6.9.3 Maker-Taker Fees and Rebates

Basic Prerequisites: 02 · Market-Maker P&L.

---

1. Intuition & Practical Objective

Every modern equity venue charges two prices for crossing: a take fee $f_t$ to the aggressive order that removes liquidity, and a make fee $f_m$ to the passive order that supplies it. Under maker-taker pricing, $f_m<0$ - the passive side is paid, and we call that payment the rebate $r=-f_m>0$. IOSCO’s definition is exactly this: “the maker of liquidity… is paid a rebate and the taker… is charged a fee.” The real numbers:

• NYSE Arca (May 2012): $0.30\ast \ast charged\ast \ast totakersperroundlot,$0.21 rebated to makers - the exchange keeps $0.09 per round lot (Colliard & Foucault 2012, fn. 3).
• Toronto Stock Exchange (1 Oct 2005): before - takers paid $2$ bps of value, makers paid nothing; after - a maker rebate of 27.5¢ and a taker fee of 40¢ per 100 shares on a pilot set of securities, with non-pilot takers at $1.8$ bps and makers still at zero (Malinova & Park 2015).

The puzzle the whole literature attacks: does the rebate improve liquidity, or just relabel it? Intuition says paying makers attracts more quotes, which tightens spreads. Theory says something more subtle. This page builds the tool that settles it - the cum-fee spread - and shows precisely when the make/take split is neutral and when it is not.

The one-sentence essence. “The rebate and the take fee are two halves of one exchange fee; what a taker actually pays is the cum-fee spread $S^{\text{raw}}+2f_t$, and moving the make/take split at constant total fee changes the raw spread but not that number.”

---

2. Mathematical Ground Truth & Derivations

Work with raw quotes: ask $A$, bid $B$, raw spread $S^{\text{raw}}=A-B$, per-share fees $f_t$ (taker) and $f_m$ (maker, $f_m=-r$ under a rebate).

Who pays what.

• A taker buying pays $A+f_t$; as a taker seller receives $B-f_t$. A round-trip taker pays

$$S^{\text{cum}}=(A+f_t)-(B-f_t)=S^{\text{raw}}+2f_t.$$

• A maker selling receives $A-f_m=A+r$; a maker buying pays $B+f_m=B-r$. The maker’s net earned spread is

$$S^{\text{net}}=(A-f_m)-(B+f_m)=S^{\text{raw}}-2f_m=S^{\text{raw}}+2r.$$

• The exchange’s net fee per share is $f_{\text{net}}=f_t+f_m=f_t-r$.

The neutrality identity. Add the two:

$$S^{\text{cum}}=S^{\text{raw}}+2f_t=S^{\text{net}}+2f_{\text{net}}=(S^{\text{raw}}+2r)+2(f_t-r).$$

The total the taker pays equals maker compensation + exchange net revenue. Consequences:

(a) Breakdown neutrality (Colliard & Foucault 2012, Prop. 1). If the exchange doubles the rebate $r$ and raises the take fee by the same amount so $f_{\text{net}}$ is constant, then competitive makers - who only care about their net compensation $h^{\text{net}}=\lambda +c_{\text{inv}}$ - narrow the raw quote by exactly $r$:

$$h^{\text{raw}}=h^{\text{net}}-r.$$

The raw spread $S^{\text{raw}}$ shrinks, but $S^{\text{cum}}$ is unchanged. The rebate is a pure relabelling of the raw spread when the tick does not bind. This is the Angel–Harris–Spatt (2011) intuition, proved without perfect competition.

(b) Only the total fee affects taker cost. Since $S^{\text{cum}}$ depends on the total fee $f_{\text{net}}$, empirical tests of make/take effects must hold the total fee constant (CF 2012) - otherwise a fee-breakdown effect is confounded with a fee-level effect.

