Event Risk: The Risk That Sits Outside the Model

Event risk is the chance that something happens to the company itself which was never in the model, so the model is not wrong by a margin but wrong about its subject. An event replaces an input rather than moving one, so a band around volume or margin cannot hold it. One purchase carries a business from net cash into borrowings of plus Rs 408 crore.

Underneath that answer is a distinction about kinds rather than sizes. A forecast carries uncertainty about what its inputs will turn out to be, and that uncertainty is measured, ranged and reported all the time. Event risk is a different creature living in the same house. Event risk is the chance that the company acquires a second business, returns a quarter of its cash to holders, issues shares to people who are not on the register yet, or hands the running of itself to somebody new. None of those is a value the model could have taken. Each is a different sheet arriving.

Sarvani Coatings Limited, an invented maker of decorative paints and industrial coatings, publishes the year three balance sheet used throughout. Four further states of that same sheet stand beside the published one, and none of the four has been carried out or proposed by anybody. Each exists to be measured against the others on a single scale.

What is event risk, when a model already carries risk of its own?

Any forecast has uncertainty already written into it. Volume growth might land at 4 per cent or at 8 per cent. Gross margin might hold at 46.0 per cent or give a point back. Somebody has usually flexed those, printed a range and called the range the risk. Flexing volume and margin is real work and it is worth doing.

Event risk is not more of that, and it is not the same measurement made less confidently. Event risk is the chance that the company being modelled stops being the company that was modelled. The sheet described a paints and coatings maker with Rs 2,415 crore of revenue, Rs 240 crore of borrowings and Rs 312 crore of cash. After a purchase it describes a paints and coatings maker plus somebody else, with different borrowings, a different revenue mix and an entry for goodwillThe accounting entry that arises when a business is bought for more than the net assets recorded inside it. Measuring it, and carrying it afterwards, is a matter for the accounting rules and is not settled here. that was not there yesterday.

Think about a household planning next year on one salary. The uncertainty everybody sizes is whether the increment lands at six per cent or at ten. A second household member moving in reshapes the year, and so does a house loan, and so does one earner leaving to start something. None of those is an increment at all, so widening the increment range never reaches them. Each one is the household becoming a different household. The finance version behaves identically, and the arithmetic below shows by how much.

What is the difference between an input moving and an input being replaced?

The mechanical heart of the whole distinction is worth taking slowly. When a demand assumption moves from 5 per cent to 8 per cent, the model is working. The line existed, it had a range, the range was explored, and the answer changed by a knowable amount. Nothing about the structure moved. Every other line still means what it meant.

When Sarvani Coatings buys another business, a different set of inputs arrives. There is a second revenue line with its own growth, a second cost base with its own behaviour, borrowings that did not exist on Monday, and an asset on the balance sheet that no operating decision produced. The old lines are all still there, still correct, and they now describe a part of something rather than the whole of it.

One is a value inside a structure and the other is a change of structure. The two are never measured on the same scale, so no width of range on the first ever reaches the second. Volume growth can be set at 4 per cent, at 9 per cent, at minus 3 per cent, and at no setting does a second business appear inside the sheet. The range explores a space. The event leaves the space.

Why can a sensitivity band not express it?

A sensitivity bandA range of outcomes produced by pushing one input up and down around a central case. How one is actually built belongs to valuation method and is not rebuilt here. is a continuousAble to take any value across a range, which is what a model input almost always is. object. A band runs smoothly from one end to the other, usually sits symmetrically around a central case, and every point inside it is a value the input could take. An event is not like that at all. An event is discreteSomething that either happens or does not, with no half-way values available in between.: it lands or it does not, and there is no such thing as forty per cent of a purchase sitting quietly inside the range.

So somebody reasonable asks the obvious question. Why not simply widen the band until the purchase fits inside it? Try it and watch what happens to the band. Work it through on the published year. At the low corner, volume growth of 3.0 per cent puts revenue at Rs 2,487.45 crore, and a gross margin of 44.0 per cent leaves about Rs 429.5 crore of earnings before interest, tax, depreciation and amortisation (EBITDA) once the Rs 665 crore of employee cost and other expenses is taken out. At the high corner, volume growth of 9.0 per cent puts revenue at Rs 2,632.35 crore, and a margin of 47.0 per cent leaves about Rs 572.2 crore. The spread between those two corners is about Rs 142.7 crore. On 24.00 crore shares that is about Rs 5.95/- a share, and every rupee of it has been allowed to land in cash with nothing taken out for tax or capital spend. The purchase moves the same measure by Rs 20.00/- a share. A band stretched to cover the second is between three and four times wider than the operating uncertainty it was built to describe.

