library(readxl)
library(rugarch)
## Warning: package 'rugarch' was built under R version 4.3.1
## Loading required package: parallel
## 
## Attaching package: 'rugarch'
## The following object is masked from 'package:stats':
## 
##     sigma
data <- read_excel('C:/Users/Thanh Lan/Documents/Zalo Received Files/data.xlsx')
SGX <- ts(data$SGX)

MÔ HÌNH ARMA

MÔ HÌNH PHÂN PHỐI BIÊN

GJR-GARCH(11)

PHÂN PHỐI CHUẨN

sgx11n.spec <- ugarchspec(variance.model = list(model = 'gjrGARCH', garchOrder = c(1,1)), mean.model = list(armaOrder = c(2,2), include.mean = TRUE), distribution.model = 'norm')

sgx11n <- ugarchfit(spec = sgx11n.spec, SGX)

print(sgx11n)
## 
## *---------------------------------*
## *          GARCH Model Fit        *
## *---------------------------------*
## 
## Conditional Variance Dynamics    
## -----------------------------------
## GARCH Model  : gjrGARCH(1,1)
## Mean Model   : ARFIMA(2,0,2)
## Distribution : norm 
## 
## Optimal Parameters
## ------------------------------------
##         Estimate  Std. Error   t value Pr(>|t|)
## mu      7.457124    0.185189  40.26758 0.000000
## ar1     0.440084    0.011552  38.09691 0.000000
## ar2     0.558340    0.011543  48.36988 0.000000
## ma1    -0.233099    0.027542  -8.46338 0.000000
## ma2    -0.635324    0.026471 -24.00063 0.000000
## omega   0.007404    0.003168   2.33721 0.019428
## alpha1  0.135357    0.032939   4.10928 0.000040
## beta1   0.867843    0.022025  39.40305 0.000000
## gamma1 -0.008400    0.038224  -0.21976 0.826058
## 
## Robust Standard Errors:
##         Estimate  Std. Error   t value Pr(>|t|)
## mu      7.457124    0.069296 107.61301 0.000000
## ar1     0.440084    0.001471 299.14694 0.000000
## ar2     0.558340    0.001552 359.72600 0.000000
## ma1    -0.233099    0.024780  -9.40670 0.000000
## ma2    -0.635324    0.024422 -26.01407 0.000000
## omega   0.007404    0.004052   1.82728 0.067657
## alpha1  0.135357    0.042251   3.20368 0.001357
## beta1   0.867843    0.027378  31.69870 0.000000
## gamma1 -0.008400    0.041963  -0.20018 0.841341
## 
## LogLikelihood : -1257.906 
## 
## Information Criteria
## ------------------------------------
##                    
## Akaike       2.5262
## Bayes        2.5703
## Shibata      2.5261
## Hannan-Quinn 2.5430
## 
## Weighted Ljung-Box Test on Standardized Residuals
## ------------------------------------
##                          statistic   p-value
## Lag[1]                       1.025 3.114e-01
## Lag[2*(p+q)+(p+q)-1][11]    28.135 0.000e+00
## Lag[4*(p+q)+(p+q)-1][19]    37.801 2.749e-13
## d.o.f=4
## H0 : No serial correlation
## 
## Weighted Ljung-Box Test on Standardized Squared Residuals
## ------------------------------------
##                         statistic p-value
## Lag[1]                      3.678 0.05514
## Lag[2*(p+q)+(p+q)-1][5]     9.154 0.01513
## Lag[4*(p+q)+(p+q)-1][9]    12.452 0.01430
## d.o.f=2
## 
## Weighted ARCH LM Tests
## ------------------------------------
##             Statistic Shape Scale P-Value
## ARCH Lag[3]    0.8285 0.500 2.000 0.36270
## ARCH Lag[5]    7.5656 1.440 1.667 0.02552
## ARCH Lag[7]    8.6271 2.315 1.543 0.03798
## 
## Nyblom stability test
## ------------------------------------
## Joint Statistic:  6.769
## Individual Statistics:              
## mu     0.01906
## ar1    0.33764
## ar2    0.31787
## ma1    4.06402
## ma2    4.32469
## omega  0.29187
## alpha1 0.16606
## beta1  0.41067
## gamma1 0.22492
## 
## Asymptotic Critical Values (10% 5% 1%)
## Joint Statistic:          2.1 2.32 2.82
## Individual Statistic:     0.35 0.47 0.75
## 
## Sign Bias Test
## ------------------------------------
##                    t-value      prob sig
## Sign Bias           1.9343 0.0533531   *
## Negative Sign Bias  3.6276 0.0003005 ***
## Positive Sign Bias  0.3483 0.7276795    
## Joint Effect       13.2898 0.0040500 ***
## 
## 
## Adjusted Pearson Goodness-of-Fit Test:
## ------------------------------------
##   group statistic p-value(g-1)
## 1    20     53.89    3.433e-05
## 2    30     69.01    4.132e-05
## 3    40     80.51    1.039e-04
## 4    50     88.08    5.191e-04
## 
## 
## Elapsed time : 0.4990029

PHÂN PHỐI STUDENT

sgx11std.spec <- ugarchspec(variance.model = list(model = 'gjrGARCH', garchOrder = c(1,1)), mean.model = list(armaOrder = c(2,2), include.mean = TRUE), distribution.model = 'std')

sgx11std <- ugarchfit(sgx11std.spec, SGX)
print(sgx11std)
## 
## *---------------------------------*
## *          GARCH Model Fit        *
## *---------------------------------*
## 
## Conditional Variance Dynamics    
## -----------------------------------
## GARCH Model  : gjrGARCH(1,1)
## Mean Model   : ARFIMA(2,0,2)
## Distribution : std 
## 
## Optimal Parameters
## ------------------------------------
##         Estimate  Std. Error   t value Pr(>|t|)
## mu      7.451364    0.183870  40.52509 0.000000
## ar1     0.441812    0.011375  38.83917 0.000000
## ar2     0.556660    0.011367  48.97258 0.000000
## ma1    -0.233484    0.027451  -8.50540 0.000000
## ma2    -0.635709    0.026522 -23.96906 0.000000
## omega   0.007053    0.003330   2.11827 0.034152
## alpha1  0.131562    0.033004   3.98631 0.000067
## beta1   0.870018    0.022605  38.48859 0.000000
## gamma1 -0.005160    0.038297  -0.13475 0.892812
## shape  99.980863  199.691326   0.50068 0.616598
## 
## Robust Standard Errors:
##         Estimate  Std. Error   t value Pr(>|t|)
## mu      7.451364    0.069266 107.57599 0.000000
## ar1     0.441812    0.001439 307.08868 0.000000
## ar2     0.556660    0.001506 369.54145 0.000000
## ma1    -0.233484    0.025120  -9.29480 0.000000
## ma2    -0.635709    0.024717 -25.71906 0.000000
## omega   0.007053    0.003810   1.85107 0.064159
## alpha1  0.131562    0.041622   3.16086 0.001573
## beta1   0.870018    0.027061  32.14969 0.000000
## gamma1 -0.005160    0.042095  -0.12259 0.902432
## shape  99.980863  197.361751   0.50659 0.612445
## 
## LogLikelihood : -1257.873 
## 
## Information Criteria
## ------------------------------------
##                    
## Akaike       2.5282
## Bayes        2.5771
## Shibata      2.5280
## Hannan-Quinn 2.5468
## 
## Weighted Ljung-Box Test on Standardized Residuals
## ------------------------------------
##                          statistic   p-value
## Lag[1]                       1.061 3.031e-01
## Lag[2*(p+q)+(p+q)-1][11]    27.772 0.000e+00
## Lag[4*(p+q)+(p+q)-1][19]    37.468 4.281e-13
## d.o.f=4
## H0 : No serial correlation
## 
## Weighted Ljung-Box Test on Standardized Squared Residuals
## ------------------------------------
##                         statistic p-value
## Lag[1]                      3.434 0.06385
## Lag[2*(p+q)+(p+q)-1][5]     8.922 0.01732
## Lag[4*(p+q)+(p+q)-1][9]    12.239 0.01600
## d.o.f=2
## 
## Weighted ARCH LM Tests
## ------------------------------------
##             Statistic Shape Scale P-Value
## ARCH Lag[3]     0.845 0.500 2.000 0.35796
## ARCH Lag[5]     7.615 1.440 1.667 0.02483
## ARCH Lag[7]     8.681 2.315 1.543 0.03693
## 
## Nyblom stability test
## ------------------------------------
## Joint Statistic:  7.1407
## Individual Statistics:              
## mu     0.01732
## ar1    0.34177
## ar2    0.32256
## ma1    3.98568
## ma2    4.30415
## omega  0.28756
## alpha1 0.15437
## beta1  0.39795
## gamma1 0.21056
## shape  0.86463
## 
## Asymptotic Critical Values (10% 5% 1%)
## Joint Statistic:          2.29 2.54 3.05
## Individual Statistic:     0.35 0.47 0.75
## 
## Sign Bias Test
## ------------------------------------
##                    t-value      prob sig
## Sign Bias           1.9399 0.0526792   *
## Negative Sign Bias  3.6320 0.0002956 ***
## Positive Sign Bias  0.3108 0.7560199    
## Joint Effect       13.2927 0.0040445 ***
## 
## 
## Adjusted Pearson Goodness-of-Fit Test:
## ------------------------------------
##   group statistic p-value(g-1)
## 1    20     52.53    5.503e-05
## 2    30     68.42    4.979e-05
## 3    40     80.03    1.187e-04
## 4    50     84.39    1.254e-03
## 
## 
## Elapsed time : 0.7286642

PHÂN PHỐI ĐỐI XỨNG (sstd)

sgx11sstd.spec <- ugarchspec(variance.model = list( model = 'gjrGARCH', garchOrder = c(1,1)), mean.model = list(armaOrder = c(2,2), include.mean = TRUE), distribution.model = 'sstd')

sgx11sstd <- ugarchfit(sgx11sstd.spec, SGX)

print(sgx11sstd)
## 
## *---------------------------------*
## *          GARCH Model Fit        *
## *---------------------------------*
## 
## Conditional Variance Dynamics    
## -----------------------------------
## GARCH Model  : gjrGARCH(1,1)
## Mean Model   : ARFIMA(2,0,2)
## Distribution : sstd 
## 
## Optimal Parameters
## ------------------------------------
##         Estimate  Std. Error    t value Pr(>|t|)
## mu      7.468130    0.185786  40.197421 0.000000
## ar1     0.440705    0.011187  39.392876 0.000000
## ar2     0.557776    0.011179  49.894418 0.000000
## ma1    -0.232592    0.027415  -8.484002 0.000000
## ma2    -0.635883    0.026589 -23.915420 0.000000
## omega   0.006533    0.003241   2.015661 0.043835
## alpha1  0.126720    0.032643   3.882031 0.000104
## beta1   0.873505    0.022536  38.760304 0.000000
## gamma1 -0.002429    0.037874  -0.064136 0.948862
## skew    1.033225    0.051112  20.215093 0.000000
## shape  59.999698   60.265545   0.995589 0.319450
## 
## Robust Standard Errors:
##         Estimate  Std. Error    t value Pr(>|t|)
## mu      7.468130    0.077612  96.224040 0.000000
## ar1     0.440705    0.001405 313.563723 0.000000
## ar2     0.557776    0.001485 375.640072 0.000000
## ma1    -0.232592    0.025464  -9.134101 0.000000
## ma2    -0.635883    0.025440 -24.995334 0.000000
## omega   0.006533    0.003980   1.641362 0.100722
## alpha1  0.126720    0.042141   3.007052 0.002638
## beta1   0.873505    0.028376  30.783530 0.000000
## gamma1 -0.002429    0.041875  -0.058008 0.953742
## skew    1.033225    0.065283  15.826941 0.000000
## shape  59.999698   50.543138   1.187099 0.235189
## 
## LogLikelihood : -1257.72 
## 
## Information Criteria
## ------------------------------------
##                    
## Akaike       2.5299
## Bayes        2.5837
## Shibata      2.5296
## Hannan-Quinn 2.5503
## 
## Weighted Ljung-Box Test on Standardized Residuals
## ------------------------------------
##                          statistic   p-value
## Lag[1]                       1.002 3.168e-01
## Lag[2*(p+q)+(p+q)-1][11]    27.642 0.000e+00
## Lag[4*(p+q)+(p+q)-1][19]    37.337 5.086e-13
## d.o.f=4
## H0 : No serial correlation
## 
## Weighted Ljung-Box Test on Standardized Squared Residuals
## ------------------------------------
##                         statistic p-value
## Lag[1]                      3.082 0.07916
## Lag[2*(p+q)+(p+q)-1][5]     8.614 0.02070
## Lag[4*(p+q)+(p+q)-1][9]    12.004 0.01811
## d.o.f=2
## 
## Weighted ARCH LM Tests
## ------------------------------------
##             Statistic Shape Scale P-Value
## ARCH Lag[3]     0.952 0.500 2.000 0.32921
## ARCH Lag[5]     7.837 1.440 1.667 0.02197
## ARCH Lag[7]     8.932 2.315 1.543 0.03239
## 
## Nyblom stability test
## ------------------------------------
## Joint Statistic:  8.1908
## Individual Statistics:              
## mu     0.01812
## ar1    0.37272
## ar2    0.35386
## ma1    3.93863
## ma2    4.31667
## omega  0.30684
## alpha1 0.16496
## beta1  0.42673
## gamma1 0.19967
## skew   0.36882
## shape  0.75242
## 
## Asymptotic Critical Values (10% 5% 1%)
## Joint Statistic:          2.49 2.75 3.27
## Individual Statistic:     0.35 0.47 0.75
## 
## Sign Bias Test
## ------------------------------------
##                    t-value      prob sig
## Sign Bias           1.9674 0.0494144  **
## Negative Sign Bias  3.6534 0.0002723 ***
## Positive Sign Bias  0.2614 0.7938676    
## Joint Effect       13.4181 0.0038145 ***
## 
## 
## Adjusted Pearson Goodness-of-Fit Test:
## ------------------------------------
##   group statistic p-value(g-1)
## 1    20     60.51    3.214e-06
## 2    30     67.58    6.454e-05
## 3    40     85.45    2.532e-05
## 4    50     94.96    9.104e-05
## 
## 
## Elapsed time : 1.14546

