Persistent Multicollinearity Issue

[John Granger]

I am working on a multivariate regression model that examines the relationship between an index of state consolidation as the dependent variable and a number of predictor variables. I have conducted some basic tests of the predictor variables and transformed them, etc.
However, the correlation matrix and other tests seem to show multicollinearity between some of the predictors. Theoretically, I do not see how they are related to each other. Accordingly, I tried to center and standardize the predictors. I used the commands beta and another called stdbeta. I also created some modified predictors the old-fashioned way subtracting from the mean and dividing by the standard deviation. The stdbeta command did not work and the multicollinearity persisted using the modified predictors.
I am at an impasse and work greatly appreciate some insight into this.
I have attached the do file.
Thank you,
J. David Granger

[Joao Santos Silva]

Do you really only have 5 observations? That is what is causing the problem!

Joao

[John Granger]

No. There are 26 time periods (1990-2015) and 16 countries. The panel is balanced. I cannot figure out why this happening.

[John Granger]

These are the Obs, etc. some of the predictors.

Variable | Obs Mean Std. Dev. Min Max
-------------+--------------------------------------------------------
GEO | 416 .375 .4847058 0 1
BOR | 416 .625 .4847058 0 1
EFR | 39 .2586196 .1642531 .023712 .6837
DEM | 416 2.122596 1.113172 .5 5.5

[John Granger]

If you don't mind, I can send you the actual file.

[Joao Santos Silva]

Check the data carefully. Maybe you have missing observations in some variables or some variables are strings. For instance, in the results above you only have 39 observations for EFR. You need to check why this happens.

Best wishes,

Joao