Representations and Warranties: What the Seller Promises Is True

A representation and warranty is a statement of fact about the target company that the seller promises is true on a stated date. The transaction document turns a false statement into a money claim against the seller, filtered by the limits that same document sets. A warranty list is therefore read as a price term rather than as description.

Here is the problem the whole mechanism exists to solve, and it is worth sitting with for a moment before any vocabulary arrives. A buyer is about to hand over a very large sum for a business it has never run, staffed by people it has never managed, holding contracts it has read summaries of and carrying obligations it has been told about by the people who want the money. Diligence narrows that gap. Diligence does not close it, and never has. Somebody who has spent six weeks in a data room knows a great deal more than they did and still does not know what the seller knows.

So the paper does something that looks almost naive at first sight. The document asks the seller to write down, in a long numbered list, that certain things about the business are true. The accounts were prepared on a stated basis. There is no litigation running other than what has been listed. The company holds the licences it needs. Nobody has claimed the company infringes their patents. Each of those sentences is a warranty, and the point is not the sentence. The point is that a false one has a price on it.

The same reflex appears in a household transaction. A buyer of a second-hand car can look at it, drive it around the block, have a mechanic spend an hour under it, and at the end of all that still not see whether it was in a flood two monsoons ago. So the buyer asks the seller to say, on the receipt, that it has never been submerged. If it turns out it was, there is no un-buying a car that has been driven for three months. The buyer gets a claim for what the flood damage cost. The written promise did not give the buyer knowledge; it gave the buyer somebody to send the bill to. That is exactly what a warranty schedule does, at a scale of hundreds of crore.

Harivansh Packaging Limited, an invented company listed on both Indian exchanges, is buying 100 per cent of Sundarban Polymers Private Limited, an equally invented unlisted maker of flexible packaging films. The enterprise value is Rs 1,320 crore, being 10.0 times the target's earnings before interest, tax, depreciation and amortisation (EBITDA) of Rs 132 crore. Taking out Sundarban Polymers' net debtBorrowings less cash. Net debt is deducted from enterprise value to reach the amount actually paid to the sellers, and how it is built up is settled in the valuation layer. of Rs 180 crore and what reaches the sellers as equity value is Rs 1,140 crore. Ashwin Rege runs the transaction team here, and Devyani Kulkarni signs off on the numbers as chief financial officer. The calendar took twenty two weeks in all, term sheet through to completion, and nine of those weeks made up the conditions period.

Every legal question in this area has a route of its own: the company law side of a share sale sits with the Ministry of Corporate Affairs at mca.gov.in, what a listed acquirer must disclose sits with the Securities and Exchange Board of India (SEBI) at sebi.gov.in, and whether a particular clause works is a question for a lawyer. The clause's effect on money, on risk and on timing is the part a banker or an analyst is actually being paid to read.

What is a representation and warranty, and what is it actually promising?

Readers often hear the phrase as a single word. The phrase is two words, and the two mean different things. A representation is a statement of fact made to induce somebody to enter into an arrangement. A warranty is a promise that a stated fact is true, given as part of the arrangement itself. In ordinary transaction paper the two travel together in one numbered list, and practitioners therefore say the phrase as though it named a single object. Which remedies the difference opens to a wronged buyer is a legal question, and a lawyer answers it.

A reader with a memo to write needs the four parts of any such clause far more than the distinction. A warranty has a speaker, a subject, a date and a consequence, and the consequence is the only one of the four with money in it. The speaker is the seller. The subject is a fact about the target. The date is the day on which the fact is asserted to be true. And the consequence is that if it turns out not to have been true on that day, a claim arises. A reader who misses any of the four has not read the clause.

Notice what a warranty is not. A warranty is not a forecast. Nobody warrants that the business will do well, that the customers will stay, or that next year's EBITDA will be Rs 145 crore. A promise about the future is a different animal with a different name. A warranty looks backwards and sideways, never forwards. A warranty says: this is how things stand.