PHÂN PHỐI Generalized Error Distribution (ged)

sgx11ged.spec <- ugarchspec(variance.model = list(model = "gjrGARCH", garchOrder = c(1, 1)), mean.model = list(armaOrder = c(2, 2), include.mean = TRUE), distribution.model = "ged")

sgx11ged <- ugarchfit(sgx11ged.spec, SGX) 
print(sgx11ged)
## 
## *---------------------------------*
## *          GARCH Model Fit        *
## *---------------------------------*
## 
## Conditional Variance Dynamics    
## -----------------------------------
## GARCH Model  : gjrGARCH(1,1)
## Mean Model   : ARFIMA(2,0,2)
## Distribution : ged 
## 
## Optimal Parameters
## ------------------------------------
##         Estimate  Std. Error  t value Pr(>|t|)
## mu      7.450129    0.185694  40.1205 0.000000
## ar1     0.446190    0.010892  40.9631 0.000000
## ar2     0.552292    0.010883  50.7469 0.000000
## ma1    -0.236650    0.028247  -8.3779 0.000000
## ma2    -0.632485    0.027481 -23.0156 0.000000
## omega   0.007304    0.003257   2.2424 0.024937
## alpha1  0.134245    0.033214   4.0418 0.000053
## beta1   0.868427    0.022360  38.8388 0.000000
## gamma1 -0.007345    0.038575  -0.1904 0.848998
## shape   1.964972    0.155907  12.6035 0.000000
## 
## Robust Standard Errors:
##         Estimate  Std. Error   t value Pr(>|t|)
## mu      7.450129    0.082958  89.80588 0.000000
## ar1     0.446190    0.001428 312.37016 0.000000
## ar2     0.552292    0.001451 380.51540 0.000000
## ma1    -0.236650    0.029336  -8.06693 0.000000
## ma2    -0.632485    0.027882 -22.68462 0.000000
## omega   0.007304    0.003845   1.89970 0.057472
## alpha1  0.134245    0.041649   3.22325 0.001267
## beta1   0.868427    0.026828  32.37016 0.000000
## gamma1 -0.007345    0.041535  -0.17683 0.859642
## shape   1.964972    0.244869   8.02459 0.000000
## 
## LogLikelihood : -1257.882 
## 
## Information Criteria
## ------------------------------------
##                    
## Akaike       2.5282
## Bayes        2.5771
## Shibata      2.5280
## Hannan-Quinn 2.5468
## 
## Weighted Ljung-Box Test on Standardized Residuals
## ------------------------------------
##                          statistic   p-value
## Lag[1]                       1.146 2.845e-01
## Lag[2*(p+q)+(p+q)-1][11]    27.889 0.000e+00
## Lag[4*(p+q)+(p+q)-1][19]    37.552 3.828e-13
## d.o.f=4
## H0 : No serial correlation
## 
## Weighted Ljung-Box Test on Standardized Squared Residuals
## ------------------------------------
##                         statistic p-value
## Lag[1]                      3.631 0.05670
## Lag[2*(p+q)+(p+q)-1][5]     9.101 0.01561
## Lag[4*(p+q)+(p+q)-1][9]    12.390 0.01477
## d.o.f=2
## 
## Weighted ARCH LM Tests
## ------------------------------------
##             Statistic Shape Scale P-Value
## ARCH Lag[3]    0.8196 0.500 2.000 0.36530
## ARCH Lag[5]    7.5451 1.440 1.667 0.02581
## ARCH Lag[7]    8.6056 2.315 1.543 0.03840
## 
## Nyblom stability test
## ------------------------------------
## Joint Statistic:  7.7851
## Individual Statistics:              
## mu     0.01724
## ar1    0.33295
## ar2    0.31433
## ma1    3.98457
## ma2    4.30822
## omega  0.29189
## alpha1 0.16291
## beta1  0.40872
## gamma1 0.22023
## shape  1.61476
## 
## Asymptotic Critical Values (10% 5% 1%)
## Joint Statistic:          2.29 2.54 3.05
## Individual Statistic:     0.35 0.47 0.75
## 
## Sign Bias Test
## ------------------------------------
##                    t-value      prob sig
## Sign Bias           1.9318 0.0536712   *
## Negative Sign Bias  3.6246 0.0003041 ***
## Positive Sign Bias  0.3383 0.7351960    
## Joint Effect       13.2594 0.0041079 ***
## 
## 
## Adjusted Pearson Goodness-of-Fit Test:
## ------------------------------------
##   group statistic p-value(g-1)
## 1    20     52.65    0.0000528
## 2    30     66.80    0.0000820
## 3    40     77.96    0.0002102
## 4    50     84.79    0.0011423
## 
## 
## Elapsed time : 0.6841369

PHÂN PHỐI Generalized Error Distribution đối xứng (sged)

sgx11sged.spec <- ugarchspec(variance.model = list(model = 'gjrGARCH', garchOrder = c(1,1)), mean.model = list(armaOrder = c(2,2), include.mean = TRUE), distribution.model = 'sged')

sgx11sged <- ugarchfit(sgx11sged.spec, SGX)
print(sgx11sged)
## 
## *---------------------------------*
## *          GARCH Model Fit        *
## *---------------------------------*
## 
## Conditional Variance Dynamics    
## -----------------------------------
## GARCH Model  : gjrGARCH(1,1)
## Mean Model   : ARFIMA(2,0,2)
## Distribution : sged 
## 
## Optimal Parameters
## ------------------------------------
##         Estimate  Std. Error  t value Pr(>|t|)
## mu      7.479284    0.193380  38.6766 0.000000
## ar1     0.437060    0.011776  37.1143 0.000000
## ar2     0.561287    0.011762  47.7203 0.000000
## ma1    -0.231529    0.028774  -8.0465 0.000000
## ma2    -0.635473    0.028331 -22.4304 0.000000
## omega   0.007083    0.003174   2.2320 0.025618
## alpha1  0.132688    0.033223   3.9938 0.000065
## beta1   0.870030    0.022094  39.3782 0.000000
## gamma1 -0.007368    0.038615  -0.1908 0.848683
## skew    1.034831    0.054037  19.1503 0.000000
## shape   2.004837    0.169221  11.8475 0.000000
## 
## Robust Standard Errors:
##         Estimate  Std. Error   t value Pr(>|t|)
## mu      7.479284    0.120130  62.25998 0.000000
## ar1     0.437060    0.001663 262.78561 0.000000
## ar2     0.561287    0.001582 354.85029 0.000000
## ma1    -0.231529    0.031763  -7.28931 0.000000
## ma2    -0.635473    0.031407 -20.23329 0.000000
## omega   0.007083    0.003922   1.80597 0.070923
## alpha1  0.132688    0.043095   3.07893 0.002077
## beta1   0.870030    0.028042  31.02632 0.000000
## gamma1 -0.007368    0.042143  -0.17482 0.861218
## skew    1.034831    0.084597  12.23251 0.000000
## shape   2.004837    0.311243   6.44139 0.000000
## 
## LogLikelihood : -1257.665 
## 
## Information Criteria
## ------------------------------------
##                    
## Akaike       2.5297
## Bayes        2.5836
## Shibata      2.5295
## Hannan-Quinn 2.5502
## 
## Weighted Ljung-Box Test on Standardized Residuals
## ------------------------------------
##                          statistic   p-value
## Lag[1]                      0.9288 3.352e-01
## Lag[2*(p+q)+(p+q)-1][11]   28.2920 0.000e+00
## Lag[4*(p+q)+(p+q)-1][19]   37.9374 2.295e-13
## d.o.f=4
## H0 : No serial correlation
## 
## Weighted Ljung-Box Test on Standardized Squared Residuals
## ------------------------------------
##                         statistic p-value
## Lag[1]                      3.465 0.06269
## Lag[2*(p+q)+(p+q)-1][5]     8.982 0.01673
## Lag[4*(p+q)+(p+q)-1][9]    12.339 0.01518
## d.o.f=2
## 
## Weighted ARCH LM Tests
## ------------------------------------
##             Statistic Shape Scale P-Value
## ARCH Lag[3]    0.9332 0.500 2.000 0.33403
## ARCH Lag[5]    7.7615 1.440 1.667 0.02290
## ARCH Lag[7]    8.8515 2.315 1.543 0.03378
## 
## Nyblom stability test
## ------------------------------------
## Joint Statistic:  9.0131
## Individual Statistics:              
## mu     0.02248
## ar1    0.36765
## ar2    0.34753
## ma1    4.10647
## ma2    4.38003
## omega  0.31126
## alpha1 0.18307
## beta1  0.44874
## gamma1 0.22120
## skew   0.31484
## shape  1.65185
## 
## Asymptotic Critical Values (10% 5% 1%)
## Joint Statistic:          2.49 2.75 3.27
## Individual Statistic:     0.35 0.47 0.75
## 
## Sign Bias Test
## ------------------------------------
##                    t-value      prob sig
## Sign Bias            1.882 0.0601941   *
## Negative Sign Bias   3.605 0.0003275 ***
## Positive Sign Bias   0.357 0.7211493    
## Joint Effect        13.126 0.0043715 ***
## 
## 
## Adjusted Pearson Goodness-of-Fit Test:
## ------------------------------------
##   group statistic p-value(g-1)
## 1    20     54.69    2.595e-05
## 2    30     68.06    5.566e-05
## 3    40     78.28    1.926e-04
## 4    50     92.96    1.526e-04
## 
## 
## Elapsed time : 1.209277

GJR-GARCH(12)

PHÂN PHỐI CHUẨN

sgx12n.spec <- ugarchspec(variance.model = list(model = 'gjrGARCH', garchOrder = c(1,2)), mean.model = list(armaOrder = c(2,2), include.mean = TRUE), distribution.model = 'norm')

sgx12n <- ugarchfit(sgx12n.spec, SGX)
print(sgx12n)
## 
## *---------------------------------*
## *          GARCH Model Fit        *
## *---------------------------------*
## 
## Conditional Variance Dynamics    
## -----------------------------------
## GARCH Model  : gjrGARCH(1,2)
## Mean Model   : ARFIMA(2,0,2)
## Distribution : norm 
## 
## Optimal Parameters
## ------------------------------------
##         Estimate  Std. Error   t value Pr(>|t|)
## mu      7.457126    0.185189  40.26770 0.000000
## ar1     0.440086    0.011549  38.10471 0.000000
## ar2     0.558338    0.011543  48.37124 0.000000
## ma1    -0.233100    0.027526  -8.46830 0.000000
## ma2    -0.635323    0.026420 -24.04740 0.000000
## omega   0.007404    0.002752   2.69031 0.007139
## alpha1  0.135356    0.017660   7.66441 0.000000
## beta1   0.867843    0.451302   1.92297 0.054483
## beta2   0.000000    0.415438   0.00000 1.000000
## gamma1 -0.008398    0.038206  -0.21982 0.826013
## 
## Robust Standard Errors:
##         Estimate  Std. Error   t value Pr(>|t|)
## mu      7.457126    0.069181 107.79077 0.000000
## ar1     0.440086    0.001432 307.36865 0.000000
## ar2     0.558338    0.001562 357.42714 0.000000
## ma1    -0.233100    0.024866  -9.37414 0.000000
## ma2    -0.635323    0.024428 -26.00770 0.000000
## omega   0.007404    0.005129   1.44345 0.148892
## alpha1  0.135356    0.075426   1.79457 0.072722
## beta1   0.867843    0.622255   1.39467 0.163114
## beta2   0.000000    0.558997   0.00000 1.000000
## gamma1 -0.008398    0.041893  -0.20047 0.841114
## 
## LogLikelihood : -1257.906 
## 
## Information Criteria
## ------------------------------------
##                    
## Akaike       2.5282
## Bayes        2.5772
## Shibata      2.5280
## Hannan-Quinn 2.5468
## 
## Weighted Ljung-Box Test on Standardized Residuals
## ------------------------------------
##                          statistic   p-value
## Lag[1]                       1.025 3.114e-01
## Lag[2*(p+q)+(p+q)-1][11]    28.135 0.000e+00
## Lag[4*(p+q)+(p+q)-1][19]    37.801 2.749e-13
## d.o.f=4
## H0 : No serial correlation
## 
## Weighted Ljung-Box Test on Standardized Squared Residuals
## ------------------------------------
##                          statistic p-value
## Lag[1]                       3.678 0.05514
## Lag[2*(p+q)+(p+q)-1][8]     11.826 0.01316
## Lag[4*(p+q)+(p+q)-1][14]    15.276 0.02199
## d.o.f=3
## 
## Weighted ARCH LM Tests
## ------------------------------------
##             Statistic Shape Scale  P-Value
## ARCH Lag[4]     8.421 0.500 2.000 0.003709
## ARCH Lag[6]     8.690 1.461 1.711 0.016117
## ARCH Lag[8]     9.761 2.368 1.583 0.025679
## 
## Nyblom stability test
## ------------------------------------
## Joint Statistic:  8.4359
## Individual Statistics:              
## mu     0.01906
## ar1    0.33764
## ar2    0.31788
## ma1    4.06402
## ma2    4.32470
## omega  0.29187
## alpha1 0.16606
## beta1  0.41067
## beta2  0.44070
## gamma1 0.22492
## 
## Asymptotic Critical Values (10% 5% 1%)
## Joint Statistic:          2.29 2.54 3.05
## Individual Statistic:     0.35 0.47 0.75
## 
## Sign Bias Test
## ------------------------------------
##                    t-value      prob sig
## Sign Bias           1.9343 0.0533538   *
## Negative Sign Bias  3.6276 0.0003005 ***
## Positive Sign Bias  0.3483 0.7276850    
## Joint Effect       13.2898 0.0040499 ***
## 
## 
## Adjusted Pearson Goodness-of-Fit Test:
## ------------------------------------
##   group statistic p-value(g-1)
## 1    20     53.89    3.433e-05
## 2    30     69.01    4.132e-05
## 3    40     80.51    1.039e-04
## 4    50     88.08    5.191e-04
## 
## 
## Elapsed time : 0.337148

