Market Concentration vs Market Share: How Each Measure Fails
A share compares one seller with a whole field. A concentration figure describes how the whole field is split. Both need somebody to have drawn the field first. After that, a share is one missing number away from being writable and a concentration figure is nine, and that difference decides which of the two actually appears in print.
Where the definitions and the formula are set out
Two measures call for two definitions and a formula, and all three are set out elsewhere. How a concentration figure is worked out, what scale the result runs on, and why the shares get squared before they are added, all belong to Concentration Risk: How Exposure Clusters and How It Is Measured, where they are set out in full. Herfindahl-Hirschman Index: Which Market Are You Measuring? takes the same figure and shifts the line around the field, worked right through on four nested boundaries. The index measures a field. Setting the two measures beside each other needs no computation at all, only a count of what each one is missing.
One question is left over, and neither of those subjects asks it. The question sounds almost too plain to be interesting until it has been sat with for a minute. Both measures describe the same field. Both are perfectly respectable. Why, then, is one of them printed in roughly every second document a reader will ever open, and the other one almost nowhere? The honest answer has nothing to do with which measure is better, more rigorous or more useful. The answer has to do with what it would physically take to write each one down.
Ranked by how each one fails rather than by what each one means, the two measures come out in the opposite order to instinct. A person who is one number short will fill in that number. A person who is nine numbers short will put the pen down and go home. So the measure that is nearly writable is the one that gets invented. Being nearly writable is a hazard, not a virtue. Hold on to that sentence. Every failure of either measure comes back to it.
What is each of these two measures actually made of?
A share is one seller's figure set over the whole field's figure. A share carries two terms, produces one ratioOne quantity divided by another. The value moves when the top moves, when the bottom moves, or when both move at once by different amounts., and compares one name with everybody. A concentration figure is a statement about how the whole field is divided among the sellers standing in it. No seller's name appears anywhere inside it. Swap one seller for another of exactly the same size and the concentration figure is the same figure it was a moment ago.
A share is about one name, and a concentration figure is about the shape of a list, and neither answer contains the other. A mall on a Saturday carries the contrast. One question is how much of the day's takings went through the shoe shop on the first floor. A different question is whether the day's takings were spread fairly evenly across the forty shops, or whether three of them took most of it while the rest sold almost nothing. Both are honest questions about the same building on the same day. Knowing the shoe shop got a large slice says nothing about whether the other thirty nine are all similar to each other or wildly different. Knowing the takings were lopsided says nothing about the position of any single shop.
A seller doubles its sales while every other seller in the field stays exactly where it was. What happens to the two measures?
Why does neither measure start until somebody draws a line?
Anjani Stationers Private Limited, an invented business, makes hard bound registers for schools. Suppose an analyst is asked, quite reasonably, for its share. Before anything can be divided by anything, a prior question has to be answered, and the question has no arithmetic in it at all. Which sellers count as being in the same field? Is the field the other register makers along the same lane? Is it every register maker in the city? Is it everyone who sells a school something to write in? Or is it everyone who sells a school anything whatsoever?
All four are honest answers. The four answers produce four different fields, four different totals, and therefore four different shares for the same business, and the numeratorThe number sitting on top of a ratio, the one being divided. In a share it is the seller's own figure, and it is usually the only part the seller can look up. did not move once while all that happened. The numerator standing still is the whole point. Nothing about Anjani Stationers changed. Somebody drew a line in a different place and the answer changed with it. The four nested boundaries themselves are drawn and worked in full under Herfindahl-Hirschman Index: Which Market Are You Measuring?, and are only named here.
Both measures inherit the same unsettled boundary decision, so neither of them can be called the more objective one. A concentration figure does not depend on any single company, and so it is tempting to argue that it escapes the boundary decision. The independence is real, and it arrives strictly after the field has been fixed. Fixing it is a judgement a person makes and then, usually, does not write down. A concentration figure computed over the lane and a concentration figure computed over everything a school buys are two different readings of two different fields, and nothing on the face of either one says which was which. The buyer side test that would settle such a line in principle belongs to Market Structures: Perfect Competition to Monopoly Compared, and watching a computed figure swing as the line is redrawn is the whole business of Herfindahl-Hirschman Index: Which Market Are You Measuring?
A concentration figure does not depend on any single company, and somebody takes that to make it more objective than a market share. What is wrong with that argument?
How many numbers would have to be found?