A warranty is also not a description. The confusion between the two costs money, and it is worth being blunt about. A reader who treats the warranty schedule as a portrait of the business will come away with a nice picture and no idea what the buyer has bought. Somebody who treats the schedule as a list of priced promises asks the only useful question about each line. If this turned out to be false, what would happen, and to whom? The schedule is not a description of the business, it is a promise about the business, and a promise that turns out false has a price attached to it.

Why does the seller agree to any of this? Because the alternative is worse for the seller. A buyer who cannot get promises will do one of two things: pay less, or walk. Every unwarranted risk is a risk the buyer prices, and the buyer prices it conservatively because it is guessing. A seller who signs the warranty schedule is, in effect, saying: do not discount the price for a risk the business does not carry, and if it turns out I was wrong, I will pay. A warranty schedule is a trade, and both sides know it is a trade. The negotiation of a warranty list is therefore a price negotiation conducted in prose.

Why is the remedy money rather than undoing the purchase?

Because by the time anybody finds out, the purchase has happened and cannot practically be reversed. Sit with what an unwinding would actually require. Harivansh Packaging Limited has paid the sellers, taken the shares, put its own people on the board, merged reporting lines, told its customers, drawn Rs 1,000 crore of new borrowing and started running the business. Three months later somebody discovers that a warranty about a customer contract was untrue. Giving the business back is not a real option. The business is not the same business. Some of its people have left, some of Harivansh Packaging's people are in it, and the money has been spent.

So this transaction's paper does the only workable thing. The document converts a false statement into a claim for rupees. The buyer still holds the target afterwards, so the live question is never whether to unwind the purchase but how much of the loss reaches the seller. Everything covered later about caps, baskets and time limits is downstream of that single decision. Once the remedy is money, the whole argument becomes an argument about how much money, and that argument gets settled in advance, in writing, by people who have not yet met the problem.

There is a second reason, and it is about who is in a position to know. The seller ran the business. The seller knows whether the licence renewal is stuck, whether a customer has been complaining, whether an old dispute is still smouldering. The buyer cannot know any of it from outside. A promise made by the side that knows, backed by money, moves the cost of being wrong onto the side that was better placed to be right. Shifting the cost of being wrong is the entire economic logic of a warranty schedule, and the schedule is negotiated line by line rather than accepted as boilerplate.

The boundary, stated plainly: what a court would do with a false warranty, what happens if the paper is silent on a point, and whether any remedy other than money might be available are legal questions. The company law side of a share sale in India is administered through the Ministry of Corporate Affairs at mca.gov.in, and what a listed acquirer such as Harivansh Packaging Limited must disclose about the transaction sits with SEBI at sebi.gov.in. A reader who needs the requirements or the consequences asks a lawyer.

One consequence of the money remedy is worth naming because it changes how the whole document reads. If the remedy is money, then the seller's capacity to pay is part of the protection. A promise from a seller who has taken the proceeds and distributed them is worth less than the same promise from a seller who has left part of the price behind. Seller capacity is precisely why an escrowPart of the price held by a neutral holder for a stated period instead of being paid over at completion, so that a later claim has something already set aside behind it. exists at all, and the size of one is a commercial term rather than an afterthought. The mechanics of an escrow, and how it compares with the alternatives, are covered under limitations on liability.

Which date does the warranty speak on?

A reader most often skips this question and should ask it first. A warranty asserts that something is true. True when? A document can test the same warranty at signing, at completion, or at both, and those are three different clauses that look almost identical on paper.

The distinction only matters because signing and completion are not the same day. On this transaction they were nine weeks apart. The calendar here covered twenty two weeks end to end, so signing landed at week 13 and completion at week 22, with the conditions period filling everything in between. Nine weeks is sixty three days of a real business trading: invoices raised, orders won and lost, staff joining and leaving, disputes started and settled. Sundarban Polymers did not pause because two sets of lawyers were exchanging drafts.

