Luận án Mối quan hệ giữa phát triển tài chính và tầng trưởng kinh tế ở các nước Asean+3 và những gợi ý chính sách cho Việt Nam

Gov -.0004798 -.0009844 .0005046 .0007099 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(6) - (b-B)'[(V b-v B) A(-1) ] (b-B) = 20.29 Prob>chi2 = 0.0025 (V b-v B is not positive definite) Modified Wald test for groupwise heteroskedasticity in fixed effect regression model HO: sigma(i)A2 = sigmaA2 for all 1 chi2 (9) = 228.86 Prob>chi2 = 0.0000 FDI tỉnh theo PCA Fixed-effects (within) regression Group variable: NoCoun R-sq: within = 0.7465 between = 0.0004 overall = 0.0808 Number of obs = 36 Number of groups = 9 Obs per group: min = 4 avg = 4.0 max = 4 F(6,21) = 10.31 corr(u_i, Xb) = -0.3302 Prob > F = 0.0000 gRGDPpc Coef. std. Err. t p> 111 [95% Conf. Interval] InYO .0136045 .0082611 1.65 0.114 -.0035755 .0307844 FDIpca -.1728251 .057836 -2.99 0.007 -.2931016 -.0525486 CAP .2296326 .0347906 6.60 0.000 .1572816 .3019836 POPtb -.0124643 .0039701 -3.14 0.005 -.0207206 -.004208 ECONFREE .000249 .0006503 0.38 0.706 -.0011034 .0016015 Gov -.0013392 .0011708 -1.14 0.266 -.003774 .0010957 _cons -.0046036 .0758188 -0.06 0.952 -.1622775 .1530702 sigma_u . 02490541 sigma_e . 00740401 rho .91879811 (fraction of variance due to u_i) F test that all u_i=0: Prob > F = 0.0000 . hausman fe re Coefficients (b) fe (B) re (b-B) Difference sqrt(diag(V_b-V_B)) S.E. InYO .0136045 -.0003958 .0140003 . 0060149 FDIpca - .1728251 -.0353732 -.1374519 . 0320404 CAP .2296326 .2370259 -.0073933 POPtb - . 0124643 -.0095953 -.0028691 ECONFREE .000249 -.0000995 .0003485 . 0003467 Gov -.0013392 -.0009545 -.0003847 .000327 b = consistent under Ho and Ha obtained from xtreg B = inconsistent under Ha, efficient under Ho obtained from xtreg Test: Ho difference in coefficient s not systematic Chi2(6) = (b-B)' (V_b-V _B)A(-1) ] (b-B) = 119.67 Prob>chi2 = 0.0000 (V_b-V_B is not positive definite) Modified Wald test for groupwise heteroskedasticity in fixed effect regression model HO: sigma(i)A2 = sigmaA2 for all i Chi2 (9) = Prob>chi2 = 400.57 0.0000 FDI tỉnh theo phương pháp DEA Fixed-effects (within) regression Group variable: NoCoun R-sq: within = 0.ỔỔ8Ổ between = 0.3775 overall = 0.4250 corr(u_i, Xb) = -0.3811 Number of Number of obs = groups = 36 9 Obs per group: min = 4 avg = 4.0 max = 4 F(6,21) = 7.06 Prob > F = 0.0003 gRGDPpc Coef. std. Err. t p> 111 [95% Conf. Interval] InYO -.0013869 . 0068478 -0.20 0.841 -. 0156277 .0128539 FDIdea -.0334052 . 0242793 -1.38 0.183 -.0838967 .0170862 CAP .1891374 . 0417529 4.53 0.000 .1023075 .2759672 POPtb -.0093864 . 0044474 -2.11 0.047 -.0186354 -.0001375 ECONFREE -.0006083 . 0006594 -0.92 0.367 - . 0019797 . 000763 Gov -.0013931 . 0013581 -1.03 0.317 - . 0042174 .0014313 _cons .1339551 . 0674563 1.99 0.060 -.006328 .2742382 sigma_u .01938378 sigma_e .00846571 rho .83981133 (fraction of variance due to u_i) F test that all u_i=o: F(8, 21) = Prob > F = 0.0008 . hausman fe re Coefficients (b) fe (B) re (b-B) Difference sqrt(diag(V_b-V_B)) S.E. InYO -.0013869 -.0076046 .0062177 .0047203 FDldea -.0334052 . 0470102 -.0804154 .0035042 CAP .1891374 .3021454 -.1130081 POPtb -.0093864 -.0090366 -.0003498 .0008664 ECONFREE -.0006083 .0001015 -.0007098 . 000376 Gov -.0013931 -.0012055 -.0001875 .0008199 b = consistent under Ho and Ha obtained from xtreg B = inconsistent under Ha, efficient under Ho obtained from xtreg Test: Ho difference i n coefficient s not systematic Chi2(6) = (b-B)' (V_b-V _B) - (-1) ] (b-B) = 145.54 Prob>chi2 = 0 0000 (V_b-V_B is not positive definite) Modified Wald test for groupwise heteroskedasticity in fixed effect regression model HO: sigma(i) A2 = sigmaA2 for all i Chi2 (9) = Prob>chi2 = 3791.88 0.0000 MÔ hình 3: . xtreg gRGDPpc FDIpca CAP POP ECONFREE Gov TRADE INF, re Random-effects GLS regression Group variable: NoCoun Number of obs Number of groups 180 R-sq: within = 0.6495 between = 0.4646 overall = 0.5919 Obs per group: min = 20 avg = 20.0 max = 20 Wald Chi2(7) Prob > chi2 311.71 0.0000 Chi2 (7) Prob>chi2 gRGDPpc Coef. std. Err. z p>|z| [95% Conf. Interval] FDIpca -.0637871 .0301919 -2.11 0.035 -.1229621 -.0046121 CAP .2552842 .0188801 13.52 0.000 .2182799 .2922886 POP -.0203305 .0028751 -7.07 0.000 -.0259656 -.0146955 ECONFREE .0001235 .0004569 0.27 0.787 -.000772 .0010189 Gov -.0029581 .0010726 -2.76 0.006 -.0050604 -.0008558 TRADE .0000776 .0000522 1.48 0.138 -.0000248 .0001799 INF -.0009753 .0003639 -2.68 0.007 -.0016886 -.000262 _cons .0970538 . 027417 3.54 0.000 .0433173 .1507902 sigma u .01680742 sigma e .01936933 rho .42953699 (fraction of variance due to u_i) corr(u_i, X) =0 (assumed) Hausman test . hausman fe re, sigmamore Coefficients (b) fe (B) re (b-B) Difference sqrt(diag(V_b-V_B)) S.E. FDIpca -.0878752 -.0637871 -.0240881 .0107322 CAP .2570322 .2552842 .001748 .0021623 POP -.0209049 -.0203305 -.0005743 .0005797 ECONFREE .0004169 .0001235 .0002934 .000374 Gov -.0038002 -.0029581 -.0008421 .0005715 TRADE .0001335 .0000776 .0000559 .0000538 INF -.001061 -.0009753 -.0000858 .0001221 b = consistent under Ho and Ha, obtained from xtreg B = inconsistent under Ha, eff icient under Ho, obtained from xtreg Test: Ho difference ir 1 coefficients not systematic (b-B) ' [ (V_b-V_B) A (-1) ] (b-B) 7.73 0.3569 gRGDPpc Coef. std. Err. z p>| z 1 [95% Conf. Interval] FDIbq -.0473466 . 0315231 -1.50 0.133 -.1091308 .0144376 CAP .2529763 .0190613 13.27 0.000 .2156168 .2903359 POP -.0200502 .0029181 -6.87 0.000 -.0257695 -.0143309 ECONFREE .0000147 .0004562 0.03 0.974 -.0008794 .0009087 Gov -.0027335 .0011587 -2.36 0.018 -.0050045 -.0004625 TRADE .000087 .000054 1.61 0.107 -.0000188 .0001927 INF -.0009801 .0003672 -2.67 0.008 -.0016999 -.0002603 _cons .0919484 . 0277997 3.31 0.001 . 0374621 . 