The assumption seems correct, but hardly significant in explaining the scale of the phenomenon.

It is probably insignificant for most wars, but it isn't uncommon in history. For example, throughout ancient Chinese history, it was common for bandits to serve in army either after they had surrendered or when government paid them to fight (mostly because these bandits knew the terrain well and had strong combat experience). Many of the bandits were rapists and killers, as one would expect.

And by scale, do you mean the amount of victims or the number of perpetrators?

The latter, as estimated from the former.