Consolidation and Fragmentation: How an Industry Concentrates

A fragmented field carries many sellers and none of them is big enough to set a price. A consolidated field carries few, and each of them can. Consolidation changes what every seller in the field faces rather than what any one of them does: fewer sellers, less pressure on price, and returns that need not stay thin. Count the sellers, then ask how the field splits between them.

Start by pinning down what the two words are attached to. Most careless use of them begins there. Fragmented and consolidated describe a place, not a business. The two words describe the field a business sells into: how many sellers stand in it, and how the money in it divides between those sellers. A seller inside a fragmented field can be superbly run or badly run. Neither word settles which. Once that is accepted, a sentence like "this business is fragmented" stops making sense, as it should.

Now the harder thing. Both of those facts sit outside everybody's accounts. A business's own statements record what happened to that business, line by line, checked by somebody. No seller keeps a record of its rivals, so a business's statements cannot say how many other sellers stand beside it, and they certainly cannot say how the money divides across a field. Both facts have to be gathered rather than read, and one of the two usually cannot be gathered at all.

Setu Bazaar, an invented marketplace, supplies four published figures for one trading year: 2,000 merchants sell on it, Rs 5,00,00,00,000/- of goods crossed it, Rs 4,80,00,00,000/- of that reached those sellers, and its own result for the year was a loss of Rs 2,50,00,000/-. Anjani Stationers Private Limited, an invented maker of hard-bound school registers, supplies two buying arrangements and nothing else: the paper it buys comes from a mill with nine alternatives sitting beside it, and the servicing of its machines comes from a contractor with exactly one.

What does the subject rest on?

First, two questions that look like one question and are not, and keeping them apart is the whole discipline of the subject. How many sellers does the field carry? One fact. How does the field divide between them? A second, independent fact. Ten sellers can be ten near equals, or one large name with nine small ones clinging to the edges, and those two fields behave nothing alike from any angle. The count is settled below and the split is settled under the Herfindahl-Hirschman index. An account that quietly treats a count as a split has claimed something it cannot support.

Second, one published population of sellers, and it is the only one with a money figure attached. Setu Bazaar has 2,000 merchants on it. Rs 5,00,00,00,000/- of goods crossed the marketplace in the year and Rs 4,80,00,00,000/- of that reached the sellers. The money reaching sellers divided by the count of them gives an average of Rs 24,00,000/- of goods sold in the year for each merchant. The average is one division on two published figures, and it describes the size of a place in that field. How the Rs 4,80,00,00,000/- actually divides between the 2,000 names is published nowhere, so the average is not a claim about any individual merchant.

Third, and the label belongs on every flow figure wherever it appears: gross flowThe value of what passes through an arrangement, counted before anything is taken out of it. It is what moved, not what anybody kept, and turning one into the other needs subtractions that may not be published at all. is not what anybody keeps. Rs 24,00,000/- a merchant is the value of goods that left. The figure sits before the cost of those goods, before the cost of selling them, and before anything the merchant paid to be on the marketplace at all. Read as income it becomes a profit figure, and it is not one.

Fourth, two input fields observed through one buyer. One set of records can show what a seller count does to a buyer without a consolidating field to point at. Anjani Stationers Private Limited buys paper in a field where nine alternatives sit beside the incumbentThe seller already supplying the buyer when the question comes up. The word says only that this one is in place today, and carries no claim about its size, its quality or how long it stays. mill, a quote is back the same day and a first delivery lands in roughly a fortnight. Anjani Stationers buys machine servicing in a field with exactly one alternative, where replacement takes 26 weeks. Same buyer, same year, two fields, and the difference between them is the number of sellers in each. Twenty six weeks against two is a ratio of 13, and that ratio is what a seller count looks like by the time it reaches a buyer's own arrangements.

Fifth, consolidation is a movement and not a state, and confusing the two produces most of the loose writing on this subject. Consolidation is a fall in the number of sellers in a field. Fragmented and consolidated are the two ends it moves between. So a sentence calling a field consolidated is describing where the field is. A sentence calling it consolidating describes a movement, and a movement always needs two dates and a direction behind it before it means anything.

