Cash Management: How Much Cash a Business Should Hold

No correct cash balance exists. Only a balance sized against named claims exists. Sankalp Industrial Systems Limited, invented, holds Rs 40,00,00,000 of operating cash, being 20.28 days of cost, beside Rs 80,00,00,000 of surplus. Its Rs 1,00,00,00,000 working capital line is drawn to the limit, so it adds no room, and anyone treating it as spare capacity has counted one borrowing twice.

Underneath that answer sits a substitution, and the substitution is the whole method. Boards ask how much cash the business should hold. The amount depends on facts about the business that no rule of thumb carries, so the question has no answer. Three narrower questions do have answers. Which claims must this balance be able to meet? How fast would each of them arrive? Which other resources could meet them? Each can be worked out from the accounts, and the three worked answers add up to a number that can be defended in a room.

Why does the question, as it is usually asked, have no answer?

The same question put to a household shows the difficulty. How much money should a person keep in a savings account? There is no answer. The answer depends on whether the rent is due weekly or yearly, whether the salary lands on the first of the month or whenever a client pays, whether there is a working scooter or a broken one, whether a parent is well. Four narrower questions put to the same person are answered easily: rent is Rs 18,000 a month, the scooter service is Rs 4,000 and overdue, one salary arrives on the second, and a hospital bill of any size would have to come out of savings. Now the balance is arguable.

A business is the same shape at a larger scale, and the reason the broad question stalls is that it collapses three different reasons for holding money into one number. Money is held because payments leave before receipts arrive. Money is held because something may land without warning. Money is held so an opportunity can be taken without going to ask for funding. The three reasons are not versions of each other. The three are sized against different things and move at different speeds, and pooling them into a single figure hides which of the three is driving it.

Claim one: what does it cost to pay before being paid?

Picture a vendor selling snacks outside a college gate. She buys her supplies on Monday morning with money in hand. The students who buy on credit through the week settle on Friday, some of them. Between Monday and Friday she has spent and not collected, and the money that carries her across those four days is not profit, is not savings and is not spare. The money is the price of running the business at all, and if it is not there on Monday the stall does not open.

Every business has that four day hole, and the only thing that varies is how wide it is and how it is measured. A manufacturer buys steel, pays for it, machines it, holds it as finished stock, ships it, waits for the customer's payment terms to run, and only then sees cash. In between, wages went out, power went out, the supplier was paid. The cash that sits in the account to absorb that timing difference is what the transaction claim means. The transaction claim is not a buffer against anything going wrong. The claim is what going right costs while it is happening.

How is the transaction requirement sized without guessing?

The balance is converted into time. Nobody in the room has an instinct for whether Rs 40,00,00,000 is a lot, so a rupee figure invites a nod and nothing else. Everybody in the room knows how long their own collections take, so a number of days invites the right argument.

Take Sankalp Industrial Systems Limited at Year 0. Cost of goods soldWhatever was spent producing the goods actually sold in the year. The cost moves with volume and sits above operating profit. for the year is Rs 7,20,00,00,000, so one day of cost is that figure over 365, being Rs 1,97,26,027. The operating half of the cash balance is Rs 40,00,00,000. Divide, and the balance is 20.28 days of cost. The 20.28 days is the transaction claim expressed as time, and the claim is now comparable with something.

The comparison is the company's cash conversion cycleCounted in days: how long money stays tied up between paying a supplier and collecting from a customer. of 47.45 days. Receivables run 65.70 days, inventory runs 73.00 days, and payables give back 91.25 days, computed on the unrounded day counts rather than on the printed ones. So Sankalp funds 47.45 days of its own operations before a rupee returns. Its operating cash covers 20.28 of those days, being 42.7 per cent of the gap and not half of it. The remaining 27.17 days are funded by borrowing and by the credit its own suppliers extend. Valued at a day of cost, those unreached days come to Rs 53,60,00,000.

