The Supply Chain: Dependency, Cost and Fragility

A supply chain is the ordered list of outside parties whose work has to happen before one unit reaches one buyer. Anjani Stationers takes paper from a merchant, sends registers to Chitra Binding Works to be bound, and pays a carrier to move them out. Every link costs money, every link can stop, and a pile of stock buys days against a link that stops.

The making of one register, and the coming apart of its price, are covered separately. The earlier working followed Rs 108.00/- from the buyer's hand down through the paper, the works and the carriage until Rs 16.60/- was left, and it reconciled to the paisa. The same sequence runs here, turned round. Instead of asking where the money went, it asks who had to do something for that money to be earned, and what happens on the morning one of them does not.

Two other established facts are doing quiet work here. The first is the published paper account. Unusually, it gives a physical quantity beside a rupee amount: 71,000 reams consumed in the year, 14,000 reams left at the year end, and Rs 28,00,000/- of value sitting on them. A chain is measured in days, and only a physical quantity turns into days. The second is the published fact that Anjani Stationers holds 70 per cent of Chitra Binding Works Private Limited, bought at the start of year two for Rs 21,00,000/-. Holding a supplier changes who keeps the margin on what it does, and it does not remove the link from the chain.

What is a supply chain, and who is actually in one?

Start on a street corner rather than in a works. A tea stall outside one office building looks like a business that depends on nothing. The stall has no contracts, no procurement department and no lorries of its own. Now list the mornings it could fail to open. The dairy could not deliver. The gas cylinder agency could not swap the empty one. The wholesaler could be out of the small clay cups. Three outside parties, three separate ways to be shut for a morning, and the stall holder controls none of them. The three of them are the stall's supply chain, and the list is complete.

The test turned on something particular. The test was not who gets paid. The stall pays a rent, and a landlord on holiday does not stop it opening. The test was not who delivers physical goods either. A licence renewal is not a thing anybody can hold. A supply chain is every outside party whose failure stops the work, and nothing else. Everybody else is a supplier of convenience: useful, paid, replaceable by tomorrow morning without a day being lost.

The test is stricter than it sounds, and it cuts in both directions. The test puts parties into the chain that nobody thinks of as suppliers, such as the single transformer that feeds a shed. The test also throws out parties that get paid a great deal and are still not links: the work would carry on without them for as long as it took to find somebody else. Mapping a chain is not making a list of vendors. Mapping a chain is making a list of ways to be stopped.

Who does Anjani Stationers depend on, link by link?

Anjani Stationers Private Limited is an invented business, and its year two figures are the ones worked on throughout. The business makes one lead product, a hard bound ruled register, and in year two it made and sold 2,50,000 of them at Rs 108.00/- each. Running the test on it brings out three outside parties, in this order.

Paper comes in from a paper merchant. Nothing published names the merchant, so it stays the paper merchant throughout, and the important fact about it is not its name but that it supplies the one input the works cannot substitute in an afternoon. Binding is done by Chitra Binding Works Private Limited, also invented. Anjani Stationers holds 70 per cent of it and bought that holding at the start of year two for Rs 21,00,000/-. Finished registers go out through a carrier, again unnamed, against a carriage and packingThe named cost of getting a finished thing boxed and moved to whoever is buying it. In a cost statement it sits on its own line rather than being folded into the price of the goods. line in the accounts.

Now look hard at the middle one. The binding link is the one most readers quietly delete. Chitra Binding Works is a subsidiaryA company that another company holds enough of to control. It keeps its own name, its own staff and its own set of accounts, and it stays a separate legal person from the company holding it., and a reader who knows that will often treat the binding as though it were happening inside Anjani Stationers, in which case it stops being a dependency at all. It is not. Holding a link does not remove it: if Chitra Binding Works stops binding, Anjani Stationers stops shipping, and the share certificate does not bind a single register. The 70 per cent is a claim on what Chitra Binding Works earns. The holding is not a substitute for the work, and it is not a promise that the work happens.

There is a second thing the 70 per cent does not do, and it is worth naming. The other 30 per cent of Chitra Binding Works belongs to holders outside the group, a non-controlling interestThe part of a controlled company held by somebody other than the company controlling it. Those holders keep their share of what it earns, and how that share is presented in a combined set of accounts is a financial accounting subject., and they keep their share of what it earns. The amount the group as a whole really carries for binding is therefore not the same as the amount Anjani Stationers is invoiced. Working that out properly asks whether a step is worth holding at all, and it is taken up under Make vs Buy: Whether a Step Is Worth Owning at All.

What does each link cost, and why do the three not add up?

Put a rupee on every link and the chain stops being a diagram. Against the Rs 108.00/- that comes in for one register, paper costs Rs 59.40/-, binding costs Rs 3.20/- and carriage and packing costs Rs 2.40/-. The binding figure is not a new number: Chitra Binding Works invoiced Anjani Stationers Rs 8,00,000/- for binding in year two, and Rs 8,00,000/- spread across 2,50,000 registers is exactly Rs 3.20/- each.