1464348 sigma_u .01676583 sigma_e .0195366 rho .42411743 (fraction of variance due to u_i) Random-effects GLS regression Group variable: NoCoun R-sq: within = 0.6432 between = 0.5242 overall = 0.6062 Number of obs = 180 Number of groups = 9 corr(u_i, X) =0 (assumed) Obs per group: min = 20 20.0 20 avg = max = Wald chi2(7) = 305.65 Prob > chi2 = 0.0000 . hausman fe re Coefficients (b-B) Difference sqrt(diag(V_b-V_B)) S.E. (b) f e (B) re FDIbq - . 0745867 - . 0473466 - . 0272401 . 0144221 CAP .2552664 .2529763 . 0022901 .0026051 POP - . 0207687 -.0200502 -.0007185 . 0006265 ECONFREE .000293 .0000147 . 0002784 .0003751 Gov -.0033954 -.0027335 - . 0006619 . 0004999 TRADE . 0001604 .000087 .0000734 .0000582 INF -.0010697 -.0009801 -.0000896 . 0001209 b = consistent under Ho and Ha, obtained from xtreg B = inconsistent under Ha, eff icient under Ho, obtained from xtreg Test: Ho difference i n coefficients not systematic Chi2(7) = (b-B)•[(V_b-V_B)A(-1)](b-B) = 7.31 Prob>chi2 = 0.3971 F (7,83) = 8.13 corr(u_i, Xb) = -0.8385 Prob > F = 0 . 0000 gRGDPpc Coef. std. Err. t p> 111 [95% Conf. Interval] FDIdea . 0162438 . 0376187 0.43 0.667 -.0585783 .091066 CAP .2560063 . 040028 6.40 0.000 . 1763922 .3356203 POP -.0301335 .0060249 -5.00 0.000 -.0421168 -.0181502 ECONFREE . 0012598 .0014658 0.86 0.393 -.0016555 .0041752 Gov -.0037742 .0034144 -1.11 0.272 -.0105654 .003017 TRADE .000419 .000152 2.76 0.007 .0001167 .0007213 INF .0010754 .0008332 1.29 0.200 -.0005818 .0027325 _cons -.0471934 .0899012 -0.52 0.601 -.2260032 . 1316164 sigma_u .05094901 sigma_e .01972993 rho .86959431 (fraction of variance due to u_i) F test that all u i=0: F(8, 83) = 4.96 Prob > F = 0 . 0000 Fixed-effects (within) regression Group variable: NoCoun R-sq: within = 0.4067 between = 0.0017 overall = 0.0169 Number of obs Number of groups Obs per group: min avg max 99 9 11 11.0 11 . hausman fe re Coefficients (b) f e (B) re (b-B) Difference sqrt(diag(V_b-V_B) ) S.E. FDIdea .0162438 . Ũ4Ũ3Ũ53 -.0240615 .0246002 CAP .2560063 .2884985 -.0324922 POP -.0301335 -.0187755 -.011358 .0039002 ECONFREE .0012598 -.0010776 .0023375 . 001363 Gov -.0037742 -.0018848 -.0018894 .0030591 TRADE . 000419 . ŨŨŨ1Ũ53 .ŨŨŨ3137 .0001405 INF .0010754 . 0004423 .000633 b = consistent under Ho and Ha, obtained from xtreg B = inconsistent under Ha, eff icient under Ho, obtained from xtreg Test: Ho difference in coefficients not systematic Chi2(7) = Prob>chi2 = (b-B)'[(V_b-V_B)A(-1)](b-B) 16.54 0.0206 (V_b-V_B is not positive definite) Phụ lục 7.2: Kết quả hồi quy mô hình 2 và Mô hình 3 sử dụng 2SLS (bổ sung cho phần 4.4.2 tr. 99) Mô hình 2 IV (2SLS) estimation Estimates efficient for homoskedasticity only Statistics consistent for homoskedasticity only Total (centered) ss Total (uncentered) ss Residual ss .0136288499 .0548931963 .0039599637 Number of obs = F( 6, 20) = Prob > F = Centered R2 = Uncentered R2 = 27 9.07 0.0001 0.7094 0.9279 Root MSE = .01211 gRGDPpc Coef. std. Err. z p>| z| [95% Conf. Interval] FDIpca .1486008 .0550597 2.70 0.007 .0406858 .2565157 LnYO -.0113435 .0064293 -1.76 0.078 -.0239446 .0012576 POPtb -.0090186 .0045859 -1.97 0.049 -.0180067 -.0000304 Gov -.0021923 .0012049 -1.82 0.069 -.0045539 .0001694 ECONFREE .0001415 .0005416 0.26 0.794 -.0009201 .001203 CAP .3073111 .0757553 4.06 0.000 .1588334 .4557889 _cons .0843138 .0276749 3.05 0.002 .0300721 .1385556 15.711 0.0004 Underidentification test (Anderson canon, corr. LM statistic): Chi-sq(2) p-val = 10% maximal IV sj Lze 19.93 15% maximal IV sj Lze 11.59 20% maximal IV sj Lze 8.75 25% maximal IV sj Lze 7.25 13.222 Weak identification test (Cragg-Donald Wald F statistic) Stock-Yogo weak ID test critical values: Source: Stock-Yogo (2005). Reproduced by permission. Sargan statistic (overidentification test of all instruments): 8.823 Chi-sq(l) p-val = 0.0030 Instrumented: FDIpca Included instruments: LnYO POPtb Gov ECONFREE CAP Excluded instruments: L.FDIpca L.gRGDPpc IV (2SLS) estimation Estimates efficient for homoskedasticity only Statistics consistent for homoskedasticity only Number of obs = 27 F( 6, 20) = 9.79 Prob > F = 0.0000 Total (centered) ss Total (uncentered) ss Residual ss .0136288499 .0548931963 .0035425749 Centered R2 = 0.7401 Uncentered R2 = 0.9355 Root MSE . 01145 gRGDPpc Coef. std. Err. z p>|z| [95% Conf. Interval] FDIbq .1002314 . 0433619 2.31 0.021 . 0152436 .1852191 LnYO -.0100277 . 006201 -1.62 0.106 -.0221814 .0021261 POPtb -.0110016 .0043312 -2.54 0.011 -.0194905 -.0025126 Gov -.0022798 .0012181 -1.87 0.061 -.0046672 .0001077 ECONFREE .0002309 .0005131 0.45 0.653 -.0007747 .0012365 CAP .3516764 .0671086 5.24 0.000 .2201459 .4832068 _cons .0905553 . 030094 3.01 0.003 .0315722 .1495385 Underidentification test (Anderson canon, corr. LM statistic): 21.170 Chi-sq(2) p-val = 0.0000 Weak identification test (Cragg-Donald Wald F statistic): 34.500 Stock-Yogo weak ID test critical values: 10% maximal IV size 19.93 15% maximal IV size 11.59 20% maximal IV size 8.75 25% maximal IV size 7.25 Source: Stock-Yogo (2005). Reproduced by permission. Sargan statistic (overidentification test of all instruments): 12.734 Chi-sq(l) p-val = 0.0004 Instrumented: FDIbq Included instruments: LnYO POPtb Gov ECONFREE CAP Excluded instruments: L.FDIbq L.gRGDPpc IV (2SLS) estimation Estimates efficient for homoskedasticity only Statistics consistent for homoskedasticity only Number of obs = 27 F( 6, 20) = 6.23 Prob > F = 0.0008 Total (centered) ss = .0136288499 Centered R2 = 0.5350 Total (uncentered) ss = .0548931963 Uncentered R2 = 0.8845 Residual ss = .0063376684 Root MSE .01532 gRGDPpc Coef. std. Err. z E>> 1 z 1 [95% Conf. Interval] FDIdea .1765001 .0584573 3.02 0.003 .061926 .2910743 LnYO -.0224699 .0096439 -2.33 0.020 -.0413716 -.0035682 POPtb -.0118936 .0058018 -2.05 0.040 -.0232649 -.0005224 Gov -.0026974 .0015101 -1.79 0.074 -.005657 .0002623 ECONFREE .0004458 .0006915 0.64 0.519 -.0009096 .0018011 CAP .3067562 .0905233 3.39 0.001 .1293339 .4841786 _cons .0881776 .0326686 2.70 0.007 . 