Make vs Buy: Whether a Step Is Worth Owning at All

Make against buy asks whether one step of the work is worth doing inside, at the volume the works actually runs. Binding at Anjani Stationers has three right answers: Rs 10,00,000/- to buy it in, Rs 8,00,000/- in Anjani's own accounts, and Rs 5,90,000/- to the group that contains both companies. Making and buying come to the same thing at 93,750 registers, far below any volume at which the business survives.

What is this particular comparison for?

Four comparisons in this sequence set two things against each other, and read quickly they can look like the same argument written four times. Each one asks a different question, and the question decides which arithmetic is relevant. Make against buy asks whether one step of the work is worth doing inside at all, at the volume the works actually runs. Which direction a business should grow in, outward into more of the same or upward into the stages above and below it, is a separate question and is answered under vertical against horizontal integration. How much paper to keep sitting in the shed, and what a stall costs when the shed runs dry, is a third question and is answered under just-in-time against just-in-case inventory.

The volume in that sentence is not decoration, and it is the whole reason the question has to be worked through rather than answered once. A step of the work can be worth doing inside at one volume and worth handing to somebody else at another, and nothing about the step itself changes when the volume moves. The number of units available to spread a fixed cost across is what changes. So the honest form of the question is never simply whether to make or to buy. The question is whether to make or to buy at the volume actually being run, and a business that answers it once and files the answer away has stopped asking the only part that moves.

Anjani Stationers Private Limited makes hard-bound registers and, in the year these notes work with, made and sold 2,50,000 of them. Binding is one step in that work. Anjani has a supplier for it, Chitra Binding Works Private Limited, and holds seventy per cent of that supplier. So the question is concrete: binding gets done, somebody gets paid for it, and what has to be established is what it costs and whether the arrangement is the right one at 2,50,000 registers a year.

Why does one step of the work have more than one right cost?

Start away from the works entirely. A tap in a house is leaking. The neighbour, who fixes taps for a living, comes across and mends it, and hands over a bill for eight hundred rupees. So what did the repair cost? To the householder, holding the bill, it cost eight hundred rupees. To the neighbour it cost the washer, the trip and an hour of a Sunday, perhaps three hundred rupees all told. And if the two of them happen to run one kitchen and one purse between them, the household is out three hundred rupees and not a paisa more. The other five hundred moved from one pocket to another inside the same house.

Nobody in that story is lying and nobody has made an arithmetic slip. Three figures, three sets of books, three perfectly correct answers to what sounds like a single question. Three answers is not a contradiction, and an account that gives one number has answered a question nobody asked. The first thing to settle is not what it cost, but whose money is being counted.

Anjani has exactly this shape. There are three sets of books in the picture. There is the open market, where a trade binderA binding workshop that takes in work from any customer that brings it, rather than being attached to one manufacturer. The workshop quotes a rate and competes on it. quotes a rate to anybody who walks in. There are Anjani's own accounts, and they record what Anjani actually paid. And there are the group accounts, covering Anjani and Chitra together and seeing straight through the invoice that passes between them. Each set of books gives a different figure for the same binding work, and each figure is the right one for somebody.

What does binding cost Anjani Stationers, and who is each answer right for?

The first answer is the market answer. A trade binder quotes Rs 4.00/- a register. Across the 2,50,000 registers Anjani made, that is Rs 10,00,000/-. The market answer is right for whoever is comparing quotes, meaning anyone testing whether the present arrangement is competitive, and it is the only one of the three that a person outside either company could ever obtain.

The second answer is what actually happened. Chitra invoiced Anjani Rs 8,00,000/- for binding in the year. Across 2,50,000 registers that is Rs 3.20/- a register. The invoice is the only binding figure that appears in Anjani's own statements, so it is the right one for anyone reading them. The invoice is also a transfer priceThe price one company inside a group charges another for goods or services. The group sets it rather than a market, so it can sit anywhere between cost and the outside rate., set between two companies under common control, and that matters a great deal further on.

