Surveying the difference of TFP between group of taiwan korea and group of thailand – malaysia; and their TFP growths determinants

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 SURVEYING THE DIFFERENCE OF TFP BETWEEN GROUP OF TAIWAN - KOREA AND GROUP OF THAILAND – MALAYSIA; AND THEIR TFP’S DETERMINANTS BY DUONG KHANH TOAN MASTER OF ARTS IN DEVELOPMENT ECONOMICS HCHIMINH CITY, DECEMBER 2013 i UNIVERSITY OF ECONOMICS HO CHI MINH CITY VIETNAM INSTITUTE OF SOCIAL STUDIES THE HAGUE THE NETHERLANDS VIETNAM - NETHERLANDS PROGRAMME FOR MA IN DEVELOPMENT ECONOMICS SURVEYING THE DIFFERENCE OF TFP BETWEEN GROUP OF TAIWAN - KOREA AND GROUP OF THAILAND – MALAYSIA; AND THEIR TFP’S DETERMINANTS A thesis submitted in partial fulfilment of the requirements for the degree of MASTER OF ARTS IN DEVELOPMENT ECONOMICS BY DUONG KHANH TOAN Academic Supervisor: Dr TRUONG DANG THUY HOCHIMINH CITY, DECEMBER 2013 ii TABLE OF CONTENTS Abstract vi CERTIFICATION vii ACKNOWLEDGMENTS viii LIST OF TABLES ix LIST OF FIGURES ix ABBREVIATIONS x Chapter I 1.1 Problem statement: 1.2 Research objectives: 1.3 Research questions: .4 1.4 Research hypothesis: .4 1.5 Study scope and data Chapter 2.1 Middle income trap and the link with TFP 2.2 TFP and Determinants to TFP level 2.2.1 TFP 2.2.2 TFP growth’s determinants: 10 2.3 Empirical review 12 2.4 Analytical Framework 19 Chapter III 20 3.1 Data source and data process 20 3.2 Methodology .22 3.2.1 TFP calculation following growth accounting approach 22 3.2.2 TFP calculation following regression approach 23 3.3 TFP’s determinant .26 3.3.1 Determinants’ definition: 26 3.3.2 Model specification: 27 Chapter IV 29 4.1 TFP computing 29 4.1.1 TFP level model and value 29 4.1.2 The contribution of TFP growth to national output growth .31 4.2 TFP’s determinant – Emperical result .34 Chapter V .36 5.1 Conclusion 36 5.2 Policy implication 37 5.3 Limitation and further research 38 REFERENCES 40 APPENDIX 44 APPENDIX 44 APPENDIX 44 APPENDIX 45 APPENDIX 47 APPENDIX 48 Abstract This study via econometric regression to calculate TFP level, and TFPG’s contribution to aggregate output growth rate of group [Taiwan – Korea] who have successfully shifted up to high-come level and group [Thailand – Malaysia] who have still stuck in middle income trap This paper also estimates how much efficiency of determinants which are education, openness, foreign direct investment, R&D result, and economy’s fluctuation effecting on the TFP level of these countries The results from the paper would contribute some implies to such Vietnam economic growth in term of handling experiments to rise up from middle income level after escaping from poverty trap CERTIFICATION “I hereby certify that this thesis has not already been submitted for any degree and have not been currently submitted for any other degree I certify that to the best of my knowledge and helps received in preparing this thesis and all sources used have been acknowledged in this thesis ” Duong Khanh Toan vii ACKNOWLEDGMENTS “The thesis at VNP is a painful process” I would like to begin my acknowledgment with the quoting from an email of Doctor Truong Dang Thuy as it is really true to my case This would be not a while paint but a going with life if I cannot achieve and reach the end of what I have dreamed, targeted and