You are on page 1of 47

STT: 34

Họ và tên: Hà Thị Thúy Nga


Lớp: D03- Kinh tế lượng ứng dụng trong Tài chính- Ngân hàng
Ngành :Tài chính- Ngân hàng
SĐT: 0338123845
Bước 0: Lý thuyết chuỗi dừng
Một chuỗi dữ liệu thời gian được xem là dừng nếu như trung bình và phương sai
của phương trình không thay đổi theo thời gian và giá trị của đồng phương sai giữa
hai đoạn chỉ phụ thuộc vào khoảng cách hay độ trễ về thời gian giữa hai thời đoạn
này chứ không phụ thuộc vào thời điểm thực tế mà đồng phương sai được tính
(Ramanathan, 2002)
E ( Yt ) = µ = Const
Var ( Yt ) = σ2 = Const
Cov ( Yt , Yt-k ) = Yt = E [ ( Yt - µ )( Yt-k - µ ) ]

File US_CPI_monthly.wf1 gồm 9 bước


Bước 1 :
CÁCH 1: DỰA TRÊN BIỂU ĐỒ GRAPH
CPI là chuỗi không dừng: chuỗi có Trend lên
Vì: khi chia đồ thị làm ba phần, nhìn bằng mắt thường ta thấy có xu hướng đi lên,
giá trị trung bình là thay đổi => vi phạm điều kiện chuỗi dừng
Cách 2: Dựa trên kiểm định nghiệm đơn vị ( unit root test )của Dickey Fuller

Null Hypothesis: CPI has a unit root


Exogenous: Constant, Linear Trend
Lag Length: 2 (Automatic - based on SIC, maxlag=13)

t-Statistic   Prob.*

Augmented Dickey-Fuller test statistic -2.974913  0.1423


Test critical values: 1% level -4.010740
5% level -3.435413
10% level -3.141734

*MacKinnon (1996) one-sided p-values.

H0: CPIt có một nghiệm đơn vị (CPIt không dừng)


H1: CPIt không có nghiệm đơn vị (CPIt dừng)
p-value < α : bác bỏ giải thiết H0 với mức ý nghĩa α=1%,5% hay 10%
Ta có : p-value=0.1423> α=0.05=5%: chấp nhận giả thiết H0
Vậy CPIt có một nghiệm đơn vị( CPIt không dừng)
Bước 2:
Dựa trên hình lý giải cách lấy sai phân DCPI
Tìm sai phân DCPI bằng cách lấy CPI vào năm n trừ cho CPI vào năm n-1 (năm
trước đó). DCPI=CPI-CPI(-1).
Ví dụ DCPI của tháng 2 năm 2000 (0.46) sẽ là CPI của tháng 2 năm 2000 (77.87)
trừ cho CPI của tháng 1 năm 2000 (77.41).
- Gõ câu lệnh “ genr dcpi= cpi- cpi(-1)” để thêm biến sai phân vào phương
trình
- Ta có được sai phân của biến CPI
- Vẽ hình xem xét biến dcpi dừng hay không dừng

Bước 3:
Cách 1:

Quan sát hình, ta thấy chuỗi DCPI là chuỗi dừng vì:


Tất cả biến động nằm trong trường từ -0.8 đến 0.8=> chuỗi biến động trong
biên. Độ biến thiên tương đối ổn định. Hầu hết đều nằm trong 180 quan sát
nhưng chỉ có 1 quan sát nằm ngoài biên, tức là < 5%=> rất ít và không đáng
kể. Giá trị trung bình không thay đổi nhiều => DCPI là chuỗi dừng.

Cách 2:
Null Hypothesis: DCPI has a unit root
Exogenous: Constant
Lag Length: 1 (Automatic - based on SIC, maxlag=13)

t-Statistic   Prob.*

Augmented Dickey-Fuller test statistic -8.335383  0.0000


Test critical values: 1% level -3.467418
5% level -2.877729
10% level -2.575480

*MacKinnon (1996) one-sided p-values.

H0: DCPIt có một nghiệm đơn vị (DCPIt không dừng)


H1: DCPIt không có nghiệm đơn vị (DCPIt dừng)
p-value < α : bác bỏ giải thiết H0 với mức ý nghĩa α=1%,5% hay 10%
Ta có : p-value=0.0000< α=0.05=5%: Bác bỏ giả thiết H0
Vậy DCPIt không có nghiệm đơn vị( DCPIt dừng)

Bước 4:
Thực hiện bậc mô hình ARIMA(p,d,q) trên chuỗi dừng.
Bước 1 CPI là chuỗi không dừng và bước 2 thì DCPI là chuỗi dừng, thực hiện mô
hình ARIMA(p,d,q).
Ta có:
- MA(q) tương ứng cột ACF (AutoCorrelation):
- AR(p) tương ứng cột PACF (partial): (nằm ngoài đường biên gạch chấm)

p = 1, 2, 5,8, 24
d=1
q = 1,3,5,6,7,8,9,11,12,20,23,24
Bước 5: Ghép mô hình ARIMA (p,d,q)
1. ARIMA(1,1,1)
2. ARIMA(1,1,1)(2,1,5)
3. ARIMA(1,1,1)(5,1,6)
4. ARIMA(1,1,1)(5,1,9)
5. ARIMA(1,1,1)(8,1,9)
6. ARI MA(1,1,1)(8,1,7)
7. ARIMA(1,1,1)(8,1,12)
8. ARIMA(1,1,1)(24,1,5)
9. ARIMA(1,1,1)(24,1,20)
10.ARIMA(1,1,1)(24,1,23)

Bước 6:
1. ARIMA(1,1,1)

Dependent Variable: DCPI


Method: Least Squares
Date: 04/23/20 Time: 09:47
Sample (adjusted): 2000M03 2014M12
Included observations: 178 after adjustments
Convergence achieved after 21 iterations
MA Backcast: 2000M02

Variable Coefficient Std. Error t-Statistic Prob.  

C 0.164393 0.046724 3.518377 0.0006


AR(1) 0.312839 0.129290 2.419665 0.0166
MA(1) 0.303959 0.128597 2.363644 0.0192
R-squared 0.292018    Mean dependent var 0.167472
Adjusted R-squared 0.283927    S.D. dependent var 0.388508
S.E. of regression 0.328760    Akaike info criterion 0.629734
Sum squared resid 18.91457    Schwarz criterion 0.683359
Log likelihood -53.04631    Hannan-Quinn criter. 0.651480
F-statistic 36.09067    Durbin-Watson stat 1.961701
Prob(F-statistic) 0.000000

Inverted AR Roots       .31


Inverted MA Roots      -.30

DCPIt =0.1644+0.3128 AR(1)+0.3040 MA(1)+ût


CPIt=0.1644+1. 3128 CPIt-1-0.3128 CPIt-2+0.3040 ût-1+ ût

2. ARIMA(1,1,1)(2,1,5)

Dependent Variable: DCPI


Method: Least Squares
Date: 03/11/20 Time: 20:27
Sample (adjusted): 2000M04 2014M12
Included observations: 177 after adjustments
Convergence achieved after 13 iterations
MA Backcast: 1999M11 2000M03

Variable Coefficient Std. Error t-Statistic Prob.  

C 0.168544 0.025332 6.653386 0.0000


AR(1) 0.904940 0.231378 3.911085 0.0001
AR(2) -0.430664 0.116096 -3.709566 0.0003
MA(1) -0.312437 0.245291 -1.273742 0.2045
MA(5) -0.145433 0.093242 -1.559739 0.1207

R-squared 0.326955    Mean dependent var 0.164802


Adjusted R-squared 0.311303    S.D. dependent var 0.387970
S.E. of regression 0.321968    Akaike info criterion 0.599110
Sum squared resid 17.83005    Schwarz criterion 0.688832
Log likelihood -48.02122    Hannan-Quinn criter. 0.635498
F-statistic 20.88874    Durbin-Watson stat 1.947715
Prob(F-statistic) 0.000000

Inverted AR Roots  .45+.48i      .45-.48i


Inverted MA Roots       .76      .27+.63i    .27-.63i -.50-.39i
-.50+.39i

DCPIt = 0.1685+0.9049 AR(1)-0.4307 AR(2)-0.3124 MA(1)-0.1454 MA(5)+ ût


CPIt=0.1685+1.9049 CPIt-1-1.3356 CPIt-2+0.4307 CPIt-3-0.3124 ût-1-0.1454 ût-5+
ût

3.ARIMA(1,1,1)(5,1,6)

Dependent Variable: DCPI


Method: Least Squares
Date: 03/11/20 Time: 20:28
Sample (adjusted): 2000M07 2014M12
Included observations: 174 after adjustments
Convergence achieved after 30 iterations
MA Backcast: 2000M01 2000M06

Variable Coefficient Std. Error t-Statistic Prob.  

C 0.164526 0.039017 4.216747 0.0000


AR(1) 0.299720 0.133042 2.252813 0.0256
AR(5) -0.133979 0.077370 -1.731661 0.0852
MA(1) 0.310629 0.132978 2.335937 0.0207
MA(6) -0.003794 0.076463 -0.049622 0.9605

R-squared 0.309580    Mean dependent var 0.164483


Adjusted R-squared 0.293238    S.D. dependent var 0.390736
S.E. of regression 0.328488    Akaike info criterion 0.639683
Sum squared resid 18.23587    Schwarz criterion 0.730461
Log likelihood -50.65245    Hannan-Quinn criter. 0.676508
F-statistic 18.94461    Durbin-Watson stat 1.961293
Prob(F-statistic) 0.000000

Inverted AR Roots  .61-.38i      .61+.38i   -.15-.63i -.15+.63i


     -.62
Inverted MA Roots       .36      .16+.33i    .16-.33i -.25+.32i
-.25-.32i          -.47

DCPIt = 0.1645+0.2997 AR(1)-0.1340 AR(5)+0.3106 MA(1)-0.0038 MA(6)+ ût


CPIt=0.1645+1.2997 CPIt-1-0.2997CPIt-2-0.1340 CPIt-5+0.1340 CPIt-6 +0.3106 ût-
1-0..0038 ût-6+ ût

4.ARIMA(1,1,1)(5,1,9)
Dependent Variable: DCPI
Method: Least Squares
Date: 03/11/20 Time: 20:31
Sample (adjusted): 2000M07 2014M12
Included observations: 174 after adjustments
Convergence achieved after 14 iterations
MA Backcast: 1999M10 2000M06

Variable Coefficient Std. Error t-Statistic Prob.  

