if the moment would create a couple between the two studs, then it probably is real. if the moment is "merely" bending the bolts, then a simply supported connection is an option; but either ...
1) you completely ignore the moment (and release it in your model), so what you're modelling is the situation after the bolt has yielded [a simple approach, but unconservative; but if you've got a large MS ...], or
2) apply both moment and axial loads to the bolt, assume the moment is equal to the limit moment strength of the bolt and this'll reduce the axial load you can apply; i'd combine as Rl + Rm = 1 (or Rl^2+Rm^2 = 1) Rl = applied axial load/ult allowable, Rm = applied moment (= limit)/ult allowable moment. to model this i'd model both full fixity and simply supported and combine the results (if yield moment is 70% of the moment generated in the fully fixed model, then the combined internal loads are 70%*fixity + 30%*SS)
clear as mud ?