Read 2 of 5
The supply of a price-taking firm
How much should one firm produce at each market price?
The last reading followed one buyer. This one follows one seller, and the structure is the same. We ask what the firm would choose at every possible price, and the answer is its supply curve.
The firm we have in mind is a . That means two things. It can sell as much as it wants at the going market price, and it's small enough that whatever it produces leaves that price unchanged. A wheat farmer is the standard picture: the world price of wheat is what it is, and one farm's harvest doesn't move it.
Both halves matter for the decision. Because the firm can't influence the price, setting a price isn't a choice it has. Because it can sell everything it makes at that price, the only question left is how much to make. Its supply curve records the quantity that earns the most profit at each price it might face.
Week 1 gave us everything we need to answer this. Economic cost tells us which resources count. The cost function tells us what each level of output costs. Average cost tells us whether production is worth undertaking at all, and marginal cost tells us whether one more unit is worth making. This reading puts those two to work, one for each half of the decision.
2.1
Supply is a collection of production decisions
Consider Auto Eco, one of many small electric-vehicle manufacturers. Let \(q\) measure its monthly output in hundreds of vehicles. So \(q=1\) means 100 vehicles, and \(q=1.23\) means 123 vehicles. We measure \(p\) in thousands of dollars per vehicle and monthly cost and profit in hundreds of thousands of dollars. A market price of $60,000 is therefore written as \(p=60\).
Suppose Auto Eco's monthly cost of producing \(q\) hundred vehicles is
This is an economic cost function of the kind we built in Week 1. The 80 is a fixed cost of $8 million, incurred only if the firm operates this month. The \(5q^2\) term captures the increasing strain on the plant as monthly production rises.
One point about the decision we're studying. Producing nothing means Auto Eco stays out of the market this month, so it incurs no cost at all, not even the 80. Its profit is then zero. Staying out is always available, and it's the benchmark every positive output has to beat.
Finding the best output at one price
Take a market price of $60,000, so \(p=60\). Write \(\pi(q)\) for Auto Eco's monthly economic profit: revenue minus cost. At \(q=5\), or 500 vehicles, revenue is \(60\times5=300\), measured in hundreds of thousands of dollars, and cost is \(80+5\times10+5\times5^2=255\), so
That's $4.5 million of profit for the month. Is 500 vehicles the best it can do? Check the neighbors. Outputs of 400 and 600 vehicles each give $4 million, and moving farther in either direction gives less. The peak is \(q=5\), so Auto Eco supplies 500 vehicles at a price of $60,000.
Doing it at every price
That's one price. Repeat the exercise at others and a pattern emerges.
| Market price | Profit-maximizing output | Maximum monthly profit | Reason |
|---|---|---|---|
| $30,000 | 0 vehicles | $0 | Every positive output makes a loss. The best positive output is 200 vehicles, which loses $6 million. |
| $60,000 | 500 vehicles | $4.5 million | Five hundred vehicles gives more profit than any other feasible output. |
| $100,000 | 900 vehicles | $32.5 million | Nine hundred vehicles gives more profit than any other feasible output. |
Each row answers the same question at a different price. Auto Eco's is that answer at every possible price: the full relationship between the market price and the quantity the firm chooses to produce.
Notice what the first row tells us. Supply isn't a promise to produce; at $30,000 the right decision is to produce nothing at all. Supply records what the firm is willing and able to sell at each price, which is the seller's counterpart to the demand curve from the last reading.
Supply is always tied to a period
Everything here is measured per month: the hundreds of vehicles Auto Eco builds, the cost of building them, and the profit it earns. That's a choice we make at the start, and we stick to it throughout. Quantities and costs measured over a year would give different numbers and a different supply curve.
So a supply statement needs three things: who is supplying, what they're supplying, and over what period. The tonnes of refined sugar Southern Mills supplies each month at different sugar prices is one such relationship. The kilograms of tomatoes Riverbend Farm supplies each year at different farm-gate prices is another. So are the hours of work one person supplies each day at different hourly wages. Change the supplier, the unit, or the period, and you've changed the question.
