Operating vs Non-Operating Asset: In the Model or the Bridge

An asset is operating when the cash it throws off is already inside the forecast, and non-operating when it is not. On Sankalp Industrial Systems Limited, invented, a surplus land parcel at Rs 45,00,00,000 and a 26.0 per cent associate holding at Rs 55,00,00,000 fed none of the Year 0 earnings before interest, tax, depreciation and amortisation (EBITDA) of Rs 2,88,00,00,000 the model starts from, so both are valued on their own and added at the bridge.

The answer rests on a test almost everybody states the wrong way round. Readers ask whether the asset feels like part of the business. One question actually settles it instead: did the forecast already count the asset's cash? The two questions agree in every easy case and part company in every hard one. The hard ones are exactly where the rupees a share are won and lost. With the test facing the right way, the sorting stops being a matter of taste about the asset and becomes a matter of fact about the model that has been built.

What was the forecast actually built on?

Before anything can be sorted, the yardstick the sorting is measured against has to be established. For Sankalp Industrial Systems Limited, Year 0 revenue is Rs 12,00,00,00,000 and EBITDATrading profit measured before depreciation, interest and tax have been taken out of it. is a flat 24.0 per cent of it, being Rs 2,88,00,00,000. Depreciation runs at 4.0 per cent of revenue, Rs 48,00,00,000, so earnings before interest and tax (EBIT) is Rs 2,40,00,00,000 and 20.0 per cent of revenue. The assumed effective tax rate is 25.0 per cent, an assumption rather than a rate read off any statute, so net operating profit after tax (NOPAT) is Rs 1,80,00,00,000, being 15.00 per cent of revenue.

The figures above are the whole reference point. Every asset on this company gets measured against one number, the Rs 2,88,00,00,000 of Year 0 EBITDA, and the only question asked of it is whether it helped make that number. Nothing else about the asset matters for the sorting: not its size, not how long it has been held, not whether somebody in the building thinks of it as core.

The shape is identical in a household, where the money is smaller. A couple in a two-income household lives in a flat they bought with a loan, and they also hold a small plot in their home town that they have never visited and never let out. Valuing that household by capitalising the salaries values the jobs, the flat they need to live near the jobs, and nothing else. The plot contributed nothing to either salary. The plot is still theirs. A valuation built off the salaries alone, with the plot forgotten, quietly tells the couple they are poorer than they are.

Side one: what does an operating asset look like here?

An operating asset is one whose cash the forecast has already picked up. On this company that is the plant, the working capital and everything else that turned into the Rs 2,88,00,00,000. Each of them can be checked by pointing at the line in the model that captures it. Pointing at a line is a far more satisfying test than arguing about whether something is core.

Net working capitalMoney tied up while goods sit in stock and customers take their time to pay, net of what the business has not yet paid its own suppliers. stands at Rs 1,80,00,00,000 and holds flat at 15.0 per cent of revenue right through the forecast. Three components build it, and the composition matters again when cash is sorted:

Point to the line: revenue rises by Rs 1,20,00,00,000 a year and fifteen per cent of that has to be funded, so the model deducts Rs 18,00,00,000 of working capital movement from cash in every single year. The deduction is the working capital being counted. Anything the model already deducts for, or already collects from, is operating by definition and must never appear again at the bridge.

Net fixed assets stand at Rs 10,20,00,00,000. Point to the line again: the forecast carries capital expenditure of Rs 1,34,80,00,000 in the first forecast year rising to Rs 1,54,00,00,000 in the fifth. The spending is what it costs to keep those assets running and to extend them as revenue grows. The plant is being paid for inside the model, year after year, and the cash it produces is inside the model too. Adding Rs 10,20,00,00,000 of plant at the bridge on top of that would be asking to be paid twice for the same factory.

Notice what has not been used to decide any of this. Not the age of the plant, not whether it is written down close to nothing, not whether the finance team calls it strategic. The only evidence that mattered was a line in the model.

Side two: what does a non-operating asset look like here?

A non-operating asset is one the forecast never touched. There are exactly two on this company. Each one fails the test for a completely different reason, and both reasons are worth spelling out.

