Matt Di Carlo goes a little deeper than most on the question of how we'd actually change salary schedules mid-stream to the flatter, front-heavy system contemporary reformers mostly advocate. Yes, higher pay for new teachers would be good, but also teachers really don't like being on top-step. That is, they like the money, but they don't like being stuck with no chance for a raise other than when the whole scale is changed for a new contract. Performance pay doesn't really fix that either, because the best teachers should in theory hit the top performance bonus fairly quickly. Adding promotion to leadership roles as teachers is expensive too, even if you don't give people financial raises for that, because you're taking those teachers out of the classroom.
My point here is not so much that the above strategies wouldn't work, but that if you really game them through, they'd result in paying more teachers more money, you might not improve retention very much (adding two years on average instead of 10, say), and find yourself paying more for the same number of inexperienced, less effective teachers.
Especially if the underlying motivation here is to get more ambitious, go-go, high achievers in the teacher pool. They are the people least likely to appreciate hitting the ceiling five years into their career.