Ứng dụng mô hình var và cvar để đo lường rủi ro các cổ phiếu ngành ngân hàng niêm yết trên sở giao dịch chứng khoán hồ chí minh

ht ếH uế - Trư ờn gĐ ại h ọc Kin – - 2022 Style Definition: TOC – ọc ại h ờn gĐ Trư – ht ếH uế Kin - 04 ăm 2022 ọc ại h ờn gĐ Trư – – i ht ếH uế Kin - – ht ếH uế - i ii v vi viii – Kin ọc 3 Đối tượng phạm vi nghiên cứu ại h 4 Phương pháp nghiên cứu tc c h n – ờn gĐ CVaR 1.1 1.1.2 1.1.3 1.2 1.2.1 m Trư 1.1.1 10 10 ii ht ếH uế - m 10 1.2.1.1 m 1.2.1.2 11 1.2.1.3 11 m 1.2.1.4 1.2.2 11 11 1.3 12 – 1.3.1 m 16 Kin 1.3.2 – Covariance) 13 1.3.3 18 1.3.4 25 ọc 27 28 ại h 1.5 30 2.1 ờn gĐ 30 2.2 31 2.3 33 2.3.1 m m 2.3.2 38 2.3.5 3.1 -GARCH 47 Trư 2.3.3 2.3.4 33 mm 49 50 54 54 iii 3.2 ht ếH uế - 56 – 59 61 Trư ờn gĐ ại h ọc Kin 62 iv ACF Autocorrelation Function ( ADF AIC ởR Dickey Và Fuller Akaike Information Criteria ARMA Autoregressive Integrated Moving Average CAPM Capital Asset Pricing Model (Mô CVaR Conditional Value At Risk (G JB Expected Shortfall ( Generalized Autoregressive Conditional Heterokedasticity (Mơ Hình Jarque-Bera NHNN Ngân P&L Profit PACF Partial Autocorrelation Function ( Loss ( TSSL R ại h R ờn gĐ Value At Risk (G Trư VaR R Kin GARCH R ọc ES G ht ếH uế - v R ht ếH uế - 31 36  37 39 -MA(5,13) 40 R -MA(5,13) 40 Kin R 41 R ọc 41 ại h 42 BID 43 R ờn gĐ 43 G R G R 45 45 46 R GARCH(1,1) 46 G R 47 Trư R -GARCH 47 48 R 99% 49 R 99% 49 vi ht ếH uế - -2021 50 - 52 R 95% 54 R GARCH(1,1) 54 R99% 56 Trư ờn gĐ ại h ọc Kin - vii ht ếH uế - DANH R R 12 14 35 Kin – 31/12/2021 37 TSSL BID 39 -2021 44 Trư ờn gĐ ại h ọc R viii -202152 -202155 R-squared Adjusted R-squared S.E of regression Sum squared resid Log likelihood Durbin-Watson stat 0.010089 0.008105 0.021821 0.237609 1222.023 2.030997 ht ếH uế - Mean dependent var 0.001757 S.D dependent var 0.02191 Akaike info criterion -4.86237 Schwarz criterion -4.8287 Hannan-Quinn criter -4.84916 Coefficient C MA(13) 0.002521 0.076855 Std Error z-Statistic ại h Variable ọc Kin Dependent Variable: MBB_ Method: ML ARCH - Normal distribution (BFGS / Marquardt steps) Date: 04/18/22 Time: 16:35 Sample: 1/03/2020 12/31/2021 Included observations: 501 Convergence achieved after 30 iterations Coefficient covariance computed using outer product of gradients MA Backcast: 12/17/2019 1/02/2020 Presample variance: backcast (parameter = 0.7) GARCH = C(3) + C(4)*RESID(-1)^2 + C(5)*RESID(-2)^2 + C(6)*GARCH(-1) 0.001014 0.044643 ờn gĐ Variance Equation 4.00E-05 0.191094 -0.08641 0.811313 R-squared Adjusted R-squared S.E of regression Sum squared resid Log likelihood Durbin-Watson stat 0.00938 0.007394 0.021829 0.23778 1236.519 2.031773 Trư C RESID(-1)^2 RESID(-2)^2 GARCH(-1) 1.64E-05 0.053081 0.055255 0.057693 2.486538 1.721552 0.0129 0.0852 2.446049 3.600068 -1.56381 14.06263 0.0144 0.0003 0.1179 Mean dependent var S.D dependent var Akaike info criterion Schwarz criterion Hannan-Quinn criter Heteroskedasticity Test: ARCH 74 Prob 0.001757 0.02191 -4.91225 -4.86175 -4.89244 F-statistic Obs*R-squared 0.219658 0.220443 ht ếH uế - Prob F(1,498) 