The paper figure is the one to sit with. The paper merchant takes 55.00 per cent of every rupee the buyer pays, and takes it at the very first link, before a single register exists. More than half the price of the finished thing has already left the business by the time the sheets are stacked in the yard. The share is not a fault, and it is not unusual in a converting business; it is simply what it means to buy a lot of one material and add work to it. But it shows immediately where a chain like this is fragile, and where a rupee saved is worth chasing.

Now the part that stops most readers cold, and it is worth slowing down for. On the face of it the working has just contradicted itself. The three link costs add up: Rs 59.40/- plus Rs 3.20/- plus Rs 2.40/- is Rs 65.00/-. But the variable cost of one register at Anjani Stationers is Rs 61.80/-, a smaller figure. Three of the numbers are outside parties being paid, one of them is a published cost per register, and Rs 65.00/- is bigger than Rs 61.80/-. Something looks broken.

Nothing is broken, and the reason is the most useful thing in this section. Where a cost sits in the accounts is decided by whether it moves with volume, not by whether it is paid to an outside party. Anjani Stationers' only bought in costs that move with volume are paper and carriage, so Rs 59.40/- and Rs 2.40/- together are the whole of the Rs 61.80/- of variable cost a register. The Rs 3.20/- of binding is not in there at all. The binding sits inside the Rs 29.60/- of works costThe cost of running the place where the making happens: the shed, the machines, the people on them, and anything bought in to keep them running. It is a cost accounting label rather than a line any buyer ever sees. a register, and the Rs 29.60/- splits into Rs 3.20/- of binding and Rs 26.40/- of everything else. Every one of the four figures is right, and they were never all in the same place.

Two honest warnings go with that paragraph. The first is that sorting costs into the ones that rise with the count of registers and the ones that stay put is a judgement made while reading the accounts, and never something a statement discloses on its face. Anybody quoting the Rs 61.80/- and the Rs 29.60/- is quoting an assumption, and ought to say so out loud. The second is that Rs 65.00/- is not a wrong number, it is a number that answers a different question. Rs 65.00/- is what three outside parties are paid for one register. The total is simply not a line in any statement, and treating it as one is what produces the apparent contradiction.

How many working days of production does the paper on the floor cover?

Here is where a chain stops being a picture and becomes a duration. Anjani Stationers consumed 71,000 reamsA standard bundle of paper, five hundred sheets of it. Paper is bought, counted and stored in reams, so a pile of paper is a count of reams long before it is an amount of money. of paper across 250 working days. Divide and the works draws 284 reams every working day it runs, and turns out 1,000 registers a working day from them. At the year end 14,000 reams were left. Divide again: 14,000 reams at 284 reams a day is 49.30 working days.

Read that number carefully. The 49.30 is not an amount of anything but a length of time. Cover answers one question and one question only: how long the making can continue if nothing arrives. A pile of 14,000 reams is not a business that is richer by 14,000 reams. A pile of 14,000 reams is a business that is 49.30 working days harder to interrupt than it would be with an empty yard.

Every household in the country understands this without being taught it. A spare gas cylinder behind the door does not make a household wealthier by one cylinder. The spare cylinder makes that household three weeks harder to stop cooking. Nobody who keeps that spare cylinder thinks of it as an asset; they think of it as three weeks. The instinct is exactly right, and a stock of paper is doing the same thing in a works.

Why does one pile of paper give two different day counts?

Because there are two rulers, both of them in common use, and they measure different things. The pile that came to 49.30 working days above comes to 68.82 days when it is measured the other way, and the other way runs as follows. The closing paper stock is carried at Rs 28,00,000/-. Materials consumedThe accounts line for the raw material actually used up during a period, as against the material bought during it. The two differ by whatever the stock pile did over the period. in the year came to Rs 1,48,50,000/-. Rs 28,00,000/- against Rs 1,48,50,000/-, spread across 365 calendar days, reads 68.82 days.

One ruler divides rupees by rupees across every day of the year, and the other divides reams by reams across the days the machines run, so they are answers to two different questions. Both are correctly computed. Neither is a mistake. The two rulers differ in both halves of the fraction: one counts money and one counts sheets, one runs over a calendar and one runs over a shift roster.

Give each of them its proper name. The names are what stop the two being swapped. 68.82 days is the accounting cover. The accounting cover belongs with the accounts, it is comparable across years and across businesses, and it is the figure that turns up in a working capital note. 49.30 working days is the operating cover, and it is the only one of the two that answers how long production can actually go on. The gap between them is more than nineteen working days of production, and nineteen working days is neither a rounding difference nor a detail.

What is Just-in-Time, and what does it need to work?