The surplus land is a parcel at Rs 45,00,00,000. The company holds it, does not use it, does not let it out and earns nothing from it. Trace any rupee of the Rs 2,88,00,00,000 back and none of it comes from that parcel. The parcel fails the test because it does nothing. There is a further detail worth carrying: the parcel sits in the accounts at its historical cost of Rs 12,00,00,000, and the Rs 45,00,00,000 is what it is judged to be worth now, a mark of Rs 33,00,00,000 over the book figure. The bridge is asking what a buyer of the company would pay rather than what the parcel once cost, so the bridge uses the current value.

The 26.0 per cent holding in Aruna Tooling Private Limited is Rs 55,00,00,000, and it fails the test for the opposite reason. Aruna Tooling is emphatically doing something. The company runs a real tooling operation, it sells, it employs, it makes a profit. The reason it is non-operating here has nothing to do with the tooling business and everything to do with how Sankalp's accounts carry it. The stake is equity accountedA stake carried as one line that moves with the investee's profit, so its turnover and its costs never reach the group's own trading lines at all.. None of Aruna Tooling's revenue and none of its EBITDA reaches the Rs 2,88,00,00,000 the forecast was built from. The forecast literally could not have counted it. Its carrying valueThe figure an asset is shown at in the accounts, which records what happened rather than estimating what the thing would fetch now. in the accounts is Rs 22,00,00,000, and the bridge again uses the current judgement of Rs 55,00,00,000.

Together the two lines come to Rs 1,00,00,00,000, and they enter the bridge as a single signed addition at their own value rather than as anything the model produces. One is idle and one is busy. The test asks about the forecast and not about the asset, so the test does not care which.

So what is the test, now that both sides are on the table?

With both sides described, the contrast falls out in a single sentence. The test is one question with one subject, and the subject is the model rather than the thing. Did this asset put cash into the Rs 2,88,00,00,000 of Year 0 EBITDA that the forecast starts from? If yes, it is operating, it is already valued inside the discounted cash flow, and it is never added at the bridge. If no, it is non-operating, it is valued on its own terms, and it arrives at the bridge as a signed line.

Two features of that question earn their place. First, it names a specific number. A specific number makes the test checkable rather than debatable. Second, it points at the forecast rather than at the asset, so the answer does not change when somebody in the room has a strong feeling about what is core to the business. The classification is a fact about the model, and the model is a document anyone can read.

Why does the wrong version of the test still feel right?

If the wrong version of the test were obviously wrong, nobody would use it. The trouble is that asking whether an asset is part of the business gives the right answer most of the time. The plant is part of the business and it is operating. The idle parcel is not part of the business and it is not operating. Two for two, and a habit forms.

Then the habit meets the associate holding and produces a confident wrong answer. A tooling company is unmistakably a business, so the habit says operating, so the analyst leaves Rs 55,00,00,000 out of the bridge and reports an equity value Rs 2.75 a share light. The wrong test does not fail loudly; it fails on precisely the cases where the money is, and it fails while sounding reasonable. A rule of thumb can have no worse failure mode. Nothing in the output flags it.

What happens to a non-operating asset once it has been sorted?

Sorting an asset out of the model is only half a job. The other half is putting it back somewhere, and the somewhere is the bridge that runs from enterprise valueThe price tag on a whole trading operation, arrived at before the funding question comes up, so lenders and shareholders both have a claim inside it. to equity value. The non-operating asset is valued on its own terms, by whatever method suits it, and enters as a signed line at that value.

The phrase on its own terms is carrying weight. The parcel of land is worth what a parcel of land is worth, and pricing land is a property question rather than a discounted cash flow question. The associate holding is worth what a 26.0 per cent stake in a tooling business is worth, and pricing that stake is a valuation exercise of its own. Neither belongs in a model whose engine is Sankalp's own EBITDA and Sankalp's own weighted average cost of capitalWhat the whole funding mix costs, blending the return shareholders look for with the rate lenders charge, scaled to how much of each the company carries. of 12.00 per cent. Neither generates Sankalp's EBITDA and neither carries Sankalp's risk. The Rs 45,00,00,000 and the Rs 55,00,00,000 are taken as given here; valuing either one is a separate exercise.