0.6395 Prob Chi-Square(1) 0.6387 TCB C MA(13) 0.002056 0.07753 C RESID(-1)^2 GARCH(-1) 6.81E-05 0.117917 0.755012 R-squared S.E of regression Sum squared resid Log likelihood Durbin-Watson stat 2.38E-05 0.031988 0.065787 Mean 0.009385 dependent var S.D 0.007399 dependent var Akaike info 0.023033 criterion Schwarz 0.264734 criterion Hannan1195.934 Quinn criter 1.957396 Trư Adjusted R-squared 0.001104 0.049739 ờn gĐ Variance Equation Std Error ọc Coefficient z-Statistic ại h Variable Kin Dependent Variable: TCB_ Method: ML ARCH Normal distribution (BFGS / Marquardt steps) Included observations: 501 after adjustments GARCH = C(3) + C(4)*RESID(-1)^2 + C(5)*GARCH(-1) 75 Prob 1.862602 1.558745 0.0625 0.1191 2.856473 3.686317 11.47667 0.0043 0.0002 0.00175 0.023119 -4.754228 -4.712146 -4.737717 Dependent Variable: TCB_ Method: ML ARCH - Normal distribution (BFGS / Marquardt steps) Date: 04/18/22 Time: 16:37 Sample: 1/03/2020 12/31/2021 Included observations: 501 Convergence achieved after 12 iterations Coefficient covariance computed using outer product of gradients MA Backcast: 12/17/2019 1/02/2020 Presample variance: backcast (parameter = 0.7) GARCH = C(3) + C(4)*RESID(-1)^2 Coefficient Std Error C MA(13) 0.001812 0.094127 z-Statistic 0.001152 0.043016 1.572602 2.18817 R-squared Adjusted R-squared S.E of regression Sum squared resid Log likelihood Durbin-Watson stat 0.01013 0.008146 0.023025 0.264535 1186.124 1.957969 2.58E-05 0.048745 17.52458 2.89457 0.0038 ại h 0.000453 0.141097 Mean dependent var 0.00175 S.D dependent var 0.023119 Akaike info criterion -4.71906 Schwarz criterion -4.68539 Hannan-Quinn criter -4.70585 ờn gĐ C RESID(-1)^2 0.1158 0.0287 ọc Variance Equation Prob Kin Variable ht ếH uế - Trư Dependent Variable: TCB_ Method: ML ARCH - Normal distribution (BFGS / Marquardt steps) Date: 04/18/22 Time: 16:38 Sample: 1/03/2020 12/31/2021 Included observations: 501 Convergence achieved after 29 iterations Coefficient covariance computed using outer product of gradients MA Backcast: 12/17/2019 1/02/2020 Presample variance: backcast (parameter = 0.7) GARCH = C(3) + C(4)*RESID(-1)^2 + C(5)*RESID(-2)^2 + C(6)*GARCH(-1) Variable Coefficient Std Error 76 z-Statistic Prob C MA(13) 0.002019 0.083704 0.001128 0.04931 1.789951 1.697488 0.0735 0.0896 Variance Equation R-squared Adjusted R-squared S.E of regression Sum squared resid Log likelihood Durbin-Watson stat 0.009703 0.007719 0.023029 0.264649 1196.919 1.95757 1.81E-05 0.048248 0.048493 0.055671 Mean dependent var S.D dependent var Akaike info criterion Schwarz criterion Hannan-Quinn criter Variable C Trư Dependent Variable: HDB_ Method: ML ARCH - Normal distribution (BFGS / Marquardt steps) Included observations: 485 after adjustments GARCH = C(3) + C(4)*RESID(-1)^2 + C(5)*GARCH(-1) 0.00175 0.023119 -4.75417 -4.70367 -4.73436 Prob F(1,498) 0.2579 Prob Chi-Square(1) 0.257 ờn gĐ 1.282986 1.284828 HDB 0.0278 0.0012 0.112 ại h Heteroskedasticity Test: ARCH F-statistic Obs*R-squared 2.199818 3.246934 -1.58941 15.1939 Kin 3.99E-05 