how interpret var-bekk results?
Posted: Fri Apr 22, 2016 7:09 am
Hello, i would like to estimate a VAR-GARCH model but i need some help
Step by step, I do:
1) a VAR system with my variables
2) proc > make a system ordered by variable > estimate > garch BEKK model.
now: i do not know how interpret results?!
this is my result for (LM, lgdp)
Coefficient Std. Error z-Statistic Prob.
C(1) 0.834258 0.101981 8.180504 0.0000
C(2) 0.259219 0.119544 2.168395 0.0301
C(3) -1.982574 0.890183 -2.227153 0.0259
C(4) -0.037918 0.069815 -0.543117 0.5870
C(5) 1.023732 0.095475 10.72251 0.0000
C(6) 0.020868 0.719952 0.028985 0.9769
Variance Equation Coefficients
C(7) 0.045151 0.033381 1.352622 0.1762
C(8) 0.015731 0.005159 3.048983 0.0023
C(9) -0.518758 0.265356 -1.954950 0.0506
C(10) -0.080221 0.545020 -0.147189 0.8830
C(11) 0.566095 0.612590 0.924100 0.3554
C(12) 0.894323 0.061216 14.60926 0.0000
Log likelihood 104.2757 Schwarz criterion -4.739637
Avg. log likelihood 1.489653 Hannan-Quinn criter. -5.088817
Akaike info criterion -5.272900
Equation: LM= C(1)*LM(-1) + C(2)*LGDP(-1) + C(3)
R-squared 0.980023 Mean dependent var 7.504566
Adjusted R-squared 0.978775 S.D. dependent var 0.530021
S.E. of regression 0.077218 Sum squared resid 0.190805
Durbin-Watson stat 1.290141
Equation: LGDP = C(4)*LM(-1) + C(5)*LGDP(-1) + C(6)
R-squared 0.971850 Mean dependent var 12.60067
Adjusted R-squared 0.970091 S.D. dependent var 0.384660
S.E. of regression 0.066524 Sum squared resid 0.141615
Durbin-Watson stat 1.029185
Covariance specification: Diagonal BEKK
GARCH = M + A1*RESID(-1)*RESID(-1)'*A1 + B1*GARCH(-1)*B1
M is a rank one matrix
A1 is a diagonal matrix
B1 is a diagonal matrix
Transformed Variance Coefficients
Coefficient Std. Error z-Statistic Prob.
M(1,1) 0.002039 0.003014 0.676311 0.4988
M(1,2) 0.000710 0.000575 1.234895 0.2169
M(2,2) 0.000247 0.000162 1.524492 0.1274
A1(1,1) -0.518758 0.265356 -1.954950 0.0506
A1(2,2) -0.080221 0.545020 -0.147189 0.8830
B1(1,1) 0.566095 0.612590 0.924100 0.3554
B1(2,2) 0.894323 0.061216 14.60926 0.0000
Step by step, I do:
1) a VAR system with my variables
2) proc > make a system ordered by variable > estimate > garch BEKK model.
now: i do not know how interpret results?!
this is my result for (LM, lgdp)
Coefficient Std. Error z-Statistic Prob.
C(1) 0.834258 0.101981 8.180504 0.0000
C(2) 0.259219 0.119544 2.168395 0.0301
C(3) -1.982574 0.890183 -2.227153 0.0259
C(4) -0.037918 0.069815 -0.543117 0.5870
C(5) 1.023732 0.095475 10.72251 0.0000
C(6) 0.020868 0.719952 0.028985 0.9769
Variance Equation Coefficients
C(7) 0.045151 0.033381 1.352622 0.1762
C(8) 0.015731 0.005159 3.048983 0.0023
C(9) -0.518758 0.265356 -1.954950 0.0506
C(10) -0.080221 0.545020 -0.147189 0.8830
C(11) 0.566095 0.612590 0.924100 0.3554
C(12) 0.894323 0.061216 14.60926 0.0000
Log likelihood 104.2757 Schwarz criterion -4.739637
Avg. log likelihood 1.489653 Hannan-Quinn criter. -5.088817
Akaike info criterion -5.272900
Equation: LM= C(1)*LM(-1) + C(2)*LGDP(-1) + C(3)
R-squared 0.980023 Mean dependent var 7.504566
Adjusted R-squared 0.978775 S.D. dependent var 0.530021
S.E. of regression 0.077218 Sum squared resid 0.190805
Durbin-Watson stat 1.290141
Equation: LGDP = C(4)*LM(-1) + C(5)*LGDP(-1) + C(6)
R-squared 0.971850 Mean dependent var 12.60067
Adjusted R-squared 0.970091 S.D. dependent var 0.384660
S.E. of regression 0.066524 Sum squared resid 0.141615
Durbin-Watson stat 1.029185
Covariance specification: Diagonal BEKK
GARCH = M + A1*RESID(-1)*RESID(-1)'*A1 + B1*GARCH(-1)*B1
M is a rank one matrix
A1 is a diagonal matrix
B1 is a diagonal matrix
Transformed Variance Coefficients
Coefficient Std. Error z-Statistic Prob.
M(1,1) 0.002039 0.003014 0.676311 0.4988
M(1,2) 0.000710 0.000575 1.234895 0.2169
M(2,2) 0.000247 0.000162 1.524492 0.1274
A1(1,1) -0.518758 0.265356 -1.954950 0.0506
A1(2,2) -0.080221 0.545020 -0.147189 0.8830
B1(1,1) 0.566095 0.612590 0.924100 0.3554
B1(2,2) 0.894323 0.061216 14.60926 0.0000