Can Brokers Have It All? On the Relation between Make‐Take Fees and Limit Order Execution Quality

• Author: ROBERT BATTALIO,ROBERT JENNINGS,SHANE A. CORWIN
• Date: 01 October 2016
• DOI: http://doi.org/10.1111/jofi.12422
• Published date: 01 October 2016

THE JOURNAL OF FINANCE •VOL. LXXI, NO. 5 •OCTOBER 2016

Can Brokers Have It All? On the Relation

between Make-Take Fees and Limit Order

Execution Quality

ROBERT BATTALIO, SHANE A. CORWIN, and ROBERT JENNINGS∗

ABSTRACT

We identify retail brokers that seemingly route orders to maximize order flow pay-

ments, by selling market orders and sending limit orders to venues paying large

liquidity rebates. Angel, Harris, and Spatt argue that such routing may not always

be in customers’ best interests. For both proprietary limit order data and a broad

sample of trades from TAQ, we document a negative relation between several mea-

sures of limit order execution quality and rebate/fee level. This finding suggests that

order routing designed to maximize liquidity rebates does not maximize limit order

execution quality and thus brokers cannot have it all.

TODAY,EVERY U. S. STOCK exchange levies fees or pays rebates that depend

on an order’s attributes. In the standard model, exchanges charge liquidity-

demanding orders (i.e., marketable orders) a “take fee” that exceeds the “make

rebate” they offer liquidity-supplying orders (i.e., nonmarketable limit orders).

More recently, a few exchanges began using inverted fee schedules, charging

liquidity suppliers a fee that exceeds the rebate they pay to liquidity deman-

ders.1In this paper, we examine the impact of these differential fee schedules

on broker routing decisions and limit order execution quality.

∗Battalio and Corwin are with the Mendoza College of Business at the University of Notre

Dame. Jennings is with the Kelley School of Business at Indiana University.The authors gratefully

acknowledge research support from the Q-Group. Wethank two anonymous referees, Jeff Bacidore,

Bruno Biais (the Editor), Peter Bottini, Colin Clark, Joe Gawronski, Alex Green, Larry Harris, Dave

Lauer, Katya Malinova, Chris Nagy, Maureen O’Hara, Steve Poser, Paul Schultz, Jamie Selway,

Jeff Smith, Chester Spatt, John Standerfer, Ingrid Werner, brownbag participants at Indiana

University and the University of Notre Dame, seminar participants at Goldman Sachs, Cornell,

Cubist Systematic Strategies, Johns Hopkins, University of Arizona, University of Cincinnati,

Vanderbilt University, Jump Trading, the Securities and Exchange Commission, and Nasdaq, and

participants at the 2014 NOIP Conference, 2014 FIRS Conference, 2014 Florida State University

SunTrust Beach Conference, and 2014 Mid-Atlantic Research Conference in Finance for their

comments. The authors have no conflict of interest to report with respect to the Journal’s disclosure

policy.

1The difference between the fee and the rebate is an important source of exchange revenue.

Given the competition between U.S. exchanges, there is a high correlation between the level of an

exchange’s fee and its rebate. Fee structures at the various exchanges are described in more detail

below.

DOI: 10.1111/jofi.12422

2193

2194 The Journal of Finance R

Although the Securities and Exchange Commission’s (SEC’s) Order Protec-

tion Rule establishes price priority in U.S. equity markets, the rule does not

specify who trades first when multiple venues have the best posted price. When

two venues offer the best price, one expects liquidity demanders to route their

marketable orders to the venue with the lower take fee, all else equal. Thus,

cross-exchange differences in fee schedules create situations in which equally

priced, nonmarketable limit orders resting on separate exchanges have differ-

ent “net price” priority. Indeed, an important advantage of routing limit orders

to inverted venues is the ability to gain queue priority.2Consider the case in

which two exchanges display limit orders at the National Best Bid (NBB). If

sufficient selling supply arrives (perhaps because the seller is informed), the

sell order walks down the limit order books at both exchanges and liquidity

at the existing NBB is exhausted. In this situation, limit orders on both ex-

changes purchase shares at the bid price. However, if the price rises before

liquidity is exhausted at the NBB, limit orders on the low-fee venue fill first

and at least some limit orders on the high-fee venue miss a favorable trad-

ing opportunity. This suggests that an exchange’s relative fee structure can

influence the conditions under which its standing limit orders execute.