2.2
The firm takes price as given
Auto Eco is a : it can sell as many vehicles as it wants at the market price, but it's too small for its own output to move that price. So if the price is \(p\) thousand dollars per vehicle and it sells \(q\) hundred vehicles, its revenue is \(pq\), measured in hundreds of thousands of dollars. No discounting to shift extra units, no premium for scarcity. Every vehicle fetches the same \(p\).
Economic profit is revenue minus economic cost:
Recall the convention from Week 1: whenever we write \(C(q)\), we mean the economic cost function relevant to the decision and period we're studying. Profit built on that cost function is economic profit, so it already accounts for everything the firm gives up, including opportunities that never appear on an invoice.
Auto Eco chooses \(q\) to make \(\pi(q)\) as large as possible. In Section 2.1 we did this by brute force, computing profit at every output and picking the winner. That works, but it won't scale, and it doesn't tell us why the answer comes out where it does. Instead we split the problem into two questions and take them in order.
Question 1: should the firm produce at all? Some prices are simply too low. If every positive output loses money, as happened at $30,000, the firm is better off staying out. Answering this gives us a threshold price: the lowest price at which some positive output earns a profit of zero or better. Below it, the firm produces nothing, and there's no second question to ask.
Question 2: if it produces, how much? Once the price clears the threshold, the firm compares positive outputs and picks the most profitable one.
The order matters. Question 2 finds the best positive output, but the best positive output can still be a loss-making one. Only Question 1 can rule that out. The next two sections take each question in turn, and Section 2.5 puts them together on Auto Eco's numbers.
2.3
Step 1: Should the firm produce at all?
Before working out how much to produce, the firm asks a simpler question: is there any positive output at all that doesn't lose money? There's a neat way to see the answer. Take the profit formula and factor out \(q\):
The middle step just divides and multiplies cost by \(q\), which changes nothing, but it turns \(C(q)\) into average cost. Now profit reads as output times a margin. The bracket is profit per vehicle in thousands of dollars. Since \(q\) counts hundreds of vehicles, the product is total profit in hundreds of thousands of dollars. Price is what each vehicle brings in; average cost is what each vehicle costs on average. The gap between them is the margin.
So at any output, profit is nonnegative exactly when \(p\geq AC(q)\). If price covers average cost, the firm makes money on each unit and therefore in total. If it doesn't, the margin is negative and every unit adds to the loss.
The firm doesn't need this to hold at every output, only at one. If there's a single quantity where price covers average cost, producing that quantity beats staying out. So the question becomes: how low can average cost go?
That lowest point is the . If price is at least as high as the cheapest average cost the firm can achieve, production is viable:
Three cases follow. Below the threshold, price is under average cost at every positive output, so every production plan loses money and the firm produces nothing. Exactly at the threshold, the firm can just break even by producing the output that minimizes average cost; its economic profit is zero. Above it, at least one positive output turns a profit, and Step 2 finds the best one.
It's worth pausing on what zero economic profit means. It isn't failure. Economic cost already includes everything the owners gave up to run the firm, so zero economic profit means the business is doing exactly as well as its next-best alternative. A firm earning zero economic profit is earning a perfectly normal return.
Let's determine the shutdown threshold
Start by setting the market price to \(p=80\), or $80,000 per vehicle. Drag the hand on the horizontal price line up or down until the control reads 80. Then change the number of EVs produced by dragging the hand on the vertical quantity line left or right.
As you change \(q\), watch the green rectangle. Its area represents Auto Eco's monthly economic profit at the selected price and output. At \(p=80\), find the range of quantities for which the rectangle is green. The two boundary points occur where the price line crosses the average-cost curve. Enter them below the graph.
The shutdown threshold is the lowest price at which some positive output can avoid a loss. To find it, gradually lower the price. After each change, move \(q\) and see whether any quantity produces a green profit rectangle. Stop when the green rectangle disappears for every quantity and the price line just touches the average-cost curve. Below that price, every positive output loses money, so Auto Eco is better off producing nothing.
