In the autumn of 1843 a young landowner named John Bennet Lawes had a field on his family’s estate at Rothamsted, in Hertfordshire north of London, sown with wheat. He had just patented a new fertilizer and built a factory to make it, and he wanted to know what each thing a farmer put on the soil really did. So the field, called Broadbalk, was cut into long strips, each given its own treatment year after year, with a chemist, Joseph Gilbert, working beside him. Wheat has grown there ever since. It is one of Rothamsted’s classical experiments, the oldest farm experiments of their kind still running anywhere.
Some of the strips were given nitrogen, a plant food, as a chemical fertilizer in steps: none, then about 48 kilograms a hectare, then twice that, three times, four times. Averaged over the thirteen harvests from 1852 to 1864, the first step raised the crop by more than half. The second step added a little less. The third added hardly anything, and the fourth almost nothing at all.
The field didn’t change, and neither did the seed or the weather from strip to strip. Only the nitrogen did, and each extra load did less than the one before. A. D. Hall, Rothamsted’s director, wrote in 1905 that these results showed “very clearly” what was known as the law of diminishing returns, and he added the part that matters to anyone running a business: sooner or later the extra crop stops paying for the extra fertilizer, and that point comes sooner when the crop sells cheap.
The same law runs Anshu’s tea shop, and the rest of this chapter follows it from the field into her room.
One room, more hands
Anshu now has a helper, Sujan. Her shop is one small room with one stove, a counter and a few benches. She can’t make the room bigger this month, or next: the lease is signed and the street is full. What she can change quickly is how many people work in it.
That split is the first idea economists use about costs. Things that can’t be changed for a while are called fixed, स्थिर: the room, the stove, a factory’s machines, a farm’s land. Things that can be changed as needed are variable, परिवर्तनशील: workers by the day, milk, sugar, gas. The short run is the time in which some things are fixed, and the long run is the time it takes until nothing is, when she could move to a bigger room or open a second one. For a tea shop the long run may be a year; for a cement factory it may be ten.
Suppose Anshu works out what her room could turn out in a busy day with different numbers of people in it. The numbers below are made up, but any shopkeeper would recognize them. A rule like this, what comes out for each amount that goes in, is what economists call a production function, उत्पादन फलन. In the short run it has one thing that can change, here the people, with the room and the stove held fixed.
| People working | Glasses a day | Extra glasses from the last person | Glasses for each person |
|---|---|---|---|
| 1 | 90 | 90 | 90 |
| 2 | 200 | 110 | 100 |
| 3 | 280 | 80 | 93 |
| 4 | 315 | 35 | 79 |
| 5 | 335 | 20 | 67 |
| 6 | 340 | 5 | 57 |
| 7 | 335 | −5 | 48 |
The second column is the total product, कुल उत्पादन, everything the room makes in a day. The third is the marginal product, सीमान्त उत्पादन: what one more person adds. The fourth is the average product, औसत उत्पादन, the total shared among the people who made it.
Alone, Anshu does everything: she boils, pours, takes the money and washes the glasses. With Sujan beside her, each of them sticks to part of the work, and together they make more than twice what she managed alone. That is the division of labor at work, and it is why the second person adds more than the first. A third person still helps: someone to wash glasses while the other two pour and serve. But after that the stove is the problem. A fourth person spends much of the day waiting for a burner. A fifth gets in the way. With seven, people are bumping into each other in the doorway, and the room makes fewer glasses than it did with six.
That is the Broadbalk field again, in a tea shop. When one thing is held fixed and more and more of another is added, with the same way of working throughout, the extra output from each addition first rises, then falls, and in the end can turn negative. Nepal’s books call it the law of variable proportions, परिवर्तनशील अनुपातको नियम, because adding one factor to a fixed one keeps changing the proportion between them. Most books call the falling part diminishing returns. It has three stages. In the first, the average is still rising, because each new person adds more than the average so far. The first stage ends where the extra person adds exactly the average, which is also where the average is at its highest. In the second stage the extra output is falling but still positive. The total is at its highest where the extra person adds nothing, here at six. In the third stage the extra output is below zero, and nobody sensible goes there: Anshu would be paying a seventh person to make the shop worse.
