Rapport de stage

Premiére regression
Dependent Variable: VRISK
Method: Panel Two-Stage EGLS (Cross-section random effects)
Date: 06/11/10 Time: 16:29
Sample (adjusted): 2001 2008Cross-sections included: 10
Total panel (balanced) observations: 80
Swamy and Arora estimator of component variances
Instrument list: C CAR(-3) REG ROA TAIL

Variable Coefficient Std.Error t-Statistic Prob.

C 0.900310 0.257019 3.502887 0.0008
REG -1.115747 0.687475 -1.622965 0.1088
ROA -0.039976 0.010776 -3.709706 0.0004
TAIL-3.574598 1.839911 -1.942810 0.0558
VCAR -0.713813 0.091361 -7.72003 0.0000
CAR1 -0.004333 0.004222 -1.026525 0.1565
EffectsSpecification

Cross-section random S.D. / Rho 0.007873 0.4348
Idiosyncratic random S.D. / Rho 0.008976 0.5652

Weighted Statistics

R-squared -0.030678 Mean dependent var-0.004016
Adjusted R-squared -0.085647 S.D. dependent var 0.046302
S.E. of regression 0.048244 Sum squared resid 0.174565
Durbin-Watson stat 2.022412 J-statistic 2.85E-19
Instrumentrank 5.000000

Unweighted Statistics

R-squared -0.086874 Mean dependent var -0.010743
Sum squared resid 0.197753 Durbin-Watson stat 1.785265
TEST DE HAUSMANTest: Ho: difference in coefficients not systematic

chi2(6) = (b-B)'[(V_b-V_B)^(-1)](b-B)
= 3.58
Prob>chi2 = 0.7331(V_b-V_B is not positive definite)

On retient le modéle a effet aléatoir

TEST DE VALIDITE DINSTRUMENT DE SARGAN
Sargan test of overid. restrictions:chi2(1) = 25.54 Prob > chi2 = 0.000

ce résultat indique que les…