How to Analyse Capex Intensity and Asset Quality

Analysing capital spend is a procedure, not a verdict. Six steps in order: establish the basis, rebuild the asset schedule, separate what maintains the base from what grows it, read the ageing against the spending, test the explanations that are not operating, and write down what would settle each remaining question. The output is questions with evidence attached, and a procedure that reaches a conclusion has skipped a step.

Underneath the order sits a chain of handovers, and the handovers are what fix the sequence. Step one hands over two years that can honestly be put beside each other. Step two hands over a set of asset arrivals that includes the ones nobody paid cash for. Step three hands over a range rather than a number. Step four hands over a direction of travel for the base. Step five hands over, for each movement, either a named cause outside operations or a clean bill. Step six converts what survives into open items with somewhere to look attached. Pull out any single step and the one below it is left holding material nobody assembled. The order is load bearing for that reason, rather than a set of tasks to be worked in whatever sequence the afternoon allows.

The mechanisms used here are settled elsewhere in their own right and taken as read: how asset turnover or capital intensity is computed, what depreciation does to a carrying value, how a right-of-use asset arrives on a balance sheet, and what an impairment is. The procedure itself adds three things only: the order of work, the rule that says when the work is finished, and an unambiguous statement of what the finished work looks like.

Capital spend is unusually easy to misread, and the reason is worth naming before the first step. In one set of accounts there are at least three different numbers a reader might reasonably call capital spend, and they can differ by a factor of two or three. There is the whole investing outflow. There is the cash actually paid for property, plant, equipment and software. And there are the additions shown in the asset schedules. Those additions include assets that arrived without any cash at all. Most capital spend analysis that goes wrong does not fail on arithmetic; it fails because the reader picked one of those three numbers without noticing there were three, and then ran four perfectly good steps on it.

Why does establishing the basis come before any arithmetic at all?

Step one reads four things and computes none of them. Whether the figures under examination are the parent business on its own or the parent added together with the businesses it controls. Whether assets are carried at cost or at a revalued amount. Whether anything arrived during the year through a purchase of another business rather than through capital spend. And whether the period contained an accounting change, meaning a change of policy or a change to the useful lives being applied. Every step after step one compares two years, and two years prepared on a different basisThe set of preparation choices behind a figure: which entities are inside it, which measurement model was applied, and which policies were in force when it was struck. are not a comparison at all, however careful the arithmetic that follows. So step one comes first.

The household version comes first. A neighbour mentions that her monthly milk bill has climbed from Rs 1,800/- to Rs 2,700/-, and before anything at all is said about dairy prices the question is how many people were drinking it in each of those months. Her sister moved in with two children in June. The subtraction was fine. The two amounts were measuring different households, and everything built on the gap between them carries that fault forward. Working out what an amount covers before setting it against another amount is not caution. The check is the entirety of step one.

Put step one to Anjani Stationers Private Limited, an invented stationery business, and four answers arrive. The figures are the standalone balance sheet, so the 70 per cent holding in Chitra Binding Works sits inside total assets as a single investment line of Rs 21,00,000 rather than as the binding operation's own machinery and stock. Assets are carried on the cost model, with no revalued amounts in either year. A business was bought at the start of year two. That answer is the holding in Chitra Binding Works restated as a warning. And no change of accounting policy is disclosed, with the useful lives applied unchanged, straight line, and nil residual value assumed on every asset. Two of those four answers do not stop the analysis but do fix what the later steps are allowed to say, and writing them down at step one is what stops them arriving as a surprise at step five.

What does step two rebuild, and why not from the cash flow statement?

Step two builds a roll-forwardA statement that starts at an opening balance, lists every movement in the period, and arrives at the closing balance with nothing unexplained left over. of the asset base from the asset note: opening gross, additions, disposals, closing gross, and the same four lines again for accumulated depreciation. The roll-forward is built from the schedule and not from the investing section of the cash flow statement, and the reason is one sentence long. The cash flow statement can only show assets that were paid for, so every asset that arrived without cash is invisible to it, and an analysis built on it is missing exactly the arrivals that nobody spent anything on.

A non-cash additionAn asset that appears on the balance sheet during the year without any money leaving the business to acquire it. is not exotic. A household that has taken a car on a long lease shows the same thing. Nothing left the bank account except the monthly payment, and yet a car now sits in the driveway and will be there for four years. The bank statement shows a monthly outgoing; what the household controls includes a car. The bank statement is not lying; it simply is not the right document for the question being asked.

