Annexe VII : Estimation du Panel des IDE
xtreg IDE Traitement_Matching pays_en_transition effet_transition
logPIBpc logPopulation ExportPIB Educexpend Tinflation QualitGovernment
Telephonelines, re
Random-effects GLS regression Number of obs = 1051
Group variable (i): countrycode Number of groups = 61
R-sq: within = 0.1276 Obs per group: min = 2
between = 0.6894 avg = 17.2
overall = 0.4173 max = 25
Random effects u_i ~ Gaussian Wald chi2(10) = 330.47
corr(u_i, X) = 0 (assumed) Prob > chi2 = 0.0000
IDE | Coef. Std. Err. z P>|z| [95% Conf. Interval]
+
|
pays_en_transition
|
|
|
-1.64e+09
|
|
effet_transition
|
|
|
1.34e+09
|
|
logPIBpc
|
|
|
2.12e+09
|
|
logPopulation
|
|
|
4.50e+08
|
|
ExportPIB
|
|
|
3.19e+07
|
|
Educexpend
|
|
|
.1011863
|
|
Tinflation
|
|
|
-4.46e+08
|
|
QualitGoverment
|
|
|
1.23e+09
|
|
Telephonelines
|
|
|
395.7187
|
|
_cons
|
|
-2.12e+10
|
|
+
sigma_u
|
|
|
2.679e+09
|
|
sigma_e
|
|
|
7.253e+09
|
|
rho
|
|
|
.12004368
|
Traitement_Matching| -1.51e+09 8.14e+08 -1.86 0.063 -3.11e+09
8.24e+07 1.17e+09 -1.40 0.162 -3.92e+09 6.54e+08 1.07e+09 1.26 0.209 -7.54e+08
3.44e+09 3.72e+08 5.71 0.000 1.39e+09 2.85e+09 4.03e+08 1.12 0.264 -3.39e+08
1.24e+09 1.97e+07 1.62 0.106 -6806383 7.06e+07 .0340846 2.97 0.003 .0343818
.1679908 1.93e+08 -2.31 0.021 -8.25e+08 -6.70e+07 1.93e+09 0.64 0.525 -2.56e+09
5.01e+09 62.63903 6.32 0.000 272.9485 518.489 7.45e+09 -2.85 0.004 -3.58e+10
-6.60e+09
(fraction of variance due to u_i)
École polytechnique de Tunis Mémoire de
Mastère Derbali Ahmed
121
** Test d'absence d'effets spécifiques
individuels**
. xttest0
Breusch and Pagan Lagrangian multiplier test for random
effects: IDE[countrycode,t] = Xb + u[countrycode] + e[countrycode,t]
Estimated results:
| Var sd = sqrt(Var)
+
IDE| 1.13e+20 1.06e+10
e | 5.26e+19 7.25e+09
u | 7.18e+18 2.68e+09
Test: Var(u) = 0
chi2(1) = 310.40
Prob > chi2 = 0.0000
**test d'hausman (choix entre FE et RE) **
. quiet xtreg IDE Traitement_Matching pays_en_transition
effet_transition logPIBpc logPopulation ExportPIB Educexpend Tinflation
QualitGovernment Telephonelines, fe
. estimates store fixed
. quiet xtreg IDE Traitement_Matching pays_en_transition
effet_transition logPIBpc logPopulation ExportPIB Educexpend Tinflation
QualitGovernment Telephonelines, re
. estimates store random . hausman fixed random
IDE | Coef. Std. Err. t
+
Traitement_Matching| -2.67e+09 8.21e+08 pays_en_transition
| -2.13e+09 9.08e+08
effet_transition | 2.52e+09 1.08e+09
logPIBpc | 1.63e+09 2.80e+08
logPopulation | 8.04e+08 2.58e+08
|
P>|t|
|
[95% Conf. Interval]
|
|
-3.25
|
0.001
|
-4.28e+09
|
-1.05e+09
|
|
-2.35
|
0.019
|
-3.91e+09
|
-3.49e+08
|
|
2.32
|
0.020
|
3.92e+08
|
4.64e+09
|
|
5.83
|
0.000
|
1.08e+09
|
2.18e+09
|
|
3.12
|
0.002
|
2.98e+08
|
1.31e+09
|
122
École polytechnique de Tunis
|
Mémoire de Mastère
|
Derbali Ahmed
|
|
+
|
|
|
|
---- Coefficients ----
(b) (B)
fixed random
|
(b-B)
Difference
|
sqrt(diag(V_b-V_B)) S.E.
