Exchange-traded fund investors love to talk about fees. After all, the expense ratios charged by ETFs are often fractions of those for competing mutual funds.
But expense ratios are just one part of the true cost of investing in ETFs. Brokerage commissions and spreads also play an important role.
Brokerage commissions are obvious—they are the $9.99 or whatever you pay your broker to execute a stock trade. Because ETFs are bought and sold like stocks, commissions apply; in contrast, many mutual funds can be bought and sold without commissions. A quick calculation will tell you if it's worth paying commissions to get the lower expense ratio.
Spreads, however, are a dirty little secret. Until recently, there was no publicly available data on ETF spreads. Monthly data is now available on IndexUniverse.com, but still, most investors ignore spreads when choosing between different investments. They do so at their peril, as spreads represent a substantial extra expense for many ETFs.
What Are Spreads?
Like stocks, ETFs are bought and sold on the market by auction. The bid/ask spread is the difference between the best price being offered for an ETF (the "bid") and the best price at which someone is willing to sell ("the ask").
Let's assume that the real market value of an ETF is halfway between the bid and the ask. If you submit a market order for an ETF and it gets filled at the "ask," the difference between that and the halfway point represents a cost: You are essentially overpaying for the ETF. The wider the spread, the more it costs you.
There's been a major debate in the ETF industry about how big ETF spreads are, and about what influences the size of those spreads. ETF promoters (especially promoters of newer, thinly traded ETFs) claim that spreads are based on the liquidity of the underlying stocks—that is, the stocks held by the ETF—and not by the level of trading in the ETF itself. The reason, proponents say, is that large institutional investors called "Authorized Participants" (or APs) can create new shares of an ETF at any time. For instance, APs in the S&P 500 SPDR ETF (SPY) can "create" new shares of SPY by buying up all 500 stocks in the S&P 500 in the right proportions and delivering them to the product issuer (State Street Global Advisors). The product issuer will give the AP shares of the ETF in return. If the bid/ask spread on SPY gets too large, the thinking goes, APs could simply create new shares, establish a better price and pocket the difference.
The caveat, of course, is that APs can only create ETF shares in large lots; typically 50,000 shares or more. If there is only demand for a few hundred shares, it's not worth the market maker's time to create an entire new group of shares. So what do spreads really look like for investors?
To find out, I examined data for all available ETFs and ETNs for the period from January 1, 2008, through March 31, 2008. That covered 666 funds, ranging from the massive (SPY) to the tiny (HealthShares Ophthalmology Fund). The data, from NYSE Arcavision, examined tick-by-tick spreads between the best bid and best offer, and weighted those spreads by volume to produce an average spread for each ETF over that time range.
There are two ways to consider spreads: in absolute dollar terms and as a percentage of the share price. First, I looked at the absolute dollar amounts.
For the time period covered, 30 ETFs had the minimum possible average spread of one penny. These included some of the largest ETFs on the market (SPY, QQQQ, EFA), all nine of the highly traded Select Sector SPDR ETFs, a number of international funds, some fixed-income ETFs and two ProShares UltraShort ETFs (which are designed to deliver -200 percent of the daily return of the underlying index). A complete list is available in Figure 1.
On the flip side, there were a handful of ETFs that reported outrageous spreads—$1-$3, and even more. These were all newly launched ETFs with very little liquidity.
A few ETFs had absurd spreads; three had spreads of $10/share or more. These were clearly anomalies, and not reflective of true investor experiences. Sometimes, when there is no market for an ETF (no shares trading), the bids and asks will become stale, and one can deviate widely from the other. For example, the iShares MSCI ACWI (ACWI) ETF launched on March 30 and traded just 600 shares during its first two days on the market (the period covered by my analysis). The average spread over that time period was $10.99/share, according to the data. But that's not really what investors paid. A spot check on April 18 showed the bid/ask spread at $0.80; still high, but well below the artificial $10.99 figure.