I think this article makes a convincing case that small changes of this particular type are a waste of time: it's much worse than throwing an equivalent amount of money at the problem.

It follows that scaling them would only waste more time.