Tomorrow I will buy 10 shares of XYZ at any price, and everyone knows this. XYZ is worth $100/share and today that is the price. If one other person in the world owns XYZ shares they can sell it to me for $200,$300, whatever price they name. If 2 people own shares, there will be competition, and I'll buy at the cheapest price available. If 1000 people own shares I will still have to pay more then $100, but if there is true competition on price then I should only pay slightly over $100.

Say company X is getting bigger, and pretty much qualifies to get into the S&P500 at the next revision of the index, it currently trades around 100.

When it gets into the S&P500 all the ETF buyers will buy this stock, that will push the price up to 110. Some of them will buy it at 100, some will buy it at 110 and some anywhere in the middle.

Now a bunch of hedge funds anticipate the ETF will do that and start buying it ahead of the stock going to the index. The price of that stock will rise to 102, 105, perhaps 107 if you have lots of people doing this arbitrage.

When the ETFs buyers pile in, they will be picking up the stock somewhere between 105 and 110, not 100 and 110.

Now one can argue this is the problem of the bank maintaining the ETF, because as far as the ETF investors are concerned, they own something that tracks the index, if it costs more to replicate the index, it is not their problem. But the way I see it is that the new stock will enter the index at an inflated price, and the price will converge again to the fair value when all the buying is done (instead of the price going up after the stock joins the index and converge again). That means that the ETF investors will have a lower performance than what they would get if no one was arbitraging them.