Mathematical Economics · Calculus
Marginal Analysis: Why Every “Marginal” Concept Is a Derivative
Marginal cost, marginal revenue, marginal utility, marginal product. Four names, one piece of mathematics. Once you see it, half of economics becomes the same problem.
Open any economics textbook and count how many times the word marginal appears. Marginal cost. Marginal revenue. Marginal utility. Marginal product of labour. Marginal propensity to consume. Marginal rate of substitution.
Students treat these as six separate things to memorise. They are not. They are one thing, applied six times.
“Marginal” means “derivative.” That is the whole translation. Every time an economist says “the marginal something of something else,” they are describing the rate at which one quantity changes as another changes — which is precisely what a derivative measures.
This post makes that connection concrete, shows you how to move between the discrete tables in your textbook and the calculus in your exam, and explains why the famous condition MR = MC is not a rule to memorise but a consequence of how derivatives behave at a maximum.
The translation table
Here is the entire vocabulary, decoded. In every row, the economics phrase on the left is describing the mathematics on the right.
| Economics says | Mathematics means | In symbols |
|---|---|---|
| Marginal cost | Derivative of total cost with respect to quantity | MC = dTC/dQ |
| Marginal revenue | Derivative of total revenue with respect to quantity | MR = dTR/dQ |
| Marginal utility | Derivative of total utility with respect to consumption | MU = dU/dx |
| Marginal product of labour | Derivative of output with respect to labour | MPL = ∂Q/∂L |
| Marginal propensity to consume | Derivative of consumption with respect to income | MPC = dC/dY |
Key idea
Marginal X of Y = the derivative of total X with respect to Y = the rate at which total X changes when Y changes by a very small amount. The word “marginal” is the economist’s way of saying “at the margin” — meaning for one more unit, in the limit as that unit becomes infinitesimally small.
Notice the pattern in the symbols. Marginal cost uses a straight d because total cost typically depends on one variable, quantity. Marginal product uses a curly ∂ because output depends on several inputs at once — labour and capital — so we take a partial derivative, holding capital constant while labour varies.
That distinction between d and ∂ is not decorative. The curly symbol is a promise that everything else is being held fixed. It is the mathematical expression of the phrase economists say constantly: ceteris paribus.
From a total function to a marginal one
Let us do this concretely. Suppose a firm’s total cost of producing Q units is:
TC(Q) = 500 + 20Q + 0.5Q²
Read the three terms before differentiating anything, because each one is telling you something economic.
The 500 is fixed cost — it does not depend on Q at all. Rent, machinery, the licence fee. Produce nothing and you still pay it.
The 20Q is a constant cost per unit. Every unit adds £20 of materials.
The 0.5Q² is the interesting one. Because it is squared, costs rise faster as output grows. This is diminishing returns showing up in the algebra: pushing more output through a fixed factory gets progressively more expensive.
Now differentiate:
MC = dTC/dQ = 20 + Q
Three things happened, and each has an economic reading.
The 500 vanished. The derivative of a constant is zero. Economically: fixed costs do not affect marginal cost. This is one of the most consequential facts in microeconomics. It is why a firm deciding whether to produce one more unit should ignore its rent entirely, and why sunk costs are irrelevant to forward-looking decisions. The mathematics enforces the economics automatically.
The 20Q became 20. A constant per-unit cost stays constant at the margin.
The 0.5Q² became Q. Marginal cost now rises with output — the upward-sloping MC curve you have drawn a hundred times, derived rather than asserted.
Worth noticing
At Q = 10, MC = 30. This means the tenth unit costs roughly £30 to produce. Not £30 on average — £30 for that specific extra unit. Average cost at Q = 10 is TC(10)/10 = (500 + 200 + 50)/10 = £75. The gap between £30 and £75 exists entirely because the £500 fixed cost is being spread across the units. Marginal and average are different questions with different answers, and confusing them is the most common error in this topic.
Why your textbook table and your calculus disagree slightly
Here is something that quietly confuses a lot of students. Early chapters teach marginal cost as a table:
| Q | Total cost | Marginal cost (table method) |
|---|---|---|
| 9 | 720.5 | — |
| 10 | 750.0 | 29.5 |
| 11 | 780.5 | 30.5 |
The table says the marginal cost of the tenth unit is 750.0 − 720.5 = 29.5. But our calculus said MC at Q = 10 is 30. Which is right?
Both. They are answering marginally different questions.
The distinction
The discrete method computes ΔTC/ΔQ — the actual cost of one whole extra unit. It is an average rate of change across an interval of width 1.
The calculus method computes dTC/dQ — the instantaneous rate of change at exactly Q = 10, the limit as the interval shrinks to nothing. It is the slope of the tangent, not the slope of a chord.
When the cost function is curved, a chord and a tangent have slightly different slopes, so the two numbers differ a little. Notice that the calculus answer of 30 sits neatly between the two table values of 29.5 and 30.5 — it is the slope at the midpoint of the curve’s behaviour there.