So follow the consequence through. A warranty tested only at signing says nothing whatever about anything that happened in those nine weeks, and a reader who never checked which date it names has read the clause without reading its most important word. Suppose the schedule at signing said there was no litigation other than what had been listed, and it was entirely true on that day. In week 17 a customer files a claim. At completion in week 22 the buyer takes a company that is being sued, and the warranty it holds is a promise about week 13.

Which is why documents often repeat the warranties. A schedule that is given at signing and given again at completion is doing twice the work: the seller is asserting the same facts on the later day too, so the nine week window is covered rather than silent. Whether a particular document repeats them, and which ones it repeats, is a matter for that document. The reader has to look, and looking takes thirty seconds.

The household version is a familiar one. A buyer inspects a flat in January and agrees to buy it, and registration happens in March. If the seller's assurance about the society dues speaks as at January, and dues pile up in February, the buyer has bought the arrears. The whole question is which month the promise is about, and it is decided by a phrase somebody wrote rather than by anything either party remembers agreeing.

Completion Accounts: what are they and what do they fix?

The two halves of the same document get confused constantly. Leave the promises for a moment and look at the other half. A warranty deals with facts that turn out to be untrue. The price mechanics deal with facts that were never asserted at all, and simply need measuring on the day.

Completion accounts are a set of accounts prepared as at the completion date, from which the adjustments to the price are computed. The parties agree in advance what they will measure, on what basis, and what happens to the price when the measurement lands above or below an agreed reference. Then completion happens, the money moves, and afterwards somebody actually prepares the accounts and works out the adjustment.

One phrase in that sentence is the thing most readers miss. Read the sentence again. The accounts are prepared after completion, so with completion accounts the price is not final on the day the money moves and the number settles later. A memo that says the sellers received Rs 1,140 crore, written on completion day, is describing a payment rather than a price. The price arrives weeks later, when the accounts are done and the adjustment is settled between the parties.

Why bother? Because two things about a business move every single day and both belong to whoever is running it. The first is working capital: the stock on the floor, the money customers owe, the money owed to suppliers. The second is net debt. Neither of them can be pinned down at the moment of an agreement made weeks earlier, and both are large enough to matter. If nobody measures them at completion, then whichever way they drift, one side gets a windfall it did not negotiate.

The working capital measurement runs against a reference figure. On this transaction the agreement set a normalised working capital of Rs 96 crore. The Rs 96 crore is a peg: a stated level deemed to be the ordinary, expected amount of working capital the business carries. The peg does two jobs at once, and both are worth stating. A peg stops a seller from stripping working capital out between signing and completion, collecting from customers, paying nobody and leaving a hollowed business behind. And it stops a buyer from claiming a windfall when the seller happens to leave more behind than expected.

The net debt measurement runs the same way against the level the transaction assumed. If net debt on the day lands above the assumption, more borrowing is coming across than the price was struck for, so the amount reaching the sellers falls. If it lands below, the amount rises. The same bridge applies on a later day: enterprise value less net debt gives equity value, so a change in net debt moves the equity value one for one.

One accounting point matters here. Completion accounts only work if both sides agree what the words mean. Working capital measured on which basis, with which provisions, treating which items as debt-like? Each of those is a measurement question. Where an accounting definition is settled in India, the Institute of Chartered Accountants of India at icai.org settles it. The document names the basis. The meaning of that basis is settled in accounting standards rather than in the agreement.

Locked Box: what does fixing the balance sheet early change?

Now the alternative, and it is genuinely an alternative rather than a variation. Under a locked box the balance sheet is frozen on a day that falls before signing. Price is agreed by reference to that frozen statement, everything the business earns or loses after that day runs for the buyer's account, and no completion adjustment is computed at all. Nothing is measured at completion, so there are no completion accounts.

The name describes the mechanism honestly. From the locked date forward, the box is shut: value is not supposed to leak out of it to the sellers. The document spells out what may and may not leave the company in the stretch that follows the locked date, and anything that does leave has its own arithmetic. The buyer is buying the business as it stood on a known day, and the days between are for the buyer's account rather than the seller's.

