The History of U.S. Equity Rebates
IEX has fired up the rhetoric in recent months about the practice of paying rebates to liquidity providers in the U.S. equity markets. Before Trading Places jumps into the current controversy, I thought it would be useful to retell the history of rebates, the maker-taker regime, and how the U.S. Equity markets evolved around them.
Brad Katsuyama, CEO of IEX, kicked up a storm recently by calling rebates “kickbacks” in his testimony before the U.S. House of Representatives Committee on Financial Services.
[Rebate] payment to brokers when not shared with the broker’s client is equivalent to a kickback. Public data shows that exchanges who pay this rebate garner a greater percentage of order flow despite providing worse execution quality.
You can sum up Katsuyama’s arguments with a meme I made using the “distracted boyfriend” meme-du-jour.
The “kickback” language is so effective because it is not exactly intuitive why rebates exist. You are using the exchange’s services, matching with another one of the exchange’s customers, and then they PAY you for it? If I buy a lot of widgets from a factory, I would expect to get a volume discount, but there is no volume at which I would expect the factory to start paying me to take the widgets off their hands. Unlike exchanges, there are no network effects from buying widgets.
Island ECN and Rebates
To better understand rebates and much of the market structure that IEX is fighting against today, you have to go back twenty years to the origin of the Island Electronic Communications Network (ECN). Island was seeded with capital and employees by Datek Securities, one of the largest day trading shops of the 1990s internet bubble.
In 1995, Datek was making $150 million a year trading U.S. equities. The Datek traders (often derided as bandits) specialized in picking off sleepy NASDAQ market makers who were still updating their quotes by hand. By 1996, the NASD regulatory body started to go after Datek for these practices, and the NASDAQ market market making desks learned a few tricks of their own to avoid trading with the Datek traders (often illegally backing away from trades). Here’s Scott Patterson from the book Dark Pools on the Island origin story:
The Nasdaq market makers, and the waves of new rules and obstacles that NASD kept ramming out, were a constant source of anxiety and outrage at Datek. Market makers routinely ignored Datek’s SelectNet orders, backing away from their posted quotes, often leading to missed opportunities or, worse, significant losses as traders, struggling to get out of positions in a fast-moving market, were left flapping in the breeze.
Josh Levine, the lead programmer at Datek, decided to fight back by creating his own trading venue, Island. On Island, traders could match orders with each other, completely bypassing the NASDAQ market makers. Patterson, again:
In its most basic form, Island was a computer program that simply matched buy and sell orders, bypassing the market makers. After Island matched the trades, it reported them to Nasdaq. The name evoked an “island” of orders where investors could retreat, a digital haven safe from Nasdaq pickpockets.
By 1997, Island matched 3% of volume in NASDAQ-listed stocks. However, future growth was uncertain. Datek traders were still providing most of the liquidity in the Island order book. Meanwhile, the Datek day trading operation was rapidly declining. Regulators were closing in, and new competition had sprung up from other day trading shops. Datek would be out of the day trading business by 1998 and would focus all its energies on the Datek Online retail portal (which was later sold to TD Ameritrade for $1.3 billion).
How could Island attract more liquidity to its venue? Traditional sales with the steak dinners and baseball games take a lot of time. What if Island just paid traders directly to send them more orders? Island, like all venues, had started out by charging traders to both add and take liquidity. Under the new “maker-taker” regime, Island would pay a rebate ($0.10/100 shares) to liquidity providers and charge a fee ($0.25/100 shares) to liquidity takers. Island would pocket the $0.15/100 share difference.
The maker-taker model was a gamer changer! By 1999, Island’s share of NASDAQ-listed stocks had jumped to 13%.
At the time, Instinet was charging $1.50/100 shares and Arca $0.50/100 shares on both sides. Island had the maker-taker model largely to itself for a few years because it seemed so unintuitive. Then, in 2002, every major competitor suddenly rolled out a variation of the maker-taker model. Matt Andresen, then CEO of Island, recently shared this with me by email:
For the next few months, literally every possible spread, cap, rebate etc was attempted by someone. Fascinatingly, as the various markets threw pricing schemes at the wall, they quickly settled on the answer, which was the $0.30/100 shares take out, $0.20/100 shares adding (with tiers take the rebate higher). That was the pricing that best optimized market share for a market. I always thought it was amazing that Josh not only hit upon the idea, but also so nearly guessed the perfect levels.
