CFOs are paid to and should be evaluated on creating/optimizing returns for the owners of the company (the shareholders).
If the return from a project for the company (and the therefore the shareholders) is greater than that of a dividend/buyback, do the project. If the dividend/buyback is the idea with the best return, do that.
This evaluation works whether on a projected/discounted look-forward basis or a backward looking basis. In both cases, you have to model assumptions about the path not taken.
Note that most established companies have a portfolio of ideas. Few operating companies return most of their net profits to shareholders. (some companies setup as financial instruments do/must, and whether a REIT is an operating company or not is a question, but Microsoft certainly is an operating company and returning 100% of net profits, even if that were the best idea they had on first-order evaluation, would be a terrible idea.)
If the return from a project for the company (and the therefore the shareholders) is greater than that of a dividend/buyback, do the project. If the dividend/buyback is the idea with the best return, do that.
This evaluation works whether on a projected/discounted look-forward basis or a backward looking basis. In both cases, you have to model assumptions about the path not taken.
Note that most established companies have a portfolio of ideas. Few operating companies return most of their net profits to shareholders. (some companies setup as financial instruments do/must, and whether a REIT is an operating company or not is a question, but Microsoft certainly is an operating company and returning 100% of net profits, even if that were the best idea they had on first-order evaluation, would be a terrible idea.)