A locked box is not a stage of a completion accounts process and is not a faster version of one, and the choice between them decides whether there is any arithmetic after completion or none at all. Knowing which of the two a document uses is the single most useful thing about the pair, and a reader should be able to answer it within a minute of opening a document. Full comparison of the two, and the argument about which suits which situation, is covered under purchase price mechanics.

Certainty and timing are the real trade between the two. Completion accounts price the business as it stood on the day itself, and charge for that in two ways: the final number is unknown until weeks after the money has gone, and somebody has to prepare and agree the accounts. A locked box gives a number that is known before anybody signs, at the cost of both sides having to live with whatever happens to working capital and net debt in the meantime.

The household analogy that fits is a rental deposit. One landlord takes a fixed deduction agreed in advance, whatever the flat looks like at the end. Another inspects on the day the tenant leaves and settles afterwards. The first is quick, certain and sometimes unfair to one side. The second is accurate and takes three more weeks and one more argument. Neither is a better system in the abstract. The two just answer different questions, and which one an agreement uses has to be known before anybody plans around the money.

Condition Precedent: why is this not a promise at all?

A condition precedent is a state of the world that must exist before completion happens. Such a condition is not a promise, it is not a warranty, and nobody is asserting that it is true. The clause is a gate.

The Sundarban Polymers purchase carried three of them: an approval from a regulator, a requirement that no material adverse changeA deterioration in the target serious enough that the paper treats it as reason not to complete. The document itself defines what counts as one, and the definition is argued about wherever it turns up. has occurred, and sign-off from the two counterparties whose contracts move across with the business. Notice what those three have in common. None of them is something the seller is claiming to be true today. Each is something that must become or remain true before the money moves. Any approval regime's own requirements belong to the regulation layer.

The difference from a warranty is not one of degree, and this is where readers slip. A broken promise produces a claim; an unmet condition produces no transaction, so no money moves in either direction and there is nothing to claim against. Those are not two sizes of the same outcome. The two are different outcomes with different arithmetic, and confusing them produces memos that are simply wrong about the buyer's exposure.

Work it through concretely. Suppose the regulatory approval never arrives. There is no completion. Harivansh Packaging Limited does not pay Rs 1,140 crore, does not take the shares, and does not hold the target. The sellers keep their company. A warranty claim is a claim by an owner about a thing it now holds, so nobody has broken a promise and no warranty claim exists. Compare that with a false warranty: there, completion happened, the money moved, and the buyer is holding a business worth less than the one it was promised. In one case the world did not change. In the other it changed for the worse and somebody has to pay.

A gate that stays shut for a very long time raises a separate question, answered by the long-stop dateThe outside date written into the paper. Once it passes with conditions still open, a side may choose to walk. Nothing ends by itself. and by the termination provisions, set out under the transaction process and under termination. For reading a warranty schedule, only one thing matters: the conditions live in a different part of the document, do a different job, and produce a different result when they fail.

The household version. A loan sanction is a condition of a flat purchase. If the loan is refused, the seller never promised a loan, so the buyer does not sue. The transaction simply does not happen. If instead the seller said the society dues were clear and they were not, that is a promise, the buyer already owns the flat, and there is a claim about money. Same document, same afternoon, two completely different mechanisms.

The worked instance: how Rs 1,140 crore became Rs 1,137 crore

Here the abstractions become a number somebody had to wire. Run the price mechanics on the transaction itself. Start where the record starts. Enterprise value Rs 1,320 crore, being 10.0 times Sundarban Polymers' EBITDA of Rs 132 crore. Deduct that company's net debt of Rs 180 crore and the equity value agreed with the sellers is Rs 1,140 crore. Quoting enterprise value as the amount paid is the error made most often about transaction figures, so the bridge is worth running explicitly every time.

Now the two adjustments the document defined. Against a normalised working capital of Rs 96 crore written into the agreement, the business closed with Rs 108 crore on its books, 12.5 per cent above the peg, so the price moves up by Rs 12 crore. Against the Rs 180 crore of net debt the transaction had assumed, the level measured on the day was Rs 195 crore, 8.33 per cent higher, so the price moves down by a Rs 15 crore reduction for net debt.

