Model specification

[Justine Borja]

Hello STATALISTERS!

I have a question regarding model specification. Is it okay if I transformed two of my independent variables into their natural logarithms and the others are not? Take for instance this one:

CO2 = LNRGDP + LNRGDP2 + LNOPEN + TFEC + POP + e


Thank you!
Justine

[Carlo Lazzaro]

Justine:
yes, it is.
The main issue rests on the interpretation of their coefficients.

[Justine Borja]

Can you give me econometric explanations why this model specification is fine. The model by the way passed all the post-estimations test, including the RAMSEY Reset Test.

[Carlo Lazzaro]

Justine:
it is not a matter of econometric justification, but of research goal.
You're dealing with a model that passed the usual postestimation tests. So far, so good.
That said, two substantive issues follow:
1) does your model give a fair and true view of the data generating process you're investigating?
2) Are you confident that the regressand/predictors relationships will be (reasonably) easy to disseminate?