Which should you use? Whichever the question gives you. If you are handed a table, subtract. If you are handed a function, differentiate. In an exam, the giveaway is the format of the data, and mixing the methods is a reliable way to lose marks.
Common error
Students sometimes think the discrete method is “the real one” and calculus is an approximation. It is better to see it the other way round: real production happens in whole units, so the discrete number is what actually occurs, but calculus is what makes the mathematics tractable. Economists model quantity as continuous because it lets us use derivatives, and the error is negligible when output is large. Nobody worries about the difference between the 10,000th and 10,001st unit of a commodity.
The marginal and average relationship
There is a relationship between marginal and average values that looks like a curiosity and is actually a general mathematical law. Once you know it, several textbook diagrams stop needing memorisation.
The rule
• If marginal is below average, the average is falling.
• If marginal is above average, the average is rising.
• If marginal equals average, the average is at a turning point — its minimum or maximum.
• If marginal is above average, the average is rising.
• If marginal equals average, the average is at a turning point — its minimum or maximum.
This is why the marginal cost curve always cuts the average cost curve at exactly its lowest point. Not by convention. By necessity.
The intuition is easier with exam marks than with costs. Suppose your average across four papers is 70. You sit a fifth paper — that is your marginal paper — and score 50. Your average must fall, because you added something below it. Score 90 instead and your average rises. Score exactly 70 and your average does not move at all.
Marginal cost cutting average cost at its minimum is the same fact wearing a different costume. As long as each additional unit costs less than the running average, the average is being dragged down. The moment an extra unit costs more than the average, the average starts climbing. The crossing point is therefore exactly where the average stops falling and starts rising — the minimum.
Why MR = MC is not a rule to memorise
Every economics student learns that firms maximise profit where marginal revenue equals marginal cost. Most learn it as an incantation. It is actually a two-line consequence of what a derivative does at a maximum.
Profit is revenue minus cost:
π(Q) = TR(Q) − TC(Q)
To find the maximum of any function, you differentiate and set the result to zero — because at a peak, the slope is flat. So:
dπ/dQ = dTR/dQ − dTC/dQ = 0
But dTR/dQ is just marginal revenue, and dTC/dQ is just marginal cost. Substituting the economics vocabulary back in:
MR − MC = 0, therefore MR = MC
That is the entire derivation. MR = MC is not an economic law discovered by observation. It is what “set the derivative of profit to zero” looks like once you translate the symbols into economics vocabulary.
The economic intuition, which matches exactly
If MR is above MC, the next unit brings in more than it costs, so making it raises profit — you should expand. If MR is below MC, the last unit cost more than it earned, so you should contract. Only when they are equal is there nothing left to gain by moving in either direction. That verbal argument and the calculus are the same argument. The mathematics is not an alternative to the intuition; it is the intuition written precisely.
One important caveat, which we develop fully in a companion post: setting the derivative to zero finds a flat point, and flat points can be maxima or minima. To be certain you have found a profit maximum rather than a profit minimum, you need to check the second derivative. That is the second-order condition, and skipping it is a genuine and commonly penalised omission.
A complete worked example
A monopolist faces the inverse demand curve P = 100 − 2Q and has total cost TC = 20 + 4Q. Find the profit-maximising output, price and profit.
Step 1 — Build total revenue. Revenue is price times quantity, and price depends on quantity:
TR = P × Q = (100 − 2Q)Q = 100Q − 2Q²
Step 2 — Differentiate for marginal revenue.
MR = dTR/dQ = 100 − 4Q
Pause on that result, because it contains a famous fact. Demand has slope −2; marginal revenue has slope −4. For a linear demand curve, MR has the same intercept but exactly twice the slope. That is why MR is always drawn below demand, and it falls straight out of the differentiation rather than needing to be remembered.
Step 3 — Differentiate for marginal cost.
MC = dTC/dQ = 4
Step 4 — Set MR = MC and solve.
100 − 4Q = 4, so 96 = 4Q, giving Q* = 24
Step 5 — Find the price from the demand curve (not from MR, a very common slip):
P = 100 − 2(24) = £52
Step 6 — Compute profit. TR = 52 × 24 = 1,248. TC = 20 + 4(24) = 116. So profit = 1,248 − 116 = £1,132.
Common error
Substituting Q* back into the marginal revenue equation to find price. MR at Q = 24 is 100 − 96 = 4, and a student who reports “price = £4” has made a serious error. Price always comes from the demand curve, because demand is what tells you what consumers will pay. Marginal revenue is an internal calculation for the firm, not something anyone is charged.
Practice questions
Q1. A firm has TC = 300 + 12Q + 2Q². Find MC, and compute the marginal cost of the fifth unit using calculus.
Q2. Total utility from consuming x units of a good is U = 40x − x². Find marginal utility, and the level of consumption at which total utility is maximised. What is MU at that point, and why does that make sense?