0241483 .1522069 Underidentification test (Anderson canon, corr. LM statistic): 10.435 Chi-sq(2) p-val = 0.0054 Weak identification test (Cragg-Donald Wald F statistic): 5.985 Stock-Yogo weak ID test critical values: 10% maximal IV size 19.93 15% maximal IV size 11.59 20% maximal IV size 8.75 25% maximal IV size 7.25 Source: Stock-Yogo (2005). Reproduced by permission. Sarqan statistic (overidentification test of all instruments): 1.319 Chi-sq(l) p-val = 0.2507 MÔ hình 3: IV (2SLS) estimation Estimates efficient for homoskedasticity only Statistics consistent for homoskedasticity only Number of obs = 171 F( 8, 162) = 42.38 Prob > F = 0.0000 Total (centered) ss = .2496021725 Centered R2 = 0.6742 Total (uncentered) ss = .4605937668 Uncentered R2 = 0.8235 Residual ss = .0813087154 Root MSE .02181 gRGDPpc Coef. std. Err. z p>|z| [95% Conf. Interval] FDIpca .0866291 .039384 2.20 0.028 .0094378 .1638203 LnY -.0113331 .0040435 -2.80 0.005 -.0192582 -.0034079 POP -.0164161 .0027693 -5.93 0.000 -.0218439 -.0109884 Gov -.0023276 .0010448 -2.23 0.026 -.0043753 -.0002799 ECONFREE .0001752 .0004561 0.38 0.701 -.0007188 .0010692 CAP .2560592 .0212071 12.07 0.000 .2144941 .2976243 TRADE .0000462 .0000383 1.21 0.228 -.0000289 .0001213 INF -.0010558 .0003931 -2.69 0.007 -.0018262 -.0002854 _cons .1208631 .0163064 7.41 0.000 .0889031 .1528232 Underidentification test (Anderson canon, corr. LM statistic): 96.406 Chi-sq(2) p-val = 0.0000 Underidentification test (Anderson canon, corr. LM statistic): 96.406 Chi-sq(2) p-val = 0.0000 Weak identification test (Cragg-Donald Wald F statistic): 104.038 Stock-Yogo weak ID test critical values: 10% maximal IV size 19.93 15% maximal IV size 11.59 20% maximal IV size 8.75 25% maximal IV size 7.25 Source: Stock-Yogo (2005). Reproduced by permission. Sarqan statistic (overidentification test of all instruments): 0.044 Chi-sq(l) p-val = 0.8330 Instrumented: FDIpca Included instruments: LnY POP Gov ECONFREE CAP TRADE INF Excluded instruments: L.FDIpca L.gRGDPpc gRGDPpc Coef. std. Err. z p>|z| [95% Conf. Interval] FDIdea .1346338 .0656625 2.05 0.040 .0059377 .2633298 LnY -.0147679 .0118131 -1.25 0.211 -.0379211 .0083853 POP -.0191112 .0045977 -4.16 0.000 -.0281225 -.0100999 Gov -.0033612 .0017462 -1.92 0.054 -.0067837 .0000612 ECONFREE .000269 .0011768 0.23 0.819 -.0020376 .0025756 CAP .3080369 .0457582 6.73 0.000 .2183524 .3977213 TRADE -3.40e-06 .0000782 -0.04 0.965 -.0001568 .00015 INF .0010384 .000955 1.09 0.277 -.0008334 .0029102 _cons .0825263 .027105 3.04 0.002 .0294015 .1356512 IV (2SLS) estimation Estimates efficient for homoskedasticity only Statistics consistent for homoskedasticity only Total (centered) ss = .2496021725 Total (uncentered) ss = .4605937668 Residual ss = .0797855647 Number of obs = 171 F( 8, 162) = 42.89 Prob > F = 0.0000 Centered R2 = 0.6803 Uncentered R2 = 0.8268 Root MSE = .0216 gRGDPpc Coef. std. Err. z p>|z| [95% Conf. Interval] FDIbq .0588779 .0380657 1.55 0.122 -.0157295 .1334853 LnY -.0115701 .0047977 -2.41 0.016 -.0209734 -.0021668 POP -.017295 .0026965 -6.41 0.000 -.02258 -.01201 Gov -.0021462 .0011492 -1.87 0.062 -.0043986 .0001061 ECONFREE .0002148 .000487 0.44 0.659 -.0007398 .0011694 CAP .2612138 .0207579 12.58 0.000 .220529 .3018986 TRADE .0000541 .00004 1.35 0.177 -.0000244 .0001326 INF -.0010777 .0003911 -2.76 0.006 -.0018442 -.0003112 _cons . 1311911 .0206766 6.34 0.000 .0906657 .1717166 Underidentification test (Anderson canon, corr. LM statistic): 134.861 Chi-sq(2) p-val = 0.0000 Weak identification test (Craqq-Donald Wald F statistic) 300.407 Stock-Yogo weak ID test critical values ■ i 10% maximal IV s ize 19.93 15% maximal IV s ize 11.59 20% maximal IV s ize 8.75 25% maximal IV s ize 7.25 Source: Stock-Yogo (2005). Reproduced by permission, Sargan statistic (overidentification test of all instruments): 0.135 Chi-sq(l) p-val = 0.7130 Instrumented: FDIbq Included instruments: LnY POP Gov ECONFREE CAP TRADE INF Excluded instruments: L.FDIbq L.gRGDPpc IV (2SLS) estimation Estimates efficient for homoskedasticity only Statistics consistent for homoskedasticity only Number of obs = 90 F( 8, 81) = 11.82 Total (centered) ss = .095724225 Total (uncentered) ss = .226759094 Residual ss = .0456885156 Prob > F = Centered R2 = Uncentered R2 = Root MSE 0.0000 0.5227 0.7985 .02253 Underidentification test (Anderson canon, corr. LM statistic): 25.264 Chi-sq(2) p-val = 0.0000 Underidentification test (Anderson canon, corr. LM statistic): 25.264 Chi-sq(2) p-val = 0.0000 Weak identification test (Cragg-Donald Wald F statistic): 15.611 Stock-Yogo weak ID test critical values: 10% maximal IV size 19.93 15% maximal IV size 11.59 20% maximal IV size 8.75 25% maximal IV size 7.25 Source: Stock-Yogo (2005) . Reproduced by permission. Sarqan statistic (overidentification test of all instruments): 2.375 Chi-sq(l) p-val = 0.1233 Instrumented: FDIdea Included instruments: LnY POP Gov ECONFREE CAP TRADE INF Excluded instruments: L.FDIdea L.gRGDPpc Phụ lục 7.3: Kết quả hồi quy mô hình 2 và Mô hình 3 sử dụng GMM (bổ sung cho phần 4.4.2 tr. 99) Mô hình 2: 2-Step GMM estimation Number of obs = F( 8, 162) = Prob > F Centered R2 = Uncentered R2 = Root MSE 171 35.99 0.0000 0.6739 0.8233 . 02182 Estimates efficient for arbitrary heteroskedasticity Statistics robust to heteroskedasticity Total (centered) ss = .2496021725 Total (uncentered) ss = .4605937668 Residual ss = .0813953979 gRGDPpc Coef. Robust std. Err z p>|z| [95% Conf. Interval] FDIpca .0885489 .0384097 2.31 0.021 .0132673 . 