Case Study 1 — The 2020 Commission Caps: What Happens When a City Regulates One Side of a Two-Sided Market

A real, public policy episode. All statements below are limited to what the public record establishes: that a number of U.S. cities adopted emergency caps on third-party delivery commissions during the COVID-19 shutdowns of 2020, that some later made caps permanent, that the platforms litigated at least one permanent cap, and that platforms in some capped markets added or raised guest-side fees. No specific ordinance text, cap percentage, market share, revenue figure, or court outcome is asserted here beyond that general shape — verify the ordinance in any market you operate in. Where this case study models arithmetic, the numbers are constructed and labeled.

Background: how a fee became a policy question

Third-party delivery marketplaces spent the second half of the 2010s building something restaurants could not build for themselves — a demand-side product with aggregation, saved payment, live tracking, and a driver network. In exchange they took a percentage of the menu subtotal. Restaurants signed the agreements because the alternative was absence from the place where an increasing share of a neighborhood decided what to eat.

For five years this was a private commercial arrangement, and the objections to it were the ordinary objections operators have to any vendor: it costs too much, the terms favor the vendor, the data is theirs. Nothing about that made it a matter for a city council.

Then, in March 2020, dining rooms across the United States were closed by public order.

The arrangement's economics did not change. Its context did, completely. A commission that had been a cost of an optional growth channel became a cost of the only channel a restaurant had. Restaurants that had never delivered were suddenly dependent on platforms for a majority of their revenue, at rates negotiated when delivery was a small share of a small number of restaurants' business, with no leverage whatsoever and no alternative to walk to.

That is the condition under which a private pricing dispute becomes a public policy question, and it did. Several U.S. cities — New York City, San Francisco, and Seattle among them — capped what third-party delivery platforms could charge restaurants. The common architecture of these ordinances was a limit around 15% of the order for delivery services, with a small number of additional percentage points permitted for other services such as marketing or payment processing. Specifics varied by city, and any operator relying on a cap should read the ordinance rather than a summary of it.

The operating issue: what the cap was actually trying to fix

It is worth being precise about the problem the caps addressed, because it was not "commissions are high."

It was the absence of an exit. Chapter 28's §28.9 makes the point that negotiating leverage on a platform comes from volume, desirability in a thin market, or a credible willingness to leave. In spring 2020, independent restaurants had none of the three simultaneously and for reasons entirely outside their control. A dining room closed by public order cannot threaten to take its business elsewhere.

Economists call this a two-sided market: the platform serves guests on one side and restaurants on the other, and its power comes from being the place where the two meet. Regulating the restaurant-facing price of such a market is a specific and limited intervention. It says nothing about what the platform may charge the guest.

That omission turned out to be the interesting part.

What it shows: three lessons that outlive the ordinances

One: a cap changes the arithmetic materially, and does not change the strategy

Take Chapter 28's modeled order and run it at both rates. [Constructed teaching example, using this book's Bellwether figures.]

A $65.00 order carrying $19.32 of plate cost, $2.24 of packaging, and $1.30 of packing labor — $22.86 of the restaurant's own variable cost:

Ten points of commission is $6.50 an order, or 25% more contribution. That is a large, real improvement, and on a restaurant doing 200 orders a week it is $67,600 a year.

And it does not change the answer to Chapter 28's central question. A capped marketplace order still returns roughly half of what the same food returns sold to two people at a table, still consumes a hearth slot, still hands the guest's identity to somebody else, and still cannot be served during a binding hour without displacing better business. A cap is a floor under a bad deal, not a reason to build a business on it.

Two: regulate one side of a two-sided market and the other side moves

Platforms in some capped markets responded by adding or raising fees charged to the guest — the one participant the ordinances did not address. This is the least surprising outcome in the entire episode and it was widely reported at the time.

For an operator, the consequence is specific and unpleasant: the guest's total rises, the restaurant's menu price is what the guest sees at the top of the receipt, and the restaurant absorbs the disappointment. A guest paying $70 in fees and tips on a $50 order does not experience that as a municipal policy interaction. They experience it as an expensive dinner from your restaurant.

The generalizable lesson is not "caps are bad." It is that a two-sided market rebalances, and any intervention on one side should be evaluated for what it does to the other. Operators who advocate for caps in their own city — which is a legitimate thing to do — should advocate with that in view.

Three: emergency policy has a durability problem, and so do the business decisions built on it

Some cities let their caps expire with the state of emergency. Some made them permanent. At least one permanent cap was challenged in court by platforms arguing that it unconstitutionally interfered with private contracts.

The practical consequence for an operator is that your channel economics may be sitting on a legal foundation that could move. A restaurant that built a delivery-dependent model on capped economics — priced the menu to it, staffed to it, budgeted to it — is exposed to a court decision or a sunset clause it does not control.

That is the same category of risk as an expiring lease concession or a promotional vendor rate: real money, a known expiry, and an operator's obligation to model the world after it changes. Chapter 4's sensitivity analysis exists precisely for this.

Outcome

The public record supports a modest summary and no more.

• Emergency caps were widely adopted in 2020 and provided real, immediate relief to restaurants at a moment when they had no alternative channel and no bargaining power.
• The caps did not survive uniformly. Some expired; some were extended; some were made permanent; at least one permanent cap was litigated.
• Platforms adjusted guest-side pricing in some capped markets.
• Off-premise volume did not return to 2019 levels after dining rooms reopened, which is the durable fact underneath the whole episode: the habit outlasted the emergency.

What is not established — and should not be asserted by anyone teaching this case — is any specific claim about how caps affected individual restaurants' survival rates, platform profitability, order volumes, or driver earnings. Those are contested empirical questions and the honest position is that we do not have a settled answer.

The lesson

Regulation can improve the terms of a channel. It cannot make a channel strategic.

Every structural weakness Chapter 28 identifies survives a commission cap intact: the missing beverage, the throughput cost at a binding hour, the guest's identity going to somebody else, the service recovery you cannot perform, and the refund you did not cause and cannot dispute. A cap addresses one line on the payout statement. It does not address the other four.

The operator's takeaway is therefore not "campaign for a cap and then relax." It is:

1. Know whether a cap applies where you operate, and whether it is permanent, expired, or in litigation. It is worth up to ten points of revenue on the channel.
2. Do not build a business model on a regulated price you do not control.
3. Build the thing a cap cannot give you — a first-party channel, an email list, and a guest who knows your name. That is the only durable form of leverage in §28.9, and it is the one you can start on Monday without a city council.

Discussion questions

1. The caps addressed the price platforms charge restaurants but not the price they charge guests. Was that a design flaw, a deliberate scope limitation, or an unavoidable feature of what a city can regulate? Argue both sides.
2. Compute the contribution improvement a 15% cap produces on your own restaurant's average order (or on Bellwether's $65 order). Does it change any decision in Chapter 28? Which ones does it not touch?
3. A cap makes a channel less bad. Under what circumstances does making a channel less bad make it more dangerous to an operator's strategy?
4. Restaurants had no exit in spring 2020 and therefore no leverage. Name three things an independent operator can do in an ordinary year to make sure they always have an exit from a vendor relationship. What does each one cost?
5. Suppose your city is debating a permanent cap and you have been asked to testify. Write the three points you would make and the one honest concession you would offer.
6. The harder question: if the habit outlasted the emergency, and guests now expect delivery from restaurants that never used to offer it, is a restaurant that declines the channel entirely making a defensible strategic choice or a nostalgic one? What evidence would settle it for a specific restaurant?