pursued Without your support, encouragement, and pressing even, I was totally fell on the road I would like to send the most gratefulness to my big family with all members who always stand by me at the harshest time Those are my dad, my sisters, my wife and my two lovely angles, especially my passed away mom Her love is with me forever, that always refill my energy all the time I fell mostly exhausted I am deeply indebted to Doctor Truong Dang Thuy, Lecturer at Faculty of Development Economics, University of Economics, Hochiminh City He has not only been willing to pick me up but also gave me substantial guidance, useful tips, and encouraged me to pursue to the end of this challenge His wholehearted discussions, and revise with patience have enabled me to achieve, enrich knowledge and experiment through the thesis Equally, I wish to express my thankfulness to Professor Doctor Nguyen Trong Hoai, Lecturer at Faculty of Development Economics, Vice Rector of University of Economics, Hochiminh City He was the first one who gave me support and significant advices as starting point of concept notes Also, without his favor granting, permission, and willing care I would be not able to continue the race I also take this chance to bring my sincere thanks to Doctor Pham Khanh Nam, Master Quan Minh Quoc Binh, Master Le Hoang Viet Phuong for right time tips and instruction to keep my study on track Finally, it is my special thanks sending to Mr Nguyen Dinh Quy (Program Librarian) and Ms Tang Thi Xuan Hong (Programme Secretary) for their warming helps during the time Duong Khanh Toan Hochiminh city, December 2013 LIST OF TABLES Table 1.1 Table 2.3 Table 4.1.1 (a) Table 4.1.2 (a) Table 4.1.2 (b): Table 4.2 Economic indicators of four countries Empirical review Summary about regression result Regressed result for model (6) & (7) Ranking of TFPG’s contribution to aggregate output growth Result of regression model of TFP level’s determinants Page Page 12- 18 Page 29 – 30 Page 31 Page 31 Page 34 LIST OF FIGURES Figure 4.1.2 a Figure 4.1.2 b The trend of TFPG’s contribution to aggregate output growth GDP and GDP per capita from 1962 – 2010 Page 32 Page 33 ABBREVIATIONS FDI Foreign Direct Investment FEM Fixed Effect Model IMF International Monetary Fund GLS Generalized Least Square OLS Ordinary Least Square REM Random Effect Model R&D Research and Development TFP Total Factor Productivity TFPG Total Factor Productivity Growth A5.2 Equation (b), (b’) with FE Group – (b) xtreg ln_rgdp ln_K ln_L t, fe Fixed-effects (within) regression Group variable: country_code Number of obs Number of groups = = 98 R-sq: Obs per group: = avg = max = 49 49.0 49 within = 0.9976 between = 1.0000 overall = 0.9433 corr(u_i, Xb) ln_rgdp ln_K ln_L t _cons F(3,93) Prob > F = -0.4715 Coef .1767478 1.270944 0246095 3.3713 sigma_u sigma_e rho 40220788 0518966 F test that all u_i=0: Std Err .0512474 1270517 0029325 5739406 (fraction F(1, 93) = t 3.45 10.00 8.39 5.87 = = 13079.3 0.0000 P>|t| [95% Conf Interval] 0.001 0.000 0.000 0.000 0749807 1.018644 0187861 2.231568 of variance due 1.52324 4.51103 to u_i) 66.34 Prob > F = 0.0000 Group – (b’) xtreg ln_Yperw ln_Kperw ln_L t, fe Fixed-effects (within) regression Group variable: country_code Number of obs Number of groups = = 98 R-sq: Obs per group: avg ma x