Limitation of Liability: Caps, Baskets and Time Limits

A limitation of liability clause decides how much of a proven warranty claim ever turns into a payment. Four filters stand between the two. Rs 1,137 crore carries all four marks on Harivansh Packaging Limited's purchase: a ceiling at Rs 227.40 crore, a counting floor at Rs 1.14 crore, a threshold at Rs 11.37 crore, and Rs 113.70 crore already in an account. A proven Rs 8 crore loss reaches none of it.

Warranties, disclosure and indemnities are settled separately: what a warranty is, what a disclosure removes, and what an indemnity promises. Each of them points forward at a number nobody has computed yet. The limitation clause is where those numbers get computed, and not one of them measures anything. All four are filters. A loss is whatever the loss is; these four sentences decide which slice of it travels.

Harivansh Packaging Limited put a small team on the purchase, and Ashwin Rege leads it. Devyani Kulkarni, who is the chief financial officer, receives the protection figure as the one line about this paperwork that ever reaches a board pack. The business being bought is Sundarban Polymers Private Limited. The whole problem sits in that journey: eight printed sides of drafting arrive at the board as a single rupee amount, and the single rupee amount is the least informative thing on the eight sides.

What is a limitation of liability clause actually limiting?

Start with what it is not limiting. The clause is not limiting the loss. If a warranty about the target's receivables turns out to be false and Rs 8 crore of them are uncollectable, the business is short Rs 8 crore whatever any document says. Nothing in a share purchase agreement reaches into the world and makes bad receivables good.

The clause limits the portion of that loss which is allowed to travel from the buyer back to the seller. Two quantities exist and a memo almost always merges them. The first is the buyer's real exposureThe amount genuinely at risk before any clause, threshold or account has been applied to it. It is a fact about the business rather than a fact about the paperwork., a fact about the business. The second is the buyer's recoverable exposure, a fact about the paper. The two exposures are different numbers, they move for different reasons, and confusing them is how a board comes to believe it has bought protection it does not have.

Think of a wedding hall taking a booking. The bill is six lakh rupees and the hall holds a refundable deposit of fifty thousand. Something goes wrong on the night and the loss to the couple is worth eight thousand rupees. The deposit is not the number that decides anything. A single line in the booking form decides it, and the line says the hall pays nothing on complaints below twenty thousand rupees. The deposit is real money and it is irrelevant. A different sentence stopped the claim two steps earlier. A limitation package is that booking form with four such lines instead of one, and with rupee amounts large enough that people stop reading them as sentences and start reading them as reassurance.

Where do all four marks come from?

Every one of the four is struck on the same figure, so the figure has to be right before anything else can be. The purchase moved through four steps to reach it, and the last two are the ones people forget.

Rs 1,137 crore is the only figure anything is struck on, and it carries all four marks by itself. Two tenths of it is the ceiling: Rs 227.40 crore. One hundredth of it is the threshold: Rs 11.37 crore. One thousandth of it is the counting floor: Rs 1.137 crore, and everybody writes that as Rs 1.14 crore. One tenth of it is the account: Rs 113.70 crore.

Rs 1,140 crore is the equity value as it stood before the two completion adjustments ran through it, and it is the figure a hurried reader reaches for because it is the one in the announcement. Put the ceiling fraction on that earlier number and Rs 912.00 crore of the price lands outside the ceiling. Put it on the number the sellers actually received and Rs 909.60 crore does. The Rs 2.40 crore between those two answers is four fifths of the Rs 3 crore the adjustments moved. Eight tenths of any base is exactly what a two tenths ceiling leaves behind, so the four fifths is forced rather than discovered.

The ceiling: what does a maximum leave outside?

A ceiling is the simplest of the four and also the most quoted, an unfortunate combination. The ceiling says that everything the buyer recovers through the warranty route, added together across every claim ever made, stops at Rs 227.40 crore. One claim of Rs 300 crore recovers Rs 227.40 crore. Forty claims summing to Rs 300 crore recover Rs 227.40 crore between them. The ceiling does not care how the loss arrived.

Nearly every summary of a ceiling implies that the rest of the price is at risk in a smaller way, and a ceiling says no such thing. The rest of the price is not at risk in a smaller way through this route. The rest of the price is not reachable through this route at all. Rs 909.60 crore of what Harivansh Packaging paid is outside the ceiling permanently, in every state of the world, whatever anybody proves about anything. Rs 909.60 crore is eight tenths of the purchase.

