How the ETF Arbitrage Pricing Mechanism Works By John Spence June 3, 2002

Research indicates that most buy-and-hold investors are primarily drawn to exchange-traded funds for their lower expense ratios and potential tax efficiency. Most passive investors are not interested in trading during the day, but the ability to do so with ETFs sets them apart from mutual funds. Many industry observers are rightly questioning if the arbitrage mechanism that keeps the price of a domestic ETF share in line with the net asset value (NAV) could temporarily break down in extreme market conditions.

First, let's look at how the ETF arbitrage mechanism works (it may also be helpful to refer to this article). Keep in mind that the whole system is designed to prevent market makers from making unfair markets.

A passive ETF share is simply a basket of securities held in the same proportion as in the index tracked. Market makers and specialists must continually decide if they want to hold those securities in packaged as an ETF share or loose - this process is called creation and redemption.

The large institutions involved in ETF market making are constantly monitoring the share prices of ETFs, and will arbitrage any differences between share price and the NAV of the underlying portfolio. Since the NAV of the underlying portfolio is updated every 15 seconds throughout the trading day, this process can happen very quickly.

Let's take a look at a theoretical example of the arbitrage pricing mechanism in action. Imagine a specialist, for example Hull Trading, sets the ask price of an ETF share at $45, while the NAV of the underlying portfolio is $46. Another specialist, perhaps Bear Hunter, spots the opportunity and starts scooping up as many shares as possible. Bear Hunter can then turn around and redeem the shares at $46. In our hypothetical example, let's assume 1,000,000 shares were purchased and subsequently redeemed. At a $1 profit per share, Hull Trading has lost a million bucks and Bear Hunter is ahead by the same amount. Institutions can also reverse the process if the ETF trades at a premium to the NAV. Therein lies the incentive for making tight markets in ETFs.

Due to the competition between specialists, the real measure of an ETF's liquidity is the liquidity of the underlying stock. This is what leads some industry observers to wonder how the arbitrage pricing mechanism will react in extreme market situations, and it's a fair question.

Setting international ETFs aside for the moment, how will domestic ETFs react in a severe market crash? In particular, Lipper analyst Don Cassidy has asked if institutional players might be unwilling to risk capital on the long side to assemble creation units at true underlying value when discounts arise. Others have harkened back to the crash of 1987 when the tight relationship between futures and stock prices came apart.

Lee Kranefuss, CEO of individual investor business at Barclays Global Investors, says there's a big difference between the 1987 situation and ETFs. Institutions were unable to arbitrage the difference between the cash and futures markets in 1987 when the markets were in free fall because the cash market was trading, but the futures market ceased trading at times. The system was built upon arbitrage between the two markets, and when one broke down the whole system unraveled. Kranefuss says ETFs are different because the arbitrage takes place in the equities market.

"Applying intuition learned from other financial instruments often leads to faulty conclusions when applied to ETFs," said Kranefuss, who also noted that domestic ETFs were stress-tested in the trading days following the incidents of 9/11 because significant premiums and discounts did not materialize.

Kranefuss points out that crashes affect all vehicles, not just ETFs.

"Market meltdowns don't serve any financial instrument well," said Kranefuss.

Therefore, says Kranefuss, concerns about ETF liquidity should involve problems specific to ETFs and not novel concerns that affect all financial instruments.

For example, ETF critics love to point to what happened to the Malaysia iShares country basket when that country imposed capital controls in 1998. The ETF was unable to honor redemption of creation units in currency other than Malaysian ringgits, and significant discounts opened up. Kranefuss argues that all instruments and not just ETFs were aversely affected by the situation, and in any case the ETF did allow trading and price discovery despite the discount.

The evidence so far shows that domestic ETFs, in particular those tied to liquid indexes, have tracked their benchmarks closely so far. How they will react in a market crash is more important for investors who like the reassurance of being able to get in and out of markets during the day. However, ETF managers and proponents have touted this benefit and should therefore deliver on it. In reality though, most skeptics won't be truly convinced until the arbitrage pricing mechanism holds up through a serious crash. Although I'm as curious as anyone, here's to hoping we never find out.

The bottom line is that investors need to understand how and why premiums and discounts arise, especially as more ETFs come to market. Although the SEC recently gave initial approval for fixed-income ETFs, one its major concerns was potential premiums and discounts in the products. The SEC and others have voiced similar concerns with regard to the possibility of upcoming actively managed ETFs. Finally, some analysts have pointed out that repetitive ETFs tracking similar market segments could reduce liquidity and increase tracking error.