Alpha Architect lives by the mission of “empowering investors through education.” Alpha Architect also operates ETF Architect, which is a low-cost white-label ETF business. Their team lives ETF operations every day and they were helpful in helping me craft this post. (For a deep dive into ETF trading, please take the time to read Wes’ post on ETFs in the secondary market.)
As Wes points out in his post, ETFs are just a basket of securities, so when you purchase an ETF share, you are purchasing a basket of securities. Specialized market makers called authorized participants (APs)(1), are constantly calculating the NAV of the basket of securities the ETF holds and creating a bid/ask spread around that NAV. Frankly, anyone with the capital, time, some fancy computers, and mental capability can trade the basket against the ETF to make arbitrage profits. However, APs have the unique ability to “create” or “redeem” ETF shares with the custodian that normal market participants do not have. If the AP has been buying the basket of shares all day and has sold 10,000 shares (2) of the ETF, the AP will place a “creation unit” with the custodian. The AP will hand the custodian a complete basket of securities, and the custodian will deliver 10,000 shares of the ETF to the AP; so the number of shares available to trade fluctuates as authorized participants (APs) can create or redeem blocks of ETF shares at any time.
Because prices are determined through market trading, they can, and often do, deviate from their net asset value (NAV). APs use this deviation from NAV to cover expenses and realize a profit. This trading should and often does push the price and NAV together, minimizing the tracking error of the ETF and trading costs to investors. However, there are instances where a wide spread can happen and stay that way for extended timeframes.
A recent study, observed the relationship between creation/redemption activity and divergences between ETF market value and NAV. Though it’s an interesting article, its conclusions rely almost exclusively on creation/redemption activity as a signal of APs taking advantage of arbitrage opportunities. However, APs do, and in fact prefer, to take advantage of arbitrage opportunities without utilizing the creation/redemption process. Therefore, the authors’ reliance on creation/redemption activity is a misguided measure of APs making markets.
Jon Fulkerson, Susan Jordan, and Denver Travis, authors of the study “ETF Arbitrage and Daily Cash Flow,” published in the April 2022 issue of The Journal of Investing, examined ETF creations and redemptions around price deviations from NAV to determine if arbitrage trades occur as expected. They collected daily shares outstanding data for equity ETFs from Bloomberg from 2001 to 2015. Their sample included 798 passively managed equity ETFs with 1,084,191 daily observations.
Following is a summary of their findings (Alpha Architect’s comments appear in clear bullet points):
- Creations occurred 13.5 percent of the time, while redemptions occurred 7.7 percent of the time—most days are “no flow” days. The higher frequency of creations over redemptions can likely be attributed to the overall growth of the ETF industry.
- More creations/redemptions follow large end-of-day price-to-NAV (P/NAV) deviations, and both the size and likelihood increase for larger P/NAV deviations.
- This implies that the APs are in fact attempting to make arbitrage profits when the discrepancy is wide.
- Despite equity ETF prices frequently diverging from NAV, APs often do not trade on the apparent arbitrage opportunity. Even when there were the most extreme price deviations (1st or 99th percentiles), flows were still relatively rare, with a less than 25 percent probability in the following day—despite what appear to be arbitrage opportunities, APs do not always respond to deviations.
- Here is where the authors run into trouble. They are relying on creation/redemption activity to indicate acting on arbitrage opportunities. AP’s and market participants can take advantage of spreads in the open market without utilizing the creation/redemption process.
- Costly exchange conditions limit arbitrage activity, while the magnitude of mispricing apparently affects the decision to arbitrage, but not the size of the arbitrage trade.
- This is mostly true. Arbitrage opportunities can exist but be difficult to take advantage of. For example, an AP will have a difficult time taking advantage of arbitrage opportunities in an emerging market ETF, the emerging market ETF is trading on US time and underlying securities trade when US markets are closed, as opposed to a U.S. technology ETF that they can easily trade the basket intraday alongside the ETF.
- ETFs with larger bid-ask spreads are less likely to have either a creation or redemption, suggesting that higher costs of trading limit arbitrage.
- It is true that ETFs with difficult-to-trade underlying securities will have wider spreads to accommodate the APs’ need to make a two-sided market. However, this shouldn’t materially impact creation/redemption activity. It may discourage investors from investing in the fund, and therefore funds with wide spreads don’t see much creation/redemption activity.
