It's not that simple, though, when you are talking about papers with multiple authors. It doesn't take into consideration to what level of involvement any particular author was. It's not uncommon for authors to be listed due to small contributions, or insight, or internal politics. At what point do you say their contribution was significant enough to warrant exclusion from impact factor calculations because of self-citation? And how do you even quantify that level of contribution?
A principal supervisor of a group may not have much involvement on a particular paper, and yet expect and receive an authorship of it. Do you reject self-citation in that case, when the authors who actually did the work are not directly affiliated with the other papers their supervisor was previously a part of?