How Hidden ETF Transaction Costs Make Billions For Market Makers

by: Logan Kane

Many popular ETFs entice investors with rock-bottom fees, even when the underlying markets have fairly wide bid/ask spreads.

Like many things in finance, this arrangement is too good to be true.

The same way bookies slant lines against popular football teams, market makers slant ETF bid/ask quotes against public order flow and profit billions of dollars per year.

Which ETFs and trading strategies to avoid and some best practices to help grow your money.

If you're not paying for a product, chances are the product is you.

With the rise of passive investing, millions of Americans sleep easier. However, just because the fees cannot be seen doesn't mean they don't exist. As passive investing gets bigger and bigger, invisible transaction costs will be an ever greater source of profits for Wall Street banks and high-frequency traders.

While ETFs aren't free, they are pretty close. However, behind the scenes, market makers make far more than the nominal ETF fees by setting the price of popular ETFs at a premium or discount to the net asset value of the fund. How do they do this? Simply by using their dominant position as ETF market makers to set the price of the ETF against the public order flow at all times. Since order flow tends to come in waves that last hours or days and is driven by news and sentiment, market makers make a mint off of households who buy and sell ETFs. When the tide turns, market makers can switch directions in milliseconds.

How much do market makers make from this? Applying these higher transaction costs to the volume of US-based ETFs means the ETF wealth transfer may be in excess of $50 billion per year from Main Street to Wall Street.

Hidden transaction costs in ETFs

Think of it this way. If the average transaction cost in, say, the Russell 2000 or emerging markets is 25 basis points, then how can the corresponding ETF have a "transaction cost" of 1 or 2 basis points? ETFs tend to need to transact in large size, driving true transaction costs higher than the classic 1/2 of bid-ask spread on each side. Additionally, in the case of international ETFs, the corresponding markets are often closed during US hours, making arbitrage somewhat difficult. There are reasons for index transaction fees to be lower at times than their underlying components, but not to the extent that bid/ask spreads imply, especially for products with relatively illiquid underlying holdings.

Source: ETF Consultants

In times of high volatility, transaction costs rise even further. As such, you tend to see large discounts to NAV in times of market stress. Have market makers decided to just give away money? I think not.

I'll illustrate how the ETF market making game is played, using data from Gary Gastineau of ETF Consultants.

ETF transaction cost – the perception

Source: ETF Consultants

Writes Gastineau:

Point D denotes the current value of the ETF’s underlying stock portfolio, and Point E represents the value of the portfolio at the best ask price. However, the ask price is usually only good for a small number of shares. Due to market impact, the cost to a market maker of acquiring the ETF’s underlying holdings in creation unit size is higher, marked F. Furthermore, the creation process is not free.

There are transaction fees involved, which pushes the cost to a market maker of actually creating shares up to G. For a market maker to make money by selling ETF shares in the market and then transacting with the ETF to create shares, the average price of the fund shares sold must exceed G. As with creations, market makers pay transaction fees in connection with redemptions, which push the proceeds to a market maker of redeeming shares down to A. For a market maker to make money by buying ETF shares in the market and then transacting with the ETF to redeem the shares, the average price of the fund shares purchased by the market maker must be less than A.

As you can see, the naive perception of investors is that their transaction cost is at most the difference between C and D. However, this is not so.

ETF transaction cost – the reality

Source: ETF Consultants

Investors tend to do the same things at the same time, so for the market maker to stay in business, it is necessary for the average transaction cost to investors to be higher than the market making firms' cost to create or redeem shares. The transaction cost to the average investor is likely to be at least the distance between A and G. Market makers systematically buy low and sell high, whereas Main Street does the opposite. Sure, sometimes market makers are selling shares out of inventory rather than having to create/redeem shares, but you can bet that they're slanting the line against the public, and profiting by doing so.

This means if you're buying and selling a Russell 2000 ETF like iShares Russell 2000 ETF (IWM) or an emerging market ETF like iShares MSCI Emerging Markets ETF (EEM), your true transaction cost is likely to average the full bid/ask spread, which is 25 to 50 basis points. So, for every $100 you trade, it's a pretty safe assumption that 25 to 50 cents is your round-trip transaction cost, not ~$0.01. In times of market volatility, like 2008 or even December 2018, take the usual transaction cost numbers and double them.

