Philosophically, "reality" is always a matter of hypothesis, so one should make one that is as useful as possible (however, if for some or other reason, you don't like that, as many things in philosophy, you can change your mind about it and do something else).
You have to realize that the question "is xxxx real", is a philosophical question, not a purely scientific one (which doesn't mean that it doesn't have any importance!). Now, only the scientific method can eventually give clear answers to only scientific questions, and hence there will never be a purely scientific answer to a question "is xxxx real".
Now, the closer the theory at hand is to daily common sense experience, the less this question comes to the foreground, because we have a strong intuition for "what is real, and what not", and that intuition is compatible with most things in basic Newtonian mechanics.
However, the more the theory at hand describes phenomena which are removed from our daily experience, the less one should rely on one's intuition to declare something "real", because doing so can put you in (unnecessary) trouble. Electromagnetism is already quite remote from our daily experience, and the concepts used in there are thus more open to pondering as to whether to take them real, depending on one's ontological a priori.
An extra difficulty comes in when two empirically equivalent formulations of the same physical phenomenon, with different formal elements, exists. For instance, it is possible to reformulate EM without EM fields, but just using retarded effects of one charge onto another. In that (rather clumsy) formulation, there are no fields present, and in as much as one assumes that all EM radiation found its origin in some movement of a charge (and was not an "initial condition" of the field itself), both are empirically equivalent. So one has to pick one's personal choice in setting up one's "reality hypothesis". One can debate over other people's choices ad infinitum.