How to Map a Corporate Restructuring, Step by Step

Mapping a restructuring means writing four things down in a fixed order: how far the debt sits above what the earnings support, what range of debt the business is assumed to carry, who ranks where when a short amount is shared, and what each available route moves. The output is a range and a distribution, never one number.

The four entries above are the finished article. The eight steps below produce them, and the order of the steps matters more than any single line in it. A restructuring usually arrives the way most people first meet one: as a pile of unsorted material, some of it arithmetic, some of it legal, a good deal of it opinion wearing the clothes of fact, and everybody in the room already holding a view about the answer. The map is what turns that pile into something a second person can check.

Be clear about the two things a map is for. The two pull in opposite directions, and both are needed. A map records what is known, in an order that lets somebody else re-derive every figure. A map also records what is not known. Without that record, an assumption gets mistaken for a measurement. A single figure hides which assumption made it and leaves no room to say what was never established, so a map that produces one number has failed at both jobs.

Three things arrive already settled and are put to work here rather than built again. One is earnings before interest, tax, depreciation and amortisation, or EBITDAWhat a business earns from trading before interest, tax and the charges for wearing out its assets are taken off., as the annual earnings figure, together with the leverage multipleBorrowings divided by an annual earnings figure, so the answer comes out in years-worth rather than in rupees. struck on it. Another is what a lender actually holds when a named asset stands behind its loan. The third is the borrower everything runs on: Meghdoot Coated Products Limited, an invented maker of coated packaging, owes Rs 900 crore against Rs 60 crore of EBITDA, of which Rs 620 crore sits with secured lenders and Rs 280 crore with unsecured ones. None of the three is taught below; all three are covered separately, and every mechanism a step points at is covered in its own right further along.

What are the eight steps, and why this order?

Here is the whole sequence in one view. The shape of the sequence is the lesson, and the individual steps are almost dull on their own. Take the sequence in whole before reading any step in detail.

1. State the two numbers and the ratio between them. Nothing else.
2. Fix the sustainable range as a range, and write down who chose it and why.
3. Convert that range into rupees and into per cent, naming the base.
4. List every claim and check the list adds back to the borrowings.
5. Apply each ranking rule to the same pot and set the outcomes side by side.
6. Mark each route against the two numbers and record what it forecloses.
7. Split the map into what is arithmetic and what is set in law.
8. Write down what is not known, inside the map rather than under it.

Now the part that is easy to miss. Five of those steps produce something: a ratio, a range, an amount, a distribution, a marked list of routes. Three of them produce nothing at all. Steps four, seven and eight add no figure to the map. The three exist purely to stop a later step, or a later reader, producing an answer that looks solid and is not. Nothing in the output announces that the checks were skipped, so a map without its three unproductive steps still generates numbers, and that is precisely the danger.

Reordering is the other failure, and it is quieter. Suppose somebody starts at step five. Who gets paid is the interesting question, and everybody wants to argue about it. Somebody starting there needs a pot to share, so they reach for a figure, and the figure they reach for was produced by a multiple nobody wrote down. The input has no stated origin, so the distribution work is real, careful and completely uncheckable. Every step exists to make the step after it checkable. The sequence is not a matter of taste.

Step one: what are the two numbers, and what is the gap?

Write down what is owed. Write down what is earned in a year. Divide the first by the second. Stop.

For Meghdoot Coated Products Limited that is Rs 900 crore of borrowings, Rs 60 crore of EBITDA, and a ratio of 15.0 times. Three figures, one line of the map, and the temptation at this point is enormous: to add a sentence about how serious that is, or which route it points to, or what a lender is likely to accept. Resist all of it. A map that starts with a view spends the next seven steps assembling evidence for the view, and the person reading it afterwards cannot tell the difference between an analysis and a case.

There is a practical reason as well as an intellectual one. Step one is the only part of the map that is pure record: these are the figures the borrower's own accounts carry, and anybody can check them against the same source. Everything after step one has some element of choice in it. Letting interpretation into the one step that has none gives up the only clean footing the map has.

