Für die Langfristgleichung stehen
folgende Gleichungen zur Verfügung:
Variante 1:
cprivp = c(10) + c(12)*yavhh/cpripi*100 + c(13)*zinskr + c(17)*dq1 + c(18)*dq2 + c(19)*dq3
Variante 2:
cprivp = c(10) + c(12)*yavm/cpripi*100 +
c(13)*yavuv/cpripi*100 + c(14)*zinsl + c(17)*dq1 + c(18)*dq2 + c(19)*dq3
'0.68
Variante 3:
cprivp
= c(10) + c(11)*cprivp(-1) + c(12)*yavhh/cpripi*100 + c(13)*zinskr +
c(17)*dq1 + c(18)*dq2 + c(19)*dq3
|
Dependent Variable: CPRIVP |
|
|
||
|
Method:
Least Squares |
|
|
||
|
Date:
05/26/13 Time: 11:33 |
|
|
||
|
Sample:
2003Q2 2013Q1 |
|
|
||
|
Included
observations: 40 |
|
|
||
|
CPRIVP = C(10) + C(12)*YAVHH/CPRIPI*100 +
C(13)*ZINSKR + C(17) |
||||
|
*DQ1 +
C(18)*DQ2 + C(19)*DQ3 |
|
|||
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Coefficient |
Std. Error |
t-Statistic |
Prob. |
|
|
|
|
|
|
|
|
|
|
|
|
|
C(10) |
-13.40797 |
3.181801 |
-4.213957 |
0.0002 |
|
C(12) |
0.944468 |
0.008274 |
114.1437 |
0.0000 |
|
C(13) |
-0.648843 |
0.114711 |
-5.656341 |
0.0000 |
|
C(17) |
-19.24845 |
0.430920 |
-44.66832 |
0.0000 |
|
C(18) |
-4.179025 |
0.430067 |
-9.717139 |
0.0000 |
|
C(19) |
-0.131800 |
0.429014 |
-0.307216 |
0.7606 |
|
|
|
|
|
|
|
|
|
|
|
|
|
R-squared |
0.998214 |
Mean dependent var |
341.2298 |
|
|
Adjusted
R-squared |
0.997951 |
S.D. dependent var |
21.18738 |
|
|
S.E.
of regression |
0.959021 |
Akaike info criterion |
2.891674 |
|
|
Sum
squared resid |
31.27055 |
Schwarz
criterion |
3.145006 |
|
|
Log likelihood |
-51.83349 |
Hannan-Quinn
criter. |
2.983271 |
|
|
F-statistic |
3800.283 |
Durbin-Watson
stat |
1.145614 |
|
|
Prob(F-statistic) |
0.000000 |
|
|
|
Variante 2
(Vorbild: Interwar Modell von L. Klein):
|
Dependent
Variable: CPRIVP |
|
|
||
|
Method:
Least Squares |
|
|
||
|
Date:
05/26/13 Time: 11:35 |
|
|
||
|
Sample:
2003Q2 2013Q1 |
|
|
||
|
Included
observations: 40 |
|
|
||
|
CPRIVP = C(10)
+ C(12)*YAVM/CPRIPI*100 + C(13)*YAVUV/CPRIPI*100 + |
||||
|
C(14)*ZINSL
+ C(17)*DQ1 + C(18)*DQ2 + C(19)*DQ3 |
||||
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Coefficient |
Std. Error |
t-Statistic |
Prob. |
|
|
|
|
|
|
|
|
|
|
|
|
|
C(10) |
18.98407 |
11.01748 |
1.723086 |
0.0942 |
|
C(12) |
0.925253 |
0.041542 |
22.27263 |
0.0000 |
|
C(13) |
0.682358 |
0.027991 |
24.37794 |
0.0000 |
|
C(14) |
-1.657911 |
0.584057 |
-2.838613 |
0.0077 |
|
C(17) |
-13.29225 |
1.407448 |
-9.444221 |
0.0000 |
|
C(18) |
2.037829 |
1.343688 |
1.516594 |
0.1389 |
|
C(19) |
5.842507 |
0.999622 |
5.844715 |
0.0000 |
|
|
|
|
|
|
|
|
|
|
|
|
|
R-squared |
0.995884 |
Mean dependent var |
341.2298 |
|
|
Adjusted
R-squared |
0.995136 |
S.D. dependent var |
21.18738 |
|
|
S.E.
of regression |
1.477664 |
Akaike info criterion |
3.776431 |
|
|
Sum
squared resid |
72.05524 |
Schwarz criterion |
4.071985 |
|
|
Log
likelihood |
-68.52861 |
Hannan-Quinn criter. |
3.883294 |
|
|
F-statistic |
1330.837 |
Durbin-Watson stat |
1.277772 |
|
|
Prob(F-statistic) |
0.000000 |
|
|
|
Variante 3
(Vorbild: siehe Literaturhinweis):
|
Dependent
Variable: CPRIVP |
|
|
||
|
Method:
Least Squares |
|
|
||
|
Date:
05/26/13 Time: 11:37 |
|
|
||
|
Sample:
2003Q2 2013Q1 |
|
|
||
|
Included
observations: 40 |
|
|
||
|
CPRIVP = C(10) + C(11)*CPRIVP(-1) +
C(12)*YAVHH/CPRIPI*100 + C(13) |
||||
|
*ZINSKR
+ C(17)*DQ1 + C(18)*DQ2 + C(19)*DQ3 |
||||
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Coefficient |
Std. Error |
t-Statistic |
Prob. |
|
|
|
|
|
|
|
|
|
|
|
|
|
C(10) |
-13.40941 |
3.229150 |
-4.152611 |
0.0002 |
|
C(11) |
0.005396 |
0.051877 |
0.104011 |
0.9178 |
|
C(12) |
0.939586 |
0.047681 |
19.70576 |
0.0000 |
|
C(13) |
-0.643518 |
0.127176 |
-5.060060 |
0.0000 |
|
C(17) |
-19.21812 |
0.525654 |
-36.56040 |
0.0000 |
|
C(18) |
-4.070425 |
1.131669 |
-3.596834 |
0.0010 |
|
C(19) |
-0.080417 |
0.658493 |
-0.122123 |
0.9035 |
|
|
|
|
|
|
|
|
|
|
|
|
|
R-squared |
0.998214 |
Mean dependent var |
341.2298 |
|
|
Adjusted
R-squared |
0.997890 |
S.D. dependent var |
21.18738 |
|
|
S.E.
of regression |
0.973284 |
Akaike info criterion |
2.941347 |
|
|
Sum
squared resid |
31.26030 |
Schwarz criterion |
3.236901 |
|
|
Log
likelihood |
-51.82693 |
Hannan-Quinn criter. |
3.048210 |
|
|
F-statistic |
3074.768 |
Durbin-Watson stat |
1.156494 |
|
|
Prob(F-statistic) |
0.000000 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Variante
3 zeigt, dass der Timelag statistisch nicht signifkant ist. Deshalb kommen nur
die ersten beiden Varianten in Frage, die mit einem Fehlerkorrekturmodell
gekoppelt werden können. Im Fall einer prognostischen Verwendung des Modells
liefern beide Gleichungen etwa die gleichen Ergebnisse. Eine ausführliche
Begründung der Variablenwahl und des FKM findet man in G. Quaas: Die
Konsumfunktion in ökonometrischen Modellen für Deutschlands Volkswirtschaft auf
Basis der VGR 2005. In: A. Wagner (Hrsg.): Empirische Wirtschaftsforschung
heute. Stuttgart 2009. S.99-110.