PHÂN PHỐI STUDENT

sgx12std.spec <- ugarchspec(variance.model = list(model = 'gjrGARCH', garchOrder = c(1,2)), mean.model = list(armaOrder = c(2,2), include.mean = TRUE), distribution.model = 'std')

sgx12std <- ugarchfit(sgx12std.spec, SGX)
print(sgx12std)
## 
## *---------------------------------*
## *          GARCH Model Fit        *
## *---------------------------------*
## 
## Conditional Variance Dynamics    
## -----------------------------------
## GARCH Model  : gjrGARCH(1,2)
## Mean Model   : ARFIMA(2,0,2)
## Distribution : std 
## 
## Optimal Parameters
## ------------------------------------
##         Estimate  Std. Error   t value Pr(>|t|)
## mu      7.450841    0.183708  40.55808 0.000000
## ar1     0.442159    0.011341  38.98774 0.000000
## ar2     0.556320    0.011335  49.07792 0.000000
## ma1    -0.233698    0.027434  -8.51857 0.000000
## ma2    -0.635576    0.026376 -24.09671 0.000000
## omega   0.007026    0.003057   2.29803 0.021560
## alpha1  0.131292    0.017294   7.59195 0.000000
## beta1   0.870203    0.475554   1.82987 0.067269
## beta2   0.000000    0.439430   0.00000 1.000000
## gamma1 -0.004989    0.038227  -0.13052 0.896155
## shape  91.818382  155.705485   0.58969 0.555397
## 
## Robust Standard Errors:
##         Estimate  Std. Error   t value Pr(>|t|)
## mu      7.450841    0.069258 107.58126 0.000000
## ar1     0.442159    0.001397 316.55632 0.000000
## ar2     0.556320    0.001507 369.13605 0.000000
## ma1    -0.233698    0.025134  -9.29797 0.000000
## ma2    -0.635576    0.024847 -25.57936 0.000000
## omega   0.007026    0.004442   1.58176 0.113704
## alpha1  0.131292    0.071911   1.82575 0.067888
## beta1   0.870203    0.627495   1.38679 0.165506
## beta2   0.000000    0.567426   0.00000 1.000000
## gamma1 -0.004989    0.041990  -0.11882 0.905415
## shape  91.818382  175.322937   0.52371 0.600480
## 
## LogLikelihood : -1257.878 
## 
## Information Criteria
## ------------------------------------
##                    
## Akaike       2.5302
## Bayes        2.5840
## Shibata      2.5299
## Hannan-Quinn 2.5506
## 
## Weighted Ljung-Box Test on Standardized Residuals
## ------------------------------------
##                          statistic   p-value
## Lag[1]                       1.065 3.021e-01
## Lag[2*(p+q)+(p+q)-1][11]    27.743 0.000e+00
## Lag[4*(p+q)+(p+q)-1][19]    37.440 4.439e-13
## d.o.f=4
## H0 : No serial correlation
## 
## Weighted Ljung-Box Test on Standardized Squared Residuals
## ------------------------------------
##                          statistic p-value
## Lag[1]                       3.416 0.06458
## Lag[2*(p+q)+(p+q)-1][8]     11.595 0.01492
## Lag[4*(p+q)+(p+q)-1][14]    15.072 0.02417
## d.o.f=3
## 
## Weighted ARCH LM Tests
## ------------------------------------
##             Statistic Shape Scale  P-Value
## ARCH Lag[4]     8.470 0.500 2.000 0.003611
## ARCH Lag[6]     8.747 1.461 1.711 0.015626
## ARCH Lag[8]     9.829 2.368 1.583 0.024798
## 
## Nyblom stability test
## ------------------------------------
## Joint Statistic:  8.8917
## Individual Statistics:              
## mu     0.01712
## ar1    0.34145
## ar2    0.32229
## ma1    3.97997
## ma2    4.30068
## omega  0.28759
## alpha1 0.15384
## beta1  0.39748
## beta2  0.42556
## gamma1 0.20944
## shape  0.85952
## 
## Asymptotic Critical Values (10% 5% 1%)
## Joint Statistic:          2.49 2.75 3.27
## Individual Statistic:     0.35 0.47 0.75
## 
## Sign Bias Test
## ------------------------------------
##                    t-value      prob sig
## Sign Bias           1.9404 0.0526111   *
## Negative Sign Bias  3.6318 0.0002958 ***
## Positive Sign Bias  0.3079 0.7582233    
## Joint Effect       13.2894 0.0040509 ***
## 
## 
## Adjusted Pearson Goodness-of-Fit Test:
## ------------------------------------
##   group statistic p-value(g-1)
## 1    20     52.57    5.428e-05
## 2    30     67.46    6.697e-05
## 3    40     79.79    1.269e-04
## 4    50     88.38    4.825e-04
## 
## 
## Elapsed time : 1.182771

PHÂN PHỐI ĐỐI XỨNG (sstd)

sgx12sstd.spec <- ugarchspec(variance.model = list( model = 'gjrGARCH', garchOrder = c(1,2)), mean.model = list(armaOrder = c(2,2), include.mean = TRUE), distribution.model = 'sstd')

sgx12sstd <- ugarchfit(sgx12sstd.spec, SGX)

print(sgx12sstd)
## 
## *---------------------------------*
## *          GARCH Model Fit        *
## *---------------------------------*
## 
## Conditional Variance Dynamics    
## -----------------------------------
## GARCH Model  : gjrGARCH(1,2)
## Mean Model   : ARFIMA(2,0,2)
## Distribution : sstd 
## 
## Optimal Parameters
## ------------------------------------
##         Estimate  Std. Error    t value Pr(>|t|)
## mu      7.468148    0.185662  40.224394 0.000000
## ar1     0.440718    0.011182  39.412714 0.000000
## ar2     0.557763    0.011178  49.899587 0.000000
## ma1    -0.232602    0.027408  -8.486502 0.000000
## ma2    -0.635873    0.026501 -23.994456 0.000000
## omega   0.006533    0.002854   2.289132 0.022072
## alpha1  0.126724    0.028506   4.445458 0.000009
## beta1   0.873504    0.556779   1.568851 0.116683
## beta2   0.000000    0.514638   0.000001 1.000000
## gamma1 -0.002433    0.037859  -0.064269 0.948756
## skew    1.033223    0.050695  20.381176 0.000000
## shape  59.999370   58.206986   1.030793 0.302638
## 
## Robust Standard Errors:
##         Estimate  Std. Error    t value Pr(>|t|)
## mu      7.468148    0.074446 100.316088  0.00000
## ar1     0.440718    0.001391 316.915564  0.00000
## ar2     0.557763    0.001509 369.635994  0.00000
## ma1    -0.232602    0.025517  -9.115399  0.00000
## ma2    -0.635873    0.025672 -24.769546  0.00000
## omega   0.006533    0.005075   1.287210  0.19802
## alpha1  0.126724    0.088621   1.429945  0.15273
## beta1   0.873504    0.866157   1.008482  0.31322
## beta2   0.000000    0.787140   0.000000  1.00000
## gamma1 -0.002433    0.041726  -0.058312  0.95350
## skew    1.033223    0.063949  16.157099  0.00000
## shape  59.999370   60.775556   0.987229  0.32353
## 
## LogLikelihood : -1257.72 
## 
## Information Criteria
## ------------------------------------
##                    
## Akaike       2.5318
## Bayes        2.5906
## Shibata      2.5316
## Hannan-Quinn 2.5542
## 
## Weighted Ljung-Box Test on Standardized Residuals
## ------------------------------------
##                          statistic   p-value
## Lag[1]                       1.002 3.168e-01
## Lag[2*(p+q)+(p+q)-1][11]    27.642 0.000e+00
## Lag[4*(p+q)+(p+q)-1][19]    37.337 5.086e-13
## d.o.f=4
## H0 : No serial correlation
## 
## Weighted Ljung-Box Test on Standardized Squared Residuals
## ------------------------------------
##                          statistic p-value
## Lag[1]                       3.082 0.07914
## Lag[2*(p+q)+(p+q)-1][8]     11.362 0.01692
## Lag[4*(p+q)+(p+q)-1][14]    14.895 0.02624
## d.o.f=3
## 
## Weighted ARCH LM Tests
## ------------------------------------
##             Statistic Shape Scale  P-Value
## ARCH Lag[4]     8.606 0.500 2.000 0.003351
## ARCH Lag[6]     8.903 1.461 1.711 0.014364
## ARCH Lag[8]    10.009 2.368 1.583 0.022615
## 
## Nyblom stability test
## ------------------------------------
## Joint Statistic:  9.9764
## Individual Statistics:              
## mu     0.01812
## ar1    0.37274
## ar2    0.35388
## ma1    3.93859
## ma2    4.31659
## omega  0.30685
## alpha1 0.16498
## beta1  0.42674
## beta2  0.45606
## gamma1 0.19968
## skew   0.36883
## shape  0.75242
## 
## Asymptotic Critical Values (10% 5% 1%)
## Joint Statistic:          2.69 2.96 3.51
## Individual Statistic:     0.35 0.47 0.75
## 
## Sign Bias Test
## ------------------------------------
##                    t-value      prob sig
## Sign Bias           1.9674 0.0494122  **
## Negative Sign Bias  3.6534 0.0002722 ***
## Positive Sign Bias  0.2614 0.7938308    
## Joint Effect       13.4182 0.0038142 ***
## 
## 
## Adjusted Pearson Goodness-of-Fit Test:
## ------------------------------------
##   group statistic p-value(g-1)
## 1    20     60.51    3.214e-06
## 2    30     67.58    6.454e-05
## 3    40     85.45    2.532e-05
## 4    50     94.96    9.104e-05
## 
## 
## Elapsed time : 1.163599

PHÂN PHỐI Generalized Error Distribution (ged)

sgx12ged.spec <- ugarchspec(variance.model = list(model = "gjrGARCH", garchOrder = c(1,2)), mean.model = list(armaOrder = c(2,2), include.mean = TRUE), distribution.model = "ged")

