Sector vs Industry: Two Units of Counting That Do Not Nest

A sector and an industry are two different units of counting, not a big box and a small box inside it. A sector is a statistical cut of a whole economy, made to answer a question about the economy. An industry is the field one business competes in, drawn around who its buyers could switch to. The two do not nest.

What is this particular comparison for?

One worry can be set down at the start. The two words are used interchangeably in ordinary writing, in careful writing, and by people who know exactly what each one means, and correcting that usage would be a waste of everybody's afternoon. There is only one thing about them that genuinely differs, and it is not the word. The difference is what got counted. Setting a sector beside an industry changes the population being counted, not the evidence being read or the boundary being tested.

The distinction is worth a minute: everything downstream of a count inherits the unit the count was made in. A number described as an average, a total, a growth rate or a share is a number about some set of businesses, and if the set was assembled to answer one question and then read as an answer to a different one, nothing later in the work can repair it. The arithmetic will be flawless and about the wrong crowd.

Picture two people describing the same tea stall on the same street. The first wants to know how many people in the district make a living selling tea, so she counts every tea stall in the district. The second counts only what a customer standing at that stall at four in the afternoon would walk to instead. The second list turns out to be the sugarcane juice cart at the corner, the two stalls on the next street, and nothing else in the district at all. Both counts are correct. Neither is a smaller version of the other. The first is built so that the parts add up to the district; the second is built so that a choice can be made between the members. Asking which of the two counts is right poses a question with no answer. Asking which of the two is in hand makes the count usable.

What is a sector, and where does its full account sit?

A sector is a cutA division of a whole into parts, made so that the parts can be counted separately and still add back to the whole. Every item lands in one part and in no more than one. of a whole economy, made so that the parts can be counted. An economy gets cut three ways: by what gets produced, by who ends up buying the output, and by how completely a measuring system records the activity. How an economy is divided, what each division reveals, and why the right one depends on the question being asked are all covered separately under Sectors in Macro Analysis: How the Economy Is Divided. The same account rules that the three are three questions rather than three rival classifications, and that none of them outranks the others.

One requirement carries the whole comparison. A cut must cover everything and cover it once, or the parts stop adding to the whole. Every property a sector has follows from that single requirement. Take the sentence apart slowly. A lot hangs on it.

Cover everything: a business left out is a hole in the aggregateA total built by adding many separate items together, such as everything produced across a whole economy in a year. An aggregate is only as complete as the parts that went into it., so there can be no business anywhere that fails to land somewhere. Cover it once: a business counted twice inflates the whole, so no business can land in two cells of the same cut. Put together, those two give a rule applied to every business alike, decided in advance and applied from outside, rather than a description of any one business's situation. A rule of that kind is why a cell can group a business with somebody it has never heard of and has nothing to do with. Nothing in the definition of a cell was ever asked to describe what its members compete with. The definition was asked to make the parts add up.

What is an industry, and what actually draws its edge?

Here the ground changes. A sector is drawn by a rule applied from outside. An industry is drawn by something no rule can supply, and its definition takes longer for that reason.

An industry is the field one business competes in: the set of sellers a buyer would choose between. Its edge is drawn by a single question: does a buyer refused at one seller find the same thing at the next? Notice who is being asked. The question is put to a buyer and never to a seller. The discipline is established under Market Structures: Perfect Competition to Monopoly Compared, as one of three tests that place a market. The other two, and the account of what the three together settle, are covered there. Only the asking travels: a seller's opinion about who its rivals are is a view, and a buyer's willingness to walk from one to the other is a fact about the field.

Three consequences fall out of that, and it is worth counting them so none goes missing. One, nobody asked an industry to add up to anything, so it need not cover a whole economy. Two, two different buyers may draw two different circles and both be right about themselves, so two industries may overlap. Three, one seller can sit in two industries at once, if it makes two things that two different sets of buyers choose between differently. An edge drawn by what a buyer would do instead has no reason to be tidy, and its untidiness is not a fault in the definition. A cut is tidy because tidiness was a requirement. An industry has no such requirement and gains nothing by pretending to.

