According to the ICM, the other players benefit if two collide and the shorter stack gets knocked out. That's essentially what happens when two stacks are combined, and if all of the other players benefit, that means the player whose stacks are combined loses equity.

The bankroll you need to play a large multi-entry tournament is not greatly affected by the chance to reach the final table twice. Nor is your equity greatly decreased. The chance to reach the final table twice is just too low for it to make a difference.

I am concerned that many players pass around their favorite numbers of buy-ins to have in a bankroll without justification. This is not safe or accurate. Most players don't play enough MTTs to see how brutal the natural swings are. A player with the skills to have a 50% ROI who starts off with a 100 buy-in downswing will probably not think he is a winning player, and may move on to another form of poker without giving bankroll advice. So, you need to check more than anecdotal evidence if you want to see how bad the swings can be. Here are a few relevant facts:

-- The downswings you see depend on your win rate, and no one can tell you what your win rate is, you have to figure that out. You should not expect the same number of buy-ins to make a marginal winner safe as to make an expert safe, and you will burn through any amount eventually if you are a losing player. Asking, "What bankroll do I need?" is close to asking, "What is my win rate?"

-- The downswings you will have also depend on your mathematical variance, which depends on the size of the field. You will have longer streaks where you don't make the final table if you play in larger tournaments. You should not expect the same number of buy-ins to apply if you play 45-player tournaments as if you play tournaments with 10,000 entries.

-- Expert low stakes MTT players tend to need more buy-ins than expert low stakes STT players, who tend to need more buy-ins than cash game players. It's simply easier to lose a MTT buy-in than a cash game buy-in. However, because these are all called buy-ins, people pass on figures they heard from cash game players, or from SNG players, and that is just wrong. The achievable ROI in MTTs is greater than in STTs, but the variance grows faster.

-- Some people say you should have 100 buy-ins for MTTs, but if that money is really your bankroll not your balance, if it is a disaster rather than an inconvenience if you lose it all, then the math says you should require much more. Check the graphs of players who grind 180 player SNGs who are on the tops of the Sharkscope leader boards (say for total profit 2010). These players are much better than typical players, and they regularly have 50+ buy-in downswings every 1000 tournaments despite having nice ROIs. They have occasional 150 buy-in downswings. If you have a 50% ROI, it is normal to break even or lose in all tournaments you don't win, and to lose big in the tournaments where you don't make the top 3. Of course, you could stay afloat in the short run from occasional large scores, or even a streak of minimum cashes, but you rely on big wins to have any profit in the long run.