0.15666 -0.07708 0.84586 ọc C RESID(-1)^2 RESID(-2)^2 GARCH(-1) ht ếH uế - Coefficient 0.002585 Std Error 0.001024 77 z-Statistic 2.524037 Prob 0.0116 AR(16) -0.081579 ht ếH uế - 0.042979 -1.898102 0.0577 4.46E-05 0.043537 0.102078 3.132176 3.296127 6.050101 0.0017 0.001 Variance Equation 0.00014 0.143505 0.61758 R-squared S.E of regression Sum squared resid ại h Log likelihood Durbin-Watson stat 0.002063 ọc Adjusted R-squared Mean 0.00833 dependent var S.D 0.006277 dependent var Akaike info 0.024215 criterion Schwarz 0.283223 criterion Hannan1131.893 Quinn criter 1.841973 Kin C RESID(-1)^2 GARCH(-1) C AR(16) Coefficient Std Error 0.002459 -0.07542 z-Statistic 0.001044 0.042201 Trư Variable ờn gĐ Dependent Variable: HDB_ Method: ML ARCH - Normal distribution (BFGS / Marquardt steps) Date: 04/18/22 Time: 16:41 Sample (adjusted): 2/03/2020 12/31/2021 Included observations: 485 after adjustments Convergence achieved after 11 iterations Coefficient covariance computed using outer product of gradients Presample variance: backcast (parameter = 0.7) GARCH = C(3) + C(4)*RESID(-1)^2 Prob 2.354487 -1.7872 0.0185 0.0739 16.51912 3.094198 0.002 Variance Equation C RESID(-1)^2 0.000474 0.195346 2.87E-05 0.063133 78 0.024292 -4.64698 -4.603845 -4.630032 R-squared Adjusted R-squared S.E of regression Sum squared resid Log likelihood Durbin-Watson stat ht ếH uế - 0.008349 0.006296 0.024215 0.283218 1125.989 1.842688 Mean dependent var 0.002063 S.D dependent var 0.024292 Akaike info criterion -4.62676 Schwarz criterion -4.59225 Hannan-Quinn criter -4.6132 Coefficient C AR(16) 0.00254 -0.08533 Std Error 0.001026 0.044148 Variance Equation 3.63E-05 0.170509 -0.10985 0.878194 R-squared Adjusted R-squared S.E of regression Sum squared resid Log likelihood Durbin-Watson stat 0.008513 0.00646 0.024213 0.283171 1132.811 1.841966 2.13E-05 0.058751 0.062521 0.05972 Trư ờn gĐ C RESID(-1)^2 RESID(-2)^2 GARCH(-1) z-Statistic 0.0133 0.0533 1.701438 2.902218 -1.757 14.70513 0.0889 0.0037 0.0789 Mean dependent var S.D dependent var Akaike info criterion Schwarz criterion Hannan-Quinn criter Heteroskedasticity Test: ARCH F-statistic 0.080954 Prob F(1,482) 79 Prob 2.476781 -1.93275 ại h Variable ọc Kin Dependent Variable: HDB_ Method: ML ARCH - Normal distribution (BFGS / Marquardt steps) Date: 04/18/22 Time: 16:42 Sample (adjusted): 2/03/2020 12/31/2021 Included observations: 485 after adjustments Convergence achieved after 28 iterations Coefficient covariance computed using outer product of gradients Presample variance: backcast (parameter = 0.7) GARCH = C(3) + C(4)*RESID(-1)^2 + C(5)*RESID(-2)^2 + C(6)*GARCH(-1) 0.7761 0.002063 0.024292 -4.64664 -4.59488 -4.6263 Obs*R-squared 0.081276 Prob Chi-Square(1) 0.7756 C MA(19) ọc ại h ờn gĐ Variable Coefficient Trư Dependent Variable: STB_ Method: ML ARCH - Normal distribution (BFGS / Marquardt steps) Date: 03/21/22 Time: 13:46 Sample (adjusted): 1/03/2020 12/31/2021 Included observations: 501 after adjustments Convergence achieved after 22 iterations Coefficient covariance