Page 1 of 1

Singular Variance Matrix in Multinomial Logit Model

Posted: Tue May 12, 2020 9:30 am
by cicada3301
Hi!
I am currently working on a multinomial logit model and have written the following code:

When running the program, I do get coefficient values. However, I do not get any standard deviations because of the error: WARNING: Singular covariance - coefficients are not unique. I tested for perfect multicollinearity and this did not seem the case. The variables displ1,..,displ6 are not summed to one per row (sometimes both displ2 and displ6 are one for example and sometimes they are all zero).

Can anybody find my fault? That would be of great help!

Thanks in advance,
Kind regards

Re: Singular Variance Matrix in Multinomial Logit Model

Posted: Tue May 12, 2020 9:36 am
by EViews Gareth
Hard to say without the data

Re: Singular Variance Matrix in Multinomial Logit Model

Posted: Tue May 12, 2020 12:41 pm
by cicada3301
This is the data!

Re: Singular Variance Matrix in Multinomial Logit Model

Posted: Tue May 12, 2020 1:03 pm
by EViews Gareth
Looks like a bug in your derivatives. Try removing the derivative code and using numerical.