C 0.166769 0.032221 5.175793 0.0000


AR(1) 0.321220 0.123530 2.600345 0.0101
AR(5) -0.143335 0.074487 -1.924286 0.0560
MA(1) 0.256598 0.124453 2.061808 0.0408
MA(9) -0.189286 0.074677 -2.534727 0.0122

R-squared 0.327561    Mean dependent var 0.164483


Adjusted R-squared 0.311645    S.D. dependent var 0.390736
S.E. of regression 0.324183    Akaike info criterion 0.613295
Sum squared resid 17.76094    Schwarz criterion 0.704072
Log likelihood -48.35664    Hannan-Quinn criter. 0.650120
F-statistic 20.58095    Durbin-Watson stat 1.960113
Prob(F-statistic) 0.000000

Inverted AR Roots  .62-.39i      .62+.39i   -.15+.63i -.15-.63i


     -.62
Inverted MA Roots       .81      .61-.53i    .61+.53i  .12+.81i
 .12-.81i     -.45-.72i   -.45+.72i -.81+.28i
-.81-.28i

DCPIt = 0.1668+0.3212 AR(1)-0.1433 AR(5)+0.2566 MA(1)-0.1893 MA(9)+ ût


CPIt=0.1668+1.3212 CPIt-1-0.3212CPIt-2-0.1433 CPIt-5-0.1433 CPIt-6 +0.2566 ût-1-
0.1843 ût-6+ ût

5.ARIMA(1,1,1)(8,1,9)

Dependent Variable: DCPI


Method: Least Squares
Date: 03/11/20 Time: 20:33
Sample (adjusted): 2000M10 2014M12
Included observations: 171 after adjustments
Convergence achieved after 25 iterations
MA Backcast: 2000M01 2000M09

Variable Coefficient Std. Error t-Statistic Prob.  

C 0.165188 0.034940 4.727737 0.0000


AR(1) 0.303086 0.127838 2.370864 0.0189
AR(8) -0.138747 0.077236 -1.796394 0.0743
MA(1) 0.291363 0.126613 2.301207 0.0226
MA(9) -0.127918 0.077222 -1.656485 0.0995

R-squared 0.329093    Mean dependent var 0.163860


Adjusted R-squared 0.312927    S.D. dependent var 0.393511
S.E. of regression 0.326181    Akaike info criterion 0.626073
Sum squared resid 17.66138    Schwarz criterion 0.717935
Log likelihood -48.52928    Hannan-Quinn criter. 0.663347
F-statistic 20.35660    Durbin-Watson stat 1.944075
Prob(F-statistic) 0.000000

Inverted AR Roots  .77+.30i      .77-.30i    .34-.72i  .34+.72i


-.26-.72i     -.26+.72i   -.69+.30i -.69-.30i
Inverted MA Roots       .77      .58+.51i    .58-.51i  .11-.78i
 .11+.78i     -.43-.68i   -.43+.68i -.79+.27i

DCPIt = 0.1652+0.3031 AR(1)-0.1387 AR(8)+0.2914 MA(1)-0.1279 MA(9)+ ût


CPIt=0.1652+1.3031 CPIt-1-0.3031CPIt-2-0.1387 CPIt-8+0.1387CPIt-9 +0.214 ût-1-
0.1279 ût-9+ ût

6.ARIMA(1,1,1)(8,1,7)

Dependent Variable: DCPI


Method: Least Squares
Date: 03/11/20 Time: 20:33
Sample (adjusted): 2000M10 2014M12
Included observations: 171 after adjustments
Convergence achieved after 34 iterations
MA Backcast: 2000M03 2000M09

Variable Coefficient Std. Error t-Statistic Prob.  

C 0.165912 0.034312 4.835348 0.0000


AR(1) 0.293636 0.132323 2.219076 0.0278
AR(8) -0.188017 0.079558 -2.363265 0.0193
MA(1) 0.304245 0.133027 2.287097 0.0235
MA(7) -0.082206 0.078185 -1.051429 0.2946

R-squared 0.324545    Mean dependent var 0.163860


Adjusted R-squared 0.308269    S.D. dependent var 0.393511
S.E. of regression 0.327284    Akaike info criterion 0.632830
Sum squared resid 17.78111    Schwarz criterion 0.724691
Log likelihood -49.10696    Hannan-Quinn criter. 0.670103
F-statistic 19.94008    Durbin-Watson stat 1.966623
Prob(F-statistic) 0.000000

Inverted AR Roots  .79-.31i      .79+.31i    .35+.74i  .35-.74i


-.28+.75i     -.28-.75i   -.72-.31i -.72+.31i
Inverted MA Roots       .66      .40+.54i    .40-.54i -.20+.67i
-.20-.67i     -.68-.30i   -.68+.30i

DCPIt = 0.1659+0.2936 AR(1)-0.1880 AR(8)+0.3042 MA(1)-0.0822 MA(7)+ ût


CPIt=0.1659+1.2936 CPIt-1-0.2936CPIt-2-0.1880 CPIt-8+0.1880CPIt-9 +0.3042 ût-1-
0.0822 ût-7+ ût

7.ARIMA(1,1,1)(8,1,12)

Dependent Variable: DCPI


Method: Least Squares
Date: 03/11/20 Time: 20:34
Sample (adjusted): 2000M10 2014M12
Included observations: 171 after adjustments
Convergence achieved after 39 iterations
MA Backcast: 1999M10 2000M09

Variable Coefficient Std. Error t-Statistic Prob.  

C 0.163725 0.041344 3.960052 0.0001


AR(1) 0.317076 0.125803 2.520411 0.0127
AR(8) -0.152984 0.074674 -2.048689 0.0421
MA(1) 0.265637 0.127436 2.084477 0.0386
MA(12) 0.133641 0.076857 1.738825 0.0839

R-squared 0.334624    Mean dependent var 0.163860


Adjusted R-squared 0.318591    S.D. dependent var 0.393511
S.E. of regression 0.324834    Akaike info criterion 0.617796
Sum squared resid 17.51579    Schwarz criterion 0.709657
Log likelihood -47.82154    Hannan-Quinn criter. 0.655069
F-statistic 20.87076    Durbin-Watson stat 1.950811
Prob(F-statistic) 0.000000

Inverted AR Roots  .78-.30i      .78+.30i    .34+.72i  .34-.72i


-.27+.73i     -.27-.73i   -.70-.30i -.70+.30i
Inverted MA Roots  .80-.22i      .80+.22i    .58+.60i  .58-.60i
 .20+.81i      .20-.81i   -.24+.81i -.24-.81i
-.62-.60i     -.62+.60i   -.84-.22i -.84+.22i

DCPIt = 0.1637+0.3171 AR(1)-0.1530 AR(8)+0.2656 MA(1)+0.1336 MA(12)+ ût


CPIt=0.1637+1.3171 CPIt-1-0.3171CPIt-2-0.1530 CPIt-8+0.1530CPIt-9 +0.2656 ût-
1+0.1336 ût-12+ ût
8.ARIMA(1,1,1)(24,1,5)

Dependent Variable: DCPI


Method: Least Squares
Date: 03/11/20 Time: 20:35
Sample (adjusted): 2002M02 2014M12
Included observations: 155 after adjustments
Convergence achieved after 15 iterations
MA Backcast: 2001M09 2002M01

Variable Coefficient Std. Error t-Statistic Prob.  

C 0.162195 0.072077 2.250302 0.0259


AR(1) 0.299223 0.126892 2.358100 0.0197
AR(24) 0.255646 0.077338 3.305564 0.0012
MA(1) 0.317324 0.130248 2.436304 0.0160
MA(5) -0.096771 0.079296 -1.220384 0.2242

R-squared 0.368981    Mean dependent var 0.170710


Adjusted R-squared 0.352154    S.D. dependent var 0.405370
S.E. of regression 0.326278    Akaike info criterion 0.629591
Sum squared resid 15.96858    Schwarz criterion 0.727766
Log likelihood -43.79331    Hannan-Quinn criter. 0.669468
F-statistic 21.92769    Durbin-Watson stat 1.981164
Prob(F-statistic) 0.000000

Inverted AR Roots       .96      .93+.24i    .93-.24i  .83+.47i


 .83-.47i      .68+.67i    .68-.67i  .49+.82i
 .49-.82i      .26-.91i    .26+.91i  .01-.94i
 .01+.94i     -.23-.91i   -.23+.91i -.46-.82i
-.46+.82i     -.66-.67i   -.66+.67i -.81+.47i
-.81-.47i     -.90+.24i   -.90-.24i      -.93
Inverted MA Roots       .57      .14+.59i    .14-.59i -.58-.36i
-.58+.36i

DCPIt = 0.1622+0.2992 AR(1)+0.2556 AR(24)+0.3173 MA(1)-0.0967 MA(5)+ ût


CPIt=0.1622+1.2992 CPIt-1-0.2992CPIt-2+0.2556 CPIt-24-0.2556CPIt-25 +0.3173 ût-
1-0.0967 ût-5+ ût
9.ARIMA(1,1,1)(24,1,20)

Dependent Variable: DCPI


Method: Least Squares
Date: 03/11/20 Time: 20:36
Sample (adjusted): 2002M02 2014M12
Included observations: 155 after adjustments
Convergence achieved after 17 iterations
MA Backcast: 2000M06 2002M01

Variable Coefficient Std. Error t-Statistic Prob.  