F(3,93) Prob > F = = = 49 49.0 49 = = 5762.86 0.0000 within = 0.9946 between 1.0000 overall = 0.8394 corr(u_i, Xb) ln_Yperw ln_Kperw ln_L t _cons sigma_u sigma_e rho = -0.2383 Coef .1767467 4476945 0246095 3.371287 40220938 0518967 F test that all u_i=0: Std Err t P>|t| [95% Conf Interval] 0512475 0875725 0029325 5739417 3.45 5.11 8.39 5.87 0.001 0.000 0.000 0.000 0749793 2737929 0187861 2.231553 (fraction F(1, 93) = 4.51102 of variance due to u_i) 66.34 Prob > F = 0.0000 Group – (b) xtreg ln_rgdp ln_K ln_L t, fe Fixed-effects (within) regression Group variable: country_code R-sq: Number of obs Number of groups within = 0.9957 between 1.0000 overall = 0.9928 corr(u_i, Xb) Obs per group: avg ma x F(3,93) Prob > F = -0.4044 ln_rgdp Coef Std Err ln_K ln_L t _cons 5902221 175822 0149643 4.958357 0487886 092027 0029354 7016247 sigma_u sigma_e rho 09049476 06261118 67627297 (fraction F test that all u_i=0: F(1, 93) = t 12.10 1.91 5.10 7.07 P>|t| 98 = = = 49 49.0 49 = = 7222.44 0.0000 [95% Conf Interval] 0.000 0.059 0.000 0.000 of variance due = = 4933376 -.0069253 0091351 3.565069 6.35164 to u_i) 1.33 Prob > F = 0.2523 Group – (b’) xtreg ln_Yperw ln_Kperw ln_L t, fe Fixed-effects (within) regression Group variable: country_code Number of obs Number of groups = = 98 R-sq: Obs per group: = avg = max = 49 49.0 49 within = 0.9888 between = 1.0000 overall = 0.9857 corr(u_i, Xb) F(3,93) Prob > F = 0.4876 ln_Yperw Coef Std Err ln_Kperw ln_L t _cons 590222 -.2339537 0149643 4.958338 0487886 0710024 0029354 7016254 sigma_u sigma_e rho 09049726 06261123 67628474 (fraction F test that all u_i=0: F(1, 93) = t 12.10 -3.30 5.10 7.07 P>|t| 0.000 0.001 0.000 0.000 of variance due 1.33 = = 2738.54 0.0000 [95% Conf Interval] 4933376 -.3749505 0091351 3.565049 -.09295 6.35162 to u_i) Prob > F = 0.2522 Remark: Equation (b) run with FE model did not fit Group when has coefficient of Labor > A5.3 Equation (c), (c’) with FE Group – (c) xtreg ln_rgdp ln_K ln_L ln_t, fe Fixed-effects (within) regression Group variable: country_code R-sq: within between overall corr(u_i, Xb) Number of obs Number of groups = 0.9961 = 1.0000 = 0.9802 = -0.3735 ln_rgdp Coef Std Err ln_K ln_L ln_t _cons 5655472 5471411 0615609 2.687189 040117 1933681 0271269 9944429 sigma_u sigma_e rho 20774251 06696548 90587215 (fraction F test that all u_i=0: F(1, 93) = t 14.10 2.83 2.27 2.70 = = 98 Obs per group: = avg = ma = x F(3,93) = Prob > F = 49 49.0 49 7842.92 0.0000 P>|t| [95% Conf Interval] 0.000 0.006 0.026 0.008 4858827 1631503 0076922 7124227 of variance due 4.66195 to u_i) 5.73 Prob > F = 0.0187 Group – (c’) xtreg ln_Yperw ln_Kperw ln_L ln_t, fe Fixed-effects (within) regression Group variable: country_code Number of obs Number of groups R-sq: Obs per group: = avg = max = within = 0.9911 between = 1.0000 overall = 0.9475 corr(u_i, Xb) F(3,93) Prob > F = -0.0459 ln_Yperw Coef ln_Kperw ln_L ln_t _cons 5655466 1126901 061561 2.687178 sigma_u sigma_e rho 20774355 06696558 90587275 F test that all u_i=0: Std Err .0401171 1559842 027127 9944445 (fraction F(1, 93) = t 14.10 0.72 2.27 2.70 P>|t| 0.000 0.472 0.026 0.008 of variance due 5.73 = = = = 98 49 49.0 49 3448.72 0.0000 [95% Conf Interval] 485882 -.1970636 0076923 7124086 4.66194 to u_i) Prob > F = 0.0187 Group –(c) xtreg ln_rgdp ln_K ln_L ln_t, fe Fixed-effects (within) regression Group variable: country_code Number of obs Number of groups = = 98 R-sq: Obs per group: = avg = max = 49 49.0 49 within = 0.9949 between = 1.0000 overall = 0.9779 corr(u_i, Xb) = -0.4717 ln_rgdp Coef F(3,93) Prob > F Std Err ln_K ln_L ln_t _cons 7295419 2022022 0545875 2.199861 0394439 1021498 0210572 376748 sigma_u sigma_e rho 21226872 06839296 90595075 (fraction F test that all u_i=0: F(1, 93) = t 18.50 1.98 2.59 5.84 = = 6047.90 0.0000 P>|t| [95% Conf Interval] 0.000 0.051 0.011 0.000 6512141 -.000647 0127721 1.451714 of variance due 2.94800 to u_i) 6.59 Prob > F = 0.0119 Group (c’) xtreg ln_Yperw ln_Kperw ln_L ln_t, fe Fixed-effects (within) regression Group variable: country_code Number of obs Number of groups = = 98 R-sq: Obs per group: = avg = max = 49 49.0 49 within = 0.9866 between = 1.0000 overall = 0.9505 corr(u_i, Xb) ln_Yperw ln_Kperw ln_L ln_t _cons sigma_ u sigma_e F(3,93) Prob > F = 0.3784 Coef .7295408 -.0682542 0545879 2.199855 2122711 0683929 90595279 F test that all u_i=0: Std Err .0394438 0682218 0210572 3767476 (fraction F(1, 93) = t 18.50 -1.00 2.59 5.84 P>|t| 0.000 0.320 0.011 0.000 of variance due 6.59 = [95% Conf Interval] 6512131 -.2037291 0127725 1.451709 2.94800 to u_i) Prob > F = 0.0119 Remarks: Equation (c’) run with FE method did not fit with both group 1, and group A5.4 Equation (a), (a’) with Generalize Least Square (GLS) Group – (a): xtgls ln_rgdp ln_K ln_L Cross-sectional time-series FGLS regression Coefficients: Panels: Correlation: generalized least squares homoskedastic no autocorrelation Estimated covariances Estimated autocorrelations Estimated coefficients = = = Log likelihood = 112.9088 ln_rgdp ln_K ln_L _cons Coef .7295331 026064 4.389026 Number of obs Number of groups Time periods Wald chi2(2) Prob > chi2 Std Err z 0095848 0266653 137236 76.11 0.98 31.98 = = = = = 98 49 19254.2 0.0000 P>|z| [95% Conf Interval] 0.000 0.328 0.000 7107471 -.026199 4.120048 4.65800 Group – (a’): xtgls ln_Yperw ln_Kperw ln_L Cross-sectional time-series FGLS regression Coefficients: Panels: generalized least squares homoskedastic Correlation: no autocorrelation Estimated covariances Estimated autocorrelations Estimated coefficients = = = Log likelihood = 112.9086 ln_Yperw ln_Kperw ln_L _cons Coef .7295332 -.2444029 4.389024 Std Err .0095849 0194292 1372362 Number of obs Number of groups Time periods Wald chi2(2) Prob > chi2 z 76.11 -12.58 31.98 P>|z| 0.000 0.000 0.000 = = = = = 98 49 8133.59 0.0000 [95% Conf Interval] 7107472 -.2824834 4.120046 -.206322 4.65800 Group – (a) xtgls ln_rgdp ln_K ln_L Cross-sectional time-series FGLS regression Coefficients: Panels: generalized least squares homoskedastic Correlation: no autocorrelation Estimated covariances Estimated autocorrelations Estimated coefficients = = = Log likelihood = 120.1835 ln_rgdp ln_K ln_L _cons Coef .8623323 -.0871881 2.48365 Number of obs Number of groups Time periods Wald chi2(2) Prob > chi2 Std Err z 0090633 0134266 1160618 95.15 -6.49 21.40 P>|z| 0.000 0.000 0.000 = = = = = 98 49 19182.80 0.0000 [95% Conf Interval] 8445685 -.1135037 2.256173 -.060872 2.71112 Group – (a’) xtgls ln_Yperw ln_Kperw ln_L Cross-sectional time-series FGLS regression Coefficients: Panels: generalized least squares homoskedastic Correlation: no autocorrelation Estimated covariances