That's rather simple. Treat the dead PCs like they're...people. They may be dead, but they might have families who expect to inherit that loot. There could be funeral costs. Maybe if they rob from the dead anyway, a ghost of the fallen PC haunts them or some curse afflicts them...
In any case, DM controls treasure output. If the party plays keepsies with the slain PC's gear, just throw them up against monsters that drop no treasure until you feel their wealth has been properly balanced (obviously, factor in the rest of the party's sudden wealth gain and the upcoming gp drought when telling the player of the deceased PC how much gear his new PC can start with so he isn't screwed over). Heck, be blatant about it, don't even try to disguise your intentions. Tell them flat out why they're not going to be fighting any dragons or NPCs with class levels (or other things that give tons of loot) for the next 10 sessions at least, until they realize trying to game the system like that is just paddling upstream.
And hence Balance by DM Fiat is perfectly described.
I don't have much to contribute, but I do feel that a part of changes in how balance is achieved in early vs later D&D editions is due to changes in the expected rate of character turnover in early vs later editions.
The older editions, the focus of balance was between the players. It didn't matter as much if you rolled up a poor character or made poor character build choices because the character will most likely die a horrible death and you get to make a new one. You get to spin the roulette a lot if you wanted, as it were.
In newer editions, there's an expectation that you'll play your character for quite some time. You may be stuck with a character with poor stats or poor built for an entire campaign. This causes a shift to focusing balance between characters, since character balance = player balance.