Orthogonality C test - matrix problem

[Ties]

Hi all,

I have to test for endogeinity and when performing the orthogonality c test using TSLS on my instruments I get an error: "Weighting matrix is of wrong size". What does this mean and how can I solve this?

My formula is:
stn = c + hicpp-2 + ygapp + r + r*(hicpp-2) + r*(ygapp) + ar(1)
and white robust standard errors

Instruments are pcr (private consumption), nxr (net exports), itr (gross investment), compr (comodity prices), urx (unemployment rate).


Kind regards,
Ties

[EViews Gareth]

To which equation object are you referring?

[Ties]

taylor_white_ar is the equation I am using

[EViews Gareth]

We don't support equations with AR terms with automatically added instruments for the ar terms for that test. The error message should be more informative, sorry.

[Ties]

So I should remove the AR(1) for the equation? But still my model needs to be corrected for serial correlation and heteroskedasticity. How do I solve that?

[EViews Gareth]

You can keep the AR term in, as long as you decheck the box that adds lagged values of regressors as instruments. But note if you do that, you'll probably need to add more instruments yourself so that the model is identified.

[Ties]

I have dechecked the box for including lagged regressors for linear equations with ARMA and added one more instrument (therefore it does not give the error "not enough instruments" anymore). However now I get an error stating: "singular matrix". How do I solve this?

-edit: could I be suffering from a dummy variable trap?

The new workfile is added to this comment (hence that I checked the box to be able to save the equation).

[EViews Gareth]

Probably need even more instruments

[Ties]

Need those instruments be in percentages? (The data for estimators are in percentages)