Go back to the four o'clock tea. A stall selling tea and a cart selling sugarcane juice stand outside the same office. For the clerk deciding what to drink before the last hour of work, those two are squarely in one industry: refused at the tea stall, she walks eight steps and buys juice. For the man buying a crate of bottled juice to take home for a wedding, the tea stall is not an alternative at all and never enters his thinking. Same two sellers, same street, same afternoon. Two buyers, two edges. The question was never asked in the abstract but of somebody, so neither edge is the true one.

What one test shows which of the two somebody means?

Somebody hands over a group of businesses and a figure computed across it. How the group was built is unknown and nobody is going to say. One question settles it, and one only: what would have to change for a business to leave this group?

If the answer is that a question asked of the economy would have to change, or that a measuring system would have to change how it records something, the group is a sector. If the answer is that a buyer's alternatives would have to change, the group is an industry. The question is the whole test, and it takes about four seconds.

The test is about what moves the boundary and not about how big the group is, and size is exactly the wrong instinct because it is the one everybody reaches for first. A sector cell can be tiny, a handful of businesses in a corner of a cut nobody thinks about. An industry can be enormous, every seller on earth of a thing that everybody buys. A mental picture of a small circle sitting inside a large one predicts that moving the large one moves the small one. The prediction fails in all three worked cases that follow, and the failure is the harm the picture does.

So the instruction that matters is short. Ask what would change the membership before asking which word somebody meant. The word is a label somebody reached for quickly. The mover is a property of the group and it cannot be faked.

How does a business end up in the same cell as its own supplier?

Anjani Stationers Private Limited, an invented manufacturer, makes hard-bound registers and exercise books. Anjani Stationers buys paper from a mill. The mill takes 46.0 per cent of everything Anjani Stationers pays out to parties beyond itself, the single largest of its outside paymentsThe money a business pays to parties other than itself: its suppliers, its contractors and the people who service its machines. Outside payments are a spend list, not a cost statement.. Nine other mills within reach make the same weight and finish, so a replacement takes about two weeks.

Now apply the first cut, by what is produced. Anjani Stationers makes things. The mill makes things. A cut cannot ask who somebody sells to without becoming a different cut, so a cut by what is produced puts a register maker and the paper mill it buys from in the same cell, and it has to.

Then ask the industry question, and ask it properly, of the buyers. The 36 schools and institutions that buy from Anjani Stationers want registers. Not one of them, refused a register, buys a reamA standard bundle of paper sheets, and the unit a mill quotes in and invoices in. A ream is an input to a register rather than a thing a school orders. of paper instead and rules the lines itself. The mill sells the wrong object to the wrong buyer for that to be possible. So the mill is never in the industry of Anjani Stationers at all, in any reading, by any buyer, on any day.

The result is already odd and it gets sharper. The two are not merely grouped differently. The mill and the register maker sit on opposite sides of the very same transaction, so a movement that helps one of them costs the other by construction. When the mill's rate rises, the mill takes more and Anjani Stationers pays more. There is no state of the world in which both of them do well out of the same rupee moving. A grouping that treats them as alike has not merely put two strangers together; it has put a buyer and a seller in one row and called them peers.

Can the sector answer move while the industry stays exactly where it is?

Anjani Stationers sells to 36 schools and institutions in one city, all of them invoiced directly, with the Sunrise Public School group the largest at 30.00 per cent, against a year's revenue of Rs 2,70,00,000/-. Now apply the second cut, by who buys the output. The cut by who buys names four buyer types: households, businesses, government and buyers abroad.

Where does a school sit among those four? The answer depends entirely on how a measuring system decides to treat a school, and that decision is made outside the business by people the business never meets. Here is the part worth pausing on. Nothing anywhere in the accounts of Anjani Stationers settles it. The accounts record 36 accounts and Rs 2,70,00,000/- of revenue and the 30.00 per cent that the Sunrise Public School group represents. The answer was never a fact about this business, so every line of them read twice still yields nothing. The answer is a convention held by somebody else.