(c) The tick-size friction (Foucault, Kadan & Kandel 2013). With finite $\tau$ the raw quote cannot move in arbitrarily small increments: the half-spread is a multiple of $\tau /2$. If $h^{\text{net}}-r$ is not on the half-tick grid, the maker cannot fully pass the rebate through, and neutrality breaks. The residual is the maker’s rent:

$$\text{pass-through gap}=\text{round-up}\left(\frac{h^{\text{net}}-r}{\tau /2}\right)\cdot \frac{\tau }{2}-(h^{\text{net}}-r)\ge 0.$$

A coarse tick therefore protects maker rents against the rebate’s competitive erosion - which is exactly why tick size and fee design must be regulated jointly (see 06 · Advanced Extensions).

---

3. Computational Implementation - neutrality vs. tick friction

Standard library only. We compute, for a fixed maker compensation $h^{\text{net}}$ and exchange net fee $f_{\text{net}}$, the raw quote, the take fee, and the cum-fee spread - once with a fine tick and once with a coarse one.

Read the two blocks against each other.

• Fine tick: perfect neutrality. As the rebate rises $0\to 0.003$, the raw half-spread falls $0.0120\to 0.0090$ exactly by $r$; the take fee rises in lock-step; the cum-fee spread is $0.0260$ in every row. The rebate moved money from the taker’s fee line to the maker’s revenue line and left the taker’s true cost untouched - the Colliard–Foucault result, reproduced numerically.
• Coarse tick: neutrality breaks, asymmetrically. The rebate of $0.003$ cannot be fully passed through ($0.0090$ is off the grid), so the maker rounds to $0.0100$; the cum-fee spread rises to $0.0280$ - the taker pays $0.002$ more for the same liquidity. And with no rebate, the grid forces the raw quote to $0.0150$, making the taker’s cum-fee spread $0.0320$. The coarse tick, not the rebate, is doing the work.

That is the entire policy content of the maker-taker literature in one table: the rebate is neutral on a fine tick, distortionary on a coarse one, and the tick - not the fee breakdown - is often the binding constraint.

---

4. Failure Modes & First-Principles Breakdowns

1. Testing the breakdown without holding the total fee constant. The single methodological error CF (2012) explicitly warns against. A rebate rise financed by an equal take-fee rise is neutral; a rebate rise financed by nothing (lower total fee) tightens true taker cost. Confounding the two produces a spurious “rebates help liquidity” result.
2. Ignoring the tick. As shown above, tick friction turns a neutral rebate into a real cost increase for the taker. Any empirical study spanning a tick-size change must re-estimate, not assume neutrality.
3. Equating the raw spread with the cum-fee spread. The raw spread falls when a rebate is introduced (mechanical pass-through) - this is not evidence of improved liquidity unless the cum-fee spread also falls. Malinova & Park (2015) find precisely this: posted quotes adjust, but taker cum-fee costs are unchanged; what does change is retail order aggressiveness.
4. Confusing maker-taker fees with payment for order flow. PFOF is a payment from a market maker/wholesaler to a broker for routing flow; maker-taker is a payment from an exchange to a liquidity supplier. They are economically distinct (see 06 · Advanced Extensions).
5. Tier gaming / wash incentives. Rebate tiers are volume-based, so a desk at a tier boundary has an incentive to trade purely to reach the next rebate - a genuine distortion, and the reason regulators scrutinise maker-taker alongside best-execution rules.

---

5. References

• Colliard & Foucault (2012), Trading fees and efficiency in limit order markets, RFS 25(11), 3389–3421
• Malinova & Park (2015), Subsidizing liquidity: the impact of make/take fees on market quality, JF 70(2), 509–536
• Foucault, Kadan & Kandel (2013), Liquidity cycles and make/take fees in electronic markets, JF 68(1)
• Angel, Harris & Spatt (2011), Equity trading in the 21st century, QJF 1(1)

---

6. Connected Graph Bridges

• Back: 02 · Market-Maker P&L
• Forward: 04 · Competition & the Race to Zero - what happens to the residual when neutrality does hold.
• Regulation: 06 · Advanced Extensions (access-fee caps, tick-size rules, PFOF)
• Mechanics: Limit Order Book Mechanics & L3