Three figures invite more reading than three figures can carry, so the limits of step one are worth stating. Step one does not say the position is unsustainable; it says the position is 15.0 times, and what that supports is settled at steps two and three. The record for Meghdoot Coated Products carries no repayment schedule at all, so step one says nothing about when payments fall due. The record holds one year of earnings and no more, and a single point cannot describe a direction, so the ratio says nothing about whether the position got worse, better or stayed still. The missing direction is easy to fill in by accident, and it is the absence to guard hardest.

Step two: what range is being assumed, and who chose it?

Step two is the step everything else is arranged around, and it is the shortest to describe. The range of leverage the business is assumed to be able to carry goes into the map, together with the name of the person who chose it and the basis they used. Then the map moves on.

For this map the range is 3.0 to 4.0 times, and the honest entry beside it reads: chosen by whoever is writing the map, on a judgement about how the earnings behave across years the record does not contain. The entry is not a coy way of saying the figure is weak. The entry records the exact standing of the figure. Nothing measured it, no authority publishes it, and a different competent person would have written 2.5 to 3.5 and been no less careful.

Step two is the only place in the whole map where a judgement enters. Record the person and the reason at step two, or both are lost for good. Once a number leaves step two it looks like every other number in the map. The number has a rupee sign, it sits in a column, it survives being copied into a summary, and by the third document nobody remembers that it began life as somebody's opinion about the future.

Now a point worth being exact about. Step four is only useful to somebody who has it right. Everything downstream of step two is often said to inherit the judgement, and the claim is wrong. Only two later steps carry the judgement: step three, whose whole output is a function of the multiple, and step five, whose pot is the supported figure step three produced. Steps four, six, seven and eight are independent of it. The claim list is Rs 620 crore secured and Rs 280 crore unsecured whatever multiple is assumed. The five routes exist whatever multiple is assumed. Nothing set in law moves, and nothing missing from the record moves either. Independence of that kind is not a technicality: it is the reason step four can catch an error that step two could never catch, and the reason the map is built in this order rather than any other.

Step three: what does the range imply for the amount that has to go?

Multiply the range by the earnings. The product is the debt the earnings are assumed to support. Subtract the product from the borrowings. The difference is the amount that has to go. Then express the difference as a percentage, and name the figure it is struck on.

Three levels, worked on the locked figures, so the range is visible rather than a point.

Two things about that table belong in the map itself rather than in a reader's head. The first is the base. Every one of those percentages is struck on the Rs 900 crore owed and on nothing else, and a percentage without its base written beside it is the single most portable error in this whole subject. The second is the rounding. The record states these three as whole numbers, 80, 77 and 73 per cent, and 690 over 900 is actually 76.667. One decimal everywhere keeps the table, the drawings and the moving panel below in agreement, and 77 per cent is then plainly the middle row rounded.

Now the bottom row, and it needs a sentence or a reader will think a figure has been pasted twice. The distance between the outer two write-downs is Rs 60 crore. The EBITDA is Rs 60 crore too. The match is not a coincidence and it is not corroboration. Setting 3.0 times beside 4.0 times sets one turn beside the next, so whatever separates the two rows has to be a single turn of earnings, and a single turn of earnings is Rs 60 crore: forced by construction rather than found by working. Widen the ends to 3.0 and 4.5 times and the gap becomes Rs 90 crore, at which point nothing looks doubled at all. A reader who spots the match and is left to guess will either distrust the whole table or, worse, read it as confirmation. Write the sentence into the map.

One rule governs the format of this step's output and it is simple to check from across a room. A step three output with one row in it has skipped step two. One row means one multiple was picked and never written down, and a figure like a write-downReducing a recorded amount owed to a smaller figure, so that the smaller figure becomes the one that stands. of Rs 690 crore then travels through every later document with its origin missing.

Step four: do the claims add back to the debt?

List every claim. Write beside each what stands behind it. Add the list up. Compare the total with the borrowings at step one. If the two agree, write down that they agree. If they do not, stop.

For Meghdoot Coated Products Limited the list is short: Rs 620 crore held by lenders who are securedA lender whose loan has particular property standing behind it, so that if repayment stops there is something specific to look to., Rs 280 crore held by lenders who are unsecuredA lender with a claim on the business and no particular property standing behind it.. Add them: Rs 900 crore. Compare with the borrowings: Rs 900 crore. Difference, nil. The entire step took one line.