sgx12ged <- ugarchfit(sgx12ged.spec, SGX) 
print(sgx12ged)
## 
## *---------------------------------*
## *          GARCH Model Fit        *
## *---------------------------------*
## 
## Conditional Variance Dynamics    
## -----------------------------------
## GARCH Model  : gjrGARCH(1,2)
## Mean Model   : ARFIMA(2,0,2)
## Distribution : ged 
## 
## Optimal Parameters
## ------------------------------------
##         Estimate  Std. Error   t value Pr(>|t|)
## mu      7.450117    0.185474  40.16804 0.000000
## ar1     0.446186    0.010889  40.97465 0.000000
## ar2     0.552295    0.010883  50.74865 0.000000
## ma1    -0.236647    0.028208  -8.38942 0.000000
## ma2    -0.632488    0.027159 -23.28850 0.000000
## omega   0.007304    0.002986   2.44622 0.014436
## alpha1  0.134245    0.011730  11.44448 0.000000
## beta1   0.868427    0.441689   1.96615 0.049281
## beta2   0.000000    0.407592   0.00000 1.000000
## gamma1 -0.007344    0.038574  -0.19039 0.849003
## shape   1.964968    0.147686  13.30504 0.000000
## 
## Robust Standard Errors:
##         Estimate  Std. Error   t value Pr(>|t|)
## mu      7.450117    0.085149  87.49466  0.00000
## ar1     0.446186    0.001376 324.21688  0.00000
## ar2     0.552295    0.001452 380.29404  0.00000
## ma1    -0.236647    0.029122  -8.12610  0.00000
## ma2    -0.632488    0.027889 -22.67881  0.00000
## omega   0.007304    0.004508   1.62038  0.10515
## alpha1  0.134245    0.068476   1.96047  0.04994
## beta1   0.868427    0.562975   1.54257  0.12294
## beta2   0.000000    0.507291   0.00000  1.00000
## gamma1 -0.007344    0.041582  -0.17662  0.85981
## shape   1.964968    0.247040   7.95406  0.00000
## 
## LogLikelihood : -1257.882 
## 
## Information Criteria
## ------------------------------------
##                    
## Akaike       2.5302
## Bayes        2.5840
## Shibata      2.5299
## Hannan-Quinn 2.5506
## 
## Weighted Ljung-Box Test on Standardized Residuals
## ------------------------------------
##                          statistic   p-value
## Lag[1]                       1.146 2.845e-01
## Lag[2*(p+q)+(p+q)-1][11]    27.889 0.000e+00
## Lag[4*(p+q)+(p+q)-1][19]    37.552 3.828e-13
## d.o.f=4
## H0 : No serial correlation
## 
## Weighted Ljung-Box Test on Standardized Squared Residuals
## ------------------------------------
##                          statistic p-value
## Lag[1]                       3.631 0.05670
## Lag[2*(p+q)+(p+q)-1][8]     11.766 0.01360
## Lag[4*(p+q)+(p+q)-1][14]    15.211 0.02267
## d.o.f=3
## 
## Weighted ARCH LM Tests
## ------------------------------------
##             Statistic Shape Scale  P-Value
## ARCH Lag[4]     8.407 0.500 2.000 0.003739
## ARCH Lag[6]     8.679 1.461 1.711 0.016208
## ARCH Lag[8]     9.740 2.368 1.583 0.025960
## 
## Nyblom stability test
## ------------------------------------
## Joint Statistic:  9.4699
## Individual Statistics:              
## mu     0.01724
## ar1    0.33296
## ar2    0.31434
## ma1    3.98450
## ma2    4.30820
## omega  0.29188
## alpha1 0.16291
## beta1  0.40871
## beta2  0.43848
## gamma1 0.22023
## shape  1.61476
## 
## Asymptotic Critical Values (10% 5% 1%)
## Joint Statistic:          2.49 2.75 3.27
## Individual Statistic:     0.35 0.47 0.75
## 
## Sign Bias Test
## ------------------------------------
##                    t-value      prob sig
## Sign Bias           1.9318 0.0536712   *
## Negative Sign Bias  3.6246 0.0003041 ***
## Positive Sign Bias  0.3383 0.7351983    
## Joint Effect       13.2594 0.0041079 ***
## 
## 
## Adjusted Pearson Goodness-of-Fit Test:
## ------------------------------------
##   group statistic p-value(g-1)
## 1    20     52.65    0.0000528
## 2    30     66.80    0.0000820
## 3    40     77.96    0.0002102
## 4    50     84.79    0.0011423
## 
## 
## Elapsed time : 0.6698198

PHÂN PHỐI Generalized Error Distribution đối xứng (sged)

sgx12sged.spec <- ugarchspec(variance.model = list(model = 'gjrGARCH', garchOrder = c(1,2)), mean.model = list(armaOrder = c(2,2), include.mean = TRUE), distribution.model = 'sged')

sgx12sged <- ugarchfit(sgx12sged.spec, SGX)
print(sgx12sged)
## 
## *---------------------------------*
## *          GARCH Model Fit        *
## *---------------------------------*
## 
## Conditional Variance Dynamics    
## -----------------------------------
## GARCH Model  : gjrGARCH(1,2)
## Mean Model   : ARFIMA(2,0,2)
## Distribution : sged 
## 
## Optimal Parameters
## ------------------------------------
##         Estimate  Std. Error   t value Pr(>|t|)
## mu      7.479288    0.193147  38.72325 0.000000
## ar1     0.437065    0.011771  37.13150 0.000000
## ar2     0.561283    0.011762  47.72125 0.000000
## ma1    -0.231532    0.028736  -8.05709 0.000000
## ma2    -0.635469    0.027987 -22.70605 0.000000
## omega   0.007083    0.002885   2.45550 0.014069
## alpha1  0.132688    0.014217   9.33334 0.000000
## beta1   0.870030    0.452961   1.92076 0.054762
## beta2   0.000000    0.418031   0.00000 1.000000
## gamma1 -0.007368    0.038614  -0.19081 0.848678
## skew    1.034831    0.054026  19.15441 0.000000
## shape   2.004836    0.159883  12.53943 0.000000
## 
## Robust Standard Errors:
##         Estimate  Std. Error   t value Pr(>|t|)
## mu      7.479288    0.122805  60.90353  0.00000
## ar1     0.437065    0.001604 272.47654  0.00000
## ar2     0.561283    0.001585 354.22740  0.00000
## ma1    -0.231532    0.031594  -7.32844  0.00000
## ma2    -0.635469    0.031619 -20.09782  0.00000
## omega   0.007083    0.004650   1.52338  0.12766
## alpha1  0.132688    0.072499   1.83020  0.06722
## beta1   0.870030    0.599006   1.45246  0.14638
## beta2   0.000000    0.539804   0.00000  1.00000
## gamma1 -0.007368    0.042169  -0.17472  0.86130
## skew    1.034831    0.084785  12.20533  0.00000
## shape   2.004836    0.316736   6.32968  0.00000
## 
## LogLikelihood : -1257.665 
## 
## Information Criteria
## ------------------------------------
##                    
## Akaike       2.5317
## Bayes        2.5905
## Shibata      2.5315
## Hannan-Quinn 2.5541
## 
## Weighted Ljung-Box Test on Standardized Residuals
## ------------------------------------
##                          statistic   p-value
## Lag[1]                      0.9288 3.352e-01
## Lag[2*(p+q)+(p+q)-1][11]   28.2920 0.000e+00
## Lag[4*(p+q)+(p+q)-1][19]   37.9373 2.295e-13
## d.o.f=4
## H0 : No serial correlation
## 
## Weighted Ljung-Box Test on Standardized Squared Residuals
## ------------------------------------
##                          statistic p-value
## Lag[1]                       3.465 0.06269
## Lag[2*(p+q)+(p+q)-1][8]     11.704 0.01407
## Lag[4*(p+q)+(p+q)-1][14]    15.195 0.02283
## d.o.f=3
## 
## Weighted ARCH LM Tests
## ------------------------------------
##             Statistic Shape Scale  P-Value
## ARCH Lag[4]     8.535 0.500 2.000 0.003484
## ARCH Lag[6]     8.820 1.461 1.711 0.015025
## ARCH Lag[8]     9.909 2.368 1.583 0.023800
## 
## Nyblom stability test
## ------------------------------------
## Joint Statistic:  10.7265
## Individual Statistics:              
## mu     0.02248
## ar1    0.36765
## ar2    0.34753
## ma1    4.10648
## ma2    4.38002
## omega  0.31126
## alpha1 0.18307
## beta1  0.44874
## beta2  0.48159
## gamma1 0.22120
## skew   0.31484
## shape  1.65185
## 
## Asymptotic Critical Values (10% 5% 1%)
## Joint Statistic:          2.69 2.96 3.51
## Individual Statistic:     0.35 0.47 0.75
## 
## Sign Bias Test
## ------------------------------------
##                    t-value      prob sig
## Sign Bias            1.882 0.0601937   *
## Negative Sign Bias   3.605 0.0003275 ***
## Positive Sign Bias   0.357 0.7211474    
## Joint Effect        13.126 0.0043714 ***
## 
## 
## Adjusted Pearson Goodness-of-Fit Test:
## ------------------------------------
##   group statistic p-value(g-1)
## 1    20     54.69    2.595e-05
## 2    30     68.06    5.566e-05
## 3    40     78.28    1.926e-04
## 4    50     92.96    1.526e-04
## 
## 
## Elapsed time : 1.526122

GJR-GARCH(21)

PHÂN PHỐI CHUẨN

sgx21n.spec <- ugarchspec(variance.model = list(model = 'gjrGARCH', garchOrder = c(2,1)), mean.model = list(armaOrder = c(2,2), include.mean = TRUE), distribution.model = 'norm')

sgx21n <- ugarchfit(sgx21n.spec, SGX)
print(sgx21n)
## 
## *---------------------------------*
## *          GARCH Model Fit        *
## *---------------------------------*
## 
## Conditional Variance Dynamics    
## -----------------------------------
## GARCH Model  : gjrGARCH(2,1)
## Mean Model   : ARFIMA(2,0,2)
## Distribution : norm 
## 
## Optimal Parameters
## ------------------------------------
##         Estimate  Std. Error     t value Pr(>|t|)
## mu      7.409277    0.135438  5.4706e+01  0.00000
## ar1     0.457257    0.010445  4.3776e+01  0.00000
## ar2     0.540432    0.004878  1.1080e+02  0.00000
## ma1    -0.242441    0.040632 -5.9668e+00  0.00000
## ma2    -0.623333    0.078023 -7.9890e+00  0.00000
## omega   0.007952    0.000298  2.6687e+01  0.00000
## alpha1  0.090683    0.002847  3.1856e+01  0.00000
## alpha2  0.000000    0.011334  1.7000e-05  0.99999
## beta1   0.882426    0.000050  1.7488e+04  0.00000
## gamma1 -0.209070    0.018021 -1.1601e+01  0.00000
## gamma2  0.259791    0.021214  1.2246e+01  0.00000
## 
## Robust Standard Errors:
##         Estimate  Std. Error     t value Pr(>|t|)
## mu      7.409277    0.482333   15.361336 0.000000
## ar1     0.457257    0.043226   10.578202 0.000000
## ar2     0.540432    0.023382   23.113535 0.000000
## ma1    -0.242441    0.180528   -1.342955 0.179286
## ma2    -0.623333    0.314488   -1.982055 0.047473
## omega   0.007952    0.003611    2.202075 0.027660
## alpha1  0.090683    0.033331    2.720655 0.006515
## alpha2  0.000000    0.038743    0.000005 0.999996
## beta1   0.882426    0.000117 7511.282693 0.000000
## gamma1 -0.209070    0.020212  -10.343891 0.000000
## gamma2  0.259791    0.036047    7.207023 0.000000
## 
## LogLikelihood : -1249.185 
## 
## Information Criteria
## ------------------------------------
##                    
## Akaike       2.5128
## Bayes        2.5667
## Shibata      2.5126
## Hannan-Quinn 2.5333
## 
## Weighted Ljung-Box Test on Standardized Residuals
## ------------------------------------
##                          statistic   p-value
## Lag[1]                       0.904 3.417e-01
## Lag[2*(p+q)+(p+q)-1][11]    24.385 0.000e+00
## Lag[4*(p+q)+(p+q)-1][19]    33.963 4.131e-11
## d.o.f=4
## H0 : No serial correlation
## 
## Weighted Ljung-Box Test on Standardized Squared Residuals
## ------------------------------------
##                          statistic p-value
## Lag[1]                       1.559 0.21184
## Lag[2*(p+q)+(p+q)-1][8]     10.729 0.02372
## Lag[4*(p+q)+(p+q)-1][14]    15.664 0.01834
## d.o.f=3
## 
## Weighted ARCH LM Tests
## ------------------------------------
##             Statistic Shape Scale  P-Value
## ARCH Lag[4]     8.739 0.500 2.000 0.003115
## ARCH Lag[6]     9.248 1.461 1.711 0.011920
## ARCH Lag[8]    12.296 2.368 1.583 0.006816
## 
## Nyblom stability test
## ------------------------------------
## Joint Statistic:  6.8514
## Individual Statistics:              
## mu     0.03081
## ar1    0.29252
## ar2    0.25877
## ma1    3.46303
## ma2    3.63152
## omega  0.28628
## alpha1 0.06594
## alpha2 0.16258
## beta1  0.35314
## gamma1 0.12787
## gamma2 0.23233
## 
## Asymptotic Critical Values (10% 5% 1%)
## Joint Statistic:          2.49 2.75 3.27
## Individual Statistic:     0.35 0.47 0.75
## 
## Sign Bias Test
## ------------------------------------
##                    t-value   prob sig
## Sign Bias           1.6171 0.1062    
## Negative Sign Bias  0.7976 0.4253    
## Positive Sign Bias  0.2712 0.7863    
## Joint Effect        3.2480 0.3549    
## 
## 
## Adjusted Pearson Goodness-of-Fit Test:
## ------------------------------------
##   group statistic p-value(g-1)
## 1    20     93.65    7.434e-12
## 2    30     85.11    2.023e-07
## 3    40    112.89    4.125e-09
## 4    50    124.27    1.813e-08
## 
## 
## Elapsed time : 0.5286181