Now assume the hard part is done. Somebody has drawn the boundary, written it down, and everybody agrees. The field carries ten sellers, and one of them is the seller under discussion. The arithmetic here is small enough to do in the head.
A share takes two figures. The first is the seller's own. A business always knows what it sold, so the first figure is already held. The second is the whole field's figure. The seller does not hold that one. One number is missing. A concentration figure is a statement about how the field divides, and a division needs a value for every part, so the measure takes what every seller in the field sold, all ten of them. The seller holds exactly one of the ten. Nine numbers are missing. Nine over one is 9.00, so on this field a concentration figure is nine times the gathering job of a share.
Concentration is nine times the gathering job of a share on the same field, and the nine are not nine easy numbers either. Each one belongs to a separate business, sits in that business's own records, and is nobody else's to look up. The only route to it is to be told. There is no clever technique waiting to be discovered here, no better dataset, no smarter method. Somebody has to hand it over, nine separate times, and each of the nine is under no obligation whatever to do so.
Pay inside an office runs on the same difference. Knowing what one employee earns takes that employee no effort at all. Knowing what the whole team costs takes one number from one person, and that person may well hand it over. Knowing how the team's total is divided across every colleague takes every single colleague disclosing what they earn, and that conversation is easy enough to imagine. Same office, same room, same arithmetic, wildly different amounts of work. The ratio between those three jobs is exactly the ratio between a figure already held, a share, and a concentration figure.
The panel below adds sellers to a field. As the count rises, what happens to the number of figures still to be obtained before a market share can be written down?
Add sellers to the field and watch what each measure still needs
One control moves, and it is the number of sellers standing inside the boundary. No share is worked out and no concentration figure is worked out. The only quantity on screen is how many figures remain to be obtained.
10
9
9.00
1
At this setting the field has 10 sellers, one of which is the seller under discussion. Writing a share still takes 1 figure that seller does not hold. Writing a concentration figure still takes 9 figures. That is 9.00 times the gathering job, and both of them need the same single boundary decision made first.
Educational illustration. This panel counts missing figures and nothing else. The panel works out no share, no concentration figure and no measure of any kind, at any setting. A business always knows its own figure, so the seller holds its own at every setting. Every other seller's figure has to be obtained from that seller, at every setting. One boundary decision is required before either measure can start, at every setting, and it is the same decision for both. The default of ten matches the paper field below, where ten sellers of an interchangeable good are known to exist and no figure for any of them is published anywhere.
On a field of ten sellers, one of which is the seller under discussion, how many figures does a concentration measure still require, and whose are they?
So which of the two actually gets produced?
The plain answer is not the interesting part. The share gets produced. Constantly, everywhere, in slides and notes and articles and pitches. The concentration figure mostly does not. A reader can go a very long professional life reading about businesses without once meeting a properly built statement of how a field divides among its sellers, and will meet a market share before lunch.
Why? Not because anybody sat down and ranked the two measures. A measure that is one number short gets that number estimated, and a measure that is nine numbers short gets abandoned. Estimating one thing feels like ordinary professional judgement. A total is needed, no total can be found, so a reasonable view of the total is taken and the day moves on. Estimating nine things is making it up, and it feels like it. People can feel the difference even when they cannot articulate it. So the nine short measure never gets written, and the one short measure gets written and then repeated.
The effect on the printed record is the heart of the matter. The written world is full of market share figures resting on one estimated denominatorThe number underneath a ratio, the one being divided by. In a share it is whatever the seller's own figure is being compared with, and it is the part a seller almost never holds. that nobody names, and nearly empty of concentration figures. The measure met most often is therefore the one with the hidden estimate inside it, and the measure almost never met is the one that would have been built out of real observations or not built at all. Where a printed share figure originates is taken up by Market Share: Where the Figure Actually Comes From.
Which of these two figures is more likely to be sitting in a published document today, and why?
Why is the easier measure the more dangerous one?
The finding deserves stating without hedging. The easier measure is the more dangerous one, and the harder measure is safer precisely because nobody can produce it. Read that once and it sounds like a paradox, or like contrarian decoration. The claim is neither, and it follows directly from what danger in a figure actually is.
The danger in a number is not how wrong it might be. The danger is how invisible the wrongness is. A share resting on a carefully measured total and a share resting on a total somebody guessed over a coffee look identical in print. Same shape, same decimal places, same confident little percentage sign. Nothing on the face of the figure tells them apart. A slide has room for a percentage and no room for a story about where the bottom of it came from, so there is usually nothing in the document either.