Run step two on Anjani Stationers. Gross block opens at Rs 45,00,000 and closes at Rs 64,00,000. Additions are Rs 19,00,000, of which Rs 12,00,000 was paid for in cash and Rs 7,00,000 is the warehouse right-of-use asset that arrived with a matching lease liability and no payment at all. Nothing was sold or scrapped, so disposals are nil. Accumulated depreciation opens at Rs 17,00,000 and closes at Rs 28,00,000 after a charge of Rs 11,00,000, and net carrying value moves from Rs 28,00,000 to Rs 36,00,000. Software adds Rs 1,00,000 of purchases against Rs 1,00,000 of amortisation, so its net figure of Rs 4,00,000 does not move. Now hold the three candidate numbers side by side: the investing outflow was Rs 34,00,000, the cash capital spend was Rs 13,00,000, and the additions across both schedules were Rs 20,00,000. Only the middle one is capital spend, and the widest of the three is more than two and a half times it.

Why is the maintenance and growth split a range rather than a number?

Step three separates the spend that keeps the existing base standing from the spend that adds to it. No set of published accounts discloses that split, so it has to be estimated, and the usual estimate treats spend up to the depreciation and amortisation charge as maintenance capital spendThe part of a period's capital spend that keeps the existing productive base standing rather than adding to it. and the excess as growth capital spendThe part of a period's capital spend that adds productive capacity the business did not have before.. The estimate is a convention, not a measurement, and it rests on a comparison that does not quite hold. The charge is struck on what the assets originally cost. Replacing them happens at whatever they cost today. So the estimate compares a historical number with a current one, and the error runs in a direction nobody can size from the statements.

On Anjani Stationers the convention gives a tidy looking answer. Cash capital spend was Rs 13,00,000 against a charge of Rs 12,00,000. The convention puts maintenance at Rs 12,00,000 and growth at Rs 1,00,000. Growth of Rs 1,00,000 is 7.7 per cent of the spend, a small number sitting on top of a large assumption. Written as a share of revenue, the spend is 4.8 per cent counting property, plant, equipment and software together, or 4.4 per cent on property, plant and equipment alone, and anyone quoting either figure has to say which one it is.

Now push on the Rs 1,00,000 from both sides and watch it stop being a number. At one edge, nothing at all left the base during the year, so not one rupee of the Rs 13,00,000 replaced something that had been retired, and on that reading the whole Rs 13,00,000 bought assets the business did not have before. At the other edge, if putting the same machinery back today costs more than the Rs 12,00,000 the charge is built on, then the maintenance requirement is above the charge and the growth share is nil or less. Both edges are supportable from what is published, so the honest output of step three is that growth lies somewhere between nil and the whole Rs 13,00,000, with the convention's point estimate of Rs 1,00,000 sitting near the bottom of that band rather than in the middle of it. A range that wide is not a failure of the step. The width is the step reporting accurately that the published figures cannot split the spend, and step six exists to write down precisely that sort of thing.

What does the ageing add that the spending cannot say on its own?

Step four reads two things together, and the whole of the step is that neither one means anything by itself. The first is how much of the base has already been charged to profit, taken as accumulated depreciation over gross block. The second is how the spend compares with the charge. A base that is ageing while spend runs at or below the charge is being consumed. A base that is young while spend runs above the charge is being built. A low spend ratio is prudence or starvation depending on how old the base already is, and a young base is a recent build or a base whose oldest assets were quietly removed. So either reading alone can be made to support almost anything.

The household version is a bicycle. Knowing that Rs 1,200/- was spent on it last year tells nothing. Knowing that it is nine years old tells nothing either. Knowing that Rs 1,200/- was spent on a nine year old bicycle tells something worth acting on, and knowing that Rs 1,200/- was spent on a bicycle bought in March tells something quite different. The pairing carries the information, and neither number on its own is the finding.

Run step four on Anjani Stationers. The base was 37.8 per cent depreciated at the end of year one, being Rs 17,00,000 of accumulated depreciation against a Rs 45,00,000 gross block, and 43.8 per cent at the end of year two, being Rs 28,00,000 against Rs 64,00,000. The base is six points older in a single year, during which spend ran at 1.08 times the charge. There is a trap sitting inside that reading, and it runs the helpful way for once. The Rs 7,00,000 right-of-use asset entered the gross block with no accumulated depreciation behind it at all, and a fresh asset pulls the average down. Strip it out and the base reads 46.1 per cent rather than 43.8 per cent. So the arrival of the lease made the base look younger and the base aged anyway. The ageing is real rather than an artefact of what came in. A base can age exactly like that while capacity is being built for revenue that has not arrived yet. So the pair, six points older while spending marginally above the charge, supports a question and nothing more.

