Market Power I: Price-Searchers and How They Price

So far we have mostly studied a seller who has no choice about price. A wheat farmer takes whatever the market offers; if he asks a penny more, buyers walk to the next stall, and he can sell all he wants at the going rate without ever cutting it. That farmer is a price-taker. But most of the firms you actually deal with are not like that. The coffee shop on your corner sets its own prices and changes them. So does your phone carrier, your airline, your pharmacy, and your university. None of them faces a single market price they must accept. Each one searches for the price that serves it best.

That difference looks small and turns out to drive almost everything in this topic. The gap between a seller who must set its own price and one who simply accepts a price explains why a sold-out concert can mean the promoter lost money, why senior citizens cruise cheaper than honeymooners, why a printer is sold near cost while the ink costs a fortune, and why several firms that agree to hold prices up keep secretly undercutting each other anyway. Part A builds the tool and turns it on how price-searchers actually price (discrimination, tie-ins, patents) and on why cartels tend to fall apart on their own. Part B turns it on the policies meant to police market power: antitrust law and economic regulation.

A Price-Searcher Faces a Downward-Sloping Demand Curve

Start with the dividing line. A price-taker faces a flat, horizontal demand for its own output: at the market price it can sell as much or as little as it likes, and one more unit fetches the same price as the last. A price-searcher faces a demand curve that slopes downward, so to sell one more unit it has to lower the price; nobody hands it a price, it has to find one.

Why would any seller face a downward-sloping demand of its own? Because buyers do not see all sellers as identical. Even goods that look interchangeable are not interchangeable to the people buying them: one gas station is closer to your route, one drugstore has a pharmacist you trust, one brand of headphones has a reputation the no-name version has not earned. These differences can be small, but they are real, and they mean a seller can raise its price a little and lose some customers, not all.

This is also where brand names and advertising earn their keep, and the case cuts both ways. One view says advertising manufactures fake differences and pushes people to overpay for the same thing. The other says brand names and advertising lower the cost of finding out which sellers are actually reliable, so buyers can be more discriminating, not less. Both contain some truth: an ad can mislead, but strip away all advertising and you have not made buyers smarter, only made it harder for them to learn who delivers. A trademark, in particular, lets a customer reward a producer who keeps quality up and punish one who lets it slip. A brand preference can be genuine and still be slight: if a near-identical substitute is a dollar cheaper, switching to it is not fickleness, it is good sense. So almost every real seller is a price-searcher in this narrow sense, which will tempt you to call almost every firm a “monopoly.” We draw that line later.

A Price-Searcher Heeds Marginal Revenue, Which Is Less Than Price

One fact does the heavy lifting all topic. When a price-searcher cuts the price to sell one more unit, it does not cut the price for that one buyer alone; it cuts the price on every unit it sells. So the extra money it takes in from selling one more unit is not the price of that unit. It is the price of the new unit minus the revenue it gave up by marking down everything else.

Economists call the extra revenue from selling one more unit marginal revenue. For a price-taker it equals the price, because selling more forces no markdown. For a price-searcher, marginal revenue is less than the price, always, as the table below shows. Read it as a seller deciding how low to set the price to sell a given daily quantity.

Price	Daily quantity	Total revenue	Marginal revenue
$1.00	1	$1.00	$1.00
.90	2	1.80	.80
.80	3	2.40	.60
.70	4	2.80	.40
.60	5	3.00	.20
.50	6	3.00	.00
.40	7	2.80	−.20
.30	8	2.40	−.40
.20	9	1.80	−.60
.10	10	1.00	−.80

Look at the move from 2 units to 3. To sell the third bottle the seller drops the price from $.90 to $.80, yet total revenue rises by only $.60, from $1.80 to $2.40. The missing $.20 is the markdown taken on the two bottles that would have sold at $.90. That is the gap between price and marginal revenue, and it widens as you go down the table.

Two questions students always raise are answered right there. Why not just set the price three times higher? Because the price you can charge is not yours to wish into being; it is set by what buyers will do, and pushing the price up shrinks the quantity so much that revenue can fall. And where is revenue largest? At six units, where marginal revenue hits zero; push past that and marginal revenue goes negative, so you are paying, in markdowns, for the privilege of selling more. If a seller’s costs were literally zero, that revenue-maximizing point, where marginal revenue equals zero, is exactly where it would stop. The responsiveness of buyers to a price change is the demand’s elasticity: demand is elastic when a small price cut raises total revenue and inelastic when a price cut lowers it. Right at the revenue-maximizing output, where marginal revenue is zero, demand is exactly unit-elastic: a small price change leaves total revenue unchanged. Above that output demand is inelastic; below it, elastic.