So take the decision both ways and watch what happens on the other side. Whichever way it goes, the 36 schools still want registers, still hold whatever alternatives they held yesterday, and could still walk to exactly the same sellers they could have walked to before. The sector answer moved and the industry did not move at all. A box inside another box moves when the outer box moves, so two boxes that move independently do not nest. The industry sat perfectly still while the outer answer changed, and sitting still is only possible for something that was never inside.

The same shape sits in an ordinary building. One canteen operates inside one office block, feeding the staff. The canteen's output may be recorded as going to households, on the ground that individuals are eating lunch, or to a business, on the ground that the block's employer settles the bill. Somebody in a statistics office makes that decision and nobody in the kitchen influences it. Meanwhile the canteen's competition is the two food carts at the gate, on both readings, on every day of the year, and the cooks know it without being told.

What does a marketplace's own cell say about what crosses it?

The third worked case runs the other way, and it needs the second invented business. Setu Bazaar runs a marketplace and holds none of the goods that move across it. Under the first cut, by what is produced, Setu Bazaar is doing things for people: it operates a place where trades happen, and the trades are not its own.

The size of that flow is the striking part. Rs 5,00,00,00,000/- of goods crosses Setu Bazaar in a year, sold by 2,000 merchantsSellers who list their own goods on somebody else's marketplace and are paid for what sells. The marketplace takes a slice; the goods and the risk stay with them. to 50,000 buyers, against Setu Bazaar's own revenue of Rs 20,00,00,000/-. The flow is 25 times the revenue. The flow is often called gross merchandise valueThe worth of everything sold across a marketplace in a period, counted before anybody's slice is taken out. The figure measures what crossed, not what was earned., and it measures what crossed rather than what anybody earned.

Now here is the sector point. Every one of those 2,000 merchants sits in whatever cell its own product puts it in: the ones making things in a making cell, the ones growing things in a growing cell, and so on across a cut that had to cover everybody. A marketplace has one sector answer of its own while sitting inside a flow that spans many, and neither of those two facts is about who it competes with.

And the industry point runs alongside it. Setu Bazaar's industry is other places the same buyers could go to buy the same things. A merchant selling across the marketplace is not an alternative to the marketplace for a buyer who is already standing on it, so the industry is never the 2,000 merchants. Selling through somebody is close to the opposite of being an alternative to them. Think of the vegetable vendors inside a covered municipal market. The vendors are not what the market building competes with. The market competes with the other covered market four streets away, and the vendors may be standing in that one too.

Where do the two boundaries disagree most sharply?

The third cut divides activity by how completely it is recorded. Some of what happens in an economy sits inside arrangements a measuring system reaches easily, with paperwork and filings and a trail. Some of it sits in arrangements that same system reaches less completely, not because anything is wrong with it but because it leaves less behind to count.

Look at what that cut is asking. The cut is asking who the measurement can see. An industry edge asks who a buyer could turn to. The two questions have no reason whatever to agree, and nothing in either one gives it any pull on the other. A seller can be easy to count and impossible to switch to. A seller can be almost invisible to a counting system and the first place a buyer walks when the usual seller says no. A buyer choosing between two sellers does not first check which of them is easy to count, and would not know how to check.

The consequence is worth stating carefully. A figure computed over one of those recording cells and a figure computed over an industry can be describing different populations of sellers even when they carry the same name in ordinary conversation, and neither figure is wrong. Each one answered the question it was built for. The gap was produced by two different questions asked of the same street. The recording cut, and what it reveals, is covered under Sectors in Macro Analysis: How the Economy Is Divided.

How many sector answers does one business actually carry?

A classification ends up with exactly as many cells as its tests produce, and this one has three tests, so the count comes from the tests rather than from assumption. Three cuts, each dividing the whole economy on its own, means every business has at least three sector answers at once, one from each cut. The three are answers to different questions, and a different question cannot contradict the one being asked, so they do not compete with one another.

