> So if men are more likely to work from the office, and managers retain a bias towards in-office workers, we should expect to see managers over-rewarding male employees at the expense of female employees
This is frustratingly wrong in at least 2 different ways.
1. If an employer over-rewards a group of employees, it's at the expensive of the employer, not the other employees.
2. Even if you believe that other employees are responsible for paying the wages of the over-rewarded employees, then if the over-rewarded employees are those who work from the office we'll just see managers over-rewarding in-office workers at the expensive of from-home workers. A woman who works from the office would be over-rewarded, and a man who works from home would be paying. To say that men will be over-rewarded at the expense of women is to throw away a high-quality signal in favour of a lower-quality signal.