Added together, the arithmetic is almost anticlimactic. Plus Rs 12 crore and that Rs 15 crore reduction net to minus Rs 3 crore, and the equity value moves from Rs 1,140 crore to Rs 1,137 crore. A quarter of one per cent. A reader told at the start that the two completion adjustments together would move the price by minus 0.26 per cent would reasonably have asked why anybody bothered.

The reason is the point of the whole exercise. Missing the working capital leg leaves the answer out by Rs 12 crore; missing the net debt leg leaves it out by Rs 15 crore; and the tininess of the net is exactly why both get computed rather than why either gets skipped. The individual legs are 1.05 per cent and 1.32 per cent of the Rs 1,140 crore agreed. The two legs happened to point in opposite directions and mostly cancel. Nothing in the mechanism made that happen, and next time they will both point the same way and the net will be Rs 27 crore rather than Rs 3 crore. A process that measures one leg because the answer usually comes out small is a process that has stopped measuring anything.

And now the base that every threshold is struck on

The indemnity package on this transaction is expressed in percentages, and percentages need a base. Every threshold in this transaction is struck on the adjusted Rs 1,137 crore rather than on the Rs 1,140 crore headline. On that base the cap is Rs 227.40 crore, the basketA threshold that the counted claims must together exceed before anything at all is paid. The full mechanics, and the two conventions for what gets paid once it is crossed, are worked under limitations on liability. is Rs 11.37 crore, the de minimisA floor below which an individual claim is not counted at all, so it never joins the pile that is tested against the basket. is Rs 1.14 crore, and the escrow is Rs 113.70 crore.

Struck on the Rs 1,140 crore headline instead, the identical percentages give Rs 228.00 crore, Rs 11.40 crore, Rs 1.14 crore and Rs 114.00 crore. Two figures three crore apart moved the cap by Rs 0.60 crore, the escrow by Rs 0.30 crore and the basket by Rs 0.03 crore. None of those is a large sum against a transaction of this size. Every one of them is a real amount of money that somebody would have to explain, and each of them appeared out of a base nobody named.

Naming the base beside every percentage is the habit worth acquiring now. A cap of 20.0 per cent is not a number. A cap of 20.0 per cent of the Rs 1,137 crore adjusted equity value is Rs 227.40 crore, and it is a number. A threshold quoted in somebody else's memo as a percentage with no base states no amount at all. The headline is the figure everybody remembers, so a reader who assumes the base will assume the headline.

How does a false warranty turn into rupees?

Most readers have never seen the route laid out and assume it is shorter than it is. Trace it once, in order. The fact is untrue. The buyer computes a loss. Then four separate filters stand between that loss and any money arriving.

The first filter is the disclosure scheduleThe document in which the seller lists the exceptions to its own promises. Anything properly listed there is carved out of the warranty it relates to. The disclosure schedule has a treatment of its own.. Before anybody argues about amounts, somebody checks whether the seller already disclosed the very fact the buyer is complaining about. A warranty says there is no litigation; the disclosure schedule lists three matters; the buyer discovers one of those three. The fact was told, so there is nothing to claim.

The second filter is the de minimis. Here it is set at 0.1 per cent of the Rs 1,137 crore adjusted equity value, so Rs 1.14 crore. A claim smaller than that is not counted at all: not reduced, not deferred, simply not counted, so it never joins the pile.

The third filter is the basket, set at 1.0 per cent of the same base, or Rs 11.37 crore. The claims that survived the first two filters are added up and tested against it. If the total does not exceed the basket, nothing is paid, however genuine each claim was.

The fourth is the cap, 20.0 per cent of the same base, or Rs 227.40 crore. The cap is the ceiling on everything that gets through. The escrow of Rs 113.70 crore sits alongside all four rather than inside the sequence. Instead of deciding whether the buyer is owed anything, the escrow decides how much of any payment is already set aside rather than having to be collected from sellers who have moved on.