Which explanations have to be ruled out before any of this is believed?

Step five puts every movement found so far to a single question: could something other than this business spending money have produced it? Five candidates are worked through in a fixed order, and the fixed order is the point. A list run from memory quietly drops whichever item the reader was least expecting. A purchase of another business bringing assets in. A lease arriving on the balance sheet. A revaluationRestating an asset's carrying value to a current valuation instead of what was originally paid for it. The balance sheet moves without any transaction. restating carrying values. An impairment shrinking the base. And a useful life changeA revision to how many years an asset is expected to be used. The annual charge changes from that point onward without any asset moving. changing the charge. Every one of those five moves capital intensity, asset turnover and the ageing reading without a single rupee being spent differently. Nothing found at steps two to four may be believed until all five have been looked at.

Run them on Anjani Stationers and they come back unevenly. Uneven is ordinary. The cost model is in use in both years, so revaluation was ruled out at step one. None was recognised in year two, so impairment is ruled out. A useful life change is not disclosed, and it is worth pausing there. The charge more than doubled from Rs 5,00,000 to Rs 12,00,000, and a doubling is exactly what a shortened life looks like from the outside. The schedule explains it without one: the vehicles and computers were bought partway through year one and carried only a part year charge then, so year two is the first year the opening base carries a full charge.

Then the other two fire, and between them they account for most of what steps two to four found. The lease put Rs 7,00,000 of assets on the balance sheet with a matching liability and no cash. And the purchase of the 70 per cent holding in Chitra Binding Works put Rs 21,00,000 inside total assets. The purchase deserves reading slowly. On these standalone figures the Rs 21,00,000 is an investment line and not a factory: it sits in the denominator of every asset efficiency measure, it pushes total assets from Rs 1,33,00,000 to Rs 1,80,00,000 alongside everything else, and it produced no revenue in the standalone revenue line at all. Of the Rs 47,00,000 by which total assets grew, Rs 21,00,000 bought a holding in another business and Rs 7,00,000 arrived under a lease, so barely a quarter of the growth is the capital spend that a reader looking only at the ratios would have assumed was all of it.

In words, the walk runs as follows. Reading four documents turns up nothing to ask about, so step one settles five basis items and opens nothing at all. The open list moves for the first time at step two, to two items, and then keeps widening: three at step three, four at step four, and five once step five has aimed each of them at a named candidate. The count of open items climbs at every step from two onward and ends at its widest. Most readers expect competent work to narrow as it goes, and the expectation is wrong. Take step five out and the count falls to nothing at step six, and in its place sits one assured sentence about a business investing heavily and getting less back. Build step two from the cash flow statement instead and the closing gross block computes to Rs 57,00,000 against the published Rs 64,00,000, a gap of exactly Rs 7,00,000. The gap is the lease asset announcing itself.

What does the last step hand over, and when is the work finished?

For every item still open, step six records the one specific thing that would resolve it, and then the work ends. Not the resolution, not the likeliest explanation, not an ordering by which explanation feels safest. The thing itself: a named note, a named column of a schedule, a named statement that would have to be obtained. The work is finished when the basis is established, the asset schedule closes, the maintenance estimate carries its range, and every movement still standing is either explained or recorded as an open item with the disclosure that would resolve it named beside it.

Conditions are the only form somebody else can audit, so the stopping rule is stated as a set of conditions rather than as a length or a deadline. In a colleague's hands every clause is testable without that colleague agreeing with a single judgement the analyst made: does the schedule close, does the split carry a range, and does every movement still standing have either an explanation or a place to look attached? If so, the work is complete whatever it turned up. If not, it is incomplete whatever tone it was written in. A rule that can be failed is the only kind worth keeping. Stop when it feels clear is worth nothing at all. Clarity arrives soonest for whoever has understood least.

Five items leave the desk at Anjani Stationers. The first asks how much of the Rs 47,00,000 growth in total assets is capacity built for revenue that has not turned up yet, for which the asset note by class is the place to look. The second asks what the short lived classes will need next. The vehicles cost Rs 6,00,000 on a four year life and the computers Rs 3,00,000 on a three year life, and the asset schedule by class would show both. The third asks how much of the Rs 13,00,000 replaced something and how much added capacity. The additions and disposals columns would narrow it. Why nothing at all left the base in a year of buying, for which the disposals column and any note on assets fully charged are the evidence. And whether the Rs 21,00,000 holding carries machinery and stock that these standalone figures never show, for which the consolidated statements and the note on the holding are the evidence. Five open items, five named places to look, and no verdict on the business anywhere.