Notice what the drawing could not do. The threshold and the counting floor are on the same bar, at 6.2 and 0.6 units out of 620, and neither is wide enough to draw. The invisibility is not a failure of the drawing. The two marks that decide almost every real claim are invisible at the scale of the price, and the one mark that is easy to see decides almost nothing. The next figure changes scale so the two working marks can be looked at properly.

The counting floor: what happens to a very small claim?

The counting floor sits at Rs 1.14 crore. Below it a claim is not counted. The word doing the work is counted, not paid, and everything the floor achieves sits in the gap between those two words. A claim under the floor is not reduced, not paid in part, not held over. A claim under the floor is treated as though it had not been made, so it never adds itself to anything else.

Why would anybody agree to that? Put yourself on the seller's side for a moment. The business has changed hands and the sellers have moved on. If every small irritation could be assembled into a claim, the sellers would spend the next several years answering letters about amounts that cost more to argue about than to concede. A counting floor stops that traffic, and it stops it by making the small items invisible rather than by making them cheap. A vegetable seller who tells regular customers not to bother returning a single bad tomato has not offered a discount on tomatoes. He has removed a category of conversation.

The genuinely surprising part appears only once the figures are computed rather than read. Taken one claim at a time, the counting floor on this transaction moves no rupee whatsoever. Anything it stops is under Rs 1.14 crore, and anything under Rs 1.14 crore is also under the Rs 11.37 crore threshold, so the threshold would have produced the same nil on its own. The floor only starts to bite once claims are being added together. At that point it decides which ones are allowed into the addition. The record sets this transaction up with each claim tested by itself, so the floor has nothing to do.

The threshold: why can a real loss recover nothing?

The threshold sits at Rs 11.37 crore, and nothing is paid until the counted claims have gone above it. Now take the middle of the three worked claims and sit with it. The middle claim is the one that changes how a reader looks at a limitation clause for good.

Sundarban Polymers warranted something. The warranty was false. Harivansh Packaging is Rs 8 crore worse off as a direct result. The loss is documented, the causal link is not in dispute, the sellers accept the facts, and no disclosure covered it. Rs 8 crore is seven thousandths of the adjusted price, so it clears the counting floor comfortably and is counted. The counted claim then meets a threshold set at Rs 11.37 crore and stops.

The buyer receives nothing. Not a reduced amount, not a contribution, not a gesture: nothing, on a real, proven, undisputed Rs 8 crore loss. And the essential thing to understand about that sentence is that no part of it describes a failure. Nobody drafted badly, nobody missed a trick, no court declined to help. The package did precisely what its two sentences said it would do. If Harivansh Packaging did not want that outcome it had one opportunity to avoid it. Testing the threshold against plausible claim sizes before signing takes four lines of arithmetic.

Read the middle row again. The middle row is the only one of the three where a real loss and a nil recovery sit in the same line, and a summary never contains that row.

Does it change anything if several claims arrive rather than one?

Adding claims together changes a great deal. Everything above rests on an assumption rather than on a property of limitation clauses: each of the three claims worked here is tested against the package by itself. Testing one claim at a time is how this transaction's record sets the exercise up, and the reading keeps the arithmetic clean enough to see.

A package can just as easily add counted claims together and test the total against the threshold. Under that reading the Rs 8 crore claim stops being a dead end and becomes a contribution: it waits, and when the next counted claim arrives the two are weighed as one amount. The moment claims are added rather than tested singly, the counting floor stops being a filter with nothing to do and becomes the rule that decides which claims are allowed into the addition. A floor that changed no outcome in the earlier figure becomes the most consequential of the four, and its new job is choosing the members of a sum.

Which of the two readings applies is settled by the wording, and a reader who cannot tell from the wording has found the single most useful question to ask the lawyer on the call. The quantumHow large a claim is in money, held separately from whether the claim is any good. A claim can have an agreed quantum and still fail on liability, or the reverse. of any individual claim is not the whole of what matters; whether the document lets one claim keep company with another matters just as much, and the difference between the two readings on a set of small losses is the difference between a full recovery and nothing.