C 0.161357 0.073973 2.181295 0.0307


AR(1) 0.318910 0.126299 2.525035 0.0126
AR(24) 0.265148 0.077764 3.409630 0.0008
MA(1) 0.290118 0.131116 2.212690 0.0284
MA(20) -0.132103 0.079662 -1.658287 0.0993

R-squared 0.372047    Mean dependent var 0.170710


Adjusted R-squared 0.355302    S.D. dependent var 0.405370
S.E. of regression 0.325484    Akaike info criterion 0.624720
Sum squared resid 15.89098    Schwarz criterion 0.722895
Log likelihood -43.41578    Hannan-Quinn criter. 0.664596
F-statistic 22.21789    Durbin-Watson stat 1.967940
Prob(F-statistic) 0.000000

Inverted AR Roots       .96      .93-.24i    .93+.24i  .83+.47i


 .83-.47i      .68+.67i    .68-.67i  .49-.82i
 .49+.82i      .26-.91i    .26+.91i  .01-.94i
 .01+.94i     -.23+.91i   -.23-.91i -.46+.82i
-.46-.82i     -.66+.67i   -.66-.67i -.81+.47i
-.81-.47i     -.90-.24i   -.90+.24i      -.93
Inverted MA Roots       .89      .85-.28i    .85+.28i  .72+.53i
 .72-.53i      .52+.73i    .52-.73i  .27-.86i
 .27+.86i     -.01+.90i   -.01-.90i -.29+.86i
-.29-.86i     -.55+.73i   -.55-.73i -.75-.53i
-.75+.53i     -.88+.28i   -.88-.28i      -.92
DCPIt = 0.1614+0.3189 AR(1)+0.2651 AR(24)+0.2901 MA(1)-0.1321 MA(20)+
ût
CPIt=0.1614+1.3189 CPIt-1-0.3189CPIt-2+0.2651 CPIt-24-0.2657CPIt-25 +0.2901ût-
1-0.1321 ût-20+ ût

10.ARIMA(1,1,1)(24,1,23)

Dependent Variable: DCPI


Method: Least Squares
Date: 03/11/20 Time: 20:36
Sample (adjusted): 2002M02 2014M12
Included observations: 155 after adjustments
Convergence achieved after 14 iterations
MA Backcast: 2000M03 2002M01

Variable Coefficient Std. Error t-Statistic Prob.  

C 0.159321 0.081867 1.946103 0.0535


AR(1) 0.312929 0.125134 2.500744 0.0135
AR(24) 0.262636 0.079383 3.308471 0.0012
MA(1) 0.301995 0.130038 2.322351 0.0216
MA(23) 0.019645 0.083652 0.234840 0.8147

R-squared 0.363697    Mean dependent var 0.170710


Adjusted R-squared 0.346729    S.D. dependent var 0.405370
S.E. of regression 0.327641    Akaike info criterion 0.637931
Sum squared resid 16.10231    Schwarz criterion 0.736106
Log likelihood -44.43962    Hannan-Quinn criter. 0.677807
F-statistic 21.43415    Durbin-Watson stat 1.989085
Prob(F-statistic) 0.000000

Inverted AR Roots       .96      .93+.24i    .93-.24i  .83-.47i


 .83+.47i      .68+.67i    .68-.67i  .49-.82i
 .49+.82i      .26-.91i    .26+.91i  .01+.94i
 .01-.94i     -.23-.91i   -.23+.91i -.46-.82i
-.46+.82i     -.66+.67i   -.66-.67i -.81-.47i
-.81+.47i     -.90+.24i   -.90-.24i      -.93
Inverted MA Roots  .82+.11i      .82-.11i    .76+.34i  .76-.34i
 .64+.53i      .64-.53i    .47-.69i  .47+.69i
 .27-.79i      .27+.79i    .04+.84i  .04-.84i
-.18+.82i     -.18-.82i   -.40+.75i -.40-.75i
-.59-.61i     -.59+.61i   -.74+.44i -.74-.44i
-.83-.23i     -.83+.23i        -.86

DCPIt = 0.1539+0.3129 AR(1)+0.2626 AR(24)+0.3020 MA(1)+0.0196 MA(23)+


ût
CPIt=0.1593+1.3129 CPIt-1-0.3129CPIt-2+0.2626 CPIt-24-0.2626CPIt-25 +0.3020 ût-
1+0.0196 ût-23+ ût

Bước 7: Từ 10 mô hình, chọn ra 3 mô hình tốt nhất


STT Mô hình R2 Adjusted AIC SBIC HQIC
R2
1 ARIMA(1,1,1) 0.2920 0.2839 0.6515*
0.6297 0.6834***
2 ARIMA(1,1,1)(2,1,5) 0.3270 0.3113 0.5991*** 0.6889** 0.6355***

3 ARIMA(1,1,1)(5,1,6) 0.3096 0.2932 0.6397 0.7305 0.6765

4 ARIMA(1,1,1)(5,1,9) 0.3276 0.3116 0.6133** 0.7041* 0.6501**

5 ARIMA(1,1,1)(8,1,9) 0.3291 0.3129 0.6261 0.7179 0.6633

6 ARI MA(1,1,1)(8,1,7) 0.3245 0.3083 0.6328 0.7247 0.6701

7 ARIMA(1,1,1)(8,1,12) 0.3346 0.3186 0.6178* 0.7097 0.6551

8 ARIMA(1,1,1)(24,1,5) 0.3690** 0.3521** 0.6296 0.7278 0.6695

9 ARIMA(1,1,1)(24,1,20) 0.3720*** 0.3553*** 0.6247 0.7229 0.6646

10 ARIMA(1,1,1)(24,1,23) 0.3637* 0.3467* 0.6379 0.7361 0.6778

Vậy chọn ba mô hình: ARIMA(1,1,1)(2,1,5), ARIMA(1,1,1)(5,1,9) và


ARIMA(1,1,1)(8,1,12)
VÌ : Ta ưu tiên lựa chọn mô hình có AIC,SBIC, HQIC càng nhỏ

Bước 8:
ARIMA(1,1,1)(2,1,5)

Inverse Roots of AR/MA Polynomial(s)


Specification: DCPI C AR(1) AR(2) MA(1) MA(5)
Date: 04/23/20 Time: 21:54
Sample: 2000M01 2014M12
Included observations: 177

AR Root(s) Modulus Cycle

  0.452470 ± 0.475327i  0.656250  7.756743

 No root lies outside the unit circle.


 ARMA model is stationary.

MA Root(s) Modulus Cycle

  0.756495  0.756495
  0.274209 ± 0.634830i  0.691520  5.402283
 -0.496238 ± 0.394674i  0.634050  2.544104

 No root lies outside the unit circle.


 ARMA model is invertible.

Mô hình AR dừng→ nghịch đảo nghiệm đặc trưng nằm trong vòng tròn đơn
vị
Trị tuyệt đối của nghịch đảo nghiệm đặc trưng mô hình nằm trong vòng tròn
đơn vị → mô hình MA nghịch đảo
Kl: Mô hình TỐT

ARIMA(1,1,1)(5,1,9)
Inverse Roots of AR/MA Polynomial(s)
Specification: DCPI C AR(1) AR(5) MA(1) MA(9)
Date: 04/23/20 Time: 21:55
Sample: 2000M01 2014M12
Included observations: 174

AR Root(s) Modulus Cycle

  0.623217 ± 0.388814i  0.734558  11.26434


 -0.150595 ± 0.634835i  0.652452  3.483478
 -0.624025  0.624025

 No root lies outside the unit circle.


 ARMA model is stationary.
MA Root(s) Modulus Cycle

 -0.813747 ± 0.282420i  0.861362  2.237970


 -0.445619 ± 0.715884i  0.843248  2.953212
  0.117022 ± 0.814963i  0.823322  4.399438
  0.611037 ± 0.532272i  0.810358  8.767872
  0.806015  0.806015

 No root lies outside the unit circle.


 ARMA model is invertible.

Mô hình AR dừng→ nghịch đảo nghiệm đặc trưng nằm trong vòng tròn đơn
vị
Trị tuyệt đối của nghịch đảo nghiệm đặc trưng mô hình nằm trong vòng tròn
đơn vị → mô hình MA nghịch đảo
Kl: Mô hình TỐT

ARIMA(1,1,1)(8,1,12)

Inverse Roots of AR/MA Polynomial(s)


Specification: DCPI C AR(1) AR(8) MA(1) MA(12)
Date: 04/23/20 Time: 21:57
Sample: 2000M01 2014M12
Included observations: 171

AR Root(s) Modulus Cycle

  0.777774 ± 0.298661i  0.833145  17.13755


  0.343694 ± 0.723386i  0.800883  5.573898
 -0.266394 ± 0.725248i  0.772625  3.267707
 -0.696537 ± 0.300746i  0.758691  2.298167

 No root lies outside the unit circle.


 ARMA model is stationary.

MA Root(s) Modulus Cycle

 -0.842561 ± 0.217658i  0.870221  2.175023


 -0.622220 ± 0.595045i  0.860951  2.641642
 -0.241275 ± 0.813512i  0.848537  3.379661
  0.197947 ± 0.814036i  0.837757  4.716189
  0.577986 ± 0.596156i  0.830344  7.845428
  0.797305 ± 0.218250i  0.826637  23.51587

 No root lies outside the unit circle.


 ARMA model is invertible.

