Yeah, I don't think that is gonna be as easy as 'throw a mesh deformer in the group and chuck the cage in an emitter'...
I suspect the problem is that the emitted cages don't exist at the time you must bind / initialize them to the high res model.
Now, with a cloned setup we can circumnavigate that problem by duplicating the high res cloner for a low res equivalent, and then using correction deformers / connect objects to make one deform the other as a whole, but not sure that is possible with an emitter. Here's the procedure for the cloned dynamics - perhaps you could try something similar with the emitters, though for some reason I don't like your chances...
Look forward to hearing what everyone else has to say about this...
CBR