In panel data, we control for firms and years fixed effects even we also have some time-variant firm-level regressors.
I am wondering whether it also happens at the industry level. If it is the case, why Alminas,2019 control for the industry fixed effect but excluded the industry control variable on column (4) of Table IA2 as below:
And I am wondering why we need to control for firm, and year x industry effects at the same time? Whether they duplicate each other ? What exactly when they control for all firm, industry x year fixed effect?
Update: I found a discussion partly relating to this but mainly about the mathematical side.