Mô hình AR dừng→ nghịch đảo nghiệm đặc trưng nằm trong vòng tròn đơn
vị
Trị tuyệt đối của nghịch đảo nghiệm đặc trưng mô hình nằm trong vòng tròn
đơn vị → mô hình MA nghịch đảo
Kl: Mô hình TỐT
Vậy có 3 mô hình tốt

Bước 9: Lựa chọn mô hình dựa trên tiêu chí đô ̣ chính xác của dự báo

Mô hình RMSE MAE MAPE


ARIMA(1,1,1)(2,1,5) 0.3883*** 0.2881*** 131.1842**
ARIMA(1,1,1)(5,1,9) 0.3893** 0.2872** 125.6214***

ARIMA(1,1,1)(8,1,12) 0.3923* 0.2906* 126.3596**

Vạy lựa chọn mô hình ARIMA(1,1,1)(2,1,5) là mô hình phù hợp nhất


VÌ:

File C2 asm.wf1 gồm 10 bước

Bước 1:
Cách 1:
CLOSEASM
45

40

35

30

25

20

15

10

5
I II III IV I II III IV I II III IV I II III IV I II III IV I II III IV
2010 2011 2012 2013 2014 2015

CLOSEASM là chuỗi không dừng. Vì:


Khi nhìn chia đồ thị thành ba phần bằng nhau, nhìn bằng mắt thường ta thấy trung
bình của chúng thay đổi, phương sai cũng thay đổi vì sự biến động quá lớn=> vi
phạm điều kiện chuỗi dừng
Cách 2:
Null Hypothesis: CLOSEASM has a unit root
Exogenous: Constant
Lag Length: 2 (Automatic - based on SIC, maxlag=23)

t-Statistic   Prob.*

Augmented Dickey-Fuller test statistic -1.656788  0.4531


Test critical values: 1% level -3.434818
5% level -2.863401
10% level -2.567809

*MacKinnon (1996) one-sided p-values.

H0: CLOSEASMt có 1 nghiê ̣m đơn vị ( CLOSEASMt không dừng)

H1: CLOSEASMt không có nghiê ̣m đơn vị (CLOSEASMt dừng)

p-value = 0.4531 > α = 0.05 = 5%: chấp nhâ ̣n giả thiết H0

Vâ ̣y CLOSEASMt có 1 nghiê ̣m đơn vị ( CLOSEASMt không dừng) với mức ý
nghĩa α = 5%
Bước 2:
Tìm sai phân CLOSEASM bằng cách lấy CLOSEASM vào năm n trừ cho
CLOSEASM vào năm n-1 ( năm trước đó ). DCLOSEASM=CLOSEASM-
CLOSEASM(-1).
DCLOSEASM
2

-1

-2

-3
I II III IV I II III IV I II III IV I II III IV I II III IV I II III IV
2010 2011 2012 2013 2014 2015

Quan sát hình , ta thấy chuỗi DCLOSEASM là chuỗi dừng vì:


- Phần lớn các biến động nằm trong trường từ -2=>2, chuỗi biến động trong
biên. Độ biến thiên tương đối ổn định. Giá trị trung bình không thay đổi
nhiều. Có một vài sự thay đổi nhưng không đáng kể
Cách 2:
Null Hypothesis: DCLOSEASM has a unit root
Exogenous: Constant
Lag Length: 1 (Automatic - based on SIC, maxlag=23)

t-Statistic   Prob.*

Augmented Dickey-Fuller test statistic -21.65076  0.0000


Test critical values: 1% level -3.434818
5% level -2.863401
10% level -2.567809

*MacKinnon (1996) one-sided p-values.

H0: DCLOSEASMt có 1 nghiê ̣m đơn vị ( DCLOSEASMt không dừng)

H1: DCLOSEASMt không có nghiê ̣m đơn vị (DCLOSEASMt dừng)

p-value = 0.00 < α = 0.05 = 5%: Bác bỏ giả thiết H0


Vâ ̣y DCLOSEASMt không có nghiê ̣m đơn vị ( DCLOSEASMt dừng) với mức ý
nghĩa α = 5%

Bước 4:
Thực hiện bậc mô hình ARIMA(p,d,q) trên chuỗi dừng.
Bước 1: CLOSEASM là chuỗi không dừng và bước 2 thì DCLOSEASM là chuỗi
dừng, thực hiện mô hình ARIMA(p,d,q).

Ta có:
-MA(q) tương ứng cột ACF (AutoCorrelation):
-AR(p) tương ứng cột PACF (partial): (nằm ngoài đường biên gạch chấm)
p = 1,2,3
d=1
q= 1,2,3,4,5,8,18,28,29
Bước 5: Ghéo mô hình ARIMA(p,d,q)
1. ARIMA(1,1,1)
2. ARIMA(1,1,1)(2,1,3)
3. ARIMA(1,1,1)(2,1,8)
4. ARIMA(1,1,1)(2,1,28)
5. ARIMA(1,1,1)(3,1,2)
6. ARIMA(1,1,1)(3,1,4)
7. ARIMA(1,1,1)(3,1,29)
8. ARIMA(1,1,1)(2,1,18)
9. ARIMA(1,1,1)(3,1,5)
10.ARIMA(1,1,1)(3,1,3)

Bước 6:
1. ARIMA(1,1,1)

Dependent Variable: DCLOSEASM


Method: Least Squares
Date: 04/19/20 Time: 22:00
Sample (adjusted): 1/20/2010 11/03/2015
Included observations: 1399 after adjustments
Convergence achieved after 7 iterations
MA Backcast: 1/19/2010

Variable Coefficient Std. Error t-Statistic Prob.  

C -0.002052 0.018651 -0.110024 0.9124


AR(1) 0.605332 0.081600 7.418296 0.0000
MA(1) -0.410381 0.093533 -4.387543 0.0000

R-squared 0.056707    Mean dependent var -0.001731


Adjusted R-squared 0.055355    S.D. dependent var 0.480156
S.E. of regression 0.466677    Akaike info criterion 1.315782
Sum squared resid 304.0309    Schwarz criterion 1.327026
Log likelihood -917.3898    Hannan-Quinn criter. 1.319986
F-statistic 41.96077    Durbin-Watson stat 1.997368
Prob(F-statistic) 0.000000
Inverted AR Roots       .61
Inverted MA Roots       .41

DCLOSEASMt=-0.0021+0.6053 AR(1)- 0.4104 MA(1)+ ût


CLOSEASMt=-0.0021+1.6053CLOSEASMt-1-0.6053CLOSEASMt-2-0.4104 ût-1
+ût
2. ARIMA(1,1,1)(2,1,3)

Dependent Variable: DCLOSEASM


Method: Least Squares
Date: 04/19/20 Time: 22:02
Sample (adjusted): 1/21/2010 11/03/2015
Included observations: 1398 after adjustments
Convergence achieved after 15 iterations
MA Backcast: 1/18/2010 1/20/2010

Coefficien
Variable t Std. Error t-Statistic Prob.  

C -0.001485 0.019287 -0.077014 0.9386


AR(1) 0.660800 0.201343 3.281959 0.0011
AR(2) -0.035391 0.053810 -0.657703 0.5108
MA(1) -0.465360 0.198928 -2.339344 0.0195
MA(3) 0.044110 0.034868 1.265064 0.2061

R-squared 0.057937    Mean dependent var -0.002240


Adjusted R-squared 0.055232    S.D. dependent var 0.479950
S.E. of regression 0.466507    Akaike info criterion 1.316484
Sum squared resid 303.1572    Schwarz criterion 1.335235
Log likelihood -915.2220    Hannan-Quinn criter. 1.323494
F-statistic 21.41743    Durbin-Watson stat 1.995513
Prob(F-statistic) 0.000000

Inverted AR Roots       .60           .06


Inverted MA Roots  .36-.22i      .36+.22i        -.25

DCLOSEASMt=-0.0015+0.6608 AR(1) -0.0354 AR(2) -0.4654 MA(1)+


0.0441MA(3)+ ût
CLOSEASMt=-0.0015+1.6608CLOSEASMt-1-0.6962CLOSEASMt-2+0.0354
CLOSEASMt-3 -0.4654ût-1 +0.0441 ût-3+ût

3. ARIMA(1,1,1)(2,1,8)
Dependent Variable: DCLOSEASM
Method: Least Squares
Date: 04/19/20 Time: 22:03
Sample (adjusted): 1/21/2010 11/03/2015
Included observations: 1398 after adjustments
Convergence achieved after 18 iterations
MA Backcast: 1/11/2010 1/20/2010

Variable Coefficient Std. Error t-Statistic Prob.  

C -0.000481 0.016877 -0.028516 0.9773

AR(1) 0.827040 0.148640 5.564036 0.0000

AR(2) -0.058624 0.049192 -1.191754 0.2336

MA(1) -0.634149 0.145371 -4.362293 0.0000

MA(8) -0.053018 0.020844 -2.543486 0.0111

R-squared 0.060619    Mean dependent var -0.002240

Adjusted R-squared 0.057922    S.D. dependent var 0.479950

S.E. of regression 0.465843    Akaike info criterion 1.313633

Sum squared resid 302.2941    Schwarz criterion 1.332384

Log likelihood -913.2292    Hannan-Quinn criter. 1.320643

F-statistic 22.47285    Durbin-Watson stat 1.995716

Prob(F-statistic) 0.000000

Inverted AR Roots       .75           .08

Inverted MA Roots       .83      .58+.45i    .58-.45i  .07+.67i

 .07-.67i     -.43-.48i   -.43+.48i      -.64

DCLOSEASMt=-0.0005+0.8270 AR(1) -0.0586 AR(2) -0.6341MA(1) -0.0530


MA(8) +ût
CLOSEASMt=-0.0005+1.8270CLOSEASMt-1-0.8856CLOSEASMt-2+0.0586
CLOSEASMt-3 -0.6341ût-1 -0.0530ût-8+ût

4. ARIMA(1,1,1)(2,1,28)
Dependent Variable: DCLOSEASM
Method: Least Squares
Date: 04/19/20 Time: 22:04
Sample (adjusted): 1/21/2010 11/03/2015
Included observations: 1398 after adjustments
Convergence achieved after 14 iterations
MA Backcast: 12/14/2009 1/20/2010

Variable Coefficient Std. Error t-Statistic Prob.  

