Two starting comments
Why are there so many different index products, under the S&P, FT, Dow Jones, MSCI…brands, that do basically the same thing? Index-tracking is a commodity product. Given a certain level of technical competence in design, to make sure that the returns are in line with the appropriate index, one product is basically the same as another. Having slightly different benchmarks with slightly different constituents, requiring slightly different holdings in the ETF/mutual fund, raises the costs for an institutional investor to switch from one provider to another. So it takes the edge off what would otherwise be brutal price competition.
Does an S&P 500 index product have 500 stocks in it? Maybe, maybe not. The ETF/fund prospectus will say for sure, but all that’s usually required is that the product have a basket of stocks in it that tracks the appropriate index with a high degree of accuracy. It’s much easier to handle inflows and outflows if you don’t have to deal with the most illiquid stocks in the index.
The ETF/fund websites should have charts or other information about “tracking error.” i.e. how closely the index product mimics its index.
Domestic indices
This process is pretty straightforward.
An index mutual fund is priced once daily, after the New York close, using closing prices of its constituent stocks. Shares are bought and sold at that price, the Net Asset Value.
In the US, index calculations are made and disseminated electronically every 15 seconds during the trading day. Index ETF market makers use this information and their sense of supply and demand to set bid and asked prices for the index ETF throughout the day. Index ETF providers publish comparisons of the last trade of the ETF with the ETF’s NAV. Differences are typically less than .5%. (This is not the same as saying that this relation holds between ETF and index throughout the trading day.)
Indices with non-US content
For mutual funds that hold non-US securities, the situation is a little more complicated.
Let’s take a Japan-only index mutual fund as an example. Daily trading in the Tokyo market, in yen, of course, starts at about 8pm EST and ends at about 2am. Under normal circumstances, the NAV is calculated using the closing prices in Tokyo and the 4pm New York $/¥ exchange rate.
This leaves a fourteen-hour gap between the time the prices were set in Tokyo and the time they enter the NAV. New information could emerge during that period that would have changed the Japanese stock prices had Tokyo investors known. If so, the NAV should be adjusted to reflect this new information. This kind of adjustment is called “fair value pricing.” The test of whether it is necessary is usually whether the S&P 500 has changed that day by more than a specified amount, say, .5% or 1.0%. If so, a third party with considerable statistical expertise estimates, stock by stock, what the closing price in Tokyo would have been if Japan had known how the S&P would close that day. The NAV is then calculated using the estimated prices. (In my experience, these service providers turn out to be amazingly accurate, judging by the following day’s opening prices in foreign markets.)
International ETFs do not use fair value pricing in calculating NAVs. They use local market closing prices and 4pm New York currency prices. Their reasoning in not making price adjustments, I think, is that the market makers will factor the new information into their spreads, and investors will do the same in setting their bids.
I understand that this is the simplest, lowest-cost way for ETFs to operate. And, if you argue that the market is always right, the procedure gives the market the opportunity to express itself through out the day. But it seems to me this gives a very big advantage to the market maker, who is able to make an informed guess–even if he doesn’t hire the kind of outside service the mutual funds employ–about what the “real” NAV should be.
It also looks as if this sense of where NAV should be gets overwhelmed by the forces of supply and demand in the market during times of stress. The i Shares Japan ETF, for examples, reported two days during October 2008 when the last trade was 11% different from NAV–once a premium, once a discount. The widest deviations from NAV during the first quarter of 2009 were +/- 5%.
I haven’t looked at this situation carefully. It may be, for example, that in the cases I mention above, the last trade comes long before the close. Or it may be that London trading in Asian stocks + Tokyo futures in Chicago + US market movements justify the wide deviations from NAV. On the other hand, even 5% seems like a lot. There may be a chance here for an astute trader to make money. Be sure you investigate the variations from NAV carefully, though, before you do anything!