Could you profit from always being on the opposite side of the trade as the public? It's possible but likely that you would have to invest in high-speed trading infrastructure to succeed in profiting from short-term fluctuations in share prices. The fact that the market makers are still in business means that most investors are not successful at paying less transaction cost than the market makers incur. While sometimes buyers and sellers agree, sharp intraday moves in ETFs seem to be the norm, not the exception, and bid/ask spreads for ETFs are much wider than the initial observation would show if you want to trade more than a few hundred shares.

For example, on the day that I wrote this, the closing auction was 0.12 percent lower for the Nasdaq 100 ETF (QQQ) than the NASDAQ Index. The two are perfect substitutes, but the public was on the other side of the market makers. The public almost certainly sold into the close, losing 0.12 percent to the market makers for the day. If this is happening for one of the most liquid products on the market, I would hate to see what happens with less liquid products.

There's an old saying in Las Vegas that you never see a bookie driving a beater, and it's true for market makers also. Bookies suffer from the same adverse selection problem as market makers, and they deal with the issue the same way, by dealing a spread and slanting the line against the teams favored by the public. Bookies get the juice and market makers get the bid/ask spread, but in both cases, the money is from slanting lines against the public.

Why can't arbitrage fix this situation? Because if you wanted to arbitrage the ETF against the underlying, you'd have to pay transaction costs on the underlying, which would cost you the difference between A and D.

It's the same thing in sports betting. Bookies set the lines so mid-major/unpopular teams cover roughly 51-52 percent of the time, but to overcome the juice, you'd have to win 52.4 percent of the time. Because 75 percent of the money typically comes in on the favorite, this boosts the bookies' profit margins anywhere from 50 to 100 percent.

Other Hidden ETF issues

There are other problems with many popular ETFs. The worst I can think of is the annual high-frequency trader frontrunning that occurs when large indexes change their component companies. Some index providers take steps to minimize this, and some don't. For example, the S&P 500 is the largest index product in the world, so there are systems in place to prevent HFT from some of the easy profits to be made from index changes. However, other index providers either don't know, don't care, or have conflicts of interest that cause them to not build reliable products. The Russell 2000 is a classic example of this.

I covered this last year and would like to share a reminder again since spring is HFT open season on the Russell 2000. Hundreds of participants will be added to the Russell in a couple of months, hundreds will drop out, and savvy traders will be buying all the winners and selling the losers to flip them back to index funds for a profit. Emerging market funds are also a popular option, but are full of rigged components and uncompensated risks. While such funds can be used tactically for trading purposes, the transaction costs are much higher than you think.

Also, don't think you can avoid these transaction costs by only investing in mutual funds. Academic research shows that hidden trading costs likely cost mutual fund investors 0.75 percent per year.

This issue isn't specific to stocks, either. I've covered in detail how some of the most popular bond funds are rigged against investors, also.

Popular commodity ETFs are rigged beyond comprehension against investors as they basically have to play poker with their cards face up by telling the market in advance when they're making billions of dollars in roll transactions.

The worst ETF offenders (among popular funds)

iShares MSCI Emerging Market ETF (EEM)

iShares Russell 2000 Index ETF (IWM)

iShares J.P. Morgan USD Emerging Markets Bond ETF (EMB)

Vanguard Total Bond Market ETF (BND)

iShares Core U.S. Aggregate Bond ETF (AGG)

SPDR Bloomberg Barclays High Yield Bond ETF (JNK)

Invesco DB Commodity Index Tracking Fund (DBC)

I'm not singling out iShares and Vanguard. They have some of the best ETF products on the market, as well as some of the worst. The underlying indexes are often the source of the problem, not the fund itself.

However, there is a pretty easy rule that's 95 percent effective in weeding out bad ETFs. If the ETF's index is named after a Wall Street bank, the product is more than likely to be rigged.

Best ETFs and some best practices for trading

iShares Core S&P 500 ETF (IVV)

iShares 20+ Year Treasury Bond (TLT) TLT tracks the ICE 20+ Year Treasury index now, not a Wall Street bank index.

iShares National Muni Bond ETF (MUB)

Vanguard High Dividend Yield ETF (VYM)

*Vanguard International High Dividend Yield ETF (VYMI)

iShares Edge MSCI Minimum Volatility USA ETF (USMV)

*iShares Edge MSCI Min. Vol. EAFE ETF (EFAV)

*International ETFs are going to suffer from higher transaction costs but can help diversify your portfolio and boost return if done right. Don't trade them actively, and don't sell into generally poor liquidity conditions. For this reason, I only recommend international investments with a strong factor tilt towards political stability and quality or value to overcome transaction costs. In this case, you're paying some transaction costs but investing in products that have structurally better risk/adjusted returns.