|
|
Traitement~g
|
|
|
-7.87e+08
|
-1.51e+09
|
7.25e+08
|
1.77e+08
|
|
effet_tran~n
|
|
|
8.45e+08
|
1.34e+09
|
-4.99e+08
|
2.07e+08
|
|
logPIBpc
|
|
|
3.93e+09
|
2.12e+09
|
1.81e+09
|
5.31e+08
|
|
logPopulat~n
|
|
|
-3.46e+09
|
4.50e+08
|
-3.91e+09
|
1.60e+09
|
|
ExportPIB
|
|
|
1.40e+07
|
3.19e+07
|
-1.79e+07
|
1.94e+07
|
|
Educexpend
|
|
|
.1176888
|
.1011863
|
.0165024
|
.0253602
|
|
Tinflation
|
|
|
-1.09e+08
|
-4.46e+08
|
3.37e+08
|
1.18e+08
|
|
QualitGove~t
|
|
|
-3.08e+08
|
1.23e+09
|
-1.53e+09
|
1.12e+09
|
|
Telephonel~s
|
|
|
532.7445
|
395.7187
|
137.0258
|
48.44583
|
b = consistent under Ho and Ha; obtained from xtreg
B = inconsistent under Ha, efficient under Ho; obtained from
xtreg
Test: Ho: difference in coefficients not systematic
chi2(7) = (b-B)'[(V_b-V_B)^(-1)](b-B)
= 33.36
Prob>chi2 = 0.0000
. reg IDE Traitement_Matching pays_en_transition effet_transition
logPIBpc logPopulation ExportPIB Educexpend Tinflation QualitGovernment
Telephonelines
Source | SS df MS Number of obs = 1051
+ F( 10, 1040) = 76.89
Model | 5.0246e+22 10 5.0246e+21 Prob > F = 0.0000
Residual | 6.7965e+22 1040 6.5351e+19 R-squared = 0.4251
+ Adj R-squared = 0.4195
Total | 1.1821e+23 1050 1.1258e+20 Root MSE = 8.1e+09
123
École polytechnique de Tunis
|
Mémoire de Mastère
|
|
Derbali Ahmed
|
|
ExportPIB
|
|
|
6.03e+07
|
1.47e+07
|
4.11
|
0.000
|
3.15e+07
|
8.90e+07
|
|
Educexpend
|
|
|
.1350692
|
.0320046
|
4.22
|
0.000
|
.0722682
|
.1978701
|
|
Tinflation
|
|
|
-5.59e+08
|
1.79e+08
|
-3.12
|
0.002
|
-9.10e+08
|
-2.07e+08
|
|
QualitGovernment
|
|
|
4.27e+09
|
1.74e+09
|
2.46
|
0.014
|
8.62e+08
|
7.68e+09
|
|
Telephonelines
|
|
|
309.1495
|
57.43063
|
5.38
|
0.000
|
196.4564
|
421.8426
|
|
_cons
|
|
-2.47e+10
|
5.00e+09
|
-4.94
|
0.000
|
-3.45e+10
|
-1.49e+10
|
**
hétéroscédasticité**
. reg IDE Traitement_Matching pays_en_transition effet_transition
logPIBpc logPopulation ExportPIB Educexpend Tinflation QualitGovernment Tel
> ephonelines
Source | SS df MS Number of obs = 1051
+ F( 10, 1040) = 76.89
Model | 5.0246e+22 10 5.0246e+21 Prob > F = 0.0000
Residual | 6.7965e+22 1040 6.5351e+19 R-squared = 0.4251
+ Adj R-squared = 0.4195
|
Total | 1.1821e+23
IDE |
+
|
1050 1.1258e+20
Coef. Std. Err.
