Maker-Taker & Rebate Economics

You’ve spent the last few lessons treating the spread as the maker’s paycheck: post a bid, post an ask, buy low, sell high, collect the difference. That model is clean, intuitive, and quietly wrong about a big chunk of the real money. On most US equity venues, the exchange itself reaches into the trade and hands the maker a cash rebate for posting liquidity — while billing the taker a fee for removing it. For a tight-spread name, that rebate can be a larger slice of the maker’s edge than the spread it captures. Once cash flows like that, behavior bends around them: makers queue for stocks where the economic spread is basically zero, brokers route to whoever pays them most, and a whole shadow economy (payment for order flow) springs up to buy the cleanest order flow in the market. This lesson is about the plumbing of fees and rebates — and how that plumbing rewires the game.

The maker-taker pricing model

Analogy. Picture a nightclub that pays attractive regulars to show up early and stand at the bar — because a full-looking room draws paying customers through the door. The regulars (makers) get cash for providing the “liquidity” of a lively crowd; the late arrivals (takers) pay a cover charge to get in and enjoy it. The club keeps the gap between the cover it collects and the appearance fee it pays. That is precisely the maker-taker exchange.

Definition. Under maker-taker pricing, the exchange charges two opposite per-share prices for the two sides of every trade:

• The maker — the trader whose resting limit order was sitting on the book and got hit — receives a rebate.
• The taker — the trader whose marketable order crossed the spread and removed that resting liquidity — pays an access fee.
• The exchange keeps the spread between the fee it collects and the rebate it pays.

Realistic US-equity numbers for a high-tier venue look like this (per share):

The taker fee is capped by regulation: Reg NMS Rule 610 caps the access fee at $0.0030 per share for stocks priced $1 and above. That cap is why “30 mils” ($0.0030, where a “mil” = $0.001/share) is the canonical taker fee, and why the top maker rebate sits just below it.

Worked example. You post 10,000 shares on the bid; a market sell hits all of them. You earn 10,000 × $0.0029 = $29.00 in rebate — on top of whatever spread your eventual round trip captures. The trader who hit you paid 10,000 × $0.0030 = $30.00. The exchange netted 10,000 × $0.0001 = $1.00. Nobody traded a single share of “spread” yet, and $29 already changed hands in your favor.

Pitfall. The rebate is only paid on the passive side that actually fills. Resting an order that never executes earns you exactly nothing — and worse, if you get impatient and cross the spread to guarantee a fill, you flip to the taker side and pay 30 mils instead of collecting 29. The sign on that per-share number swings by ~60 mils depending purely on whether you posted or hit.

When it matters

Maker-taker dominates the largest US equity exchanges (the Nasdaq and NYSE-family “Tape A/B/C” books). The economics matter most for high-turnover, tight-spread names where the rebate is a meaningful fraction of total edge, and least for wide-spread, low-volume names where the spread dwarfs a third of a cent.

Effective captured spread, with rebates

Analogy. A waiter’s real income isn’t the menu price of the food — it’s the menu price plus tips minus what they tip out to the busser. Quote the spread without the rebate and you’re reading only the menu price; you’re ignoring the tip that often makes the shift worthwhile.

Definition / formula. Decompose the maker’s per-share economics on a completed round trip (buy passively, later sell passively):

where $s$ is the quoted spread, $r$ the per-share rebate, $a$ the average adverse-selection loss, and $c$ inventory-holding cost. The first term is the spread you set out to earn; the second is the two rebates — one for the passive buy, one for the passive sell — that the simple model omits entirely.

Worked per-share P&L stack. Take a penny-spread name (s = $0.01, so the half-spread is $0.005), with a maker rebate of $0.0029, average adverse selection of $0.004/share, and inventory cost of $0.001/share:

Net edge is $0.0108/share. Of that, the rebate contributes $0.0058 — about 54% of the take. Strip the rebate out and your edge collapses from +$0.0108 to +$0.0050, nearly halving it.

Pitfall. It’s tempting to model the rebate as “+0.29¢ of free margin.” But the rebate is the fragile part of the stack: it only lands if both legs fill passively, and adverse selection (the $a$ term) tends to spike exactly when fills come fast. A name can be rebate-positive on paper and net-negative the moment toxic flow eats your $a$.

When to use it

Always price the net number, never the gross spread. In ultra-tight names the rebate is the difference between a viable and an unviable quote, so it belongs in your fair-value math from the start — not as an afterthought tacked on at month-end.

How rebates reshape behavior

Analogy. Imagine a toll road that pays you a few cents to drive on it during off-peak hours. Suddenly people take trips they had no real reason to take, just to bank the credit. The payment manufactures activity that the underlying need never justified. Rebates do the same to liquidity provision.

The mechanism — rebate capture. Because the rebate is paid regardless of whether the spread itself was profitable, a maker will rationally post and queue even when the economic spread is essentially zero — the trade pays for itself through the rebate alone. This is rebate capture: strategies whose entire edge is the +29 mils, run at scale across names where the spread offers little or nothing. The maker is, in effect, being paid to stand in line.

