For two chapters, something has stayed fixed: the oven, the lease, the building. That is what made it the short run. The long run is long enough that every input can change: sign a bigger lease, buy a second oven, build a new kitchen, or sell the whole thing. Neither is a set number of months. A food truck's long run might be a few weeks; a chip factory's is years. The long run is however long it takes to change the slowest input.

One consequence follows directly: in the long run, there are no fixed costs. The rent you're locked into today becomes, at renewal, one option among several. That is the formal version of the "cost as opportunity" reading from Chapter 9: given enough time, a firm can rethink any cost it now treats as fixed.

Choosing a plant. Suppose the workshop is planning ahead and can choose among four sizes of shop. Once built, each has its own fixed cost and its own bottleneck, so each has its own short-run ATC curve — its own U. The small shop is the workshop from Chapter 9, which does best at 4 units. Bigger shops cost more to run at low output but can handle more before they jam. For whatever output it plans to produce, the firm picks the shop with the lowest average cost. Read down any column of the table: the lowest number is the long-run average cost of that output.

Traced across every output, those lowest costs form the long-run average total cost curve (LRATC). Its shape has three parts.

Economies of scale. Where LRATC falls, bigger is cheaper per unit: the firm has economies of scale. A bigger operation can split work into narrower specialties, run larger and faster machines, and buy inputs in bulk. Costco stocks only about 4,000 items, so it buys each in enormous volume and negotiates with suppliers in a way a corner store never could. Where up-front costs are huge and each extra copy is cheap, the effect is dramatic: a drug may cost a billion dollars to develop and pennies to manufacture, and the same holds for a film, a video game, or a semiconductor fab.

Constant returns to scale. Where LRATC is flat, doubling every input roughly doubles output, so cost per unit holds steady and a firm can grow without getting cheaper or costlier per unit. The lowest output at which LRATC reaches its minimum is the minimum efficient scale — 8 units in the table. When that scale is small relative to the market, as for restaurants, salons, and contractors, small and large firms survive side by side.

Diseconomies of scale. Where LRATC rises, the firm has grown past its best size and bigger is costlier per unit: diseconomies of scale. The bottleneck returns in a new form. In the long run you can buy another oven, but you can't easily buy more of the founder's attention. As a firm adds layers, information passes through more hands, decisions wait on more approvals, and more of the payroll goes to coordinating the work rather than doing it.

Keep the two time horizons separate. A rising short-run curve comes from piling variable inputs onto a fixed one; a rising long-run curve comes from an organization too large to coordinate. Both raise costs, for different reasons. Your exam checks whether you can tell them apart.

Choosing a plant is one long-run decision. The bigger one is whether to be in the industry at all, and that is where profit comes in.

In the short run, profit is a scoreboard: it tells the firm whether it is beating its next-best alternative. In the long run, profit is also a flare. Everyone else can see it, and in the long run they can act on it.

Entry is what happens when profit draws in competitors: new firms start up and existing ones expand, so market supply rises. Exit is what happens when losses push firms out: they close, sell their equipment, and move their money and people elsewhere, so market supply falls. Exit is not the same as the short-run shutdown from Chapter 10. A firm that shuts down still owes its fixed costs and can reopen next week; a firm that exits has escaped them for good.

Trace the profit case on two graphs at once. On the firm's graph, price is above ATC, so there is a profit rectangle. That profit draws in new firms, which shifts the market supply curve to the right. The market price falls, and since each firm takes that price as given, its horizontal price line falls too. Each step down shrinks the profit rectangle. Entry keeps going until the rectangle is gone: price has fallen to the bottom of ATC, and economic profit is zero.

Losses run the same story backward. Firms exit, market supply shifts left, the price rises, and the remaining firms' losses shrink until price is back at the bottom of ATC. Easy profits come and go for the same reason: anything easy to earn is easy to copy.

Entry continues while firms earn economic profit, and exit continues while they take losses. The movement stops only when economic profit is zero. Every firm is then producing where price equals marginal cost (its profit-maximizing quantity) and where price equals average total cost (zero profit). Both can be true only at the bottom of ATC, where MC crosses it. That gives the long-run equilibrium condition for a perfectly competitive firm: P = MR = MC = min ATC.