The word to hold onto is that this is a reconciliationChecking that a list of parts adds back to the total it is supposed to add back to, and hunting down the difference when it does not. and not a summary, because the two look identical on paper and behave completely differently. A summary describes what somebody has said. A reconciliation asserts that the description is complete, and it earns that assertion by closing to nil. If the claim list came to Rs 880 crore against borrowings of Rs 900 crore, there are only two possibilities: a claim of Rs 20 crore is missing from the list, or Rs 20 crore has been double counted somewhere else and the borrowings figure is wrong. Both are found in ten minutes here. Neither is found at all once the map moves on.

Why this step is placed fourth rather than second is worth a moment. The claim list is not affected by the multiple, so it could sit anywhere. The claim list sits fourth because it is the last thing that can be checked before the map starts distributing money, and because a reader who has just watched three steps of arithmetic is in exactly the right frame of mind to notice that step four produces no number. Step four is a hinge: everything before it is about how much, everything after it is about to whom.

Step five: what does each ranking rule do to the same pot?

Take the pot. Apply the first rule and write the outcome for each group with the base of every percentage named. Apply the second rule to the identical pot and write that outcome beside it. Do not choose between them. The comparison is the output of the step.

Work it on the middle level, where the earnings support Rs 210 crore. Under strict ranking the secured lenders take the whole pot: Rs 210 crore against their own claim of Rs 620 crore, a recovery of 33.9 per cent of that Rs 620 crore. The unsecured lenders receive nothing at all against their Rs 280 crore. Under a pro rata rule each claim is cut back in the same proportion: Rs 144.7 crore goes to the secured lenders and Rs 65.3 crore to the unsecured, both figures being 23.3 per cent of what that group holds. The two receipts add back to Rs 210 crore, and adding back is the check that the split is complete.

Two figures will be misread if the map does not defend them, and defending them costs one clause each. The first is 33.9 per cent, struck on the secured lenders' own Rs 620 crore. The figure is not anybody's recovery on the whole Rs 900 crore, and a reader holding unsecured paper who sees an undefended 33.9 per cent will take it as theirs. The second is subtler. Whichever rule is applied, the total handed out is Rs 210 crore, so the recovery measured across all Rs 900 crore of claims comes to 23.3 per cent either way: ranking moves a fixed quantity between two groups without adding a rupee to it or removing one. Say the base every single time and the two figures cannot be confused; leave the base off once and they will be.

The rule of the step is short: never run one rule. A single rule produces a table of receipts that looks like the answer to a question of fact. Two rules produce two tables that cannot both be the answer, and the real question then shows itself. Step three settled how much is available. On what principle the money is shared is a question no amount of arithmetic settles. The pair worked above sits at the two ends of what ranking can do, so the arithmetic in between shows up plainly. Neither end describes what any statute lays down.

Step six: which routes are available, and what does each one move?

Name the routes. For each, mark what it does to the amount owed, what it does to the earnings, whether any cash arrives from outside the business, and what taking it forecloses. Mark; do not score.

Marking rather than scoring is the whole of step six, and it is harder to follow than it sounds. The map is being built to solve a problem, and a route that solves nothing looks like clutter, so the impulse when a route moves neither number is to strike it off. But the map records what each route moves and ranks nothing. A route that moves neither number and brings in no outside cash may still be the right thing to do, and the map records that honestly instead of scoring it as a fix. Separating a business into two is the standing example: on the day it happens the same earnings and the same debt exist, merely in two containers rather than one, yet it can put a sound half in a position to borrow on its own merits once it stands alone. The benefit is real and it is downstream, and a map that struck the route off has hidden it.

The fourth column, what a route forecloses, is the one people leave blank and the one that matters most in a room. Selling the whole business is not one option among five; it consumes all the others. After a whole sale there is nothing left to separate, rebuild or sell in part. Selling a particular part cannot be done twice. Recording that in a column, at the moment the routes are first listed, is what stops a later conversation treating the routes as a menu that can be ordered from several times. The fourth column is also where the perimeterThe line drawn around what a transaction includes, with everything else deliberately left outside it. of any part sale gets its first written definition, and the later work needs it.

