In the 2x3+ team game we had a great debate over whether they should go exclusively to the best units, or whether all units should have some sort of minimum percentage strength which was far more than just enough to keep them ready. In the latter case the overall average statistics would be lower. Note that replacements do not simply fully dilute the experience and morale of the unit they arrive in. By sending replacements to better units rather than worse units they in effect become higher rated replacements themselves. see https://www.matrixgames.com/forums/tm.asp?m=4699518
To illustrate the difference we created the following example with hard numbers
There are three groups of units which I have called Elite, Workhorses and Dregs. There are 280k men in the pool and I have assumed we want to keep a buffer of 43k.
Two methods
i) We aim to maximise the morale/xp/cv regardless of unit distribution. So we start switching units with the best stats up to max 100% ToE lowering the threshold for them until all the pool will be used except the buffer we want to leave. First we set all the Elite units to a maximum of 100% ToE, but find there will still be many more replacements in the pool than we need. So we lower the threshold some more and next we move on to the workhorses. Now we find all Elite and Workhorses set at maximum 100% ToE exhaust all our manpower pool except the margin we want to keep and so we stop lowering the threshold. All units below (the Dregs) are turned down to 33%, just enough to prevent them from being unready.
ii) We aim to maintain a minimum ToE for all units. In this case giving every unit at least 67% maximum ToE will still leave enough to get all the Elite units to a maximum 100% ToE.
Let us assume that there is a "North" sector consisting of five hexes each facing the enemy on two hexsides. With the second method each hex could be defended by a Dreg unit of 12,000 men with a defensive CV of 10.4 per enemy hexside. In the first method you would need each hex to be defended by a Dreg unit plus at least a regiment from a Workhorse division to do as well. This would be exactly the same number of men, 12,000, but would have a slightly higher defensive cv per enemy hexside of 11.0
The first method does produce higher average morale, cv and experience per unit and higher total cv for the whole army overall. In our assumed North it would also have slightly higher cv for exactly the same number of men although using more units. The rest of the front would also have exactly the same number of men for roughly the same cv but they will be concentrated in fewer units. This is an illustrative example only. But slightly different numbers could mean the North would be defended by the same CV for a smaller manpower for example
So essentially the debate is do you want to maximise total cv/morale/xp of your total manpower and feel you can split up and relocate units to do whatever is needed? Or are there some combinations of split up and relocated units that could not do that like was done in this example. Or is there something intrinsic about a division that cannot be replicated by some other combination with the same CV and manpower in the same hex? Or do you need to sacrifice some of the maximum possible potential of your manpower in favour of particular units? I think is a global view versus a local view and probably the answer is a bit of both but the judgement is to what degree of each.