There are other ETFs that aren't rigged, but these are some of the best. Note that if you are trading international ETFs, you have no way of knowing your transaction costs, so you need to hold for the long term and not trade them. Also, consider hedging the FX risk or using an ETF that does so on your behalf. Research shows that this will help your performance if you're US-based.

Also note, when trading, I now recommend using "immediate or cancel," or "fill or kill orders." A colleague of mine ran some tests in Python and we discovered that market orders were actually outperforming limit orders for small traders. I spent much of this week trying to figure out how this could be and more of the dynamics of market microstructure, and I now have a much better understanding of it, as well as of why HFT is so keen to buy retail limit orders from companies like Robinhood.

The classic research shows that traders who use limit orders outperform traders who use market orders. Back in the days when bid/ask spreads were in 1/8s, this was 100 percent true. However, since the decimalization of stock trading and the rise of HFT, it may not be true anymore.

When you use a market order, you're guaranteed to pay a penny or two higher than the current price of the stock. However, you get your order immediately filled. This can be dangerous in fast-moving markets and can cause unexpected things to happen. Not caring about what price you get is dangerous.

However, when you use a limit order, you are exposed to a different kind of risk, which is the risk of adverse selection. When you submit a limit order to the market at 5 cents below the current price, you essentially are selling a put option on the stock. If the stock goes up, you don't get filled. If the stock goes down, you do get filled. You can profit if noise traders buy stock from you, but if the traders are informed by order flow, you have a problem. The issues start to happen when you let a limit order sit. The longer you let a limit order sit, the greater the chance that someone else will get information on the stock and trade with you (either actual news or the information that large buyers or sellers are transacting). Stale limit orders around earnings announcements are a fountain of HFT profits for this reason. The study I’m about to share shows that people get picked off for over 250 basis points on average in these situations.

Households’ limit orders suffer significant losses: the average same-day return is −2.5% for the orders executed during the first two minutes after the (earnings) announcement.

That's the essence of adverse selection, someone else knows something you don't, and they can profit off you because of it.

This is also why HFT submits and cancels thousands of orders on stocks, to avoid this adverse selection problem. If the stock is going down through the limit price that you set, you're buying it. If the stock is going up, you're not getting a piece. In the aggregate, letting limit orders sit for hours is worse than simply using market orders. UCLA researchers call this the "limit order effect."

I'm guessing here, but what I've seen in tests indicate that adverse selection might cost you somewhere between 2 and 10 cents per share, per trade, rather than saving you money as is commonly assumed.

Best practices

1. For small orders, use immediate or cancel limit orders at either the current bid or a penny higher. Keep doing this until you get filled. Check with your broker to make sure that you're not getting double charged commission on this. If it's a problem, use fill-or-kill orders.

2. For larger orders ($100,000+), consider using a trading algorithm to break up trades. IB's accumulate/distribute seems to work well for liquid issues, and the adaptive algorithm seems to work best for less liquid issues. Even larger traders may want to consider using a custom algorithm.

3. If it's a domestic ETF, try to only buy when the ETF is showing a discount and sell when it's showing a premium. Otherwise, know that your overall transaction cost is higher.

4. No one is going to be successful in doing any kind of short-term trading without heavy investment in trading infrastructure and an ultra-low-cost structure (commissions, interest, clearing, etc). If these terms aren't familiar to you, you have no business doing short-term trades. Honestly, I'd be shocked if 20 percent of Robinhood customers make money in any given year.

5. Never, ever use stop orders or good-til-canceled limit orders.


It's a jungle out there. Wall Street has to make a living somehow, and with the rise of passive investing, this is the new way they're doing it. The perception of costs for index funds and the reality are two distinctly different things. Fortunately, you can protect yourself by avoiding problematic products and controlling trading costs. Sometimes the best investment options are the simplest. Sophisticated traders have ways of managing costs through the derivatives market, but for retail investors, simple may be best.

Did you enjoy this article? Consider following me for future research updates!

Disclosure: I am/we are long VYM, VYMI, MUB, IVV. I wrote this article myself, and it expresses my own opinions. I am not receiving compensation for it (other than from Seeking Alpha). I have no business relationship with any company whose stock is mentioned in this article.