PHÂN PHỐI STUDENT

sgx21std.spec <- ugarchspec(variance.model = list(model = 'gjrGARCH', garchOrder = c(2,1)), mean.model = list(armaOrder = c(2,2), include.mean = TRUE), distribution.model = 'std')

sgx21std <- ugarchfit(sgx21std.spec, SGX)
print(sgx21std)
## 
## *---------------------------------*
## *          GARCH Model Fit        *
## *---------------------------------*
## 
## Conditional Variance Dynamics    
## -----------------------------------
## GARCH Model  : gjrGARCH(2,1)
## Mean Model   : ARFIMA(2,0,2)
## Distribution : std 
## 
## Optimal Parameters
## ------------------------------------
##         Estimate  Std. Error   t value Pr(>|t|)
## mu      7.421743    0.172987  42.90358 0.000000
## ar1     0.450276    0.007514  59.92324 0.000000
## ar2     0.548180    0.007505  73.03737 0.000000
## ma1    -0.238192    0.025547  -9.32382 0.000000
## ma2    -0.630331    0.025043 -25.17004 0.000000
## omega   0.009064    0.004132   2.19343 0.028276
## alpha1  0.122662    0.061718   1.98747 0.046871
## alpha2  0.011505    0.080309   0.14325 0.886089
## beta1   0.848661    0.030611  27.72379 0.000000
## gamma1 -0.118465    0.063155  -1.87577 0.060686
## gamma2  0.150808    0.082101   1.83686 0.066231
## shape  98.192315  161.053787   0.60969 0.542070
## 
## Robust Standard Errors:
##         Estimate  Std. Error   t value Pr(>|t|)
## mu      7.421743    0.066276 111.98172 0.000000
## ar1     0.450276    0.001154 390.21262 0.000000
## ar2     0.548180    0.001229 445.98629 0.000000
## ma1    -0.238192    0.025244  -9.43540 0.000000
## ma2    -0.630331    0.024963 -25.25109 0.000000
## omega   0.009064    0.005129   1.76728 0.077182
## alpha1  0.122662    0.067482   1.81771 0.069109
## alpha2  0.011505    0.101705   0.11312 0.909937
## beta1   0.848661    0.041615  20.39315 0.000000
## gamma1 -0.118465    0.076419  -1.55020 0.121094
## gamma2  0.150808    0.096897   1.55637 0.119620
## shape  98.192315  121.770693   0.80637 0.420029
## 
## LogLikelihood : -1251.617 
## 
## Information Criteria
## ------------------------------------
##                    
## Akaike       2.5197
## Bayes        2.5784
## Shibata      2.5194
## Hannan-Quinn 2.5420
## 
## Weighted Ljung-Box Test on Standardized Residuals
## ------------------------------------
##                          statistic   p-value
## Lag[1]                      0.8815 3.478e-01
## Lag[2*(p+q)+(p+q)-1][11]   25.4133 0.000e+00
## Lag[4*(p+q)+(p+q)-1][19]   35.3309 7.069e-12
## d.o.f=4
## H0 : No serial correlation
## 
## Weighted Ljung-Box Test on Standardized Squared Residuals
## ------------------------------------
##                          statistic p-value
## Lag[1]                    0.006643  0.9350
## Lag[2*(p+q)+(p+q)-1][8]   7.810995  0.1036
## Lag[4*(p+q)+(p+q)-1][14] 11.280854  0.1232
## d.o.f=3
## 
## Weighted ARCH LM Tests
## ------------------------------------
##             Statistic Shape Scale  P-Value
## ARCH Lag[4]     6.971 0.500 2.000 0.008285
## ARCH Lag[6]     7.028 1.461 1.711 0.039247
## ARCH Lag[8]     8.561 2.368 1.583 0.047067
## 
## Nyblom stability test
## ------------------------------------
## Joint Statistic:  7.943
## Individual Statistics:              
## mu     0.01213
## ar1    0.33264
## ar2    0.30686
## ma1    3.59223
## ma2    4.10058
## omega  0.30158
## alpha1 0.10699
## alpha2 0.22498
## beta1  0.34575
## gamma1 0.14047
## gamma2 0.27087
## shape  0.98083
## 
## Asymptotic Critical Values (10% 5% 1%)
## Joint Statistic:          2.69 2.96 3.51
## Individual Statistic:     0.35 0.47 0.75
## 
## Sign Bias Test
## ------------------------------------
##                    t-value    prob sig
## Sign Bias           1.6643 0.09637   *
## Negative Sign Bias  2.2701 0.02341  **
## Positive Sign Bias  0.2623 0.79316    
## Joint Effect        5.8585 0.11870    
## 
## 
## Adjusted Pearson Goodness-of-Fit Test:
## ------------------------------------
##   group statistic p-value(g-1)
## 1    20     77.70    4.621e-09
## 2    30     75.35    5.424e-06
## 3    40     90.08    6.421e-06
## 4    50    101.54    1.551e-05
## 
## 
## Elapsed time : 1.217238

PHÂN PHỐI ĐỐI XỨNG (sstd)

sgx21sstd.spec <- ugarchspec(variance.model = list( model = 'gjrGARCH', garchOrder = c(2,1)), mean.model = list(armaOrder = c(2,2), include.mean = TRUE), distribution.model = 'sstd')

sgx21sstd <- ugarchfit(sgx21sstd.spec, SGX)

print(sgx21sstd)
## 
## *---------------------------------*
## *          GARCH Model Fit        *
## *---------------------------------*
## 
## Conditional Variance Dynamics    
## -----------------------------------
## GARCH Model  : gjrGARCH(2,1)
## Mean Model   : ARFIMA(2,0,2)
## Distribution : sstd 
## 
## Optimal Parameters
## ------------------------------------
##         Estimate  Std. Error    t value Pr(>|t|)
## mu      7.436206    0.172865  43.017502 0.000000
## ar1     0.452003    0.007526  60.057545 0.000000
## ar2     0.546553    0.007518  72.699668 0.000000
## ma1    -0.238992    0.025529  -9.361680 0.000000
## ma2    -0.629874    0.025127 -25.067597 0.000000
## omega   0.008272    0.004157   1.989893 0.046603
## alpha1  0.124212    0.062090   2.000513 0.045445
## alpha2  0.000002    0.082810   0.000019 0.999985
## beta1   0.855260    0.032089  26.652973 0.000000
## gamma1 -0.119424    0.063356  -1.884959 0.059435
## gamma2  0.158041    0.082440   1.917037 0.055233
## skew    1.040537    0.054180  19.205146 0.000000
## shape  59.999711   56.759277   1.057091 0.290470
## 
## Robust Standard Errors:
##         Estimate  Std. Error    t value Pr(>|t|)
## mu      7.436206    0.066427 111.945561 0.000000
## ar1     0.452003    0.001231 367.095558 0.000000
## ar2     0.546553    0.001303 419.617995 0.000000
## ma1    -0.238992    0.025355  -9.425941 0.000000
## ma2    -0.629874    0.025372 -24.825323 0.000000
## omega   0.008272    0.006138   1.347548 0.177804
## alpha1  0.124212    0.072300   1.718016 0.085794
## alpha2  0.000002    0.122571   0.000013 0.999990
## beta1   0.855260    0.051605  16.573065 0.000000
## gamma1 -0.119424    0.081370  -1.467660 0.142197
## gamma2  0.158041    0.112496   1.404865 0.160062
## skew    1.040537    0.081943  12.698361 0.000000
## shape  59.999711   41.120522   1.459118 0.144533
## 
## LogLikelihood : -1251.406 
## 
## Information Criteria
## ------------------------------------
##                    
## Akaike       2.5212
## Bayes        2.5849
## Shibata      2.5209
## Hannan-Quinn 2.5454
## 
## Weighted Ljung-Box Test on Standardized Residuals
## ------------------------------------
##                          statistic   p-value
## Lag[1]                       0.836 3.606e-01
## Lag[2*(p+q)+(p+q)-1][11]    25.134 0.000e+00
## Lag[4*(p+q)+(p+q)-1][19]    35.106 9.463e-12
## d.o.f=4
## H0 : No serial correlation
## 
## Weighted Ljung-Box Test on Standardized Squared Residuals
## ------------------------------------
##                          statistic p-value
## Lag[1]                    0.003225  0.9547
## Lag[2*(p+q)+(p+q)-1][8]   7.767723  0.1058
## Lag[4*(p+q)+(p+q)-1][14] 11.334264  0.1206
## d.o.f=3
## 
## Weighted ARCH LM Tests
## ------------------------------------
##             Statistic Shape Scale  P-Value
## ARCH Lag[4]     7.171 0.500 2.000 0.007411
## ARCH Lag[6]     7.243 1.461 1.711 0.034992
## ARCH Lag[8]     8.814 2.368 1.583 0.041492
## 
## Nyblom stability test
## ------------------------------------
## Joint Statistic:  8.6596
## Individual Statistics:              
## mu     0.01186
## ar1    0.36525
## ar2    0.33928
## ma1    3.55601
## ma2    4.14021
## omega  0.32916
## alpha1 0.11374
## alpha2 0.21821
## beta1  0.36879
## gamma1 0.12825
## gamma2 0.23386
## skew   0.40919
## shape  0.81020
## 
## Asymptotic Critical Values (10% 5% 1%)
## Joint Statistic:          2.89 3.15 3.69
## Individual Statistic:     0.35 0.47 0.75
## 
## Sign Bias Test
## ------------------------------------
##                    t-value    prob sig
## Sign Bias           1.6118 0.10732    
## Negative Sign Bias  2.2794 0.02285  **
## Positive Sign Bias  0.3133 0.75409    
## Joint Effect        5.8595 0.11865    
## 
## 
## Adjusted Pearson Goodness-of-Fit Test:
## ------------------------------------
##   group statistic p-value(g-1)
## 1    20     72.23    3.896e-08
## 2    30     72.06    1.574e-05
## 3    40     93.35    2.369e-06
## 4    50     94.86    9.345e-05
## 
## 
## Elapsed time : 1.662677

PHÂN PHỐI Generalized Error Distribution (ged)

sgx21ged.spec <- ugarchspec(variance.model = list(model = "gjrGARCH", garchOrder = c(2,1)), mean.model = list(armaOrder = c(2,2), include.mean = TRUE), distribution.model = "ged")

sgx21ged <- ugarchfit(sgx21ged.spec, SGX) 
print(sgx21ged)
## 
## *---------------------------------*
## *          GARCH Model Fit        *
## *---------------------------------*
## 
## Conditional Variance Dynamics    
## -----------------------------------
## GARCH Model  : gjrGARCH(2,1)
## Mean Model   : ARFIMA(2,0,2)
## Distribution : ged 
## 
## Optimal Parameters
## ------------------------------------
##         Estimate  Std. Error   t value Pr(>|t|)
## mu      7.429268    0.176237  42.15510 0.000000
## ar1     0.448433    0.007604  58.97578 0.000000
## ar2     0.549954    0.007594  72.42249 0.000000
## ma1    -0.237511    0.026294  -9.03299 0.000000
## ma2    -0.629894    0.026176 -24.06405 0.000000
## omega   0.009569    0.004108   2.32951 0.019832
## alpha1  0.120364    0.060832   1.97865 0.047856
## alpha2  0.020413    0.079913   0.25544 0.798386
## beta1   0.844891    0.030032  28.13284 0.000000
## gamma1 -0.116177    0.062259  -1.86603 0.062037
## gamma2  0.142569    0.082453   1.72909 0.083793
## shape   2.004572    0.162686  12.32175 0.000000
## 
## Robust Standard Errors:
##         Estimate  Std. Error   t value Pr(>|t|)
## mu      7.429268    0.079159  93.85262 0.000000
## ar1     0.448433    0.001210 370.64981 0.000000
## ar2     0.549954    0.001261 436.08123 0.000000
## ma1    -0.237511    0.029065  -8.17159 0.000000
## ma2    -0.629894    0.028140 -22.38428 0.000000
## omega   0.009569    0.005202   1.83960 0.065828
## alpha1  0.120364    0.067969   1.77087 0.076582
## alpha2  0.020413    0.103182   0.19783 0.843177
## beta1   0.844891    0.040757  20.72981 0.000000
## gamma1 -0.116177    0.076708  -1.51454 0.129888
## gamma2  0.142569    0.099340   1.43517 0.151239
## shape   2.004572    0.257869   7.77359 0.000000
## 
## LogLikelihood : -1251.573 
## 
## Information Criteria
## ------------------------------------
##                    
## Akaike       2.5196
## Bayes        2.5783
## Shibata      2.5193
## Hannan-Quinn 2.5419
## 
## Weighted Ljung-Box Test on Standardized Residuals
## ------------------------------------
##                          statistic   p-value
## Lag[1]                      0.8851 3.468e-01
## Lag[2*(p+q)+(p+q)-1][11]   25.7916 0.000e+00
## Lag[4*(p+q)+(p+q)-1][19]   35.6502 4.665e-12
## d.o.f=4
## H0 : No serial correlation
## 
## Weighted Ljung-Box Test on Standardized Squared Residuals
## ------------------------------------
##                          statistic p-value
## Lag[1]                    0.009228  0.9235
## Lag[2*(p+q)+(p+q)-1][8]   7.857314  0.1014
## Lag[4*(p+q)+(p+q)-1][14] 11.272823  0.1236
## d.o.f=3
## 
## Weighted ARCH LM Tests
## ------------------------------------
##             Statistic Shape Scale  P-Value
## ARCH Lag[4]     6.904 0.500 2.000 0.008598
## ARCH Lag[6]     6.954 1.461 1.711 0.040808
## ARCH Lag[8]     8.470 2.368 1.583 0.049254
## 
## Nyblom stability test
## ------------------------------------
## Joint Statistic:  8.8282
## Individual Statistics:              
## mu     0.01364
## ar1    0.33440
## ar2    0.30743
## ma1    3.63764
## ma2    4.12716
## omega  0.30786
## alpha1 0.11886
## alpha2 0.25003
## beta1  0.36047
## gamma1 0.15154
## gamma2 0.30267
## shape  1.82882
## 
## Asymptotic Critical Values (10% 5% 1%)
## Joint Statistic:          2.69 2.96 3.51
## Individual Statistic:     0.35 0.47 0.75
## 
## Sign Bias Test
## ------------------------------------
##                    t-value    prob sig
## Sign Bias            1.656 0.09804   *
## Negative Sign Bias   2.251 0.02462  **
## Positive Sign Bias   0.236 0.81347    
## Joint Effect         5.731 0.12547    
## 
## 
## Adjusted Pearson Goodness-of-Fit Test:
## ------------------------------------
##   group statistic p-value(g-1)
## 1    20     83.16    5.271e-10
## 2    30     78.35    2.017e-06
## 3    40     96.38    9.228e-07
## 4    50     98.55    3.509e-05
## 
## 
## Elapsed time : 1.181122