Estimated autocorrelations Estimated coefficients = = = Log likelihood = 120.1835 ln_Yperw ln_Kperw ln_L _cons Coef .8623324 -.224856 2.48365 Std Err .0090633 0088267 1160619 Number of obs Number of groups Time periods Wald chi2(2) Prob > chi2 z 95.15 -25.47 21.40 P>|z| 0.000 0.000 0.000 = = = = = 98 49 9236.54 0.0000 [95% Conf Interval] 8445687 -.2421561 2.256173 Remarks: Equation (a) run with GLS method did not fit with Group -.207556 2.71112 A5.5 Equation (b), (b’) with Generalize Least Square (GLS) Group – (b): xtgls ln_rgdp ln_K ln_L t Cross-sectional time-series FGLS regression Coefficients: Panels: generalized least squares homoskedastic Correlation: no autocorrelation Estimated covariances Estimated autocorrelations Estimated coefficients = = = Log likelihood = 127.0594 ln_rgdp ln_K ln_L t _cons Coef .4529043 2897943 0211999 6.973163 Number of obs Number of groups Time periods Wald chi2(3) Prob > chi2 Std Err z 0490007 0515025 003701 4665093 9.24 5.63 5.73 14.95 P>|z| 0.000 0.000 0.000 0.000 = = = = = 98 49 25733.50 0.0000 [95% Conf Interval] 3568647 1888513 013946 6.058822 5489438 3907374 0284539 7.887505 Group – (b’): xtgls ln_Yperw ln_Kperw ln_L t Cross-sectional time-series FGLS regression Coefficients: Panels: generalized least squares homoskedastic Correlation: no autocorrelation Estimated covariances Estimated autocorrelations Estimated coefficients = = = Log likelihood = 127.0592 ln_Yperw ln_Kperw ln_L t _cons Coef .4529043 -.2573014 0212 6.973163 Std Err .0490008 016967 0037011 4665105 Number of obs Number of groups Time periods Wald chi2(3) Prob > chi2 z 9.24 -15.16 5.73 14.95 P>|z| 0.000 0.000 0.000 0.000 = = = = = 98 49 10889.57 0.0000 [95% Conf Interval] 3568645 -.290556 013946 6.058819 5489441 -.2240468 0284539 7.887507 Group – (b): xtgls ln_rgdp ln_K ln_L t Cross-sectional time-series FGLS regression Coefficients: Panels: generalized least squares homoskedastic Correlation: no autocorrelation Estimated covariances Estimated autocorrelations Estimated coefficients = = = Log likelihood = 134.3552 ln_rgdp ln_K ln_L t _cons Coef .6106301 0761498 0158858 5.472755 Number of obs Number of groups Time periods Wal chi2(3) Prob > chi2 Std Err z 044599 0307688 0027709 530969 13.69 2.47 5.73 10.31 = = = = = 98 49 25649.25 0.0000 P>|z| [95% Conf Interval] 0.000 0.013 0.000 0.000 5232177 0158441 0104549 4.432074 6.51343 Group – (b’): xtgls ln_Yperw ln_Kperw ln_L t Cross-sectional time-series FGLS regression Coefficients: Panels: generalized least squares homoskedastic Correlation: no autocorrelation Estimated covariances = Estimated autocorrelations = Estimated coefficients = Log likelihood ln_Yperw ln_Kperw ln_L t _cons = Coef .6106305 -.3132202 0158858 5.472751 Number of obs Number of groups Time periods Wald chi2(3) Prob > chi2 134.3551 Std Err .044599 0172021 0027709 5309696 z 13.69 -18.21 5.73 10.31 P>|z| 0.000 0.000 0.000 0.000 = = = = = 98 49 12367.17 0.0000 [95% Conf Interval] 5232181 -.3469356 0104549 4.43207 -.2795048 6.51343 Remarks: Both (b) and (b’) run with GLS method are fit with both group 1, and group A5.6 Equation (c), (c’) with Generalize Least Square (GLS) Group – (c): xtgls ln_rgdp ln_K ln_L ln_t Cross-sectional time-series FGLS regression Coefficients: Panels: Correlation: generalized least squares homoskedastic no autocorrelation