Refinancing Risk: When the Maturity Wall Arrives

Refinancing risk is the risk that money borrowed today has to be borrowed again on terms nobody can see yet. Sankalp Industrial Systems Limited, invented, has Rs 7,25,00,00,000 outstanding at the end of Year 5, of which Rs 3,00,00,00,000 falls due in one instalment, being 41.38 per cent. The company's own cash generation in that year covers 56.67 per cent of it.

Refinancing risk is an awkward idea, and an awkward idea lands faster on something small. Start with a tailoring unit run out of two rooms behind a main road. The household that runs it borrowed Rs 4,00,000 three years ago to buy machines. The arrangement was simple and it is the arrangement most small borrowers would take if offered it: pay the interest every month, and pay the whole Rs 4,00,000 back on one date three years out.

The unit has done well. The machines run six days a week, the interest has never been late, and the household has more income now than it had when it signed. The money went into cloth, into a third machine and into fees for a child's schooling. And on the morning the Rs 4,00,000 falls due, the household does not have Rs 4,00,000. So on that morning it has to ask for the loan again.

Whether it gets the loan again, how much of it, and at what rate, is decided that morning by things that have almost nothing to do with how well the household stitches. The unit does not have a cash problem and it has never missed a payment, and it is still exposed to something real: a date it chose three years ago has arrived, and it is not the one deciding what happens on it. Refinancing risk is that situation, done to a company instead of a household, and worked out to the rupee.

What exactly is refinancing risk, and why is it not the same as running out of money?

Because they are two different problems that happen to arrive through the same door. Refinancing riskThe risk that borrowing which must be replaced can only be replaced on different terms, or not in full. is the risk that borrowing which has to be replaced can only be replaced on different terms, or cannot be replaced in full. Notice what that sentence does not say. The definition does not say the company is short of money, it does not say the business is deteriorating, and it does not say anybody has missed a payment.

The distinction between the two is collapsed constantly, and collapsing it makes the whole subject unreadable. Put the two side by side. A company that cannot pay has a cash problem. A company facing refinancing risk may have no cash problem whatever, and has a date problem instead. On one particular day it must replace borrowing it already has, and the terms it will be offered on that day are set by conditions that are invisible from where it is standing now.

A cash problem and a date problem are found in different places. A cash problem shows up in the cash flow statement and in the working capital lines, and an analyst who reads those carefully will see it coming. A date problem shows up nowhere in the profit and loss account at all. A date problem lives in a schedule of dates attached to the borrowings. A company can look excellent on every measure of profitability and coverage while carrying one.

The word risk is doing real work in that definition and is worth being precise about. The risk is not that the company deserves to be turned down. The risk is that the terms available on a single morning are decided by the appetite of lenders on that morning, and appetite is not a property of the borrower. Appetite is a genuinely different kind of exposure from anything a margin or a coverage ratio measures, and it needs to be looked for on purpose.

What is a maturity wall, and how is one spotted?

A maturity wallA date on which an unusually large share of a company's borrowing falls due at once. is a date on which an unusually large share of a company's borrowing falls due at once. The word that matters in that sentence is share. A wall is a concentration, not a total. A company with Rs 7,25,00,00,000 of borrowing spread evenly over ten dates has no wall anywhere. A company with the same Rs 7,25,00,00,000 where two fifths of it lands on one morning has one, and it has one whether or not the total is comfortable.

Two features of a loan decide whether it builds a wall or not. The first is how the principal comes back. A loan with a bullet maturityThe whole principal falling due in one instalment on one date. repays the whole principal in one instalment on one date, having paid only interest until then. An amortisingPrincipal repaid in instalments across the life of the loan. loan repays principal in instalments across its life, so the balance falls as the business generates cash and there is much less of it left at the end. The same amount of borrowing at the same rate creates a wall in one form and no wall at all in the other, and the only difference is where the principal was scheduled.

The second feature is what everything else in the borrowing is doing on the same date. A single bullet is one thing. A bullet arriving in the same year as several other events is another, and the total of those events is what actually has to be dealt with.

A wall is spotted by reading the maturity profileThe amounts falling due plotted against the dates they fall due on.. The profile is simply the amounts falling due plotted against the dates they fall due on. A wall is a shape and shapes are not visible in a list, so reading the profile as a picture rather than as a column of numbers is the whole technique. The shape to look for is a year that towers over the years around it, and years that are empty because nothing was ever scheduled into them rather than because data is missing.

The four empty years are the reason a wall survives every check an analyst runs, so look at them first. Nothing is missing from Years 1 to 4. Nothing was ever scheduled into them. The company pays interest through those years and repays no principal at all, so on every annual measure it looks like a company with a very quiet borrowing profile, right up to the year it is not.