C -0.000618 0.020891 -0.029572 0.9764


AR(1) 0.753754 0.161124 4.678097 0.0000
AR(2) -0.040128 0.051343 -0.781560 0.4346
MA(1) -0.560797 0.158034 -3.548573 0.0004
MA(28) 0.040867 0.022744 1.796802 0.0726

R-squared 0.059201    Mean dependent var -0.002240


Adjusted R-squared 0.056499    S.D. dependent var 0.479950
S.E. of regression 0.466194    Akaike info criterion 1.315141
Sum squared resid 302.7505    Schwarz criterion 1.333892
Log likelihood -914.2837    Hannan-Quinn criter. 1.322151
F-statistic 21.91401    Durbin-Watson stat 1.994783
Prob(F-statistic) 0.000000

Inverted AR Roots       .70           .06


Inverted MA Roots  .92+.10i      .92-.10i    .87-.29i  .87+.29i
 .78-.47i      .78+.47i    .65-.62i  .65+.62i
 .50+.75i      .50-.75i    .31+.84i  .31-.84i
 .12+.88i      .12-.88i   -.08+.88i -.08-.88i
-.28+.84i     -.28-.84i   -.46+.75i -.46-.75i
-.61+.63i     -.61-.63i   -.74-.47i -.74+.47i
-.83-.29i     -.83+.29i   -.87-.10i -.87+.10i

DCLOSEASMt=-0.0006+0.7538 AR(1) -0.0401 AR(2) -0.5608 MA(1) +0.0409


MA(28) +ût
CLOSEASMt=-0.0006+1.7538CLOSEASMt-1-0.7939CLOSEASMt-2+0.0401
CLOSEASMt-3 -0.5608ût-1 +0.0409ût-28+ût

5. ARIMA(1,1,1)(3,1,2)

Dependent Variable: DCLOSEASM


Method: Least Squares
Date: 04/19/20 Time: 22:12
Sample (adjusted): 1/22/2010 11/03/2015
Included observations: 1397 after adjustments
Convergence achieved after 15 iterations
MA Backcast: 1/20/2010 1/21/2010

Variable Coefficient Std. Error t-Statistic Prob.  


C -0.000522 0.019172 -0.027209 0.9783
AR(1) 0.529235 0.205319 2.577630 0.0101
AR(3) 0.036611 0.040037 0.914420 0.3607
MA(1) -0.336232 0.207061 -1.623826 0.1046
MA(2) 0.003726 0.047138 0.079038 0.9370

R-squared 0.059929    Mean dependent var -0.001664


Adjusted R-squared 0.057228    S.D. dependent var 0.479638
S.E. of regression 0.465712    Akaike info criterion 1.313074
Sum squared resid 301.9075    Schwarz criterion 1.331836
Log likelihood -912.1819    Hannan-Quinn criter. 1.320088
F-statistic 22.18484    Durbin-Watson stat 1.997958
Prob(F-statistic) 0.000000

Inverted AR Roots       .62     -.05+.24i   -.05-.24i


Inverted MA Roots       .32           .01

DCLOSEASMt=-0.0005+0.5292 AR(1)+0.0366 AR(3)-0.3362MA(1)+0.0037


MA(2) +ût
CLOSEASMt=-0.0005+1.5292CLOSEASMt-1-0.5292CLOSEASMt-2+0.0366
CLOSEASMt-3-0.3362 CLOSEASMt-4 -0.3362ût-1 +0.0037ût-2+ût

6. ARIMA(1,1,1)(3,1,4)

Dependent Variable: DCLOSEASM


Method: Least Squares
Date: 04/19/20 Time: 22:07
Sample (adjusted): 1/22/2010 11/03/2015
Included observations: 1397 after adjustments
Convergence achieved after 10 iterations
MA Backcast: 1/18/2010 1/21/2010

Variable Coefficient Std. Error t-Statistic Prob.  

C -0.000553 0.019266 -0.028718 0.9771


AR(1) 0.525976 0.125488 4.191440 0.0000
AR(3) 0.033112 0.033910 0.976464 0.3290
MA(1) -0.331208 0.134801 -2.457008 0.0141
MA(4) 0.012662 0.030652 0.413079 0.6796

R-squared 0.060027    Mean dependent var -0.001664


Adjusted R-squared 0.057326    S.D. dependent var 0.479638
S.E. of regression 0.465688    Akaike info criterion 1.312970
Sum squared resid 301.8761    Schwarz criterion 1.331732
Log likelihood -912.1093    Hannan-Quinn criter. 1.319984
F-statistic 22.22330    Durbin-Watson stat 2.001732
Prob(F-statistic) 0.000000

Inverted AR Roots       .61     -.04+.23i   -.04-.23i


Inverted MA Roots  .34+.20i      .34-.20i   -.17-.22i -.17+.22i

DCLOSEASMt=-0.0006+0.5260 AR(1)+0.0331 AR(3)-0.3312 MA(1)+0.0127


MA(4) +ût
CLOSEASMt=-0.0006+1.5260CLOSEASMt-1-0.5260CLOSEASMt-2+0.0331
CLOSEASMt-3-0.0331 CLOSEASMt-4 -0.3312ût-1 +0.0127ût-4+ût

7. ARIMA(1,1,1)(3,1,29)

Dependent Variable: DCLOSEASM


Method: Least Squares
Date: 04/19/20 Time: 22:08
Sample (adjusted): 1/22/2010 11/03/2015
Included observations: 1397 after adjustments
Convergence achieved after 8 iterations
MA Backcast: 12/14/2009 1/21/2010

Variable Coefficient Std. Error t-Statistic Prob.  

C -0.000296 0.020225 -0.014633 0.9883


AR(1) 0.531080 0.119225 4.454430 0.0000
AR(3) 0.037130 0.033148 1.120126 0.2629
MA(1) -0.338873 0.128094 -2.645505 0.0082
MA(29) 0.040865 0.025410 1.608223 0.1080

R-squared 0.061751    Mean dependent var -0.001664


Adjusted R-squared 0.059055    S.D. dependent var 0.479638
S.E. of regression 0.465260    Akaike info criterion 1.311134
Sum squared resid 301.3225    Schwarz criterion 1.329896
Log likelihood -910.8270    Hannan-Quinn criter. 1.318148
F-statistic 22.90359    Durbin-Watson stat 2.000351
Prob(F-statistic) 0.000000

Inverted AR Roots       .63     -.05+.24i   -.05-.24i


Inverted MA Roots  .90+.10i      .90-.10i    .86+.28i  .86-.28i
 .78-.46i      .78+.46i    .66+.61i  .66-.61i
 .52-.74i      .52+.74i    .34-.83i  .34+.83i
 .16-.88i      .16+.88i   -.04+.89i -.04-.89i
-.23-.86i     -.23+.86i   -.41-.79i -.41+.79i
-.57-.68i     -.57+.68i   -.70+.54i -.70-.54i
-.80-.38i     -.80+.38i   -.86-.19i -.86+.19i
     -.89

DCLOSEASM=-0.0003+0.5311 AR(1)+0.0371 AR(3)-0.3389


MA(1)+0.0409 MA(29) +ût

CLOSEASMt=-0.0003+1.5311CLOSEASMt-1-0.5311CLOSEASMt-
2+0.0371 CLOSEASMt-3-0.0371 CLOSEASMt-4 -0.3389ût-1 +0.0409ût-
29+ût

8. ARIMA(1,1,1)(2,1,18)
Dependent Variable: DCLOSEASM
Method: Least Squares
Date: 04/19/20 Time: 22:09
Sample (adjusted): 1/21/2010 11/03/2015
Included observations: 1398 after adjustments
Convergence achieved after 44 iterations
MA Backcast: 12/28/2009 1/20/2010

Variable Coefficient Std. Error t-Statistic Prob.  

C -0.001009 0.017545 -0.057535 0.9541


AR(1) 0.729095 0.170449 4.277486 0.0000
AR(2) -0.033932 0.053195 -0.637878 0.5237
MA(1) -0.534175 0.167632 -3.186598 0.0015
MA(18) -0.037910 0.023249 -1.630651 0.1032

R-squared 0.058953    Mean dependent var -0.002240


Adjusted R-squared 0.056250    S.D. dependent var 0.479950
S.E. of regression 0.466256    Akaike info criterion 1.315405
Sum squared resid 302.8303    Schwarz criterion 1.334156
Log likelihood -914.4680    Hannan-Quinn criter. 1.322415
F-statistic 21.81642    Durbin-Watson stat 1.994414
Prob(F-statistic) 0.000000

Inverted AR Roots       .68           .05


Inverted MA Roots       .88      .82+.28i    .82-.28i  .67+.53i
 .67-.53i      .45-.71i    .45+.71i  .17-.81i
 .17+.81i     -.12+.81i   -.12-.81i -.39+.72i
-.39-.72i     -.61+.53i   -.61-.53i -.76+.28i
-.76-.28i          -.81

DCLOSEASMt=-0.0010+0.7291 AR(1)-0.0339AR(2)-0.5342 MA(1)


-0.0379 MA(18) +ût

CLOSEASMt=-0.0010+1.7291 CLOSEASMt-1-0.763CLOSEASMt-2+0.0339
CLOSEASMt-3 -0.5342ût-1 -0.0379ût-18+ût

9. ARIMA(1,1,1)(3,1,5)

Dependent Variable: DCLOSEASM


Method: Least Squares
Date: 04/19/20 Time: 22:10
Sample (adjusted): 1/22/2010 11/03/2015
Included observations: 1397 after adjustments
Convergence achieved after 11 iterations
MA Backcast: 1/15/2010 1/21/2010

Variable Coefficient Std. Error t-Statistic Prob.  