1638305 LnY -.0114454 .0033671 -3.40 0.001 -.0180448 -.004846 POP -.0163683 .0025871 -6.33 0.000 -.021439 -.0112976 Gov -.0023209 .0010602 -2.19 0.029 -.0043988 -.0002429 ECONFREE .0001687 .0004064 0.41 0.678 -.000628 . 0009653 CAP .2561106 .0196803 13.01 0.000 .2175379 .2946833 TRADE .0000458 .0000415 1.10 0.270 -.0000355 . 0001271 INF -.0010548 .000559 -1.89 0.059 -.0021505 . 0000409 _cons .1213088 .0165093 7.35 0.000 .0889512 . 1536664 Underidentification test (Kleibergen- -Paap rk LM statistic): 47.101 Chi-sq(2) P-val = 0.0000 Weak identification test (Cragg-Donald Wald F statistic) 104.038 (Kleibergen- -Paap rk Wald F statistic): 81.697 Stock-Yogo weak ID test critical values: 10% maximal IV size 19.93 15% maximal IV size 11.59 20% maximal IV size 8.75 25% maximal IV size 7.25 Source: Stock-Yogo (2005). Reproduced by permission. NB: Critical values are for Cragg-Donald F statistic and i.i.d. errors. Hansen J statistic (overidentification test of all instruments): 0.021 Chi-sq(l) p-val = 0.8850 Instrumented: FDIpca Included instruments: LnY POP Gov ECONFREE CAP TRADE INF Excluded instruments: L.FDIpca L.gRGDPpc 2—Step GMM estimation Estimates efficient for arbitrary heteroskedasticity Statistics robust to heteroskedasticity Number F( 8, Prob > of obs 162) F 171 37.50 0.0000 Total (centered) ss Total (uncentered) ss Residual ss .2496021725 .4605937668 .0798045167 Centered R2 = 0.6803 Uncentered R2 = 0.8267 gRGDPpc Coef. Robust std. Err. z p> 1 z 1 [95% Conf. Interval] FDIbq . 0579544 .0316913 1.83 0.067 - .0041595 .1200683 LnY - . 0114609 .0036828 -3.11 0.002 - . 0186791 - .0042427 POP - . 0173231 .0025542 -6.78 0.000 - . 0223292 - . 012317 Gov - . 0021927 .0010969 -2.00 0.046 - .0043425 - .0000429 ECONFREE . 0002302 .0004096 0.56 0.574 - .0005727 .0010331 CAP .260957 . 0191972 13.59 0.000 .2233311 .2985829 TRADE . 000054 .0000414 1.30 0.192 - .0000272 .0001352 INF - . 0010857 .0004983 -2.18 0.029 - .0020624 -.000109 _cons .1303397 .0199157 6.54 0.000 .0913057 .1693736 . 0216 Underidentification test (Kleibergen-Paap rk LM statistic): 55.373 Ch±-sq(2) p-val = 0.0000 Underidentification test (Kleibergen-Paap rk LM statistic): 55.373 Ch±-sq(2) p-val = 0.0000 Weak identification test (Cragg-Donald Wald F statistic): 300.407 (Kleibergen-Paap rk Wald F statistic): 286.755 Stock—Yogo weak ID test critical values: 10% maximal IV size 19.93 15% maximal IV size 11.59 20% maximal IV size 8.75 25% maximal IV size 7.25 Source: Stock-Yogo (2005) . Reproduced by permission. NB: Critical values are for Cragg—Donald F statistic and i.i.d. errors. Hansen J statistic (overidentification test of all instruments): 0.064 Chi-sq(l) p-val = 0.8010 Instrumented: FDIbq Included instruments: LnY POP Gov ECONFREE CAP TRADE INF Excluded instruments: L.FDIbq L.gRGDPpc 2-Step GMM estimation Estimates efficient for arbitrary heteroskedasticity Statistics robust to heteroskedasticity Number of obs = 90 F( 8, 81) = 6.78 Prob > F 0.0000 Total (centered) ss = . 095724225 Centered R2 = 0.5311 Total (uncentered) ss = .226759094 Uncentered R2 = 0.8021 Residual ss = . 0448824325 Root MSE .02233 gRGDPpc Coef. Robust std. Err z p>l z| [95% Conf. Interval] FDldea .1071478 .0815517 1.31 0.189 -.0526906 .2669861 LnY -.0109393 . 0127409 -0.86 0.391 -.035911 .0140325 POP - . 0211445 . 0052127 -4.06 0.000 -.0313612 - . 0109278 Gov -.0034767 .0016542 -2.10 0.036 -.0067189 -.0002346 ECONFREE .0002135 .0011131 0.19 0.848 -.0019681 .0023951 CAP .3183508 .0565857 5.63 0.000 .2074449 . 4292567 TRADE .0000292 .0000757 0.39 0.700 -.0001192 .0001776 INF .0008798 .001051 0.84 0.403 -.0011801 . 0029397 _cons .0750359 .022217 3.38 0.001 .0314913 .1185804 Underidentification test (Kleibergen- Paap rk LM statistic): 13.457 Chi-sq(2) p-val = 0.0012 Weak identification test (Cragg-Donald Wald F statistic) 15.611 (Kleibergen- Paap rk Wald F stat istic): 7.029 Stock-Yogo weak ID test critical values: 10% maximal IV size 19.93 15% maximal IV size 11.59 20% maximal IV size 8.75 25% maximal IV size 7.25 Source: Stock-Yogo (2005). Reproduced by permission. NB: Critical values are for Cragg-Donald F statistic and i.i.d. errors. Hansen J statistic (overidentification test of all instruments): 1.182 Chi-sq(l) P-val = 0.2771 MÔ hình 3: 2—step GMM estimation Estimates efficient for arbitrary heteroskedasticity Statistics robust to heteroskedasticity Number of obs = 27 F( 6, 20) = 5.96 Prob > F = 0.0011 Total (centered) ss = .0136288499 Centered R2 = 0.7346 Total (uncentered) ss = . 0548931963 Uncentered R2 = 0.9341 Residual ss = .0036166095 Root MSE = .01157 gRGDPpc Coef. Robust std. Err. z p>| z 1 [95% Conf. Interval] FDIbq .1070761 .0406496 2.63 0.008 . 0274044 . 1867477 LnYO -.010101 .0053229 -1.90 0.058 - . 0205338 .0003318 POPtb -.0096335 .0044426 -2.17 0.030 - . 0183409 - .0009261 Gov -.0021921 .0013318 -1.65 0.100 - . 0048024 .0004181 ECONFREE .0000989 .0005914 0.17 0.867 - . 0010602 .0012581 CAP .3556844 .0703202 5.06 0.000 .2178593 .4935095 _cons .0936622 .0229016 4.09 0.000 . 0487759 . 1385486 Underidentification test (Kleibergen-Paap rk LM statistic): 8.565 Chi—sq(2) p-val = 0.0138 Weak identification test (Cragg-Donald Wald F statistic) : 34.500 (Kleibergen-Paap rk Wald F statistic): 57.322 Stock—Yogo weak ID test critical values: 10% maximal IV size 19.93 15% maximal IV size 11.59 20% maximal IV size 8.75 25% maximal IV size 7.25 NB: Critical values are for Cragg-Donald F statistic and i.i.d. errors. Source: Stock-Yogo (2005). Reproduced by permission. 11.148 0.0008 Hansen J statistic (overidentification test of all instruments): Chi—sq(l) p—val = Instrumented: FDIbq Included instruments: LnYO POPtb Gov ECONFREE CAP Excluded instruments: L.FDIbq L.gRGDPpc 2-Step GMM estimation Estimates efficient for arbitrary heteroskedasticity Statistics robust to heteroskedasticity Number of obs = 27 F( 6, 20) = 6.21 Prob > F = 0.0008 Total (centered) ss = .0136288499 Centered R2 = 0.6293 Total (uncentered) ss = . 