Material Information in a Deal: What Cannot Be Shared

Information is material when a reasonable investor would want it before dealing in the shares. Harivansh Packaging Limited is listed, so a live purchase of another business is that kind of information about it. The information counts from the first approach rather than from the announcement. The Securities and Exchange Board of India (SEBI), at sebi.gov.in, sets what counts as material, what must be disclosed and when, and what may not be done with the information before publication.

The finance version is only an ordinary situation with more zeroes on it, so start with the ordinary situation. A cousin who works at a bank knows on Tuesday that the shop across the road has agreed to sell itself to a larger chain. Nobody has told the neighbourhood. On Wednesday the shopkeeper next door offers to buy that same shop's stock at last month's prices. The cousin knows a fact that would change what the neighbour offers, and the neighbour does not know that fact. The gap between the two of them is the whole subject. Everything technical that follows is machinery built around that one uncomfortable moment.

The worked example throughout is Harivansh Packaging Limited, an invented manufacturer. The company is listed on both Indian exchanges and it makes rigid and flexible packaging for food and personal care customers. Its promoter and promoter groupThe founding holders of an Indian listed company and the people and entities associated with them, disclosed as a single block on the shareholding statement. hold 58.0 per cent of it, so the free floatThe part of a listed company's shares held by anybody outside the promoter and promoter group, and therefore the part that actually changes hands in the market. is 42.0 per cent. On an illustrative share price of Rs 300/-, carrying its stated as-of date, that is a market capitalisation of Rs 5,400 crore, of which the free float is Rs 2,268 crore. The target it is buying is Sundarban Polymers Private Limited, unlisted. Sundarban Polymers makes flexible packaging films and sells to some of the same customers.

Hold on to the free float figure. The free float is what makes the controls necessary rather than merely tidy. There are people out there, right now, buying and selling Rs 2,268 crore worth of a company on the information they have. Every one of them is trading against whoever knows about the purchase, and the only thing standing between those two groups is a set of controls that somebody has to actually operate. The controls are the subject. Whether they work is a question about people, dates and one list.

Four things follow, in order. First, what makes information material as a mechanism. Second, a first reading on how large this particular purchase is against its buyer, with the base of every ratio named and the arithmetic stopped at the exact point where it runs out. Third, who holds the information and how that is recorded. Fourth, the two written records a transaction leaves behind, the decision logA dated record of what a transaction chose, who chose it and what they were looking at when they chose. and the assumption registerA dated record of what a transaction believed to be true but had not yet verified, with a named holder for each item., and why they are two documents and not one.

What makes information about a transaction material?

Take the mechanism first and leave the legal standard entirely alone. The mechanism and the standard are different objects that share a word. As a mechanism, information is material when knowing it would change what somebody is willing to pay. That is it. Not whether it is interesting, not whether it is confidential, not whether somebody would be embarrassed if it got out. The test is about price, and it is about the price a reasonable person would arrive at with the information against the price they would arrive at without it.

Run the test on Harivansh Packaging Limited. A purchase is being negotiated that would add Rs 880 crore of revenue to its own Rs 3,180 crore and Rs 132 crore of earnings before interest, tax, depreciation and amortisation (EBITDA) to its own Rs 477 crore. Rs 1,000 crore of new borrowing at a contracted 9.0 per cent is going into the funding alongside Rs 140 crore of the company's own cash, so borrowings would go from Rs 740 crore to Rs 1,740 crore. On the locked figures it moves earnings per share from Rs 12.50/- to Rs 12.14/-, a dilution of 2.9 per cent, and the target's profit after tax of Rs 61 crore inside that chain is a rounded figure, stated as rounded and not used as the base of an exact bridge.

A purchase that changes the size, the borrowings and the earnings of a listed company is precisely the kind of information a reasonable investor would want before dealing in its shares, and it is that kind of information from long before anybody announces it. Somebody deciding today whether to buy Harivansh Packaging shares at Rs 300/- would want to know that its borrowings are about to nearly triple in absolute terms. Whether they would then buy or not buy is their business. The point is only that they would want to know, and the mechanism asks nothing more than that.

