sq: Number of obs Number of groups z Fixed-effects (within) regression Group variable: id = = 9.59 0.0000 k t P>|t| 0212231 028273 1078179 013314 0389221 3.339713 sigma_u sigma_e rho 4.5106778 3.3426731 64550832 (fraction of variance due to u_i) -.0355922 0698555 -.0302818 0911899 092574 10.54397 ey t re Prob > F = 0.0000 n 4.99 -.1191339 -.041437 -.4546915 0387815 -.0606373 -2.602329 va F(31, 290) = 0.000 0.616 0.025 0.000 0.682 0.235 an Lu -.0773631 0142092 -.2424867 0649857 0159684 3.97082 om CREDIT INF GOV TO EDU _cons -3.65 0.50 -2.25 4.88 0.41 1.19 [95% Conf Interval] l.c Coef F test that all u_i=0: Std Err gm GROWTH 99 Figure D.3-3: The REM MODEL t to ng Random-effects GLS regression Group variable: id Number of obs Number of groups = = 327 32 R-sq: Obs per group: = avg = max = 10.2 12 hi within = 0.1073 between = 0.1683 overall = 0.0974 ep Random effects u_i ~ Gaussian corr(u_i, X) = (assumed) Wald chi2(5) Prob > chi2 = = 31.28 0.0000 w n GROWTH lo Coef ad P>|z| [95% Conf Interval] 0102772 0272808 0730493 0066426 0203186 1.544365 2.0892602 3.3426731 28091619 (fraction of variance due to u_i) yi -2.63 1.23 -3.06 2.48 1.20 3.20 0.009 0.218 0.002 0.013 0.229 0.001 -.0471214 -.019878 -.3663445 0034212 -.0153713 1.918462 -.0068355 0870606 -.0799966 0294597 0642762 7.972261 pl n ua al sigma_u sigma_e rho z -.0269785 0335913 -.2231705 0164404 0244525 4.945361 ju y th CREDIT INF GOV TO EDU _cons Std Err n va ll fu at nh Figure D.4-1: The pooled OLS model oi m Estimation results of model z SS df MS z Source Total 4352.74964 260 16.7413448 0049619 0306011 0621749 0059755 0168323 1.111186 -1.37 1.56 -3.64 1.00 1.62 4.09 P>|t| 0.171 0.120 0.000 0.319 0.107 0.000 [95% Conf Interval] -.0165799 -.0125418 -.3484623 -.0058014 -.0059567 2.356837 002963 1079845 -.1035789 017734 0603393 6.733378 an Lu -.0068085 0477213 -.2260206 0059663 0271913 4.545108 t om STOCK INF GOV TO EDU _cons Std Err 261 5.20 0.0001 0.0925 0.0747 3.9358 l.c Coef = = = = = = gm GROWTH k 80.5499015 15.4901966 jm 255 ht 402.749508 3950.00013 vb Model Residual Number of obs F( 5, 255) Prob > F R-squared Adj R-squared Root MSE n va ey t re 100 t to Figure D.4-2: The FEM MODEL ng hi ep Number of obs Number of groups = = 261 25 R-sq: Obs per group: = avg = max = 10.4 12 w Fixed-effects (within) regression Group variable: id n lo within = 0.1168 between = 0.0499 overall = 0.0354 ad corr(u_i, Xb) y th GROWTH Coef ju -.0097339 0336772 -.4759169 0522821 0019463 5.808138 pl 009788 0338324 1243896 016863 0462949 4.201582 P>|t| 0.321 0.321 0.000 0.002 0.967 0.168 6.11 0.0000 [95% Conf Interval] -.029019 -.0329823 -.7210002 0190573 -.0892679 -2.470184 0095512 1003367 -.2308336 085507 0931605 14.08646 n (fraction of variance due to u_i) fu F(24, 231) = 4.20 ll F test that all u_i=0: = = va 4.2388812 3.4500397 60152594 t -0.99 1.00 -3.83 3.10 0.04 1.38 n ua al sigma_u sigma_e rho Std Err yi STOCK INF GOV TO EDU _cons F(5,231) Prob > F = -0.8270 Prob > F = 0.0000 