Cash Pooling and Concentration: Notional, Physical and What Each Moves

Cash pooling connects bank accounts that were being run separately. Physical concentration sweeps the balances into one header account, so the cash genuinely moves and one entity ends up owed money by another. Notional pooling leaves every balance untouched and combines them only inside the interest computation. One arrangement creates positions between entities and the other creates none, and that single mechanical difference decides almost everything else.

Start where a treasurer actually starts: with a set of bank accounts nobody ever planned. A business opens one because a plant needs it, another because a large customer pays into a particular city, a third because a subsidiary was incorporated and had to have its own. Nobody sat down and decided to run twenty six accounts. The accounts arrived one at a time, each for a good reason, and by the time somebody finally looks at them together the interest running across them has stopped being an accident and started being a number worth a treasurer's afternoon.

Nirjhar Industries Limited, an invented steel and alloys maker, is where that number can be looked at hard. Girish Talwalkar is its group treasurer. The group holds Rs 540 crore of cash across 26 bank accounts with 4 banks in 3 currencies, and it makes about 1,840 payments a month. Every rupee, rate, count and limit below belongs to that group and to nothing else.

What problem does cash pooling actually solve?

Asked what pooling is for, most people answer in terms of tidiness: fewer places to look, one number instead of many, a treasurer who can finally see the position. All true, and none of it is the reason anybody signs the paperwork. The reason is a rupee figure, and the rupee figure comes from a group paying one rate on money it has borrowed while earning a lower rate on money it already holds, at the same moment, inside itself.

The shape is identical in a household and costs nothing to feel there, so the household version comes first. A household keeps a fixed deposit of Rs 4,00,000/- at one bank because that is where the salary account has always been. The car loan came through another bank, and at that bank the household runs a credit line of Rs 1,00,000/- against which it has drawn the full amount. The deposit earns a modest rate. The drawn credit line costs three times that rate. The household is comfortably in credit once the two are added together, and it is still losing money every month, and nobody in the house is doing anything wrong. The two amounts have simply never met.

Cash poolingConnecting bank accounts that were being managed separately, so that the position is treated as one rather than as many. is that meeting, arranged deliberately and at scale. At Nirjhar, 18 of the 26 accounts, all of them belonging to E1 Nirjhar Industries Limited and E2 Nirjhar Alloys Private Limited, went into an arrangement of exactly this kind. Before it started, those 18 accounts between them held gross credit balances of Rs 486 crore and gross overdrawn balances of Rs 162 crore. The overdrawn figure is exactly one third of the credit figure. The relation is arithmetic and not a fact about anything.

At the group's own contracted rates, invented and never market rates, of 3.0 per cent earned on a credit balance and 9.5 per cent paid on an overdrawn one, the year looked like this. Paid: Rs 162 crore times 9.5 per cent = Rs 15.39 crore. Earned: Rs 486 crore times 3.0 per cent = Rs 14.58 crore. The group paid Rs 0.81 crore a year in net interest on its own cash, across a set of accounts that held three rupees of credit for every rupee overdrawn. The Rs 0.81 crore is the whole reason for what follows.

One phrase needs pinning down before it does damage. The statement that the pooled accounts were in credit throughout means those 18 bank accounts and nothing else. The group itself carries net debt of Rs 4,320 crore less Rs 540 crore of cash, being Rs 3,780 crore. Net debt of Rs 3,780 crore is 2.92 times its earnings before interest, tax, depreciation and amortisation of Rs 1,296 crore. A set of bank accounts in credit and a business without borrowings are two entirely different statements, and Nirjhar is the first and very much not the second.

Notice what that picture does and does not say. The picture does not say the group was short of money. The picture says the group was paying an overdraft rate somewhere inside itself while earning a deposit rate somewhere else inside itself, and the two rates were far enough apart that the smaller balance won. Pooling makes those two balances see each other. There are two ways of arranging it, they look similar on a bank's proposal, and they are not remotely the same thing.

Cash Concentration: what actually happens to the money at the end of each day?

Cash concentrationPhysically moving balances into one account, so the cash actually leaves the accounts it came from. is the physical answer, and the word physical is doing real work. At an agreed cut-off each evening, every account in the arrangement is emptied. Whatever balance it holds is transferred out and lands in one nominated account. The next morning, each account is funded back up to what it needs for the day's payments. The daily round trip is a zero balance sweepAn arrangement that empties every participating account into one account each day and funds it back the next morning., and the account everything lands in is the header accountThe single account a physical sweep concentrates balances into..