Worked example. Suppose a name’s true edge from spread-minus-adverse-selection is exactly $0.0000/share — a coin flip. With a $0.0029 rebate per fill, a maker who turns over 50 million shares/day passively earns 50,000,000 × $0.0029 = $145,000 per day from rebate alone, on a strategy with zero spread edge. That is why so much resting volume exists on names that look unprofitable on a pure-spread basis.

Consequence — queue intensification. Tie this back to the queue lesson: if the rebate alone justifies posting, more makers crowd the front of the FIFO queue, $Q_{\text{ahead}}$ balloons, and fill rates per maker fall. Rebate capture turns the queue into a land grab — everyone races for early position to harvest the same +29 mils, which is exactly why the bulk of displayed volume in liquid US equities is passive, rebate-seeking liquidity.

Pitfall. Rebate-capture quoting is not riskless income. The same flow that fills you also adversely selects you; chasing the rebate on toxic names is how firms turn a +29-mil credit into a net loss. The rebate lowers your breakeven — it does not remove the risk beneath it.

Inverted (taker-maker) and flat-fee venues

Analogy. Most clubs pay the cool regulars and charge the latecomers. But some clubs flip it: they charge the regulars for a guaranteed spot at the front bar and pay the eager latecomers to rush in — because what that club is selling is speed of access. Same trade, opposite cash flows, different product.

Definition. An inverted (or taker-maker) venue reverses the signs: the taker receives a rebate and the maker pays a fee. Why would a maker ever pay to post? Because inverted books usually have short queues — far fewer makers want to pay — so an order there can jump to the front and fill fast. You’re buying queue priority and fill certainty with the fee. There are also flat-fee (or low-fee, “member-tier-pricing”) venues that charge a small symmetric fee to both sides and pay no rebate, marketing themselves on simplicity and neutral routing incentives.

Worked example. You urgently need a passive fill before a number prints. On the standard book you’re 200,000 shares deep behind rebate-seekers and may never fill. On an inverted venue you pay, say, $0.0010/share to post — but the queue ahead is 5,000 shares, so you fill in seconds. Paying $0.0010 to guarantee a fill you’d otherwise miss (and avoid having to cross and pay 30 mils as a taker) is a bargain. The fee bought you queue position, the scarcest thing at a one-tick spread.

Pitfall. Don’t read “inverted = you lose money.” A maker pays the inverted fee deliberately, trading rebate income for fill speed and front-of-queue priority. The mistake is routing by fee schedule alone: the cheapest-looking venue is worthless if your order never fills there.

When to use each

Use standard maker-taker when you can afford to wait and want the rebate; use inverted when fill certainty/speed beats a third of a cent (closing auctions imminent, inventory you must offload, races you must win); use flat-fee when you want minimal routing conflict and don’t care about rebates.

Tiered pricing and economies of scale

Analogy. Airlines don’t charge every flyer the same. Rack up enough miles and you get into a status tier with free upgrades and priority everything. Exchanges run loyalty programs too — except the “miles” are share volume, and the reward is a fatter rebate. The biggest flyers fly cheapest, which keeps them flying.

Definition. Exchange fee schedules are tiered: the more volume (often measured as a percentage of total market or “consolidated” volume) a member adds or removes, the better its per-share rebate and the lower its fee. A small firm might earn the base rebate; a top-tier firm earns the maximum, plus extra-credit tiers for specific behaviors (adding displayed liquidity, hitting volume in particular tapes, setting the NBBO).

Worked example. Compare two makers on the same 100-million-share day, same fills:

Identical trading, but the top-tier firm pockets $120,000 more per day — purely for being big enough to hit the top tier. Over a year (~252 trading days) that gap is ~$30 million in rebate edge the small firm cannot access.

Pitfall — entrenchment. Tiered pricing is a textbook economy of scale: the cheapest cost structure goes to whoever already trades the most, which lets them quote tighter, win more, and lock in the tier. New entrants face a structurally worse rebate, so tiers quietly entrench the incumbents — a competitive moat built out of fee schedules.

Trade-off note

Tiers reward concentration of flow on one venue, which can improve displayed depth there but reduces a small firm’s ability to compete on net edge. When you read a venue’s headline rebate, ask which tier it assumes — the advertised number is almost always the top tier few firms ever reach.

Conflicts of interest and regulation

Analogy. Imagine hiring a travel agent who’s supposed to book your best flight — but each airline secretly pays the agent a different kickback. Will the agent route you to the cheapest flight, or the one that pays them most? When the person choosing isn’t the person paying, you have an agency problem. Broker order routing under maker-taker is that travel agent.

The conflict. A broker must obtain best execution for its client. But the broker also earns (or pays) the rebate/fee on where it routes. A broker can be tempted to send marketable orders to whichever venue pays it the most rebate or charges it the least fee — even if a different venue offered the client a better realized price or fill. The interests of the router and the client diverge: that’s the maker-taker agency problem in one sentence.

The regulatory frame.