Now the boundary, stated in the same breath. Teaching the mechanism and then drifting into the standard does real damage. Whether a given fact is material as SEBI uses the word, what a listed company must disclose about a transaction, when it must do so, and what may not be done with the information before it is published, are set by the Securities and Exchange Board of India and published at sebi.gov.in. Each of those is settled at the source rather than here, and that boundary is the teaching rather than a gap in it. A reader who leaves here able to describe the mechanism and knowing exactly where the standard lives has got the useful half of this subject. A reader who leaves with a remembered threshold has got something worse than nothing.

The word that goes with material through the rest of this guide is price sensitiveDescriptive of information that would move what somebody is willing to pay if they had it. The everyday phrase for the mechanism described here., and the word that goes with it in the other direction is unpublishedNot yet made available to everybody at once through the route a listed company uses to tell the market things. Told to a few people is not published.. Information can be price sensitive and published, in which case the market has it and there is nothing to control. Information can also be price sensitive and unpublished, and that is the state a live transaction sits in for months. The controls exist for exactly one of those four combinations, and they exist for the whole time it lasts.

When does information about a transaction start being material?

A transaction runs through seven milestones in one fixed order: approach and confidentiality, indicative offer and term sheet, confirmatory diligence, documentation, signing, the conditions period, completion. The instinct most people arrive with is that the information becomes serious somewhere in the middle, once it looks likely. The instinct is exactly backwards, and correcting it changes how somebody reads every one of those milestones.

Think about what an outsider would pay for. An outsider would not pay for certainty; certainty is free to everybody once the transaction is announced. An outsider would pay for the early, uncertain, nobody-else-knows version. Only that version has any advantage attached to it. The information is at its most valuable to somebody who should not have it at the exact moment it is least certain, and that moment is the approach. The controls therefore start at the first conversation rather than at the point where the transaction starts to look real.

Follow it through the milestones on this transaction. At the approach, Harivansh Packaging Limited has done nothing but ask whether Sundarban Polymers Private Limited would consider a conversation. There is no price and no structure. There is, however, a listed buyer with a live interest in a specific target, and that alone would change what a reasonable person would pay for the buyer's shares. By the term sheet there is a shape: Rs 1,320 crore of enterprise value, being 10.0 times the target's EBITDA of Rs 132 crore. By diligence there are findings. By documentation there is paper. At signing there is a commitment. Not one of those steps makes the information material for the first time. The information already was material at the approach.

Where publication falls in that line is not fixed by the milestone order. The point at which a listed company must publish, and what it must publish, is set by SEBI at sebi.gov.in. Until publication happens, everything before it is unpublished and price sensitive, and the controls run across the whole of it.

Two other things are worth naming about the shape of that line. Twenty two weeks ran from the term sheet to completion on this transaction, of which the conditions period was nine. Transactions vary enormously in how long they take, so twenty two weeks is this transaction's own elapsed time and not a benchmark for any other. And the milestones are not evenly spaced in time even though they have to be drawn evenly to be read. How long the approach and the diligence ran is not recorded anywhere on this transaction, and an unrecorded duration is better named than estimated.

How is a first reading taken on how large this transaction is?

Here is where a transaction team reaches for arithmetic, and here is where the arithmetic has to be handled carefully. The reading is called relative sizeA ratio putting the transaction, or the target, against the buyer, used as a first sense of scale rather than as a test that settles anything., and it is a first sense of scale. Relative size is not a test, it decides nothing, and it produces a different number every time the denominator changes. The last of those three properties is the one that causes trouble.

Do it properly for Harivansh Packaging Limited, and name the base on every single line. On what is being paid, the equity value of Rs 1,140 crore that actually reaches the sellers is 21.1 per cent of a market capitalisation of Rs 5,400 crore. The market capitalisation is illustrative and carries its stated as-of date. Note the figure used there: Rs 1,140 crore, not Rs 1,320 crore. Enterprise value of Rs 1,320 crore less the target's net debt of Rs 180 crore is the Rs 1,140 crore of equity value, and it is the equity value that reaches the sellers. On the enterprise side, comparing like with like, Rs 1,320 crore against Harivansh Packaging's enterprise value of Rs 6,000 crore is 22.0 per cent.

