This shifted the race of trainees, but it doesn't seem to have changed the more important metric of how many people were actually hired. The author claims it had an effect, but as far as I can tell he's never quantified it.

The real issue is just insufficient slots.