The third answer is the group answer, and it is the one that surprises people. Chitra's own cost of doing Anjani's binding is Rs 5,00,000/-. On the Rs 8,00,000/- it invoiced, that leaves Chitra a margin of Rs 3,00,000/- on this work. Inside a group, the money behind a margin charged by one member to another has not left, so the margin is not by itself a cost to anybody. But Anjani holds only seventy per cent of Chitra. Thirty per cent of Chitra belongs to holders outside the group, and thirty per cent of that Rs 3,00,000/- margin is theirs, or Rs 90,000/-. The Rs 90,000/- genuinely leaves. So the work costs the group Rs 5,00,000/- of resources plus Rs 90,000/- of margin that walks out of the door, a total of Rs 5,90,000/-.

All three figures are right, and each one answers a different question, so the useful move is to settle which reader is being served before going looking for the number. The share belonging to those outside holders has a name in a set of group accounts, non-controlling interestThe part of a subsidiary that the parent does not hold. Group accounts show it separately because that slice of the subsidiary's assets and profit belongs to somebody else., and how it is presented there is a financial accounting matter rather than an operating one. Only the arithmetic consequence matters: control is not the same thing as holding all of it, and the gap between the two has a price.

Written as a ladder, the three answers are one figure walked down and then partly back up. From Rs 10,00,000/- at the trade rate, taking off the Rs 2,00,000/- that Anjani saves by not going to the market gives the Rs 8,00,000/- invoice. Taking off Chitra's Rs 3,00,000/- margin, money that only moves inside the group, gives the Rs 5,00,000/- of resources the work actually consumes. Adding back the Rs 90,000/- of that margin that belongs to the outside holders lands on Rs 5,90,000/-. The last step back up is short, and it is the one people forget.

One thing that ladder does not do is move anything else. Chitra's total profit for the year stays exactly where it was published, at Rs 10,00,000/-. Of that, Rs 3,00,000/- comes from Anjani's work and the other Rs 7,00,000/- comes from Chitra's own customers, who have nothing to do with registers and are not part of this decision. Splitting Chitra's profit between two sources describes it; it does not change it.

What shape does the make against buy decision actually have?

With the rupees stripped away for a moment, what remains is the shape, and the shape is the part that carries to any other step of the work. Doing binding inside means Chitra's binding department exists: a shed, a machine, people who are there whether or not a register arrives. The block of cost, Rs 3,00,000/- a year, is paid before a single register is bound. On top of it sits a per-register cost of Rs 0.80/-, the glue and board and thread and the time that one more register consumes.

Buying has no block at all. OutsourcingPaying an outside supplier to perform a step of the work that a business could otherwise do itself. The supplier carries the workshop and the payroll; the buyer pays a rate. the step means the trade binder carries the shed and the people, and Anjani pays Rs 4.00/- for each register bound and nothing whatever for the ones it does not bind. Buying converts a fixed cost into a rate, and that conversion is the whole trade.

Two consequences fall out of that sentence before any arithmetic is done, and a reader who holds them will predict the answer rather than wait for it. At low volumes, buying wins. The fixed block has almost nothing to spread itself across, and the buyer pays for nothing it does not use. At high volumes, making wins. The block is spread thinner and thinner while the per-register gap between Rs 0.80/- and Rs 4.00/- keeps accumulating on every single unit. Somewhere between those two there is a volume at which the two come to exactly the same thing, and that volume is the only genuinely new number this comparison produces.

Which costs is the crossing point comparing?

One error is worse than any other in this comparison, and it is worth slowing down for. Three costs for binding are now on the table, along with two descriptions of how binding behaves as volume moves. The Rs 8,00,000/- invoice is the figure in the accounts and feels the most solid of the three, so it is extremely tempting to reach for. Reaching for it here would be wrong.

The crossing point compares what the work costs in resources, Rs 3,00,000/- of fixed cost plus Rs 0.80/- a register, against what the outside rate costs, Rs 4.00/- a register. Both sides of that comparison are resources given up in exchange for binding getting done. The Rs 8,00,000/- invoice is not a resource cost at all. The invoice is a price agreed between two companies under one holding, and a group can set it anywhere it likes between the cost of the work and the rate the market would charge. Change the invoice tomorrow to Rs 9,00,000/- and not one thing about whether binding should be done inside has changed. No shed has moved and no machine has run any faster.