Now put the two businesses against each other instead of the two prices. Sundarban Polymers Private Limited's revenue of Rs 880 crore is 27.7 per cent of Harivansh Packaging's Rs 3,180 crore. Its EBITDA of Rs 132 crore is 27.7 per cent of Rs 477 crore. Its earnings before interest and tax (EBIT) of Rs 98 crore is 28.9 per cent of Rs 339 crore. Its net worth of Rs 320 crore is 19.4 per cent of Rs 1,650 crore.

The same purchase reads as 19.4 per cent on one base and 28.9 per cent on another, a spread of nearly ten percentage points, and every one of those numbers is correct. That is why more than one base gets computed. A reader who has been handed a single ratio and has not asked what sits underneath it has not looked at the transaction; they have looked at somebody's choice of denominator. The habit worth building is simple and slightly tedious: never say a percentage of this transaction out loud without saying of what in the same sentence.

One more reading belongs here because it is the one an existing holder would care about most, and it is not a percentage of the buyer at all. The funding brings Rs 1,000 crore of new borrowing at a contracted 9.0 per cent against opening borrowings of Rs 740 crore. The new borrowing is 135.1 per cent of what the company already owed. Borrowings go from Rs 740 crore to Rs 1,740 crore and cash goes to nil, so leverage on the acquirer's own EBITDA of Rs 477 crore moves from 1.26 times to 3.65 times. On a consolidated basis, where the target's Rs 180 crore of net debt comes across too, Rs 1,920 crore over combined EBITDA of Rs 609 crore is 3.15 times. Both of those are honest and they are struck on different bases, so each one is named every time it is used.

Why do two of those readings land on exactly the same number?

Look again at the revenue line and the EBITDA line. Sundarban Polymers Private Limited is 27.7 per cent of Harivansh Packaging Limited on revenue, and 27.7 per cent on EBITDA. Two different measures, two different pairs of figures, one identical answer. A reader who notices that and shrugs has missed something they can actually use. A reader who notices it and can say why has just demonstrated the difference between reading figures and repeating them.

The reason is a property of the two businesses that was chosen deliberately in this record. Both run a 15.0 per cent EBITDA margin. Harivansh Packaging earns Rs 477 crore of EBITDA on Rs 3,180 crore of revenue, a margin of 15.0 per cent. Sundarban Polymers earns Rs 132 crore on Rs 880 crore, the same 15.0 per cent. So the EBITDA ratio is not an independent reading at all. The EBITDA ratio is the revenue ratio with both the top and the bottom multiplied by the same 0.150.

Multiplying both sides of a ratio by the same number cannot move the ratio. The second reading therefore carries no information the first one did not already carry. It looks like corroboration and it is arithmetic repeating itself. A case that this target is about 28 per cent of this buyer has one observation supporting it, not two.

Turn that around and the point becomes useful rather than merely tidy. Two ratios landing on different numbers is the interesting case. A difference between them shows the two businesses are shaped differently. The EBIT line does exactly that here, at 28.9 per cent rather than 27.7 per cent. Depreciation and amortisation sits differently across the two businesses. Sundarban Polymers carries Rs 34 crore of it against Rs 132 crore of EBITDA. Harivansh Packaging carries Rs 138 crore against Rs 477 crore. The gap between 27.7 and 28.9 is a fact about capital intensity, and it survives being noticed.

The habit here generalises well past transactions. When two independent-looking measures agree exactly, the question worth asking is whether they are actually independent. Two shops in the same street reporting identical growth might both be growing, or there might be one bus route that feeds both of them. The finance version and the street version are the same question, and the answer is worth having in both.

Where does the arithmetic stop, and who answers the rest?

The most important move in the whole of this subject is knowing where to stop. Every reading in the last two sections says the same thing: this purchase is large relative to Harivansh Packaging Limited. Somewhere between a fifth and rather more than a quarter of the buyer, on any of the bases available, funded by new borrowing worth 135.1 per cent of what the company already owed. The finding is real and it narrows the question usefully.