Now reveal the marginal-cost curve. What do you notice where average cost reaches its minimum?
2.4
Step 2: How much should the firm produce?
We know Auto Eco produces only if the price is at least $50,000. So fix a price above that, say $80,000, and set \(p=80\) on the graph. The price line sits well above the lowest point of the average-cost curve, so production is clearly worth it. The question is how many vehicles.
Rather than test every quantity, pick one and ask whether the firm could do better by moving.
Start high: 1,000 vehicles
Drag output to \(q=10\), or 1,000 vehicles. The marginal-cost curve is at 110, far above the price line at 80.
Read that literally. At this output, marginal cost is about $110,000 per vehicle, while each vehicle sells for $80,000. Moving from \(q=10\) to \(q=9\) means producing 100 fewer vehicles. The firm gives up $8 million in revenue but saves $10.5 million in cost, so profit rises by $2.5 million.
The same logic applies at \(q=9\) and \(q=8\), where marginal cost is still above 80. Wherever the marginal-cost curve sits above the price line, the firm is producing vehicles that cost more than they earn. Cutting back raises profit.
Start too low: 500 vehicles
Now drag output to \(q=5\), or 500 vehicles. Marginal cost is 60, below the price line.
Here the logic runs the other way. Expanding from 500 to 600 vehicles adds $6.5 million to cost and $8 million to revenue, so profit rises by $1.5 million. The firm is leaving money on the table. Wherever the price line sits above the marginal-cost curve, there are profitable vehicles still unbuilt. Expanding raises profit.
Where it stops
Above \(q=7\), cutting back helps. Below \(q=7\), expanding helps. The two forces meet where the price line crosses the marginal-cost curve:
For Auto Eco at \(p=80\), that's \(80=10+10q\), so \(q^*=7\), or 700 vehicles. The arithmetic confirms it:
| Output \(q\) (hundreds of vehicles) | 5 | 6 | 7 | 8 | 10 |
|---|---|---|---|---|---|
| Monthly profit ($ million) | 14.5 | 16 | 16.5 | 16 | 12 |
What the equality does and doesn't say
It's a stopping rule, nothing more. It tells the firm where to stop expanding, not what it earns. Profit comes from average cost. At \(q=7\), \(AC(7)\) is about $56,400 per vehicle, so the margin is about $23,600 per vehicle. Across 700 vehicles, that gives the $16.5 million in the table. Price equals marginal cost at the chosen output, but the firm earns a positive margin on the vehicles it produces.
The comparison is more general than the crossing
The graph shows a textbook case: the price line crosses the rising marginal-cost curve inside the firm's feasible range, \(0\leq q\leq \bar q\), where \(\bar q\) is the firm’s capacity constraint. Below the crossing, \(p>MC(q)\), so another unit adds more to revenue than to cost and expanding raises profit. At the crossing, \(p=MC(q)\). Above it, \(p\lt MC(q)\), so contracting raises profit. The change from \(p>MC\) to \(p\lt MC\) is what makes the crossing a profit maximum.
But price need not intersect marginal cost before the firm reaches capacity. If \(p>MC(q)\) at every feasible output, profit keeps rising all the way to \(\bar q\), so the firm produces at capacity. The best output is then a boundary even though price never equals marginal cost. If \(p\lt MC(q)\) at every positive feasible output, every expansion lowers profit, so no positive output is optimal and the firm produces nothing.
This gives a rule that works in all three cases. When \(p>MC(q)\), expand if possible. When \(p\lt MC(q)\), contract. When \(p=MC(q)\), stop only if marginal cost crosses price from below to above. Apply the comparison over the full feasible range, including the endpoints \(0\) and \(\bar q\).
Step 1 still comes first. Even the best positive output must cover average cost; otherwise the firm chooses zero. And if marginal cost is falling at a crossing, equality doesn't identify a peak. It marks the worst output in its neighborhood, so the firm must compare the feasible boundary outputs.
