Thank you for clearly enumerating some of the issues with ETFs. (Though the hidden cost of NTF platforms is orthogonal - that's a problem with the platform regardless of the vehicle, ETF or OEF - and one that doesn't plague most ETFs.)
A couple of smaller downsides of ETFs:
4. Non-commission trading costs - the bid/ask spread and the (petty) SEC Section 3
1 fee (
currently 0.184 basis points, usually passed through to investors by brokerages).
5. Tracking error - this is the "mini" version of your #2 - structural pricing issues. Even when the market isn't in free fall, there is a divergence between NAV and trading price, due in part to liquidity costs (and I guess also due to the fees that authorized participants pay to the sponsor to buy and sell creation units). This is different from the tracking error of the fund with respect to its benchmark index, which is inherent in all index funds.
On the plus side, ETFs may be more tax-efficient than their OEF counterparts. Only "maybe" for a couple of reasons. One is that well run cap weighted index funds rarely distribute capital gains, regardless of the funds' structure. The other is that Vanguard OEFs share the same advantage as their sibling ETFs, because they are merely different share classes of the same portfolios, not clones.
I completely agree with you regarding S&P indexes (not so for Russell, Wilshire, FTSE). This has been obvious since 2000, when S&P methodically swapped out "old economy" stocks for "new economy" stocks, just in time to see its index (supposedly a measure of market performance) underperform the market by several percent.
http://www.thestreet.com/story/10029393/1/the-sp-500-is-a-mutual-fund--and-a-bad-one-at-that.htmlFinally, a note on the CNBC link - Usually, when an article is written saying how wonderfully cheap ETFs are, it gives an "average" equity OEF ER of somewhere around
1.3%. That's an unweighted average and a rather silly figure. Since the purpose of this article is to show how expensive (some) ETFs are, it did the opposite, and gave the dollar weighted average ER of 0.70%. A much better figure IMHO, but without labeling, it seems chosen more to support the thesis than to be objective. (Since no one really cares what numbers mean, let's just pick the "best" one for our point.)