Sixth, and the second half rests on it: a consolidating field can only be shown by a count taken at two dates, and no set of statements anywhere contains one. Setu Bazaar and Anjani Stationers are each observed once, so neither business stands in a field that can be called consolidating. Two fields observed once can still show what a falling count does to a buyer, and the facts somebody would need to gather to establish a fall can still be named.

What are fragmentation and consolidation actually describing?

Every question about a field comes back to the same pair. How many sellers, and how does the field divide between them. In ordinary conversation the two arrive glued together, and the glue is where the trouble lives. Ask them as two questions with a gap between them.

Here is the everyday version. Ten shops in one market sell the same model of phone. A walk down the road counting shutters settles in ten minutes that there are ten sellers. The first question is answered, cheaply, from outside. Now the second one. Which of those ten sells four phones in a day and which sells forty? A shop records its own sales and not its neighbour's, so nothing on the road settles it and nothing in any one shop's own books settles it either. A week of standing there still yields only a count.

The asymmetry between the two questions runs through everything below. The count is the fact that can usually be got. The split is the fact that usually cannot. And because writing is easier than gathering, the word fragmented gets written far more often than it gets checked.

Fragmentation: what makes a field fragmented?

A field is fragmented when it carries many sellers and no one of them is large enough to move the price by deciding to. The price is made by the field and taken by everybody standing in it. The definition ends there. Notice that it is a statement about what a seller can and cannot do, not about how many sellers there happen to be in absolute terms.

The mechanism underneath is set out under market structures, and one clause of it is enough: where many sellers offer near-identical goods, no one of them is able to move the price. A seller who lifts a rate watches the buyer walk four steps sideways to somebody selling the same thing. A seller who cuts one has given money away for volume that would probably have come anyway. A seller in that position is left with doing the same thing more cheaply or more reliably than the seller beside it. Running cheaper is real work, and it is a different thing from setting a price.

Now work it on Setu Bazaar's published population of sellers. Setu Bazaar has 2,000 merchants on it. In its published year, Rs 5,00,00,00,000/- of goods crossed the marketplace and Rs 4,80,00,00,000/- of that reached those sellers. Divide the second figure by the count and each merchant averages Rs 24,00,000/- of goods sold in the year.

Say what that average is and what it is not in the same breath, every time, or it will be misread within a paragraph. It is one division on two published figures. The average describes the size of a place in that field, and it is gross flow rather than anything kept. How the money divides across the 2,000 names is published nowhere at all, so the average is not a claim about any individual merchant. Some of those merchants may be many times that figure and some a fraction of it, and the published figures cannot settle which.

Two hundred stalls stand in one wholesale market. The length of it walks in ten minutes, and at the end of the walk nobody could name the biggest stall in the place. That is a fragmented field seen from outside, and the honest description of it is a count and a shrug.

Why does a count of sellers settle nothing on its own?

Because a count says how many names stand in a field and says nothing whatever about how the field divides between them. The whole limit is in that one line, and it is the sentence worth carrying away even if everything else goes.

Two populations of sellers appear above, and neither comes with a split. Setu Bazaar's 2,000 merchants are the first. The Rs 4,80,00,00,000/- that reaches them is not published as a division across those names anywhere, in any form. The second population sits in the field Anjani Stationers Private Limited buys paper from: nine alternatives beside the incumbent mill, so ten sellers of an identical paper are known to exist, and not one of their shares is published either.

Ten sellers could be one name at nine tenths of the field and nine names scrapping over the rest, or ten near equals, and a count cannot tell those two apart. Those fields would feel completely different to a buyer standing in either of them, and the count is identical in both. So a count cannot even rank two fields against each other, let alone describe one.

How a field's split is measured, and what happens to that measurement when nobody agrees where the field ends, is covered under the Herfindahl-Hirschman index. Even an approximate split answers a question a count cannot answer.

Thin returns: what do they look like, and what are they a fact about?

Returns are thin where what any seller keeps is small against what passes through its hands, and where nothing in the field lets one seller charge more for the same thing than the next one does. Both halves of that matter. The first half is about the shape of the money. The second half is about why the shape does not change by itself.