The picture in most people's heads at this point is wrong in a way worth correcting immediately. Most people imagine the operating accounts being closed, or drained and left useless, or a plant manager finding no money to pay a supplier. None of that happens. The accounts keep working exactly as they did; the only thing the sweep takes is the balance that would otherwise have sat there overnight doing nothing. A supplier paid on Tuesday morning is paid out of Tuesday morning's funding, and the sweep never sees that money at all. The sweep changes the overnight position, and the overnight position is precisely what the bank computes interest on.

A sweep runs on a rhythm, and the rhythm is worth stating because it is what makes the arrangement operationally boring rather than dramatic. Evening: balances out. Overnight: one position. Morning: balances back. Repeat tomorrow. Nobody has to decide anything on any given day. Reliability, not cleverness, is where the value comes from. The single decision was taken once, at the point of signing, and the arithmetic then runs itself three hundred and sixty five times a year.

What does moving the cash actually do inside the group?

Here is the part that separates people who have run a pool from people who have read about one. A group is not one pocket. A group is a set of separate legal entities, each with its own accounts, its own auditor, its own directors and its own tax position. E1 Nirjhar Industries Limited, E2 Nirjhar Alloys Private Limited and E3 Nirjhar Trading FZE are three companies, not three departments. So when cash physically leaves an account held by one of them and lands in a header account held by another, something has happened that is not an internal transfer at all.

Money that leaves one entity's bank account and arrives in another entity's header account is a debt from that evening, and it is a debt again tomorrow evening, and the evening after that. If E1's balance lands in a header account held by E2, then E2 is holding money that belongs to E1, and E2 owes it. Nobody signed anything. Nobody called it a loan. Moving cash between two legal persons does exactly that. The cash moved, and the obligation arrived with it.

A document, a price and a repayment schedule are what make an internal loan a loan rather than a floating balance, and all three are settled under the pricing and documentation of internal loans. E1 already lends E2 Rs 240 crore on exactly that basis, documented, priced at the group's own contracted rate and carrying a schedule. A physical sweep manufactures internal lending as a by-product, at speed, without anybody choosing to. Whether the price on it works for tax is a transfer pricing question for the group's own tax advisers on its own facts.

Notional Pooling: what moves when it runs?

Nothing. The one-word answer is the whole of it, and the urge to complicate it is worth resisting.

In notional poolingAn arrangement in which balances stay where they are and only the interest computation treats them as combined., every balance stays exactly where it is. The account in credit is still in credit at the same bank on the same night. The overdrawn account is still overdrawn. The bank agrees, for the purpose of computing interest, to treat the balances as though they were one. The treatment is called offsettingTreating a credit balance and an overdrawn balance as netting against each other for a stated purpose., and the phrase for the stated purpose is doing all the work in that sentence. Notional pooling is an arrangement about arithmetic, not about money, and everything that follows from movement simply never arises.

The consequence follows quickly. If nothing moves, then no entity's balance sheet is touched, no internal position is created, nothing has to be priced between entities and nothing has to be documented between them. The arrangement is between the group and its bank, and it stops there. The simplicity is genuinely attractive, and the simplicity is also why the arrangement is a legal and tax question rather than a treasury one. Somebody has to be satisfied that balances belonging to different companies, possibly in different places, may properly be offset against each other for the purpose the arrangement states.

The Reserve Bank of India at rbi.org.in states what an Indian entity may do about offsetting, about operating a pool, and about moving cash across a border, and the rest is a matter for a group's own tax and legal advisers on its own facts. Anyone who offers the general answer without asking whose entities, whose banks and which places is claiming something they cannot know.

Cash Pooling vs Notional Pooling: which one creates a debt inside the group?

Both arrangements are sold under the same word and they are drawn on the same sheet of the same proposal. Both promise a treasurer one position instead of many. The only mechanical difference between them is whether cash moves, and every other difference in the table below is a consequence of that one. Read the first row, then read the other four as its children rather than as separate facts to memorise.

Two of the consequences decide who ends up in the room, so follow them out one more step. First, documentation. A group running a physical sweep has to record the internal positions it keeps creating, price them, and be able to explain both to an auditor. A group running a notional pool records an arrangement with its bank and nothing else internally. There is nothing else to record. The physical arrangement is cheaper on interest and more expensive on paperwork, and the paperwork is the part that arrives eighteen months later when nobody remembers the decision.