A Price-Searcher Produces Where Marginal Revenue Equals Marginal Cost

Revenue is only half the story. Producing each unit costs something, and the extra cost of one more unit is its marginal cost. A price-searcher maximizes profit where marginal revenue equals marginal cost: not where average cost is lowest, not where revenue is highest, but where the last unit’s extra revenue just covers its extra cost. Add this water-seller’s costs to the same demand and the optimum jumps out.

Price	Quantity	Total revenue	Marginal revenue	Marginal cost	Total cost	Profit
$1.00	1	$1.00	$1.00	$0.10	$0.30	$0.70
.90	2	1.80	.80	.20	.50	1.30
.80	3	2.40	.60	.30	.80	1.60
.70	4	2.80	.40	.40	1.20	1.60
.60	5	3.00	.20	.50	1.70	1.30
.50	6	3.00	.00	.60	2.30	.70
.40	7	2.80	−.20	.70	3.00	−.20
.30	8	2.40	−.40	.80	3.80	−1.40

Profit peaks at three bottles, sold at $.80, for a profit of $1.60. A fourth bottle would add $.40 of revenue and $.40 of cost, a wash; a fifth would cost more than it brings in. Notice what is absent: average cost tells you nothing about what price to set or whether you are even profitable. Only marginal revenue and marginal cost decide the output, and the demand curve then tells you the highest price at which that output will sell.

A useful consequence falls right out of this. Suppose the government slaps a flat annual license fee on the seller, fixed regardless of how much it sells. A fixed charge changes neither marginal revenue nor marginal cost, so it changes neither the profit-maximizing price nor the quantity; the seller does exactly what it did before and simply earns less profit. Keep this straight from a per-unit tax, which adds to the cost of each unit, shifts marginal cost, and does change price and output. We handle the per-unit tax, and how its burden splits between buyers and sellers, back when we did competition and the tax wedge.

The same logic clears up real puzzles. A concert promoter thrilled that the show sold out four months early should think twice: a clean sellout that fast is a sign the price was set too low, marginal revenue still above marginal cost. A private tutor who is always booked solid, turning students away, has almost certainly priced too low. Idle time, by contrast, does not automatically mean the price is too high; it depends on how sensitive demand is. A price-searcher cannot read its demand curve off a chart; it searches, by trial and error, raising and lowering and watching what happens. Getting it wrong is punished: a price far from the profit-maximizing one invites rivals to undercut the margin or grab the business it drove away.

Market Power Distorts Output Because Marginal Revenue Sits Below Price

Now the part that worries people about market power. Because a price-searcher stops where marginal revenue equals marginal cost, and marginal revenue is below the price, it stops where the price still sits above marginal cost. There are units worth more to consumers than they cost to produce, and they go unmade.

Picture it concretely. Suppose marginal revenue equals marginal cost at five units, and to sell those five the firm charges sixteen dollars. A sixth unit would cost only about twelve dollars to make, and some buyer values it near the sixteen-dollar price. That sixth unit does not get made; resources that could have produced something worth sixteen dollars go off to produce something worth less. That forgone gain is what economists mean by the distortion of market power. I am describing it in words on purpose: we are not measuring a deadweight-loss triangle, only noting that valuable units go unproduced.

Two cautions keep this honest. The benchmark we compare against is the competitive, price-taking outcome, where the seller has no power to mark up and output is pushed until cost rises to meet the price; that competitive standard is the heart of our competition-and-efficiency material. And do not oversell the harm: some of what looks like restricted output is really the cost of variety, a hundred slightly different products made in smaller batches, what buyers paid for in differentiation. The distortion is real but milder than it first looks, and it tends to erode on its own.

What Matters Is Whether the Market Is Open or Closed, Not the Number of Sellers

Here is where most loose talk about “monopoly” goes wrong. In the trivial sense we built above, almost every seller has a downward-sloping demand and so a sliver of pricing discretion. By that test a corner grocery is a “monopoly” and so is a doctor’s practice; the word that loose is useless. The distinction that earns its keep is whether the market is open or closed.

