I think you should not worry about it as long as you are able to explain it. Basically, you are removing the effect of unobserved heterogeneity that does not change over time.

Dear Justine.

It is normal for coefficient to change in this type of regression.

I also revommend5@Stephen Potitzer-Ahle.

Regards

You can rely solely on the fixed effects regression result. No one will put much stock into the pairwise Pearson correlation coefficient, as it does not account for any other factors that may impact FRQ.

You can find lots of weird correlations in the world (e.g. ice cream sales correlated with robberies). They don’t really tell us much, however, so it makes sense to include a well designed set of controls.

The other thing you might consider exploring if you haven’t already is, does firm-level FRQ change when the CEO changes to an older or more-experienced executive. If you find evidence of this, it would help you with a causal argument.

Thanks for the energetic discussion.

@Steven Kozlowski @ Kehinde Mary Bello@ Stephen Politzer-Ahles's and @ José-Ignacio Antón Do you have any literature reference that will support your arguments? just curious to know.

Thank you everyone for the discussion and helpful answers!