Average Cost vs. Marginal Cost: The Two Numbers That Drive Every Output Decision

A bakery owner gets an offer she will probably regret turning down: 200 extra loaves for a school event, priced at $2.50 each. Her accountant runs the numbers and reports that each loaf costs the bakery $3.10 to make, so she declines. Selling below cost is madness. Except the accountant is using the wrong number, and that refusal likely burned real money.

The $3.10 is an average. It blends rent already paid, equipment already depreciated, a manager already on salary, and the actual flour and gas for those loaves. The question the order actually poses is different: what does it cost to bake 200 more loaves on an oven that is already hot, with staff already on the clock? That is a marginal question, and the answer is almost certainly well below $3.10. Get these two numbers straight and most output decisions get simpler. Confuse them and you will keep making the bakery's mistake.

What each one actually measures

Average total cost is total cost divided by units produced. If it costs $12,000 to make 1,000 chairs, average cost is $12 per chair. It is a backward-looking summary: a single figure that spreads every fixed and variable dollar evenly across output, including costs that were sunk before production ever started.

Marginal cost is the change in total cost from producing one more unit.¹ As the standard microeconomics treatment defines it, marginal cost equals the change in total cost divided by the change in quantity. The critical thing is what it leaves out: fixed costs. Rent and machinery do not change when you make one extra chair, so they contribute nothing to marginal cost. It is built purely from the variable inputs the next unit actually consumes. That is exactly why the marginal figure, not the average, is the relevant one for an incremental decision.² Only marginal cost reflects what the next unit truly adds to the bill.

What they share

Both numbers are extracted from the same total-cost data: two different ways of slicing one cost curve. Both are expressed per unit. And in most real production processes, both eventually rise as output climbs, because firms run into capacity limits: machines get crowded, workers trip over each other, overtime kicks in. The divergence is in what each number is for, and in a strict mathematical relationship between them that trips people up constantly.

The iron rule that links them

Here is the relationship worth locking in: when marginal cost sits below average cost, it drags the average down. When marginal cost rises above average cost, it pushes the average up. It works exactly like a test average. If your running grade is 85 and you score a 70 on the next exam, your average falls. Score a 95 and it rises. The next score, the marginal one, pulls the average toward itself.

This produces a sharp consequence: the average cost curve hits its lowest point precisely where the marginal cost curve crosses it from below. To the left of that crossing, marginal is under average and the average is still falling. To the right, marginal is above average and the average is climbing. The two curves can only intersect at the bottom of the U. That is not a coincidence of graph paper; it is forced by the arithmetic of averages.

The practical payoff is that you can read a firm's efficiency off a single comparison. If the cost of your next unit is below your current average, you are still getting cheaper per unit; keep going. The moment the next unit costs more than the running average, you have passed your most efficient output level, and every additional unit makes the whole batch a little more expensive on average. No need to draw a curve: just compare the marginal number to the average number and note which is bigger.

Run the numbers

Take a small firm assembling drones. Fixed cost is $1,000. Watch both columns as output grows:

Through unit 5, marginal cost sits at or below average cost, and the average keeps falling, bottoming at $430. At unit 6, marginal cost ($450) finally exceeds average cost ($433), and the average ticks up for the first time. The crossover lands right at the trough, exactly as the iron rule predicts.

Now watch what the average column would get wrong. Suppose drones sell for $440 each. Average cost at 7 units is $457, so a manager running off the average sees a loser and stops early. But should the firm make the 6th drone? Its marginal cost is $450, above the $440 price, so no: that unit loses $10. And the 5th? Marginal cost $300, price $440, a $140 gain. The marginal column tells you exactly where to stop: produce up to the point where the next unit's cost exceeds its price. The average column, when used for that job, points to the wrong quantity.

Back to the bakery, and where else this bites

Return to the 200-loaf order at $2.50 against a $3.10 average. That $3.10 carries a slice of rent, equipment depreciation, and a manager's salary: costs already being paid regardless of this order. The marginal cost of 200 more loaves on an already-running line might be flour, yeast, a little extra labor and gas, perhaps $1.40 per loaf. At $2.50 each, every loaf throws off roughly $1.10 toward profit. Taking the order adds about $220 the bakery would not otherwise see. Refusing it on average-cost grounds simply burns that $220.

The same confusion scales up to industries where it costs real billions. The Bureau of Labor Statistics tracks unit labor costs across the economy,³ a close cousin of average cost, and analysts who read a rising average as a blanket signal to cut output can miss that the marginal economics of an extra shift still pencil out. Utilities, airlines, and software firms run on this distinction every quarter: enormous average costs, near-zero marginal costs, and pricing decisions that only make sense once you stop dividing total cost by total units and start asking what the next unit actually costs.⁴

The Federal Reserve's own economic research frames this as the gap between average-cost pricing, which is common in regulated industries, and marginal-cost pricing, which is what competitive theory says firms ought to do.⁵ In practice most firms blend both: average cost for long-run viability, marginal cost for the incremental call. Knowing which question each answers is what keeps you from making the bakery's mistake at any scale.

The rule to carry out

Divide for diagnosis; decide at the margin. Average cost tells you how the business is doing at its current scale. Marginal cost tells you what to do next. They are built from the same data and they answer different questions, and the day you start routing each question to the right number is the day a lot of output decisions get cleaner.

So: what is the next unit going to cost you?