Marginal Cost: The Cost of One More Unit

Marginal cost is what one more unit costs to make, counting only what changes. Marginal cost decides output through one rule: keep making more while the price covers the cost of the next unit, and stop when it does not. What has already been spent never enters that decision, however large it is. No choice available now can bring it back.

Underneath that sits a fact about decisions in general, not about farming or about cost sheets. A decision can only be made on the amounts that the decision itself moves. Every other amount appears identically on both sides of the comparison, so it cancels, and an amount that cancels cannot tip a choice one way or the other. Marginal cost is what remains once the cancelling is done. Marginal cost is therefore a smaller number than most people expect, and the number people reach for instead, the one that includes everything the operation costs, is nearly always the wrong one to hold up against a price.

The grower worked through below farms onions in the Republic of Sankhya, an invented country, and faces a market price of Rs 2,000/- a quintalA unit of weight used in wholesale trade for farm produce, equal to one hundred kilograms. Onions are bought and sold by the quintal rather than by the kilogram at the market yard. that no single grower can shift. Four things come out of the work below: how to read the grower's cost ladder, how to apply the stopping rule to it, how to hold that stopping point steady when the season's standing cost doubles, and what the standing cost genuinely does settle instead.

What is marginal cost, and what does it leave out?

Dinner is already being cooked for four when a fifth person calls to say they are coming. What does that fifth plate cost? Not one fifth of the gas cylinder, not a share of the kitchen rent, not part of what the pressure cooker cost three years ago. The fifth plate costs a handful more rice, a little more dal, and the two extra minutes standing at the stove. Everything else was going to be spent tonight anyway, and the fifth guest changed none of it.

Marginal cost is defined by what it excludes, not by what it contains, so the honest way to compute it is to start with every rupee the operation costs and take out everything that does not move when one more unit is made. Learn marginal cost as a formula to be filled in and the ordinary cases come out right, then the first unusual case comes out badly wrong. The formula does not say which rupees go into it. The exclusion test does. Ask of each cost: if the next lot is not grown, does this amount change? If the answer is no, that cost is outside, whatever it is called in the accounts and however large it is.

Two categories fall outside almost always. The first is a fixed costAn amount that stays the same across the output levels being considered. The rupees owed do not move with output., which is there at every level of output over the range being considered: the season lease on the field, the annual maintenance on the borewell, the licence for the market yard. The second is anything already committed and not recoverable. No present choice can change it. Inside falls a variable costAn amount that goes up and down with how much gets produced. Produce nothing at all and it disappears entirely. and only the part of it that the extra output actually causes: the extra seed, the extra hours of picking, the extra fuel to carry the extra bags to the market.

What does the next lot actually cost this grower?

Now put numbers on it. The invented Sankhya grower plants in lots of 200 quintals, and each lot costs more per quintal than the lot before. The ladder runs like this: the lot that takes output to 200 quintals costs Rs 1,400/- a quintal, the lot that takes it to 400 costs Rs 1,700/-, the lot that takes it to 600 costs Rs 2,000/-, the lot that takes it to 800 costs Rs 2,400/-, and the lot that takes it to 1,000 costs Rs 2,900/-. Read each rung as a rate for that lot and nothing else. The Rs 1,400/- does not describe the whole crop; it describes the first 200 quintals only.

Turning a ladder of rates into a running total is one multiplication and one addition per rung, and doing it in that order is what keeps marginal cost and total cost from being confused with each other. Two hundred quintals at Rs 1,400/- is Rs 2,80,000. Add the second lot, 200 at Rs 1,700/- for Rs 3,40,000, and the growing cost so far is Rs 6,20,000. Add the third, 200 at Rs 2,000/- for Rs 4,00,000, and the growing cost reaches Rs 10,20,000 at 600 quintals. The fourth lot adds Rs 4,80,000 and the fifth Rs 5,80,000, so growing the full 1,000 quintals costs Rs 20,80,000 before the season's standing cost of Rs 8,00,000 is counted at all.

Why does marginal cost climb as output climbs?

Because the good stuff gets used first. Nobody plants their worst land before their best. The grower puts the first lot on the level plot near the borewell, where the water reaches easily and a bullock cart can turn. The second lot goes on the sloped patch that needs more water because more of it runs off. The third goes on the far corner where every bag has to be carried further before it reaches a road. Nothing about the grower got worse. The land did.