A concentration figure cannot fail that way. Being nine numbers short, it either exists with a real distribution standing behind it, or it is never written at all. Its failure mode is absence. A blank announces itself from across the room and a plausible figure announces nothing, so absence is a safer failure than plausibility. A reader who finds no concentration figure knows immediately that they do not have one. A reader who finds a share does not know what they have. The same ruling governs a missing cost filled in with a trade average, a line better described as an invention wearing the clothes of research: the blank was the finding, and filling it would have destroyed the finding without improving anything.
One qualification, and it matters. The claim here is not that published shares are usually wrong. Such a claim would be a statement about the world with no evidence behind it, and it is not what follows from the argument. The argument yields something narrower and sharper: the measure that almost works is the measure that gets invented. Most of the invented ones may well be close. Which ones cannot be told from the outside, and neither can the person who published the figure tell.
Why is the easier measure the more dangerous one?
What does each measure tell, once it has been obtained?
An account that only attacks measures teaches distrust, and distrust is not judgement. Both measures answer a real question, and the questions are different, so each of the two is owed its due.
A share answers a question about position. Where does one seller stand relative to everybody else? Position is a genuinely useful thing to know, and a share comes with a property worth remembering: it moves when either of its two terms moves. A seller's own figure can sit perfectly still for a year while every other seller grows, and its share will fall, and nothing about that seller will have changed. A share is therefore a ratio of two moving numbers rather than a fact about the seller, and reading one that has already moved is taken up by How to Interpret Market Share Changes: Reading a Ratio.
A concentration figure answers a question about structure. Is this field divided among a few, or spread thinly across many? A concentration figure carries no name inside it at all. Change which seller is under consideration and it does not move; change how the whole list divides and it does. The measure that never names a seller is the one that says most about where that seller is standing. The claim sounds backwards and is not. A field where three sellers take most of the business is a particular kind of place to trade in, and knowing that is a much heavier fact than knowing one seller's own slice. The count of sellers in a field, and what a consolidating field does to that count, sits on Consolidation and Fragmentation: How an Industry Concentrates. The wider set of pressures bearing on a field sits on How to Apply Porter's Five Forces to an Industry.
Suppose the question is whether the field a business sells into is divided among a few sellers or spread across many. Which measure answers that, and what does the answer do if the business in question doubles?
What does either measure actually produce on such a field?
One specific field is instructive precisely because it produces neither measure. Anjani Stationers Private Limited buys its paper from a mill. There are nine further mills within reach beside that one, so ten sellers of an interchangeable goodSomething a buyer would accept from any of several sellers without minding which. What arrives is the same either way, and the sameness makes switching easy and makes the sellers hard to tell apart. are known to exist. The paper is the same weight and the same finish, and quotes come back in about a day.
A count of ten is a published fact about a field, and a genuinely useful one. On its own, the count is also completely insufficient for either measure. A count is not a share distribution. Applied to this field, that ruling settles both questions at once. There is no total to divide by, so no share can be produced. Nine of the ten values have never been recorded anywhere, so no concentration figure can be produced.
The Herfindahl-Hirschman Index treatment reaches this same field and refuses it too, for a different reason from the one given here. The index treatment refuses because there is no distribution over sellers whose count is known. The refusal here rests on the gathering cost that made the distribution unavailable in the first place: nine separate businesses would each have to hand over a figure, and none of them has. One refusal is about what is on the table; the other is about why nothing was ever put on it.
So what is left, and how should it be written? Not as a lament, and not as a caveat under a number. The honest output is the count, written as a count, with the words and nothing else is published beside it. Ten interchangeable sellers, quotes back in about a day, and no figure for any of them. Put that sentence in front of somebody who has worked in the trade for twenty years and they can push back on it usefully. A veteran of the trade may know that a tenth mill shut its gates last spring, or that two of the ten sit under one holding with different names painted on them. A figure invites no such conversation. A figure can be believed or disbelieved, and neither of those is worth much to anybody.
A field carries ten interchangeable sellers and nothing else about any of them is published. What is the honest output?
The share that was repeated until it became a fact
Nobody in this story does anything foolish, and that is exactly why it is worth telling. A business writes a slide saying it holds a certain share of its market. The numerator is its own revenue, auditedChecked by somebody outside the business who is then required to say so in writing. The word describes how a figure was tested, and says nothing about whether the figure is large or small., correct, and nobody could fault it. The denominator is a market total, and the total came from somewhere: a report, a trade estimate, a figure somebody in the industry mentioned once at a conference. The estimate is not marked anywhere on the slide, and not because anybody hid it. A slide has room for a percentage and no room for a provenance.