PHÂN PHỐI Generalized Error Distribution đối xứng (sged)

sgx21sged.spec <- ugarchspec(variance.model = list(model = 'gjrGARCH', garchOrder = c(2,1)), mean.model = list(armaOrder = c(2,2), include.mean = TRUE), distribution.model = 'sged')

sgx21sged <- ugarchfit(sgx21sged.spec, SGX)
print(sgx21sged)
## 
## *---------------------------------*
## *          GARCH Model Fit        *
## *---------------------------------*
## 
## Conditional Variance Dynamics    
## -----------------------------------
## GARCH Model  : gjrGARCH(2,1)
## Mean Model   : ARFIMA(2,0,2)
## Distribution : sged 
## 
## Optimal Parameters
## ------------------------------------
##         Estimate  Std. Error   t value Pr(>|t|)
## mu      7.454337    0.180311  41.34155 0.000000
## ar1     0.443181    0.008279  53.52990 0.000000
## ar2     0.555130    0.008263  67.18121 0.000000
## ma1    -0.234663    0.026493  -8.85760 0.000000
## ma2    -0.631647    0.026635 -23.71475 0.000000
## omega   0.009081    0.004018   2.26025 0.023806
## alpha1  0.123883    0.060704   2.04079 0.041272
## alpha2  0.009697    0.081558   0.11889 0.905361
## beta1   0.849930    0.030683  27.70072 0.000000
## gamma1 -0.118439    0.061976  -1.91105 0.055998
## gamma2  0.148793    0.082671   1.79982 0.071889
## skew    1.045825    0.056139  18.62912 0.000000
## shape   2.050184    0.173796  11.79649 0.000000
## 
## Robust Standard Errors:
##         Estimate  Std. Error    t value Pr(>|t|)
## mu      7.454337    0.093820  79.453347 0.000000
## ar1     0.443181    0.001393 318.115068 0.000000
## ar2     0.555130    0.001410 393.796897 0.000000
## ma1    -0.234663    0.030075  -7.802640 0.000000
## ma2    -0.631647    0.029367 -21.508522 0.000000
## omega   0.009081    0.005871   1.546926 0.121881
## alpha1  0.123883    0.071631   1.729472 0.083725
## alpha2  0.009697    0.122535   0.079134 0.936926
## beta1   0.849930    0.049627  17.126458 0.000000
## gamma1 -0.118439    0.079962  -1.481194 0.138555
## gamma2  0.148793    0.113225   1.314137 0.188800
## skew    1.045825    0.097797  10.693884 0.000000
## shape   2.050184    0.316711   6.473364 0.000000
## 
## LogLikelihood : -1251.217 
## 
## Information Criteria
## ------------------------------------
##                    
## Akaike       2.5209
## Bayes        2.5845
## Shibata      2.5205
## Hannan-Quinn 2.5451
## 
## Weighted Ljung-Box Test on Standardized Residuals
## ------------------------------------
##                          statistic   p-value
## Lag[1]                      0.7295 3.930e-01
## Lag[2*(p+q)+(p+q)-1][11]   26.0119 0.000e+00
## Lag[4*(p+q)+(p+q)-1][19]   35.8795 3.459e-12
## d.o.f=4
## H0 : No serial correlation
## 
## Weighted Ljung-Box Test on Standardized Squared Residuals
## ------------------------------------
##                          statistic p-value
## Lag[1]                     0.01652  0.8977
## Lag[2*(p+q)+(p+q)-1][8]    7.86507  0.1010
## Lag[4*(p+q)+(p+q)-1][14]  11.37018  0.1190
## d.o.f=3
## 
## Weighted ARCH LM Tests
## ------------------------------------
##             Statistic Shape Scale  P-Value
## ARCH Lag[4]     7.127 0.500 2.000 0.007594
## ARCH Lag[6]     7.191 1.461 1.711 0.035983
## ARCH Lag[8]     8.741 2.368 1.583 0.043033
## 
## Nyblom stability test
## ------------------------------------
## Joint Statistic:  9.7217
## Individual Statistics:              
## mu     0.01723
## ar1    0.38413
## ar2    0.35421
## ma1    3.74735
## ma2    4.26853
## omega  0.33821
## alpha1 0.13394
## alpha2 0.25827
## beta1  0.39915
## gamma1 0.14210
## gamma2 0.28657
## skew   0.32613
## shape  1.80294
## 
## Asymptotic Critical Values (10% 5% 1%)
## Joint Statistic:          2.89 3.15 3.69
## Individual Statistic:     0.35 0.47 0.75
## 
## Sign Bias Test
## ------------------------------------
##                    t-value    prob sig
## Sign Bias           1.5318 0.12590    
## Negative Sign Bias  2.2232 0.02643  **
## Positive Sign Bias  0.3382 0.73525    
## Joint Effect        5.5543 0.13543    
## 
## 
## Adjusted Pearson Goodness-of-Fit Test:
## ------------------------------------
##   group statistic p-value(g-1)
## 1    20     75.27    1.200e-08
## 2    30     77.33    2.830e-06
## 3    40     91.12    4.691e-06
## 4    50     94.46    1.037e-04
## 
## 
## Elapsed time : 1.792915

GJR-GARCH(22)

PHÂN PHỐI CHUẨN

sgx22n.spec <- ugarchspec(variance.model = list(model = 'gjrGARCH', garchOrder = c(2,2)), mean.model = list(armaOrder = c(2,2), include.mean = TRUE), distribution.model = 'norm')

sgx22n <- ugarchfit(sgx22n.spec, SGX)
print(sgx22n)
## 
## *---------------------------------*
## *          GARCH Model Fit        *
## *---------------------------------*
## 
## Conditional Variance Dynamics    
## -----------------------------------
## GARCH Model  : gjrGARCH(2,2)
## Mean Model   : ARFIMA(2,0,2)
## Distribution : norm 
## 
## Optimal Parameters
## ------------------------------------
##         Estimate  Std. Error   t value Pr(>|t|)
## mu      7.428427    0.173531  42.80743 0.000000
## ar1     0.449141    0.007568  59.34578 0.000000
## ar2     0.549254    0.007559  72.65842 0.000000
## ma1    -0.237939    0.025372  -9.37791 0.000000
## ma2    -0.629574    0.024756 -25.43093 0.000000
## omega   0.009540    0.002381   4.00592 0.000062
## alpha1  0.120515    0.059254   2.03388 0.041963
## alpha2  0.020041    0.067699   0.29603 0.767209
## beta1   0.845033    0.486408   1.73729 0.082335
## beta2   0.000000    0.431700   0.00000 1.000000
## gamma1 -0.116347    0.061982  -1.87711 0.060503
## gamma2  0.143023    0.080507   1.77653 0.075646
## 
## Robust Standard Errors:
##         Estimate  Std. Error   t value Pr(>|t|)
## mu      7.428427    0.066430 111.82368 0.000000
## ar1     0.449141    0.001172 383.16046 0.000000
## ar2     0.549254    0.001264 434.55981 0.000000
## ma1    -0.237939    0.024739  -9.61780 0.000000
## ma2    -0.629574    0.024912 -25.27185 0.000000
## omega   0.009540    0.007462   1.27841 0.201106
## alpha1  0.120515    0.066805   1.80399 0.071232
## alpha2  0.020041    0.127190   0.15757 0.874800
## beta1   0.845033    0.551493   1.53227 0.125457
## beta2   0.000000    0.469029   0.00000 1.000000
## gamma1 -0.116347    0.076919  -1.51260 0.130382
## gamma2  0.143023    0.099369   1.43932 0.150060
## 
## LogLikelihood : -1251.573 
## 
## Information Criteria
## ------------------------------------
##                    
## Akaike       2.5196
## Bayes        2.5783
## Shibata      2.5193
## Hannan-Quinn 2.5419
## 
## Weighted Ljung-Box Test on Standardized Residuals
## ------------------------------------
##                          statistic   p-value
## Lag[1]                      0.8955 3.440e-01
## Lag[2*(p+q)+(p+q)-1][11]   25.7544 0.000e+00
## Lag[4*(p+q)+(p+q)-1][19]   35.6144 4.888e-12
## d.o.f=4
## H0 : No serial correlation
## 
## Weighted Ljung-Box Test on Standardized Squared Residuals
## ------------------------------------
##                          statistic p-value
## Lag[1]                    0.009153  0.9238
## Lag[2*(p+q)+(p+q)-1][11]  9.521247  0.1165
## Lag[4*(p+q)+(p+q)-1][19] 14.515139  0.1163
## d.o.f=4
## 
## Weighted ARCH LM Tests
## ------------------------------------
##             Statistic Shape Scale P-Value
## ARCH Lag[5]   0.01374 0.500 2.000  0.9067
## ARCH Lag[7]   0.49925 1.473 1.746  0.8974
## ARCH Lag[9]   2.05147 2.402 1.619  0.7485
## 
## Nyblom stability test
## ------------------------------------
## Joint Statistic:  9.7639
## Individual Statistics:              
## mu     0.01347
## ar1    0.33359
## ar2    0.30682
## ma1    3.63095
## ma2    4.12594
## omega  0.30728
## alpha1 0.11890
## alpha2 0.25003
## beta1  0.36067
## beta2  0.38964
## gamma1 0.15151
## gamma2 0.30275
## 
## Asymptotic Critical Values (10% 5% 1%)
## Joint Statistic:          2.69 2.96 3.51
## Individual Statistic:     0.35 0.47 0.75
## 
## Sign Bias Test
## ------------------------------------
##                    t-value    prob sig
## Sign Bias           1.6566 0.09792   *
## Negative Sign Bias  2.2516 0.02456  **
## Positive Sign Bias  0.2379 0.81200    
## Joint Effect        5.7384 0.12506    
## 
## 
## Adjusted Pearson Goodness-of-Fit Test:
## ------------------------------------
##   group statistic p-value(g-1)
## 1    20     84.80    2.733e-10
## 2    30     77.45    2.720e-06
## 3    40     97.90    5.722e-07
## 4    50     97.35    4.837e-05
## 
## 
## Elapsed time : 0.514236

PHÂN PHỐI STUDENT

sgx22std.spec <- ugarchspec(variance.model = list(model = 'gjrGARCH', garchOrder = c(2,2)), mean.model = list(armaOrder = c(2,2), include.mean = TRUE), distribution.model = 'std')

sgx22std <- ugarchfit(sgx22std.spec, SGX)
print(sgx22std)
## 
## *---------------------------------*
## *          GARCH Model Fit        *
## *---------------------------------*
## 
## Conditional Variance Dynamics    
## -----------------------------------
## GARCH Model  : gjrGARCH(2,2)
## Mean Model   : ARFIMA(2,0,2)
## Distribution : std 
## 
## Optimal Parameters
## ------------------------------------
##         Estimate  Std. Error   t value Pr(>|t|)
## mu      7.418710    0.172148  43.09489 0.000000
## ar1     0.453328    0.007414  61.14179 0.000000
## ar2     0.545169    0.007405  73.61767 0.000000
## ma1    -0.240535    0.025286  -9.51262 0.000000
## ma2    -0.628764    0.024792 -25.36188 0.000000
## omega   0.008980    0.002581   3.47878 0.000504
## alpha1  0.122509    0.060849   2.01333 0.044080
## alpha2  0.009827    0.067052   0.14656 0.883478
## beta1   0.849362    0.519444   1.63514 0.102020
## beta2   0.000000    0.462378   0.00000 1.000000
## gamma1 -0.117960    0.063203  -1.86637 0.061990
## gamma2  0.152563    0.081298   1.87658 0.060575
## shape  80.059232  106.267579   0.75337 0.451225
## 
## Robust Standard Errors:
##         Estimate  Std. Error    t value Pr(>|t|)
## mu      7.418710    0.064741 114.591366 0.000000
## ar1     0.453328    0.001134 399.869721 0.000000
## ar2     0.545169    0.001201 453.901420 0.000000
## ma1    -0.240535    0.025027  -9.610947 0.000000
## ma2    -0.628764    0.025094 -25.056275 0.000000
## omega   0.008980    0.006980   1.286640 0.198220
## alpha1  0.122509    0.067074   1.826483 0.067777
## alpha2  0.009827    0.121211   0.081076 0.935382
## beta1   0.849362    0.585586   1.450448 0.146934
## beta2   0.000000    0.504081   0.000000 1.000000
## gamma1 -0.117960    0.076365  -1.544694 0.122420
## gamma2  0.152563    0.094563   1.613359 0.106666
## shape  80.059232   80.909984   0.989485 0.322426
## 
## LogLikelihood : -1251.649 
## 
## Information Criteria
## ------------------------------------
##                    
## Akaike       2.5217
## Bayes        2.5854
## Shibata      2.5214
## Hannan-Quinn 2.5459
## 
## Weighted Ljung-Box Test on Standardized Residuals
## ------------------------------------
##                          statistic   p-value
## Lag[1]                      0.8969 3.436e-01
## Lag[2*(p+q)+(p+q)-1][11]   25.2650 0.000e+00
## Lag[4*(p+q)+(p+q)-1][19]   35.2075 8.298e-12
## d.o.f=4
## H0 : No serial correlation
## 
## Weighted Ljung-Box Test on Standardized Squared Residuals
## ------------------------------------
##                          statistic p-value
## Lag[1]                    0.006388  0.9363
## Lag[2*(p+q)+(p+q)-1][11]  9.443098  0.1204
## Lag[4*(p+q)+(p+q)-1][19] 14.559853  0.1145
## d.o.f=4
## 
## Weighted ARCH LM Tests
## ------------------------------------
##             Statistic Shape Scale P-Value
## ARCH Lag[5]   0.01464 0.500 2.000  0.9037
## ARCH Lag[7]   0.49180 1.473 1.746  0.8994
## ARCH Lag[9]   2.11451 2.402 1.619  0.7362
## 
## Nyblom stability test
## ------------------------------------
## Joint Statistic:  10.3358
## Individual Statistics:              
## mu     0.01149
## ar1    0.33006
## ar2    0.30484
## ma1    3.61858
## ma2    4.13586
## omega  0.29757
## alpha1 0.10314
## alpha2 0.21565
## beta1  0.33729
## beta2  0.36469
## gamma1 0.14105
## gamma2 0.26441
## shape  0.96256
## 
## Asymptotic Critical Values (10% 5% 1%)
## Joint Statistic:          2.89 3.15 3.69
## Individual Statistic:     0.35 0.47 0.75
## 
## Sign Bias Test
## ------------------------------------
##                    t-value    prob sig
## Sign Bias            1.660 0.09725   *
## Negative Sign Bias   2.275 0.02314  **
## Positive Sign Bias   0.262 0.79337    
## Joint Effect         5.862 0.11855    
## 
## 
## Adjusted Pearson Goodness-of-Fit Test:
## ------------------------------------
##   group statistic p-value(g-1)
## 1    20     75.94    9.206e-09
## 2    30     75.00    6.100e-06
## 3    40     90.24    6.119e-06
## 4    50     96.05    6.825e-05
## 
## 
## Elapsed time : 1.644891