A market is open when nothing but cost stands between a would-be competitor and entry. It is closed, or restricted, when entry is blocked by something other than the ordinary difficulty of competing: a government license, a patent, an exclusive franchise, an import tariff, an acreage limit, a professional board that controls who may practice. The number of sellers today is not the issue; what disciplines a seller is the threat of who could enter tomorrow. An open market keeps a lone seller honest, because charging too much simply invites entry; a closed market lets the protected seller hold the price up and let costs drift, because the door behind it is locked.

This is why “the grocery business is a monopoly” and “medicine is a monopoly” are not the same claim. Both face downward-sloping demand, but anyone with capital can open a grocery, so the grocer is disciplined by open entry, while becoming a licensed physician means passing through a gate the profession itself guards, so medicine is genuinely closed — and that is where you expect higher prices and weaker cost control.

Two refinements matter. Closing a market does not automatically turn a price-taker into a price-searcher: you can restrict entry, say by capping acreage or licensing a trade, and still leave so many small sellers that each remains too tiny to budge the price; restricted wheat or tobacco growers can still be price-takers. And a price rise driven by higher costs is not an exercise of market power at all: if everyone’s input prices rise, every producer’s costs and prices rise with them, with no one “wielding” anything. Cost-driven and power-driven price increases look alike on a receipt and are completely different underneath.

When a closed market does pay the protected seller above what it costs to do the work, that surplus has a name: a monopoly rent, collected not because you created extra value but because access to the activity has been artificially restricted. A taxi medallion that sells for a fortune, a license that lets a few firms supply a city, an exclusive grant to serve a stadium: each can throw off rents, a transfer from buyers won by locking the door. This is why I am careful when people say “every profit is just society rewarding someone for moving resources to a higher-valued use.” Usually true, but monopoly rents from a closed market are a real exception: there the gain comes from contrived scarcity, not from creating value.

Long-Run Entry Drives a Price-Searcher’s Profit Toward Zero

If the market is open, market power is a fast-melting asset. A price-searcher earning fat profits is a flare in the night sky for imitators: rivals copy the product, open nearby, and chip away at the demand facing the original seller, the price the first firm can charge falls, and its profit shrinks. In the long run, entry pushes that profit down toward zero, where the price just covers full average cost.

Where did the profit go? Partly it gets competed away in lower prices, and partly it gets absorbed into costs: if a prime location, a star chef, or a skilled manager is what made the profits, competitors bid up the price of that scarce ingredient until the extra earnings show up as a cost of hiring it rather than as profit. This is the same absorption-into-costs idea we met before, applied to market power: durable profit in an open market is hard to hold because the door keeps opening behind you. Sustained, comfortable profit is a symptom of a closed market.

Price Discrimination Is Marginal-Revenue Equalization, Not a Moral Failing

Now to the family of pricing tactics that make market power interesting. The first is price discrimination: charging different buyers different prices for essentially the same thing. The name sounds sinister; the mechanics are not. A seller with two groups of buyers who differ in price-sensitivity can earn more by charging each its own price, the rule being to equalize marginal revenue across the groups, then produce where that common marginal revenue equals marginal cost. Work it through with two customers, A and B, who value the good differently.

Price	A’s quantity	A’s total revenue	A’s marginal revenue	B’s quantity	B’s total revenue	B’s marginal revenue
$12	1	12	$12	0	0	—
11	2	22	10	0	0	—
10	3	30	8	0	0	—
9	4	36	6	0	0	—
8	5	40	4	1	8	8
7	6	42	2	2	14	6
6	7	42	0	3	18	4
5	8	40	−2	4	20	2
4	9	36	−4	5	20	0

Suppose the seller is charging both customers $7. From A it sells 6 units; from B, 2. But the last unit sold to A brings in only $2 at the margin, while the last to B brings in $6, so the seller can do better. Raise A’s price to $8 and A buys 5 instead of 6, costing $2 of revenue; take that freed-up unit and sell it to B by dropping B’s price to $6, where B buys 3, adding $4. The $4 gained beats the $2 lost, total revenue rises, and the two marginal revenues now sit at $4 apiece. That is the rule: shift units toward the buyer with the higher marginal revenue until the margins are equal. The different prices are the result, not the goal.

