Annexe 5: Test d'homoscédasticité des
résidus
White Heteroskedasticity Test:
F-statistic 1.041280 Probability 0.564627
Obs*R-squared 16.94819 Probability 0.388957
|
Test Equation:
|
Dependent Variable: RESID^2
|
Method: Least Squares
|
Date: 08/24/12 Time: 03:52
|
Sample: 1991 2010
|
Included observations: 20
|
Variable Coefficient Std. Error t-Statistic Prob.
|
C -66.95781
|
33.38731
|
-2.005487
|
0.1386
|
STDPIB 9.994740
|
16.84427
|
0.593361
|
0.5947
|
STDPIB^2 -5.629154
|
5.668238
|
-0.993105
|
0.3939
|
STDEXP 10.78392
|
7.549572
|
1.428415
|
0.2485
|
STDEXP^2 -0.892668
|
0.607893
|
-1.468462
|
0.2383
|
VTE -9.784409
|
38.48448
|
-0.254243
|
0.8157
|
VTE^2 16.82434
|
28.68102
|
0.586602
|
0.5987
|
SEDEXP -198.9039
|
158.8844
|
-1.251878
|
0.2993
|
SEDEXP^2 468.3380
|
446.0078
|
1.050067
|
0.3708
|
TCDEM 18.61626
|
15.13952
|
1.229647
|
0.3065
|
TCDEM^2 -3.504035
|
3.132709
|
-1.118532
|
0.3448
|
BCPIB -0.380841
|
30.21753
|
-0.012603
|
0.9907
|
BCPIB^2 -88.94709
|
332.5404
|
-0.267478
|
0.8064
|
SOBPPIB 104.2977
|
79.23899
|
1.316242
|
0.2796
|
SOBPPIB^2 901.5496
|
483.3530
|
1.865199
|
0.1590
|
EXPPIB 228.4820
|
135.3351
|
1.688269
|
0.1899
|
EXPPIB^2 -416.2977
|
294.0307
|
-1.415831
|
0.2518
|
R-squared 0.847410
|
Mean dependent var 2.616316
|
Adjusted R-squared 0.033594
|
S.D. dependent var 3.272927
|
S.E. of regression 3.217482
|
Akaike info criterion 4.977955
|
Sum squared resid 31.05657
|
Schwarz criterion 5.824328
|
Log likelihood -32.77955
|
F-statistic 1.041280
|
Durbin-Watson stat 2.952847
|
Prob(F-statistic) 0.564627
|
Source : Nos calculs à l?aide du logiciel E-views
3.1
Ce tableau présente le test White, il a permis de savoir
si les résidus (termes
d?erreur) du modèle ont la même variance.
L?hypothèse nulle est celle d'homoscédasticité contre
l?hypothèse alternative
d?hétéroscédasticité. On accepte H0 si la
probabilité du test est supérieure à 0,05.
Annexe 6 : Test d'ARCH
ARCH Test:
F-statistic 1.030197 Probability
|
0.324341
|
Obs*R-squared 1.085609 Probability
|
0.297446
|
Test Equation:
|
|
Dependent Variable: RESID^2
|
|
Method: Least Squares
|
|
Date: 08/24/12 Time: 03:53
|
|
Sample(adjusted): 1992 2010
|
|
Included observations: 19 after adjusting endpoints
|
|
Variable Coefficient Std. Error t-Statistic
|
Prob.
|
C 3.026299 0.964850 3.136550
|
0.0060
|
RESID^2(-1) -0.231357 0.227941 -1.014986
|
0.3243
|
R-squared 0.057137
|
Mean dependent var
|
2.389149
|
Adjusted R-squared 0.001675
|
S.D. dependent var
|
3.196520
|
S.E. of regression 3.193842
|
Akaike info criterion
|
5.259627
|
Sum squared resid 173.4107
|
Schwarz criterion
|
5.359042
|
Log likelihood -47.96646
|
F-statistic
|
1.030197
|
Durbin-Watson stat 2.030595
|
Prob(F-statistic)
|
0.324341
|
Source : Nos calculs à l?aide du logiciel E-views
3.1
Nous avons utilisé ce test vient corroborer
l?homoscédasticité de White des érreurs, car selon ce test
les erreurs du modèle sont homoscédastiques (H0) si la
probabilité de Obs* R-squared est supérieure à 0,05.
|