Hei,

I am doing a fixed effect model on a panel of N = 1089 and T = 30 000 (ish).


ESGscore = beta1*Ownership concentration +beta2* identity of the largest owner+beta3*Ownership concentration*identity of the largest owner + control variables.

ESG score is measured from 0 to 100.
Ownership concentration is measured as the relative ownership share of the largest owner ( largest owner share/(second largest ownership share + third largest ownership share)).
Identity of the largest owner is a dummy variable taking the value of 1 if the largest owner is a corporation, 2 if it is the government, 3 if it is an individual, 4 if it is an insider, 5 if it is an institution and 6 for PE firm.

I would like to do an interaction effect between ownership concentration (continuous variable) and identity (dummy variable) to see if the effect of ownership concentration has varying affect on ESG score conditional on the identity of the largest owner.

The problem however, is that the identity of the largest owner changes multiple times within unit over time. Thus, creating gaps within units when stata is generating the results.

Therefore, I am thinking that the interaction effect (and the dummy variables independent of the interaction) does not work for a fixed effect model.
Is that correct? What should I do?