Read as marginal-revenue equalization, a long list of everyday pricing is all the same move. Senior, student, and off-peak discounts charge a lower price to the more price-sensitive buyer; cruise lines and resorts go further on idle capacity, since empty cabins and rooms cost almost nothing to fill, which is why a luxury hotel late in the day may price below a budget motel. “Dumping,” selling cheaper abroad than at home, is the same move across two markets, not selling below cost: every unit still covers its marginal cost, since a seller never knowingly sells below it. Coupons and rebates deliver a lower price only to the price-sensitive customers who bother to clip and redeem; a wider Regular-to-Premium gap at the full-service gas island is the same judgment that one set of buyers is less price-sensitive, not a cost difference. Tuition scholarships pair a high sticker price with targeted markdowns, charging each family closer to what it will pay; calling it “discriminatory” describes the practice, it does not condemn it.

The old rail-rate dispute between New York, Denver, and San Francisco makes the same point without a cost difference. Railroads charged more for the shorter trip to Denver because San Francisco freight faced water competition through the Panama Canal. When law required equal rates, the railroads could not raise the San Francisco price without losing traffic to ships, so they cut Denver’s rate. Had Denver supplied most of the revenue, they would instead have raised San Francisco’s. The identical mile does not set the price; the demand and alternatives facing each buyer do.

The same logic can be turned on a single buyer, one unit at a time, and this is where discrimination captures the most. A buyer values the first unit most and each later unit a little less, so a seller who posts one price for everything leaves a lot of that value on the table. Sequential pricing, also called declining-block pricing, claws some back by offering the early units dear and successive units cheaper. Take a buyer landscaping a yard who would buy five trees at $6, a sixth at $5, and two more at $3. Sell them in that descending sequence and the buyer pays $41 for eight trees (five at $6, one at $5, two at $3). Post a single price of $3 and that same buyer buys the same eight trees for $24. Same trees, same final per-unit price, $17 more in the seller’s pocket; what changed is how much of the gap between the buyer’s total value and the price the seller captures rather than leaves to the buyer.

A two-part tariff does the same job more simply: a fixed entry fee plus a low per-unit price. A warehouse club charges an annual membership and prices goods near cost; an amusement park charges admission and runs the rides cheap. A buyer who would pay a $5 membership to then buy trees at $3 each ends up in the same place, unit for unit, as buying at $3 with no fee, except the seller has now also collected the $5, the lump it could not get by posting a single per-unit price. When the per-unit price is set at marginal cost, this arrangement produces exactly the efficient quantity that ordinary marginal-cost pricing would; only the fixed fee changes who captures the gains from trade.

One condition makes all of this possible: the low-price buyers must not be able to resell to the high-price buyers, or they would buy cheap, undercut the seller, and collapse the two prices into one. That is why discrimination thrives where resale is hard: on services, on personalized deals, on tickets tied to an ID.

Tie-Ins Meter Demand and Tend to Raise Output, Not Restrict It

The second tactic is the tie-in: sell one product on the condition that the buyer also buy a second from you, often a consumable like paper or ink. The folk theory is that a firm uses its grip on the first product to “leverage” a monopoly into the second. The economics says otherwise: a tie-in is usually a way to meter how intensely each customer uses the main product and charge accordingly.

Take a copier sold to two customers, Big and Small, who differ in how much they copy. Start with the worth of the machine to each, before any tie-in:

	Big	Small
Worth of copies	$430	$250
Cost of labor	$100	$90
Cost of paper (1¢/sheet)	$30 (3,000 sheets)	$10 (1,000 sheets)
Worth of the copier	$300	$150

The seller’s cost to make a copier is $30. With a single price to both, the best it can do is charge $150, the most Small will pay, sell two machines, and earn $240. Charging Big $300 and Small $150 separately would earn $390, but that is naked price discrimination, which courts both resale and a lawsuit. And the seller does not even know each buyer’s exact worth; it knows only that the worths differ, and a tie-in lets it sort them without asking.

So the seller offers the machine to anyone for $55 and requires that all paper be bought from it at 10¢ a sheet, nine cents above the going rate.

	Big	Small
Worth of copier’s product	$410	$240
Cost of labor	$100	$90
Cost of paper (10¢/sheet)	$250 (2,500 sheets)	$95 (950 sheets)
Cost of copier	$30	$30
Worth of the copier	$60	$55

The dear paper makes each customer use a little less, so the machine is worth a bit less, but both still buy. The seller earns only $50 on the two machines. The action is in the paper: Big buys $250 of paper that costs the seller $25, a $225 profit; Small buys $95 of paper that costs $9.50. Total profit comes to $365.50, far above the $240 a single machine price could earn. The tie-in let the seller charge Big more than Small without ever posting two prices, because Big, the heavier user, automatically buys more of the metered paper.