PHÂN PHỐI ĐỐI XỨNG (sstd)

sgx22sstd.spec <- ugarchspec(variance.model = list( model = 'gjrGARCH', garchOrder = c(2,2)), mean.model = list(armaOrder = c(2,2), include.mean = TRUE), distribution.model = 'sstd')

sgx22sstd <- ugarchfit(sgx22sstd.spec, SGX)

print(sgx22sstd)
## 
## *---------------------------------*
## *          GARCH Model Fit        *
## *---------------------------------*
## 
## Conditional Variance Dynamics    
## -----------------------------------
## GARCH Model  : gjrGARCH(2,2)
## Mean Model   : ARFIMA(2,0,2)
## Distribution : sstd 
## 
## Optimal Parameters
## ------------------------------------
##         Estimate  Std. Error     t value Pr(>|t|)
## mu      7.439139    0.179206   41.511652 0.000000
## ar1     0.465164    0.003747  124.131508 0.000000
## ar2     0.532715    0.003796  140.340724 0.000000
## ma1    -0.245374    0.023012  -10.662751 0.000000
## ma2    -0.620988    0.023572  -26.344675 0.000000
## omega   0.006791    0.002999    2.264646 0.023534
## alpha1  0.080195    0.000246  325.811287 0.000000
## alpha2  0.000000    0.018072    0.000004 0.999997
## beta1   0.890223    0.000330 2694.147046 0.000000
## beta2   0.000000    0.010951    0.000013 0.999989
## gamma1 -0.200008    0.001580 -126.602530 0.000000
## gamma2  0.256084    0.034220    7.483371 0.000000
## skew    1.071306    0.057404   18.662726 0.000000
## shape  59.999809   51.988249    1.154103 0.248458
## 
## Robust Standard Errors:
##         Estimate  Std. Error     t value Pr(>|t|)
## mu      7.439139    0.086540   85.961776 0.000000
## ar1     0.465164    0.001179  394.451044 0.000000
## ar2     0.532715    0.001371  388.655702 0.000000
## ma1    -0.245374    0.025474   -9.632145 0.000000
## ma2    -0.620988    0.024951  -24.888631 0.000000
## omega   0.006791    0.003919    1.732953 0.083104
## alpha1  0.080195    0.002005   40.002682 0.000000
## alpha2  0.000000    0.024698    0.000003 0.999998
## beta1   0.890223    0.000386 2306.304795 0.000000
## beta2   0.000000    0.014711    0.000010 0.999992
## gamma1 -0.200008    0.001673 -119.539262 0.000000
## gamma2  0.256084    0.040633    6.302376 0.000000
## skew    1.071306    0.079991   13.392822 0.000000
## shape  59.999809   40.570318    1.478909 0.139165
## 
## LogLikelihood : -1248.421 
## 
## Information Criteria
## ------------------------------------
##                    
## Akaike       2.5173
## Bayes        2.5858
## Shibata      2.5169
## Hannan-Quinn 2.5433
## 
## Weighted Ljung-Box Test on Standardized Residuals
## ------------------------------------
##                          statistic   p-value
## Lag[1]                        1.00 3.172e-01
## Lag[2*(p+q)+(p+q)-1][11]     23.63 0.000e+00
## Lag[4*(p+q)+(p+q)-1][19]     33.29 9.764e-11
## d.o.f=4
## H0 : No serial correlation
## 
## Weighted Ljung-Box Test on Standardized Squared Residuals
## ------------------------------------
##                          statistic p-value
## Lag[1]                       1.857 0.17300
## Lag[2*(p+q)+(p+q)-1][11]    14.025 0.01393
## Lag[4*(p+q)+(p+q)-1][19]    21.578 0.00624
## d.o.f=4
## 
## Weighted ARCH LM Tests
## ------------------------------------
##             Statistic Shape Scale P-Value
## ARCH Lag[5]    0.1131 0.500 2.000  0.7366
## ARCH Lag[7]    2.2193 1.473 1.746  0.4577
## ARCH Lag[9]    5.2491 2.402 1.619  0.2418
## 
## Nyblom stability test
## ------------------------------------
## Joint Statistic:  12.1513
## Individual Statistics:              
## mu     0.02880
## ar1    0.32523
## ar2    0.28849
## ma1    3.23686
## ma2    3.53882
## omega  0.31915
## alpha1 0.05350
## alpha2 0.13216
## beta1  0.36470
## beta2  0.40749
## gamma1 0.08192
## gamma2 0.15664
## skew   0.32132
## shape  0.72057
## 
## Asymptotic Critical Values (10% 5% 1%)
## Joint Statistic:          3.08 3.34 3.9
## Individual Statistic:     0.35 0.47 0.75
## 
## Sign Bias Test
## ------------------------------------
##                    t-value   prob sig
## Sign Bias           1.3753 0.1694    
## Negative Sign Bias  0.7451 0.4564    
## Positive Sign Bias  0.2513 0.8016    
## Joint Effect        2.2545 0.5213    
## 
## 
## Adjusted Pearson Goodness-of-Fit Test:
## ------------------------------------
##   group statistic p-value(g-1)
## 1    20     87.39    9.593e-11
## 2    30     96.23    3.885e-09
## 3    40    116.32    1.275e-09
## 4    50    114.60    3.570e-07
## 
## 
## Elapsed time : 1.850571

PHÂN PHỐI Generalized Error Distribution (ged)

sgx22ged.spec <- ugarchspec(variance.model = list(model = "gjrGARCH", garchOrder = c(2,2)), mean.model = list(armaOrder = c(2,2), include.mean = TRUE), distribution.model = "ged")

sgx22ged <- ugarchfit(sgx22ged.spec, SGX) 
print(sgx22ged)
## 
## *---------------------------------*
## *          GARCH Model Fit        *
## *---------------------------------*
## 
## Conditional Variance Dynamics    
## -----------------------------------
## GARCH Model  : gjrGARCH(2,2)
## Mean Model   : ARFIMA(2,0,2)
## Distribution : ged 
## 
## Optimal Parameters
## ------------------------------------
##         Estimate  Std. Error   t value Pr(>|t|)
## mu      7.429281    0.176179  42.16893 0.000000
## ar1     0.448422    0.007604  58.97240 0.000000
## ar2     0.549964    0.007594  72.41939 0.000000
## ma1    -0.237496    0.026097  -9.10040 0.000000
## ma2    -0.629900    0.026004 -24.22335 0.000000
## omega   0.009571    0.002716   3.52384 0.000425
## alpha1  0.120335    0.059369   2.02691 0.042672
## alpha2  0.020472    0.069970   0.29258 0.769845
## beta1   0.844873    0.483702   1.74668 0.080692
## beta2   0.000000    0.429672   0.00000 1.000000
## gamma1 -0.116163    0.062175  -1.86832 0.061717
## gamma2  0.142539    0.081897   1.74046 0.081779
## shape   2.004638    0.162124  12.36483 0.000000
## 
## Robust Standard Errors:
##         Estimate  Std. Error   t value Pr(>|t|)
## mu      7.429281    0.078620  94.49651 0.000000
## ar1     0.448422    0.001206 371.80558 0.000000
## ar2     0.549964    0.001259 436.86417 0.000000
## ma1    -0.237496    0.028706  -8.27326 0.000000
## ma2    -0.629900    0.028056 -22.45121 0.000000
## omega   0.009571    0.007043   1.35894 0.174166
## alpha1  0.120335    0.066974   1.79673 0.072378
## alpha2  0.020472    0.123219   0.16614 0.868047
## beta1   0.844873    0.533360   1.58406 0.113181
## beta2   0.000000    0.454683   0.00000 1.000000
## gamma1 -0.116163    0.076050  -1.52746 0.126648
## gamma2  0.142539    0.097267   1.46543 0.142804
## shape   2.004638    0.260253   7.70265 0.000000
## 
## LogLikelihood : -1251.573 
## 
## Information Criteria
## ------------------------------------
##                    
## Akaike       2.5216
## Bayes        2.5852
## Shibata      2.5212
## Hannan-Quinn 2.5458
## 
## Weighted Ljung-Box Test on Standardized Residuals
## ------------------------------------
##                          statistic   p-value
## Lag[1]                      0.8855 3.467e-01
## Lag[2*(p+q)+(p+q)-1][11]   25.7922 0.000e+00
## Lag[4*(p+q)+(p+q)-1][19]   35.6505 4.663e-12
## d.o.f=4
## H0 : No serial correlation
## 
## Weighted Ljung-Box Test on Standardized Squared Residuals
## ------------------------------------
##                          statistic p-value
## Lag[1]                     0.00911  0.9240
## Lag[2*(p+q)+(p+q)-1][11]   9.52552  0.1163
## Lag[4*(p+q)+(p+q)-1][19]  14.51810  0.1162
## d.o.f=4
## 
## Weighted ARCH LM Tests
## ------------------------------------
##             Statistic Shape Scale P-Value
## ARCH Lag[5]   0.01343 0.500 2.000  0.9078
## ARCH Lag[7]   0.50064 1.473 1.746  0.8970
## ARCH Lag[9]   2.05548 2.402 1.619  0.7477
## 
## Nyblom stability test
## ------------------------------------
## Joint Statistic:  11.2675
## Individual Statistics:              
## mu     0.01364
## ar1    0.33443
## ar2    0.30745
## ma1    3.63701
## ma2    4.12683
## omega  0.30806
## alpha1 0.11894
## alpha2 0.25032
## beta1  0.36071
## beta2  0.38964
## gamma1 0.15153
## gamma2 0.30299
## shape  1.82936
## 
## Asymptotic Critical Values (10% 5% 1%)
## Joint Statistic:          2.89 3.15 3.69
## Individual Statistic:     0.35 0.47 0.75
## 
## Sign Bias Test
## ------------------------------------
##                    t-value    prob sig
## Sign Bias           1.6559 0.09806   *
## Negative Sign Bias  2.2504 0.02464  **
## Positive Sign Bias  0.2357 0.81371    
## Joint Effect        5.7289 0.12557    
## 
## 
## Adjusted Pearson Goodness-of-Fit Test:
## ------------------------------------
##   group statistic p-value(g-1)
## 1    20     83.16    5.271e-10
## 2    30     78.35    2.017e-06
## 3    40     96.38    9.228e-07
## 4    50     98.55    3.509e-05
## 
## 
## Elapsed time : 1.272338

