Mark

Introduction#

You start off a little bit colloquially. Maybe you can start from an optimal contracting perspective (take e.g. a textbook like Tirole, Theory of Corp Finance, or Bolton & Dewatripont, Contract Theory, and read the introduction) to see how they word the same ideas you enounce a little bit more formally.

Based on this reasoning, this paper will aim to obtain a better understanding of the factors influencing forced CEO turnover

Living up to the contract can encompass many specific things; corporate governance books usually talk about entrenchment, empire building, using resources to favor interest groups, etc. etc. as ways in which CEOs can waste resources. Try to provide a bit more context.

Can you explain why you test the conjectures in a US sample? Why is it a good setting?

On average, there do not seem to be any negative consequences for a forced turnover. Denis and Denis (1995) found that when a turnover is forced there tends to be a positive abnormal return upon announcement, which does not seem to be the case when the CEO left due to other reasons. This, therefore, indicates that shareholders tend to approve of forced turnovers. A possible explanation for this could be that forced turnover are more likely to occur after a poor performance, especially when the stock performed badly (Gao, Harford & Li, 2012).

I think it is a little bit confusing to start talking about the consequences of CEO turnovers, because that is not part of your study: it is just the factors leading to CEO turnover, and not the consequences of CEO turnovers.

Hypotheses#

  • They are not specific enough - you have to take a stand and say how they influence the relationship? Do they mediate the relationship (make the relationship less negative? and for what ethnicity? (ethnicity) and in what direction (% of women)?)

  • Secondly, what about the gender/ethnicity of the CEO itself? Do they have a smaller/higher probability of being replaced with equal performance decrease/increase?

I think your data, and also your theory, allows you to hypothesize about this as well.

Method#

  • I think you can use several other asset pricing models to test the robustness of the results to the specific asset pricing model used.

I would advise to do Probit (easy to estimate panel models):

  • Stata: xtprobit
  • R: pglm

and also using OLS (as you write): Two-way FE estimator (Time and Firm-specific effects)

  • This R Package contains a slightly different estimator (which is more robust than the TWFE estimator), Stata might have it as well.