Which three moves are never part of this procedure?

Three moves look like the natural continuation, feel like the reward for the work, and belong to no part of this sequence. Measuring the spend against an industry average is the first, and it fails for the reason step one exists: an average is built from many businesses on many bases, and if two bases are not a comparison then dozens of them are not one either. Calling a business under investing or over investing is the second. Under against what target, set by whom, is disclosed nowhere in a set of accounts. Deciding whether the spend will pay off is the third.

Deciding whether the spend will pay off is capital budgeting, a different subject with its own methods, and it needs information no filing contains. A procedure that runs on published statements cannot reach it, however carefully the first six steps were done. None of that argues against ever forming a view. The argument is that forming a view is a separate activity drawing on separate inputs, and that the seam between the two is exactly where a reader has to say out loud which of them is now being done. Writing that the business is investing too little steps off published figures onto something else entirely, carrying none of the evidence the previous six steps were built to assemble.

Who runs this procedure, and what does each of them do with five open items?

The same accounts get walked through these six steps by three different readers in one week, and no two of them want the same thing at the end. A lender runs the procedure to size how much cash the business must give up before anything is left for interest, an analyst runs it to work out which parts of a forecast rest on an assumption rather than a figure, and Vaidehi Rao, the finance controller, runs it from inside to learn what somebody outside would make of what she is about to publish.

Watch the lender first. The lender's use turns on one computation that step two makes possible and nothing else does. Cash from operating activities was Rs 36,30,000. Cash capital spend was Rs 13,00,000. Free cash flow is therefore Rs 23,30,000, and that is the figure a lender puts against its interest and its repayments. The wrong way to reach it is to add the investing section to the operating section. That sum gives Rs 2,30,000 and a completely different picture of the same year. The investing section carries the Rs 21,00,000 paid for the holding in Chitra Binding Works. Buying another business is a decision the lender may care about a great deal, but it is not the annual cost of keeping this one running, and netting the two together answers neither question.

The analyst cannot ask, so step six is where she records which parts of her model rest on nothing. The Rs 1,00,000 of estimated growth spend is not a number to her; it is a band running the whole width of the spend, and writing it into a forecast as a point would be claiming a precision the accounts do not hold. Vaidehi Rao puts the procedure to the oddest use of the three. She runs it against her own accounts before anybody outside gets the chance. All five resolutions are already known to her. She signed the invoices, so she can see which part of the Rs 13,00,000 replaced and which part added. The test is something else: whether somebody working only from the published accounts lands anywhere defensible, and which way they tip when they are pressed for time. A reader who comes back reporting heavy investment in machinery nobody bought has shown her a disclosure gap she can close with a note this year, instead of an argument she has to win next year.

What is the honest last line of this work?

Not a verdict. Anjani Stationers spent Rs 13,00,000 in cash on assets against a Rs 12,00,000 charge, its base aged from 37.8 per cent to 43.8 per cent depreciated, its asset turnover fell from 1.80 to 1.50 times while total assets grew Rs 47,00,000 of which Rs 21,00,000 bought a holding in another business and Rs 7,00,000 arrived under a lease, and the split between maintaining and growing sits in a band as wide as the spend itself. Those figures are the resting place of six steps worked properly. The procedure works from published figures, and those figures simply do not hold the resolutions. Finishing with open items is the right outcome and not a shortfall, and a closing line that supplied a resolution would have come from whoever wrote it rather than from the accounts.

Everything the six steps are for lives in the gap between the two candidate endings, so the endings are worth setting beside each other. The first says the business is investing heavily and getting less back. That ending sounds complete and rests on nothing. The second says cash capital spend was Rs 13,00,000, roughly a tenth above the charge, on a base that aged six points, alongside Rs 28,00,000 of asset growth that no capital spend produced, and here are the five disclosures that would tell a reader what any of it means. The second takes longer to write, gives the reader less, and is the one still standing after somebody checks it. So the honest closing line is not a statement at all. The honest closing line is the item left exactly as it stands: how much of the growth in this asset base is capacity waiting to be filled, and how much of what is already standing there is nearer the end of its life than the average lets on?

Anjani Stationers Private Limited, Chitra Binding Works Private Limited and Vaidehi Rao are invented.
Educational material. Not advice on any investment, tax, budget or market position.