With those 1 or 2 percent raises being a norm almost everywhere, it's hardly a reason to actually leave the company. And saving for not doing 5 to 10 percent on everyone every year are much bigger than an occasional hiring and training costs.
But it is a reason to leave the company because you can easily get 10% or more switching jobs.
>occasional hiring and training costs.
Hiring and training costs are enormous, and they aren't occasional when you're talking about software companies with very high turnover. If you're incentivizing job hopping every few years, the average programmer at your company could be spending 25% or more of their tenure ramping up to full capacity.
Depends on the company. Mine is 100% remote since COVID, has generous benefits, a strict 9-5 M-F culture, and plenty of interesting work to do as much (or as little) as I have an appetite for. I make maybe 10% under market rate, but it would take at least a 50% raise to get me to leave. That's never happening, as it would be well over what anyone short of FAANG would pay for a senior dev, so there are definitely intangibles about a work environment that are worth more to me than another $15-$20k a year.
Obviously you can get away with paying below market rates if you provide some other benefits to make up for it.
Only if you have people who crave stability more than they need/want more compensation or who are invested in the work with you (i.e. they find it _really_ interesting). Otherwise it just creates churn, as people need to leave to get much of a bump up and leave the next place next time they want a bump up too.
So it only works overall if most people are in the "wanting stability" group.
This is probably part of the driver towards contract work rather than perm: if you need to switch jobs regularly to keep improving your conditions, then why not have the extra flexibility and/or extra compensation of working contract?