There is a neat rule hidden in the table. Whenever the extra person adds more than the average, the average goes up; whenever she adds less, it goes down. So the marginal line always crosses the average line at the average’s highest point.
The marginal is also the steepness of the line. Draw glasses a day against people working and the curve climbs steeply at first, then flattens, then tips over. How steep it is at any point is the marginal product there: steep means one more person adds a lot, flat means she adds nothing. Grade 11 economics, which uses calculus, calls that steepness the derivative. The idea is just this: how fast the total changes when one thing changes a little.
What a glass really costs
Now put rupees on the table.
Anshu pays Rs. 15,000 a month for the room, about Rs. 500 a day, and sets aside about Rs. 100 a day for the stove, pans and glasses that wear out and the interest on what she borrowed to buy them. That Rs. 600 a day is her fixed cost. It is the same whether she sells ten glasses or three hundred, and she pays it even on a day she doesn’t open.
Everything else is variable cost, and it rises with the glasses. Milk, tea leaf, sugar and gas come to about Rs. 10 a glass. Sujan is paid Rs. 21,000 a month, a made-up wage like every number of this shop, but a little above the legal minimum wage, which since July 2025 has been Rs. 19,550 a month. Spread over the thirty days the shop is open, that is Rs. 700 a day. Every extra person she takes on would cost the same.
Then there is a cost that never reaches her account book. Anshu works in the shop all day without paying herself a wage, though she could be earning one, at least as much as Sujan, in someone else’s shop. Giving that up is a cost, the opportunity cost of her own time. Economists split costs this way. An explicit cost, व्यक्त लागत, is money actually paid: rent, wages, milk. An implicit cost, अव्यक्त लागत, is the value of something the owner already has and uses in the business instead of earning from it elsewhere: her own hours, her own savings, a room she owns and could have let. Accounting profit, what a tax office sees, takes away only the explicit costs. Economic profit takes away both. In what follows, Anshu’s own day is counted at Rs. 700, as if she were her own first worker.
From the table of glasses and these prices, every cost follows.
| Glasses a day | Total cost a day | Cost for each glass | Cost of each extra glass |
|---|---|---|---|
| 90 | Rs. 2,200 | Rs. 24.4 | Rs. 17.8 |
| 200 | Rs. 4,000 | Rs. 20.0 | Rs. 16.4 |
| 280 | Rs. 5,500 | Rs. 19.6 | Rs. 18.8 |
| 315 | Rs. 6,550 | Rs. 20.8 | Rs. 30.0 |
| 335 | Rs. 7,450 | Rs. 22.2 | Rs. 45.0 |
Total cost, कुल लागत, is the fixed cost plus the variable cost. Divide it by the glasses and you get the average cost, औसत लागत, what each glass costs her on average. The last column is the marginal cost, सीमान्त लागत: what the last few glasses added to the bill, for each glass. Going from 200 glasses to 280 means one more person (Rs. 700) and 80 more glasses of milk and sugar (Rs. 800), so Rs. 1,500 for 80 glasses, about Rs. 18.8 each. (The first row’s figure is for the 90 glasses she makes alone, from nothing.)
Two things shape the average cost. The first is the fixed cost spread more thinly. At 90 glasses each one carries Rs. 6.7 of rent and wear; at 280, only about Rs. 2. That share, the average fixed cost, keeps falling as long as output grows, though more and more slowly, and never reaches zero. The second is the room. Once the stove is crowded, each new person adds fewer glasses for the same wage, so each extra glass costs more to make. The marginal cost is the law of diminishing returns, turned from glasses into rupees: when the extra output from a person falls, the cost of the extra output rises.