A reader will meet two further comparisons in somebody else's paper and should be able to place them. Set against revenue of Rs 12,00,00,00,000, the operating cash is 3.33 per cent and the whole balance is 10.00 per cent. Set against net working capital of Rs 1,80,00,00,000, the operating cash is 22.22 per cent. Revenue says nothing about when money leaves, and net working capital is a stock rather than a speed. Neither ratio is wrong, and neither is much use alone. The days measure earns its place by being the only one of the three carried in the same units as the difficulty it describes.

The days measure is where the discussion actually starts. A treasury team that says the company holds Rs 40,00,00,000 of operating cash has said nothing anyone can push back on. A treasury team that says the cycle runs 47.45 days and the cash covers 20.28 of them has invited the only two useful questions in the subject: should the cycle be shorter, and who is comfortable funding the rest of it.

Claim two: what could arrive, how big would it be, and how fast?

The precautionary claim is the one that people size with a percentage, and the percentage is where the thinking stops. Somebody says three per cent of revenue, or one month of costs, and the figure goes into the paper and is never argued with again. The figure came from nowhere, so there is nothing to argue with.

The honest way to size a precaution is to write down what the year actually contains and add it up, rather than to apply a share of a number that has nothing to do with what falls due. The list is short and every item is in the accounts already. Interest. Capital spending the board has committed to. The extra working capital that growth consumes. The dividend, if the company pays one. The list is not everything a business owes in a year, and no list is, but the four items on it have sizes that are known rather than assumed.

What does one real year of claims look like on this company?

Here is Sankalp's Year 1, set out as rows rather than as a sentence. Interest runs on opening gross debt of Rs 6,00,00,00,000 at the company's own blended cost of debtOne rate standing in for several borrowings at once, each carrying weight in proportion to its size. of 8.00 per cent. The 8.00 per cent is a contracted rate on these particular borrowings rather than a statement about borrowing costs generally. The dividend is the Year 0 declaration of Rs 3.60 a share. The other two rows come straight out of the year's plan, and every amount below is Year 1 unless the row says otherwise.

Two things are worth pulling out of that table before moving on. The first is that the Rs 72,00,00,000 is really two payments: a regular Rs 52,00,00,000 and a special dividendDeclared once, outside the regular cycle, and carrying no promise about the year after. of Rs 20,00,00,000 declared once. The regular part has risen every year, and the company will be judged on it next year. Strip the special part out and the recurring version of the list is Rs 2,52,80,00,000, a total the balance covers to 47.5 per cent. Both numbers are useful and they answer slightly different questions, so a paper that prints one should say which.

The second is what 44.0 per cent does not mean. An operating business meets a year of claims out of the year's own cash generation and not out of the balance it started with. A balance covering under half of one year of claims is therefore ordinary rather than alarming. Set the four claims against Year 1 earnings before interest, tax, depreciation and amortisation (EBITDA) of Rs 3,16,80,00,000 and they take 86.1 per cent of it. The comparison against EBITDA is the one worth making. EBITDA is what arrives during the year, and the balance is only what happened to be there at the start of it. The same thing appears directly in Sankalp's own cash flow. Year 1 free cash flow to equity, after the interest, the tax, the capital expenditure and the working capital movement have been taken out, is Rs 87,00,00,000. Against the Rs 72,00,00,000 dividend that leaves Rs 15,00,00,000. The year pays for the year.

There is a catch inside that, and it is worth seeing because it is the kind of thing a reader is meant to notice. Rs 25,00,00,000 of new borrowing sits inside that Rs 87,00,00,000. Take it out and the year generates Rs 62,00,00,000 against a Rs 72,00,00,000 dividend, a shortfall of Rs 10,00,00,000. So the year pays for the year only with the help of a further drawing on the same working capital line the treasury paper is about to count as spare. Once it is made, the line stands at Rs 1,25,00,00,000 where it stood at Rs 1,00,00,00,000 a year earlier.

Claim three: what does being able to act without asking actually buy?

The third reason for holding cash is the one that resists measurement, and pretending otherwise is how treasury papers acquire numbers nobody can defend. A company holds some cash so that when a supplier goes under and its plant comes up for sale, or a competitor's order book becomes available, the answer can be given in a week rather than in a quarter.