0548931963 Uncentered R2 = 0.9080 Residual ss = .0050524811 Root MSE = .01368 gRGDPpc Coef. Robust std. Err. z p>| z 1 [95% Conf. Interval] FDIpca .2081449 .0574266 3.62 0.000 . 0955909 .3206989 LnYO -.0161161 .0062776 -2.57 0.010 -.02842 - .0038122 POPtb -.008451 .0041089 — 2.06 0.040 - . 0165043 -.0003977 Gov -.0024217 .0012446 -1.95 0.052 - . 0048611 .0000176 ECONFREE .0002515 .0005581 0.45 0.652 - . 0008424 .0013454 CAP .2804354 .0745684 3.76 0.000 . 1342841 . 4265868 _cons .0962332 .0211667 4.55 0.000 . 0547473 . 1377191 Underidentification test (Kleibergen-Paap rk LM statistic): 5.846 Ch±-sq(2) p-val = 0.0538 Weak Identification test (Cragg-Donald Wald F statistic) : 13.222 (Kleibergen-Paap rk Wald F statistic): 10.524 Stock-Yogo weak ID test critical values: 10% maximal IV size 19.93 15% maximal IV size 11.59 20% maximal IV size 8.75 25% maximal IV size 7.25 NB: Critical values are for Craqq—Donald F statistic and i.i.d. errors. Source: Stock-Yoqo (2005). Reproduced by permission. Hansen J statistic (overidentification test of all instruments): 7.847 Chi-sq(l) p-val = 0.0051 2—Step GMM estimation Estimates efficient for arbitrary heteroskedasticity Statistics robust to heteroskedasticity Number of obs = Total (centered) ss Total (uncentered) ss Residual ss .0136288499 . 0548931963 . 0073627171 F( 6, 20) = Prob > F = Centered R2 = Uncentered R2 = 27 8.08 0.0002 0.4598 0.8659 . 01651 gRGDPpc Coef. Robust std. Err. z p>| z 1 [95% Conf. Interval] FDIdea .1954777 . 0363947 5.37 0.000 . 1241454 .2668099 LnYO -.0218059 . 0087693 —2.49 0.013 -.0389934 - . 0046184 POPtb - . 0100686 . 0046924 —2.15 0.032 -.0192656 -.0008717 Gov -.0031544 . 0015442 —2.04 0.041 -.0061809 -.0001278 ECONFREE . 0002456 .0006357 0.39 0.699 -.0010003 . 0014916 CAP .3040604 . 0830431 3.66 0.000 . 1412989 .466822 _cons . 082051 . 021838 3.76 0.000 . 0392493 . 1248526 Underidentification test (Kleibergen—Paap rk LM statistic): 6.170 Chi—sq(2) p—val = 0.0457 Weak identification test (Craqq—Donald Wald : F statis tic) : 5 . 985 (Kleibergen—Paap rk Wald F statistic): 12 . 384 Stock—Yogo weak ID test critical values: 10% maximal IV size 19 . 93 15% maximal IV size 11 .59 20% maximal IV size 8 . 75 25% maximal IV size 7 .25 Source: Stock-Yogo (2005) . Reproduced by permission. NB: Critical values are for Cragg—Donald F statistic and i.i.d. errors. Hansen J statistic (overidentification test of all instruments): 0.865 Chi—sq(l) p—val = 0.3523 Instrumented: FDIdea Included instruments: LnYO POPtb Gov ECONFREE CAP Excluded instruments: L.FDIdea L.gRGDPpc ivhettest 2SLS vs GMM (FDIpca) . ivhettest, all IV heteroskedasticity test(s) using levels of IVs only Ho: Disturbance is homoskedastic Pagan-Hall general test statistic : 13.212 Chi-sq(9) p-value = 0.1532 Pagan-Hall test w/assumed normality : 18.129 Chi-sq(9) p-value = 0.0337 White/Koenker nR2 test statistic : 15.485 Chi-sq(9) p-value = 0.0784 Breusch-Pagan/Godfrey/Cook-Weisberg : 25.028 Chi-sq(9) p-value = 0.0029 ivhettest 2SLS vs GMM (FDIbq) . ivhettest, all IV heteroskedasticity test(s) using levels of IVs only Ho: Disturbance is homoskedastic Pagan-Hall general test statistic : 11.529 Chi-sq(9) p-value = 0.2412 Pagan-Hall test w/assumed normality : 16.919 Chi-sq(9) p-value = 0.0500 White/Koenker nR2 test statistic : 12.408 Chi-sq(9) p-value = 0.1913 Breusch-Pagan/Godfrey/Cook-Weisberg : 20.823 Chi-sq(9) p-value = 0.0135 Phụ lục 7.4: Kết quả hồi quy mô hình 4 (2SLS và GMM) (bổ sung cho phần 4.4.2 tr. 99) IV (2SLS) estimation Estimates efficient for homoskedasticity only Statistics consistent for homoskedasticity only Number of obs = 171 F( 9, 161) = 35.57 Prob > F 0.0000 Total (centered) ss = .2496021725 Centered R2 = 0.6528 Total (uncentered) ss = .4605937668 Uncentered R2 = 0.8118 Residual ss = .0866704057 Root MSE .02251 gRGDPpc Coef. std. Err. z p>|z| [95% Conf. Interval] FDIpca .3824492 .1557718 2.46 0.014 .0771421 .6877563 FDIpca2 -.4165662 .2150997 -1.94 0.053 -.8381538 .0050214 LnY -.0045404 .0055439 -0.82 0.413 -.0154063 .0063256 POP -.0166504 .0028632 -5.82 0.000 -.0222621 -.0110387 Gov -.0024009 .001078 -2.23 0.026 -.0045137 -.000288 ECONFREE -.0002074 .0005141 -0.40 0.687 -.0012149 .0008001 CAP .2589862 .0219671 11.79 0.000 .2159316 .3020408 TRADE .0000452 .0000395 1.14 0.253 -.0000323 .0001227 INF -.0009989 .000407 -2.45 0.014 -.0017965 -.0002012 _cons .0419255 .0443789 0.94 0.345 -.0450555 .1289066 Underidentification test (Anderson canon, corr. LM statistic): 53.122 Chi-sq(2) p-val = 0.0000 Weak identification test (Cragg-Donald Wald F statistic): 24.035 Stock-Yogo weak ID test critical values: 10% maximal IV size 13.43 15% maximal IV size 8.18 20% maximal IV size 6.40 25% maximal IV size 5.45 Source: Stock-Yogo (2005). Reproduced by permission. Sargan statistic (overidentification test of all instruments): 0.400 Chi-sq(l) p-val = 0.5273 Instrumented: FDIpca FDIpca2 Included instruments: LnY POP Gov ECONFREE CAP TRADE INF Excluded instruments: L.FDIpca2 L.FDIpca L.gRGDPpc IV (SSLS) estimation Estimates efficient for homoskedasticity only Statistics consistent for homoskedasticity only 171 F( 9, 161) = 39.34 Total (centered) ss = .2496021725 Total (uncentered) ss = .4605937668 Residual ss = .0805627163 Prob > F = Centered R2 = Uncentered R2 = Root MSE = 0.0000 0.6772 0.8251 . 02171 gRGDPpc Coef. std. Err. z p>|z| [95% Conf. Interval] FDIbq .392075 . 0883475 4.44 0.000 .218917 .5652329 FDIbq2 -.5329996 . 1339425 -3.98 0.000 - .7955221 - . 2704772 LnY .0039505 . 0064314 0.61 0.539 -.0086547 .0165558 POP - .0164145 .0027167 -6.04 0.000 -.0217392 -.0110899 Gov -.0017025 . 0011686 -1.46 0.145 -.0039929 .0005878 ECONFREE -.0007981 . 