Widening a band until it covers an event destroys the band for its own job. Event risk is listed rather than modelled for exactly that reason. The band existed to answer a narrow question: how much does the answer move if the operating year goes well or badly. Stretch it far enough to swallow a purchase and the band answers nothing. A reader can no longer tell whether the width came from a plausible bad year or from a transaction nobody has proposed. Two instruments, two jobs. Cramming both into one produces neither.

How does event risk actually break a thesis?

A thesis rests on a handful of named variables, and beside each one somebody writes the disconfirming evidenceThe observation, written down in advance, that would show a stated view to be wrong. Settled in the reading before this one and applied rather than rebuilt here. that would show it wrong. The worked thesis this reader arrived with names three: whether gross margin holds at 46.0 per cent, whether volume growth holds above the sector's 4.5 per cent, and whether the 0.13 percentage point share gain repeats. All three are year three figures and all three are quoted here rather than rebuilt.

Now imagine the day after a purchase. Gross margin on the old business is still 46.0 per cent. Volume on the old business still beat 4.5 per cent. The share gain still repeated. Every disconfirming test written in advance comes back negative. No evidence anybody thought to name has broken the view.

An event can leave every named variable intact and still make the thesis irrelevant. The entity the thesis was about has changed shape, and no list of disconfirming evidence catches a change of shape. The list was built to test statements about a business. The list has nothing to say about the business becoming a different business. A well built instrument pointed at the wrong object keeps reporting cleanly the whole time.

Which is why a disconfirming list needs a companion sitting beside it rather than a better version of itself. Sharpening the tests does not help. There is no wording of the sentence about gross margin that also detects the arrival of a second business, and trying to write one produces a sentence that tests nothing well.

How large is a single event against a whole plausible forecast range?

Direction is cheap. Here is the size, built only from figures the shared record already publishes, with every step written out so that it can be repeated.

Start where Sarvani Coatings Limited actually sits. At the end of year three, net worth is Rs 1,486 crore, total borrowings are Rs 240 crore, of which Rs 90 crore is long term and Rs 150 crore is short term, and cash and investments are Rs 312 crore. Borrowings less cash is minus Rs 72 crore, so the business sits in net cash rather than in a state of gearingCarrying more borrowing than cash, so lenders hold a claim standing ahead of the holders of the shares.. On the assumed 24.00 crore shares that is net debt per shareBorrowings less cash, divided by the share count, so a balance sheet position can be set beside a per-share figure. of minus Rs 3.00/-.

Now four states beside it. Every one is hypothetical, not one has been carried out and not one has been proposed. A purchase, at Rs 480 crore, of an industrial coatings maker the record leaves without a name takes all Rs 312 crore of cash and needs Rs 168 crore of new borrowing, so borrowings become Rs 408 crore against no cash at all and net debt lands at plus Rs 408 crore, or plus Rs 17.00/- a share. A buyback of Rs 240 crore at Rs 600/- retires 0.40 crore shares, leaves Rs 72 crore of cash against unchanged borrowings, and gives plus Rs 168 crore of net debt on 23.60 crore shares, or plus Rs 7.12/- a share. A rights issue of one new share for every twenty held at Rs 350/- raises Rs 420 crore on 1.20 crore new shares which, simply held as cash, gives minus Rs 492 crore on 25.20 crore shares, or minus Rs 19.52/- a share. A follow-on placementAn issue of new shares to selected buyers rather than to existing holders, so nobody already on the register receives an entitlement. of the same 1.20 crore shares at Rs 460/- raises Rs 552 crore and gives minus Rs 624 crore, or minus Rs 24.76/- a share.

Every one of those four states leaves all three thesis variables untouched, and every one of them moves net debt per share by more than the entire plausible range of any of them. Now the question the opening asks. Double volume growth, from the sector's 4.5 per cent to 9.0 per cent, hold gross margin at 46.0 per cent and let the whole difference land in cash: that is about Rs 50.0 crore, or Rs 2.08/- a share. The purchase moves the same measure by Rs 20.00/-, about 9.6 times as far. Widen the flex to the full band of Rs 5.95/- a share and the purchase is still 3.36 times larger. The placement at Rs 21.76/- is 3.66 times larger, and even the smallest of the four, the buyback at Rs 10.12/-, is 1.70 times larger. The narrowest of these comparisons still has the event winning comfortably.

Can event risk be forecast, and if not, what can honestly be done with it?