Put the two together and the average cost falls, bottoms out, and rises again, a U. The marginal cost crosses it right at the bottom, for the same reason as before. While the extra glass costs less than the average, it pulls the average down; once it costs more, it pulls it up. In the table, Anshu’s cheapest glass, Rs. 19.6, comes at 280 a day. The average of the variable cost alone, without the rent, makes a smaller U of its own, lowest at about 200 glasses, where it is Rs. 17.
Takings work the same way. Economists call them revenue, आय. Total revenue is the price times the glasses sold. Average revenue is total revenue divided by the glasses, which is just the price. Marginal revenue is what one more glass sold adds to total revenue.
How marginal revenue behaves depends on the seller. Suppose for now that the bazaar is full of tea shops charging the same Rs. 25, so Anshu can sell as many glasses as she makes at that price, and none at all if she charges more. Then every extra glass brings in exactly Rs. 25, and marginal revenue, average revenue and the price are all the same number. On a graph they are one flat line. That is how revenue looks for a seller among very many, the case Nepal’s books call perfect competition.
But a seller whose customers come for her in particular has to lower her price to sell more, and the lower price goes to everyone, not only the extra customer. When Anshu once tried charging Rs. 30, she sold 180 glasses; at Rs. 25 she sells 200. Those 20 extra glasses don’t bring in 20 × Rs. 25. Her takings go from Rs. 5,400 to Rs. 5,000, so the extra glasses bring in less than nothing. For such a seller, average revenue falls as she sells more, and marginal revenue falls faster and sits below it. A monopoly, the only seller of something, the case Nepal’s books set beside perfect competition, faces revenue of this shape. The next chapter follows the sellers who face each kind of market. Here, to see costs plainly, the price stays at Rs. 25.
How many glasses?
Now the question every firm faces. Should Anshu take on a third person?
The third person would add 80 glasses a day. Each sells for Rs. 25 and needs Rs. 10 of milk, sugar and gas, so together they bring in Rs. 1,200 more than their ingredients. The person costs Rs. 700. Yes, then. A fourth would add 35 glasses, worth Rs. 525 over their ingredients, for the same Rs. 700. No.
Put another way, she should go on making glasses as long as one more brings in more than it costs. At Rs. 25 a glass, the cost of an extra glass stays below the price up to about 280 glasses and then shoots past it. So about 280 is her best day. In economists’ shorthand, profit is largest where marginal revenue equals marginal cost. With the rent, the wages and her own day all counted, Anshu’s economic profit at 280 glasses is about Rs. 1,500 a day. Her account book, which leaves out her own day, says Rs. 2,200.
The fixed cost doesn’t enter that decision at all. The rent is the same however many glasses she makes, so it can’t change which number is best. It matters for something else: whether the shop is worth having at all.
Suppose a new shop opens across the road and the going price falls to Rs. 18. Now even her cheapest glass, at Rs. 19.6, costs more than it sells for, and every day open is a day at a loss. Should she close at once? Not yet. If she stays open and makes about 200 glasses, she loses about Rs. 400 a day. If she shuts, she still owes the rent and the stove’s loan, and loses Rs. 600 a day. A business that covers its variable costs, and a bit of its fixed costs, does better to keep going in the short run. Only when the price falls below the lowest average variable cost, here Rs. 17, does each glass lose money before the rent is even counted. Then it is better to shut the door. In the long run, when the lease ends, she can leave for good if things don’t improve.
Economists usually assume a firm wants as much profit as it can make. Real owners also want to get through a bad year, to grow, or to keep a favorite place open, but a business that loses money for long can’t pay its rent.
A bigger shop
In the long run nothing is fixed. Anshu can take the room next door, buy a second stove and hire three more people. Then the question changes. It is no longer what one more person adds to a fixed room. It is what happens when everything grows together.
