A Wall Street insider has published a report that could upend conventional thinking about payment for order flow, the controversial practice that helped Robinhood grow explosively on its way to an initial public offering.
After tracking the controversy over trading in GameStop this winter, Hitesh Mittal, the founder of BestEx Research, used his expertise from working at one of the world's largest hedge funds and consulting with institutional clients to analyze recent trades. His report, "The Good, the Bad and the Ugly of Payment for Order Flow," undermines claims made by market makers and Robinhood's defenders on Wall Street and Sand Hill Road, and could help critics seeking to build a case for banning the payments market makers send to retail brokers for directing trades to them, as market regulators in the U.K., Australia and Canada already have.
The main argument for payment for order flow, the system that Robinhood, Schwab and others use to subsidize the cost of trading so they can offer zero-commission trades, is that payment for order flow gives retail investors better prices, as market makers compete to do better than the price mandated by SEC rules known as the National Best Bid or Offer or NBBO. Bloomberg Intelligence estimates that investors received $3.7 billion in price improvement last year.
Mittal thinks it's not so simple. He says prices would be better if trades moved to exchanges. He likens the promises of price improvement to "getting a 30% discount on an item after the shopkeeper raises the price by 40%."
The analysis is timely: The SEC under new chair Gary Gensler is bringing greater scrutiny to trading and market structure after Robinhood temporarily froze buying in GameStop amid a social media-fueled run-up in the game retailer's stock. Robinhood's move brought attention to its close relationship with market maker Citadel Securities, which is a dominant player in handling trades from retail brokers. Gensler recently indicated to lawmakers that the agency would look at how to address the "inherent" conflict in payment for order flow and the current market structure.
Mittal, who was global head of trading at quantitative hedge fund AQR Capital before starting his own consultancy, doesn't work on behalf of retail investors; his firm's goal is to minimize execution costs for institutional investors, he said. But payment for order flow ends up affecting his clients as well. The dramatic increase in retail volume prodded him to write about the issues with payment for order flow.
Payment for order flow happens when brokers send orders to market makers with the expectation that they can deliver better prices than what's available on the exchanges. Those small price differences get split three ways: The market maker takes some as profit, the broker gets a payment from the market maker, and the retail investor who placed the trade gets a slightly better price, or price improvement.
Defenders of payment for order flow generally make three interrelated claims:
- It delivers better prices than exchanges, particularly for retail investors.
- It doesn't benefit market makers beyond the profit they make on a trade.
- It doesn't lead to market concentration and lessened competition.
Mittal's analysis, which examined aggregate trading data of all trades recorded on the public NYSE Trade and Quote system during December, undermines all three claims.
On pricing, his analysis shows that the top five market makers seem to get 24.5% price improvement for retail investors' trades, compared to the NBBO price on exchanges. That seems good for market makers, brokers and retail investors. But Mittal also found that those trades would get even better prices if the retail trades were all done through exchanges.
Mittal believes that under the current market structure, brokers can't change because those who don't take payment for order flow are at a competitive disadvantage — so only regulation can address the situation.
Mittal's analysis shows that overall spreads on trades would drop if all trades were moved to exchanges, which would mean both retail and institutional investors would get better pricing.
How does he show this?
First, it helps to understand what the NBBO is — and isn't. Despite the name, it doesn't actually capture the market's best pricing. The National Best Bid and Offer, which is the price that the SEC requires brokers to get on certain trades, is only based on trades of 100 shares or more on certain public exchanges. It also doesn't include trades with wholesalers, or on alternative trading systems.
Because of that, the 24.5% price improvement market makers get compared to exchanges is actually more like 15%, Mittal found. That's because certain prices on exchanges are already better than what's captured in the NBBO price.
Certain exchange trades aren't included in NBBO, such as hidden orders and odd lots, both of which get tighter spreads — meaning better prices. Hidden orders — which are limit orders placed with instructions not to publish them — make up 16.7% of daily volume in liquid stocks and 20.8% of illiquid stocks. Odd-lot orders of fewer than 100 shares are "extremely common," particularly for high-priced stocks.