What if it is operating but a slice of it belongs to somebody else?

Sankalp Coatings Private Limited is held at 75.0 per cent and is fully consolidatedEvery rupee of a subsidiary's revenue and cost added into the group accounts as though the whole of it belonged to the group.. Under full consolidation the whole of the coatings business shows up in the group's revenue and cost lines, so one hundred per cent of its EBITDA is inside the Rs 2,88,00,00,000 and one hundred per cent of its cash flow is inside the forecast. Apply the test and the answer comes back operating, without hesitation.

Which leaves an obvious problem. The model has valued cash flows the group only has a three quarter claim on. The repair is not to strip the extra quarter out of the operating figure. Stripping it would mean rebuilding the whole forecast on partial subsidiaries and would make a mess of every margin in the model. The repair is to leave the operating figure alone and take the outside claim out at the bridge as minority interestOutside shareholders' stake in a subsidiary whose numbers were added in full, carried through the accounts as a claim of its own. of MINUS Rs 60,00,00,000. The asset stays in the model and the claim against it leaves at the bridge. The move is different from anything the two-way sort does. How that Rs 60,00,00,000 is itself measured, and whether it is a book figure or a current one, is a separate question, settled elsewhere.

Why is the associate holding the clearest case in the whole subject?

Put the subsidiary and the associate side by side and something useful happens. Both are stakes in industrial businesses. Both were bought with the group's money. Both are shown somewhere on the same balance sheet. And a difference that has nothing to do with either business gives them opposite treatments at the bridge, one negative and one positive.

The subsidiary is consolidated, so all of its cash entered the model and a quarter of that value has to leave at the bridge. The associate is equity accounted, so none of its cash entered the model and all of its value has to arrive at the bridge. Same kind of asset, opposite direction, and the accounting treatment is the only thing that decided it. The pair is the sharpest possible demonstration that the sorting is not a judgement about the asset. The accounting treatment did all of the work.

There is a small warning attached. Because the treatment decides the direction, a change of treatment changes the direction, and treatments do change when a shareholding moves across a threshold or when control arrangements are rewritten. The thresholds themselves, and the current accounting text, are settled by the authorities named below.

Where does cash sit, and is all of it the same?

Cash is the item that will not sit still. Cash produces no trading profit, so cash fails the test and belongs at the bridge. Yet part of a cash balance is genuinely needed to run a business, and calling all of it non-operating therefore feels wrong. Sankalp holds Rs 1,20,00,00,000, of which Rs 40,00,00,000 is judged to be needed to run the business and Rs 80,00,00,000 is surplus to that.

The bridge here adds the whole Rs 1,20,00,00,000, and the way the forecast is assembled forces that rather than any taste for tidiness. Look again at the three components that build net working capital here. Money customers still owe, stock sitting on the floor, and money the company still owes its own suppliers. A cash line is absent from all three, and the absence is a consequence of defining working capital that way rather than an accident. Because the model holds no cash anywhere, picking up the entire balance at the bridge duplicates precisely nothing, and that is what makes the whole balance convention defensible here.

A second convention is in live use and it is not a lesser one. Where the Rs 40,00,00,000 gets folded into the working business, only the surplus is picked up, so the addition is Rs 80,00,00,000 and the finish is Rs 16,48,13,79,094 of equity value, being Rs 82.41 on each share against the Rs 84.41 the whole-balance convention gives. The gap is Rs 40,00,00,000, or Rs 2.00 on each of 20,00,00,000 shares. Both conventions are used in practice, and neither ranks above the other; what actually matters is that a reader is told which one is in front of them.

What is the sorting actually worth on this company?

Abstract distinctions are easy to nod along to and hard to remember. A rupee figure fixes that. The model produces an enterprise value of Rs 21,28,13,79,094 for the trading business, and the bridge then runs five signed lines to reach what a share is worth.

