A stereo subgroup is a tool. It's like a hammer. If you need to pound a nail into a piece of wood, then use a hammer. If you don't need to pound a nail into the wood there's no good reason to be banging on the wood with a hammer.
Submix groups are there for combining signals and treating them as a group. For example, you can apply compression or eq or level automation to the whole batch together. Just assigning one track to a group doesn't do anything that can't be done to the track directly.
A small distinction- I don't know if this true in most DAW's, and it isn't in hardware mixers, but in Sonar all track and bus paths are stereo (more accurately 'dual path) unless you force them to act like mono. All the difference is in the content of the audio. If you feed a mono track, as soon as you pan it or insert a verb or stereo effect in there, it behaves as it now is- two paths with a L/R difference. You can pan mono content in a single bus as opposed to needing a pair of sub-buses on hardware mixers (that I'm familiar with.