The SEC transaction fee pilot has received considerable attention recently. The main issues involved are well-known.[1] But in a nutshell, the fee pilot aims to gauge how a fee cap reduction and (separately) an elimination of rebates would impact markets. Many on the buyside view the proposal favorably, noting that both would reduce the incentives of brokers to route orders to high rebate venues. Critics have noted the pilot could lead to increased spreads and/or decreased depths due to reduced incentives to provide liquidity. Plus, implementing the pilot involves adding complexity to an already extremely complex marketplace. And from a competitive perspective, leaving dark pools and other non-exchange entities out of the pilot puts them at a relative advantage in pilot stocks, which could have a detrimental effect on price discovery as volume shifts to off-exchange venues.
Can rebates actually help align broker and client incentives?
While much has been made of the adverse incentive effects rebates have on broker routing, relatively little attention has been paid to the potential beneficial aspects of rebates on broker incentives. Specifically, in the same way rebates give market makers greater incentives to provide liquidity, rebates provide brokers with similar incentives to provide liquidity via greater use of passive “make” orders relative to aggressive “take” orders. A key consequence of increased passive trading is the potential for better fill prices for clients, leading to more favorable performance on average.[2]
An example of this can be seen in the context of algorithmic trading. Algorithms often have discretion over the mix of “make” and “take” child orders they send out to execute a parent order. To the extent that rebates incentivize brokers to trade passively via “make” orders, they can indirectly result in improved algo performance if the “bias” toward passive pricing leads to better fill prices and reduced market impact.[3] Contrast this to a child order pricing strategy where “make” and “take” orders have identical exchange fees. In that setting, a broker would essentially be indifferent from a fee perspective between sending “make” or “take” orders. And when other incentives are taken into account, the broker may actually prefer aggressive “take” orders over passive “make” orders since “take” orders generate commissions for the broker with certainty, while an unfilled “make” orders generate no commissions.[4]
Will this favorable incentive go away if the SEC bans rebates?
It should be noted that a “make” rebate is not entirely necessary for a broker to be incentivized to use passive “make” orders. All that is required is that the cost of a “take” order exceed the cost of a “make’ order. Of course, the magnitude of the difference has an impact on the strength of the incentive. Eliminating rebates and/or reducing the fee cap would shrink the difference between the “make” and “take” fees, thereby reducing – but not eliminating – the incentive for brokers to prefer “make” orders to “take” orders.
But these incentives alone may not be enough to improve performance
To be clear, the argument above does not necessarily imply that incentivizing passive trading would actually improve buyside trading performance. For example, a broker who simply routes its “make” orders to the highest fee venues will not improve performance, should the lower fill rates lead to sufficiently high opportunity costs. But, in most cases, the cost differential between the optimal and the suboptimal “make” venues is typically significantly smaller than the fee differential between “make” and “take” orders.[5] Consequently, it is more likely that other considerations could sway a broker to route its “make” orders optimally.
For example, suppose a buyside trader submits an order to a broker’s algorithm with an instruction to complete the order fully (e.g., unconstrained VWAP). The “must complete” instruction actually incentivizes the algorithm to choose the venue that maximizes fill rates for its “make” orders. To see how, consider what would happen if the algorithm routed to a high rebate venue that had a low fill rate. If the “make” order does not fill, the algorithm would eventually need to resort to higher-cost “take” orders to stay on schedule and to ensure completion. And because the optimal “make” venue is likely to be cheaper than the cheapest “take” venue, the algorithm will still prefer passive trading to aggressive trading, on average.
An interesting corollary to this argument is that, if each exchange set its “make” and “take” fees to be identical, brokers would not only lose the incentive to submit “make” orders, but the perverse broker incentives motivating the fee pilot could still exist. Specifically, if each exchange’s “make” and “take” fee were equal, but these fees still varied across exchanges, brokers would still prefer routing to the low fee venue. So long as differences in fees exist across venues, simply abolishing rebates and/or reducing the fee cap will not eliminate broker incentives to route to the lowest cost venue.[6]
A second corollary is that inverted venues may provide incentives for brokers to prefer “take” orders, even if the broker could have filled the order passively at a better price. For example, suppose an inverted market is at the best opposite side quote and the “optimal” make venue is also an inverted venue. In this case, the broker is incentivized to take liquidity and earn a rebate rather than route to the “optimal” make venue and pay a fee.