No, it cannot be forecast, and pretending otherwise is where the work most often goes wrong. Nobody outside a boardroom knows whether a purchase is being discussed, and even inside one the answer changes month to month. Anybody publishing a probability that Sarvani Coatings buys something next year has invented that number, and the invention is not improved by being written to one decimal place.

Different work can be done instead, and it is honest work. Name the events the business is capable of, given its cash, its borrowings, its stated intentions and its shape. Size the largest of them. State which named assumptions each would replace. Each of those four is checkable, and not one of them is a forecast.

The output is a capability list, never a prediction, and writing it as a prediction is exactly where this goes wrong. Saying that Sarvani Coatings could fund a purchase of roughly Rs 480 crore from cash plus Rs 168 crore of new borrowing is a statement about a balance sheet. Saying that Sarvani Coatings will make one is a statement about people whose minds nobody has read. The first is arithmetic anybody can check against the published figures. The second is fiction wearing the clothes of the first, and the clothes are what make it dangerous.

A household does this without any vocabulary at all. The balance in an account supports the honest statement that a car could be bought with it. The same account supports no statement that a car will be bought. The account has no opinion. Everybody understands this perfectly until a balance sheet is involved.

What does a large cash balance actually establish?

Nothing whatever about intention, and this is worth stating flatly because the mistake is so easy and so common. The balance sheet is a statement of position on one date. The sheet says what the business held and what it owed. The sheet permits certain things and forbids others. The sheet has nothing at all to say about which of the permitted things anybody wants to do.

Cash and investments of Rs 312 crore against borrowings of Rs 240 crore says what Sarvani Coatings Limited could pay for, and says nothing whatever about what it will do. The same Rs 312 crore funds a purchase, funds a buyback, sits untouched for three years, or goes into the coatings line already held in capital work in progress at Rs 118 crore. All four are permitted. The statement ranks none of them. Nor does adding a second measure help: capital employed, meaning net worth plus borrowings, is Rs 1,726 crore, and that figure is just as silent about intention as the first one.

A researcher who reads intention into capacity has done something worse than guess. The researcher has written somebody else's plan for them and then treated their own writing as evidence. A note ends up saying a company is likely to buy something, sourced ultimately to nothing but the fact that it could. The chain looks like reasoning and contains no observation anywhere in it.

How is event risk written down in a research note?

As a named list, and each entry carries four things. The event itself, stated plainly. The balance sheet capacity that permits it, quoted from the published statement. The model lines it would replace, named specifically rather than gestured at. And the first observable, meaning the earliest thing a researcher outside the company could actually see that would tell them it had happened.

The first observable is the column people leave out and the column that does the most work. A list without it is a list of worries. A list with it is a monitoring instrument. Each row now tells the reader where to look and what would count. Without a first observable, an entry cannot ever be closed, and a risk register that only grows is a register nobody reads.

No single figure can carry any of those four columns, and that is the entire reason the instrument is a list rather than a number. A number can say twenty rupees a share. A number cannot say which twenty rupees, permitted by what, replacing which assumptions, visible first in which place. Compressing the four columns into one figure throws away everything a reader would need in order to act on it, and keeps only the part that looks precise.

Who actually keeps a list like this, and what do they do with it on a Tuesday?

Meghna Iyer, covering the coatings names, keeps hers as four rows in the same file as the model, and she rewrites it once a quarter rather than once a year. The trigger is the results release, when the balance sheet capacity changes. Cash moves, borrowings move, and a row that was permitted last quarter may not be permitted now. When she writes a note, the list travels with the range rather than instead of it, and the two sit under separate headings so that nobody reads one as the other.

A lender does the identical work under a different name and with more force behind it. Before extending a working capital limit, a credit officer looks at what the borrower could do to the balance sheet that has nothing to do with trading badly. Could the borrower buy something and gear itself? Could it return cash it currently holds? Both questions get written into the loan documents as things the borrower agrees not to do without asking. A list of events has become a set of permissions. The lender does not forecast any of them either. The lender simply refuses to let capacity and intention drift apart without notice.

Everybody serious about this keeps two instruments side by side, one that ranges and one that lists, and the discipline is never letting either one pretend to be the other. Ravindra Setlur, sitting on the company side as chief financial officer, reads the same pair in reverse: he knows exactly which of his own capacities an outside reader can see, and he knows the reader cannot tell which one he intends to use. A household version exists too. A household budget carries a range for the electricity bill and a separate short list headed things that would change everything. On a kitchen table the difference is obvious, and nobody has ever confused the two.