Because of my misbelieve in capitalism and the belief in the corruption of governments. Or as the saying goes: “All power corrupts” and all that. On top of that, I am no believer in meritocracy (that is I don’t believe people advance in position based on merit).
Now lets do some thought statistics. I—as a bayesean—assign the uniform distribution as a prior to competence, that is I assume that competence is equally distributed across occupations and positions independent of responsibility. Now, lets allow some movement between positions and assume correlation between responsibility and salaries (that is people being payed more are responsible for a greater proportion of the system). Now assume there is not a 100% correlation between competence and being moved into a greater position with a greater salary (that is we assume some level of corruption; people moved into greater position because of favors, family ties, wealth, gender, etc. Even misattributed skills of those promoted [i.e. incompetence of those responsible promoting and hiring; i.e. suspect positive feedback of incompetence]).
Finally we acknowledge the fact that being moved into a position with more responsibility and you don’t gain skills as you are moved, your competence will decrease (that is competence is a function of responsibility and skill).
It should be easy to see that your posterior distribution of competence to salaries and the posterior distribution of competence to responsibility should both be skewed towards the right. That is you as you increase responsibility and salaries, you will find relatively fewer people competent at their jobs.