0005628 -1.42 0.156 -.0019011 . 000305 CAP .265779 . 0209102 12.71 0.000 .2247957 .3067623 TRADE .0000658 . 0000405 1.62 0.104 -.0000136 .0001453 INF -.0009122 .0003956 -2.31 0.021 -.0016876 -.0001368 _cons .0144618 . 0371151 0.39 0.697 -.0582824 .0872061 Underidentification test (Anderson canon, corr. LM statistic): 103.556 Chi-sq(2) p-val = 0.0000 Underidentification test (Anderson canon, corr. LM statistic): 103.556 Chi-sq(2) p-val = 0.0000 Weak identification test (Cragg-Donald Wald F statistic): 81.890 Stock-Yogo weak ID test critical values: 10% maximal IV size 13.43 15% maximal IV size 8.18 20% maximal IV size 6.40 25% maximal IV size 5.45 Source: Stock-Yogo (2005). Reproduced by permission. Sarqan statistic (overidentification test of all instruments): 0.813 Chi-sq(l) p-val = 0.3672 Instrumented: FDIbq FDIbq2 Included instruments: LnY POP Gov ECONFREE CAP TRADE INF Excluded instruments: L.FDIbq2 L.FDIbq L.gRGDPpc IV (2SLS) estimation Estimates efficient for homoskedasticity only Statistics consistent for homoskedasticity only Number F( 9, Prob > of obs 80) F 90 11.56 0.0000 Total (centered) ss = .095724225 Total (uncentered) ss = .226759094 Residual ss = .0420270865 Centered R2 = Uncentered R2 = Root MSE = 0.5610 0.8147 . 02161 gRGDPpc Coef. std. Err. z p> 1 z 1 [95% Conf. Interval] FDldea . 8279677 . 7301479 1.13 0.257 -.6030958 2.259031 FDldea2 -.4482376 . 4764784 -0.94 0.347 -1.382118 . 4856429 LnY -.0059988 . 0149426 -0.40 0.688 -.0352856 .0232881 POP -.0178856 . 0046162 -3.87 0.000 -.0269333 -.0088379 Gov -.0044499 . 0018911 — 2.35 0.019 - . 0081564 -.0007435 ECONFREE .0001626 . 0010951 0.15 0.882 -.0019839 . 002309 CAP .3063453 . 043796 6.99 0.000 .2205067 .3921839 TRADE -.0000312 . 0000739 -0.42 0.673 -.0001759 .0001136 INF .0011634 . 0009121 1.28 0.202 -.0006242 .0029511 _cons - .2283304 .3318825 -0.69 0.491 -.8788082 . 4221473 Underidentification test (Anderson canon, corr. LM statistic) : 7.718 Chi-sq(2) p-val = 0.0211 Underidentification test (Anderson canon, corr. LM statistic) : 7.718 Chi-sq(2) p-val = 0.0211 Weak Identification test (Cragg-Donald Wald F statistic) : 2.470 Stock—Yogo weak ID test critical values: 10% maximal IV size 13.43 15% maximal IV size 8.18 20% maximal IV size 6.40 25% maximal IV size 5.45 Source: Stock-Yogo (2005). Reproduced by permission. Sarqan statistic (overidentification test of all instruments): 1.782 Chi-sq(l) p-val = 0.1819 gRGDPpc Coef. Robust std. Err. z p> 1 z 1 [95% Conf. Interval] FDIpca .3877773 . 1400432 2.77 0.006 . 1132976 .662257 FDlpca2 - . 4201563 .1974649 — 2.13 0.033 - . 8071805 -.0331321 LnY - . 0045681 .0051393 -0.89 0.374 - . 014641 . 0055048 POP -.0165409 .0026605 -6.22 0.000 - . 0217554 -. 0113265 Gov - .0023296 .0011012 — 2.12 0.034 - . 0044879 -. 0001713 ECONFREE - . 00025 .0004768 -0.52 0.600 - . 0011845 . 0006844 CAP .2600789 .0209842 12.39 0.000 .2189506 .3012072 TRADE .0000464 . 0000434 1.07 0.284 - .0000386 . 0001315 INF - .0010186 .0005994 -1.70 0.089 - .0021935 .0001563 _cons . 0420272 . 0428054 0.98 0.326 - .0418698 . 1259241 Total (centered.) ss Total (uncentered) ss Residual ss . 2496021725 .4605937668 . 0868881055 Number of obs = F( 9, 161) = Prob > F = Centered R2 = Uncentered R2 = Root MSE = 28.91 0.0000 0.6519 0.8114 . 02254 Underidentification test (Kleibergen-Paap rk LM statistic): 21.953 Chi-sq(2) p-val = 0.0000 Weak, identification test (Cragg-Donald Wald F statistic): 24.035 (Kleibergen-Paap rk Wald F statistic): 9.997 stock—Yogo weak ID test critical values: 10% maximal IV size 13.43 15% maximal IV size 8.18 20% maximal IV size 6.40 25% maximal IV size 5.45 NB: Critical values are for Cragg—Donald F statistic and i.i.d. errors. Source: stock—Yogo (2005) . Reproduced by permission. Hansen J statistic (overidentification test of all instruments): 0.141 Chi-sq(l) p-val = 0.7068 Instrumented: FDlpca FDlpca2 Included instruments: LnY POP Gov ECONFREE CAP TRADE INF Excluded instruments: L.FDlpca2 L.FDlpca L.gRGDPpc 2—step GMM estimation Estimates efficient for arbitrary heteroskedasticity Statistics robust to heteroskedasticity Total (centered) ss Total (uncentered) ss Residual ss .2496021725 .4605937668 .0810248402 Number of obs = F( 9, 161) = Prob > F = Centered R2 = Uncentered R2 = 31.03 0.0000 0.6754 0.8241 gRGDPpc Coef. Robust std. Err. z p> 1 z 1 [95% Conf. Interval] FDIbq .3999731 .0875302 4.57 0.000 . 228417 . 5715292 FDIbq2 - .5486704 .1360358 — 4.03 0.000 -.8152957 - .2820451 LnY .0046365 .0057564 0.81 0.421 -.0066458 .0159188 POP - .0162738 .0026315 -6.18 0.000 -.0214314 -.0111161 Gov -.001497 .0011116 -1.35 0.178 -.0036757 .0006817 ECONFREE - .0009028 .0005062 -1.78 0.074 -.0018949 .0000892 CAP .2669208 .0202989 13.15 0.000 .2271358 .3067058 TRADE .0000698 .0000431 1.62 0.106 -.0000148 .0001544 INF -.000921 .0005181 -1.78 0.075 -.0019366 .0000945 _cons .011428 .0370355 0.31 0.758 -.0611602 .0840163 Root MSE = .02177 Underidentification test (Kleibergen-Paap rk LM statistic): 44.693 Chi-sq(2) p-val = 0.0000 Underidentification test (Kleibergen-Paap rk LM statistic): 44.693 Chi-sq(2) p-val = 0.0000 Weak Identification test (Cragg-Donald Wald F statistic): 81.890 (Kleibergen-Paap rk Wald F statistic): 65.866 stock—Yogo weak ID test critical values: 10% maximal IV size 13.43 15% maximal IV size 8.18 20% maximal IV size 6.40 25% maximal IV size 5.45 NB: Critical values are for Cragg-Donald F statistic and i.i.d. errors. Source: Stock-Yogo (2005). Reproduced by permission. Hansen J statistic (overidentification test of all instruments): 0.329 Chi-sq(l) p-val = 0.5663 2-Step GMM estimation Estimates efficient for arbitrary heteroskedasticity Statistics robust to heteroskedasticity Number of obs = 90 F( 9, 80) = 9.98 Prob > F = 0.0000 Total (centered) ss = .095724225 