Maglaras, Moallemi, and Zheng (2015) examine the impact of fee sched-

ules on order routing and limit order execution quality in an equilibrium

model with sophisticated investors who make their own order routing deci-

sions and pay/receive the fees/rebates generated by their trades. In this setting,

Maglaras, Moallemi, and Zheng (2015) show that standing limit orders directed

to high-fee venues trade less frequently and face greater adverse selection costs

than orders sent to low-fee venues. Thus, sophisticated investors must trade

off potential rebate revenue and enhanced limit order execution quality when

deciding where to route their limit orders. In some circumstances, cross-venue

differences in fill rates are small enough to justify order routing designed to

capture liquidity rebates. In others, the expected rebate revenue offered by

high-fee venues is too small to offset the enhanced limit order execution qual-

ity available on low-fee venues.

In contrast to the setting modeled in Maglaras, Moallemi, and Zheng (2015),

most retail investors delegate order routing decisions to brokers who charge

fixed commissions and do not pass fees/rebates through to their customers.

Angel, Harris, and Spatt (2011) argue that this can lead to a conflict of interest

in the broker’s order routing decision. In particular, if investors choose brokers

based primarily on commissions, perhaps because they lack the sophistication

and/or information necessary to evaluate limit order execution quality, brokers

may make order routing decisions that maximize their liquidity rebate revenue

rather than their customers’ limit order execution quality.3

2As noted in Harris (2013), the “net” prices resulting from maker-taker fees provide a form of

subpenny pricing that sophisticated electronic traders can use to step in front of other traders.

3This issue is highlighted in an April 2010 letter from the Investment Company In-

stitute to the SEC, which states that “brokers may refrain from posting limit orders on

a particular exchange because it offers lower liquidity rebates than other markets, even

Can Brokers Have It All? 2195

Are brokers routing limit orders in a manner that maximizes liquidity re-

bates? Toinvestigate this question, we use SEC-mandated Rule 606 disclosures

for the fourth quarter of 2012 to examine the routing decisions of 10 well-known

national retail brokers. Consistent with the conjectures of Angel, Harris, and

Spatt (2011), the routing reports suggest that four of these brokers route orders

to capture liquidity rebates. In particular, these brokers sell market orders to

market makers and route limit orders exclusively to either market makers or

the exchanges offering the highest liquidity rebates (and charging the highest

take fees) during our sample period.

Because fees may not be the only factor affecting the order routing decision,

the impact of fee structure on limit order execution quality is an empirical

question.4We address this question using three measures of execution quality:

the likelihood of a fill, the speed of fills, and the realized spread generated by

fills.5We begin by analyzing proprietary limit order data obtained from a major

investment bank that uses a sophisticated algorithm to route orders. To control

for stock and market conditions that may affect limit order execution quality,

we identify identically priced limit orders to buy (or sell) shares of the same

stock displayed concurrently on multiple venues. For these concurrent orders,

market conditions are held constant and differences in fill rates, execution

speeds, and realized spreads can be linked directly to exchange characteristics

such as the rebate/fee schedule. Using these order pairs, we conduct “horse

races” between different exchanges. In nearly every comparison we make, the

low-fee venue wins more horse races (i.e., fills when the high-fee venue does not

or fills more rapidly) and enjoys higher average realized spreads (i.e., trades in

more favorable conditions) than the high-fee venue.

The results of the horse races demonstrate that there are instances in which

routing to the high-fee venue provides diminished limit order execution qual-

ity. However, because simultaneous order display is endogenously determined

by the smart router, it is not clear whether these conclusions generalize to

a broader sample of orders and trades. To analyze more general market con-

ditions and order characteristics, we conduct a multivariate analysis of the

though that exchange offers the best possibility of an execution for those limit orders.” (See

http://www.sec.gov/comments/s7-02-10/s70210-138.pdf.) This type of behavior would appear to be

a clear violation of the broker’s best execution obligations. For example, NASD Notice to Members

01-22, which governs broker best execution obligations, notes that “in evaluating its procedures

for handling limit orders, the broker dealer must take into account any material differences in

execution quality (e.g., the likelihood of execution) among various market centers to which limit

orders may be routed” and “must not allow an order routing inducement, such as payment for

order flow . . . to interfere withits duty of best execution.”

4As suggested by the high proportion of trades executed on standard fee venues, the ability to

gain queue priority on lower-fee or inverted venues may be important only under certain conditions.

5Following prior literature and the definition codified in SEC Rule 11Ac1-5 (now SEC Rule 605),

we define the realized spread for an executed limit sell order as twice the difference between the

execution price and the midpoint of the spread prevailing five minutes after the trade. For limit

buy orders, we multiply by negative one. As noted by the SEC and illustrated in our motivating

example below, the average realized spread provides a summary measure that reflects both the

probability of execution and the conditions under which orders execute.