An analyst reads the slide and puts the share in a note, correctly attributedCredited in writing to a stated source, so a reader can see where the figure came from. Crediting a figure correctly says nothing at all about whether it was measured well. to the business. A second analyst reads that note and puts the share into a model, correctly attributed to the first note. By that point the share looks like a plain fact about the industry, and a journalist reads the model and writes it as one. Four steps, nobody lied, and the single estimated number in the whole chain is now completely invisible.
Now ask what would have happened to a concentration figure travelling the same road. The contrast is the whole teaching. The concentration figure would never have started. Writing the first slide needed nine figures the business did not hold and could not get, so the slide could not have been written at all. The measure that could not be produced produced no false record, and the measure that almost could produced four.
Land the cost somewhere specific. A general worry about bad data teaches nothing. The share is now being used to size a decision, and the decision is sensitive to the denominator, and the denominator is the one term in the entire chain that nobody has ever checked. Worse: if the original estimate was flattering, the error runs in a consistent direction every single time it is repeated. Repetition does not average it out. Repetition entrenches it. Each retelling adds a source and removes a provenanceThe record of where something came from and whose hands it passed through on the way. Applied to a number, it is the trail back to whoever first worked it out..
The fix is not a better estimate. The fix is to print the denominator beside every share, with its source. Then the one estimated number travels with the figure instead of falling off it at the first handover.
Four questions to ask of any published figure of either kind, in this order
An equity analyst, a lender or anybody reading a pitch puts four questions to a figure that arrived rather than one they worked out. The four take about ninety seconds and are worth every one of them.
One, which sellers are inside the field? Wanted as a sentence, not as a word. A figure described as a share of the stationery market says nothing; a figure described as a share of hard bound registers sold to schools in this district says a great deal, including how to disagree with it. If the document cannot answer this, the remaining three questions are about a figure that does not yet mean anything.
Two, where did the denominator come from? Specifically: was it measured, was it estimated, or was it taken from somebody else who estimated it? The third of those is the most common and the least often stated. A denominator with a named source that can be checked is a different animal from one with no source at all, even when the two produce the same percentage.
Three, how many of the component figures were actually observed? For a share the honest maximum is two, and it is usually one. For a concentration figure it is as many as there are sellers, and a properly built one is genuinely worth reading for exactly that reason. Counting observations rather than trusting decimal places is the single most useful habit of the four.
Four, who is served by the figure being this size? The question is not cynicism. Somebody chose to produce the number, chose to publish it, and chose that particular form, and asking who is served by those choices is the ordinary response. A share is usually published by a party with a view about how large it should look, and noticing that is simply reading carefully.
A figure that survives all four questions is rare, and knowing which of the four it failed is far more useful than discarding it. A share that fails question two is still evidence of something. The share is just evidence about the seller's own revenue and about somebody's view of a total, a smaller and more honest claim than the percentage was making.
Which of the four questions on the closing card asks nothing about how the figure was built?
India for the setting, everywhere for the mechanism
India supplies the setting: the digit grouping, and the words Private Limited sitting at the end of one invented name above. The table below names a single Indian institution to establish that a competition regime operates, and for nothing further than that. The mechanism itself travels anywhere. All the counting above turns on two things only: how many sellers stand inside the line, and the fact that each seller's figure sits in that seller's own records. Neither of those changes at a border. The present wording of any rule about how fields get measured or reported sits on the institution's own site.
Where can the one claim that is not arithmetic be checked?
Neither source supplies a number. A competition regime is the reason anybody bothers measuring how a field divides at all, and the only arithmetic in the comparison is a tally of figures that are missing.
| Row | What that row establishes, and what does not travel from it | Where | Date checked |
|---|---|---|---|
| Competition Commission of India | Establishes that a competition regime operates in India, and that the way a field divides among its sellers is among the matters somebody working inside such a regime attends to. | cci.gov.in | 25 August 2026 |
| The counting done above | Establishes what the arithmetic above consists of: a tally of how many figures each measure still requires, and the ratio between two such tallies. Not one share was worked out for any seller. Not one concentration figure was worked out for any field. | the site these notes sit on | 25 August 2026 |
Anjani Stationers Private Limited is invented.
Educational material. Not advice on any investment, tax, budget or market position.