p value for J-statistics for simultaneous equation (gmm)

For technical questions regarding estimation of single equations, systems, VARs, Factor analysis and State Space Models in EViews. General econometric questions and advice should go in the Econometric Discussions forum.

Moderators: EViews Gareth, EViews Moderator

kelaynak
Posts: 33
Joined: Fri Nov 07, 2008 7:02 am

p value for J-statistics for simultaneous equation (gmm)

Postby kelaynak » Sat Sep 04, 2010 7:40 am

Hello,

Could you please help me how I can find p-value for J-statistics in simulataneous equation with GMM?

Specifically, I am looking "the degrees of freedom (df)" number for simulatneous equation with GMM method. I could not find how to calculate it.

IF I know that, I think I can find p-value by using
the p value: pval=chisq(j-statistics,df).

Thanks in advance

EViews Gareth
Fe ddaethom, fe welon, fe amcangyfrifon
Posts: 12503
Joined: Tue Sep 16, 2008 5:38 pm

Re: p value for J-statistics for simultaneous equation (gmm)

Postby EViews Gareth » Mon Sep 06, 2010 4:46 pm

In general the degrees of freedom are equal to the number of moment conditions (instruments) minus the number of coefficients.
Follow us on Twitter @IHSEViews

kelaynak
Posts: 33
Joined: Fri Nov 07, 2008 7:02 am

Re: p value for J-statistics for simultaneous equation (gmm)

Postby kelaynak » Mon Sep 06, 2010 10:58 pm

Gareth,
Thanks for reply and your help.

I am using Eviews system -GMM-Hac
For eviews system-gmm HAC, for the following model , total moment conditions is just total instrument number?
Could you please help me for it?

In 2 equations, I have total 12 instruments (including constant), so total moments condition is just 12? and
thus degree of freedom: 12-9 =3 ?

Thanks

The result is in below

System: SYSTEM1
Estimation Method: Generalized Method of Moments
Date: 09/07/10 Time: 08:46
Sample: 1971 2007
Included observations: 37
Total system (balanced) observations 74
Kernel: Bartlett, Bandwidth: Fixed (4), No prewhitening
Linear estimation after one-step weighting matrix

Coefficient Std. Error t-Statistic Prob.

C(1) -2472.962 553.9893 -4.463917 0.0000
C(2) 1.190021 0.180319 6.599544 0.0000
C(3) -147.1880 43.58633 -3.376929 0.0012
C(4) 21.72342 9.186045 2.364828 0.0210
C(5) -97.10365 19.77433 -4.910591 0.0000
C(6) 2332.957 498.0215 4.684450 0.0000
C(7) -0.008666 0.122840 -0.070548 0.9440
C(8) 4.058377 0.323720 12.53668 0.0000
C(9) 1.344488 0.803866 1.672528 0.0992

Determinant residual covariance 1.20E+12
J-statistic 0.182100


Equation: FDI = C(1) + C(2)*GDP + C(3)*OPEN + C(4)*EXCH +C(5)
*INF
Instruments: OPEN EXCH INF GFCF TB C
Observations: 37
R-squared 0.863444 Mean dependent var 1764.658
Adjusted R-squared 0.846375 S.D. dependent var 4127.667
S.E. of regression 1617.842 Sum squared resid 83757200
Durbin-Watson stat 1.046572

Equation: GDP = C(6) + C(7)*FDI + C(8)*GFCF + C(9)*TB
Instruments: OPEN EXCH INF GFCF TB C
Observations: 37
R-squared 0.971717 Mean dependent var 9670.406
Adjusted R-squared 0.969146 S.D. dependent var 4395.521
S.E. of regression 772.0905 Sum squared resid 19672084
Durbin-Watson stat 0.756237

maman.setiawan
Posts: 1
Joined: Sat Jul 07, 2012 3:08 am

Re: p value for J-statistics for simultaneous equation (gmm)

Postby maman.setiawan » Sat Jul 07, 2012 3:24 am

Dear Gareth and Kelaynak,

I also have a problem in calculating Degree of freedom in GMM methods for the system of equations. I have 4 equations with 4 endogenous variables with 36 total exogenous variables in the system (4 variables and 32 fixed effect dummies). I use all the exogenous variables for the instrumental variables. Total of parameters to be estimated in system of equations are 15 parameters including intercept (excluding 32 dummies).

Another important questions is that how you calculate the HansenJ-statistic in this system of equations? (Do you calculate Hansen J-statistic for every equation and then make all of the Hansen J-statistic from every equation as an average?)

I am really looking forward to hearing your response.

Thank you very much

Best regards,

Maman


Return to “Estimation”

Who is online

Users browsing this forum: No registered users and 8 guests