computed using outer product of gradients MA Backcast: 12/09/2019 1/02/2020 Presample variance: backcast (parameter = 0.7) GARCH = C(3) + C(4)*RESID(-1)^2 + C(5)*GARCH(-1) Kin STB ht ếH uế - 0.003134 -0.090104 Std Error z-Statistic Prob 0.001062 0.04369 2.951226 -2.062351 0.0032 0.0392 4.14E-05 2.715173 0.0066 Variance Equation C 0.000112 80 0.136473 0.705058 0.045467 0.085707 Mean 0.008845 dependent var S.D 0.006859 dependent var Akaike info 0.026882 criterion Schwarz 0.360598 criterion Hannan1123.93 Quinn criter R-squared Adjusted Rsquared S.E of regression Sum squared resid Log likelihood Durbin-Watson stat 1.957104 3.001583 8.226343 0.026975 -4.466787 -4.424705 ờn gĐ ại h ọc Dependent Variable: STB_ Method: ML ARCH - Normal distribution (BFGS / Marquardt steps) Date: 04/18/22 Time: 16:44 Sample: 1/03/2020 12/31/2021 Included observations: 501 Convergence achieved after 28 iterations Coefficient covariance computed using outer product of gradients MA Backcast: 12/09/2019 1/02/2020 Presample variance: backcast (parameter = 0.7) GARCH = C(3) + C(4)*RESID(-1)^2 Variable Coefficient Std Error C MA(19) 0.003303 -0.09806 z-Statistic 0.001047 0.045764 Prob 3.154805 -2.14264 0.0016 0.0321 14.31469 3.355721 0.0008 Mean dependent var S.D dependent var Akaike info criterion Schwarz criterion 0.002616 0.026975 -4.44132 -4.40766 Variance Equation 0.000538 0.249388 3.76E-05 0.074317 Trư C RESID(-1)^2 R-squared Adjusted R-squared S.E of regression Sum squared resid 0.008589 0.006602 0.026885 0.360691 81 0.0027 0.002616 Kin RESID(-1)^2 GARCH(-1) ht ếH uế - -4.450276 Log likelihood Durbin-Watson stat 1116.551 1.95676 Hannan-Quinn criter ht ếH uế - -4.42811 Coefficient C MA(19) 0.003116 -0.08964 Std Error 0.001067 0.043557 0.000116 0.128847 0.013752 0.693199 R-squared Adjusted R-squared S.E of regression Sum squared resid Log likelihood Durbin-Watson stat 0.008867 0.006881 0.026882 0.36059 1123.947 1.957146 Trư 0.0035 0.0396 0.019 0.0297 0.8495 Mean dependent var S.D dependent var Akaike info criterion Schwarz criterion Hannan-Quinn criter 0.002616 0.026975 -4.46286 -4.41237 -4.44305 4.97E-05 0.059272 0.072464 0.110318 ờn gĐ C RESID(-1)^2 RESID(-2)^2 GARCH(-1) 2.919033 -2.05798 Prob ại h Variance Equation z-Statistic ọc Variable Kin Dependent Variable: STB_ Method: ML ARCH - Normal distribution (BFGS / Marquardt steps) Date: 04/18/22 Time: 16:45 Sample: 1/03/2020 12/31/2021 Included observations: 501 Convergence achieved after 23 iterations Coefficient covariance computed using outer product of gradients MA Backcast: 12/09/2019 1/02/2020 Presample variance: backcast (parameter = 0.7) GARCH = C(3) + C(4)*RESID(-1)^2 + C(5)*RESID(-2)^2 + C(6)*GARCH(-1) 2.345453 2.173815 0.189785 6.283623 Heteroskedasticity Test: ARCH F-statistic Obs*R-squared 0.006841 0.006868 Prob F(1,498) 0.9341 Prob Chi-Square(1) 0.934 82 ht ếH uế - Variable C AR(22) ọc ại h ờn gĐ Dependent Variable: TPB_ Method: ML ARCH - Normal distribution (BFGS / Marquardt steps) Date: 03/21/22 Time: 13:49 Sample (adjusted): 2/11/2020 12/31/2021 