If you have a per-turn pool of roadbuilding capacity that has to be used that turn or it's lost, it will create a couple of related problems:
1. Micromanagement of roadwork. If the player doesn't always have each infrastructure pool building new roads, points will be lost to no benefit.
2. Proliferation of roads. With roads as a waste product of cities, they'll be everywhere and not special.
There are two ways I can think of to band-aid these problems, and both of them are bad:
1. Road maintenance. Now the road system is crufted up with road micro, and the budget is crufted up with the fallout from building roads.
2. Slow growth. Balance infrastructure points so that full-bore road building has restrained results. This is bad because now the player's painstaking road management has only trifling consequences.
The bottom line for me is that road mechanics should reward attention but not demand it, and certainly not demand it consistently. Even a simple Simcity 1 model where you lay out a road and it is built in exchange for money accomplishes this in a modest way -- the player who is concerned with roads can build them extensively, but it's also possible (in a MoM-like game, not Simcity) to play without roads and use other tricks, like a combination of fast scouting parties, large forces massed in cities, and dimensional gates.
But the better way to handle this is with organic roads, where players can, when concerned with roads, build and manage cities with an eye to the roads that will arise in consequence (or cast relevant spells). When NOT concerned with roads, players can disregard the process and still get a road network, although maybe not an optimal road network for their purposes. And players can deliberately sacrifice roads for other things, for example, by building a recruitment-oriented city that's full of houses and mustering grounds but short on anything that would assist road-forming trade links. It's a system of meaningful decisions without being a system of chores.