Four separate filters sit between a real loss and a rupee received, and each one of them is a sentence somebody wrote into a document during a negotiation. Not one of them is arithmetic that fell out of the transaction. Somebody proposed a percentage, somebody else countered, and the result is now the difference between a claim that pays and a claim that does not. How each filter behaves in detail, and what happens on the far side of the basket, is worked in full under limitations on liability.

What does a reader check first?

An analyst is handed a warranty schedule. The schedule runs to forty numbered paragraphs across eleven printed sides. Where does the analyst start?

Not by reading it in order. Order is the instinct and the wrong one: reading forty warranties in order gives forty warranties and nothing at all about what the buyer holds. Three questions tell a reader more about what the buyer actually holds than reading every warranty in order would, and none of the three is answered inside the numbered paragraphs.

Question one: which date does each warranty speak on? Signing, completion, or both. One clause sitting outside the schedule usually settles it, and that clause decides whether the nine week conditions period is covered or silent.

Question two: is the schedule repeated at completion? If only part of it repeats, which part? A document that repeats the fundamental warranties and not the operational ones has made a choice, and the choice is visible in about a minute.

Question three: which price mechanic did the paper choose? Completion accounts or a locked box. The choice settles whether there is arithmetic still to come after completion, whether the quoted price is final, and what base the thresholds are struck on. On this transaction the answer was completion accounts, and Rs 1,140 crore became Rs 1,137 crore, so every threshold is struck on the later figure.

There is a fourth thing to look at, and it is the one everybody looks at last: what does the disclosure schedule say against each warranty being relied on? A warranty and its exceptions are one object, not two, and reading the promise without its carve-outs is like reading a price without a currency. The disclosure schedule is covered next.

A warranty does not have a range: it is true or it is not, and the two positions differ in kind rather than in degree. The relationship that does vary continuously in this subject, between the size of a claim and the rupees that actually arrive, is worked where the thresholds live, under limitations on liability.

How does a lender, an analyst or a household read this?

The same schedule put in front of three readers is opened at three different places. A lender, an analyst and a household reflex do not want the same thing out of it at all.

A lender reads it as backing for a loan already committed. Rs 1,000 crore is going out at Harivansh Packaging Limited's own contracted rate of 9.0 per cent, against a business the lender will never control, so what it wants to know is what happens to that business if a promise turns out false. The lender's questions are blunt. Is the remedy money, so the borrower keeps the asset and the loan keeps its collateral? Is any part of the price held back, leaving a claim with funds behind it rather than a lawsuit? And is the cap large enough that a serious problem is not simply absorbed by the borrower? Rs 227.40 crore against an equity value of Rs 1,137 crore is a fifth of the price, and a lender can form a view on whether a fifth is a lot in the context of what it has lent.

An analyst who follows Harivansh Packaging Limited reads it as a question about what has quietly been assumed into the model. The purchase brings goodwill of Rs 820 crore, being the Rs 1,140 crore paid against the target's net worth of Rs 320 crore, before anything is allocated to identified intangibles, and the allocation exercise that follows belongs to the accounting layer. The analyst wants from the warranty package a sense of how much of the price is exposed to something turning out to have been untrue, and whether the answer is bounded. A capped, basketed package with an escrow behind it is a bounded exposure. Whether the terms were well negotiated is not knowable from outside. The analyst is noting the size of the box.

Somebody already holding the shares reads it as a disclosure question. A purchase that later produces a claim against the sellers is a thing the company may have to explain, and what a listed acquirer has to put out about a transaction, and by when, belongs to SEBI at sebi.gov.in.

And at household scale, the same reflex arriving without the vocabulary. Anybody buying something that cannot be fully inspected is choosing between three things: pay less to cover what cannot be seen, take a written promise about it, or accept the risk. Most people do all three without naming any of them. A written promise is worth exactly what the person behind it can pay and what the fine print allows to be collected, so the promise and its limits are read together or not at all.