Essay · The derivative

"Marginal" is just the economist's word for a derivative

one more unit of input, how much more output?

Updated September 2026

In economics, marginal means "per one more unit of input". Marginal cost is the extra cost of one more unit, marginal utility the extra satisfaction from one more unit. That is exactly what a derivative measures, so marginal cost is dTC/dQ and marginal utility is dU/dQ. Same object, two vocabularies.

Economics courses and calculus courses usually introduce this idea twice, in different languages, and never say they're the same. One says "the extra cost of the next unit"; the other says "the limit of the difference quotient". Students end up holding two unrelated skills, and the derivative turns into something you do to a polynomial rather than something that means anything.

What a derivative actually compares

Forget tangent lines for a moment. A function is a machine: push the input a little, and the output moves by some amount. Those two movements — the input nudge dx and the output nudge df — are the only things a derivative cares about. It compares them.

And a comparison is a ratio: call one of them 1 and read off the other. Call the input nudge 1, and the size of the output nudge is the derivative. That's what df/dx says out loud: per one unit of input, this much output.

So why is that the same as "marginal"?

Because "marginal" is that sentence with the calculus removed. One more unit of labour, how much more output? That's the marginal product of labour, and it's dQ/dL. One more unit sold, how much more revenue? Marginal revenue, dTR/dQ. The economics word describes the ratio; the calculus symbol writes it down.

Why profit is maximised where MR = MC

The most quoted rule in microeconomics becomes obvious once the words are translated. Profit is revenue minus cost, so the derivative of profit is dTR/dQ − dTC/dQ, which is MR minus MC. A maximum sits where the derivative is zero. Set MR − MC to zero and you have MR = MC.

Read as a sentence rather than a formula: keep producing while the next unit brings in more than it costs, and stop when it stops. No memorisation required — it's the first-derivative test wearing a business suit.

Why the textbook's version looks slightly different

Introductory courses often define marginal cost as ΔTC / ΔQ with ΔQ = 1 — literally the cost of the next whole widget. That isn't a different concept, it's the same ratio measured over a step of size one instead of an infinitesimal one.

The two agree exactly when the cost curve is straight across that step, and drift apart when it curves. Suppose TC(Q) = Q². The derivative at Q = 10 is 2Q = 20, while the eleventh unit actually costs 121 − 100 = 21. Close, not identical — and the gap is exactly the curvature you skipped over.

The gap stays at 1 here because Q² has constant curvature; for gentler curves it shrinks, and for a straight-line cost curve it vanishes entirely. Which is the honest version of "the derivative is the marginal quantity": it's the marginal quantity with the step shrunk until the curvature stops interfering.

Diminishing marginal utility, in one line

The second glass of water is worth less than the first. In calculus that sentence is: dU/dQ is positive but falling, so the second derivative is negative and the total utility curve is concave. Diminishing returns is not an extra law of economics — it's a statement about the shape of a graph, and the derivative is what reports the shape.