• Reg NMS Rule 610 caps the access fee at $0.0030/share (≥ $1 stocks), bounding how large the fee-vs-rebate distortion can get.
• Reg NMS Rule 611 (Order Protection Rule) forbids “trading through” a better-displayed price on another venue, forcing routers to honor the NBBO (National Best Bid and Offer) — which partly constrains routing-for-rebate, since you can’t ignore a better price just to chase a rebate.
• The SEC’s long-running maker-taker debate questions whether rebates corrupt routing. Its proposed Transaction Fee Pilot (2018) would have experimentally lowered/banned rebates on test groups to measure the effect — courts vacated it in 2020, so it was shelved, leaving the model intact.

Worked illustration. Two venues both show the NBBO of $50.00. Venue X pays the broker a +$0.0029 rebate to post; Venue Y charges $0.0030 to take. A broker with a marketable order that could be worked passively might park it on X to harvest the rebate, delaying the client’s fill, rather than take immediately on Y for a certain fill. Same NBBO, different broker incentive — the price looked equal, but the routing wasn’t neutral.

Pitfall. “The NBBO protected the client, so routing is fine” is too generous. Rule 611 protects against trading through a better price; it does not police speed, fill probability, or which equal-priced venue you chose. Plenty of routing conflict lives in the gaps the Order Protection Rule doesn’t cover.

Payment for order flow and internalization

Analogy. A casino will happily comp the flights and hotel of a tourist who plays slots for fun, but bars the card-counter at the door. Why? The tourist’s bets are uninformed — pure house edge — while the counter’s are informed and toxic. Wholesalers pay brokers for retail flow for the exact same reason: retail orders are the tourist, and they’re worth paying to get.

Definition. Under payment for order flow (PFOF), a wholesale market maker (an internalizer) pays a retail broker to route the broker’s customer orders to it instead of to a public exchange. The wholesaler then internalizes — fills the order against its own book off-exchange — typically offering price improvement versus the NBBO. The broker gets paid; the customer (usually) gets a slightly better price than the public quote; the wholesaler keeps the residual edge.

Why retail flow is so valuable — tie back to adverse selection. Recall the P&L stack: a maker’s worst enemy is the adverse-selection term $a$ — getting filled by someone who knows the price is about to move. Retail order flow is overwhelmingly uninformed: small, uncorrelated, not predictive of short-term price moves. So the $a$ on retail flow is tiny. Low adverse selection means the maker keeps far more of the spread, which is exactly why a wholesaler will pay to be the one who fills it. Uninformed flow is the cleanest, most profitable flow in the market — the opposite of the toxic flow that eats exchange makers alive.

Worked price-improvement example. The NBBO on a stock is $10.00 bid / $10.02 ask (2-cent spread). A retail customer sends a market buy of 500 shares.

• On a public exchange they’d pay the $10.02 ask: 500 × $10.02 = $5,010.00.
• The wholesaler internalizes at $10.015 — $0.005 of price improvement — so the customer pays 500 × $10.015 = $5,007.50, saving $2.50.
• The wholesaler still captures roughly half the spread (it bought low / sold here above its fair value) against flow it knows is benign, and may pay the broker, say, $0.0010/share = $0.50 as PFOF.

Everyone at the table is better off than the public quote — because the flow is uninformed enough to fund all three slices.

Pitfall / controversy. PFOF is genuinely contentious. The same payment that funds price improvement is a best-execution and conflict minefield: does the broker route to the wholesaler offering the best customer price, or the biggest broker payment? Critics argue PFOF incentivizes brokers to maximize their own payment and segments the cleanest flow off public exchanges, widening displayed spreads for everyone else. Regulators (and a 2022–2023 SEC reform push) have scrutinized exactly this trade-off.

When it matters

PFOF/internalization is the dominant path for US retail equity and options orders. It’s the mechanism that makes “commission-free” brokerage economically possible — the customer stopped paying a commission, but their flow is the product being sold.

Misconception: “rebates are free money”

The claim. “The exchange just pays me 29 mils for posting — it’s free money, so post everything everywhere and collect.”

Why it’s wrong. Three reasons, in order of bite:

1. Someone pays. The rebate isn’t conjured — it’s funded by the taker’s access fee (and the exchange’s cut). It’s a transfer from aggressors to providers, not a subsidy from nowhere. Net of fees, the system is roughly zero-sum plus the exchange’s rake.
2. It only lands on fills, against risk. You earn the rebate only when your passive order executes — and an execution is precisely the moment adverse selection can hit you. Rebate-positive and net-negative coexist comfortably on toxic names. The rebate lowers your breakeven; it doesn’t remove the risk.
3. It distorts. Chasing rebates crowds the queue (raising $Q_{\text{ahead}}$ and cutting everyone’s fill rate), and on the routing side it tempts brokers to route for their rebate over the client’s best execution. “Free money” behavior creates the very congestion and conflicts that erode the edge.

Pitfall in one line. Treating the rebate as costless margin leads you to over-post on names where adverse selection quietly exceeds the rebate — you’ll proudly collect 29 mils per fill while losing 40 to the people picking you off.

Key Takeaways