The reach of the cash can be sized, even though the worth of the speed cannot. The board set this year's investment spending at Rs 2,50,00,00,000. One project inside it is mandatory and takes Rs 45,00,00,000, so Rs 2,05,00,00,000 is left to choose between. The Rs 80,00,00,000 of surplus funds 32.0 per cent of one year's investment programme, so the surplus buys speed on a fraction of the list rather than independence from funding. Nothing in this company's record puts a value on that speed.

The absence of that valuation matters more than it looks. The benefit is unobservable and the cost is spread thinly, so the opportunity claim is the one under which almost any balance can be justified. Where the worth of the speed cannot be stated, the honest move is to state what the cash can reach and let the reader weigh it, rather than to attach a number that would look precise and mean nothing.

What else can meet a claim, if not cash?

Cash is not the only thing that can answer a claim, and a company that behaves as though it were will carry a balance it does not need. A committed facilityWritten into an agreement rather than offered informally, so the lender must lend up to the stated amount once the agreed conditions are met. can answer a claim. So can an asset that could be sold quickly, and so can the simple ability to pay a supplier a week later than usual.

Think about a household again. Two people have the same salary and the same rent. One keeps Rs 2,00,000 in a savings account. The other keeps Rs 50,000 and has an overdraft on the salary account with nothing drawn against it. The second person is not less prepared. The second person has moved the readiness from an owned balance to a promise from somebody else, and the promise costs less to carry as long as it is there when it is needed.

The words as long as it is there carry the whole subject, and most of the damage in corporate treasury is done exactly there. A promise from somebody else is only as good as the two conditions attached to it, and both conditions are easy to state and easy to skip.

Which two conditions must a facility meet before it counts as room to move?

The first is that it must be undrawn. A line already borrowed against is not capacity, it is a balance the company owes. The money went into the business at some earlier date and is now sitting inside inventory or receivables or a paid supplier invoice. The line has become an obligation, and nothing is left to draw.

The second is that the conditions for drawing it must still be met on today's position, not on the position at the date the agreement was signed. The tests attached to a facility are usually tied to the borrower's own performance. The second condition is the one that gets skipped, and it is the one that decides outcomes. A line sized against a borrowing baseThe pool of assets a lender sizes a line against, most often receivables and inventory. shrinks when receivables shrink, and receivables shrink in exactly the conditions that make a company want to draw. A line with a financial test attached stops being available at the moment the test is failed. The moment the test is failed is the moment the company most wanted the money.

Put those together and the rule is unsentimental. A facility counts towards what a company can do only while both conditions hold, and a facility failing either one is worth nothing to a plan, however impressive the limit looks in the notes to the accounts.

What does Sankalp's own facility give it, worked through?

Sankalp has three borrowings at Year 0. A secured term loan of Rs 3,00,00,00,000, listed debentures of Rs 2,00,00,00,000, and a working capital facility of Rs 1,00,00,00,000 secured against receivables and inventory. Gross debt is Rs 6,00,00,00,000. The working capital facility is drawn in full.

Work the three numbers a treasury paper ought to carry. Total facilities are Rs 1,00,00,00,000. The drawn balance is Rs 1,00,00,00,000. Undrawn capacity is therefore Rs 0. Total facilities, the drawn balance and undrawn capacity are three different figures, the difference between the first two is exactly the third, and only the third can be spent. Add the undrawn capacity to the cash balance and what Sankalp can deploy today is Rs 1,20,00,00,000, which is the cash and nothing else.

The drawn balance is easy to picture as money sitting somewhere. It is not there. The Rs 1,00,00,00,000 was borrowed and spent into the working capital cycle, and it now sits inside the Rs 2,16,00,00,000 of receivables and the Rs 1,44,00,00,000 of inventory the facility is secured against. Receivables and inventory less payables of Rs 1,80,00,00,000 give net working capital of Rs 1,80,00,00,000, and no cash line sits inside that figure at all. The cash sits separately. So the whole Rs 1,20,00,00,000 balance is picked up as a separate item wherever the company's capital is added up, and nothing is double counted when it is.

What happens when the drawn line is counted as though it were spare?

How does a lender or an analyst actually use all this?