|
Root MSE
t P>|t|
|
= 8.1e+09
[95% Conf. Interval]
|
|
Traitement_Matching | -2.67e+09
|
8.21e+08
|
-3.25
|
0.001
|
-4.28e+09
|
-1.05e+09
|
|
pays_en_transition
|
|
|
-2.13e+09
|
9.08e+08
|
-2.35
|
0.019
|
-3.91e+09
|
-3.49e+08
|
|
effet_transition
|
|
|
2.52e+09
|
1.08e+09
|
2.32
|
0.020
|
3.92e+08
|
4.64e+09
|
|
logPIBpc
|
|
|
1.63e+09
|
2.80e+08
|
5.83
|
0.000
|
1.08e+09
|
2.18e+09
|
|
logPopulation
|
|
|
8.04e+08
|
2.58e+08
|
3.12
|
0.002
|
2.98e+08
|
1.31e+09
|
|
ExportPIB
|
|
|
6.03e+07
|
1.47e+07
|
4.11
|
0.000
|
3.15e+07
|
8.90e+07
|
|
Educexpend
|
|
|
.1350692
|
.0320046
|
4.22
|
0.000
|
.0722682
|
.1978701
|
|
Tinflation
|
|
|
-5.59e+08
|
1.79e+08
|
-3.12
|
0.002
|
-9.10e+08
|
-2.07e+08
|
|
QualitGovernment
|
|
|
4.27e+09
|
1.74e+09
|
2.46
|
0.014
|
8.62e+08
|
7.68e+09
|
|
Telephonelines
|
|
|
309.1495
|
57.43063
|
5.38
|
0.000
|
196.4564
|
421.8426
|
|
_cons
|
|
-2.47e+10
|
5.00e+09
|
-4.94
|
0.000
|
-3.45e+10
|
-1.49e+10
|
. hettest
Breusch-Pagan / Cook-Weisberg test for heteroskedasticity Ho:
Constant variance
École polytechnique de Tunis Mémoire de
Mastère Derbali Ahmed
124
Variables: fitted values of IDE
chi2(1) = 1149.46
Prob > chi2 = 0.0000
quiet reg IDE Traitement_Matching pays_en_transition
effet_transition logPIBpc logPopulation ExportPIB Educexpend Tinflation
QualitGovernment Telephonelines
. predict residu,resid
(1259 missing values generated)
. gen resid2=residu^2
(1259 missing values generated)
. reg resid2 Traitement_Matching pays_en_transition
effet_transition logPIBpc logPopulation ExportPIB Educexpend Tinflation
QualitGovernment Telephonelines
Source | SS df MS Number of obs = 1051
|
+
|
|
F( 10, 1040)
|
= 6.61
|
|
Model | 3.3311e+43
|
10 3.3311e+42
|
Prob > F
|
=
|
0.0000
|
|
Residual| 5.2449e+44
|
1040 5.0432e+41
|
R-squared
|
=
|
0.0597
|
+ Adj R-squared = 0.0507
Total | 5.5780e+44 1050 5.3124e+41 Root MSE = 7.1e+20
|
resid2
+
|
|
|
Coef. Std. Err. t
|
P>|t|
|
[95% Conf. Interval]
|
|
Traitement_Matching| -2.39e+20
|
7.21e+19
|
-3.31
|
0.001
|
-3.80e+20
|
-9.73e+19
|
|
pays_en_transition
|
|
|
-2.04e+20
|
7.98e+19
|
-2.55
|
0.011
|
-3.60e+20
|
-4.71e+19
|
|
effet_transition
|
|
|
2.28e+20
|
9.51e+19
|
2.39
|
0.017
|
4.11e+19
|
4.14e+20
|
|
logPIBpc
|
|
|
8.67e+19
|
2.46e+19
|
3.53
|
0.000
|
3.85e+19
|
1.35e+20
|
|
logPopulation
|
|
|
5.61e+19
|
2.27e+19
|
2.48
|
0.013
|
1.17e+19
|
1.01e+20
|
