MULTICOLINEARITY

[Lol38]

Hello,

Im trying to run a programme but it shows the message "near singular matrix. Regressors may be perfect collinear". I have calculated the level of multicolinearity and it is more than 210.

Is there any way to solve the multicolinearuty problem?

[EViews Gareth]

210?

[startz]

Near perfect multicollinearity almost always indicates a specification mistake. In the rare instances where that isn't the case, the only cure is to get more data.

You might want to post your EViews workfile, including the equation you wish to estimate.

[Lol38]

I mean that the multicolinearity is very high.

I have tried with Variance Inflation Factors but even deleting the serie that have the highest VIF, the problem persist.

[Lol38]

It is very difficult to post the workfile because it is a mix from different data sources but here i share my programme.

tesis.prg

(28.72 KiB) Downloaded 315 times

[startz]

You don't need to post the workfile if you don't want to, but no one can run the program without it. Nor is there anyway to tell what the data looks like or which of the 100+ equations has the problem.

[Lol38]

Ok, the equation looks like this.

equation.png (104.68 KiB) Viewed 5369 times

[startz]

Something doesn't make any sense. When EViews says "near singular matrix," it won't estimate the equation.