Payment for Order Flow: the Hidden Cost

This report contains detail compiled from discussions with many industry experts and practitioners. To those who contributed to my research, thank you!

It seems like everyone is either trading stocks or talking about it right now. Thanks to commission-free trading and user-friendly mobile apps, trading is now cheaper and easier than ever for the average joe. Data from Piper Sandler indicates that average daily equities volume this year is at $14.7B, twice what it was in 2019. Surge in market participation from retail traders plays a big part — retail volumes are now 20–25% of total equities volumes, compared to 10% historically, as estimated by Joe Mecane, head of execution services at Citadel Securities.

The decision to drop commissions to 0 also invited public scrutiny to the way retail brokerages make money. An obscure but common industry practice known as payment for order flow (PFOF) has become the focal point of a hot-button issue. Echoing Michael Lewis’s Flash Boys, some of the online commentary and media coverage incorrectly pointed to PFOF as proof that the stock market is “rigged” or that Wall St is “front-running” the hapless clients. The truth is not that simple.

How do brokerages make money?

There are primarily two ways in which retail brokerages make money.

1. Net Interest Revenue: Brokerages charge retail rates for margin loans and stock loans used to fund levered long and short positions while accessing core deposits and wholesale funding markets to refinance these positions at much cheaper rates.
2. Transactional Revenue: Brokerages no longer charge a commission, but many receive payment from wholesale market makers in exchange for routing clients’ orders toward them (i.e. PFOF).

Some brokerages charge extra for add-on services (like trading tools, market data, research reports). Some brokerages are part of companies that also generate income from non-brokerage businesses (like asset management or wealth management).

The revenue breakdown spans widely among companies in the brokerage business. Commissions represents about 5% of Charles Schwab’s revenue in 2019, compared to 22% of TD Ameritrade’s revenue and 15% of E*Trade’s revenue, implying that the industry-wide move to drop commissions in late 2019 had a stronger impact on certain brokers than others, offering a clue to the industry consolidation that ensued. PFOF is not a strategically important profit centre for every brokerage firm: only 1.3% of Charles Schwab’s comes from order flow revenue in 2019, compared to 8.2% of TD Ameritrade’s and 6.5% of E*Trade’s.

How does PFOF work?

When you tell your brokerage you want to buy or sell some stock, by default, you are allowing the brokerage to decide how to execute your trades. What brokerages typically do is to aggregate clients’ orders and channel them toward their choice of wholesale market makers and (less commonly) exchanges. Wholesale market makers have been executing retail trades at prices often better than if those trades were sent to an exchange (“price improvement”). Furthermore, market makes offer to compensate the broker for the opportunity to trade against these orders (“payment for order flow”).

Market makers profit from charging a spread between what they offer to buy and sell a security. When they execute your trade they are allowed to do so at no worse than the best bid or offer available on public exchanges (the “NBBO”), but how are they able to offer you prices better than that and still make a profit? The reasoning has to do with the nature of order flow coming from retail brokerages: it is an aggregation of small individual orders submitted by uninformed traders. Wholesale market makers can afford to show retail trades a tighter spread than exchanges do because making markets for retail orders is less likely to result in subsequent losses (versus if market makers were executing orders from informed traders or traders whose orders are sizable enough to move the market against the market maker).

When wholesale market makers offer to trade with a brokerage’s retail flow, they offer the brokerage a given sum of any combination of PFOF and price improvement, giving the brokerage the discretion over the split of how much to keep for itself versus how much to pass onto the clients. In other words, after the market maker takes its cut of the spread, the broker gets to “slice the pie”. 

Even though PFOF is legally permissible in the US, brokerages face a conflict of interest: making money from PFOF vs. fulfilling the duty to seek the best deal for clients (“best execution”).1

Rule 606: Shining a light on PFOF arrangements

Thanks to the revised Rule 606 of Regulation NMS, brokers have to publish quarterly reports featuring granular detail regarding their PFOF arrangements. These reports allow us to compare the amount that brokers make from PFOF; their contents reveal a surprisingly large variance among brokerages in how much PFOF they make on a per-share basis, not to mention that certain brokerages do not partake in this practice, passing along any savings to the clients in the form of price improvement. Questions to consider:

How would brokerages justify how much PFOF they take in the context of best execution?

What is the appropriate amount of PFOF, if there is one?

The debate about PFOF is distracting us from the core issue: how much your broker makes from PFOF does not reveal the actual level of execution quality you’re getting from that broker. There are metrics like E/Q spread ratio and frequency of price improvement that directly measure execution quality; unfortunately, brokerages are not disclosing this data in a timely, accurate, and standardised fashion, nor are they required to. As a result, retail traders do not have the data they need to accurately benchmark and compare brokers on execution quality.

Putting PFOF into perspective

As a result of price competition and technology advancement, the explicit cost to trade has shrunk dramatically for the average person — commissions, once $45/trade in the 1970’s, are gone. Regulatory guardrails also ensure that any of these brokers today deliver a baseline level of execution quality when their clients trade stocks.2 Data shows that most retail traders simply do not trade at frequencies or amounts large enough for PFOF to materially affect their PnL (more than 65% of retail orders are for trades under 500 shares and fewer than 3% of retail orders are for trades greater than 5,000 shares, according to Nasdaq Economic Research). Having said that, PFOF is no trivial matter. People want to know if they’re getting a great deal from their brokers, both in absolute and relative terms. When brokers take PFOF, do they have their clients’ best interests at heart?

[1] Beyond the scope of this article, routing retail flow toward wholesale market makers, away from exchanges, has broader implications for the markets as a whole. For instance: since almost all retail flow is executed this way, the surge in retail trading has caused the share of off-exchange trading (which is reported to the Trade Reporting Facility) to approach 50% of equities volume; off-exchange activity does not contribute to price discovery on exchanges, theoretically leading to less informative market prices.

[2] Compare with crypto spot markets in the US: bitcoin and ether are not securities and do not fall within the purview of the SEC. Hence, bitcoin and ether traders do not enjoy the investor protections conferred by regulations overseeing the behaviour of securities market participants. Bitcoin and ether derivatives trading in the US is governed by the CFTC.