Controlling for industries

[Rachel Vos]

Good afternoon!

I'm currently writing my thesis on the moderating effect of national culture on the relationship between CSR and firm performance. Since industry heavily affects firm performance, I want to control for it. In terms of data, I have the first 2 digits of the NACE Rev 2 classifications. In order to employ this as a control variable, would I need to create about a 100 dummy variables, or is there any other way?

[Øyvind Snilsberg]

sysuse auto, clear

*create indicator variables
tab rep78, gen(D)
reg price weight length D*

*prefix the variable with i.
reg price weight length i.rep78

*use xtreg with fe option
xtreg price weight length, fe i(rep78)

[Carlo Lazzaro]

Rachel:
welcome to this forum.
Please note that if you're dealing with panel data, continuous regerssand and -fe- specification (ie, -xtreg,fe-) all time-invariant variables (such as -i.industry-, if -panelid- do not change -industry- during the time-horizon the panel dataset stretches over) will be wiped out by the -fe- estimator.