ECOLE SUPERIEURE MEDITERANEENNE PRIVEE DES
SCIENCES INFORMATIQUES, ECONOMIQUES ET DE GESTION DE
NABEUL
Epreuve principale en ANNEE UNIVERSITAIRE
Durée 2H 2017-2018
Nombre de pages : 6 ECONOMETRIE DES
DONNEES DE PANEL Enseignant : Fethi Amri
Niveau : M1 finance
Recommandations :
Tous les documents sont interdits. Il sera tenu compte dans la notation des justifications et de la qualité de la rédaction.
Exercice 1(6pts)
On souhaite expliquer l’ouverture commerciale (TR) en fonction de l’énergie non-
renouvelable (NREN) pour 72 pays durant la période 1990-2012.
Ecrire les modèles permettant de spécifier une telle relation.
Exercice 2(7pts)
On souhaite expliquer l’ouverture commerciale (TR) en fonction de l’énergie renouvelable
(REN) pour 72 pays durant la période 1990-2012.
1.
Ecrire les modèles permettant de spécifier une telle relation.
2.
Ecrire le modèle pour un individu sous forme matricielle.
3.
Ecrire le modèle agrégé (celui de l’ensemble des individus) sous forme matricielle.
4.
Nous considérons dans un premier temps un model des données de panel statique, quel
modèle nous choisissons entre un modèle à effets fixes et un modèle à effets
aléatoires.
5. Interpréter les résultats.
6. Nous considérons dans deuxième temps un model des données de panel statique,
présenter le modèle sous forme dynamique.
7. Interpréter les résultats.
Annexes exercice 2
. . xtset countrycode ANNEE
panel variable: countrycode (strongly balanced)
time variable: ANNEE, 1990 to 2012
delta: 1 unit
.
. xtreg TR REN, fe
1
Fixed-effects (within) regression Number of obs = 1656
Group variable: countrycode Number of groups = 72
R-sq: within = 0.4868 Obs per group: min = 23
between = 0.4574 avg = 23.0
overall = 0.4590 max = 23
F(1,1583) = 1501.36
corr(u_i, Xb) = -0.2375 Prob > F = 0.0000
------------------------------------------------------------------------------
TR | Coef. Std. Err. t P>|t| [95% Conf. Interval]
-------------+----------------------------------------------------------------
REN | 4.30e+09 1.11e+08 38.75 0.000 4.08e+09 4.51e+09
_cons | 7.23e+10 5.27e+09 13.71 0.000 6.20e+10 8.27e+10
-------------+----------------------------------------------------------------
sigma_u | 3.096e+11
sigma_e | 1.410e+11
rho | .82816721 (fraction of variance due to u_i)
------------------------------------------------------------------------------
F test that all u_i=0: F(71, 1583) = 104.60 Prob > F = 0.0000
. estimate store fixed
. xtreg TR REN, re
Random-effects GLS regression Number of obs = 1656
Group variable: countrycode Number of groups = 72
R-sq: within = 0.4868 Obs per group: min = 23
between = 0.4574 avg = 23.0
overall = 0.4590 max = 23
Wald chi2(1) = 1552.50
corr(u_i, X) = 0 (assumed) Prob > chi2 = 0.0000
------------------------------------------------------------------------------
TR | Coef. Std. Err. z P>|z| [95% Conf. Interval]
-------------+----------------------------------------------------------------
REN | 4.24e+09 1.08e+08 39.40 0.000 4.03e+09 4.45e+09
_cons | 7.44e+10 3.58e+10 2.08 0.038 4.28e+09 1.45e+11
-------------+----------------------------------------------------------------
sigma_u | 3.001e+11
sigma_e | 1.410e+11
rho | .81913239 (fraction of variance due to u_i)
------------------------------------------------------------------------------
. hausman fixed
2
---- Coefficients ----
| (b) (B) (b-B) sqrt(diag(V_b-V_B))
| fixed . Difference S.E.
-------------+----------------------------------------------------------------
REN | 4.30e+09 4.24e+09 5.84e+07 2.69e+07
------------------------------------------------------------------------------
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(1) = (b-B)'[(V_b-V_B)^(-1)](b-B)
= 4.70
Prob>chi2 = 0.0302
. xtabond TR REN, lags(1) maxldep(2) maxlags(2) twostep artests(2)
Arellano-Bond dynamic panel-data estimation Number of obs = 1512
Group variable: countrycode Number of groups = 72
Time variable: ANNEE
Obs per group: min = 21
avg = 21
max = 21
Number of instruments = 43 Wald chi2(2) = 5.13e+08
Prob > chi2 = 0.0000
Two-step results
------------------------------------------------------------------------------
TR | Coef. Std. Err. z P>|z| [95% Conf. Interval]
-------------+----------------------------------------------------------------
TR |
L1. | .960855 .000054 1.8e+04 0.000 .9607491 .9609609
|
REN | 4.63e+08 547706 844.45 0.000 4.61e+08 4.64e+08
_cons | 2.23e+09 3.73e+08 5.97 0.000 1.50e+09 2.96e+09
------------------------------------------------------------------------------
Warning: gmm two-step standard errors are biased; robust standard
errors are recommended.