Where does this company's wall sit, and what is it made of?

Sankalp Industrial Systems Limited, invented, is a listed manufacturer of industrial valves, precision castings and the aftermarket parts and service that go with them. Its borrowing at Year 0 is Rs 6,00,00,00,000 in three tranches, and the shape of the wall is entirely decided by the three rows below. The composition of these tranches is settled earlier in this subject and is restated here rather than rebuilt.

Every rate in that table is this invented company's own contracted rate and none of them is a statement about what borrowing costs anybody in India. The blended figure is worth checking once. A check like that shows the rest of the record was built rather than assembled: Rs 3,00,00,00,000 at 7.80 per cent is Rs 23,40,00,000 of interest, Rs 2,00,00,00,000 at 8.50 per cent is Rs 17,00,00,000, and Rs 1,00,00,00,000 at 7.60 per cent is Rs 7,60,00,000. The three interest amounts add to Rs 48,00,00,000 on Rs 6,00,00,00,000 of borrowing, and Rs 48,00,00,000 over Rs 6,00,00,00,000 is exactly 8.00 per cent.

Now watch what each tranche does over the five years. Tranche 1 does nothing at all until its date: the tenorThe length of time a borrowing runs for. runs to the end of Year 5 and the whole Rs 3,00,00,00,000 arrives then. Tranche 2 does nothing within the forecast at all. Its date is the end of Year 7. Tranche 3 is the one that moves. The facility has no final date, only an annual renewalThe annual re-agreement of a facility that has no fixed final maturity., and it is the tranche the company draws on: the Rs 25,00,00,000 of net new borrowing in each of the five years is taken there, so it goes Rs 1,25,00,00,000, Rs 1,50,00,00,000, Rs 1,75,00,00,000, Rs 2,00,00,00,000 and Rs 2,25,00,00,000.

Add the three at the end of Year 5 and the company has Rs 7,25,00,00,000 outstanding, of which the Rs 3,00,00,00,000 term loan falls due in one instalment on that day. The wall is that single instalment, and everything below is measurement of it.

How big is the instalment as a share of everything outstanding?

Rs 3,00,00,00,000 over Rs 7,25,00,00,000 is 41.379 per cent, written 41.38 per cent. Two fifths of everything the company has borrowed falls due on one morning, and the arithmetic that gets there is one division. The reason to do that division rather than quote the amount is that an amount travels nowhere. Rs 3,00,00,00,000 is enormous for one company and trivial for another, and until it is expressed as a share of the borrowing it sits inside, nobody can tell which.

The other two slices are worth naming. Each carries a different kind of exposure, and only one of them is a wall. The Rs 2,25,00,00,000 facility, 31.03 per cent of the total, has no final date at all, so it never matures and it is never repaid in the ordinary course. The facility comes up for renewal every single year instead. An annual renewal is a smaller exposure than the bullet and a more frequent one: the facility asks the question once a year, every year, and the term loan asks it once, very loudly.

The Rs 2,00,00,00,000 of debentures, 27.59 per cent of the total, is the third case and it is the most interesting one for anybody reading a model. The debentures mature at the end of Year 7. The explicit forecast stops at Year 5. So more than a quarter of the borrowing outstanding at the end of the forecast has a maturity the forecast never reaches, and no amount of care inside those five years will surface it. The blind spot is not a defect in this particular model. A blind spot past the last column is a property of every finite forecast, and the only defence against it is to read the borrowing schedule out to its own last date rather than out to the model's last column.

How much of the instalment does the year's own cash flow cover?

The coverage test is one division and it is the second number anybody should compute after the share. Take the free cash flow to the firmCash the business generates after tax and after reinvestment, before any payment to lenders. the company produces in the year the wall arrives, and put it against the instalment. Year 5 free cash flow to the firm is Rs 1,70,00,00,000. The instalment is Rs 3,00,00,00,000. Rs 1,70,00,00,000 over Rs 3,00,00,00,000 is 56.667 per cent, written 56.67 per cent.

The test is harsher than it looks, so be careful about what that Rs 1,70,00,00,000 already is. The figure is the cash the business generates after tax and after every rupee of the Rs 1,00,00,00,000 of net new capital it reinvests that year, and it is measured before any payment to lenders. Free cash flow to the firm is not spare money sitting in an account. Devoting the entire figure to one instalment would mean paying no dividend, buying nothing back, adding nothing to cash and still servicing the interest on the other two tranches out of something else.