Setu Bazaar offers two readings of thinness in one year, from two different heights, and they need to be kept apart. On the seller side, Rs 24,00,000/- of goods a merchant in a year is gross flow, so what any merchant finally keeps is smaller than that figure by an amount nobody publishes. The cost of the goods came out of it. The cost of selling them came out of it. Whatever the merchant paid to be on the marketplace came out of it. Three subtractions, none of them published, and no published line turns the turnoverThe value of sales made in a period. It sits at the top of a statement of profit and loss and every cost of making those sales is still to come out of it, which is why a large turnover and a loss sit together comfortably. figure into a kept figure.

Setu Bazaar publishes its own year, so on the marketplace's own side the reading is different and rather sharper. Of the Rs 5,00,00,00,000/- that crossed it, Rs 4,80,00,00,000/- went to the sellers, so Rs 20,00,00,000/- stayed behind as the marketplace's own revenue. Its standing baseThe costs a business keeps paying at much the same level whether trade is heavy or light. Rent, wages and systems belong here, and a slow month makes not one of them any smaller. is 62.50 per cent of that revenue, and its published result for the year is a loss of Rs 2,50,00,000/-.

Here is the claim that matters most: thin is a description of a field and not a verdict on anybody's competence. There are two conclusions a reader reaches for at this point and neither one is available. The first is that the merchants are doing badly. A flow figure is not an earnings figure and the subtractions are missing, so nothing published supports it. The second is that the marketplace is doing badly. The second is closer to something, but it is one published year of one business, and a single year describes a year.

Think of a road with twelve tea stalls on it. All twelve charge ten rupees a cup. All twelve are busy from six in the morning. Not one of them can charge eleven. The customer simply walks twenty steps further. Twelve busy stalls at ten rupees a cup are thin returns, and none of the twelve stall holders is bad at their trade. The thinness belongs to the road.

How Industry Consolidation Changes Competitive Dynamics: what moves, and who feels it first?

Define it before anything else, and define it as a movement. Consolidation is a fall in the number of sellers in a field, however that fall comes about. Fragmented and consolidated are the two ends it travels between. A field can lose sellers without anybody buying anybody, so nothing in the definition mentions a transaction.

Four things change as the count comes down. Each is a claim read off published structure rather than asserted from the general shape of the idea.

One, the number of alternatives a buyer can actually reach falls, and this is the change a buyer sees first because it is the only one visible from inside a buyer's own arrangements. Noticing that the list of sellers still worth calling has got shorter takes no field-wide data at all, only the buyer's own records.

Two, the replacement timeHow long it takes, from the decision to move, until a different seller is supplying normally. It covers finding one, agreeing terms and getting the first delivery working, and it is measured in weeks rather than in money. lengthens, and this is the change that costs money rather than the price change everybody watches for. A stoppage is expensive whether or not anybody raised a rate.

Three, price stops being made by the field and starts being set by somebody. The fragmentation mechanism runs backwards. Many sellers of a near-identical thing leaves no one of them able to move the price. Take enough of them away and that sentence stops holding.

Four, who has to be persuaded changes. In a field of many sellers a buyer negotiates by leaving, or by being visibly able to leave. In a field of few, leaving is not available, so the buyer has to negotiate by talking. Leaving and talking are different skills on different timescales, and a buying operation built for the first one does badly at the second.

Now work the first two from the published case, the part worth remembering. Anjani Stationers Private Limited buys in two fields at once, in the same year, out of the same set of records. In the paper field, nine alternatives sit beside the incumbent mill, a quote is back inside a day, and a first delivery arrives in roughly two weeks. In the machine servicing field there is exactly one alternative, and replacement takes 26 weeks.

The two fields differ in the number of sellers, and what reaches the buyer's own arrangements is the replacement time: 26 weeks against 2 is a ratio of 13. The ratio is the most useful point in the whole subject. A seller count shows up in a buyer's accounts as time long before it ever shows up as a rate. Time is the early signal and rate is the late one, and almost everybody watches the late one.