A credit officer looking at Sankalp does not start from the cash line. She starts from what has to be paid and asks what is available to pay it with, and she builds the same two line habit into her own spreadsheet because she has been burned by the one line version. Cash held, then capacity that could be drawn on today's position, and a note of what the drawing test is. A bad two quarters takes receivables down by roughly thirty per cent. So if the test is a borrowing base, she reruns it on a receivables figure thirty per cent lower and looks at what the line gives then rather than what it gives now.

An equity analyst uses the same three numbers differently. He is not worried about whether the company survives. He wants to know what management expects, and the balance is his evidence. A company sitting on a surplus while its interest coverHow many times over the year's operating profit would pay the year's interest charge. is thin is behaving defensively, and the surplus is being held because somebody insisted. A company sitting on a surplus with comfortable cover is holding it for a reason of its own, and the interesting question becomes what that reason is.

Both of them refuse to read a single number and ask instead what the number is being held against. The refusal is the whole discipline of this subject, and it is available to anybody who can read a set of accounts, without a model and without a data service.

What can be read backwards out of a cash balance?

A balance is evidence. Somebody chose it, and the choice was made against a view of what the company was going to face. So the reasoning can be run in reverse: given this balance and these other readings, what would the company have to be expecting?

Run it on Sankalp. Interest cover, being earnings before interest and tax (EBIT) of Rs 2,40,00,00,000 over interest of Rs 48,00,00,000, is exactly 5.00 times. Net debtWhat is left of the borrowings once the cash on hand is set against them. is gross debt of Rs 6,00,00,00,000 less the cash of Rs 1,20,00,00,000, being Rs 4,80,00,00,000, and against Year 0 EBITDA of Rs 2,88,00,00,000 that is 1.67 times. Neither reading describes a company under pressure from anybody. A company with cover at 5.00 times and leverage at 1.67 times is not holding Rs 80,00,00,000 of surplus because its lenders are nervous, so the surplus reads as a choice rather than as a condition imposed on it.

The inference stops at a definite point. A balance read backwards is a choice, and that is all it is. The reading does not say the choice is a good one, does not say the surplus should be spent or returned, and does not say what the company should do next. Reading a balance backwards leads to what the number implies about expectations, and then it hands over to a different subject entirely.

What can a cash balance never solve, however large it is?

Two things, and both are worth naming because a treasury team under pressure will be asked to solve them with cash and will not be able to.

The first is a claim far larger than any balance a working business could carry. Sankalp's secured term loan of Rs 3,00,00,00,000 amortises not at all. The whole of it is repayable on a single day, and that day sits at the close of Year 5. Two and a half times the entire cash balance arrives at once. A company could try to accumulate against it, and would spend five years carrying a balance whose funding costs it far more than the deposit return the balance earns, and would still arrive short. The instalment is a question about how the company funds itself, and cash management is the wrong tool for it.

The second is more subtle. Past the point where every claim on the list is covered, another rupee of cash meets nothing new, and it costs exactly what the rupee before it cost. Rank Sankalp's four Year 1 claims from smallest to largest and the shape is easy to see. The first Rs 18,00,00,000 covers the working capital movement. Reaching Rs 66,00,00,000 covers interest as well. Reaching Rs 1,38,00,00,000 covers the dividend too, and only at Rs 2,72,80,00,000 is capital expenditure covered as well. The Rs 1,20,00,00,000 actually held covers two of the four whole and reaches 75.0 per cent of the way into the third. Beyond the top of that list, the next rupee changes nothing on it.

The flat top of that list is the honest end of the method, and it is also why no curve of cash held against the chance of running short can be drawn for this company. Such a curve exists in principle. Drawing it would require knowing how often an unplanned claim arrives, how large it tends to be, and whether the arrivals come together or independently. None of the three is recorded anywhere for this company. Drawing the curve anyway would mean choosing three numbers and then presenting a picture that looked as though it had followed from the business. Naming a correct number of days of cash for a stranger's company is the same failure.

Sankalp Industrial Systems Limited is invented.
Educational material. Not advice on any investment, tax, budget or market position.