The game balance would depend on what the devs would decide, so what follows is hand-wavy illustrative waffle...
I guess that having, say a couple of Sov Vs together would make them mutually less expensive to run, because of the local feeling of security (among the ordinary people at least) that extensive lands would bring. Likewise, having confederate or allied towns immediately nearby helps, and conversely enemy towns or armies would make Sov less practical (I'd be nervous working on the east edge of a territory where you can see the smoke of enemy fires, and provided I'm not a slave, I'd be wanting reasonable compensation).
At the moment, real security is not much use, because armies can ride roughshod over any terrain or ownership to reach any square, so a pathfinding weighting based on Sov would help implement a practical aspect to security, and bring the cost down that way.
I write all this in the knowledge that the Sov cost is not purely bribery money, mostly reflecting the cost of central provisioning, but even then, I'm sure that there are economies of scale, which would be reflected in the modified Sov cost mentioned above.