Disclosure, covered separately, works an example in which several claims are weighed together and one of them is removed by a disclosure, so the aggregate arithmetic is set out there.

Tipping or deductible: which convention is the document using?

Once the threshold is reached, one identical claim can be settled two ways, and both of the ways are unremarkable pieces of drafting.

Under a tipping convention, reaching the threshold makes the whole claim recoverable. Rs 15 crore of proven loss produces Rs 15 crore. The threshold behaves like a qualifying condition: once satisfied, it steps out of the way.

Under a deductible convention, only the part above the threshold is paid. Rs 15 crore of proven loss produces Rs 15 crore less Rs 11.37 crore, or Rs 3.63 crore. The threshold behaves like the excess on a motor policy: it is subtracted from every settlement rather than merely opened.

Nothing about the business, the loss, the evidence or the arithmetic differs between those two paragraphs. The document decides, and the document alone. The paper settles the outcome far more thoroughly than the numbers do, and nowhere more sharply than here.

The number everybody quotes, and why it is the wrong one

On a Rs 15 crore claim the two answers are Rs 11.37 crore apart. Set that difference against the claim and it reads 75.80 per cent. The 75.80 per cent gets rounded to seventy six per cent and repeated, and the repetition quietly turns the figure into a property of the two conventions. The seventy six per cent is no such property. The gap is a reading taken at one claim size, and it was taken near the top of its own range.

Work the gap at several claim sizes and the shape appears immediately. In rupees the gap is Rs 11.37 crore at a Rs 15 crore claim, Rs 11.37 crore at a Rs 50 crore claim, and Rs 11.37 crore at a Rs 227.40 crore claim. The gap does not move. Subtracting a fixed threshold from a claim removes a fixed amount. As a share of the claim, meanwhile, it collapses: 75.80 per cent, then 22.74 per cent, then 5.00 per cent.

So the honest statement about a convention is a rupee statement, not a percentage one. Choosing a deductible convention over a tipping one costs the buyer exactly one threshold, every time, on any claim between the threshold and the ceiling. The percentage is not a fact about the drafting at all; it is a fact about how big the claim happened to be. On a Rs 12 crore claim the same two conventions pay Rs 12 crore and Rs 0.63 crore, a gap of 94.75 per cent, and nobody quotes that one because nobody chose that claim size to quote.

What shape does recovery make as the claim grows?

Plot recovery against claim size for both conventions and the relationship turns out not to be a line at all. Recovery has a flat floor, then a break, then a rise, then a second flat stretch where the ceiling takes over.

Both conventions pay nothing on any claim up to Rs 11.37 crore. The flat stretch covers a large part of the range in which real warranty claims actually arrive, and that is the practical reason the threshold matters more than the ceiling does. Then the two conventions part company in a way worth watching closely. The difference between them is not a difference of amount but a difference of kind. The tipping line jumps. At a claim of Rs 11.35 crore it pays nothing, and at Rs 11.40 crore it pays Rs 11.40 crore, so five hundredths of a crore of extra loss buys eleven and a third crore of extra recovery. The deductible line does no such thing: it leaves the floor smoothly and climbs from zero, paying Rs 0.03 crore at that same Rs 11.40 crore claim.

The record describes the shape as a floor followed by a step, and that is right for one of the two lines and wrong for the other. Only the tipping convention has a step in it, and having a step is exactly what makes it a tipping convention. Once both lines are drawn the distinction stops being a definition to memorise and becomes something visible.

Zoom into the first Rs 30 crore and the two behaviours separate properly. The step on the tipping line is eleven and a third crore tall at a single point on the axis. The deductible line passes through the same point at zero and walks upward. Every rupee of vertical distance between them, everywhere on this chart, is the same Rs 11.37 crore.

Rs 238.77 crore is worth a sentence of its own, and it is the one place where the convention stops mattering. The tipping line reaches the ceiling at a claim of Rs 227.40 crore. The deductible line has to climb an extra threshold to get there, so it reaches the ceiling at Rs 238.77 crore. Above a claim of Rs 238.77 crore both conventions pay Rs 227.40 crore and the choice of drafting is worth nothing at all. Rs 238.77 crore is twenty one hundredths of Rs 1,137 crore, and the figure is forced rather than being a separate fact about the deal. The ceiling fraction and the threshold fraction added together on one base give exactly that.