- Arbitrageurs are less likely to trade on a mispricing that requires trading the underlying assets against broad market movements—APs find it riskier to arbitrage premiums in a bull market and discounts in a bear market, likely due to higher execution costs associated with trading in the same direction as the market.
Their findings led Fulkerson, Jordan, and Travis to conclude:
“In practice, arbitrage trading may or may not occur. Our data find that creations and redemptions are rare or frequently delayed around price deviations. Further, several studies find that deviations sometimes persist in ETFs for long periods, suggesting that the arbitrage trading either does not always occur or does not always work quickly.”
Fulkerson, Jordan and Travis’ findings are consistent with those of Antti Petajisto, author of the study “Inefficiencies in the Pricing of Exchange-Traded Funds,” published in the first quarter 2017 issue of the Financial Analysts Journal. He found that while the average premium was only 6 basis points (bps), the time-series volatility of the premium was 49 bps—ETF prices fluctuate considerably around NAVs even if the average level is correct. The value-weighted average volatility was comparable at 40 bps—the result was not limited to smaller funds.
Economically, the equal-weighted volatility showed that the typical fund is trading within a range of -96 bps to +96 bps around its NAV, with a 95 percent probability.
The evidence suggests that investors attracted to the low expense ratios (and tax efficiency) of ETFs also need to seriously consider the premium or discount to NAV the fund is trading at. We at Alpha Architect believe Antti Petajisto said it best:
“Inefficiencies in the pricing of ETFs are not limited to a few funds or a particular time period. In fact, the cross-section of ETFs routinely exhibits some economically significant differences between the ETF share price and the value of the underlying portfolio, especially in some asset classes, indicating that the unsophisticated investor may face an unexpected additional cost (or benefits) when trading ETFs.”
“The efficiency of ETF prices therefore would be expected to depend on the transaction costs and any other limits to arbitrage that might deter arbitrageurs from trying to profit from a mispricing.”
Luckily, this is an instance in which the SEC is looking out for the little guy, as funds are required to post historical data on the discrepancy between the fund’s market value and NAV on their website(3)
Additionally, it should be pointed out that if you’re purchasing a large ETF order (a creation unit size or greater), it’s best to contact the fund or an AP directly, as they will likely be able to implement the trade as close to NAV as possible(4).
Often, the trading costs are limited to the bid-offer spread of an ETF, as market makers will not typically incorrectly value one side of the basket of securities. It is possible, though, that an AP could have an imbalance they want to adjust out of (possibly with another fund) and will take a loss or require compensation to take certain positions. As ETF managers ourselves, Alpha Architect tries to be intimately aware of spreads in our funds, but it is clearly best for investors to be aware that differences between NAV and the market price of an ETF can and do happen.
The question is: What is a fair price for investors to pay for the convenience of purchasing ETF shares, and are there alternative products that don’t have this issue? If a fund has a NAV of $25 a share and the bid-ask spread is $24.95 to 25.05, that’s a 20-basis-point spread away from NAV, which seems reasonable considering many individual securities can trade that wide. That said, some funds can have much wider spreads between NAV and the market value that last for extended periods, and those costs can be material to investors. It’s best to check first and see the premium/discount to NAV on a fund in which you are looking to transact.
For informational and educational purposes only and should not be construed as specific investment, accounting, legal, or tax advice. Certain information is based upon third party data which may become outdated or otherwise superseded without notice. Third party data is deemed to be reliable, but its accuracy and completeness cannot be guaranteed. By clicking on any of the links above, you acknowledge that they are solely for your convenience, and do not necessarily imply any affiliations, sponsorships, endorsements, or representations whatsoever by us regarding third-party websites. We are not responsible for the content, availability, or privacy policies of these sites, and shall not be responsible or liable for any information, opinions, advice, products, or services available on or through them. Neither the Securities and Exchange Commission (SEC) nor any other federal or state agency have approved, determined the accuracy, or confirmed the adequacy of this document.
The opinions expressed by featured authors are their own and may not accurately reflect those of Buckingham Strategic Wealth® or Buckingham Strategic Partners®, collectively Buckingham Wealth Partners.LSR-22-282
|↑1||Can also be called Lead Market Makers|
|↑2||creation/redemption unit sizes can vary but typically land between 10,000-600,000 shares|
|↑3||See TAIL’s website as an example|
|↑4||Just only make the brokers aware of a creation unit size trade, not which side of the trade you’re going to be on when asking for a bid/ask|