2.5
Worked example: Auto Eco
We can now derive the choices from Section 2.1 properly, instead of by checking outputs one at a time. Auto Eco's monthly total cost, in hundreds of thousands of dollars, is
where \(q\) is measured in hundreds of vehicles. Dividing by \(q\) gives average cost, and marginal cost follows from the rule for quadratic cost functions in Week 1. Both are in thousands of dollars per vehicle:
Step 1: the threshold price
Step 1 said the firm produces only if price covers average cost at some output, so we need the lowest point of the average-cost curve. Week 1 gave us a shortcut for finding it: average cost is at its minimum where it equals marginal cost. Setting the two expressions equal:
At \(q=4\), or 400 vehicles, average cost is \(80/4+10+20=50\), that is, $50,000 per vehicle. No output is cheaper to produce per vehicle than this. So $50,000 is Auto Eco’s threshold price: below it, nothing it can do covers its costs.
Step 2: the quantity, at any price above the threshold
Step 2 said to set price equal to marginal cost. Doing it once with \(p\) left as a letter solves the problem for every price at once:
Checking two prices
At \(p=70\), Auto Eco produces \((70-10)/10=6\), or 600 vehicles. Average cost there is \(80/6+10+30=\frac{160}{3}\), about $53,333 per vehicle, so the margin is roughly $16,667 on each one:
In hundreds of thousands of dollars, that’s $10 million for the month. Note that price equals marginal cost on the last vehicle built, and the firm still clears $10 million, because every earlier vehicle cost less than the last.
At \(p=40\), the same formula gives \(q=3\), or 300 vehicles. But $40,000 is below the $50,000 threshold, and sure enough, average cost at \(q=3\) is about $51,667 per vehicle, above the price. Producing those 300 vehicles would lose about $3.5 million. Auto Eco supplies nothing.
This is why Step 1 comes first. The equation \(p=MC(q)\) finds the best positive output, and the best positive output can still be a loss. Only Step 1 catches that.
2.6
The firm's supply curve
Putting the two steps together gives Auto Eco’s supply, with \(p\) in thousands of dollars per vehicle and \(q\) in hundreds of vehicles:
Look at what the third line is. It’s \(p=MC(q)\) rearranged. So above the threshold, the supply curve and the marginal-cost curve are the same curve, read in opposite directions: marginal cost answers “what does the next vehicle cost at this output?”, and supply answers “at this price, how many vehicles?” That’s why supply slopes upward here. A higher price brings more vehicles within reach of being worth building, because each can be sold for more.
At exactly \(p=50\), Auto Eco’s optimal-output set is \(\{0,4\}\): either no vehicles or 400 vehicles. Both give zero economic profit, so there’s no fact of the matter about which it does. What happens just above and just below the threshold is what matters.
2.7
A movement along supply versus a shift of supply
When the market price changes and everything else stays fixed, Auto Eco faces the same problem with a different price line. Its chosen quantity moves to another point on the same supply curve. Economists call this a .
On the graph, a rise from $70,000 to $80,000 takes the firm from 600 vehicles to 700. The right way to describe this is “Auto Eco’s quantity supplied increases.” Not “Auto Eco’s supply increases,” which is reserved for something else.
That something else is a shift of supply, and it happens when one of the things we’ve been holding fixed changes instead. Auto Eco’s supply curve was drawn under one condition: its cost function is \(C(q)=80+10q+5q^2\). Change that and we have to redo the whole exercise. Suppose battery prices fall, so every vehicle costs less to build. Marginal cost is lower at every output, so at $70,000 the firm might now choose 700 vehicles instead of 600, at $80,000 it might choose 800 instead of 700, and so on at every price. The result is a new curve sitting to the right of the old one. That’s a shift of supply: not a new point on the curve, but a new curve.
So the two cases differ in what caused the change. A change in the market price moves Auto Eco along its curve. A change in its costs gives it a different curve. The distinction prevents the same mistake we met with demand. The supply curve is already a complete answer to “how much does the firm produce at each price?” So a price change can never explain why the curve moved; the curve hasn’t moved. Anything that rewrites the cost function can: technology, input prices, taxes and regulation are the usual sources, and they matter most in the next reading, when we add firms together into market supply.