Instinet threw in the towel and purchased Island in 2002 for $508 million. The merger was essentially a takeover of Instinet by the Island executives and technologists. Three years later, NASDAQ acquired the Island ECN for $934 million and a similar takeover of NASDAQ’s U.S. equity business was done by the core Island team. The NASDAQ exchange you know today is a descendant of the Java rewrite done by Island developer Brian Nigito. The IEX matching engine was built by a protegee of Nigito, Constantine Sokoloff, using the same architectural principals as Island. Trading is a small world.
A Changing Market
Levine could have never imagined the current fury over rebates in 1997. Rebates were just a marketing tool to attract brokers to the new trading venue. In the late nineties, bank broker-dealers had large NASDAQ market-making desks. When a buy-side firm wanted to sell 100,000 shares of Apple, the bank would buy the shares for their own account. It was then the bank’s responsibility to trade out of the shares on venues like Island. Since the bank was then trading for its own account, it made the rebates or paid the fees when executing on Island and other trading venues.
By the mid-2000s, banks started getting out of the principal trading business in cash equities and started offering agency algorithmic execution to customers. The rise of high-frequency trading firms meant that human market makers were no longer viable. Now instead of the bank buying the 100,000 shares of Apple for their own account, the bank would instead execute the shares on the buy-side firm’s behalf on venues like Island using an automated execution algorithm. Because the bank broker-dealer maintained the relationship with the exchange, the bank continued to keep the rebates.
As you might imagine, an incentive was created for brokers to optimize their execution to generate more rebate revenue. There was no collusion or top-down orders, and unlike the movie Margin Call, there was no dramatic meeting where the head of the desk gave an explicit order to screw over clients. Instead, there were a series of decisions, made over years, about a complex system that each give slight advantage to the broker. These cumulative decisions turned rebates into a small, but important, source of revenue to the brokers.
A Regulatory Solution
There is a very simple solution to removing the conflict of interest rebates create between brokers and customers. The SEC could mandate that brokers pass along to customers all fees and rebates incurred in agency execution. The broker is then no longer incentivized to collect rebates for its own account. This is not what IEX is arguing for, which is a complete end to rebates, but we will get to that next week.
Brokers (and even some buy-side firms) argue against this regulation because of the complexity of accommodating passed back rebates/fees in back-office systems. I think this fear is a bit overblown given enough time to accommodate the change. More likely, brokers are just concerned about a lost line of revenue.
Dave Lauer, a market reform advocate who is in favor of banning rebates, told me he consults with a lot of buy-side firms that prefer not to have fees/rebates passed back: “[Buy-side firms] want predictable commissions and the bundled services that come with it.” Buy-side firms receive a lot of services from their brokers including balance sheet, research, meetings with CEOs, and even steak dinners. The primary way buy-side firms pay for those services is through execution. A single per share commission makes budgeting much simpler for the buy side. Moreover, the current strategy of focusing on execution costs is largely working for the buy-side, as execution fees have fallen 40% since the financial crisis according to Greenwich Associates.
Viewed in this light, the change in how rebates get credited may make sense for regulators to pursue, but the “kickback” charge is just a headline grabber. The goal of IEX is not to have more transparent fees/rebates but to remove maker-taker from the market structure entirely. IEX has designed their fee schedule around this idea. IEX charges $0.09/100 shares to both sides to execute in their dark order book (the “speed bump”) and charges nothing to trade in their displayed order book. Free sounds great, but remember that other Exchanges are paying a rebate of $0.28+/100 shares. As such, IEX’s market share has been limited to date, hovering a bit above 2%.
Would banning rebates entirely actually improve market share for IEX? Would it make the U.S. equities markets better? Next week, Trading Places will dive into the contemporary debate around rebates and maker-taker.