Here is the point that kills the folk theory: the seller has not “monopolized paper.” It is making its money on the copier; the paper is just the meter that reads each customer’s intensity of demand. The profit source is the tying good, not the tied one, and metering lets the seller serve both customers, including the small one it might otherwise have priced out. Bundling and tie-ins, including a software maker shipping a browser inside its operating system, are best read this way: attempts to capture more of the value buyers place on the main product, not schemes to take over a second market.

Directly metered pricing does the same job more precisely than overpricing a tied input: a per-mile charge on a rental car, a per-copy fee on a leased machine, a membership fee plus a low per-unit price all read off how much a customer actually uses the good. The same instinct explains two habits that look like leaving money on the table. A late-night hotel that turns down $50 cash for a room that cost $20 to clean is not being foolish: a stable, posted price spares buyers the uncertainty and haggling that would erode demand over the long run. Deliberately under-priced tickets, like cheap student seats, get made non-transferable for the same reason: the seller wants the low price to reach the intended buyer, not a scalper. (When underpricing comes from a price control rather than the seller’s own choice, scalping is the market’s way around the control, a story from our price-controls topic.)

Patents Reward Costly, Risky Discovery

A patent is a deliberately closed market, granted on purpose. It hands an inventor a temporary exclusive right that lets the inventor charge above marginal cost for a while. People look at a pill that costs pennies to manufacture, see it selling for dollars, and conclude the profits are obscene. The mistake is measuring cost as the cost of the last pill. The real cost is the cost of every action that produced the drug, including the years of research, and including the many research programs, at this company and its rivals, that failed and produced nothing sellable. Reward only the cheap manufacturing and you stop paying for the expensive, risky discovery. When entry into the research contest is open, the prospect of a patent draws in many competitors, the failures eat up much of the winnings, and the expected return is not the windfall the per-pill math suggests. The patent is the prize that makes anyone enter the race; the sunk research is its real price.

Cartels Are Unstable Because Every Member Has a Reason to Cheat

Now suppose several firms try to behave like a single monopolist: agree to hold output down and prices up, sharing among themselves the gains a monopolist would earn. That is a cartel, and the striking fact about cartels is how reliably they fall apart on their own.

To work, a cartel has to do several hard things at once. It must identify all the producers and get them to join. It must agree on how much each will cut. It must block new firms from entering and undercutting the propped-up price. It must police non-price competition, since members barred from cutting prices will compete on service, terms, and quality instead. And above all it must detect and punish cheating, because every single member has a powerful private incentive to cheat. If the cartel holds the price high, any one member can quietly shave its price, win a flood of new business, and pocket the gain while the others keep restricting. The holdout captures the extra sales precisely because everyone else is behaving; the reward for defecting is large and immediate, the punishment slow and uncertain, so members chisel.

History is a graveyard of cartels for exactly this reason. When nineteenth-century railroads ran competing “trunk lines” between major cities, the lines drove prices down toward the bare cost of adding freight to a train already running, so they banded together to fix rates, and the cartels kept breaking down. Once a train is rolling, an extra ton of freight is almost pure profit, so the temptation to offer a shipper a secret rebate to grab that traffic was irresistible. One secret rebate triggered another, a price war followed, and the cartel collapsed; steamboat operators had failed the same way earlier. The pattern is general: a homogeneous product, lumpy capacity, and the chance to cheat undetected are poison to collusion.

What can keep a cartel alive? One thing is a way to make cheating visible and punishable, which is why some cartels funnel all sales through a central pool, so no member can secretly cut a side deal. The other, decisive, force is government: a cartel that cannot police itself can survive if the law forces every producer to join and obey, turning a fragile private agreement into an enforced one. When European coal producers pooled their sales, the pool existed to detect and stop secret price-cutting, and members could not simply stay out because the law compelled them to join. That last move, government enforcement of what private collusion cannot sustain, is the door into our political-economy topic, where we will see how competition that is barred from the marketplace reappears in the political arena. Left to themselves, cartels cheat themselves to death.