How long do the promises last, and what does this record actually publish?

A time limit does something none of the other three filters does. The ceiling, the floor and the threshold all ask how big. A time limit asks when, and it converts a live promise into a dead one at a date. A claim that would have recovered Rs 50 crore on a Tuesday recovers nothing on the Wednesday after the window shuts. A time limit is the only filter whose effect is total and instantaneous.

The honest part sits uncomfortably with what the heading above promises. The record for this transaction publishes no warranty survival period at all. There is exactly one period in it, the eighteen months for which the escrow account is held, and an escrow hold is a different quantity from a claim window. One says how long money stays parked. The other says how long a right stays alive. A package can easily have a window longer than its hold, leaving the unfunded part of the ceiling exposed for the remainder, or a hold longer than its window, leaving money sitting after the last claim could have been brought. Which of those describes Harivansh Packaging's purchase cannot be settled from what the record publishes, and a survival period arrived at by guesswork would sit in a memo looking exactly like one that had been read.

The two clocks weighed against each other, and against the transaction calendar, are covered under termination. The point needed here is narrower, and it is a point about reading. When a summary states that the warranties are protected for eighteen months, the first question is whether the eighteen months is a claim window or an account. A claim window and an escrow hold are printed in the same font and they answer different questions.

Which of these numbers is money, and which is a promise?

The ceiling is a limit on an obligation. The escrow account is cash that has already been set aside and is sitting where neither side can spend it. A limit and a funded account are not two versions of the same thing, and a reader who treats them alike has skipped the only question about a limitation package that concerns collection rather than entitlement.

On this transaction the account holds Rs 113.70 crore, being one tenth of Rs 1,137 crore, and the ceiling stands at Rs 227.40 crore, being two tenths of it. So the record, and the brief that describes it, arrive at a tidy statement: exactly half the ceiling is money and the other half is a promise from sellers the buyer no longer has any hold over. The half and half statement is arithmetically correct, and it describes the one outcome that almost never happens, a recovery that reaches the ceiling.

Half the ceiling is money only if the recovery is the ceiling. Below that, the account is not paying half of anything. An escrow funds the first Rs 113.70 crore of whatever gets recovered, so a recovery of Rs 3.63 crore is a hundred per cent funded, a recovery of Rs 15 crore is a hundred per cent funded, and a recovery of Rs 113.70 crore is a hundred per cent funded. The funded share only starts to fall once recovery passes Rs 113.70 crore, and it reaches fifty per cent at exactly one point, the ceiling itself.

Rebuild the claim on the measure that matters, the funded share of an actual recovery rather than of a hypothetical maximum, and the naive reading reverses. On all three worked claims the funded share is a hundred per cent. Two of them recover nothing and the third recovers Rs 3.63 crore or Rs 15 crore depending on the convention. The worry about an unsecured half is real, but it is a worry about the top of the range, and it does not touch any claim the earlier blocks looked at. To reach it under a tipping convention the buyer needs a proven claim above Rs 113.70 crore, a tenth of the price. Under a deductible convention it needs a claim above Rs 125.07 crore, eleven hundredths of the price, and that figure is forced too: it is the account fraction and the threshold fraction added together on one base.

Do these four marks reach every promise in the document?

Here is the check that gets skipped most often, and it is skipped because it cannot be done with arithmetic. A limitation package applies to the promises the clause says it applies to, and a share purchase agreement contains several different kinds of promise. Statements of fact about the business are one kind. Promises about who validly holds the shares and whether they are free of anybody else's claim are another. Promises to do or not do something after completion are a third. A covenant to pay a tax bill that relates to a period before completion is a fourth.

Nothing in the four rupee amounts indicates which of those the ceiling, the floor and the threshold actually cover, and the clause can carve any of them out. A package can put every promise inside the same ceiling. A package can lift some of them out and give them a higher one, or none at all. A package can send one category to a different threshold entirely. All of those are ordinary drafting positions. Which of them is usual is a negotiating question, and no figure in this record answers it.

A reader can still notice the size of the consequence. If a promise sits inside the package, a proven Rs 8 crore breach of it recovers nothing on this transaction. If the same promise sits outside the package, the same Rs 8 crore breach recovers Rs 8 crore. The identical loss, on the identical facts, is worth either everything or nothing depending on a list that appears nowhere in any of the four numbers. A summary quoting the ceiling and stopping there is therefore not a summary of the protection at all.

