Of all the regulatory requirements that traders need to follow, “best execution” is typically the vaguest and the least understood. Rules such as Regulation NMS, which only require firms to “protect” the best quote at each official venue, add to the confusion since some have interpreted this to mean that other liquidity can be ignored. The following conversation, which happened in the spring of 2008, when I was responsible for Lava Trading LLC, typifies the way best execution has been handled poorly by some in the industry:
“Let me get this straight. You are telling me that our technology is doing TOO GOOD a job of filling client orders???” I asked, somewhat dumbfounded. “Exactly” said Greg, my head of product development. “The problem is that NASDAQ’s “STGY” order type only routes a liquidity removing order to locked markets when it HAS to under Regulation NMS. Our equivalent order type always tries to complete each client’s order when it can and our router reacts to all of the available liquidity in the market. As you know, we use the full depth of available liquidity in our market data so that makes a big difference.” After his explanation, I nodded and directed my next comment to the salesman on the account.
Click Here for More from RegOne
“Ok. I understand that, but since our client (broker name omitted) can post an order in our system that will STAY on the bid without routing, I don’t understand the problem. If our client is concerned with the more than half cent difference between paying a take fee and collecting a rebate, then why don’t we configure them for posting without routing whenever their order is not marketable?”
At this point, our salesman said, “Well, I don’t know why, but they are telling me that our system is costing them a lot of money. They want us to pay them back for the past month and want us to know that this would cost more than a million dollars per year in access fees if it continues.”
I turned back to Greg and repeated my question and he answered: “These are cases where our client’s client did not specify anything and the TRADER at our client picked the price. In those cases, if NASDAQ or ARCA locks our ECN, which is where the system is posting the order, the trader gets angry if the order is not filled. The trader actually likes the fact that we have a higher fill rate with this order type, but the “bean counters” don’t. You see, these orders fill 15% more stock, but that half cent difference is costing them a lot of money and they can’t pass access fees on to their client.”
After getting over the shock of how the management of this broker dealer was willing to ignore their own trader’s views of best execution, I shook my head. “Okay, if they put it in writing, put their configuration in our front end system back to posting on NASDAQ, using the STGY order type.” Let’s think about building an order type that tries to maximize rebates for our clients, if they opt in explicitly… This makes me uncomfortable, but this is a software company and if our clients want something, we need to consider building it.”
To be clear, we never built such an order type. After several internal meetings and discussions with our largest clients, we decided that Lava should not offer a routing option that ignored liquidity. Despite the differential costs, most of Lava’s clients did not want us to build such an order type. We decided that our competitive position was dependent on our software’s ability to maximize the liquidity it found.
This story is a perfect example of what many commentators, in the debate over “maker/taker” pricing, call the “agency” conflict. Broker dealers, due to intense pressure to preserve margins, amid continued downward pressure on commissions, continue to build and enhance strategies that minimize fees or generate rebates. At the same time, exchanges, who built complex routing algorithms into their order types, cater to their cost conscious clients by offering order types that eschew pure liquidity maximization, in favor of economics.
This conflict does not affect all agency situations in the same way. There is a marked difference between strategies built to service broker dealer clients and those offered directly to the “buy side,” including asset management companies, hedge funds and pension funds. While there is certainly downward commission pressure for buy side clients, the sell side business is much more sensitive to explicit costs. In many firms, algorithms and routing strategies, offered to other brokers, are priced for fees that approach zero.
The result is that such strategies are extremely sensitive to explicit costs and often make profits based on the ability to be a net collector of rebates. I do not believe that this is necessarily the fault of the executing broker. Such brokers typically offer multiple strategies with different explicit fees and levels of performance, and leave it to their clients to choose what best suit their needs. There is nothing wrong with the wide diversity of order types and routing strategies, per se. The problem is that many broker dealers have insufficient performance reporting on specific strategies and arguably insufficient incentive to act upon the reporting that is available.
Given the amount of commentary related to “maker taker” or “fee caps” related to the agency conflict, I want to be explicit about two key points:
1) The conflict of interest, being discussed in this note, would be reduced, but not eliminated, if regulations either eliminated “maker taker” pricing or severely capped access fees. However, there would likely be serious side effects to either the imposition of fee caps or eliminating rebates. Rebates are a major incentive to displayed liquidity, so there could be significant increases in spreads and higher transaction costs, particularly for mid and small cap stocks, if they were banned or reduced.
2) The only way to actually eliminate the agency conflict of interest is to have PERFECT performance reporting on execution quality, coupled with full disclosure of all fees and rebates. While perfection is not possible, I strongly believe we are capable of generating high quality, actionable execution analysis. If all broker dealers analyzed all executions and performance metrics become a standard consideration for choosing different trading and routing strategies, it would dramatically improve the management of the “agency” conflict. This is a preferable solution to fee caps or other price controls.
The good news is that the recent focus on market structure i.e., “Flash Boys,” has increased attention to best execution. FINRA, for example, has announced one of their exam priorities is to analyze where retail brokers route non-marketable limit orders. It is their intent to evaluate whether the order routing practices are putting too much value on brokers’ own economics.
Specifically, they want to look at the tradeoff between high rebate exchanges and potential fill rates of client orders. (This issue was raised in a paper by Professors Robert Battalio and Shane Corwin of the University of Notre Dame and Professor Robert Jennings of Indiana University last year) While this is not an unwelcome development, the bigger issue is evaluating where both algorithms and institutional traders post limit orders. Retail orders are usually placed and left at a single price during a trading day.
Thus, for a retail order to be “disadvantaged,” it would, typically, have to have been placed at the low price of the day, if it were a buy order. (Or at the high of the day, if it were a sell order). Based on analysis that I have seen, this does not happen a high percentage of the time, with most of these examples represented by a small number of very liquid, low priced stocks. Institutional algorithmic orders, however, are often placed at or near the bid or offer side of the market and then modified repeatedly to stay near the market. In addition, most of these institutional orders are excluded from the “official” monitoring of Rule 605.
We believe that it is time to start addressing this situation and agree with recent comments by both the FIA and BATS; improved best ex rules are needed. These rules should be applied broadly and should not be limited to retail brokers or even brokers. Rules should also be applied to stock exchange routers as well as to strategies designed by asset managers or vendors, regardless, of who implements them.
Specifically, we believe that best execution regulation should be based on a few key principles:
Best execution should be interpreted as requiring broker dealers to fill client orders whenever possible, unless there are specific instructions from their client to avoid or minimize fees. Broker dealers should only attempt to avoid such fees when doing so results in a lower NET cost.
All firms that send orders to algorithms or routing strategies should be held accountable for the parameters and trading instructions they provide for these orders. Instructions that opt for lower explicit commissions or fees should be scrutinized.
Improve transparency by eliminating most exemptions from rule 605; over 65% of executed orders are currently excluded from the rule. We would propose that all market and limit orders sent to market centers which are eligible for execution without constraints from special handling instructions be covered by the rule. (In particular, the child orders created from large institutional parent orders should be included)
In conclusion, while the principles stated above are clear, there is a concern that encouraging regulators to implement prescriptive rules could end up being counter-productive. There are many nuances to both the investment process and the operation of markets and there isn’t a “one size fits all” rule that would work. That being said, emphasizing transparency and performance reporting, would be a great start.
David Weisberger is the Managing Director and Head of Market Structure Analysis of RegOne Solutions