If doubling everything exactly doubles the tea, a second room like the first making another 280 glasses, the business has constant returns to scale. If it more than doubles the tea, it has increasing returns: with two rooms, perhaps one person can do nothing but boil milk all day, and gets fast at it. If it less than doubles it, decreasing returns: Anshu can’t be in both rooms at once, so less gets noticed. Nepal’s books call this the law of returns to scale, स्तर प्रतिफलको नियम.
Returns to scale is about output. A closely related idea is about cost. When a firm grows and its cost for each unit falls, it has economies of scale; when growth makes each unit dearer, diseconomies. They usually go together, but not always: a firm that buys a hundred sacks of sugar at once may get a better price, so its cost falls even if its tea per worker doesn’t rise.
Economies of scale that come from inside the firm, internal economies, have several sources. A bigger firm can divide its work more finely. It can buy in bulk. It can use machines that only pay at a large size: a machine that bottles tea pays for itself in a factory, never in a stall. And costs that don’t grow with output, like designing a product or advertising it, are spread over more sales.
Container ships show both sides. In the decade before 2015, the largest ships doubled in size, and the cost of carrying each container fell by roughly a third. But the savings were shrinking: the newest giants saved four to six times less than the round of growth before them, and more than half of that came from better engines, not size. The giants also pushed costs onto everyone else. Ports had to dig deeper channels and buy taller cranes, and when one giant unloads thousands of boxes at once, the trucks, barges and trains waiting for them can be held up. The report’s verdict was that making the ships bigger still would raise the cost of moving goods, once everyone’s costs are counted.
Diseconomies of scale come mostly from managing. The bigger a firm, the more layers of bosses, the slower the messages, and the more mistakes nobody notices. Where the lowest cost for each unit comes at a small size, a trade is full of small firms; where it comes only at a huge one, a few giants.
Some economies come from outside the firm, from its neighbors. When many firms in one trade gather in one place, each gains: skilled workers live nearby, suppliers set up close, and buyers know where to come. Alfred Marshall, whose Principles of Economics (1890) was for a generation the book every student of the subject read, called these external economies. In such a place, he wrote, “the mysteries of the trade become no mysteries; but are as it were in the air, and children learn many of them unconsciously.” Crowding has a price too: rents rise, roads clog, and the same neighbors become competitors for the same workers. Those are external diseconomies.
The long run also lets a firm choose how to make things: with many hands and few machines, labor-intensive, श्रमप्रधान, or with many machines and few hands, capital-intensive, पुँजीप्रधान. Neither is the modern way and the other the backward one. The cheaper way depends on what workers and machines cost, and how much each can do. Where workers are plentiful and machines dear, hands make sense; where wages rise, machines start to pay. So how many workers or machines a firm wants depends on how well its product sells and on what each costs and can do.
Nepal’s hills face that choice now. With so many young men abroad, farm workers there are harder to find and rural wages have risen. One answer is the mini-tiller, a small two-wheeled tractor that can plow terraces bullocks used to plow. In a survey of mid-hill farms in 2017, about one household in eight said it was hard to find workers, and a day’s farm work cost about Rs. 630. But a mini-tiller cost about Rs. 55,000, and on average the farmers were willing to pay about a third less than that. A machine can make sense in principle and still be out of reach for a family with little cash, which is part of why the change is slow.
So a shop’s costs come down to one law and a few sums. With the room fixed, each extra worker or dose of fertilizer adds less than the last, so each extra glass or kilo costs more to make. The average cost falls while the fixed cost is spread more thinly and rises once the crowding bites, and the marginal cost crosses it at its lowest point. A firm makes the most profit by going on while one more unit brings in more than it costs, and keeps going at a loss in the short run while it covers its variable costs. In the long run it can grow, and find bigger cheaper up to a point, or choose machines when wages rise. Anshu now knows how many glasses to make at Rs. 25. What sets that Rs. 25, and whether she could charge more, depends on the shops across the road.
Found a mistake? Please tell us.
