Here is an additional issue:
Province 7; regent A has a holding 5, regent B has a holding 1. Regent B transfers his holding to regent A. Regent A now has a holding 6.
As you can see, 5+1 is 6 in this case, even though it certainly involves a LOT less effort than for A to rule from 5 to 6...
It involves 5 GB less effort for B to rule a lvl 0 to a lvl 1 holding than it would for A to rule a lvl 5 to a lvl 6 holding.
On the other hand if regent A were to transfere one level of his holding to B, that holding would have cost 3GB more to rule for A, than it would have for B just to rule his own holding.
I think it seems to complex to create a general rule for transferance cost based on individual holding levels.
Perhaps you could just say that getting your ally to rule and transfere holdings to you because it is cheaper than if you had to do it yourself is against the spirit of the game. And if someone do it anyway you could deal with it on a case by case basis, either using negative events, failed actions, or both?
I think it should be possible to convince the other holding owners in a province to surrender their holdings to you before you have ruled all empty holding slots though.(Even thoug it would be cheaper for yourself to rule your holdings first and then get the most expensive level transfered to you)
As an afternote. I think it is a bit much to make the transferance of holdings cost the same as rule holdings. Sure, temples might have that problem Jon describes, as it is hard to convince a clergy to change faith. But a city guard is in it for the pay. So is a guilder or a trader. (I am not exactly sure about Manors)
Perhaps you could insert a period of decreaced income from the holding, based on how many holding levels were given, after each transferance to reflect the problems of incorperating old enemies into your workforce?...
All in all this is a tough question.
(Sorry for bad gramma, long sentances and general incomprehensibility. I am tired.)