Zero economic profit sounds like failure, but it is the normal profit from Chapter 10. The owner covers every cost, including what her time and money could earn elsewhere. She is doing exactly as well as her next-best option: well enough to stay, not well enough to draw in anyone new.

The end point of entry is good news for buyers. Economists don't usually respond to profit from a good idea by saying, "Cap it." They ask, "Can competitors enter?" If entry is open, the first firm's profit summons rivals, supply expands, and the price falls until buyers capture most of the gain. A firm wants its profit to last; consumers want the value to spread. Free entry is how a private discovery becomes a public gain.

Long-run perfect competition also hits two kinds of efficiency at once. Allocative efficiency means P = MC: the value buyers place on one more unit equals the cost of producing it, so no unit worth more than it costs goes unmade, and no unit is made that costs more than it is worth. Productive efficiency means producing at min ATC: each unit is made at the lowest cost per unit the technology allows.

Long-run equilibrium is a resting point, not a permanent address. Tastes change, input prices change, and governments add taxes and fees. Every shock plays out in two acts. In the short run, the number of firms is fixed, so each firm can only slide along its MC curve to a new quantity. In the long run, entry and exit change the number of firms until the price is back at the bottom of ATC, which may be a new bottom.

Before you draw anything, ask one question: did the shock change the price, or did it change the firm's costs? The answer tells you which curve moves on the firm's graph.

A demand shock moves the price line. Suppose protein bars catch on and market demand rises. The market price climbs from P₁ to P₂ along the market supply curve. On the firm's graph, the horizontal price line rises with it, and the firm slides up its MC curve to a larger quantity. Nothing about the firm's costs has changed: same kitchen, same ingredients, same ATC and MC. The profit rectangle appears only because the price moved above ATC.

Then the flare goes up. Entry shifts market supply right until the price is back at the bottom of ATC, and each firm is back at its original quantity. The extra output comes from more firms, not bigger ones. A fall in demand runs the story backward: the price line drops, firms take losses, some exit, and the price climbs back to the bottom of ATC with fewer firms.

Connect A and C in Fig 11.5 and you have the industry's long-run supply curve. If new firms can enter with the same costs as the old ones, the industry can grow without its costs rising, so in the long run it will supply any quantity at the same price: the bottom of ATC. Its long-run supply curve is horizontal. Economists call that a constant-cost industry. If growth bids up the price of something the industry needs (land, specialized workers), each new firm costs a little more to run, and long-run supply slopes upward instead: an increasing-cost industry. That is the only way a change in demand reaches the firm's cost curves: not because buyers want more, but because the industry's expansion raises the prices of its inputs.

A cost shock moves the cost curves. A cost shock starts on the firm's graph instead. Which curves move depends on which kind of cost changed, and you already have the rule: fixed cost never enters marginal cost.

Say a tariff raises the cost of steel for a competitive industry of tool makers by $15 per tool. MC shifts up, market supply shifts left, and the price rises, but by less than $15, because buyers cut back as the price climbs. Each firm's ATC is now $15 higher, so firms take losses. In the long run, losses cause exit. Supply keeps shifting left until the price equals the bottom of the new ATC. A per-unit cost lifts ATC straight up without moving its lowest point left or right, so the price ends exactly $15 higher and each surviving firm makes the same quantity as before. The industry shrinks through fewer firms, and in a constant-cost industry buyers end up paying the whole $15.

Now say the city doubles the annual license fee for food trucks. Nothing changes at the margin: the next taco costs what it did, so MC doesn't move and neither does market supply. In the short run, the price and each truck's quantity stay put. Every truck simply earns less: the price is now below its higher ATC, so it takes a loss. It keeps operating, because closing wouldn't refund the fee. In the long run, some trucks exit, supply shifts left, and the price rises to the bottom of the new ATC. That bottom sits higher and farther right, because MC still crosses ATC at its lowest point and MC hasn't moved. Spreading a bigger fixed cost over more tacos is the cheapest way to cover it. The trucks that remain are fewer and busier.

If entry erodes profit, a firm that wants its profit to last has two broad options. The first is to keep creating value: lower costs, improve quality, earn trust, solve the next problem before anyone else does. The second is to make entry harder, either by building an advantage rivals can't easily copy or by getting the government to keep them out (Concept 6). To see the difference clearly, start with the businesses most often accused of creating nothing.

