I think it is unavoidable, because if you are using division of labour to generalize towards comparative advantage, then you are basically describing interactions between (at least two [wool? wine?] different "employments." In that case at minimum, unemployment is another (third) "employment" kind that needs another variable to represent it.