Not one of those readings says whether the information is material as SEBI uses that word. Materiality in that sense is a standard set at sebi.gov.in and not a ratio anybody can compute. The arithmetic reaches the door. The arithmetic does not open it. A transaction team that computes six ratios and concludes from them that the information must be material has done a reasonable thing and has not done a sufficient one, and a team that computes six small ratios and concludes the information is therefore not material has done something considerably worse.

Now the practical half, meaning what a person actually holding this information needs. Three things are addressed by every regime that deals with unpublished price sensitive information anywhere: dealing in the shares oneself, passing the information on to somebody else, and suggesting that somebody else deal without telling them why. Dealing, passing on and suggesting are the three shapes the problem takes. SEBI sets what is actually prohibited, for whom, with what exceptions and with what consequences, and publishes all of it at sebi.gov.in.

The one instruction that can honestly be given is this: assume the information may not be acted on, and read the current text. That is not a hedge and it is not laziness. The instruction follows from where the consequences land. If a company gets its process wrong, the company deals with it. If an individual gets the use of unpublished information wrong, the consequence attaches to that individual personally, and no amount of having been told something reassuring by a colleague, a website or a teaching note transfers it back. When the cost of being wrong sits on one person, the source that person relies on should be the one that sets the rule.

There is a household version of this that makes the shape obvious. Where it is unclear whether an electrical job needs a licensed person, the answer is not to reason from how simple the wiring looks. The answer is to check. The consequence of being wrong falls on the householder and is not proportionate to how confident they felt. The transaction version has the same structure, with the added feature that the person who might be harmed by that confidence is a stranger on the other side of a trade.

Who is on the list of people holding it, and how is that list kept?

A transaction spreads. It has to. Harivansh Packaging Limited cannot buy Sundarban Polymers Private Limited with two people, and every additional person who is told is one more place the information can go. Trying to keep the number down to nothing fails. The response is to know exactly who has the information, when they got it, and why. The record that does this is usually called an insider listThe maintained record of every person who has been given information about a transaction before it is published, with the date each was told and the reason., and the mechanism is worth understanding whatever it is called locally.

The rule of maintenance is one sentence. Every person who is told is added at the moment they are told, with the date and the reason they were told. Devyani Kulkarni, chief financial officer of Harivansh Packaging Limited, goes on at the approach because she is running it. Ashwin Rege goes on at the approach because he leads the transaction team. A finance analyst goes on in week two of this transaction because somebody needed the funding modelled and the analyst could not model it without knowing what was being funded. Each of those is a name, a date and a reason, written down while it is still true.

The list works in two directions at once, and both of them fail if it is not kept as it happens. Looking forward, it tells the company who to remind, who to include in the next set of instructions, and who eventually needs to be told the matter is over. Looking backward, it is the record that exists if anybody ever asks how information travelled. The second use is the one people underrate. Looking backward only matters on the days it matters enormously.

Then there is the part nobody builds properly: coming off the list. A releaseThe act of formally taking somebody off the list of people holding unpublished information, recorded with a date, once the information no longer constrains them. is an act somebody performs, on a date, for a reason, exactly as being added was. A release does not happen because a person changed roles, or left the company, or simply stopped being invited to meetings. A list that only ever grows and from which nobody is ever removed stops carrying any meaning at all. The list stops describing who actually holds anything and becomes a list of everyone who was ever in the room.

The everyday version of this is a household keeping spare keys. The useful record is not how many keys were cut. The record worth having is who holds a key now, and when each of them got it. A household that cut twelve keys over eight years and never wrote down who returned theirs does not have a key record; it has a number. The transaction version has the same failure mode and a much larger cost attached.

What is a Decision Log, and what does it record?

A transaction runs for months and makes dozens of decisions. Most of them feel obvious on the day and none of them feels worth writing down at the time. On the day, everybody in the room knows why. The shared understanding in the room is exactly the illusion the record exists to defeat. A decision log is a dated record of what was decided, by whom, and on what basis, kept while the decision is being made rather than reconstructed after it has consequences.

