ETF Net Asset Value Not Always Best Indicator of Short-Term Value


Initially, critics of exchange-traded funds (ETFs) decried their trading at a significant premium/discount to net asset value (NAV), the underlying stock values they represent. This is a problem that regularly plagues closed-end mutual funds, where prices of the funds are generally lower (sometimes by 50% or more) than the NAV of their portfolios due to a lack of liquidity.

So far most critics appear misinformed. The biggest flaw in their argument is their selection of end-of-day NAV figures as the measure of the true worth of a portfolio. End-of-day NAV will always be an imprecise measure of a stock portfolio, especially one with small capitalization and international equities.

Consider an ETF trading in the U.S. that represents an index for a country on the other side of the world. The U.S. exchange trading the ETF is open while the exchange trading the underlying stocks is closed. The official end-of-day NAV is many hours old and will reflect yesterday's news, while the ETF trades actively and reflects today's news. Which is going to be a more accurate reflection of the value of the underlying stocks? We put our money on the ETF.

To test this theory, Morgan Stanley Dean Witter conducted a study on July 5, 2000 following after-hours price movements in U.S. markets of the iShares MSCI Japan ETF and after-hours Nikkei 225 futures contracts, two instruments representing the same Japanese stock index.

The study started at the close of Tokyo's exchange, when the official end-of-day NAV was recorded and Japanese traders there went to bed. The ETF and futures soon began to trade in New York and meandered throughout the day, incorporating changes in investor sentiment, currency exchange rates, and a myriad of other information. The ETF and futures remained nearly in lockstep, moving inexorably away from the official end-of-day NAV. Lo and behold, when the Japanese stock traders arrived for business the next day, NAV at the open market resumed almost exactly where the ETF and U.S. futures left off, and far from the previous day's official NAV. All of this is consistent with basic trading theory.

Chart courtesy of Morgan Stanley Dean Witter. Originally appeared in an August 2000 Global Equity and Derivative Markets feature "MSCI iShares and their Relationship to Net Asset Value," and is reproduced with permission.

A similar situation occurs with a U.S. ETF replicating U.S. small cap stocks. Both types of instruments trade during the same hours, but they certainly don't all trade as frequently. Many small stocks have relatively few interested investors and trades are sometimes separated by hours, so that NAV figures - on average - represent relatively old information. ETFs, on the other hand, tend to trade by the minute, if not the second.

In such cases bid/ask spreads of the small stocks tell more about where the next trade is likely to take place than the last trades used to construct official NAV figures.

More importantly, the ETF also rarely strays from within the bid/ask spread. In other words, the last price paid for an ETF that claims to accurately represent a portfolio generally lies between what all the sellers of stocks of that portfolio want to receive and buyers want to pay. This is all just as it should be.

This can be seen from a study by Salomon Smith Barney of ETF activity during September 2000. The study shows that ETF prices during taken at multiple "snapshots" mid-day stray on average an incredibly small 0.014% from the midpoint of the bid/ask spread of underlying stocks. The midpoint has no magical meaning, but seems like a sensible spot to locate fair pricing in absence of recent actual trades. As expected, the variation around that midpoint is higher for small cap and sector ETFs with lower trading volumes.