Labour behaves the same way. The first hours of the day are the productive ones, done in daylight by people who are fresh. Pushing the harvest wider means hiring at the peak of the season when everybody else is hiring, paying for a second trip because the tempo is full, and paying overtime for hours that produce less than the morning hours did. Seen from the cost side, that pattern is diminishing returnsThe pattern where each extra dose of an input, added to something that is fixed in supply, produces less additional output than the dose before it.: the same rupee buys less output than it used to, and the same output now costs more rupees.

A producer will only add output when the price rises high enough to cover the cost of the lot being added, so a rising marginal cost and an upward sloping supply curve are one fact seen from two sides. That is why a supply curve slopes the way it does. The slope is not a habit of drawing and not a convention. Ask what the market has to pay to bring out one more lot, and the answer climbs for exactly the reasons above, and the curve that plots those answers climbs with it.

How does the ladder decide how much to grow?

The grower is a price takerA seller too small to move the going price on its own, so it treats the price as given and decides only how much to sell at it. and cannot argue with the market yard. Onions in Sankhya clear at Rs 2,000/- a quintal, the point where the market's demand and supply meet, and this grower is one of thousands. Growing more does not lower the price, and growing less does not raise it. So the price is a flat line at Rs 2,000/- and the only choice left is how much to bring to it.

Lay the ladder against that line and the answer falls out. The first lot costs Rs 1,400/- a quintal and every quintal of it sells for Rs 2,000/-, so it leaves Rs 600/- a quintal behind and the whole lot adds Rs 1,20,000. The second lot costs Rs 1,700/- and adds Rs 60,000. The third lot costs Rs 2,000/- exactly, so it sells for precisely what it cost and adds nothing at all. The fourth lot costs Rs 2,400/- against the same Rs 2,000/-, so growing it takes Rs 80,000 off the season, and the fifth takes off Rs 1,80,000 more.

The grower stops at 600 quintals, where marginal cost meets the price exactly, and that answer came out of the ladder and the price alone with nothing else consulted. Notice what was never opened to get there. Not the lease agreement, not the loan statement, not last year's accounts. Two columns of numbers, one comparison per rung, and the decision is made. The third lot is the boundary case and it is worth sitting with: at Rs 2,000/- against Rs 2,000/- the grower is exactly indifferent, so 600 quintals and 400 quintals leave the season in the same place, and 600 is named as the stopping point because it is the last rung the price still covers.

Why does the Rs 8,00,000 already spent not enter the decision?

Here is the part readers resist, and it is worth being blunt about why. The grower has committed Rs 8,00,000 to the season: the lease, the borewell, the loan interest, the licence. Rs 8,00,000 is a large amount of real money and it hurts. The same Rs 8,00,000 appears, unchanged, next to every single output on the ladder. Grow 200 quintals and it is Rs 8,00,000. Grow 1,000 and it is Rs 8,00,000. An amount that is identical in every option cannot make one option better than another, and a number that cannot separate the options has no work to do in choosing between them.

Test it rather than take it on trust. Double the standing cost to Rs 16,00,000 and run the whole rule again. Lot one still costs Rs 1,400/- against a price of Rs 2,000/- and still pays for itself. Lot two still adds Rs 60,000. Lot four still costs Rs 2,400/- and still takes Rs 80,000 off the season. The doubling entered none of the comparisons, so nothing in any of them changed. The season's total is worse by exactly Rs 8,00,000 at every single output level. The shape of the picture is identical and only its height moved.

A cost the decision cannot change is not a reason for the decision or against it, and few ideas in economics are resisted harder. The name for money already committed and not recoverable is a sunk costAn amount already committed that no present choice can recover. Because it is the same whichever option is picked, it cannot make one option better than another., and the resistance to leaving it out is not stupidity. The resistance is the very reasonable feeling that a big cheque ought to count for something. The cheque did count, at the moment it was written, and it cannot reach forward and change which lot is worth growing today.

Take the four standing costs one at a time. With no standing cost at all, the best output is 600 quintals and the season total is Rs 1,80,000. At Rs 4,00,000 the best output is 600 quintals and the total is minus Rs 2,20,000. At the case standing cost of Rs 8,00,000 the best output is 600 quintals and the total is minus Rs 6,20,000. At Rs 16,00,000 the best output is 600 quintals and the total is minus Rs 14,20,000. Four standing costs, four very different totals, one unchanged answer to how much to grow.

If the standing cost does not decide how much to grow, what does it decide?

