This strategy can be sub-optimal from the employer's perspective.
Consider yourself as an employer and assume your employees have perceived switching cost C > 0.
Your employee will approach you with bids b_1, ..., b_n all of which are substantially larger than C. So to keep your employee you will need to pay C plus some premium. In fact, not just any premium, but probably something close to max(b_1, ..., b_n).
If you can make good guesses at C and b_1..b_n, then you can retain employees by ensuring you're always giving raises between C and max(b_1, ..., b_n).
And all of that assumes that employees are fungible at any particular price point. They're not; employers also have switching costs. Among two otherwise equivalent employees you'd much prefer the one you already have.
This might be true for FAANG employers or others employing people in high demand, but based on the behavior of the vast majority of employers, it is clearly less costly of a problem than the savings of not having to pay prevailing market prices for all their employees.
Consider yourself as an employer and assume your employees have perceived switching cost C > 0.
Your employee will approach you with bids b_1, ..., b_n all of which are substantially larger than C. So to keep your employee you will need to pay C plus some premium. In fact, not just any premium, but probably something close to max(b_1, ..., b_n).
If you can make good guesses at C and b_1..b_n, then you can retain employees by ensuring you're always giving raises between C and max(b_1, ..., b_n).
And all of that assumes that employees are fungible at any particular price point. They're not; employers also have switching costs. Among two otherwise equivalent employees you'd much prefer the one you already have.