Included observations: 479 after adjustments Convergence achieved after 25 iterations Coefficient covariance computed using outer product of gradients Presample variance: backcast (parameter = 0.7) GARCH = C(3) + C(4)*RESID(-1)^2 + C(5)*GARCH(-1) Kin TPB Coefficient 0.003003 -0.087103 Std Error z-Statistic Prob 0.001018 0.040862 2.949538 -2.131621 0.0032 0.033 5.94E-05 0.066161 0.122027 3.93667 3.861479 2.991792 0.0001 0.0001 0.0028 C RESID(-1)^2 GARCH(-1) R-squared Trư Variance Equation 0.000234 0.255478 0.36508 Mean 0.008629 dependent var 83 0.002761 Adjusted Rsquared S.D 0.00655 dependent var Akaike info 0.024596 criterion Schwarz 0.288563 criterion Hannan1116.083 Quinn criter S.E of regression Sum squared resid Log likelihood Durbin-Watson stat ht ếH uế - 0.024677 -4.639176 -4.59563 -4.622058 1.946903 Variable ại h ọc Kin Dependent Variable: TPB_ Method: ML ARCH - Normal distribution (BFGS / Marquardt steps) Date: 04/18/22 Time: 16:47 Sample (adjusted): 2/11/2020 12/31/2021 Included observations: 479 after adjustments Convergence achieved after 12 iterations Coefficient covariance computed using outer product of gradients Presample variance: backcast (parameter = 0.7) GARCH = C(3) + C(4)*RESID(-1)^2 Coefficient Std Error 0.002664 -0.08593 0.001033 0.040294 2.579027 -2.13245 ờn gĐ C AR(22) z-Statistic Prob 0.0099 0.033 Variance Equation 0.00043 0.292782 R-squared Adjusted R-squared S.E of regression Sum squared resid Log likelihood Durbin-Watson stat 0.008738 0.00666 0.024594 0.288531 1114.348 1.947046 2.91E-05 0.074691 14.79614 3.91991 0.0001 Mean dependent var 0.002761 S.D dependent var 0.024677 Akaike info criterion -4.63611 Schwarz criterion -4.60127 Hannan-Quinn criter -4.62241 Trư C RESID(-1)^2 Dependent Variable: TPB_ Method: ML ARCH - Normal distribution (BFGS / Marquardt steps) 84 ht ếH uế - Date: 04/18/22 Time: 16:48 Sample (adjusted): 2/11/2020 12/31/2021 Included observations: 479 after adjustments Convergence achieved after 23 iterations Coefficient covariance computed using outer product of gradients Presample variance: backcast (parameter = 0.7) GARCH = C(3) + C(4)*RESID(-1)^2 + C(5)*RESID(-2)^2 + C(6)*GARCH(-1) Coefficient C AR(22) 0.002896 -0.08726 Std Error z-Statistic 0.001033 0.04011 2.80432 -2.17542 0.008721 0.006643 0.024595 0.288536 1116.428 1.947094 1.355789 3.793324 -0.77794 1.8873 Mean dependent var S.D dependent var Akaike info criterion Schwarz criterion Hannan-Quinn criter ại h R-squared Adjusted R-squared S.E of regression Sum squared resid Log likelihood Durbin-Watson stat 0.000118 0.073048 0.131814 0.30039 ờn gĐ 0.000159 0.277096 -0.10254 0.566927 0.005 0.0296 ọc Variance Equation C RESID(-1)^2 RESID(-2)^2 GARCH(-1) Prob Kin Variable 0.1752 0.0001 0.4366 0.0591 0.002761 0.024677 -4.63645 -4.58419 -4.6159 Heteroskedasticity Test: ARCH EIB Dependent Variable: EIB_ Method: ML ARCH - Normal distribution (BFGS / Marquardt steps) 0.019357 0.019437 Prob F(1,476) 0.8894 Prob Chi-Square(1) 0.8891 Trư F-statistic Obs*R-squared 85 Included observations: 490 