Three fields, and the third is the one that does the work. The decision itself is usually recoverable later anyway. The outcome leaves traces in the offer, the paper and the bank statement. Who decided it is sometimes recoverable and sometimes not. The basis they decided on is never recoverable. The basis lived in somebody's head and in a version of a spreadsheet that has been overwritten nine times since.

A transaction makes its decisions over months, the people who made them move on, and without a log the reasons are gone while the consequences remain. That asymmetry is the entire argument. Consequences are durable and reasons are not, so the reasons have to be written down. Here is what the log carries on this transaction, in the milestone order the sequence uses.

Read the last column on its own and the log's purpose becomes visible. Every entry there is a belief about the world held on a particular day. Some of them turned out to be right. If one of them turned out to be wrong, the log shows that the decision followed reasonably from what was believed. A decision that followed from nothing at all is a completely different finding. The basis column has also started doing a second job, and doing it badly. Separating the two jobs is the problem the assumption register solves.

What is an Assumption Register, and why is it a separate document?

An assumption register is a dated record of what the transaction believed to be true and had not verified, with a named holder for each item and, crucially, a written statement of what would confirm it and what would disprove it. The register sits beside the decision log rather than inside it, and it is a separate document because it answers a different question and has a different life cycle.

Start with what an entry is. An item on an assumption register is an open question, not a fact, and writing it there is an admission that nobody has checked it yet. That admission is the whole value. A transaction is full of things everybody is quietly assuming: that the earnings figure being paid a multiple on is a level rather than a peak, that consents will be given, that a lender's indicative terms will survive documentation, that customers shared by both businesses will stay. Each of those is load bearing, each of them is unverified at the point the shape is agreed, and each of them is invisible unless somebody writes it down as unverified.

Four fields, and the fourth is the one that gets dropped: what is believed, who holds it, what would confirm it, and what would disprove it. The last field is the one that turns a register into a working document. A written disproof test is what lets a diligence finding actually close an item. Nobody said in advance what finding would count, so an entry with no disproof written down cannot be closed by anything diligence finds. The entry drifts all the way to completion as an unexamined belief wearing the clothes of a recorded one.

Look hard at the last row. The last row is the most instructive one in the table. Its confirming evidence is not available yet and cannot be available before completion. A weak register would either delete that row for being uncloseable, or quietly mark it confirmed because nobody objected. A working register keeps it open and says plainly that nothing available before completion settles it. An assumption that cannot be tested in time is a different and more useful thing to know than an assumption that has been tested and passed, and only a register that is willing to hold open items shows which of the two is in hand.

The everyday version is a household buying a second-hand scooter. The decisions are the price and the day. The assumptions are that the service history is complete, that the registration transfers without a problem, and that the seller is the person on the papers. The first two can be checked before paying and the third is checked by looking at one document. The purchase is never ruined by the decision about price. The ruin comes from an assumption nobody wrote down and therefore nobody checked.

Decision Log vs Assumption Register: which record answers which question?

Both records have now been taught in their own right, so the comparison can be made properly. A decision log records what was chosen. An assumption register records what was believed. The two descriptions sound close enough that merging the records looks like sensible housekeeping, and merging them destroys most of what either record was for.

The two documents differ on four things at once. The two records answer different questions: what was chosen, against what was taken as true. The two are written by different people at different moments. A log entry is written when a choice is made. A register entry is written when somebody notices they are relying on something unverified, usually earlier and often by somebody junior. The two close differently. A decision does not close at all, it simply stands. An assumption closes when evidence arrives that matches the disproof field. And they are read by different people afterwards, for different purposes.

Now the reason the separation earns its keep. The reason is entirely about what happens when something goes wrong. A decision that went badly because a well-tested assumption turned out false is a completely different failure from a decision made against an assumption nobody ever checked, and only two separate records let anybody tell those two apart afterwards. The first is a business that took a reasonable risk and lost. The second is a process that does not work. The remedies are not remotely alike, and a company that cannot distinguish them will apply the wrong one, usually by tightening the part that was already fine.

Put the two records on two axes and every outcome lands in one of four cells. The decision log answers the vertical question, was the decision sound on what was known at the time. The assumption register answers the horizontal one, did the belief underneath it hold. A single merged document answers neither axis cleanly, so nothing can be placed in a cell at all, and the review turns into a conversation about who remembers what. The same separation is what a transaction review depends on, when the quality of a process is assessed independently of how the transaction turned out.