Q3. A firm in perfect competition sells at a fixed market price of £25. Its total cost is TC = 100 + 5Q + 0.5Q². Find the profit-maximising output. Why is MR equal to £25 here?
Q4. Explain, without using calculus, why the marginal cost curve must cross the average cost curve at the minimum of average cost.
Q5. A production function is Q = 12L − 0.5L² where L is labour. Find the marginal product of labour, the level of L at which output is maximised, and the level of L beyond which MPL becomes negative.
Worked answers
A1. Differentiate term by term. The constant 300 vanishes, 12Q becomes 12, and 2Q² becomes 4Q. So MC = 12 + 4Q. At Q = 5, MC = 12 + 20 = £32. Note that the fixed cost of 300 played no part — as it never does at the margin.
A2. MU = dU/dx = 40 − 2x. Total utility is maximised where its derivative is zero: 40 − 2x = 0, giving x = 20. At that point MU = 40 − 40 = 0.
This makes perfect sense, and it is worth pausing on. Marginal utility is zero at the peak of total utility because you are completely satiated — one more unit adds nothing. Consume beyond x = 20 and MU turns negative, meaning additional units actively make you worse off. Think of a fourth slice of cake. Total utility rising, then peaking, then falling is exactly the story that diminishing marginal utility tells.
A3. In perfect competition the firm is a price taker, so it sells every unit at £25 regardless of how much it produces. Total revenue is therefore TR = 25Q, and MR = dTR/dQ = 25 — a constant. This is the mathematical statement of “the firm faces a horizontal demand curve.”
Now set MR = MC. Here MC = dTC/dQ = 5 + Q, so 25 = 5 + Q, giving Q* = 20. Profit is TR − TC = 500 − (100 + 100 + 200) = £100.
A4. Use the marks analogy. The average cost is the running average of all units produced so far. Marginal cost is the cost of the newest unit. Whenever the newest unit costs less than the current average, it drags the average down. Whenever it costs more, it pulls the average up. So average cost falls while MC is below it and rises once MC is above it. The switch from falling to rising is the minimum, and it occurs precisely at the crossing point where MC equals AC. No calculus required — it is a property of how averages work.
A5. MPL = dQ/dL = 12 − L. Output is maximised where MPL = 0, so L = 12. Beyond L = 12, MPL is negative — each additional worker reduces total output. That sounds absurd until you picture a small kitchen with forty chefs in it: they get in each other’s way. This is the region of negative marginal returns, and a rational firm never operates there, because it could produce more by hiring fewer people.
The short version
• Marginal means derivative. Marginal X of Y = d(total X)/d(Y).
• Straight d for one variable; curly ∂ when holding others constant.
• Fixed costs always disappear when you differentiate — which is why sunk costs are irrelevant at the margin.
• Table method gives ΔTC/ΔQ; calculus gives dTC/dQ. Small differences are expected on a curve.
• Marginal below average means average falls. Marginal above means average rises. Equal means turning point.
• MR = MC is just the profit derivative set to zero, translated into economics vocabulary.
• For linear demand, MR has the same intercept and twice the slope.
• Always read price off the demand curve, never off MR.
• Straight d for one variable; curly ∂ when holding others constant.
• Fixed costs always disappear when you differentiate — which is why sunk costs are irrelevant at the margin.
• Table method gives ΔTC/ΔQ; calculus gives dTC/dQ. Small differences are expected on a curve.
• Marginal below average means average falls. Marginal above means average rises. Equal means turning point.
• MR = MC is just the profit derivative set to zero, translated into economics vocabulary.
• For linear demand, MR has the same intercept and twice the slope.
• Always read price off the demand curve, never off MR.
Once marginal analysis is secure, the natural next step is checking whether the flat point you found is genuinely a maximum — the first and second order conditions — and then extending the same logic to problems with a budget constraint using Lagrange multipliers.
References
1. Chiang, A.C. & Wainwright, K. (2005) Fundamental Methods of Mathematical Economics. 4th edn. New York: McGraw-Hill.
2. Varian, H.R. (2014) Intermediate Microeconomics: A Modern Approach. 9th edn. New York: W.W. Norton.
3. Sydsaeter, K., Hammond, P. & Strom, A. (2016) Essential Mathematics for Economic Analysis. 5th edn. Harlow: Pearson.
4. Nicholson, W. & Snyder, C. (2016) Microeconomic Theory: Basic Principles and Extensions. 12th edn. Boston: Cengage.
5. Marshall, A. (1890) Principles of Economics. London: Macmillan.
Take this further
Marginal analysis runs through nearly every chapter of intermediate economics. Our full treatment — with the diagrams, the derivations and worked exam questions — is in Economics Made Simple: A Supplementary Guide for AP and Cambridge A-Level Students, a 914-page guide covering thirty chapters of the AP and A-Level syllabus.
If you want the theory alongside real applications, Economics Case Studies: 100 Real-World Examples shows these tools being used on actual firms and markets. Both are available together in the Economics Made Simple bundle.