Commissions speak louder than rebates
The discussion thus far has focused entirely on how broker incentives relate to fees. But there is another, equally critical aspect of this discussion. Brokers are concerned not simply with minimizing costs. Rather, they are concerned with maximizing net commissions. For a fixed level of commissions, brokers do have an incentive to reduce exchange fees by maximizing rebates (or minimizing costs), as noted above. But using that logic, brokers also have incentives to hire cheap inexperienced staff, deploy subpar technology, use higher latency SIP data, understaff the support desk, etc. But brokers ignore these incentives if other stronger offsetting incentives exist, namely that taking these actions could have a significant, negative effect on commission revenue.
The implication of this is that, even in the absence of regulation, buyside traders themselves can help mitigate the adverse incentives brokers have to route inefficiently by using commissions as a carrot… and a stick. Buyside traders can reward those brokers who focus on execution quality over rebate maximization by routing more flow. And for brokers that put their own interests ahead the client’s, buyside traders can simply reduce flow or stop allocating orders to those brokers entirely.[7] But to do this effectively, buyside traders must figure out which brokers are providing superior performance even in the presence of these potential conflicts.
Key Takeaway
While brokers can have incentives to take actions at the expense of client performance, buyside traders have an even stronger tool to combat these incentives: performance-based broker order allocation. Determining which brokers are focusing on performance can be done partially on a qualitative basis, by better understanding how brokers route their orders. The increased level of disclosure mandated by the SEC surrounding routing incentives will be quite useful in this regard.[8] But to really incentivize brokers to prioritize performance over rebate maximization, buyside traders must allocate orders based on performance itself . And this requires buyside traders to create a process to measure performance meaningfully and on an ongoing basis. For example, a client could use an “Algo Wheel” to randomize orders across brokers in a controlled manner, as discussed in a prior post.[9] Many clients have already set up in-house expertise to evaluate broker performance. Others rely on external vendors, like The Bacidore Group, LLC, to assist them in developing performance measurement systems or to measure performance on their behalf. But regardless, performance measurement is key to ensuring brokers focus on trading performance and not on cost minimization.
One last thing…
Thank you for read this post. If you have any feedback, we'd love to hear it. Also, if anyone has comments on any of our posts, suggestions for future topics, or would like more information on The Bacidore Group, please contact us using the form at the bottom of this page or email us at info@bacidore.com.
And please take a look at our other posts available here.
The author is the Founder and President of The Bacidore Group, LLC. The Bacidore Group provides research and consulting services to buyside and sellside clients, including assistance with broker algo customizations, "algo wheel" design and analysis, custom TCA frameworks, custom algo development, and bespoke quantitative analysis.
Copyright 2019, The Bacidore Group, LLC. All Rights Reserved.
[1] For a complete discussion of the issues, see https://www.sec.gov/rules/final/2018/34-84875.pdf.
[2] This could also lead to enhanced market quality as well, as more passive trading of institutional orders increases the posted quantity in the market and provides competition to market makers and other liquidity providers.
[3] Of course, for orders with high alpha, traders or PMs with high risk aversion, etc., the increased passive trading could actually increase cost (in the former case) or generate a poor risk-return trade off (in the latter case).
[4] Furthermore, algorithmic providers may have less incentive to invest in their passive smart routers, as such systems require continuous monitoring of markets, frequent updating of limit prices and sizes, more sophisticated models, etc.
[5] For example, suppose a take order cost upward of 30 mils, while a make order could generate a rebate of 29 mils, for a 59 mil difference. Even if an inverted venue turns out to be the “optimal” market, the broker is typically still better off posting at an inverted venue and paying a fee than using a “take” order and paying an even steeper take fee. (An exception, of course, is when the inverted venue is offering the best “take” price, since the broker may be incentivized to take liquidity and earn a rebate with certainty).
[6] Unless of course, all exchanges converge to the same fee schedule.
[7] Some buyside clients may not be able to avoid routing at least some flow to a given broker, however. In the U.S., for example, buyside traders may need to route to a specific broker to pay for research, to compensate their prime broker for services, etc.
[8] See https://www.sec.gov/rules/final/2019/34-85714.pdf for details.
[9] See our earlier blog post on algo wheels here: https://www.bacidore.com/post/algo-wheel-of-fortune.