|
ExportPIB
|
|
|
4.30e+18
|
1.29e+18
|
3.34
|
0.001
|
1.78e+18
|
6.83e+18
|
|
Educexpend
|
|
|
7.19e+08
|
2.81e+09
|
0.26
|
0.798
|
-4.80e+09
|
6.24e+09
|
|
Tinflation
|
|
|
-1.69e+18
|
1.57e+19
|
-0.11
|
0.914
|
-3.26e+19
|
2.92e+19
|
|
QualitGove~t
|
|
|
2.53e+20
|
1.53e+20
|
1.66
|
0.097
|
-4.60e+19
|
5.53e+20
|
|
Telephonel~s
|
|
|
-9.48e+12
|
5.05e+12
|
-1.88
|
0.060
|
-1.94e+13
|
4.15e+11
|
|
_cons
|
|
-1.55e+21
|
4.39e+20
|
-3.53
|
0.000
|
-2.41e+21
|
-6.88e+20
|
École polytechnique de Tunis Mémoire de
Mastère Derbali Ahmed
125
scalar R2= e(r2)
scalar N =_N
scalar NR2= N*R2
scalar CHI29= invchi2(10, 0.95)
scalar list
CHI29 = 18.307038
NR2 = 137.95119
N = 2310
R2 = .05971913
** Autocorrélation**
. xtserial IDE Traitement_Matching pays_en_transition
effet_transition logPIBpc logPopulation ExportPIB Educexpend Tinflation
QualitGovernment
> Telephonelines
Wooldridge test for autocorrelation in panel data H0: no
first-order autocorrelation
F(1, 53) = 0.174
Prob > F = 0.6778
École polytechnique de Tunis Mémoire de
Mastère Derbali Ahmed
126
** Régression finale **
. reg IDE Traitement_Matching pays_en_transition effet_transition
logPIBpc logPopulation ExportPIB Educexpend Tinflation QualitGovernment
Telephonelines, robust
|
Linear regression Number of obs = 1051
|
|
F( 10, 1040)
|
= 96.11
|
|
Prob > F
|
=
|
0.0000
|
|
R-squared
|
|
= 0.4251
|
|
Root MSE
|
|
= 8.1e+09
|
|
IDE
+
|
|
|
|
Robust
Coef. Std. Err.
|
t
|
P>|t|
|
[95% Conf. Interval]
|
|
Traitement_Matching | -2.67e+09
|
1.42e+09
|
-1.88
|
0.061
|
-5.45e+09
|
1.21e+08
|
|
pays_en_transition
|
|
|
-2.13e+09
|
1.20e+09
|
-1.77
|
0.077
|
-4.49e+09
|
2.30e+08
|
|
effet_transition
|
|
|
2.52e+09
|
1.38e+09
|
1.82
|
0.070
|
-2.01e+08
|
5.23e+09
|
|
logPIBpc
|
|
|
1.63e+09
|
4.17e+08
|
3.91
|
0.000
|
8.12e+08
|
2.45e+09
|
|
logPopulation
|
|
|
8.04e+08
|
2.67e+08
|
3.02
|
0.003
|
2.81e+08
|
1.33e+09
|
|
ExportPIB
|
|
|
6.03e+07
|
2.06e+07
|
2.93
|
0.003
|
1.99e+07
|
1.01e+08
|
|
Educexpend
|
|
|
.1350692
|
.0386908
|
3.49
|
0.001
|
.0591482
|
.2109901
|
|
Tinflation
|
|
|
-5.59e+08
|
1.09e+08
|
-5.11
|
0.000
|
-7.74e+08
|
-3.44e+08
|
|
QualitGovernment
|
|
|
4.27e+09
|
1.79e+09
|
2.39
|
0.017
|
7.62e+08
|
7.78e+09
|
|
Telephonelines
|
|
|
309.1495
|
76.37086
|
4.05
|
0.000
|
159.291
|
459.0081
|
|
_cons
|
|
-2.47e+10
|
7.08e+09
|
-3.49
|
0.001
|
-3.86e+10
|
-1.08e+10
|
| 