C -0.000500 0.019199 -0.026061 0.9792


AR(1) 0.535301 0.126937 4.217065 0.0000
AR(3) 0.035765 0.033575 1.065215 0.2870
MA(1) -0.341150 0.136575 -2.497893 0.0126
MA(5) 0.001697 0.027465 0.061795 0.9507

R-squared 0.059924    Mean dependent var -0.001664


Adjusted R-squared 0.057223    S.D. dependent var 0.479638
S.E. of regression 0.465713    Akaike info criterion 1.313079
Sum squared resid 301.9091    Schwarz criterion 1.331841
Log likelihood -912.1855    Hannan-Quinn criter. 1.320093
F-statistic 22.18292    Durbin-Watson stat 2.000056
Prob(F-statistic) 0.000000

Inverted AR Roots       .63     -.05+.23i   -.05-.23i


Inverted MA Roots  .32+.12i      .32-.12i   -.04+.25i -.04-.25i
     -.23

DCLOSEASMt=-0.0005+0.5353AR(1)+0.0358AR(3)-0.3412MA(1)+0.0017
MA(5) +ût
CLOSEASMt=-0.0005+1.5353CLOSEASMt-1-0.5353CLOSEASMt-
2+0.0358CLOSEASMt-3-0.0358CLOSEASMt-4-0.3412ût-1 +0.0017ût-5+ût

10.ARIMA(1,1,1)(3,1,3)

Dependent Variable: DCLOSEASM


Method: Least Squares
Date: 04/19/20 Time: 22:10
Sample (adjusted): 1/22/2010 11/03/2015
Included observations: 1397 after adjustments
Convergence achieved after 14 iterations
MA Backcast: 1/19/2010 1/21/2010

Variable Coefficient Std. Error t-Statistic Prob.  

C -0.000471 0.018670 -0.025209 0.9799


AR(1) 0.621698 0.134144 4.634543 0.0000
AR(3) -0.070488 0.090814 -0.776179 0.4378
MA(1) -0.439536 0.144840 -3.034626 0.0025
MA(3) 0.112138 0.088162 1.271954 0.2036

R-squared 0.060945    Mean dependent var -0.001664


Adjusted R-squared 0.058247    S.D. dependent var 0.479638
S.E. of regression 0.465460    Akaike info criterion 1.311992
Sum squared resid 301.5811    Schwarz criterion 1.330754
Log likelihood -911.4263    Hannan-Quinn criter. 1.319006
F-statistic 22.58551    Durbin-Watson stat 1.978577
Prob(F-statistic) 0.000000

Inverted AR Roots  .45+.22i      .45-.22i        -.28

DCLOSEASMt= -0.0005+0.6217 AR(1)-0.0705 AR(3)-0.4395 MA(1)+0.1121 MA(3)+


ût
CLOSEASMt= -0.0005+1.6217CLOSEASMt-1-0.6217CLOSEASMt-2-
0.0705CLOSEASMt-3+0.0705CLOSEASMt-4-0.4395ût-1 +0.1121ût-3+ût
Bước 7: Từ 10 mô hình, chọn ra 3 mô hình tốt nhất

STT Mô hình R2 Adjusted R2 AIC SBIC HQIC


1 ARIMA(1,1,1) 0.0567 0.0554 1.3152 1.3270*** 1.3200*

2 ARIMA(1,1,1)(2,1,3) 0.0579 0.0552 1.3165 1.3352 1.3235

3 ARIMA(1,1,1)(2,1,8) 0.0606* 0.0579* 1.3136 1.3324 1.3206

4 ARIMA(1,1,1)(2,1,28) 0.0592 0.0565 1.3151 1.3340 1.3222

5 ARIMA(1,1,1)(3,1,2) 0.0599 0.0572 1.3131* 1.3318 1.3201

6 ARIMA(1,1,1)(3,1,4) 0.0600 0.0573 1.3130 1.3317 1.3200


7 ARIMA(1,1,1)(3,1,29) 0.0618*** 0.0591*** 1.3111*** 1.3299** 1.3182***

8 ARIMA(1,1,1)(2,1,18) 0.0590 0.0563 1.3154 1.3341 1.3224

9 ARIMA(1,1,1)(3,1,5) 0.0599 0.0572 1.3131* 1.3318 1.3201

10 ARIMA(1,1,1)(3,1,3) 0.0609** 0.0582** 1.3120** 1.3308* 1.3190**

Vậy chọn 3 mô hình: -ARIMA(1,1,1), ARIMA(1,1,1)(3,1,29) và ARIMA(1,1,1)


(3,1,3)
Vì ta thấy đáp ứng được tiêu chí R2 lớn và AIC,SBIC,HQIC NHỎ

Bước 8. Kiểm định tính dừng và nghịch đảo của mô hình


Mô hình ARIMA(1,1,1)(2,1,8)

Inverse Roots of AR/MA Polynomial(s)


Specification: DCLOSEASM C AR(1) AR(2) MA(1) MA(8)
Date: 04/23/20 Time: 21:49
Sample: 1/18/2010 11/03/2015
Included observations: 1398

AR Root(s) Modulus Cycle

  0.748743  0.748743
  0.078297  0.078297

 No root lies outside the unit circle.


 ARMA model is stationary.

MA Root(s) Modulus Cycle

  0.829821  0.829821
  0.582915 ± 0.449597i  0.736157  9.563609
  0.067332 ± 0.666584i  0.669976  4.273907
 -0.430438 ± 0.477655i  0.642986  2.726787
 -0.635291  0.635291

 No root lies outside the unit circle.


 ARMA model is invertible.
Mô hình AR dừng- nghịch đảo nghiệm đặc trưng nằm trong vòng tròn đơn
vị
Trị tuyệt đối của nghịch đảo nghiệm đặc trưng mô hình nằm trong vòng tròn
đơn vị -- mô hình MA nghịch đảo
Kl: Mô hình phù hợp

Mô hình ARIMA(1,1,1)(3,1,29)

Inverse Roots of AR/MA Polynomial(s)


Specification: DCLOSEASM C AR(1) AR(3) MA(1)
        MA(29)
Date: 04/23/20 Time: 21:41
Sample: 1/18/2010 11/03/2015
Included observations: 1397

AR Root(s) Modulus Cycle

  0.625870  0.625870
 -0.047395 ± 0.238914i  0.243569  3.556593

 No root lies outside the unit circle.


 ARMA model is stationary.

MA Root(s) Modulus Cycle

  0.904817 ± 0.096429i  0.909940  59.17909


  0.862908 ± 0.284828i  0.908701  19.70772
  0.781113 ± 0.460026i  0.906511  11.80507
  0.663339 ± 0.613844i  0.903782  8.415006
  0.515151 ± 0.739055i  0.900879  6.530970
  0.343503 ± 0.829757i  0.898048  5.332431
  0.156420 ± 0.881666i  0.895434  4.503401
 -0.037365 ± 0.892325i  0.893107  3.896197
 -0.228813 ± 0.861218i  0.891096  3.432531
 -0.408992 ± 0.789794i  0.889409  3.067035
 -0.569497 ± 0.681391i  0.888043  2.771611
 -0.702837 ± 0.541084i  0.886991  2.527933
 -0.802791 ± 0.375443i  0.886245  2.323542
 -0.864693 ± 0.192221i  0.885801  2.149678
 -0.885653  0.885653
 No root lies outside the unit circle.
 ARMA model is invertible.

Mô hình AR dừng- nghịch đảo nghiệm đặc trưng nằm trong vòng tròn đơn vị
Trị tuyệt đối của nghịch đảo nghiệm đặc trưng mô hình nằm trong vòng tròn đơn vị
-- mô hình MA nghịch đảo
Kl: Mô hình TỐT

ARIMA(1,1,1)(3,1,3)

Inverse Roots of AR/MA Polynomial(s)


Specification: DCLOSEASM C AR(1) AR(3) MA(1) MA(3)
Date: 04/23/20 Time: 21:46
Sample: 1/18/2010 11/03/2015
Included observations: 1397

AR Root(s) Modulus Cycle

  0.450673 ± 0.221253i  0.502055  13.76763


 -0.279647  0.279647

 No root lies outside the unit circle.


 ARMA model is stationary.

MA Root(s) Modulus Cycle

  0.405656 ± 0.370229i  0.549206  8.493430


 -0.371776  0.371776

 No root lies outside the unit circle.


 ARMA model is invertible.