oi m at nh z Figure D.4-3: The REM MODEL z vb Number of obs Number of groups = = 261 25 R-sq: Obs per group: = avg = max = 10.4 12 jm ht Random-effects GLS regression Group variable: id k within = 0.0873 between = 0.1530 overall = 0.0869 Wald chi2(5) Prob > chi2 -.0069336 0538534 -.3011605 0109858 0367918 4.297176 0068862 0315876 0847686 008949 0249 1.858546 sigma_u sigma_e rho 1.9587931 3.4500397 24377109 (fraction of variance due to u_i) -.0204302 -.0080571 -.467304 -.006554 -.0120112 6544919 006563 115764 -.135017 0285256 0855948 7.93986 ey t re STOCK INF GOV TO EDU _cons 0.314 0.088 0.000 0.220 0.140 0.021 [95% Conf Interval] n Coef va GROWTH -1.01 1.70 -3.55 1.23 1.48 2.31 P>|z| 22.13 0.0005 an Lu z = = om Std Err l.c gm Random effects u_i ~ Gaussian corr(u_i, X) = (assumed) 101 APPENDIX E: RESULTS OF BREUSCH – PAGAN LM TEST t to Figure E.1: Breusch – Pagan LM Test for Pooled OLS and REM of model 1- ng hi DEPTH ep Breusch and Pagan Lagrangian multiplier test for random effects w GROWTH[id,t] = Xb + u[id] + e[id,t] n lo Estimated results: GROWTH e u Test: Var(u) = ju y th ad Var sd = sqrt(Var) 16.87021 11.23787 4.217359 4.107336 3.352294 2.053621 yi pl chi2(1) = Prob > chi2 = 70.73 0.0000 ua al n Figure E.2: Breusch – Pagan LM Test for Pooled OLS and REM of model - va n BANK fu ll Breusch and Pagan Lagrangian multiplier test for random effects m GROWTH[id,t] = Xb + u[id] + e[id,t] oi Var 16.58523 10.24685 4.376714 4.072497 3.201071 2.09206 z z Var(u) = 86.68 0.0000 jm chi2(1) = Prob > chi2 = ht vb Test: sd = sqrt(Var) at GROWTH e u nh Estimated results: k Figure E.3: Breusch – Pagan LM Test for Pooled OLS and REM of model Breusch and Pagan Lagrangian multiplier test for random effects om GROWTH[id,t] = Xb + u[id] + e[id,t] 4.143401 3.342673 2.08926 65.40 0.0000 102 ey chi2(1) = Prob > chi2 = t re Var(u) = n Test: 17.16777 11.17346 4.365008 va GROWTH e u sd = sqrt(Var) an Lu Estimated results: Var l.c gm CREDIT Figure E.4: Breusch – Pagan LM Test for Pooled OLS and REM of model – STOCK t to Breusch and Pagan Lagrangian multiplier test for random effects ng hi GROWTH[id,t] = Xb + u[id] + e[id,t] ep Estimated results: Var w GROWTH e u n lo ad Test: sd = sqrt(Var) 16.74134 11.90277 3.83687 4.091619 3.45004 1.958793 Var(u) = 45.33 0.0000 ju y th chi2(1) = Prob > chi2 = yi pl al n ua APPENDIX F: RESULTS OF HAUSMAN TEST n va ll fu sqrt(diag(V_b-V_B)) S.E z -.0174571 0393201 -.2218614 0184767 0139279 -.0285371 -.0211745 -.0693706 0484179 -.015126 0171398 0082768 0795393 0116737 0325236 z jm ht vb -.0459941 0181456 -.2912321 0668945 -.0011982 (b-B) Difference at DEPTH INF GOV TO EDU nh Coefficients (b) (B) fixed random oi m Figure F.1: Hausman test for FEM and REM of model – DEPTH k b = consistent under Ho and Ha; obtained from xtreg B = inconsistent under Ha, efficient under Ho; obtained from xtreg Ho: difference in coefficients not systematic om l.c chi2(5) = (b-B)'[(V_b-V_B)^(-1)](b-B) = 20.16 Prob>chi2 = 0.0012 