The wholesaler, the broker, the ride app, the reseller: they "make nothing," and the popular verdict is that they are parasites skimming a cut between the people who do the real work. "Cut out the middleman" is one of the most common pieces of business advice there is. The economist's reply is that the middleman is often the one creating the most value in the chain. You just can't see it, because what they produce isn't a thing.

The clearest account comes from the Duke economist Michael Munger, who argues that firms like Uber don't really sell rides. They sell reductions in transaction costs. A transaction cost is everything that stands between you and a deal, beyond the price itself, and Munger sorts it into three: triangulation (finding the right partner and agreeing on a price), transfer (actually paying and delivering), and trust (believing the other side will come through). Before Uber, a stranger's car was useless to you: you couldn't find it, couldn't price it, couldn't trust the driver. The app didn't build a car; it manufactured triangulation, transfer, and trust, and that is a real product. To a customer, a dollar of hassle removed is worth as much as a dollar off the price.

Which gives us the honest line. Picture two ways to make money on a river. One person dredges the channel so barges pass faster and cheaper: a real cost falls, value is created, and the fee is earned. Another stretches a chain across the water and charges boats to lower it: nothing moves faster and no cost falls; they have added an obstacle and rented out the key. Same fee collected, opposite economics. The first creates value; the second only extracts it, which economists call rent-seeking . The test for any middleman, any platform's cut, any toll: did a real cost fall because you're here, or did you just hang a chain across the river?

Concept 5 ended with a chain across the river. It's tempting to conclude that anything keeping rivals out is a chain. But some of the most valuable things a firm can build are barriers, and some barriers are deals society makes on purpose. The useful question isn't whether a barrier exists. It's where the barrier came from.

A barrier to entry lets economic profit last because the flare goes up but rivals can't answer it. That sounds like bad news for buyers, and sometimes it is. But barriers come from three very different places, and economists treat them very differently.

Earned barriers. Some barriers are what's left behind by creating value. Walmart spent decades driving down the cost of moving goods (Concept 1), and a newcomer can't match that cost on its first day. A reputation for showing up on time, a brand customers trust, know-how that took years to learn, a network that gets more useful as more people join: each makes entry harder, and each was built by serving customers better than the alternatives. Call it a moat. A moat is the reward for creating value, and the chance to build one is a big part of why firms keep investing in lower costs and better products. Earned moats also stay contestable. A rival who finds a cheaper way or a better product can still cross, as Sears, Blockbuster, and BlackBerry each learned.

Traded barriers. Some barriers are deals that society makes on purpose. A new drug can cost more than a billion dollars to develop and pennies to copy (Concept 1). If rivals could copy it the day it launched, the profit flare would go out at once, and no one would pay to discover the next one. A patent blocks copying for about 20 years: buyers pay higher prices for a while, and in exchange the invention gets made. Copyright does the same for books, songs, and software. A license can be a fair trade too, when buyers can't judge quality until it's too late: surgery, electrical wiring, a bridge. In each case the barrier is the price, and the invention or the safety is what society buys with it.

Protected barriers. Some barriers buy nothing for buyers. A cap on the number of taxi licenses, a three-year permit wait, a tariff, or a subsidy only incumbents get keeps rivals out without making the product cheaper, better, or safer. That is the chain across the river, stretched across a whole market.

So the economist's question is never "barrier or no barrier?" It's "where did this barrier come from?" If customers built it by choosing a firm again and again, let rivals try to beat it. If it's a trade, ask whether society is getting what it pays for: does the patent last longer than the research needs, and does the license demand more than safety requires? If it mainly protects insiders, remove it.

That points to the policy question. A contestable market is one that rivals can enter and challenge, and even the threat of entry keeps incumbents honest. So when profit persists, don't ask only "why are these firms greedy?" Ask "why hasn't entry happened?" Sometimes the industry is genuinely hard. Sometimes entry is blocked by law, paperwork, or privilege, and then the cure is to remove the block. Trade is one of the most powerful forms of entry: imports are competitors. A tariff (a tax on imports) or a quota (a limit on them) is a barrier to entry by another name. It protects domestic producers' profits from competition and makes buyers pay more. Trade gets its own chapter later in the course.