Now take line 3 away and run the same five lines. Equity value comes out at Rs 15,88,13,79,094 and Rs 79.41 a share. The sorting of two balance sheet lines is worth exactly Rs 5.00 a share, being Rs 2.25 for the parcel and Rs 2.75 for the stake, and that is 5.92 per cent of the answer. Both figures are printed to the rupee because the record holds them to the rupee. On an answer weighted this heavily towards a terminal value, nobody should read meaning into the final digits, and that caution is worth repeating every time such a figure appears.

Two equity values now stand side by side for the same company, and they must never be swapped for one another. Rs 16,88,13,79,094 is the equity value of the company. Rs 15,88,13,79,094 is the equity value of the trading business on its own. The narrower subject makes it genuinely useful, and it is the figure a cash flow route that also excludes the parcel and the stake should be compared against. The two values differ by exactly the Rs 1,00,00,00,000 that was sorted, and the difference is one of subject rather than of method.

What single check catches both errors, in about a minute?

Both errors are invisible from inside the model that made them, so the check has to put the model and the bridge in the same field of view. Every asset on the balance sheet goes down a sheet, each one marked either in the model or at the bridge, and the marks are then read rather than the amounts.

Every row must carry exactly one mark: two marks is a double count and no marks means the asset has vanished. The whole power of the check is in that constraint. The constraint converts an argument about judgement into a clerical inspection that anybody can perform on somebody else's model.

One refinement matters. The list is a list of assets, and a claim is not an asset. Minority interest is a claim against assets that are already in the model, so it does not get a row and does not get a mark. Confusing the two produces a row that seems to break the rule and sends people looking for an error that is not there.

Who decides all this, and does anybody publish it?

Here is the uncomfortable part. Several of the inputs the sorting depends on are judgements rather than disclosures, and they move the answer.

Some things can be read straight off a set of accounts. The cash total, the gross debt, the size of a shareholding, whether a holding is consolidated or equity accounted, the historical cost the land sits at. All of those are reported and can be checked. Other things nobody publishes. Which parcel of land the company is not using. How much of a cash balance the business actually needs to run. The price the parcel would fetch today. The price the stake would fetch today. On Sankalp the Rs 45,00,00,000, the Rs 55,00,00,000, the Rs 40,00,00,000 and the Rs 80,00,00,000 are fixed by assumption; for any real business every one of those four would be an estimate somebody defended in a meeting.

The practical consequence is that two careful analysts working from the same published accounts can reach different equity values without either of them making a mistake. The spread is not a flaw. An honest method looks exactly like this when several of its inputs were never published by anybody in the first place. The method demands only that a valuation state its judgements rather than bury them, so a reader can disagree with the input instead of arguing with the answer.

How does any of this get used in a room?

An equity analyst uses the sorting twice on the same model. Once on the way in, to keep the forecast clean. A forecast that quietly carries associate dividends or rental income from an idle parcel is no longer a forecast of the trading business, and its margins stop meaning anything. Once on the way out, at the bridge, to put back what was correctly taken out. The clean version of the discipline is that the two passes have to agree with each other, and the two column list is how anybody proves that they do.

A lender reads the same balance sheet with a different question and reaches a different emphasis. A lender is asked whether the loan gets repaid, and repayment comes from trading cash first and from asset sales second. So the parcel at Rs 45,00,00,000 that is worth nothing to the forecast can be worth a great deal to the security package. The associate holding at Rs 55,00,00,000 may be far harder to realise, and selling a minority stake in somebody else's tooling business is not the same exercise as selling a parcel of land. The sorting that matters to a valuation and the sorting that matters to a lender are two different sorts, run on the same list.

An investor reading somebody else's published valuation has one habit worth building: finding the bridge, then counting its lines. Where a note quotes an equity value and never shows a bridge, there is no way to tell whether the non-operating assets were added, and on this company that ambiguity is Rs 5.00 on a Rs 84.41 figure. The question to ask of any valuation is not whether the analyst knows what a non-operating asset is, but whether the write-up shows where each one ended up.

A sorting has no dial to turn. An asset is in the model or it is at the bridge, and the cases that are hard were settled by an accounting treatment rather than by a quantity. The one quantity anybody could move, the value of the two sorted lines, runs through to the answer in a straight line the arithmetic already states exactly.