Even on that heroic assumption the year falls Rs 1,30,00,00,000 short, and that shortfall has to be refinanced or funded from somewhere the operating business did not generate. The coverage test does not say whether the company is in trouble. The coverage test says how much of the date the business itself can carry, and the answer is a little over half.

What else is happening in Year 5 that a maturity table does not show?

Renewals are the sharpest point in the whole subject, and they are the reason a careful analyst can still get the year wrong. Tranche 3 is renewed annually. Its Rs 2,25,00,00,000 balance comes up for renewal in Year 5 exactly as it does in every other year. A renewal is not a contractual maturity. A renewal has no due date to put in a column, so it does not appear in a maturity table at all. And yet it needs a decision from a lender in that year just as surely as the term loan does.

Add the two together. Rs 3,00,00,00,000 falls due and Rs 2,25,00,00,000 comes up for renewal, so Rs 5,25,00,00,000 of the Rs 7,25,00,00,000 outstanding is either due or up for renewal in one year. Rs 5,25,00,00,000 over Rs 7,25,00,00,000 is 72.41 per cent. A maturity table for that year reports 41.38 per cent, is technically correct, and understates by three fifths again what actually has to be agreed with somebody.

Which figures are locked and which are worked out is worth knowing, so state plainly where that 72.41 per cent comes from. The tranche balances are locked. The 41.38 per cent and the 56.67 per cent are locked. The Rs 5,25,00,00,000 and the 72.41 per cent are arithmetic performed here on those locked balances, and they are checkable in one line: 300 plus 225 is 525, and 525 over 725 is 72.414 per cent.

Run the coverage test again on the honest denominator and the picture changes shape. Year 5 free cash flow to the firm of Rs 1,70,00,00,000 against Rs 5,25,00,00,000 is 32.38 per cent, also arithmetic rather than a locked figure. Neither number is a verdict. The pair shows that the answer depends entirely on which denominator was chosen, and choosing the smaller one without saying so is how a reasonable analysis quietly becomes a misleading one.

Does five years of cash flow settle the question?

Five years of cash flow is the obvious objection, and the money genuinely is there in total, so the objection deserves a proper answer rather than a brush off. Across the five forecast years free cash flow to the firm is Rs 98,00,00,000, Rs 1,16,00,00,000, Rs 1,34,00,00,000, Rs 1,52,00,00,000 and Rs 1,70,00,00,000. The five years add to Rs 6,70,00,00,000. Against that the company pays after-tax interest of Rs 36,00,00,000, Rs 37,50,00,000, Rs 39,00,00,000, Rs 40,50,00,000 and Rs 42,00,00,000, totalling Rs 1,95,00,00,000. Both totals are arithmetic on the locked yearly figures rather than locked figures themselves, and both foot.

So the business generates Rs 4,75,00,00,000 across the five years after serving its lenders, against an instalment of Rs 3,00,00,00,000. Rs 4,75,00,00,000 is 1.58 times the instalment. On any reading, the money exists.

The money exists in aggregate and it is absent on the day, and the whole subject of refinancing risk lives in the distance between those two statements. Cash that was generated in Year 2 is only available in Year 5 if somebody deliberately held it back rather than reinvesting it, distributing it or spending it. Nothing in this record says anybody did. The company reinvests Rs 1,00,00,00,000 of net new capital in every one of the five years. Reinvestment is precisely how the business grows, and cash that has been turned into a machine is not available to repay a loan.

The everyday version is a household that earns comfortably more over three years than the cost of the roof it eventually has to replace, and still cannot replace the roof on the day it fails. The three years of surplus went on school fees and a scooter. Nobody in that household made a mistake. An aggregate is simply not a balance, and the two are read from different documents.

What could actually differ on the day the borrowing is replaced?

Three separate things, and almost every treatment of this subject collapses them into one. The question is very rarely whether the borrowing gets replaced. The question is what the replacement looks like.

The first is the rate. Nothing on the record fixes the rate at which a replacement for tranche 1 could be arranged, and it is worth resisting the assumption that it would be the same 7.80 per cent. Sankalp Industrial Systems Limited carries a schedule of what it would pay at different levels of borrowing, and even within that one schedule the rate runs from 8.00 per cent where the company currently sits to 13.00 per cent when it has borrowed a great deal more. The schedule belongs to Sankalp Industrial Systems Limited alone and describes nobody else.

The second is the amount. A lender may be willing to replace part of the Rs 3,00,00,00,000 rather than all of it. A partial replacement is still a replacement, and it changes the problem into a smaller version of itself: whatever is not replaced has to come out of cash, out of an asset, or out of somebody else.

