Re: The GMM estimation specifiction does not work, please he

[somaiiehmr]

Hi guys,

I'm almost new with Eviews and am trying to estimate a non-linear GMM model with some instruments, but when I estimate the model using estimate and choose the GMM -Time series (HAC) the following error would come up:

No valid observation in equation, can you please let me know where is the problem?

Many thanks in advance for your help.

[EViews Gareth]

Do you have any valid observations?

[somaiiehmr]

Yes, I think all of them are valid observations.

[EViews Gareth]

You should post your workfile.

[somaiiehmr]

Here it is my workfile.

[startz]

Try changing the starting values in the C vector.

[somaiiehmr]

Hi,

Thanks for your reply, I tried to estimate GMM using GMM1 system in the attached workfile and I got table01 results, but surprisingly when I tried to re-estimate the same system the foolowig erroe came up: near singular matrix! why is that so?

[startz]

EViwes uses the values in your C vector as starting values and also places the final estimates in C. You probably have unreasonable values in your C vector.

[somaiiehmr]

Yes, you are right I have changed the starting values and it now works. Thanks a lot.

Just one more question. I also would like to find the following values and their standard errors using the estimated c(1), c(2) and c(3):

phi= c(2)+c(1)*(1-c(2)*(1-c(3)))
lambda= ((1-c(1))*(1-c(2))*(1-c(3)*c(2)))/phi
gammaf=c(3)*c(2)/phi
gammab=c(1)/phi

But I have no idea how to do that. Shall I add them to the gmm1 system? and how? if not how I can calculate them.

tahnks again.

[startz]

EViews won't do this directly, but you can fake it by going View/Coefficient diagnostics/Wald and entering tests along the lines of 0= c(2)+c(1)*(1-c(2)*(1-c(3)))

[somaiiehmr]

Thanks it works.

[startz]

I have no idea about this one.