Estimated covariances Estimated autocorrelations Estimated coefficients = = = Log likelihood = 125.5313 ln_rgdp ln_K ln_L ln_t _cons Coef Number of obs Number of groups Time periods Wald chi2(3) Prob > chi2 Std Err .6530831 0885221 1061687 5.032661 z 0165523 0261733 0197852 170128 39.46 3.38 5.37 29.58 = = = = = 98 49 24940.37 0.0000 P>|z| [95% Conf Interval] 0.000 0.001 0.000 0.000 6206411 0372234 0673905 4.699217 5.36610 Group – (c’): xtgls ln_Yperw ln_Kperw ln_L ln_t Cross-sectional time-series FGLS regression Coefficients: Panels: Correlation: generalized least squares homoskedastic no autocorrelation Estimated covariances Estimated autocorrelations Estimated coefficients = = = Log likelihood = 125.5312 ln_Yperw ln_Kperw ln_L ln_t _cons Coef .653083 -.2583948 106169 5.032662 Std Err .0165524 017279 0197852 1701282 Number of obs Number of groups Time periods Wald chi2(3) Prob > chi2 z 39.46 -14.95 5.37 29.58 = = = = = 98 49 10552.2 0.0000 P>|z| [95% Conf Interval] 0.000 0.000 0.000 0.000 6206409 -.292261 0673907 4.699217 -.2245286 5.36610 Group – (c): xtgls ln_rgdp ln_K ln_L ln_t Cross-sectional time-series FGLS regression Coefficients: Panels: generalized least squares homoskedastic Correlation: no autocorrelation Estimated covariances Estimated autocorrelations Estimated coefficients = = = Log likelihood = ln_rgdp ln_K ln_L ln_t _cons Coef .8146214 -.0557685 0513963 2.969581 Number of obs Number of groups Time periods Wald chi2(3) Prob > chi2 123.0401 Std Err z 0215505 0183812 02119 2298801 37.80 -3.03 2.43 12.92 = = = = = 98 49 20340.2 0.0000 P>|z| [95% Conf Interval] 0.000 0.002 0.015 0.000 7723832 -.091795 0098646 2.519025 -.019742 3.42013 Group – (c’): xtgls ln_Yperw ln_Kperw ln_L ln_t Cross-sectional time-series FGLS regression Coefficients: Panels: Correlation: generalized least squares homoskedastic no autocorrelation Estimated covariances Estimated autocorrelations Estimated coefficients = = = Log likelihood = ln_Yperw ln_Kperw ln_L ln_t _cons Coef .8146212 -.2411474 0513967 2.969585 Number of obs Number of groups Time periods Wal chi2(3) Prob > chi2 123.0401 Std Err .0215505 0108909 02119 2298801 z 37.80 -22.14 2.43 12.92 P>|z| 0.000 0.000 0.015 0.000 = = = = = 98 49 9796.91 0.0000 [95% Conf Interval] 772383 -.2624933 009865 2.519028 -.219801 3.42014 Remarks: Equation (c ) run with GLS method did not fit group for negative coefficient of Labor; also Equation (c’) did not fit group for negative coefficient of Labor Conclusion: Choosing the equations which that fit to both groups in terms of both aggregate and per capita model, that are (b), (b’) A5.7 Equation (6), (7) run with Generalize Least Square (GLS) Group – (6): xtgls g_Y g_K g_L Cross-sectional time-series FGLS regression Coefficients: Panels: Correlation: generalized least squares homoskedastic no autocorrelation Estimated covariances Estimated autocorrelations Estimated coefficients = = = Log likelihood = 202.7011 g_Y g_K g_L _cons Coef .432764 8015684 0128891 Number of obs Number of groups Time periods Wald chi2(2) Prob > chi2 Std Err z 0896445 157082 0081458 4.83 5.10 1.58 P>|z| 0.000 0.000 0.114 = = = = = 98 49 91.99 0.0000 [95% Conf Interval] 2570641 4936934 -.0030765 1.10944 0288546 Group – (7): xtgls g_Yperw g_Kperw g_L Cross-sectional time-series FGLS regression Coefficients: Panels: Correlation: generalized least squares homoskedastic no autocorrelation Estimated covariances Estimated autocorrelations Estimated coefficients = = = Log likelihood = 203.3227 g_Yperw g_Kperw g_L _cons Coef .4361735 2272552 0122865 Std Err .0903809 1402668 0081918 Number of obs Number of groups Time periods Wald chi2(2) Prob > chi2 z 4.83 1.62 1.50 P>|z| 0.000 0.105 0.134 = = = = = 98 49 23.97 0.0000 [95% Conf Interval] 2590302 -.0476627 -.003769 0283421 Group – (6): xtgls g_Y g_K g_L Cross-sectional time-series FGLS regression Coefficients: Panels: Correlation: generalized least squares homoskedastic no autocorrelation Estimated covariances Estimated autocorrelations Estimated coefficients = = = Log likelihood = 188.7607 g_Y g_K g_L _cons Coef .6501268 4679094 0051572 Number of obs Number of groups Time periods Wald chi2(2) Prob > chi2 Std Err z 0969156 1301705 0080752 6.71 3.59 0.64 P>|z| 0.000 0.000 0.523 = = = = = 98 49 70.42 0.0000 [95% Conf Interval] 4601757 21278 -.0106698 0209842 Group – (7): xtgls g_Yperw g_Kperw g_L Cross-sectional time-series FGLS regression Coefficients: Panels: Correlation: generalized least squares homoskedastic no autocorrelation Estimated covariances Estimated autocorrelations Estimated coefficients = = = Log likelihood = 190.0581 g_Yperw g_Kperw g_L _cons Coef .6660256 118602 0040956 Number of obs Number of groups Time periods Wald chi2(2) Prob > chi2 Std Err .0967247 1467454 008078 z 6.89 0.81 0.51 P>|z| 0.000 0.419 0.612 = = = = = [95% Conf Interval] 4764487 -.1690137 -.0117369 Remarks: equation (6) is more better than equation (7) in terms of closeness to the estimated value of K, and rejection of null hypothesis 98 49 57.57 0.0000 0199281 A5.8 Equation (8a) and (8b) run with Fixed Effect method Run Model (8a) : xtreg ln_TFP ln_edu ln_FDI CPI ln_patent ln_export, fe Fixed-effects (within) regression Group variable: country_code Number of obs Number of groups = = 122 R-sq: Obs per group: avg ma x F(5,113) Prob > F = = = 30 30.5 31 = = 376.82 0.0000 within = 0.9434 between 0.8790 overall = 0.6793 corr(u_i, Xb) = 0.5807 ln_TFP Coef Std Err ln_edu ln_FDI CPI ln_patent ln_export _cons 5363595 0103647 -.0018455 050792 1480144 4.408151 2522038 0125649 0020826 0111345 0473492 2091204 sigma_u sigma_e rho 77794242 07934768 98970376 (fraction F test that all u_i=0: t 2.13 0.82 -0.89 4.56 3.13 21.08 P>|t| 0.036 0.411 0.377 0.000 0.002 0.000 of variance due F(3, 113) = [95% Conf Interval] 0366983 -.0145286 -.0059716 0287326 0542072 3.993846 1.03602 4.82245 to u_i) 411.91 Prob > F = 0.0000 Run model (8b) with dropping FDP and CPI: xtreg ln_TFP ln_edu ln_export ln_patent, fe Fixed-effects (within) regression Group variable: country_code Number of obs Number of groups = = 122 R-sq: Obs per group: = avg = max = 30 30.5 31 within = 0.9423 between = 0.8983 overall = 0.6977 corr(u_i, Xb) F(3,115) Prob > F = 0.5992 ln_TFP Coef Std Err ln_edu ln_export ln_patent _cons 4738333 1666449 0547197 4.38205 2078085 0353763 0107895 2038735 sigma_u sigma_e rho 76750724 07940359 98941011 (fraction F test that all u_i=0: F(3, 115) = t 2.28 4.71 5.07 21.49 = = 626.42 0.0000 P>|t| [95% Conf Interval] 0.024 0.000 0.000 0.000 0622047 0965713 0333478 3.978216 4.78588 of variance due to u_i) 530.62 Prob > F = 0.0000 A.5.9 Hauseman Test Estimates store