The third is the terms, and it is the one that gets least attention and often matters most. Security can change, so a replacement might take a charge over assets the current loan does not touch, leaving less to pledge the next time. The tenor can change, and a shorter tenor does something specific and unpleasant: it moves the next wall closer. The promises written into the document can change. None of that shows up in a rate, and none of it shows up in an answer that is only yes or no.

A refinancing that happens on worse terms has still happened, so the honest question is never whether the date was survived but what the company looks like on the other side of it.

What does a company do about a wall, and when does it have to start?

There is a list, and the striking thing about the list is not what is on it. The striking thing is that almost every item has to be started years before the date. The last year before a maturity is the year with the least room in it. The date is closest, any lender being asked can see that it is closest, and every response that involves rearranging the shape of the borrowing needs a shape that can still be rearranged.

Take them in order. StaggeringArranging maturities so that no single date carries a large share of the borrowing. means arranging maturities so that no single date carries a large share of the borrowing, and it is a decision made when the borrowing is put in place rather than later. Amortising rather than bulleting is the same kind of decision: schedule the principal to come back in instalments and the balance falls as the business generates cash, so there is far less of it left at the end.

Extending the maturity before the date arrives rather than at it is the middle response, and the reason it is a middle response is that it needs a lender to agree while the date is still comfortably far off. Pre-fundingRaising the replacement borrowing before the maturity date and holding it. means raising the replacement early and holding it. Holding it costs the difference between what the money earns while it waits and what it costs to have raised it, and that cost is the price of certainty about the date.

Keeping undrawn headroomThe part of an agreed facility that has not been drawn and could still be. on a facility is a different kind of answer, and it is capacity a company holds without using it. Undrawn headroom is a response a company can have, and this record discloses none on tranche 3.

The last two are the ones a company reaches for when the earlier ones were not taken. The company can sell something that is not part of the operating business, and the non-operating assets set out below are exactly that. Or it can reduce the amount it needs, by cutting reinvestment in the years before the date and holding the cash instead. Cutting reinvestment is a real option with a real cost. The Rs 1,00,00,00,000 a year of net new capital is exactly what makes the forecast grow, and a company that stops reinvesting to build a repayment fund has bought its date with its own growth.

What is on the balance sheet that is not part of the operating business?

Rs 1,00,00,00,000, and it is worth separating carefully because it changes what is available without changing the business at all. Sankalp Industrial Systems Limited holds non-operating assetsAssets that produce none of the operating profit being forecast. of Rs 1,00,00,00,000: a surplus land parcel at Rs 45,00,00,000 and a 26.0 per cent holding in Aruna Tooling Private Limited, invented, at Rs 55,00,00,000.

The point about both is structural rather than incidental. Neither produces a rupee of the earnings before interest, tax, depreciation and amortisation (EBITDA) that was forecast for any of the five years. The land is surplus, meaning nothing is made on it, and the holding in the associate is accounted for outside the operating result. So in principle each could be realised without touching the valves, the castings or the aftermarket business at all, and being realisable that way is exactly what makes them different from selling a factory.

Set the Rs 1,00,00,00,000 against the Rs 1,30,00,00,000 the year of cash flow does not reach and it is 76.92 per cent of it. The 76.92 per cent is arithmetic on the record's own figures and not a plan. Nothing in this record says the land was sold, says the holding was sold, or says either was ever considered. The arithmetic establishes only that the company has something outside its operating business of roughly the size of the gap. The size of that holding is a fact about the balance sheet rather than a forecast about anybody's behaviour.

So what may be concluded about how this one ended?

Nothing, and the reason is worth stating precisely rather than treating as modesty. The record contains the instalment, the share, the coverage, the tranche balances and the dates. The record contains no statement whatever about whether tranche 1 was refinanced, extended, repaid, repaid in part or dealt with in any other way. There is no outcome in the record because none was ever set down.

Supplying an ending would invent the one fact least entitled to be invented. The invented fact would then travel into every other treatment that used this company. A favourable ending would quietly teach that walls resolve themselves. An unfavourable one would quietly teach that a bullet maturity is a defect. Neither is supported by anything here.

The method remains, and the method is complete without the ending. The method reads the profile rather than the total. It states the concentration as a share. It tests that share against the cash the business itself generates in that year. It adds anything else that needs a decision in the same year, even if it has no due date. It asks which of the three things could differ on the day. And it looks at when each available response has to start, which is almost always earlier than the year everybody is worried about.

Refinancing risk is a funding risk rather than a solvency one, and that single sentence is probably the most portable thing here. A company with ample interest cover, rising cash flow and no difficulty visible anywhere in its accounts can still be exposed to one, because the exposure is to a date and to the appetite of others on that date. Reading it any other way turns a question about timing into a judgement about quality, and they are not the same question.