Centered R2 = 0.5574 Total (uncentered) ss = .226759094 Uncentered R2 = 0.8132 Residual ss = .0423698685 Root MSE .0217 gRGDPpc Coef. Robust std. Err. z p>|z| [95% Conf. Interval] FDIdea .8495531 .7565869 1.12 0.261 -.6333299 2.332436 FDldea2 -.4712867 .5068639 -0.93 0.352 -1.464722 .5221482 LnY -.0031608 .0162181 -0.19 0.845 -.0349476 .028626 POP -.0189803 .0060913 -3.12 0.002 -.0309189 -.0070416 Gov -.0048017 .0014941 -3.21 0.001 -.0077301 -.0018733 ECONFREE .0001592 .0009786 0.16 0.871 -.0017588 .0020772 CAP .3037504 .0581919 5.22 0.000 .1896964 .4178043 TRADE -.0000169 .0000647 -0.26 0.794 -.0001438 .00011 INF .0010499 .0009676 1.08 0.278 -.0008466 .0029463 _cons -.2502002 .3488873 -0.72 0.473 -.9340068 .4336064 Underidentification test (Kleibergen-Paap rk LM statistic): 4.362 Chi-sq(2) p-val = 0.1129 Underidentification test (Kleibergen-Paap rk LM statistic): 4.362 Chi-sq(2) p-val = 0.1129 Weak identification test (Cragg-Donald Wald F statistic): 2.470 (Kleibergen-Paap rk Wald F statistic): 1.324 Stock-Yogo weak ID test critical values: 10% maximal IV size 13.43 15% maximal IV size 8.18 20% maximal IV size 6.40 25% maximal IV size 5.45 NB: Critical values are for Cragg-Donald F statistic and i.i.d. errors. Source: Stock-Yogo (2005). Reproduced by permission. Hansen J statistic (overidentification test of all instruments): 0.737 Chi-sq(l) p-val = 0.3907 Instrumented: FDIdea FDIdea2 Included instruments: LnY POP Gov ECONFREE CAP TRADE INF Excluded instruments: L.FDIdea2 L.FDIdea L.gRGDPpc ivhettest 2SLS vs GMM (FDIpca)- MH bậc 2 . ivhettest, all IV heteroskedasticity test(s) using levels of IVs only Ho: Disturbance is homoskedastic Pagan-Hall general test statistic : 13.699 Chi-sq(lQ) P-value = 0.1872 Pagan-Hall test w/assumed normality : 19.353 Chi-sq(lQ) p-value = Q.Q36Q White/Koenker nR2 test statistic : 21.998 Chi-sq(lQ) P-value = Ũ.Q151 Breusch-Pagan/Godfrey/Cook-Weisberg : 38.232 Chi-sq(lQ) P-value = Q.QQQQ ivhettest 2SLS vs GMM (FDIbq)- MH bậc 2 . ivhettest, all IV heteroskedasticity test(s) using levels of IVs only Ho: Disturbance is homoskedastic Pagan-Hall general test statistic : 16.253 Chi-sq(lO) P-value = 0.0926 Pagan-Hall test w/assumed normality : 25.440 Chi-sq(lO) p-value = 0.0046 White/Koenker nR2 test statistic : 19.398 Chi-sq(lO) P-value = 0.0355 Breusch-Pagan/Godfrey/Cook-Weisberg : 35.763 Chi-sq(lũ) P-value = Ũ.ŨŨŨ1 Phụ lục 7.5: Kết quả hồi quy với các nhóm chỉ số thành phần (bổ sung cho phần 4.4.2 tr. 99) 2-Step GMM estimation Estimates efficient for arbitrary heteroskedasticity Statistics robust to heteroskedasticity Number of obs = 171 F( 8, 162) = 34.61 Prob > F 0.0000 Total (centered) ss = . 2496021725 Centered R2 = 0.6643 Total (uncentered) ss = .4605937668 Uncentered R2 = 0.8181 Residual ss = .0837930196 Root MSE .02214 gRGDPpc Coef. Robust std. Err. z p> 1 z 1 [95% Conf. Interval] FDIi .0308459 .0216245 1.43 0.154 -.0115372 .0732291 LnY -.0081058 .0027827 -2.91 0.004 -.0135598 -.0026518 POP -.0171573 .0026456 -6.49 0.000 -.0223426 -.0119721 Gov -.0017448 .0009839 -1.77 0.076 -.0036731 .0001836 ECONFREE .0000352 .0003967 0.09 0.929 -.0007423 .0008127 CAP .2699504 .0197149 13.69 0.000 .23131 .3085908 TRADE .0000667 .0000382 1.75 0.081 -8.14e-06 .0001415 INF -.0009938 .0005498 -1.81 0.071 -.0020715 .0000838 _cons .112827 .0156671 7.20 0.000 .0821201 .143534 Underidentification test (Kleibergen- Paap rk LM statistic): Chi-sq(2) p-val = 51.140 0.0000 Weak identification test (Cragg-Donald Wald : F statistic) 131. 992 (Kleibergen-Paap rk Wald F statistic): 106. 620 Stock-Yogo weak ID test critical values : 10% maximal IV size 19 . 93 15% maximal IV size 11 .59 20% maximal IV size 8 .75 25% maximal IV size 7 .25 Source: Stock-Yogo (2005) . Reproduced by permission NB: Critical values are for Cragg-Donald F statistic and i.i.d. errors. 0.079 Hansen J statistic (overidentification test of all instruments): Chi-sq(l) p-val = 0.7788 Instrumented: FDIi Included instruments: LnY POP Gov ECONFREE CAP TRADE INF Excluded instruments: L.FDIi L.gRGDPpc Estimates efficient for homoskedasticity only Statistics consistent for homoskedasticity only Number of obs = 171 F( 8, 162) = 43.98 Prob > F 0.0000 Total (centered) ss = .2496021725 Centered R2 = 0.6882 Total (uncentered) ss = .4605937668 Uncentered R2 = 0.8310 Residual ss = .0778200465 Root MSE .02133 gRGDPpc Coef. std. Err. z p>|z| [95% Conf. Interval] FDIm .0496773 .0313481 1.58 0.113 -.0117639 .1111185 LnY -.0094563 .0038369 -2.46 0.014 -.0169766 -.001936 POP -.0169792 .0026866 -6.32 0.000 -. 0222449 - .0117136 Gov -.0013351 . 000843 -1.58 0.113 -. 0029874 .0003173 ECONFREE .0000263 .0004365 0.06 0.952 -.0008292 .0008817 CAP .2484065 .0232276 10.69 0.000 .2028813 .2939317 TRADE .0000796 .0000326 2.45 0.014 .0000158 .0001434 INF -.001123 .0003902 -2.88 0.004 -.0018877 -.0003583 _cons .1191384 .0162801 7.32 0.000 . 0872299 .1510469 Underidentification test (Anderson canon . corr. LM statistic): Chi-sq(2) p-val = 92.718 0.0000 Weak identification test (Cragg-Donald Wald F statistic) 95.345 Stock-Yogo weak ID test critical values : 10% maximal IV size 19.93 15% maximal IV size 11.59 20% maximal IV size 8.75 25% maximal IV size 7.25 Source: Stock-Yogo (2005). Reproduced by permission. 0.027 Sargan statistic (overidentification test of all instruments): Chi-sq(l) p-val = 0.8693 Instrumented: FDIm Included instruments: LnY POP Gov ECONFREE CAP TRADE INF Excluded instruments: L.FDIm L.gRGDPpc Estimates efficient for arbitrary heteroskedasticity Statistics robust to heteroskedasticity Number of obs = 171 F( 8, 162) = 35.10 Prob > F = 0.0000 Total (centered) ss = .2496021725 Centered R2 = 0.6687 Total (uncentered) ss = .4605937668 Uncentered R2 = 0.8205 Residual ss = .0826899803 Root MSE .02199 gRGDPpc Coef. Robust std. Err. z p>|z| [95% Conf. Interval] FDIibq .0788915 . 