Stop right there. If you're doing anything in the mixing stage for the sake of loudness, you're not helping your mix. Do what makes the mix SOUND best. If it's not loud enough, turn up your monitors.
That said, I usually prefer mixes with no buss compression for several reasons. #1 is consistency; one song may have more compression than another, usually too much, and I have to try to "undo" the damage with expansion to help it match the less compressed songs, which is pretty hit & miss. The other reason is 99% of the masters I've done require some EQ. If the compressor is reacting to say, some heavy bass that will get reduced in mastering to sit right, the compression will sound very unnatural. I have two choices there, either let the song stay somewhat bass heavy and make the other songs match or correct the bass and have odd compression that serves no purpose.
I'm not saying compression on the main buss is always bad. I'm saying it's something that I'd never encourage, because only the best mix engineers seem to do it well. Bear in mind, also, that compression on the stereo mix is something the mastering engineer can do very easily with all the mixes in context. So why would you feel the need to use it in the mix?
I didn't catch that. I'll stand by what I said though, if you're doing it for loudness, you're not helping yourself. If you're using compression on the tracks because it sounds better that way, go for it.
I don't think there is a problem with some good mix buss compression for groove, mojo, and some leveling if needed, but not for squashing things for loudness. Too many top mix engineers use buss compression and even buss limiting (Pensado swears by L2 on the '2 buss'), so its hard to argue that you can't use compression/limiting on the mix buss. I refuse to believe that its some sort of standard. Actually, I think if the track is mixed right and the mix-e puts a compressor on the buss, sometimes the mastering engineer may have less to do. Just my opinion.