The carve-out list also matters for how the recourseThe ability to go after somebody for money once things have gone wrong. Where it is absent, the disappointed party simply keeps the loss. works in practice. A claim that falls inside the package and inside the escrow window can be taken out of money already sitting in an account. A claim on a promise carved out of the package has no account behind it and has to be collected from the sellers themselves, who by then have been paid and have moved on. Two claims of identical size can therefore be worth very different amounts of trouble, and neither the ceiling nor the threshold hints at it.

Do the neat multiples between the four marks carry any meaning?

Set the four marks beside each other and some pleasing relationships appear. The ceiling is twenty times the threshold. The account is exactly half the ceiling. The ceiling is two hundred times the counting floor. Reading those multiples as evidence about what the two sides were worried about is tempting, and the temptation is worth resisting hard.

All four marks are fixed fractions of one base, so every one of those relationships is forced: the ratios are just the fractions divided by each other, and nothing else got a vote. Two tenths over one hundredth is twenty. One tenth over two tenths is a half. The marks would stand in the same relationships if the price had been Rs 400 crore or Rs 4,000 crore. A ratio between two figures that are both slices of the same cake carries information about the slicing instruction and none at all about the cake, the baker or the appetite.

There is a small trap hidden inside the two hundred as well. The two hundred holds against the unrounded Rs 1.137 crore. Against the Rs 1.14 crore everybody actually writes, the ceiling is 199.47 times the floor. The mismatch is not an error in anybody's arithmetic; it is a rounding sitting where a checking reader will divide two printed numbers and get a third one. The fractions themselves, together with the base they were struck on, are the whole of the available evidence about how a package was designed. Ratios taken between the marks add nothing to it.

How does a practitioner actually use this?

Take the three people who read this clause for a living, and notice that each of them is asking a different question of the same four marks.

An investment banking associate, handed the agreement and asked for a summary on one sheet, is not being asked to say whether the package is good. The summary that earns its place has four rows and a worked line: the base, the four marks in rupees, whether the threshold tips or deducts, and what the package pays on a claim of Rs 5 crore, Rs 15 crore and Rs 50 crore. On this transaction that last row reads nil, Rs 3.63 crore and Rs 38.63 crore under a deductible convention. Three numbers say more than three paragraphs, and they are the three numbers a partner will ask for.

A credit analyst at a lender is asking a narrower question: how much of this can turn into a cash outflow, and when. The ceiling is the outer bound on the buyer's exposure through this route, and Rs 113.70 crore of it is already parked and therefore already gone from usable cash. A covenant test turns on the unfunded portion, the part that could become a payment out of operating cash at a moment nobody controls. Note the direction here: for the buyer, the funded half is the comfortable half; for the buyer's lender, the funded half is the half that has already left.

An equity analyst covering Harivansh Packaging Limited is asking whether any of this ever becomes visible. A ceiling on somebody else's obligation is not a liability of the buyer and appears nowhere. An amount held in an escrow account is real and restricted. And a claim that has been made but not settled is a contingent item whose reporting is settled by the Institute of Chartered Accountants of India. The practical instruction is to read a disclosed protection figure as a bound on a possible inflow and never as an asset.

The household version of the same three readings is easier to feel. A tenant who pays a returnable security deposit to a landlord knows exactly where that money is and cannot spend it. The landlord treats the same amount as somebody else's money he happens to hold. The bank looking at the tenant's account sees a balance that has already left. One amount, three completely different readings, and none of them is wrong.

What does a reader compute before believing a protection figure?

Four checks, and a figure quoted without all four is not yet a number.

There is a fifth thing, and it is not a check because nothing on this transaction answers it. How long the promises stay live is the missing input, and the honest way to hold a missing input is to name it in the memo rather than to fill it with a plausible period. A summary that says the survival period is not stated in the papers reviewed is more useful than one that quietly assumes a common one. The first invites somebody to go and look; the second stops anybody looking.

Harivansh Packaging Limited, Sundarban Polymers Private Limited, Ashwin Rege and Devyani Kulkarni are invented.
Educational material. Not advice on any investment, tax, budget or market position.