GMM Error "Order condition violated-Insufficient instrumens"

[osnola]

Hello. Please help me with this problem.

I'm estimating a equation with 4 coefficients using GMM [with the following options: (1) HAC (Newey-West); (2) Iterate to Convergence;] and using 7 instrumens (in especific i'm using lags of variables like instruments).
I have more instruments that coeficients and Eviews send me the error message "Order condition violated-Insufficient instrumens", I don't know what is happend with this?

And in some cases, if I change the order of the instrument (the same instruments but in different order) Eviews can do the estimation.
The order of the instruments it's important?
Can you help me? Thanks

[EViews Gareth]

Post the workfile, along with the equation specification.

[osnola]

Hello, here are more specific details and the Workfile in Eviews 8.

Specification:
GMM(METHOD=CONVERGE,INSTWGT=HAC,NOCINST) INF=C(1)*INF(1)+C(2)*GAP+C(3)*PT+(1-C(1)-C(3))*INF(-1)+C(4)*D2010M1

1) So, with this instruments order i can't estimate, because Eviews send me the error "Order condition violated-Insufficient instrumens";
@ infus(-7 to -10) inf(-1 to -1) gap(-0 to -0) dner(-7 to -7) d(d2010m1)

2) when i change the order of instruments Eviews can do it without a problem
@ inf(-1 to -1) infus(-7 to -10) gap(-0 to -0) dner(-7 to -7) d(d2010m1)

I'm confused about this problem. This is happening me with differents sets of instruments.
Thank you.

[EViews Gareth]

Looks like the weighting matrix is very close to singular, and is just failing the singularity check with one order, and just passing with another order.

[osnola]

OK, but then the error message is incorrect, didn't?
There are a problem with set of instruments but isn't this "Order Condition Violated - Insufficient Instruments".
Thank you

[EViews Gareth]

When the weighting matrix is singular, EViews drops instruments so that estimation can occur. If it drops too many, eventually you hit insufficient instruments.

[osnola]

It has been a great help. Greetings and thank you very much.