The curve stays put; quantity supplied changes.
The whole curve moves; the quantity chosen at every price may change.
2.8
Capacity and discrete output
The rule \(p=MC(q)\) is easiest to apply when output can vary by any amount and the firm has room to expand. Real production is often less accommodating. A plant has a hard limit, and output may only move in blocks. Neither changes the logic, but both change how the supply curve looks.
A capacity limit
Suppose Auto Eco can build at most 800 vehicles a month, so \(q\leq 8\). Up to that point nothing changes: the firm follows \(p=MC(q)\) as before. From \(q=(p-10)/10\), it reaches capacity at a price of $90,000.
Above $90,000, the rule would call for more than 800 vehicles, and the firm simply can’t build them. Its quantity supplied stays at 800 however high the price goes. On the graph, the supply curve rises along marginal cost up to $90,000 and then turns vertical at \(q=8\).
The firm isn’t indifferent to the higher price. At $100,000 it earns far more on its 800 vehicles than at $90,000. What it can’t do is respond by producing more, at least until it can expand the plant, which is a decision over a longer period than this month.
Output in blocks
Now suppose Auto Eco can only adjust production in blocks of 100 vehicles, and take a market price of $60,000. Each block brings in $6 million of revenue, so the question for each block is whether it costs less than that to build.
| Block | Added cost | Added revenue | Effect on profit |
|---|---|---|---|
| 400 → 500 vehicles | $5.5 million | $6 million | +$500,000 |
| 500 → 600 vehicles | $6.5 million | $6 million | −$500,000 |
The firm takes the first block and declines the second, settling at \(q=5\). Notice that price doesn’t equal marginal cost at either end: $6 million sits between the two block costs, never on one. With discrete output, the equality often has nowhere to happen.
The underlying rule survives, though, in the form it had all along: build every block whose added revenue covers its added cost, and stop at the first one that doesn’t. The equality \(p=MC(q)\) was only ever a convenient way of saying that when output is divisible.
When capacity sets the threshold
Capacity can also change Step 1. Consider a firm whose economic cost function has a constant marginal cost and a positive fixed cost, as in the pharmaceutical exercise. Average cost then falls at every output, because the fixed cost spreads over more and more units and nothing pushes back. There’s no interior minimum to find.
So the cheapest output per unit is the largest one the firm can produce, and capacity is what defines it. The threshold price is average cost at full capacity. Below it, the firm produces nothing; at or above it, it goes straight to capacity, since every unit up to the limit adds to profit.
The supply curve is then a step rather than a smooth rise: flat at zero below the threshold, then vertical at capacity. This is a useful case to keep in mind. Industries with heavy up-front costs and cheap production, like software and pharmaceuticals, tend to look like this, and they behave very differently from Auto Eco.
2.9
What profit tells the manager
The supply decision keeps apart three numbers that are easy to blur in practice. Price is what each unit brings in. Marginal cost answers one question only: is the next unit worth making? Average cost answers a different one: are we making money on what we produce?
Keeping them separate matters because they can disagree. A firm can sit exactly at \(p=MC\) and still be losing money, if price is below average cost at that output. We saw it with Auto Eco at $40,000: the marginal rule pointed to 300 vehicles, and 300 vehicles would have lost $3.5 million. Marginal cost found the best positive output, which happened to be the least bad loss. It took average cost to reveal that the whole plan was worse than doing nothing.
This is the practical content of the two steps. Marginal cost is the right tool for “how much,” and the wrong tool for “whether.” A manager who reaches for it first can optimize their way into a loss.
One more thing to watch. Both steps rest on the cost function, and which costs belong in it depends on the decision being made. A plant manager setting this month’s output and an investor deciding whether to build the plant at all are asking different questions, so they work from different cost functions and can reach different conclusions about the same factory. Supply depends on the technology and on the decision the firm is actually facing.
Summary