0446974 1.77 0.078 -.0087139 .1664969 LnY -.0089724 . 0028729 -3.12 0.002 -.0146031 -.0033417 POP -.0170712 .0026005 -6.56 0.000 -.022168 -.0119743 Gov -.0020526 .0010439 -1.97 0.049 -.0040985 -6.68e-06 ECONFREE .0001094 .0003951 0.28 0.782 -.000665 .0008839 CAP .2707726 .0197695 13.70 0.000 .2320252 .30952 TRADE .0000598 .0000385 1.55 0.120 -.0000156 .0001352 INF -.0010058 .0005297 -1.90 0.058 -.0020439 .0000324 _cons .1173099 .016202 7.24 0.000 .0855545 .1490653 Underidentification test (Kleibergen- Paap rk LM statistic): Chi-sq(2) p-val = 63.066 0.0000 Weak identification test (Cragg-Donald Wald F statistic): (Kleibergen-Paap rk Wald F statistic): 463.541 654.238 Stock-Yogo weak ID test c ritical values: 10% maximal IV size 19.93 15% maximal IV size 11.59 20% maximal IV size 8.75 25% maximal IV size 7.25 Source: Stock-Yogo (2005). Reproduced by permission. NB: Critical values are for Cragg-Donald F statistic and i.i.d. errors. 0.092 Hansen J statistic (overidentification test of all instruments): Chi-sq(l) p-val = 0.7613 Instrumented: FDIibq Included instruments: LnY POP Gov ECONFREE CAP TRADE INF Excluded instruments: L.FDIibq L.gRGDPpc Estimates efficient for arbitrary heteroskedasticity Statistics robust to heteroskedasticity Number of obs = 171 F( 8, 162) = 38.55 Prob > F = 0.0000 Total (centered) ss = .2496021725 Centered R2 = 0.6781 Total (uncentered) ss = .4605937668 Uncentered R2 = 0.8256 Residual ss = .0803492368 Root MSE .02168 gRGDPpc Coef. Robust std. Err. z p>|z| [95% Conf. Interval] FDImbq .0304682 .0510383 0.60 0.551 -.069565 .1305015 LnY -.007775 .0037013 -2.10 0.036 -.0150295 -.0005205 POP -.0177141 .0026057 -6.80 0.000 -.0228211 -.0126071 Gov -.0012134 .0009123 -1.33 0.183 -.0030015 .0005746 ECONFREE -9.00e-06 . 0004065 -0.02 0.982 -.0008057 .0007877 CAP .261427 . 0214619 12.18 0.000 . 2193624 .3034915 TRADE .0000824 .0000383 2.15 0.032 7.27e-06 .0001576 INF -.0010741 .000473 -2.27 0.023 -.0020011 -.000147 _cons .1167859 .0203397 5.74 0.000 .0769208 .1566509 Underidentification test (Kleibergen- Paap rk LM statistic): Chi-sq(2) p-val = 48.823 0.0000 Weak identification test (Cragg-Donald Wald : F statistic) 155 . 728 (Kleibergen-Paap rk Wald F statistic): 123. 414 Stock-Yogo weak ID test critical values : 10% maximal IV size 19 . 93 15% maximal IV size 11 .59 20% maximal IV size 8 . 75 25% maximal IV size 7 .25 Source: Stock-Yogo (2005) . Reproduced by permission NB: Critical values are for Cragg-Donald F statistic and i.i.d. errors. Hansen J statistic (overidentification test of all instruments): 0.163 Chi-sq(l) p-val = 0.6864 Instrumented: FDImbq Included instruments: LnY POP Gov ECONFREE CAP TRADE INF Excluded instruments: L.FDImbq L.gRGDPpc Phụ lục 7.6: Kết quả hồi quy với các chỉ số tài chính đơn lẻ (bổ sung cho phần 4.4.2 tr. 99) IV (2SLS) estimation Instrumented: Included instruments: Excluded instruments: Estimates efficient for homoskedasticity only Statistics consistent for homoskedasticity only Number of obs = 167 F( 10, 156) = 36.30 Prob > F 0.0000 Total (centered) ss = .2483139065 Centered R2 = 0.7007 Total (uncentered) ss = .4507615019 Uncentered R2 = 0.8351 Residual ss = .0743308296 Root MSE . 0211 gRGDPpc Coef. std. Err. z p>|z| [95% Conf. Interval] LnM3 .0000376 . 0098375 0.00 0.997 -.0192435 .0193187 LnPricred .0032795 . 0081249 0.40 0.686 -. 0126451 .019204 LnMkvalue .0049556 . 0015083 3.29 0 .001 . 0019993 .0079119 LnY -.0084123 . 0036348 -2.31 0 . 021 -.0155363 -.0012883 POP -.0176148 .0027561 -6.39 0 .000 -.0230166 -.0122129 Gov -.0026518 .0010754 -2.47 0 .014 -.0047596 -.0005441 ECONFREE -.0002975 . 0005023 -0.59 0.554 -.0012819 .000687 CAP .2536074 . 0218942 11.58 0 .000 .2106955 .2965192 TRADE .0000716 . 0000404 1.77 0 . 077 -7.66e-06 .0001509 INF -.0014498 . 0004029 -3.60 0 .000 -. 0022394 -.0006601 _cons .1354143 . 0272981 4.96 0 .000 .081911 .1889177 Underidentification test (Anderson canon, cor r. LM statistic): 143.640 Chi -sq(2) p-val = 0.0000 Weak identification test (Cragg-Donald Wald F statistic) 238.276 Stock-Yogo weak ID test critical values <not available> Sargan statistic (overidentification te st of all instruments): 0.003 Chi -sq(l) p-val = 0.9534 LnM3 LnPricred LnMkvalue LnY POP Gov ECONFREE CAP TRADE INF L.LnM3 L.LnPricred L.LnMkvalue L.gRGDPpc Estimates efficient for homoskedasticity only Statistics consistent for homoskedasticity only Chi-sq(2) p-val = 0.0000 Weak identification test (Cragg-Donald Wald F statistic): Stock-Yogo weak ID test critical values: 238.276 Sargan statistic (overidentification test of all instruments): 0.003 Chi-sq(l) p-val = 0.9534 Numbe r of obs = 167 F( 10 , 156) = 36.30 Prob > F 0.0000 Total (centered) ss = .2483139065 Cente red R2 0.7007 Total (uncentered) ss = . 4507615019 Uncentered R2 = 0.8351 Residual ss = .0743308296 Root : MSE .0211 gRGDPpc Coef. std. Err. z p>| z 1 [95% Conf. Interval] LnM3 .0000376 .0098375 0.00 0.997 -.0192435 .0193187 LnPricred .0032795 .0081249 0.40 0.686 -.0126451 .019204 LnMkvalue .0049556 .0015083 3.29 0.001 .0019993 .0079119 LnY -.0084123 .0036348 -2.31 0.021 -.0155363 -.0012883 POP -.0176148 .0027561 -6.39 0.000 -.0230166 -.0122129 Gov -.0026518 .0010754 -2.47 0.014 -.0047596 -.0005441 ECONFREE -.0002975 .0005023 -0.59 0.554 -.0012819 .000687 CAP .2536074 . 0218942 11.58 0.000 .2106955 .2965192 TRADE .0000716 .0000404 1.77 0.077 -7.66e-06 .0001509 INF -.0014498 .0004029 -3.60 0.000 -.0022394 -.0006601 _cons . 1354143 . 0272981 4.96 0.000 .081911 .1889177 Underidentification test (Anderson canon, corr. LM statistic): 143.640 Instrumented: LnM3 LnPricred LnMkvalue Included instruments: LnY POP Gov ECONFREE CAP TRADE INF Excluded instruments: L.LnM3 L.LnPricred L.LnMkvalue L.gRGDPpc