Work it on the case, and never state a sector answer without naming which cut produced it. Under what is produced, Anjani Stationers is making things, and so is the paper mill. Under who buys the output, the answer turns on how a school is recorded and nothing in the accounts settles it. Under how completely activity is recorded, it is a private limited company that files, and a school choosing a supplier has never once asked that question. All three are simultaneously true. A business does not have a sector. A business has as many sector answers as there are cuts, and each one is complete on its own terms.

Now refuse the reading nearly everybody arrives at. Somebody reads all of the above, nods, and concludes that the fix is a finer classification: cut the cells smaller and eventually a cell will contain register makers and not paper mills. The thought is reasonable and it does not work.

A finer cut is still a cut, moved by a rule rather than by a buyer, so making the cells smaller brings a sector cell closer in size to an industry while leaving it exactly as far away in kind. Subdivide once and the cells are smaller and still have to cover everything and cover it once. Subdivide four times and the cells are much smaller with the same requirement and the same mover. Nothing in the subdivision ever asks a buyer anything, so nothing in it can produce a boundary drawn by what a buyer would do instead. The size gap closes and the kind gap does not narrow at all.

There is a second thing to say plainly here, and it saves a great deal of pointless argument. Two data providers can put the same business in different cells without either of them being wrong. A cut is a decision, more than one reasonable decision exists, and two careful people asking slightly different questions of the same economy will draw slightly different cells. Disagreement of that kind is the nature of a cut, not a defect to be reported.

What happens when one business is watched through all three cuts?

Reading that the two boundaries move independently is easy. Believing it is harder. The picture of a small box inside a big one is stubborn, and it survives being told it is wrong. So put one business through the three cuts yourself and keep an eye on the ring at the bottom of the panel rather than on the cell at the top.

What gets written down when either word is used?

The thing to carry away from all of this fits on the back of an envelope. Whenever either word is used, or read in somebody else's writing, two lines are worth writing and nothing else.

Line one, what got counted: name the population, never the word. Line two, what would change the membership: either a question asked of an economy, or a buyer's alternatives. Fill both in and there is nothing left to argue about, so the argument about which word was meant simply stops happening.

Worked on the case, twice. Under the sector unit, line one reads: every business that makes things, including the mill Anjani Stationers buys its paper from. Line two reads: the question being asked of the economy, or a change in how a measuring system records something. Under the industry unit, line one reads: the sellers those 36 schools could buy registers from instead, none of them yet named. Line two reads: a change in what those schools could turn to.

A card that ends in a population and a mover is finished, and a card that ends in the word sector or the word industry has not started. Notice that the industry card is the one with an honest hole in it, and that the hole is information rather than a gap in the work. The hole says the population exists and has not been enumerated. A true statement about the state of the evidence is a great deal more useful downstream than a full-looking list assembled from the wrong unit.

How a working analyst, a lender and a household use the same two lines

An equity analyst builds a list of comparable businesses more often than almost anything else in the job, and the two lines are what stops the list being assembled out of whatever grouping is nearest to hand. The card asks what got counted before any figure is set beside any other figure, and where the honest answer is that the population could not be enumerated, that is what gets reported. The output is shorter and truer.

A lender reading a proposal from a maker of registers meets the same question in a different coat. A figure quoted as typical for businesses like this one is a figure about some population, and the two lines force the question of which. If the population was a cell built to add up, it contains this borrower's own suppliers and a great deal else besides, and it is not a description of what this borrower faces when it tries to hold a price.

And a household does it without any of the vocabulary. When somebody says the shops on this road are doing badly, the useful reply is always the same: which shops, and what would have to change for one of them to stop counting? Asked at the dinner table, that is the card being run. The whole comparison is that question dressed up for work.

Anjani Stationers Private Limited, the Sunrise Public School group and Setu Bazaar are invented.
Educational material. Not advice on any investment, tax, budget or market position.