fixed Run REM xtreg ln_TFP ln_edu ln_FDI CPI ln_patent ln_export, re Random-effects GLS regression Group variable: country_code Number of obs Number of groups = = 122 R-sq: Obs per group: = avg = max = 30 30.5 31 within = 0.7066 between = 0.9819 overall = 0.9272 Random effects u_i ~ Gaussian corr(u_i, X) = (assumed) ln_TFP Coef Std Err ln_edu ln_FDI CPI ln_patent ln_export _cons 1620227 -.1520429 0118793 3406275 -.0319778 7.05014 2239583 0339727 0063733 0202882 0746258 5308704 sigma_u sigma_e rho 07934768 (fraction Estimate store random Wald chi2(5) Prob > chi2 z 0.72 -4.48 1.86 16.79 -0.43 13.28 P>|z| 0.469 0.000 0.062 0.000 0.668 0.000 of variance due = = 1477.03 0.0000 [95% Conf Interval] -.2769275 -.2186282 -.0006122 3008635 -.1782417 6.009653 to u_i) -.0854576 8.09062 Hauseman test: hausman fixed random, sigmamore Note: the rank of the differenced variance matrix (3) does not equal the number of coefficients being tested (5); be sure this is what you expect, or there may be problems computing the test Examine the output of your estimators for anything unexpected and possibly consider scaling your variables so that the coefficients are on a similar scale ln_edu ln_FDI CPI ln_patent ln_export Coefficients (B) (b) rando fixed m 5363595 1620227 0103647 -.1520429 -.0018455 0118793 050792 3406275 1480144 -.0319778 (b-B) Difference 3743367 1624076 -.0137249 -.2898355 1799922 sqrt(diag(V_b-V_B)) S.E .8302964 0261049 0031323 0320915 1431707 b = consistent under Ho and Ha; obtained from xtreg B = inconsistent under Ha, efficient under Ho; obtained from xtreg Test: Ho: difference in coefficients not systematic chi2(3) = (b-B)'[(V_b-V_B)^(-1)](b-B) = 106.28 Prob>chi2 = 0.0000 (V_b-V_B is not positive definite) hausman fixed random, sigmaless Note: the rank of the differenced variance matrix (3) does not equal the number of coefficients being tested (5); be sure this is what you expect, or there may be problems computing the test Examine the output of your estimators for anything unexpected and possibly consider scaling your variables so that the coefficients are on a similar scale ln_edu ln_FDI CPI ln_patent ln_export Coefficients (B) (b) rando fixed m 5363595 1620227 0103647 -.1520429 -.0018455 0118793 050792 3406275 1480144 -.0319778 (b-B) Difference 3743367 1624076 -.0137249 -.2898355 1799922 sqrt(diag(V_b-V_B)) S.E .2435012 0076558 0009186 0094115 0419877 b = consistent under Ho and Ha; obtained from xtreg B = inconsistent under Ha, efficient under Ho; obtained from xtreg Test: Ho: difference in coefficients not systematic chi2(3) = (b-B)'[(V_b-V_B)^(-1)](b-B) = 1235.72 Prob>chi2 = 0.0000 (V_b-V_B is not positive definite) Conclusion: Should not use REM but FEM ... TFP BETWEEN GROUP OF TAIWAN - KOREA AND GROUP OF THAILAND – MALAYSIA; AND THEIR TFP? ??S DETERMINANTS A thesis submitted in partial fulfilment of the requirements for the degree of MASTER OF ARTS IN... much significance of each determinant on TFP? ??s level? 1.4 Research hypothesis:  Hypotheses 1: It is existed the shift of TFP of Taiwan and Korea by time while the TFP of Thailand and Malaysia moves...UNIVERSITY OF ECONOMICS HO CHI MINH CITY VIETNAM INSTITUTE OF SOCIAL STUDIES THE HAGUE THE NETHERLANDS VIETNAM - NETHERLANDS PROGRAMME FOR MA IN DEVELOPMENT ECONOMICS SURVEYING THE DIFFERENCE OF TFP BETWEEN