On the selection of variables for testing panel data assumptions

[volkan kale]

Hello. I would like to get some information from you about something I'm curious about. When performing heteroskedasticity and autocorrelation tests, should we use the stationary forms of the variables or their original forms? I would greatly appreciate your help on this matter. Thank you very much in advance.

[Maxence Morlet]

The form of the variable that will be used in the regression

[volkan kale]

The form of the variable that will be used in the regression

Thank you so much!