Dear Michael Olsson,

I appreciate your response and I also agree with you on this point. However, I heard of some technique that gives us optimal selection of variables in diverse models.

Thank you very much.

Dear Daniel,

Thank you very much for your response. I appreciate.

First, you need to identify what you trying to test, that is, which of your variables does your theoretical model suggest impacts C02. For example if you are thinking about a Kuznets' type of curve, your variables of interest might by pcRGDP and pcRGDP^2. The remaining variables are just controls. Then sk yourself if the estimated coefficients are robust to different sets of controls, that is the coeffiecient on pcRGDP always positive and the one on pcRGDP^2 < 0, regardless of which controls you use.

I second Michael's answer above, the controls should be determined by theory and what other researchers have done.

Dear Robert Sonora,

I appreciate your valuable response.

Warm regards,

Dear Atif

I hope you will able to get reasonable methodology for your research because recently there are lots of researchers are trying to show the impact of CO2 emissions on various indicators. Hope to get a slice of knowledge from you in future.

I agree with Dr Olsson and Dr Sonora, First develop model base on theory and existing empirical evidence. If you want to test some specific variables include them in the model. before that you have to justify the inclusion of those variables.

Dear Muhammad,

did you find a good way to do so? In my research question I am not interested in making any statement about causality, but just in finding the best fitting model (Designing a put-option).

Thank you and best regards,

Lukas

https://www.ndsu.edu/faculty/horsley/Stepwise_regression_(HZAU).pdf

http://www3.stat.sinica.edu.tw/sstest/oldpdf/A22n36.pdf