Somebody who has just learned the sunk cost point tends to overcorrect, and the overcorrection costs as much as the original mistake. Having been told the Rs 8,00,000 does not belong in the output decision, it is very easy to conclude that it does not matter. The Rs 8,00,000 matters enormously. The standing cost simply answers a different question, and the two questions have different inputs.

Work the season out in full at the chosen output. Six hundred quintals at Rs 2,000/- a quintal brings in Rs 12,00,000. Growing those 600 quintals cost Rs 10,20,000, the sum of the three lots the grower actually planted. The difference, Rs 1,80,000, is what the season leaves behind for everything else, and that contributionWhat is left of the money coming in once the costs caused by producing are taken off, available to go towards costs that were there anyway. is the number the output rule was quietly maximising all along. Now bring in the Rs 8,00,000 that the output rule ignored. Against a contribution of Rs 1,80,000, the season ends Rs 6,20,000 short.

Marginal cost answers how much to grow, and the whole picture answers whether the season was worth committing to at all, and treating those as one question is the other half of the mistake. Both answers are correct together and they do not fight. Given that the Rs 8,00,000 is already committed, growing 600 quintals is still much better than growing nothing: growing nothing leaves the grower Rs 8,00,000 down, and growing 600 leaves the grower Rs 6,20,000 down, so the season as run recovered Rs 1,80,000 of a hole it could not otherwise have touched. And at the same time, a standing cost of Rs 8,00,000 against a best possible contribution of Rs 1,80,000 says plainly that committing to this field on these terms was not worth doing. The verdict is about next season, when the lease is still a choice.

How is marginal cost different from average cost?

Marginal cost and average cost get confused precisely when they are introduced together, so define each one on its own before setting them against each other. Marginal cost is the cost of the next unit and counts only what that unit causes. Average costEvery rupee an operation spent, divided by the number of units it produced. Average cost is a per unit figure worked backwards from a total. is every rupee the operation spent divided by every unit it produced, so it includes the standing cost, the growing cost, and anything else in the ledger. One is forward looking and answers what the next step costs. The other is backward looking and reports what the whole thing cost per unit once it is done.

Watch them on the same grower. At 600 quintals the total spend is Rs 10,20,000 of growing cost plus Rs 8,00,000 of standing cost, or Rs 18,20,000, and dividing by 600 quintals gives an average of about Rs 3,033/- a quintal to the nearest rupee. The marginal cost of the third lot was Rs 2,000/-. The two figures describe the same grower in the same season and differ by more than a thousand rupees a quintal. The difference is built into the definitions: the average carries a share of the Rs 8,00,000 and the marginal carries none of it.

Average cost is a report and marginal cost is a decision, so an output choice made on average cost is being made on a number that was never built to answer that question. The consequence of the grower using the wrong one is easy to trace. Average cost at 600 quintals is Rs 3,033/- against a price of Rs 2,000/-, a comparison that reads as a disaster and suggests growing nothing at all. Growing nothing would have cost the grower the Rs 1,80,000 the season did recover. Average cost is genuinely good at one thing: it reports where the season ended up. In this ladder the average keeps falling until output is well past the stopping point, reaching Rs 2,875/- a quintal at 800 quintals and then turning back up to Rs 2,880/- at 1,000, and the turn happens where marginal cost has climbed past the average and started pulling it up.

Where does this show up outside a field?

A household runs a small tiffin service from its own kitchen, thirty boxes a day to an office nearby. One more office worker calls and asks to be added. What does that thirty first box cost? The vegetables, the extra atta and dal, the container, and perhaps twenty minutes more at the stove. Those amounts together are the number to hold against what the box sells for. The rent on the flat, the fridge bought last year and the monthly gas connection charge do not appear. Every one of them is the same amount whether the box is made or not.

The same test runs for a lender sizing a working capital limit, an analyst asking whether a factory should take a discounted order, and a household deciding on one more tiffin box, and in all three the question is what changes, never what the whole thing costs. The extra borrowing is serviced by the extra production, so a lender wants to know whether that production covers the extra cost it causes. The plant is being paid for anyway, so an analyst looking at a discounted bulk order wants the same comparison and nothing else. And a household wants to know whether one more box leaves anything behind at the end of the month. Where each of them goes wrong is identical: pull the average cost off last month's ledger, hold it against the price, and refuse an order that would have left money behind.

The Republic of Sankhya, its onion market and the grower worked through above are invented.
Educational material. Not advice on any investment, tax, budget or market position.