PHÂN PHỐI Generalized Error Distribution đối xứng (sged)

sgx22sged.spec <- ugarchspec(variance.model = list(model = 'gjrGARCH', garchOrder = c(2,2)), mean.model = list(armaOrder = c(2,2), include.mean = TRUE), distribution.model = 'sged')

sgx22sged <- ugarchfit(sgx22sged.spec, SGX)
print(sgx22sged)
## 
## *---------------------------------*
## *          GARCH Model Fit        *
## *---------------------------------*
## 
## Conditional Variance Dynamics    
## -----------------------------------
## GARCH Model  : gjrGARCH(2,2)
## Mean Model   : ARFIMA(2,0,2)
## Distribution : sged 
## 
## Optimal Parameters
## ------------------------------------
##         Estimate  Std. Error   t value Pr(>|t|)
## mu      7.454066    0.180040  41.40230 0.000000
## ar1     0.443320    0.008266  53.63262 0.000000
## ar2     0.554993    0.008251  67.26121 0.000000
## ma1    -0.234788    0.026287  -8.93188 0.000000
## ma2    -0.631565    0.026471 -23.85884 0.000000
## omega   0.009058    0.002488   3.64009 0.000273
## alpha1  0.123982    0.059148   2.09614 0.036069
## alpha2  0.009555    0.071889   0.13291 0.894268
## beta1   0.849995    0.513953   1.65384 0.098160
## beta2   0.000000    0.457953   0.00000 1.000000
## gamma1 -0.118517    0.061917  -1.91412 0.055605
## gamma2  0.149012    0.081773   1.82227 0.068415
## skew    1.045834    0.056098  18.64294 0.000000
## shape   2.049751    0.173518  11.81291 0.000000
## 
## Robust Standard Errors:
##         Estimate  Std. Error    t value Pr(>|t|)
## mu      7.454066    0.091670  81.314059 0.000000
## ar1     0.443320    0.001388 319.294632 0.000000
## ar2     0.554993    0.001410 393.723319 0.000000
## ma1    -0.234788    0.029425  -7.979223 0.000000
## ma2    -0.631565    0.029128 -21.682222 0.000000
## omega   0.009058    0.007993   1.133262 0.257104
## alpha1  0.123982    0.069658   1.779862 0.075098
## alpha2  0.009555    0.146110   0.065393 0.947861
## beta1   0.849995    0.616468   1.378814 0.167952
## beta2   0.000000    0.524520   0.000000 1.000000
## gamma1 -0.118517    0.079004  -1.500144 0.133577
## gamma2  0.149012    0.109506   1.360769 0.173587
## skew    1.045834    0.096895  10.793430 0.000000
## shape   2.049751    0.319503   6.415432 0.000000
## 
## LogLikelihood : -1251.217 
## 
## Information Criteria
## ------------------------------------
##                    
## Akaike       2.5229
## Bayes        2.5914
## Shibata      2.5225
## Hannan-Quinn 2.5489
## 
## Weighted Ljung-Box Test on Standardized Residuals
## ------------------------------------
##                          statistic   p-value
## Lag[1]                      0.7281 3.935e-01
## Lag[2*(p+q)+(p+q)-1][11]   25.9961 0.000e+00
## Lag[4*(p+q)+(p+q)-1][19]   35.8652 3.524e-12
## d.o.f=4
## H0 : No serial correlation
## 
## Weighted Ljung-Box Test on Standardized Squared Residuals
## ------------------------------------
##                          statistic p-value
## Lag[1]                      0.0163  0.8984
## Lag[2*(p+q)+(p+q)-1][11]    9.5767  0.1138
## Lag[4*(p+q)+(p+q)-1][19]   14.7160  0.1083
## d.o.f=4
## 
## Weighted ARCH LM Tests
## ------------------------------------
##             Statistic Shape Scale P-Value
## ARCH Lag[5]    0.0109 0.500 2.000  0.9169
## ARCH Lag[7]    0.5271 1.473 1.746  0.8899
## ARCH Lag[9]    2.1117 2.402 1.619  0.7367
## 
## Nyblom stability test
## ------------------------------------
## Joint Statistic:  12.2201
## Individual Statistics:             
## mu     0.0172
## ar1    0.3842
## ar2    0.3543
## ma1    3.7504
## ma2    4.2720
## omega  0.3365
## alpha1 0.1343
## alpha2 0.2590
## beta1  0.3992
## beta2  0.4308
## gamma1 0.1424
## gamma2 0.2881
## skew   0.3262
## shape  1.8022
## 
## Asymptotic Critical Values (10% 5% 1%)
## Joint Statistic:          3.08 3.34 3.9
## Individual Statistic:     0.35 0.47 0.75
## 
## Sign Bias Test
## ------------------------------------
##                    t-value    prob sig
## Sign Bias           1.5330 0.12560    
## Negative Sign Bias  2.2256 0.02626  **
## Positive Sign Bias  0.3395 0.73430    
## Joint Effect        5.5669 0.13469    
## 
## 
## Adjusted Pearson Goodness-of-Fit Test:
## ------------------------------------
##   group statistic p-value(g-1)
## 1    20     75.27    1.200e-08
## 2    30     77.33    2.830e-06
## 3    40     91.12    4.691e-06
## 4    50     95.55    7.783e-05
## 
## 
## Elapsed time : 2.00112

LỰA CHỌN MÔ HÌNH GJR-GARCH

LỰA CHỌN MÔ HÌNH BIÊN PHÙ HỢP NHẤT CHO CHUỖI SGX

sgx.model.list <- list(garch11n = sgx11n, 
                        garch11std = sgx11std,
                        garch11sstd = sgx11sstd,
                        garch11ged = sgx11ged,
                        garch11sged = sgx11sged,
                        
                        garch12n = sgx12n, 
                        garch12std = sgx12std,
                        garch12sstd = sgx12sstd,
                        garch12ged = sgx12ged,
                        garch12sged = sgx12sged,
                        
                        garch21n = sgx21n, 
                        garch21std = sgx21std,
                        garch21sstd = sgx21sstd,
                        garch21ged = sgx21ged,
                        garch21sged = sgx21sged,
                        
                        garch22n = sgx22n, 
                        garch22std = sgx22std,
                        garch22sstd = sgx22sstd,
                        garch22ged = sgx22ged,
                        garch22sged = sgx22sged)

sgx.info.mat <- sapply(sgx.model.list, infocriteria)

rownames(sgx.info.mat) <- rownames(infocriteria(sgx11n))

sgx.info.mat
##              garch11n garch11std garch11sstd garch11ged garch11sged garch12n
## Akaike       2.526233   2.528161    2.529850   2.528179    2.529741 2.528227
## Bayes        2.570298   2.577122    2.583707   2.577140    2.583597 2.577188
## Shibata      2.526074   2.527965    2.529613   2.527983    2.529504 2.528031
## Hannan-Quinn 2.542978   2.546767    2.550317   2.546785    2.550207 2.546833
##              garch12std garch12sstd garch12ged garch12sged garch21n garch21std
## Akaike         2.530165    2.531844   2.530173    2.531735 2.512831   2.519675
## Bayes          2.584022    2.590597   2.584030    2.590487 2.566688   2.578428
## Shibata        2.529928    2.531562   2.529936    2.531453 2.512594   2.519393
## Hannan-Quinn   2.550632    2.554171   2.550639    2.554062 2.533298   2.542002
##              garch21sstd garch21ged garch21sged garch22n garch22std garch22sstd
## Akaike          2.521248   2.519586    2.520872 2.519587   2.521734    2.517290
## Bayes           2.584897   2.578339    2.584520 2.578340   2.585383    2.585835
## Shibata         2.520918   2.519304    2.520541 2.519305   2.521403    2.516908
## Hannan-Quinn    2.545436   2.541913    2.545059 2.541914   2.545921    2.543338
##              garch22ged garch22sged
## Akaike         2.521580    2.522865
## Bayes          2.585229    2.591410
## Shibata        2.521250    2.522483
## Hannan-Quinn   2.545768    2.548913
sgx.inds <- which(sgx.info.mat == min(sgx.info.mat), arr.ind = TRUE)

model.sgx <- colnames(sgx.info.mat)[sgx.inds[,2]]

model.sgx
## [1] "garch21n"

THAM SỐ ƯỚC LƯỢNG MÔ HÌNH BIÊN PHÙ HỢP NHẤT

sgx21n
## 
## *---------------------------------*
## *          GARCH Model Fit        *
## *---------------------------------*
## 
## Conditional Variance Dynamics    
## -----------------------------------
## GARCH Model  : gjrGARCH(2,1)
## Mean Model   : ARFIMA(2,0,2)
## Distribution : norm 
## 
## Optimal Parameters
## ------------------------------------
##         Estimate  Std. Error     t value Pr(>|t|)
## mu      7.409277    0.135438  5.4706e+01  0.00000
## ar1     0.457257    0.010445  4.3776e+01  0.00000
## ar2     0.540432    0.004878  1.1080e+02  0.00000
## ma1    -0.242441    0.040632 -5.9668e+00  0.00000
## ma2    -0.623333    0.078023 -7.9890e+00  0.00000
## omega   0.007952    0.000298  2.6687e+01  0.00000
## alpha1  0.090683    0.002847  3.1856e+01  0.00000
## alpha2  0.000000    0.011334  1.7000e-05  0.99999
## beta1   0.882426    0.000050  1.7488e+04  0.00000
## gamma1 -0.209070    0.018021 -1.1601e+01  0.00000
## gamma2  0.259791    0.021214  1.2246e+01  0.00000
## 
## Robust Standard Errors:
##         Estimate  Std. Error     t value Pr(>|t|)
## mu      7.409277    0.482333   15.361336 0.000000
## ar1     0.457257    0.043226   10.578202 0.000000
## ar2     0.540432    0.023382   23.113535 0.000000
## ma1    -0.242441    0.180528   -1.342955 0.179286
## ma2    -0.623333    0.314488   -1.982055 0.047473
## omega   0.007952    0.003611    2.202075 0.027660
## alpha1  0.090683    0.033331    2.720655 0.006515
## alpha2  0.000000    0.038743    0.000005 0.999996
## beta1   0.882426    0.000117 7511.282693 0.000000
## gamma1 -0.209070    0.020212  -10.343891 0.000000
## gamma2  0.259791    0.036047    7.207023 0.000000
## 
## LogLikelihood : -1249.185 
## 
## Information Criteria
## ------------------------------------
##                    
## Akaike       2.5128
## Bayes        2.5667
## Shibata      2.5126
## Hannan-Quinn 2.5333
## 
## Weighted Ljung-Box Test on Standardized Residuals
## ------------------------------------
##                          statistic   p-value
## Lag[1]                       0.904 3.417e-01
## Lag[2*(p+q)+(p+q)-1][11]    24.385 0.000e+00
## Lag[4*(p+q)+(p+q)-1][19]    33.963 4.131e-11
## d.o.f=4
## H0 : No serial correlation
## 
## Weighted Ljung-Box Test on Standardized Squared Residuals
## ------------------------------------
##                          statistic p-value
## Lag[1]                       1.559 0.21184
## Lag[2*(p+q)+(p+q)-1][8]     10.729 0.02372
## Lag[4*(p+q)+(p+q)-1][14]    15.664 0.01834
## d.o.f=3
## 
## Weighted ARCH LM Tests
## ------------------------------------
##             Statistic Shape Scale  P-Value
## ARCH Lag[4]     8.739 0.500 2.000 0.003115
## ARCH Lag[6]     9.248 1.461 1.711 0.011920
## ARCH Lag[8]    12.296 2.368 1.583 0.006816
## 
## Nyblom stability test
## ------------------------------------
## Joint Statistic:  6.8514
## Individual Statistics:              
## mu     0.03081
## ar1    0.29252
## ar2    0.25877
## ma1    3.46303
## ma2    3.63152
## omega  0.28628
## alpha1 0.06594
## alpha2 0.16258
## beta1  0.35314
## gamma1 0.12787
## gamma2 0.23233
## 
## Asymptotic Critical Values (10% 5% 1%)
## Joint Statistic:          2.49 2.75 3.27
## Individual Statistic:     0.35 0.47 0.75
## 
## Sign Bias Test
## ------------------------------------
##                    t-value   prob sig
## Sign Bias           1.6171 0.1062    
## Negative Sign Bias  0.7976 0.4253    
## Positive Sign Bias  0.2712 0.7863    
## Joint Effect        3.2480 0.3549    
## 
## 
## Adjusted Pearson Goodness-of-Fit Test:
## ------------------------------------
##   group statistic p-value(g-1)
## 1    20     93.65    7.434e-12
## 2    30     85.11    2.023e-07
## 3    40    112.89    4.125e-09
## 4    50    124.27    1.813e-08
## 
## 
## Elapsed time : 0.5286181

TRÍCH XUẤT CHUỖI PHẦN DƯ CỦA CHUỖI LỢI SUẤT SGX

sgx.res <- residuals(sgx21n)/sigma(sgx21n)

fitdist(distribution = 'norm' , sgx.res, control = list())
## $pars
##         mu      sigma 
## 0.04993222 0.99128476 
## 
## $convergence
## [1] 0
## 
## $values
## [1] 1414.416 1414.416 1414.416
## 
## $lagrange
## [1] 0
## 
## $hessian
##      [,1]     [,2]
## [1,]    1    0.000
## [2,]    0 2038.455
## 
## $ineqx0
## NULL
## 
## $nfuneval
## [1] 30
## 
## $outer.iter
## [1] 2
## 
## $elapsed
## Time difference of 0.00407505 secs
## 
## $vscale
## [1] 1 1 1
u <- pdist(distribution = 'norm' , q = sgx.res, mu = 0.04993222 , sigma = 0.99128476)
print(u)
##            m.c.seq.row..seq.n...seq.col..drop...FALSE.
##    1-01-01                                  0.48864362
##    2-01-01                                  0.45082182
##    3-01-01                                  0.79073131
##    4-01-01                                  0.43807505
##    5-01-01                                  0.45516812
##    6-01-01                                  0.58845281
##    7-01-01                                  0.30344977
##    8-01-01                                  0.47224690
##    9-01-01                                  0.39971841
##   10-01-01                                  0.43730514
##        ...                                            
##  994-01-01                                  0.59530410
##  995-01-01                                  0.85782320
##  996-01-01                                  0.79584407
##  997-01-01                                  0.03479422
##  998-01-01                                  0.04563937
##  999-01-01                                  0.76541969
## 1000-01-01                                  0.09994958
## 1001-01-01                                  0.29459338
## 1002-01-01                                  0.05455010
## 1003-01-01                                  0.76806037

LỰA CHỌN MÔ HÌNH COPULA VÀ ƯỚC LƯỢNG THAM SỐ

?BiCopSelect
## No documentation for 'BiCopSelect' in specified packages and libraries:
## you could try '??BiCopSelect'