A comparison has to be made on one basis all the way through, and the invoice is a different basis from the resource cost. The Rs 5,90,000/- group figure is a third basis again, useful for stating what the arrangement costs the group as it stands but not for asking whether the arrangement should exist. Mixing the three produces a number that belongs to nobody. The basis has to be named and then held to.

Where do the two lines actually cross?

Buying costs Rs 4.00/- for every register. Doing it inside costs Rs 0.80/- for every register. So each register bound inside rather than bought saves Rs 3.20/-, and that saving has one job: paying off the Rs 3,00,000/- block. Rs 3,00,000/- divided by Rs 3.20/- is 93,750 registers. Below that, buying is cheaper. Above it, making is cheaper. At exactly 93,750 registers both cost Rs 3,75,000/-, and there is nothing to choose between them.

Now the warning that has to travel with that arithmetic: the Rs 3.20/- in the division above is Rs 4.00/- less Rs 0.80/-, and it is not the Rs 3.20/- a register that Chitra invoices. Two entirely different quantities here happen to be the same size, and the coincidence is a property of this case and of nothing else. One is a price charged by one company to another. The other is the per-register distance between two ways of getting a job done and is not a price at all. At a trade rate of Rs 4.10/- the second becomes Rs 3.30/- while the first stays at Rs 3.20/-, and the difference tells the two apart.

There is a second coincidence of the same kind sitting nearby, and it is worth naming now rather than letting a reader discover it as a puzzle. Rs 0.80/- is Chitra's variable cost of binding one register. Rs 0.80/- is also, quite separately, the amount by which making beats buying on Anjani's own books: the Rs 2,00,000/- gap between the Rs 10,00,000/- trade cost and the Rs 8,00,000/- invoice, spread across 2,50,000 registers, is Rs 0.80/- a register. Neither pair is a rule and both are coincidences of these particular figures. Where two quantities in one argument share a value, the coincidence has to be named. A reader who assumes they are the same thing will eventually put one where the other belongs.

What happens when the volume moves?

Volume is the one thing worth moving. Everything else is held still: the Rs 3,00,000/- block, the Rs 0.80/- a register inside, the Rs 4.00/- a register outside. Two things there are worth attention. The shaded gap flips from one side to the other as the marker passes 93,750, and that flip is the crossing doing its work in the open. And the two ends behave oddly: at nil volume one option costs nothing at all while the other still costs Rs 3,00,000/-, and above 4,00,000 registers the inside line stops describing anything real.

Does the crossing point settle it, and does it matter?

A crossing point becomes useful only once it is clear which side of it the business lives on. Placed next to the other volumes that describe Anjani, 93,750 registers stops being a decision and turns into a footnote.

Anjani's published break even is Rs 1,72,98,701/- of revenue. At Rs 108.00/- a register that is 1,60,173 registers, and below that the business does not cover its own fixed costs at all. The crossing at 93,750 registers sits far below that, at not much more than half of it. And Anjani made 2,50,000 registers, comfortably above both.

A business that cannot survive below 1,60,173 registers never operates below 93,750, so at every volume where Anjani is a going concern, binding inside is right, and right by a wide margin. At the 2,50,000 registers it actually ran, the inside route costs Rs 5,00,000/- of resources against Rs 10,00,000/- at the trade rate. The inside route is not narrowly cheaper: it is half the price.

Now say the thing most treatments of this arithmetic will not say. Knowing a decision is not close is worth as much as the arithmetic that showed it. Squeezing a close call out of a settled one teaches a reader to distrust their own numbers, and it wastes review time on the wrong question every year. The useful output here is not the 93,750 figure by itself. The useful output is the sentence that follows it: binding inside is settled at any volume the business will ever see, and the review time it absorbs belongs to something that is actually in play.

Where is the line between made and bought drawn?

There is a quieter question underneath everything above, and a careful reader will already have tripped over it. Binding is done by a separate legal company, so binding is bought in. Yet Anjani's published split of its own costs treats only paper and carriage as varying with volume. Where, then, does the Rs 3.20/- a register of binding sit?