Second, what each entity's own accounts show. After a sweep, the entity whose cash moved reports less at the bank and more owed to it by a related company. The entity's picture differs from the one it presented before, and a reader of that single entity's accounts, a lender to it for instance, is now looking at a receivable from a related party where cash used to be. Nothing improper has happened. But the picture changed, and it changed because of a treasury arrangement rather than because of anything the business did.

What was physical concentration actually worth at this group?

Now the arithmetic, in full, on the locked figures. Before pooling, across the 18 accounts: earned Rs 486 crore times 3.0 per cent = Rs 14.58 crore, paid Rs 162 crore times 9.5 per cent = Rs 15.39 crore, so the year's net interest position on the group's own cash was a cost of Rs 0.81 crore. After physical concentration there is one balance and one rate: Rs 486 crore less Rs 162 crore = Rs 324 crore, times 3.0 per cent, earning Rs 9.72 crore.

Hold that Rs 324 crore carefully. A second Rs 324 crore lives in this case, and the two have nothing to do with each other. The pooled net cash position across the 18 bank accounts is Rs 324 crore, and Nirjhar's profit after tax for the year is also Rs 324 crore, and they are two entirely different objects that happen to share a number. One is where money sat on a particular night. The other is what the year earned after tax. Never let them touch.

The swing is Rs 9.72 crore plus the Rs 0.81 crore that is no longer being paid, being Rs 10.53 crore a year, every year. The running cost of the arrangement is a daily instruction nobody has to think about. The Rs 10.53 crore is the number that appears on the proposal, and the number the treasurer takes to the board.

Almost nothing changed between the two halves of that table. Look at what did. Nobody put a rupee of new money in. Nobody negotiated a better rate on anything. Nobody sold an asset, drew a facility or reached an agreement with a customer. The balances are the same balances and the two rates are the same two rates. The only thing that changed is that the Rs 162 crore and the Rs 486 crore were finally allowed to sit in the same place, and that alone turned a Rs 0.81 crore annual cost into a Rs 9.72 crore annual receipt.

So where did the Rs 10.53 crore actually come from?

Ask a room of people this and most of them give the same answer: the cash is finally in one place where it can be put to work properly, so the group is now earning more on it. The answer is a completely reasonable one. The answer is also wrong, and wrong in a way that matters. A treasurer who believes it will go looking for the benefit in the wrong column and will over-promise on the next arrangement.

Predict the answer before the algebra arrives, then move the controls underneath and see whether the prediction survives contact with the arithmetic.

Now the algebra. Four lines settle the argument permanently. Write C for the gross credit balances, D for the gross overdrawn balances, c for the rate earned on a credit balance and c plus s for the rate paid on an overdrawn one. The letter s is then the spreadThe difference between the rate paid on an overdrawn balance and the rate earned on a credit balance, which is where the value of connecting accounts comes from. between the two rates. Before pooling the year is C times c, less D times c plus s. After pooling there is one balance, so the year is C less D, all times c. Subtract the first from the second and every single term in c disappears. D times s survives, the overdrawn balance multiplied by the spread, and the rate earned on the cash is not in the answer at all.

Check the identity against the locked figures and it lands exactly. Rs 162 crore times 6.5 percentage points is Rs 10.53 crore. The long way through Rs 15.39 crore paid, Rs 14.58 crore earned and Rs 9.72 crore after gives the same swing. Two routes, one number, and the short route does not mention the credit rate. The truth is far less glamorous than earning more: the group stopped paying 9.5 per cent on Rs 162 crore that it already had sitting somewhere else in its own name. Pooling did not make the cash work harder; it stopped the cash being borrowed back at a punishing rate by its own group.

Two consequences follow that a treasurer can use immediately. First, D is zero for a group with no overdrawn balances anywhere, and D times s is then zero, so such a group gets nothing from pooling however untidy its accounts are. Second, the value scales with the spread and not with the size of the cash pile, so a group with a modest balance and a wide spread beats a group with an enormous balance and a narrow one. Both statements come straight off the identity, and neither is obvious until the credit rate is seen to vanish.

Which accounts sit outside the pool, and why?

A pool has a boundary, and the boundary is a decision rather than an oversight. Of the group's 26 accounts, 18 are participating accountsAccounts included in a pooling arrangement, as against those deliberately left outside it. in the sweep and 8 are not. The 18 belong to E1 Nirjhar Industries Limited and E2 Nirjhar Alloys Private Limited, both in India. The 8 belong to E3 Nirjhar Trading FZE, an overseas trading entity. Moving cash across a border is a decision of a different kind from moving it across a city, and it carries its own consents, its own documentation and its own approvals.