Instruments for differenced equation
GMM-type: L(2/3).TR
Standard: [Link]
Instruments for level equation
Standard: _cons
. estat sargan
Sargan test of overidentifying restrictions
H0: overidentifying restrictions are valid
chi2(40) = 41.89827
Prob > chi2 = 0.3885
3
. estat abond
Arellano-Bond test for zero autocorrelation in first-differenced errors
+-----------------------+
|Order | z Prob > z|
|------+----------------|
| 1 |-2.1462 0.0319 |
| 2 |-1.7719 0.0764 |
+-----------------------+
H0: no autocorrelation
..
Exercice 3(7pts)
On souhaite expliquer l’ouverture commerciale (TR) en fonction de l’énergie non-
renouvelable (REN), le capital (K) et le travail (L) pour 72 pays durant la période 1990-2012.
1. Quels sont les avantages de l’utilisation des données de panel.
2. Interpréter les résultats de l’estimation du modèle à effets fixes pour les pays
développés (countrycode<=23).
3. Interpréter les résultats de l’estimation du modèle à effets fixes pour les pays en
développement (countrycode>23).
4. Interpréter les résultats de l’estimation du modèle à effets fixes pour l’ensemble des
pays.
5. Comparer les 3 estimations.
Annexes exercice 3
Fixed-effects (within) regression Number of obs = 529
Group variable: countrycode Number of groups = 23
R-sq: within = 0.8049 Obs per group: min = 23
between = 0.8703 avg = 23.0
overall = 0.7865 max = 23
F(3,503) = 691.53
corr(u_i, Xb) = -0.9907 Prob > F = 0.0000
------------------------------------------------------------------------------
TR | Coef. Std. Err. t P>|t| [95% Conf. Interval]
-------------+----------------------------------------------------------------
NREN | 1.88e+09 2.21e+08 8.50 0.000 1.44e+09 2.31e+09
K | -.2182954 .0953127 -2.29 0.022 -.4055555 -.0310354
L | 41088.14 6724.407 6.11 0.000 27876.75 54299.53
_cons | -7.99e+11 6.94e+10 -11.51 0.000 -9.35e+11 -6.62e+11
-------------+----------------------------------------------------------------
sigma_u | 1.987e+12
sigma_e | 1.080e+11
rho | .99705581 (fraction of variance due to u_i)
------------------------------------------------------------------------------
F test that all u_i=0: F(22, 503) = 103.84 Prob > F = 0.0000
4
. xtreg TR NREN K L, fe
Fixed-effects (within) regression Number of obs = 1656
Group variable: countrycode Number of groups = 72
R-sq: within = 0.7594 Obs per group: min = 23
between = 0.8514 avg = 23.0
overall = 0.8193 max = 23
F(3,1581) = 1663.13
corr(u_i, Xb) = -0.7468 Prob > F = 0.0000
------------------------------------------------------------------------------
TR | Coef. Std. Err. t P>|t| [95% Conf. Interval]
-------------+----------------------------------------------------------------
NREN | 3.40e+08 3.56e+07 9.54 0.000 2.70e+08 4.10e+08
K | 1.261012 .0519961 24.25 0.000 1.159024 1.363001
L | 217.3385 375.3565 0.58 0.563 -518.9103 953.5873
_cons | 2.33e+10 1.01e+10 2.31 0.021 3.48e+09 4.32e+10
-------------+----------------------------------------------------------------
sigma_u | 2.498e+11
sigma_e | 9.661e+10
rho | .8698936 (fraction of variance due to u_i)
------------------------------------------------------------------------------
F test that all u_i=0: F(71, 1581) = 65.43 Prob > F = 0.0000
. xtreg TR NREN K L if countrycode>23, fe
Fixed-effects (within) regression Number of obs = 1127
Group variable: countrycode Number of groups = 49
R-sq: within = 0.8763 Obs per group: min = 23
between = 0.6984 avg = 23.0
overall = 0.6967 max = 23
F(3,1075) = 2538.03
corr(u_i, Xb) = -0.7616 Prob > F = 0.0000
------------------------------------------------------------------------------
TR | Coef. Std. Err. t P>|t| [95% Conf. Interval]
-------------+----------------------------------------------------------------
NREN | 6.98e+08 7.84e+07 8.91 0.000 5.45e+08 8.52e+08
K | .300509 .1684445 1.78 0.075 -.0300082 .6310263
L | 1089.893 252.1684 4.32 0.000 595.0949 1584.691
_cons | -8.42e+09 8.70e+09 -0.97 0.333 -2.55e+10 8.64e+09
-------------+----------------------------------------------------------------
sigma_u | 2.076e+11
sigma_e | 5.999e+10
5
rho | .92294037 (fraction of variance due to u_i)
------------------------------------------------------------------------------
F test that all u_i=0: F(48, 1075) = 48.50 Prob > F = 0.0000
Bonne chance