Step seven: what here is arithmetic, and what is law?

A line runs down the middle of the map. On the left, everything anybody can recompute from the figures already written. On the right, everything that is settled by text somebody else publishes. The left is filled in. The right is not.

The left column of this map is already complete, built step by step above: the ratio at 15.0 times, the supported range of Rs 180 crore to Rs 240 crore, the Rs 660 crore to Rs 720 crore that has to go, the claim list closing at Rs 900 crore, the Rs 210 crore pot and both distributions of it. A reader can take a pencil to every one of those and land in the same place. Every figure on the left is worth writing precisely because it can be recomputed.

The right column stays blank on purpose. Its questions decide whether anything on the left can actually happen: whether a reduction binds a party that never agreed to it, which parties it reaches, along which route it has to travel, in what sequence the money is applied, and across what period the whole thing runs. The answers are published and they change, and a figure recalled from memory is worth nothing to the person who has to act on it. Not one of them goes into the map from memory. What goes in the right column is a name and a site.

There is a reason this is a step rather than a caveat at the bottom. A caveat is read once and forgotten; a column is looked at every time the map is opened, and its emptiness is visible. Somebody who wants a timetable can see at a glance that the map does not have one, and where to go for it. A plausible sentence that turns out to be six months out of date is far worse.

Step eight: what is not known, and how is it recorded?

List the gaps in the record, and put the list inside the map with the other seven steps, not underneath as a note.

For Meghdoot Coated Products Limited the list is five lines long and every one of them closes off work somebody will otherwise attempt. There is no repayment schedule and no maturity, so nothing on this map can describe a recovery spread over time. There is no interest rate and no facility terms, so no coverage of interest and no cover of debt service can be computed here at all. There is no register of the assets behind the Rs 620 crore of secured claims, so nothing can be said about what stands behind them beyond the fact that something does. There is no bid and no sale price, so there is nothing to set against the Rs 900 crore. And there is only one year of earnings, so no rebuild path can be drawn. A single point carries no direction, so no sentence may say the business deteriorated.

An unknown that is not written down becomes an assumption within about a week, and afterwards nobody can tell which figures rested on it. Recording the gaps is a step rather than a caveat for exactly that reason. The mechanism is ordinary and entirely human. Somebody needs a number for a slide. The absence is not visible anywhere, so a reasonable placeholder is used. The placeholder is copied into a second document without the word placeholder. By the third document it is a figure with a history, and unpicking it costs more than producing the map did.

There is one more reason to keep this list visible, and it is the one that matters when the room gets tense. The gaps in the record are the honest limit of what anybody in the conversation knows, and they belong to everybody equally. A map that shows its gaps invites the parties to fill them; a map that hides them invites the parties to distrust the parts that are filled in.

Notice what the control cannot be made to do. Drag it to the far right and 12.5 per cent of the bar is still open. The last step is a register rather than a resolution. A finished map is not a map with nothing open in it; it is a map where everything still open has been written down and named. The standard is much lower than most people expect from an analysis, and it is the only one anybody can actually clear.

How is a map like this actually used, and by whom?

Four people pick the same map up and read four different parts of it first. The four choices say a good deal about the purpose of the map.

A lender turns to step four before anything else. The claim list is where a lender finds out what position it is actually in, as opposed to what it is owed, and those are different facts: Rs 100 crore of secured claim and Rs 100 crore of unsecured claim are the same amount and not the same thing. A lender then reads step five twice, once for each rule, and the question it is really asking is which of the two outcomes it is being invited to accept, and on what basis. A lender who reads only step three has learned how large the problem is and nothing about its own place in it.

An analyst goes straight to step two and will not be moved off it. Handed a map whose third step reads Rs 690 crore, the first three things an analyst wants are the multiple sitting behind that figure, the name against it, and what the row looks like a few tenths of a turn in either direction. None of that is doubt for show. The three questions are the only way to separate a result that would have survived a different assumption from a result that is really a description of the assumption. Writing step three as three rows in the first place is what makes those questions take ten seconds instead of a day.