gm Test: an Lu n va ey t re 103 Figure F.2: Hausman test for FEM and REM of model – BANK t to Coefficients (b) (B) fixed random ng hi BANK INF GOV TO EDU ep 0712947 -.022958 -.6250647 0760589 -.0589488 (b-B) Difference 0284564 0137372 -.3425755 0273702 0169124 sqrt(diag(V_b-V_B)) S.E .0428382 -.0366952 -.2824892 0486887 -.0758612 0193673 0090826 0921328 0131 0322229 w n b = consistent under Ho and Ha; obtained from xtreg B = inconsistent under Ha, efficient under Ho; obtained from xtreg lo ad Test: Ho: difference in coefficients not systematic ju y th chi2(5) = (b-B)'[(V_b-V_B)^(-1)](b-B) = 24.82 Prob>chi2 = 0.0002 yi Figure F.3: Hausman test for FEM and REM of model –CREDIT pl al ua Coefficients (b) (B) fixed random n (b-B) Difference va -.0269785 0335913 -.2231705 0164404 0244525 -.0503846 -.0193821 -.0193161 0485453 -.0084841 ll fu 0185688 0074244 0793001 0115385 0331976 oi m -.0773631 0142092 -.2424867 0649857 0159684 n CREDIT INF GOV TO EDU sqrt(diag(V_b-V_B)) S.E Ho: difference in coefficients not systematic at Test: nh b = consistent under Ho and Ha; obtained from xtreg B = inconsistent under Ha, efficient under Ho; obtained from xtreg z z chi2(5) = (b-B)'[(V_b-V_B)^(-1)](b-B) = 23.55 Prob>chi2 = 0.0003 ht vb -.0028003 -.0201762 -.1747564 0412963 -.0348455 0069559 0121183 0910333 0142925 0390283 an Lu -.0069336 0538534 -.3011605 0109858 0367918 om -.0097339 0336772 -.4759169 0522821 0019463 sqrt(diag(V_b-V_B)) S.E l.c STOCK INF GOV TO EDU (b-B) Difference gm Coefficients (b) (B) fixed random k jm Figure F.4: Hausman test for FEM and REM of model – STOCK Test: Ho: difference in coefficients not systematic ey 104 t re chi2(5) = (b-B)'[(V_b-V_B)^(-1)](b-B) = 12.40 Prob>chi2 = 0.0297 n va b = consistent under Ho and Ha; obtained from xtreg B = inconsistent under Ha, efficient under Ho; obtained from xtreg APPENDIX G: THE REGRESSION MODEL RESULTS OF t to FIRST – DIFFERENCE GMM ng Figure G.1: First difference GMM of model –DEPTH hi ep Instrumental variable (GMM) regression Number of obs = 214 Wald chi2(6) = 57.36 Prob > chi = 0.0000 R – squared = 0.1932 Root MSE = 3.8832 w GMM weight matrix: Robust n Coef GROWTHi, t-1 0503047 Robust Std Err 4350783 -.1382538 0133705 lo GROWTHit ad P> |z| [ 95% Conf Interval] 0.12 0.908 -.8024331 9030425 561286 -2.46 0.014 -.2482639 -.282437 0382841 0.35 0.727 -.0616649 0884059 -.5657275 268372 -2.11 0.035 -1.091727 -.039728 2.82 0.005 0312592 1737142 -1.10 0.271 -.6503223 1824403 0.15 0.879 -.5820668 6805224 pl 1024867 0363412 EDUit -.233941 2124433 _cons 0492278 n TOit ua al GOVit yi INFit ju y th DEPTHit z n va ll fu m Instrumented: GROWTHi, t-1 322095 oi Instruments: DEPTHit INFitGOVitTOitEDUit GROWTH_n4 nh Figure G.2: First difference GMM of model –BANK at z Instrumental variable (GMM) regression z 2422392 BANKit 0583932 INFit GOVit z P> |z| = 190 = 32.56 = 0.0000 = 0.0156 = 4.3118 [ 95% Conf Interval] k GROWTHi, t-1 Robust Std Err 4683615 Coef jm ht vb GMM weight matrix: Robust GROWTHit Number of obs Wald chi2(6) Prob > chi R – squared Root MSE 0.605 -.6757325 gm 1.160211 049697 1.17 0.240 -.0390112 1557976 0148773 0489025 0.30 0.761 -.0809698 -.7909463 2816325 -2.81 0.005 -1.342936 -.2389569 847065 0504605 1.68 0.093 -.0141942 1836072 EDUit -.2400032 2484416 -0.97 0.334 -.7269398 2469335 _cons -.0194185 343102 -0.06 0.955 -.6918861 6530491 n ey t re 105 va Instruments: BANKit INFitGOVitTOitEDUit GROWTH_n4 1107243 an Lu Instrumented: GROWTHi, t-1 om TOit l.c 0.52 Figure G.3: First difference GMM of model –CREDIT Instrumental variable (GMM) regression t to Number of obs = 210 Wald chi2(6) = 64.35 Prob > chi = 0.0000 R – squared = 0.2383 Root MSE = 3.7978 ng hi GMM weight matrix: Robust ep z P> |z| [95% Conf Interval] -.023368 Robust Std Err 3624047 -0.06 0.949 -.7336683 -.1732655 0584235 -2.97 0.003 -.2877734 -.0587575 INFit 0016365 0319282 0.05 0.959 -.0609417 0642146 GOVit -.5458838 2519901 -2.17 0.030 -1.039417 -.0519922 1171291 032153 3.64 0.000 0541104 1801478 -.1983322 1994108 -0.99 0.320 -.5891703 1925058 1109649 334401 0.33 0.740 -.5444491 7663788 GROWTHit Coef w GROWTHi, t-1 n CREDITit lo ad pl al _cons yi EDUit ju y th TOit 6869322 n ua Instrumented: GROWTHi, t-1 n va Instruments: CREDITit INFit GOVit TOit EDUit GROWTH_n4 fu ll Figure G.4: First difference GMM of model –STOCK m Number of obs Wald chi2(6) Prob > chi R – squared Root MSE oi Instrumental variable (GMM) regression at nh z GMM weight matrix: Robust = 173 = 75.75 = 0.0000 = 0.3277 = 3.5458 vb z Robust Std Err 6331143 [ 95% Conf Interval] -0.61 0.544 -1.624663 8570996 012684 0138801 0.91 0.361 0446467 0761401 0.59 0.558 -.8256605 3323747 1284711 0516158 2.49 0.013 0273061 2296361 EDUit -.1385895 2070761 -0.67 0.503 -.5444512 2672721 _cons 34006616 2675744 -1.27 0.204 -.8644978 1843746 GROWTHit 0.013 an Lu n va Instruments: STOCKit INFit GOVit TOit EDUit GROWTH_n4 om Instrumented: GROWTHi, t-1 -1.477102 17442189 l.c TOit -2.48 -.1045851 1938785 gm GOVit -.0145205 0398886 k INFit P> |z| jm STOCKit -.3837816 z ht GROWTHi, t-1 Coef ey t re 106 Figure G.5: Test the relevance of instrument variable for model –DEPTH t to ng Test of endogeneity (orthogonality conditions) Ho: variables are exogenous hi ep GMM C statistic chi2(1) = 672249 (p = 0.4123) w n Figure G.6: Test the relevance of instrument variable for model –BANK lo ad y th ju Test of endogeneity (orthogonality conditions) Ho: variables are exogenous yi pl GMM C statistic chi2(1) = 1.77781 (p = 0.1824) ua al n Figure G.7: Test the relevance of instrument variable for model –CREDIT n va fu ll Test of endogeneity (orthogonality conditions) Ho: variables are exogenous oi m 46497 (p = 0.4953) at nh GMM C statistic chi2(1) = z z (p = 0.8700) om l.c 026778 gm GMM C statistic chi2(1) = k Test of endogeneity (orthogonality conditions) Ho: variables are exogenous jm ht vb Figure G.8: Test the relevance of instrument variable for model -STOCK an Lu n va ey t re 107