In part I, we examined exchange-traded funds' (ETFs) bid-offer spreads through the “investor’s lens” explaining what they are, where they come from and why they differ from fund to fund and across asset classes.
In Part II we will look through the “trader’s lens” analysing what actually happens at a trading desk as buy and sell orders hit the wire. Specifically, we will seek to answer the questions: what is the role of the authorised participant, the creation/redemption process, the ETF arbitrage mechanism and how does all of this feeds back into the spreads quoted on the exchanges?
Behind the Scenes: The Primary Market
Within the primary market, authorised participants (APs) can create new ETF shares by delivering cash or securities to an ETF provider in exchange for new ETF shares. APs can hold these shares and ultimately redeem them for cash or securities from the provider, or they can turn around and sell them on the secondary market. Market makers (MMs), or so-called liquidity providers (LPs), are broker-dealer firms that usually hold a certain number of ETF shares in order to facilitate trading on the secondary market, hence providing liquidity. It should be noted that all APs can also act as MMs to facilitate trading, however not all MMs can act as APs as they are not eligible to create or redeem ETF-shares with the ETF-provider.
As the graph below shows, APs have three options to provide liquidity on the secondary market before actually turning to the ETF provider to create or redeem new shares in the primary market.
The AP can act as an agent and buy the ETF on the exchange. Or it can buy the actual underlying index constituents and exchange the basket with the ETF provider in exchange for ETF shares. Alternatively, the AP can buy a correlating hedge to cover its short position in the ETF. Which option is used obviously depends on various factors.
Under option A, the AP is probably buying the ETF from another liquidity provider (LP) and is therefore acting only as a middle man. However, at some point in the order flow someone would have to buy the actual index constituents (for most physically replicated ETFs). In the case of option B, the AP would be short the ETF and long the basket which means that he will turn to the ETF issuer to exchange the basket for ETF shares. Option C is usually used for synthetic ETFs or in the case that the underlying market is closed. As there will be financing costs involved maintaining the long hedge and short ETF position, this is usually only a short-term solution and the AP ultimately tries to either unwind its position or turns to the ETF issuer as soon as possible. In addition, some market makers even buy the index constituents for synthetic ETFs in case this is the best and most cost efficient way to hedge their positions. In general, market makers do have the same options to provide liquidity on the secondary market. However, to flatten their book at the end of day, non-AP MMs would have to turn to an AP rather then the ETF-issuer directly.
The Creation/Redemption Process
Now, let’s have a look at where ETF shares are born--the creation/redemption process. Trades on the exchange or over the counter (OTC) are considered to be secondary market trades whereas the primary market is where new shares are issued. When a creation is done by an AP, the requisite shares or cash matching the creation unit are delivered to the ETF-provider in return for new ETF shares. The ETF provider does not hold an inventory of shares but rather issues new ETF shares as part of the creation process; hence this will increase the number of shares outstanding. The opposite holds true for redemptions; meaning ETF shares are delivered to the ETF provider in return for the underlying basket or cash. As the ETF provider does not intend to hold any inventory, these shares are “destroyed” and the number of outstanding shares will decrease.
Creations and redemptions are transacted at the official end-of-day net asset value (NAV) of the ETF. As the NAV is calculated from the basket of securities called the creation unit and a cash component, the basket and the ETF should be priced equally. Disregarding transaction costs, one should be indifferent between holding the creation unit or the ETF share.
There is also a creation and redemption fee, which is, in particular in the US, often a single flat fee independent of how many units are transacted. This obviously creates great scale benefits to a large market-making business. As at is unknown how many ETF shares the AP will trade in future, these fees are built into the standard spreads in an ETF market. However, market makers in Europe tend to quote their creation and redemption fees in basis points relative to the size of the creation or redemption.
The creation/redemption fees are either paid by the investor directly, in case of large OTC-trades, or embedded in the spread as market makers will include the costs of flattening their books in the spreads. Therefore--holding all else equal--ETFs with higher levels of turnover tend to have tighter spreads as it is more likely that the AP will quickly recoup the creation costs by making a market in the shares in the secondary market.
Arbitrage Mechanism
The creation/redemption mechanism allows ETFs to trade very close to NAV. If for some reason the market price of an ETF diverges from its NAV, arbitrageurs will step in to bring the NAV and market price back in line. The following graph illustrates the situation where the market price of an ETF surpasses its NAV plus the transactions cost of purchasing all of its underlying securities. In this case, it becomes profitable for a market maker or arbitrageur to buy the underlying basket and exchange it with the ETF provider for new shares of the ETF.
Investors interested in a more detailed analysis of NAV, premium and discount should read our Terms of the Trade: NAV, Premiums, and Discounts.
Once the entire creation/redemption and trading process for ETFs is understood, it is relatively easy to understand the underlying source of the bid-offer spreads for ETF shares trading in the secondary market. If the AP is able to buy the ETF from another LPs (option A in the previous graph), he passes on the spread costs and probably adds something on top of it as compensation for his effort. If the AP buys the underlying basket (option B), the ETF’s spread will be dependent upon the aggregate asset weighted average spreads of its underlying index constituents. In some instances, an ETF’s spreads can actually be less than the aggregate asset weighted average spread of its underlying. The hedging costs under option C depend on different factors. The higher the market volatility the more expensive the hedge is going to be. Also, for illiquid markets implementing a hedge is more difficult to achieve and the risk of hedging errors increase, hence hedging costs will rise. In addition, if the underlying market is closed, the price will likely be based on trading in the relevant futures market--assuming that there is one. Therefore, as it is difficult to attach a “fair value” to a London-listed ETF tracking the NASDAQ 100 Index during the morning trading hours spreads tend to be a little wider compared to the afternoon trading hours. Investors will probably get a tighter spread later in the day once Wall Street has opened. This is especially important in those instances where high impact news arises when local markets are still closed, news that will impact once trading opens. As an example, let’s assume you want to buy a London-Listed ETF tracking the NASDAQ 100 Index which allocates about 18% to shares of Apple. After the market closed, Apple released headline figures that missed market expectations. Therefore, the true value of Apple and ultimately the index, given Apple’s large weight, is only an estimate until the stock market opens the following day. As discussed earlier, sentiment can vary across different exchanges as well. Therefore, when the markets that an ETF’s underlying constituents trade on are closed, there is an increased risk of hedging errors for the AP which result in higher hedging costs. The total hedging costs are another parameter in the spread-equation.
Summary
Spreads are an component of the total cost of ownership for ETF investors—and those with short-term time horizons in particular. Spreads are a very complex topic and there are many more factors feeding into the equation. However, spreads are mainly determined by the liquidity of the underlying and the depth of the market makers’ order books and pricing structure. We hope that looking at spreads from the perspective of both an “investor” and a “trader” has helped to deepen your knowledge of this important topic.
