The problem with assigning perk value according to the percentage of kills gained is that you would have to flood the arena with 262's, Ta's, Tempests, Ar234's, and whatever comes in the future before their proper ratio of kills stabilizes to the point you could take an accurate measure. Is this something you would want to see?
Ask how many WB vets what they thought of jet day in the MA. Ask them how many of them even bothered to log on. Then ask yourself, "Would people really want to see 262's unperked?" Undoubtedly some would, until they saw 262's in every low furball, in every nook and cranny of the game. It would be a one-plane show in a hurry. Still, in order to bring our jet up to the proper level of kills (to meet the kill percentage perk requirement) you have to allow that to happen.
I prefer to think there are several factors that make perking relevant. How exotic? How powerful? How much potential to dominate? How few produced? All of the factors need to be considered.
Kill percentage alone cannot be the measure of perk, because it would completely upend the game as it is now. Would it be bad? Dunno, but it is a bad idea to radically change anything that is successful without controlled testing first.