If you assume that people are promoted to their level of incompetence -- terminal responsibility level, then you would expect that level adjusted performance should approach a Gaussian?

No, because there simply aren't enough high-level employees at the top in any given company for a meaningful sample. You'd have to compare across companies; I guess the stock market does that indirectly.