Just-in-Time is an arrangement in which an input arrives close to the moment it is used. Cover is deliberately short, very little money sits on the floor as stock, and the yard at the start of any given week holds not much more than that week needs. Applied to Anjani Stationers it would mean a paper pile measured in a handful of working days rather than in the high forties.

Just-in-Time has requirements, and they are requirements rather than preferences. Just-in-Time needs a supplier that delivers reliably and a short lead timeThe waiting a supplier imposes: an order placed today is filled only after however many days that supplier and that route need. A buyer can bargain over it but never simply decide it., and where either of those is missing the arrangement does not work as described. A merchant who is dependable but takes six weeks to ship cannot supply a business running on a week of cover, and neither can a merchant next door who is unreliable. Both conditions have to hold at once. The definition stops there. Nothing in it makes Just-in-Time better, cheaper, more modern or more correct than a long cover; it makes it one setting of the same quantity.

What is Just-in-Case Inventory, and is it a different system?

Just-in-Case inventory is an arrangement in which cover is deliberately long enough to outlast an interruption the business expects. Because the flow comes off the floor rather than off a lorry, the input keeps reaching the machines through a stretch when the supplier is not supplying anything at all. Anjani Stationers at 49.30 working days of cover is plainly running this way, whether anybody in the business ever used the phrase or not.

Now hold the two definitions next to each other and look at what actually differs. Not the machines, not the accounting, not the industry, not the era. Just-in-Time and Just-in-Case are two settings of one dial and not two different systems, and the dial reads in days of cover. Turned down, less money sits on the floor and less time stands between the works and a stall. Turned up, both of those reverse. Cover is one quantity, held short or held long.

How far to turn that dial for a given business is a separate exercise with its own arithmetic, and it is worked through under Just-in-Time vs Just-in-Case Inventory: How Much Cover to Hold.

How Supply-Chain Disruption Affects a Business: what does a stall actually do?

Almost nobody predicts the shape of this correctly, including people who have run works for years, and the reason is that the shape is unlike anything else in a set of accounts. Ask what a supplier stalling costs and the instinct is to reach for something proportional: half as long a stall, half as much damage. The damage is not proportional.

Nothing visible happens at all, for a long time, and then output stops dead. A stall costs nothing until it outlasts the cover, and from that moment it costs a whole production day for every further day. While the buffer is being eaten, the machines run at their normal rate, the registers come out at their normal count, the statements look exactly as they always looked, and the only thing changing is a pile in the yard that nobody outside the works can see. Then the pile ends, and the next morning the count is zero.

Put the published posture under it. Anjani Stationers finished the year with 49.30 working days of cover. Let the paper merchant stop for 21 working days, a month of trading days and a serious event by any reading. The cost, in production, is nothing at all. Not one register is lost. 21 sits comfortably inside 49.30, and the pile absorbs the whole interruption. After those 21 days the works has drawn 5,964 reams and still has 8,036 reams in the yard, another 28.30 working days of cover.

Now take the sentence that people draw from an experience like that. The real damage lives in it. Having survived a 21 day stall untouched, a works can very reasonably conclude that stalls of that kind do not hurt it. The conclusion is false and the evidence is genuine, the worst combination there is. The absence of damage during a stall is not evidence that the stall was harmless, only evidence that the buffer had not run out yet. A chain can look completely healthy right up to the day it is not, and it will look healthiest on exactly the days it is being quietly eaten.

Why can paper be stockpiled and binding cannot?

Look back at the chain drawing and count the buffers. The paper link has one, 14,000 reams deep. The carriage link has something like one: finished registers can wait on a floor for a few days without anybody minding much. The binding link has nothing at all, and no amount of money or space could give it one.

The reason is a difference in what is being bought rather than a difference in how well anybody is managing. A ream is a thing. A ream can be stacked in a corner, and next March it is still a ream. An hour of binding is not a thing, it is work being done at a moment. A service cannot be stockpiled, so a service link carries no cover at all. If Chitra Binding Works does not bind on Tuesday, Tuesday's binding is gone, and no amount of Wednesday brings Tuesday back.

The claim is about how the chain is shaped rather than about how busy the works happens to be, and the two get confused constantly. Machine rates, rated capacity and how full the works runs are a separate matter; how much of the works is in use, and how that is measured, is taken up under Capacity Utilisation: How to Compute It and What It Hides. The claim here is narrower and more structural: the paper link and the binding link fail in different ways, and only one of the two can be defended by holding stock.

What goes wrong when the wrong cover figure is quoted?

The mistake below is not made by a careless person. A competent planner makes it, reading a real number correctly off a real statement, and that is precisely why it survives review meetings.

How does anyone map a chain they have never seen before?

Three questions, asked in order, and the map is drawn. A lender doing a site visit is asking those three while walking round the yard, an analyst is asking them when a filing mentions a single supplier, and a household is asking them without calling them anything when it decides whether to keep a spare cylinder.

What is settled elsewhere