An investor weighing whether to put fresh money behind a business at this point reads step six and step eight, and frequently skips step three altogether. An investor cares about the shape on the far side: what will be owed, what will be earned, and how much of the company the new money buys. Read that list again and see what is absent from it: the size of the write-down. Then see what step eight closes off here. The record for Meghdoot Coated Products carries no share count and no shareholders whatever, so this map cannot say how much of the company a lender converting debt would end up with, and no honest version of it invents a figure.

And somebody advising the borrower reads step seven first. The right column decides what is even possible. Everything on the left of the map can be agreed in a room by people who want to agree. Whether that agreement binds a party who does not want to agree is the other column, and a plan built without looking at it is a plan among the willing.

One more use, and it is the one that keeps the map honest over months rather than days. Because diligenceThe examination a party runs over a business before committing, to find out what it is actually taking on. keeps producing material after the map is first written, step eight doubles as the work list. Each line in the register is a thing somebody could go and find, and when one is found the map gains a figure and loses an unknown, in that order and visibly. A map without a register has nowhere to put an answer when one arrives.

What does this look like away from the crore?

The eight steps are not a finance invention, and most people have watched somebody do them, so the machinery works away from the balance sheet entirely.

A wedding is three weeks away in a small town, and the money set aside has run out. The instinct in the room is to start deciding: pay the tent people first because they will not deliver otherwise, or the caterer because everybody will notice the food. Deciding first is starting at step five, and it is why these conversations go around in circles for two days.

The person who has done this before starts differently. The experienced person takes a sheet of paper and writes down what is left in hand and what is committed: step one. Next comes what the household can realistically add before the date, written as a range rather than a figure, with whose income the range assumes said out loud: step two. Then they work out what has to give at each end of the range: step three. Then, and this is the step everybody skips, they list every single person who is owed something and add the list up, and they keep going until the list matches the committed total. Adding the list up is step four, and it is where the forgotten advance to the decorator surfaces, in the kitchen with three weeks to go rather than in the hall on the day.

Only then does the room decide who is paid what, and the person with the sheet insists on doing it twice: once as everybody takes a share of what is left, and once as the people with a written agreement are paid first. Two arrangements set side by side turn an argument about fairness into a visible choice. One arrangement produced quietly turns a choice into a fact somebody else has to accept. Then they mark which options remain open, they write down the one thing they cannot settle themselves, and they write down the two things nobody has yet checked. Eight steps, one sheet of paper, and no crore anywhere.

What does the finished map hand the room?

Everybody who walks into a restructuring wants one number. The lender wants to know what it will get back. The borrower wants to know what it will be left owing. Whoever is advising wants a figure that can be written in one line and defended. The map disappoints all three of them, and that is the map working rather than the map failing.

The map hands over a range and a distribution. The range is Rs 660 crore to Rs 720 crore having to go, or 73.3 to 80.0 per cent of the Rs 900 crore owed, with the assumption that produced each end written beside it. The distribution is both ranking rules run on the same Rs 210 crore, with the base of every percentage named, and no view offered as to which rule ought to apply. A single figure is a judgement that stopped announcing itself, so a map that ends in one has failed at step two.

And the last three columns of it are the ones that will still be useful in six months: what is arithmetic, what is law, and what nobody knows yet. The arithmetic can be recomputed by anyone. The law is where it is published. The unknowns are the work list. Three columns are a modest output for eight steps, and they are the only kind of output a second person can take apart and put back together without having to ask the author what was meant.

Where these figures come from

Meghdoot Coated Products Limited is a made-up borrower and every rupee in the eight steps was written for teaching, including the Rs 900 crore of borrowings, the Rs 60 crore of EBITDA and the split of the claim list. The three multiples at step two are somebody's judgement: nothing measured them. A proceeding's timetable, a class of creditor, a proportion of consent and the order in which claims are paid are settled by text that changes and are read at the source rather than recalled. Strict ranking and pro rata appear here as two arithmetic extremes that make the effect of ranking visible; neither is the rule in force anywhere. The write-downs print to one decimal place throughout, so the table, the drawings and the moving panel cannot quietly disagree with each other, and the whole numbers the record uses, 80, 77 and 73 per cent, are the rounded form of the same three figures.