Being honest about what the boundary costs is worth a moment. Leaving E3 out means the group is running a pool that does not reach every rupee it holds, and a purist would call that an incomplete job. A treasurer would call it a completed one. A project that tries to solve two problems at once, the domestic one and the cross-border one, usually delivers neither. The domestic sweep was straightforward, the arithmetic was known in advance and it went live. The cross-border question was left where it belongs, with the group's own advisers and with the source that governs it.

Why did this group look at a notional pool and not take it?

The most instructive part of the story is the part most treatments leave out. A bank did offer this group a notional pool. The arithmetic on it was attractive for exactly the reasons the comparison above sets out: no sweep to run, no internal positions to create, nothing to document between entities, one arrangement with one bank and an interest computation that does the work. On paper it is the cleaner of the two arrangements by some distance.

The group reviewed it and did not adopt it, and the reason had nothing to do with the arithmetic. The offsetting the arrangement depended on would have run across two jurisdictions, and that turns a treasury question into a legal and tax one. The group put it to its own tax and legal advisers, and on these particular facts those advisers did not support it. So the group kept the physical sweep it already had, and recorded the notional pool as considered and declined.

Read that sequence carefully for what it does not contain. The sequence contains no rule. The sequence does not say that a cross-border notional pool is permitted, and does not say that one is refused. The sequence records one invented group's decision on its own facts. The only honest general statement available is that the answer depends on whose entities, whose banks and which places are involved, and a question of that kind belongs with advisers and with the source. The Reserve Bank of India at rbi.org.in states what binds an Indian entity, including anything at all about cash crossing a border. Anybody who offers a general answer to this without asking those three questions is claiming something they are not in a position to know.

What new problem did the pool create on the day it started working?

Here is the sentence that separates a good treasury project from a finished one. The sweep worked. The sweep delivered exactly the Rs 10.53 crore a year the proposal promised, on the arithmetic the proposal set out, with no surprises in the interest computation at all. And it created a fresh problem the same evening, one that was sitting in the arithmetic the whole time and that nobody had looked at because everybody was looking at the interest.

There is a second consequence, quieter than the first and more expensive to fix late. Every evening the sweep runs, it manufactures internal positions between E1 and E2 of exactly the kind described earlier. A group that starts a sweep without deciding in advance how those positions will be recorded, priced and repaid has acquired internal lending that nobody chose, at no stated price, on no schedule. The person who eventually asks about it is an auditor or a tax adviser, eighteen months later, and by then there is a year of undocumented daily positions to reconstruct. Neither problem is an argument against pooling. Both are arguments for reading the second consequence of a good decision before the first one starts paying.

Who actually picks this up, and what do they do with it?

Four people read a pooling arrangement and each of them reads it for something different. Knowing which is which is worth doing before sitting in a room with any of them.

The treasurer reads it as an identity, not as a project. D times s can be computed from balances the treasury already has on its own statements, so Girish Talwalkar does not need a bank's proposal to work out what a pool is worth to this group. The overdrawn balances across the accounts, multiplied by the difference between the two contracted rates, is the answer before anybody has been asked for a quote. Knowing the number in advance matters commercially: a treasurer who walks into the meeting already holding it is negotiating about fees and operations rather than about value.

A lender to one entity reads it as a change in what it is looking at. After a sweep starts, the entity whose cash moves reports less at the bank and a receivable from a related company instead. Nothing improper has happened and the group is no weaker than it was. But a lender that had cash as its comfort now has an internal receivable as its comfort, and those are not the same asset. The swap of cash for an internal receivable is the single most common surprise a physical pool delivers to somebody outside the treasury, and the conversation is worth having before the arrangement starts rather than at the next review.

An analyst reads a consolidated cash figure and cannot see the pool at all. Consolidation nets the internal positions away, so the sweep leaves no trace in a set of consolidated accounts. The interest line is what the analyst can see, and the interest line is where a pool actually shows up. A group that has stopped paying an overdraft rate to itself reports less interest for reasons that have nothing to do with its borrowings. The obvious question, whether the interest fell because debt fell or because internal borrowing stopped, is the useful one.

And a household reads it as the thing it should have done years ago. The Rs 4,00,000/- deposit at one bank and the fully drawn Rs 1,00,000/- credit line at another is the same arrangement in miniature, and the same identity applies: what the household saves by clearing the drawn line from the deposit is the drawn amount times the difference between the two rates, and it is nothing whatever to do with what the deposit earns. The identity is the whole subject in one sentence a person can act on this week.