Most people have heard of the New York Stock Exchange and Nasdaq, but there are several other U.S. exchanges, including the Investors Exchange (or IEX, referenced in Michael Lewis’s 2014 book Flash Boys) and the Better Alternative Trading System (BATS, now part of Cboe Global Markets). There also are trading venues that are not formally stock exchanges—the so-called “dark pools” run by Wall Street firms. And a significant amount of order flow simply gets “internalized” in a brokerage’s own system for matching orders.

When a brokerage buys or sells stock on behalf of a client, the brokerage is legally obligated to try to obtain the best execution possible. This requires it to have access to a wide variety of data on prices and quantities traded on an exchange, as well as the various bids and asks that are current. The exchanges charge for that information and their prices have risen dramatically over the last decade, a trend that was helped along by a provision in the 2010 Dodd–Frank Act. The question is whether those price increases can be justified under normal standards employed by the Securities and Exchange Commission. The SEC seems to have some concerns about that; it recently pushed back—at least temporarily—on the increases by overturning prior fee increase approvals given to Nasdaq and the NYSE. That has given brokerages some hope for more price relief in the future.

This is more than just a fight between exchanges and brokerages. If the price of these data keeps increasing, it may lead to a reconsideration of the present execution standards. The more brokerages and hedge funds have to pay to execute trades, the higher they will set their fees for ordinary investors who have retirement funds invested with such entities. And thanks to the magic of compound interest, even small reductions in net returns can result in a significant reduction in the ultimate size of an investor’s nest egg.

Making data more available / The Securities Acts Amendments of 1975 charged the SEC to develop a “national market system” that would link the numerous financial markets. The SEC concluded that this law required the exchanges (and other non-exchange execution venues as well) to publish current bid and offer prices—and their quantities—as well as the price and quantities of recent trades.

To that end, the SEC required that the industry provide a consolidated information source known as a securities information processor (SIP), which would be run jointly by the exchanges and their private regulator, the Financial Industry Regulatory Authority (FINRA). This information is referred to as core data.

Exchanges also are free to sell non-core data that customers might also find valuable. This includes the so-called depth-of-book data—that is, the bids and offers currently on the limit order book that are not at the best prices. For example, assume that the current market for Acme Inc. is $10.00 bid for 5,000 shares, while 3,000 shares are offered at $10.01. In deciding how to route orders, firms might find it useful to know the quantity of Acme shares demanded below $10.00 and quantity supplied above $10.01—in other words, enough information to draw approximate supply and demand curves. Suppose there is a high demand for Acme at $9.99 and below, but not much supply at $10.02 and $10.03. Market participants might find this information useful. It would be particularly important information for institutional brokers, who might wish to transact 250,000 shares when the current quote is relevant for only a small transaction. In many cases, these brokers are trading stocks for mutual funds owned by common investors.

In the last decade, the cost of obtaining these data has increased considerably. Table 1 contains some of the NYSE fees in both 2008 and 2018. Access fees nearly tripled over that time and new fees were introduced.

The SEC’s targets for its recent order are the fees for depth-of-book products charged by the NYSE’s ArcaBook and Nasdaq’s Level 2, both of which were implemented in 2010. Dodd–Frank permitted exchanges to immediately implement fee increases while the SEC determined their appropriateness. Before then, the exchanges had to file a formal proposal to increase fees and then wait for SEC approval, an often-drawn-out process that included a public comment period.

Determining fair and reasonable fees / Giving the SEC such a vague mandate has made its adjudications a bit problematic, but there are two commonly accepted ways to determine fair and reasonable in this context. One would be to compare the fee to the marginal cost of providing the service, and the other would be to show that the fee is based on some sort of market price. The exchanges embrace the latter approach—understandably so because their incremental cost of providing those data is almost assuredly close to zero.

The exchanges claim that competition constrains their ability to impose fees in two ways. First, if their fees are too high, then customers will simply reroute orders to other exchanges. That would cost high-priced exchanges both data fees and trading fees for transactions carried out on their platforms. Second, the exchanges argue that the market data of other exchanges are substitutes for their data product. In this scenario, if the ArcaBook fee gets too high, then customers will simply drop ArcaBook and pick up market data from some other exchange because the limit order books across exchanges are highly correlated. In 2014 an administrative law judge agreed with the exchanges on this point.

However, a 2018 court case and an associated 2018 SEC finding concluded otherwise. While agreeing that the competition for order flow is intense, both the court and the SEC found that the exchanges had not done an adequate job of demonstrating that the link between order flow and market data fees was strong. In fact, in 2014 the SEC concluded that while depth-of-book market data drives order flow, most market participants do not purchase the depth-of-book market data, contrary to the competition argument. In addition, the current findings question whether exchanges’ depth-of-book products are, in fact, close substitutes. No one has apparently examined the data all that closely to determine the truth.

Our conclusion is that there are market participants that need depth-of-book data from all exchanges in order to meet their obligations to seek best execution for clients. This does not necessarily hold for every participant in the market, but an important subset cannot stop sending orders to a particular market or merely assume that the book on Exchange A is perfectly correlated with the book on Exchange B.

Justifying data price increases / Complicating the market price standard is the fact that a few firms control the bulk of the order flow, which gives each a modicum of market power when negotiating fees. Nasdaq claims that 90% of its order flow comes from 100 firms—way too many for any sort of collusion to occur.

The court did not believe that the exchanges demonstrated that firms would divert order flow elsewhere if fees went up. In particular, the court found the exchange “evidence” to be merely anecdotal, as it amounted to a couple of examples of what happened to order flow when fees increased, which included one firm substantially diverting order flow.

There are multiple alternatives to the non-core data of an exchange: the exchange’s own core data, the non-core data of other exchanges, purchasing non-core data from data vendors, or merely “pinging” orders (which entails sending an oversized order to an exchange to see what the depth behind the best price is). However, the existence of various substitutes does not imply that the exchanges lack market power; all of these alternatives are lacking in some way.

Several experts, working for the exchanges, produced an event study of market share of trading around data fee increases, complete with a regression analysis of market share of trading as a function of the fees. They found no significant correlation. However, the 2018 court ruled that these experts had not properly controlled for various important independent variables. The court eventually concluded that the best approach would be to estimate the product’s elasticity of demand without specifying a particular methodology. We believe that a serious economic analysis should be conducted that does precisely this task, as we do not believe such a study currently exists.

What is best for retail investors? / Exchanges have brokerages over a barrel with these fees, to some extent. In order to get the best execution for their clients as required by law, brokerages must obtain data from multiple exchanges. The exchanges have taken advantage of this requirement by dramatically increasing the price of their data beyond what can be justified by either costs or market competition, the two commonly accepted methods to justify data price increases. The data substitutes for non-core data suggested by the exchanges—their own core data, non-core data from other exchanges, or pinging the market—are not viable substitutes.

We do not want to return to a world where retail investors would not necessarily get their trades executed in a timely manner and unwitting traders received prices that were hours or even a day out-of-date. However, it is not clear to us that the cost of the data that brokerages must obtain in order to comply with data execution rules are worth the benefits. As long as the SEC holds that these data must be acquired, then the commission should also ensure that data fees are appropriate. We do not believe that is currently the case. We suggest that the SEC look more closely at the fees as well as the broader information acquisition requirements for funds.