Mô hình AR dừng→ nghịch đảo nghiệm đặc trưng nằm trong vòng tròn đơn
vị
Trị tuyệt đối của nghịch đảo nghiệm đặc trưng mô hình nằm trong vòng tròn
đơn vị → mô hình MA nghịch đảo
Kl: Mô hình TỐT
Vâ ̣y có 3 mô hình phù hợp
Bước 9: Lựa chọn mô hình dựa trên tiêu chí đô ̣ chính xác của dự báo
Mô hình RMSE MAE MAPE
ARIMA(1,1,1)(2,1,8) 0.4798 * 0.3307* 82.3138*
ARIMA(1,1,1)(3,1,3) 0.4789 *** 0.3296*** 82.1472***
ARIMA(1,1,1) 0.4790 ** 0.3298** 82.2190**
(3,1,29)
KL: Mô hình phù hợp nhất là ARIMA(1,1,1)(3,1,3)
Lý do: vì có ba chỉ tiêu RMSE, MAE, MAPE đều nhỏ nhất trong ba mô hình
Bước 10: Dự báo mô hình ARIMA(1,1,1)(3,1,3)
DCLOSEASM 11/04/2015 max=1.03
DCLOSEASM 11/04/2015=0.09
DCLOSEASM 11/04/2015 min=-0.84
Ta có: DYt= Yt – Yt-1 => Yt = DYt + Yt-1
Vâ ̣y:
CLOSEASM 11/04/2015 max = 1.03 + 13 = 14.03
CLOSEASM 11/04/2015 = 0.09 + 13 = 13.09
CLOSEASM 11/04/2015 min = -0.82 + 13 = 12.18

ĐỐI VỚI Vnindex


Bước 1:
Cách 1: DỰA TRÊN BIỂU ĐỒ GRAPH
Quan sát hình, ta thấy chuỗi VNINDEX là chuỗi dừng vì:
Tất cả biến động nằm trong trường từ -5 đến 5=> chuỗi biến động trong
biên. Độ biến thiên tương đối ổn định. Giá trị trung bình không thay đổi
nhiều => VNINDEX là chuỗi dừng.
CÁCH 2: Dựa trên kiểm định nghiệm đơn vị ( unit root test )của Dickey
Fuller

Null Hypothesis: VNINDEX has a unit root


Exogenous: Constant
Lag Length: 3 (Automatic - based on SIC, maxlag=23)

t-Statistic   Prob.*

Augmented Dickey-Fuller test statistic -15.86468  0.0000


Test critical values: 1% level -3.434526
5% level -2.863271
10% level -2.567740

*MacKinnon (1996) one-sided p-values.

H0: VNINDEXt có một nghiệm đơn vị (VNINDEXt không dừng)


H1: VNINDEXt không có nghiệm đơn vị : (VNINDEXt dừng)
p-value < α : bác bỏ giải thiết H0 với mức ý nghĩa α=1%,5% hay 10%
Ta có : p-value=0.0000< α=0.05=5%: Bác bỏ giả thiết H0
Vậy VNINDEXt không có nghiệm đơn vị(VNINDEXt dừng)
Vì VNINDEX dừng nên làm tiếp bước 4
Bước 4:
TH1: Nếu Bước 1 VNINDEX là chuỗi dừng ,mô hình sẽ là ARMA(p,q)
Ta có:
- MA(q) tương ứng cột ACF (AutoCorrelation):
- AR(p) tương ứng cột PACF (partial): (nằm ngoài đường biên gạch
chấm)

p = 1,2,4,5,13,22
q = 1,4,5,12,13,14,18,19,21,22,23,28,32
Bước 5: Ghép mô hình ARMA (p,q)
1. ARMA(1,1)(5,4)
2. ARMA(1,1) (2,4)
3. ARMA(1,1) (4,4)
4. ARMA(1,1) (5,4)
5. ARMA(1,1) (5,12)
6. ARMA(1,1) (5,13)
7. ARMA(1,1) (13,12)
8. ARMA(1,1) (5,1)
9. ARMA(1,1) (13,14)
10. ARMA(1,1) (13,5)
BƯỚC 6:
1. ARMA(1,1)(5,4)

Dependent Variable: VNINDEX


Method: Least Squares
Date: 04/26/20 Time: 09:26
Sample (adjusted): 1/10/2006 12/30/2011
Included observations: 1491 after adjustments
Convergence achieved after 7 iterations
MA Backcast: 1/04/2006 1/09/2006

Variable Coefficient Std. Error t-Statistic Prob.  

C 0.009372 0.074636 0.125574 0.9001


AR(1) 0.106956 0.086394 1.237994 0.2159
AR(5) 0.095852 0.027470 3.489389 0.0005
MA(1) 0.191842 0.085607 2.240979 0.0252
MA(4) 0.099732 0.025782 3.868309 0.0001

R-squared 0.100869    Mean dependent var 0.009239


Adjusted R-squared 0.098449    S.D. dependent var 1.873924
S.E. of regression 1.779292    Akaike info criterion 3.993656
Sum squared resid 4704.497    Schwarz criterion 4.011453
Log likelihood -2972.270    Hannan-Quinn criter. 4.000288
F-statistic 41.67671    Durbin-Watson stat 1.998855
Prob(F-statistic) 0.000000

Inverted AR Roots       .65      .22+.59i    .22-.59i -.49-.37i


-.49+.37i
Inverted MA Roots  .35-.39i      .35+.39i   -.45+.39i -.45-.39i

vnindext=0.0094+0.1070AR(1)+0.0959AR(5)+0.1918MA(1)+0.0997M
A(4)+ ût
vnindext=0.0094+0.1070vnindext-1+0.0959vnindext-5+0.1918ût-1
+0.0997 ût -4 + ût

2. ARMA(1,1) (2,4)

Dependent Variable: VNINDEX


Method: Least Squares
Date: 04/26/20 Time: 09:28
Sample (adjusted): 1/05/2006 12/30/2011
Included observations: 1494 after adjustments
Convergence achieved after 18 iterations
MA Backcast: 12/29/2005 1/04/2006

Variable Coefficient Std. Error t-Statistic Prob.  

C 0.009034 0.071687 0.126018 0.8997


AR(1) 0.750675 0.160363 4.681107 0.0000
AR(2) -0.183966 0.050114 -3.670986 0.0003
MA(1) -0.454844 0.159017 -2.860352 0.0043
MA(4) 0.130064 0.024233 5.367276 0.0000

R-squared 0.100363    Mean dependent var 0.009488


Adjusted R-squared 0.097946    S.D. dependent var 1.872128
S.E. of regression 1.778082    Akaike info criterion 3.992289
Sum squared resid 4707.586    Schwarz criterion 4.010057
Log likelihood -2977.240    Hannan-Quinn criter. 3.998909
F-statistic 41.52787    Durbin-Watson stat 1.994982
Prob(F-statistic) 0.000000

Inverted AR Roots  .38-.21i      .38+.21i


Inverted MA Roots  .56-.39i      .56+.39i   -.33+.41i -.33-.41i

Vnindext=0.0090+0.7507AR(1)-0.1840AR(2)-
0.4548MA(1)+0.1301MA(4)+ ût
Vnindext=0.0090+0.7507vnindext-1-0.1840vnindext-2-0.4548ût-1+0.1301
ût-4+ ût
3. ARMA(1,1) (4,4)

Dependent Variable: VNINDEX


Method: Least Squares
Date: 04/26/20 Time: 09:29
Sample (adjusted): 1/09/2006 12/30/2011
Included observations: 1492 after adjustments
Convergence achieved after 6 iterations
MA Backcast: 1/03/2006 1/06/2006

Variable Coefficient Std. Error t-Statistic Prob.  

C 0.009024 0.076382 0.118145 0.9060


AR(1) 0.115933 0.078099 1.484441 0.1379
AR(4) 0.282220 0.079228 3.562126 0.0004
MA(1) 0.185295 0.077521 2.390249 0.0170
MA(4) -0.187756 0.080136 -2.342963 0.0193

R-squared 0.100253    Mean dependent var 0.008973


Adjusted R-squared 0.097833    S.D. dependent var 1.873324
S.E. of regression 1.779329    Akaike info criterion 3.993696
Sum squared resid 4707.861    Schwarz criterion 4.011483
Log likelihood -2974.297    Hannan-Quinn criter. 4.000324
F-statistic 41.42177    Durbin-Watson stat 2.002544
Prob(F-statistic) 0.000000

Inverted AR Roots       .76      .03-.73i    .03+.73i      -.70


Inverted MA Roots       .62     -.05+.65i   -.05-.65i      -.71

Vnindext=0.0090+0.1159AR(1)+0.2822AR(4)+0.1853MA(1)-
0.1878MA(4)+ût

Vnindext=0.0090+0.1159vnindext-1+0.2822vnindext-4+0.1853ût-1
-0.1878 ût-4+ût

4. ARMA(1,1) (5,4)
Dependent Variable: VNINDEX
Method: Least Squares
Date: 04/26/20 Time: 09:29
Sample (adjusted): 1/10/2006 12/30/2011
Included observations: 1491 after adjustments
Convergence achieved after 7 iterations
MA Backcast: 1/04/2006 1/09/2006

Variable Coefficient Std. Error t-Statistic Prob.  

C 0.009372 0.074636 0.125574 0.9001


AR(1) 0.106956 0.086394 1.237994 0.2159
AR(5) 0.095852 0.027470 3.489389 0.0005
MA(1) 0.191842 0.085607 2.240979 0.0252
MA(4) 0.099732 0.025782 3.868309 0.0001

R-squared 0.100869    Mean dependent var 0.009239


Adjusted R-squared 0.098449    S.D. dependent var 1.873924
S.E. of regression 1.779292    Akaike info criterion 3.993656
Sum squared resid 4704.497    Schwarz criterion 4.011453
Log likelihood -2972.270    Hannan-Quinn criter. 4.000288
F-statistic 41.67671    Durbin-Watson stat 1.998855
Prob(F-statistic) 0.000000

Inverted AR Roots       .65      .22+.59i    .22-.59i -.49-.37i


-.49+.37i
Inverted MA Roots  .35-.39i      .35+.39i   -.45+.39i -.45-.39i

5. ARMA(1,1) (5,12)

Dependent Variable: VNINDEX


Method: Least Squares
Date: 04/26/20 Time: 09:30
Sample (adjusted): 1/10/2006 12/30/2011
Included observations: 1491 after adjustments
Convergence achieved after 9 iterations
MA Backcast: 12/22/2005 1/09/2006

Variable Coefficient Std. Error t-Statistic Prob.  

