On the last point: does that actually happen, though? That's what idealized economic theory would predict, but large corporations' internal structure is quite far from idealized economic theory. My guess would be that if you were able to set up some experiment where one employee cost $100k while somehow having $0 payroll taxes, while another employee cost $90k plus $10k in payroll taxes, the employer would, more often than not, hire the $90k employee.

This happens already, I believe, as regards families. Employees with families cost more in health-insurance benefits than single employees, if the employer offers to subsidize a percentage of health insurance (as most do). But employers don't pay single employees of equal qualifications commensurately more, despite what economics would predict.