The lane version: six tailors work on one street, and a bad fitting costs a walk to the next door and a week's delay. The one man in the district who repairs a particular model of machine is a different situation entirely, and the difference has nothing to do with what either of them charges.

Two boundaries sit one sentence away at all times. The terms the servicing contractor could demand are a separate question with a separate answer, covered under supplier power. And neither field is being called consolidated: both are states observed once. The contrast reasons about what a falling count would do. Claiming a count fell is a different thing.

What has to be true before a field can consolidate at all?

Start from the obvious objection, a good one. Sellers leave fields all the time. Shops shut every week. And somebody else opens in the same doorway within a month, so most fields stay exactly as crowded as they always were. So the count only falls where something stops a new seller taking the place of the one that went.

Three things do that, and each is familiar enough to be named rather than re-explained. Being in the field at all costs a great deal, so the entry costWhat somebody has to put up before they can trade in a field at all: the plant, the licence, the stock, the first year of losses. It is spent before the first sale and it does not come back if the venture stops. keeps newcomers out. Something in the field cannot be added at will, so no amount of willingness produces another seller. Or the right to sell is granted by a body standing outside the sale altogether.

Then the second condition, and it is the one almost everybody misses: capacityThe output an operation is built to produce in a period, set by whichever stage of it is slowest. It is a physical property of plant and people rather than a figure in a set of accounts. has to leave the field rather than change hands. A works bought by a rival and kept running has moved a name on a list without removing a seat at the table. The same output still reaches the same buyers, made by the same people, on the same machines, and the only thing that changed is the name above the gate. A field whose capacity all stays in place has consolidated on paper and not in fact.

A market carries ten fruit sellers and one of them gives up his stall on a Tuesday. If somebody else is selling from that stall on Thursday, the count did not fall. If the stall stands empty because the licence went with the man who left, it did. Same event, two completely different fields, and only the second week settles which of the two it is.

Neither Setu Bazaar nor Anjani Stationers has been observed at two dates, so none of the three fields above can be called consolidating. An invented example would supply exactly the evidence nobody has, and a manufactured example teaches how to recognise a drawing rather than how to recognise a field.

How would anybody tell whether a field is consolidating, when no seller publishes a share?

Five things are worth watching, in the order somebody could actually gather them rather than in the order they matter. Two of the five can be read from the records of Anjani Stationers.

One, is the count of sellers falling? A falling count needs the field observed at two dates, and it is the fact people most often assume rather than gather. Somebody remembers there used to be more, and the memory becomes a finding.

Two, is anybody buying anybody? A purchase is public wherever such things are recorded, and it establishes that a name moved. A recorded purchase does not establish whether any capacity moved with it.

Three, is capacity leaving? Capacity leaving is the observableA fact somebody can go out and record, as against one that is inferred, remembered or assumed. Naming a fact as observable is a claim about how it would be got, not about how important it is. that decides whether the second one meant anything at all. A works closed is capacity gone. A works sold is a name gone.

Four, is a buyer's list of alternatives getting shorter? Any buyer can read this off its own arrangements, and this case supports it: nine alternatives in one field and one in another, on the same buyer's books, in the same year.

Five, is replacement time lengthening? The same buyer can read this too, and it is the earliest honest warning available to anybody. Two weeks against 26 weeks in one year is what the two ends of that observable look like side by side.

Why is a fragmented field not automatically a competitive one?

One last correction remains, and it is a short one. Many sellers of a near-identical thing means no one of them can move the price by deciding to. The definition is a statement about the price, and about nothing else at all.

Nothing in it says buyers can move between those sellers easily. Nothing in it says the sellers are interchangeable for every buyer's particular purpose. Nothing in it says any given buyer has an alternative it could actually reach this month. Ease of movement, interchangeability and reach are three separate questions, and each has its own answer.

The case above makes the point without needing another one. Both of Anjani Stationers' input fields carry more than one seller. In one of them a replacement is supplying normally in about two weeks. In the other it takes 26. Both fields have alternatives in them; only one of them has alternatives a buyer could use inside a quarter.

The street version is three banks standing within a hundred metres of each other and a loan already drawn from one of them. The count of banks on that street is three, and it does very little for the borrower once the papers are signed.