C 0.009310 0.068275 0.136365 0.8916


AR(1) 0.113090 0.083719 1.350823 0.1770
AR(5) 0.076312 0.025830 2.954369 0.0032
MA(1) 0.183217 0.083092 2.204985 0.0276
MA(12) 0.012132 0.025541 0.475027 0.6348

R-squared 0.091971    Mean dependent var 0.009239


Adjusted R-squared 0.089527    S.D. dependent var 1.873924
S.E. of regression 1.788074    Akaike info criterion 4.003503
Sum squared resid 4751.054    Schwarz criterion 4.021301
Log likelihood -2979.612    Hannan-Quinn criter. 4.010136
F-statistic 37.62781    Durbin-Watson stat 1.998633
Prob(F-statistic) 0.000000
Inverted AR Roots       .62      .21+.57i    .21-.57i -.46-.35i
-.46+.35i
Inverted MA Roots  .66-.18i      .66+.18i    .48+.49i  .48-.49i
 .16+.67i      .16-.67i   -.19-.67i -.19+.67i
-.51+.49i     -.51-.49i   -.69+.18i -.69-.18i

Vnindext=0.0093+0.1131vnindext-1+0.0763vnindext-5+0.1832 ût-
1+0.0121 ût-12+ ût

6. ARMA(1,1) (5,13)

Dependent Variable: VNINDEX


Method: Least Squares
Date: 04/26/20 Time: 09:30
Sample (adjusted): 1/10/2006 12/30/2011
Included observations: 1491 after adjustments
Convergence achieved after 6 iterations
MA Backcast: 12/21/2005 1/09/2006

Variable Coefficient Std. Error t-Statistic Prob.  

C 0.009216 0.071197 0.129441 0.8970


AR(1) 0.130908 0.082608 1.584693 0.1132
AR(5) 0.073733 0.025779 2.860173 0.0043
MA(1) 0.162193 0.082275 1.971351 0.0489
MA(13) 0.063507 0.025583 2.482378 0.0132

R-squared 0.095243    Mean dependent var 0.009239


Adjusted R-squared 0.092808    S.D. dependent var 1.873924
S.E. of regression 1.784850    Akaike info criterion 3.999893
Sum squared resid 4733.932    Schwarz criterion 4.017691
Log likelihood -2976.920    Hannan-Quinn criter. 4.006525
F-statistic 39.10758    Durbin-Watson stat 1.996592
Prob(F-statistic) 0.000000

Inverted AR Roots       .62      .21+.56i    .21-.56i -.46-.35i


-.46+.35i
Inverted MA Roots  .77-.19i      .77+.19i    .59+.54i  .59-.54i
 .27+.76i      .27-.76i   -.11-.80i -.11+.80i
-.47-.66i     -.47+.66i   -.73+.38i -.73-.38i
     -.82

vnindext-1=0.0092+0.1309vnindext-1+0.0737vnindext-5+0.1622 ût-
1+0.0635 ût-13+ ût

7. ARMA(1,1) (13,12)
Dependent Variable: VNINDEX
Method: Least Squares
Date: 04/26/20 Time: 09:31
Sample (adjusted): 1/20/2006 12/30/2011
Included observations: 1483 after adjustments
Convergence achieved after 9 iterations
MA Backcast: 1/04/2006 1/19/2006

Variable Coefficient Std. Error t-Statistic Prob.  

C 0.008202 0.069749 0.117589 0.9064


AR(1) 0.117779 0.085175 1.382778 0.1669
AR(13) 0.074957 0.026369 2.842668 0.0045
MA(1) 0.184989 0.084535 2.188303 0.0288
MA(12) 0.024740 0.025993 0.951799 0.3414

R-squared 0.091742    Mean dependent var 0.008368


Adjusted R-squared 0.089284    S.D. dependent var 1.878678
S.E. of regression 1.792850    Akaike info criterion 4.008856
Sum squared resid 4750.752    Schwarz criterion 4.026731
Log likelihood -2967.567    Hannan-Quinn criter. 4.015519
F-statistic 37.32255    Durbin-Watson stat 2.000561
Prob(F-statistic) 0.000000

Inverted AR Roots       .83      .74-.38i    .74+.38i  .47+.67i


 .47-.67i      .11+.81i    .11-.81i -.28-.77i
-.28+.77i     -.60+.54i   -.60-.54i -.79+.20i
-.79-.20i
Inverted MA Roots  .70+.19i      .70-.19i    .51-.52i  .51+.52i
 .18-.71i      .18+.71i   -.21-.71i -.21+.71i
-.54+.52i     -.54-.52i   -.73+.19i -.73-.19i

Vnindext=0.0082+0.1178vnindext-1+0.0750vnindext-13+0.1850 ût-
1+0.0247 ût-12+ût

8. ARMA(1,1) (5,19)

Dependent Variable: VNINDEX


Method: Least Squares
Date: 04/26/20 Time: 09:32
Sample (adjusted): 1/10/2006 12/30/2011
Included observations: 1491 after adjustments
Convergence achieved after 9 iterations
MA Backcast: 12/13/2005 1/09/2006

Variable Coefficient Std. Error t-Statistic Prob.  

C 0.009282 0.068226 0.136053 0.8918


AR(1) 0.109587 0.083558 1.311501 0.1899
AR(5) 0.075719 0.025854 2.928726 0.0035
MA(1) 0.187093 0.082821 2.259009 0.0240
MA(19) 0.013538 0.025552 0.529827 0.5963
R-squared 0.092007    Mean dependent var 0.009239
Adjusted R-squared 0.089563    S.D. dependent var 1.873924
S.E. of regression 1.788039    Akaike info criterion 4.003464
Sum squared resid 4750.866    Schwarz criterion 4.021261
Log likelihood -2979.582    Hannan-Quinn criter. 4.010096
F-statistic 37.64402    Durbin-Watson stat 1.998942
Prob(F-statistic) 0.000000

Inverted AR Roots       .62      .21+.57i    .21-.57i -.46-.35i


-.46+.35i
Inverted MA Roots  .78+.13i      .78-.13i    .69+.38i  .69-.38i
 .53+.59i      .53-.59i    .31-.73i  .31+.73i
 .06-.79i      .06+.79i   -.21-.77i -.21+.77i
-.45-.67i     -.45+.67i   -.64-.49i -.64+.49i
-.77+.26i     -.77-.26i        -.81

vnindext=0.0093+0.1096vnindext-1+0.0757vnindext-5+0.1871 ût-
1+0.0135 ût-19+ût

9. ARMA(1,1) (13,14)

Dependent Variable: VNINDEX


Method: Least Squares
Date: 04/26/20 Time: 09:33
Sample (adjusted): 1/20/2006 12/30/2011
Included observations: 1483 after adjustments
Convergence achieved after 9 iterations
MA Backcast: 12/30/2005 1/19/2006

Variable Coefficient Std. Error t-Statistic Prob.  

C 0.008230 0.068557 0.120050 0.9045


AR(1) 0.104235 0.086177 1.209542 0.2266
AR(13) 0.068192 0.025974 2.625445 0.0087
MA(1) 0.198404 0.084914 2.336520 0.0196
MA(14) 0.020458 0.026251 0.779333 0.4359

R-squared 0.091584    Mean dependent var 0.008368


Adjusted R-squared 0.089126    S.D. dependent var 1.878678
S.E. of regression 1.793005    Akaike info criterion 4.009029
Sum squared resid 4751.574    Schwarz criterion 4.026904
Log likelihood -2967.695    Hannan-Quinn criter. 4.015692
F-statistic 37.25214    Durbin-Watson stat 1.998843
Prob(F-statistic) 0.000000

Inverted AR Roots       .82      .73-.38i    .73+.38i  .47-.67i


 .47+.67i      .11-.81i    .11+.81i -.28-.76i
-.28+.76i     -.60+.54i   -.60-.54i -.78-.19i
-.78+.19i
Inverted MA Roots  .73-.17i      .73+.17i    .58-.47i  .58+.47i
 .32+.68i      .32-.68i   -.01+.76i -.01-.76i
-.34+.68i     -.34-.68i   -.61-.47i -.61+.47i
-.75-.17i     -.75+.17i

vnindext=0.0082+0.1042vnindext-1+0.0682vnindext-13+0.1984 ût-
1+0.0205 ût-14+ ût

10. ARMA(1,1) (13,5)

Dependent Variable: VNINDEX


Method: Least Squares
Date: 04/26/20 Time: 09:33
Sample (adjusted): 1/20/2006 12/30/2011
Included observations: 1483 after adjustments
Convergence achieved after 8 iterations
MA Backcast: 1/13/2006 1/19/2006

Variable Coefficient Std. Error t-Statistic Prob.  

C 0.008226 0.071713 0.114705 0.9087


AR(1) 0.128795 0.082286 1.565205 0.1177
AR(13) 0.068273 0.025785 2.647768 0.0082
MA(1) 0.166001 0.081870 2.027627 0.0428
MA(5) 0.074036 0.025774 2.872473 0.0041

R-squared 0.095981    Mean dependent var 0.008368


Adjusted R-squared 0.093534    S.D. dependent var 1.878678
S.E. of regression 1.788661    Akaike info criterion 4.004178
Sum squared resid 4728.580    Schwarz criterion 4.022053
Log likelihood -2964.098    Hannan-Quinn criter. 4.010841
F-statistic 39.23015    Durbin-Watson stat 1.999202
Prob(F-statistic) 0.000000

Inverted AR Roots       .82      .73-.38i    .73+.38i  .47+.67i


 .47-.67i      .11+.81i    .11-.81i -.28-.76i
-.28+.76i     -.60-.54i   -.60+.54i -.78+.19i
-.78-.19i
Inverted MA Roots  .45+.35i      .45-.35i   -.22-.56i -.22+.56i
     -.63

vnindext=0.0082+0.1288vnindext-1+0.0683vnindext-13+0.1660 ût-1+0.0740 ût-5+ ût


Bước 7 :
STT Mô hình R2 Adjusted AIC SBIC HQIC
R2
1
2
3
4
5
6
7
8
9
10

You might also like