There are also trades available on alternative trading systems that are priced at the NBBO midpoint — which is theoretically the best price available because there is no spread between buy and sell prices. Retail trades sent to market makers don't get these better prices.
Mittal calculates that if all retail trades were moved from market makers to exchanges, spreads between trades would drop by more than 25%.
Making a market
That 25% price improvement of moving trades to exchanges would be even bigger if you accounted for private information that market makers have, Mittal said.
Big market makers with private information have an immense advantage — raising the risks of market concentration, MIttal says. That's because they're seeing more of the trading happening — both retail trades coming direct to them and public trades on exchanges — and know who is behind those trades.
Big market makers on exchanges also know where their own trades are coming from, whereas other market makers don't since trades are anonymous on public exchanges. So market makers have a better "picture of the supply and demand in a stock," Mittal said. They can use information to get the best execution not only for their off-exchange market making, but also they can "price stocks more accurately than other" market makers on exchanges, jumping in when they expect the price to move their way, and backing off when they expect prices to move against them, he said.
Beyond retail order flow, market makers also have private information from electronic liquidity providers that large wholesalers provide to institutional brokers.
Mittal believes the spread improvement could be even higher than that 25% he calculated if there was a "level playing field" of trading on exchanges. "We think the spread improvement would be even bigger," Mittal said. "The reason we think it'd be bigger especially in illiquid stocks is because if wholesalers get 50% of the volume they have a lot of information about that stock that other market makers do not.'
Another example of private information: Brokers send retail limit orders to market makers but market makers don't execute them; they send them to exchanges, he noted. Brokers get exchange rebates on these trades, since exchanges typically provide rebates on limit orders and charge for market orders.
It's a strange practice since market makers do not trade against non-marketable limit orders, since there would be no spread. But market makers benefit from this additional private information. Mittal believes this is done for "optics" reasons, because brokers don't want to answer questions about when and why they send orders to market makers.
It takes concentration
The private information market makers get from order flow doesn't just have short-term benefits. Over time, it can erode the competitiveness of the market, Mittal writes — a dynamic where having more information gives leading market makers an edge, compounding as they take more share.
As the number of market makers decline, spreads increase, which means worse prices for investors. That market concentration effect isn't included in Mittal's calculations.
The five largest wholesalers had retail volume equal to 47% of the volume during regular trading on all 16 exchanges in December — and this doesn't include odd lots, which aren't reported by market makers. Citadel had 50% of the top five's volume and Virtu had 26%.
The concentration is even higher for certain stocks. In the highest traded Russell 2000 stocks, Citadel traded 69% to 188% of the volume of these stocks on exchanges during December.
Some industry analysts praised the report.
"The BestEx report advances our understanding of the impacts of captive retail order flows executed by a concentrated group of wholesale market makers in the dark," said Paul Rowady, director of research at Alphacution Research Conservatory.
Mittal says he wants to "represent both sides," despite the implicit critique of the industry. "There is nothing evil about allowing retail market order flow to go to wholesalers," he writes, and he notes that without payment for order flow, a "lack of revenue for retail brokers would likely lead to increased commissions charged to retail investors."
Robinhood is expected to file a prospectus soon for its initial public offering. That S-1 will likely reveal just how dependent the company is on payment for order flow, though other required regulatory disclosures already suggest it's taking in considerable sums and growing quickly.
Brokers aren't really to blame for how they route trades because they face a prisoner's dilemma, Mittal said. If only one broker sent its trades to exchanges, other brokers would still get the benefit of better prices. Because you can't coordinate in a competitive market, the only way for all trades to move to exchanges would be through regulation, he said.
That's the conclusion regulators already reached in other countries, though the U.S. is a larger market. With billions on the line, Mittal's analysis promises to add fuel to an already explosive debate.