One cell deserves naming on its own: the cell nobody investigates. A weak decision resting on an assumption that happened to hold produces a good outcome, and a good outcome is never reviewed. The process failure sits there untouched and gets repeated, with a different assumption underneath it next time. Two separate records at least make that cell visible to anybody who reads them side by side rather than waiting for a bad result to prompt the reading.

When does information about a transaction stop being material?

Two routes out, and no third one. The first is publication: once the information has been made available to everybody at once, there is no longer an inside and an outside, and the whole apparatus described here has nothing left to protect. The second is that the transaction ends without ever being published and the fact of it stops bearing on anything, at which point what remains is history rather than an advantage.

Neither route runs itself, and this is where transactions get careless at exactly the moment everybody has stopped concentrating. Release is an act somebody has to perform: a name, a date, a reason, recorded exactly as the addition was recorded, and it does not happen because the transaction ended or because a person moved on. By the time a transaction completes, the team is tired, the interesting part is over, and the last administrative job on the list is the one that formally tells eleven people that a constraint they have been living under has lifted. The release gets skipped, and skipping it damages the record in both directions.

Forward, the people concerned do not know where they stand. Somebody who was told about this transaction in week three and heard nothing since has no way to know whether the constraint still applies. Either they behave as though it applies forever and carry an unfair burden, or they decide for themselves that it has lapsed and carry a worse risk. Backward, a list that nobody was ever removed from cannot answer the question it was built to answer. The list no longer describes who holds anything.

Note the one thing that is not on the list of routes out. Time passing is not one. A transaction that was negotiated and abandoned quietly eighteen months ago may still be information nobody outside has, and whether it has stopped bearing on anything is a judgement somebody has to make and record rather than something that happens by default. Nor is a person leaving the company a release. A departure is the moment a release matters most, and the company has just lost every other means of reminding them.

When it stops being material is set by SEBI at sebi.gov.in, as is what publication means and what must be published. What stands here is the mechanism: two routes out, both of which require somebody to act, and a record that is only worth keeping if coming off it is as deliberate as going on it was.

How does a lender, an analyst or a board member actually use any of this?

Four readers, four first questions, and none of them is asking what the others are asking. Seeing that is the fastest way to understand why the records are shaped the way they are.

A lender being asked to provide the Rs 1,000 crore of new borrowing reads the assumption register before it reads the decision log. Its exposure is to whether the Rs 132 crore of EBITDA it is lending against is a level that repeats, and that is an entry on the register with a confirming test attached, not a decision anybody made. A lender that is handed a clean register with named holders and written disproof tests is looking at a borrower that knows what it does not know. A lender handed a register with four rows all marked confirmed and no disproof fields learns something too, and it is not reassuring.

An analyst covering Harivansh Packaging Limited after publication reads the size readings first and immediately checks the base of every one. A purchase reported as 27.7 per cent and a purchase reported as 19.4 per cent are the same purchase described two ways. Then the funding: Rs 1,000 crore of new borrowing against opening borrowings of Rs 740 crore, leverage from 1.26 times to 3.65 times on the acquirer's own EBITDA. The size and the funding are the two facts that change a model, and neither of them requires an opinion about whether the transaction was a good idea.

A board member reads the decision log, and reads it before the outcome is known rather than after. The question a board is actually answering is whether the process that produced this transaction is one the company should keep using, and the log is the only artefact that carries the answer. Somebody reading it eighteen months later, after a result, cannot help reading backwards from the result. Somebody reading it during the conditions period is reading the process on its own terms.

And an existing holder of Harivansh Packaging Limited shares, who is the person all of this is ultimately for, uses none of these documents and is affected by all of them. An existing holder owns part of the Rs 2,268 crore of free float. Their entire interest in this machinery is that the information reached the market properly. Whatever price they dealt at was then a price formed with the same facts everybody else had. The controls are not a compliance overhead sitting on top of the transaction. The controls are what make the price on the screen mean anything at all to somebody who is not in the room.