kcowan wrote:When bond issues are reopened, is it always because of insufficient demand by the original closing date?
I don't understand what you mean.
like_to_retire wrote:So my thought was always, "why don't the sellers to the ETF's see them coming and charge a higher price, since (unlike individual investors) they can't walk away"?
They can't walk away but they have a lot more choice than the Fed in their investments.
They can, for instance, participate in a new issue. Or (in most cases) they can park the money in a benchmark bond and gradually swap out of that into what they want. Or they can simply pay up and the difference will be their tracking error.
Note as well that when they issue new units, they will typically get a basket of securities in exchange, and so don't have to go to market.Added Later:
Which is not to say that they don't overpay and never get taken to the cleaners. Any systemic tracking error (in excess of MER) is evidence that they're overpaying (or being underpaid). Typically, an ETF's trading is done by a clerk (or jumped-up clerk with regulatory status) who aren't much good, don't much care, and have very little discretion anyway.
The most egregious example I know of with respect to an ETF being taken to the cleaners is CPD buying POW.PR.C last January
... and we only know about that because the market is small, CPD reports positions daily, and the market reports transaction prices.
This is my big problem with the regulators' beloved Trading Expense Ratio (TER). It only reports the direct, reportable, costs of trade - the commission and settlement expense. Unless the fund is paying a ridiculous commission, the less visible Market Impact Costs (which include the bid/ask spread, and can only be estimated in any case) are going to dwarf the commission. Which is not to say that commission costs should not be reported of course; only that TER doesn't even begin to tell the story.
I should also note that tracking error doesn't usually tell the whole story either, since the index itself can experience "meta-tracking-error" (vs. a notional constant superset which contains all issues that are added and deleted over the period). This is because index changes are typically announced a week (or so) in advance of the change; in the interim, relative prices move with the biases one might expect. <ad> I discussed this with respect to TXPR and preferred shares in the September, 2010, edition of PrefLetter </ad>