Binding sits inside the works cost. Anjani's Rs 29.60/- a register of works cost splits into Rs 3.20/- of binding and Rs 26.40/- of everything else. The placement is worth stating plainly because the alternative reading looks broken: paper at Rs 59.40/- plus binding at Rs 3.20/- plus carriage at Rs 2.40/- comes to Rs 65.00/-, more than the Rs 61.80/- of variable cost a register, and a reader who adds those three and compares will conclude something is wrong. Nothing is wrong. Paper and carriage are the only bought-in inputs that move with volume, so they alone make up the Rs 61.80/-, and binding sits above the line with the works.

Read a different way, the line falls in a different place. Ask what value Anjani itself adds, and binding is bought in from another company and comes out with the paper and the carriage: Rs 108.00/- less Rs 59.40/- less Rs 2.40/- less Rs 3.20/- is Rs 43.00/- a register. Ask how costs behave when volume moves, and binding does not move with volume, so it stays inside and the figure is Rs 46.20/-. Both readings are right, they answer different questions, and any use of either figure must name which line was drawn.

The Rs 16.60/- of operating profit a register at the bottom does not move, and it is the same under either reading, so the invariance is the point rather than a problem to hide. Rs 43.00/- less Rs 26.40/- is Rs 16.60/-. Rs 46.20/- less Rs 29.60/- is Rs 16.60/-. Where the line is drawn changes the name of the middle and changes nothing at all at the bottom.

Keeping binding inside the works cost has a direct consequence. Because binding does not vary with volume in Anjani's own cost behaviour, the Rs 8,00,000/- is not something Anjani can switch off by making fewer registers this month. The arrangement is a standing one with a shed behind it, and that is one more reason the base of the year is not the part of the decision genuinely in play.

How does an operator actually run this decision through a year?

A lender, an analyst and the person running the works all want the same thing from this arithmetic, and none of them wants a yes or a no. All three want the split. Anjani's own binding is rated for 4,00,000 registers a year, the most its slowest stage can produce across two lines, eight hours and 250 working days. Below that, binding can be done inside. Above it, it cannot, whatever anybody would prefer.

The question is almost never which one, it is how much of each. Make the base and buy the peak. The base is the volume that runs steadily and comfortably inside the shed, where the Rs 3,00,000/- block is already paid and every register bound inside saves against the trade rate. The peak is whatever arrives on top of that. Above 4,00,000 registers the choice disappears entirely: the inside route has nothing left to give, and every further register must be bought at whatever rate the market will quote for it. A full stage is a constraint in the sense Goldratt described in 1984: it sets the pace of the whole works, and no amount of spare cutting or printing capacity changes it.

An analyst reading a manufacturer's accounts can look for the same shape without any of these figures. Find the step that is bought in, find whether the business also does that step itself, and ask what the rated capacityThe most a plant or a stage can produce in a period when it runs at its designed rate for its full working hours. The ceiling is set by machinery and hours, not by demand. of the inside route is against the volume the business actually runs. A business buying in a step it has idle capacity for is either paying for a convenience it has not named, or has a capacity figure that does not survive contact with a real week. A business making a step it has no capacity for is quietly buying the difference and may not have priced it.

What does the trade binder's quote quietly assume?

Everything above rests on a rate of Rs 4.00/- a register, and it is time to ask what kind of object that rate is. The Rs 4.00/- is a quote for the whole year, taken as a steady run. A whole year is the only basis on which a binder can quote at all. A binder has its own shed and its own people, and has to know roughly how much work will arrive in each month to plan around it. A whole-year rate is a promise about a smooth flow in both directions.

Anjani's year is not smooth. Its seasonalityThe tendency of a business to earn or produce unevenly across the year, with the same months busy and the same months quiet, for reasons outside its control. is published in its own accounts, not as a chart of monthly volumes but in the shape of its borrowing: a cash credit facilityA running bank borrowing a business can draw on and repay as it needs, usually secured on stock and receivables. The facility is the standard way an Indian manufacturer funds a busy season. drawn through the school-supply season and cleared before the year end. A business that draws and clears like that is a business whose work arrives in months rather than weeks apart.

Anjani Stationers Private Limited, Chitra Binding Works Private Limited and the trade binder quoting Rs 4.00/- a register are invented.
Educational material. Not advice on any investment, tax, budget or market position.