after adjustments GARCH = C(5) + C(6)*RESID(-1)^2 + C(7)*GARCH(-1) C AR(11) MA(11) MA(15) Coefficient 0.000388 -0.709907 0.653408 -0.041204 Std Error z-Statistic 0.000776 0.122864 0.130247 0.025871 Variance Equation S.E of regression Sum squared resid Log likelihood Durbin-Watson stat 0.6169 0 0.1112 6.444466 5.334774 21.40623 0 ọc 3.91E-06 0.046089 0.034798 0.50029 -5.778008 5.016687 -1.592644 Mean 0.016052 dependent var S.D 0.009978 dependent var Akaike info 0.022739 criterion Schwarz 0.251299 criterion Hannan1223.504 Quinn criter ại h R-squared Adjusted Rsquared 2.52E-05 0.245875 0.744884 ờn gĐ C RESID(-1)^2 GARCH(-1) 1.826373 Trư Dependent Variable: EIB_ Method: ML ARCH - Normal distribution (BFGS / Marquardt steps) Date: 04/18/22 Time: 16:52 Sample (adjusted): 1/20/2020 12/31/2021 Included observations: 490 after adjustments Convergence achieved after 23 iterations Coefficient covariance computed using outer product of gradients MA Backcast: 12/30/2019 1/17/2020 Presample variance: backcast (parameter = 0.7) 86 Prob Kin Variable ht ếH uế - 0.001562 0.022854 -4.965321 -4.905401 -4.941789 GARCH = C(5) + C(6)*RESID(-1)^2 Variable Coefficient Std Error C AR(11) MA(11) MA(15) 0.000569 -0.5206 0.407674 -0.05423 z-Statistic 0.001004 0.229908 0.241783 0.036626 Prob 0.566443 -2.26439 1.686117 -1.48062 0.5711 0.0235 0.0918 0.1387 19.05104 3.796013 0.0001 0.01863 0.012572 0.02271 0.250641 1182.39 1.822809 Mean dependent var 0.001562 S.D dependent var 0.022854 Akaike info criterion -4.80159 Schwarz criterion -4.75023 Hannan-Quinn criter -4.78142 ọc R-squared Adjusted R-squared S.E of regression Sum squared resid Log likelihood Durbin-Watson stat 1.99E-05 0.068702 ại h 0.000378 0.260796 Kin Variance Equation C RESID(-1)^2 ht ếH uế - C AR(11) MA(11) MA(15) Coefficient Trư Variable ờn gĐ Dependent Variable: EIB_ Method: ML ARCH - Normal distribution (BFGS / Marquardt steps) Date: 04/18/22 Time: 16:52 Sample (adjusted): 1/20/2020 12/31/2021 Included observations: 490 after adjustments Convergence achieved after 35 iterations Coefficient covariance computed using outer product of gradients MA Backcast: 12/30/2019 1/17/2020 Presample variance: backcast (parameter = 0.7) GARCH = C(5) + C(6)*RESID(-1)^2 + C(7)*RESID(-2)^2 + C(8)*GARCH(-1) 0.000281 -0.7007 0.639031 -0.04075 Std Error 0.000771 0.125207 0.134581 0.028237 Variance Equation 87 z-Statistic 0.364435 -5.59633 4.748284 -1.44314 Prob 0.7155 0 0.149 2.30E-05 0.283673 -0.05396 0.763074 R-squared Adjusted R-squared S.E of regression Sum squared resid Log likelihood Durbin-Watson stat 0.015449 0.009371 0.022746 0.251453 1223.764 1.824767 4.05E-06 0.083961 0.066043 0.034918 5.67882 3.378646 -0.81706 21.85311 Mean dependent var S.D dependent var Akaike info criterion Schwarz criterion Hannan-Quinn criter Heteroskedasticity Test: ARCH 0.001562 0.022854 -4.9623 -4.89382 -4.93541 ọc Prob F(1,487) 0.3798 Prob Chi-Square(1) 0.3788 ờn gĐ ại h 0.772791 0.774736 Trư F-statistic Obs*R-squared 0.0007 0.4139 Kin C RESID(-1)^2 RESID(-2)^2 GARCH(-1) ht ếH uế - 88