Capital Structure: The Debt and Equity Mix and What Decides It

A capital structure is the mix of borrowing and shareholders' money a business runs on. Sankalp Industrial Systems Limited, invented, carries Rs 6,00,00,00,000 of gross debt against Rs 18,00,00,00,000 of equity at market, a 25.0 per cent debt share. Modigliani and Miller showed the mix cannot matter without taxes or distress. Both exist here, so it does.

The idea underneath capital structure is easier to feel than to define, so it is worth starting on an ordinary street. A workshop on the edge of a small industrial estate needs a new lathe costing eight lakh rupees. The owner has two ways to get it. She can take eight lakh out of the money the workshop has already earned and kept, or she can put in three lakh and borrow five from a bank against the machine itself. Either way the same lathe arrives on the same lorry, is bolted to the same floor, and cuts the same metal for the same customers.

Ask what changed about the lathe between those two versions and the honest answer is nothing whatever. The change is in who has a claim on what the lathe earns, and in what order they get paid. In the second version the bank goes first, at a rate written into a document, whatever kind of month the workshop has. The owner goes second and takes whatever is left after the bank has been served. The whole subject is that one move, done at the scale of a listed company and measured properly.

What is a capital structure, and what is actually being mixed?

Two claims are being mixed, and they are not two flavours of the same thing. A capital structureThe mix of borrowing and shareholders' money a company uses to fund itself. is the proportion in which a business funds the assets it works with using money it has promised to repay and money it has not. The proportion is the whole definition. Everything else about capital structure is a consequence of it.

Debt vs Equity Financing: a fixed claim against a residual one

A lender's claim is fixed. The amount is written down, the rate is written down, the dates are written down, and none of those three move because the year went well or badly. If the company earns twice as much, the lender still gets exactly what the document says. If it earns half as much, the lender still expects exactly what the document says, and the remedies for not being paid are also written down.

A shareholder's claim is a residual claimA claim that is paid only after every fixed claim has been served in full.. Nobody has promised a shareholder anything at all. A shareholder gets what remains after the lenders, the suppliers, the staff and the tax authority have been served. The residual position is worse in a bad year and better in a good one, and the entire difference in what the two sources cost comes from that one ordering and nothing else.

Think about the household version for a moment. The shape is exactly the same. A household with a home loan pays the instalment first, every month, before it decides anything else. The household spends the residual on itself. Nobody sends the household a bill for the residual, and nobody promises it either.

What is this company actually borrowing, and how do three loans become one rate?

Real borrowing is almost never one loan. Borrowing is several separate loans, contracted at different times with different lenders on different terms, and each one is called a trancheOne separately contracted piece of a company's borrowing, with its own rate, security and maturity.. Sankalp Industrial Systems Limited has three at Year 0, and every rate below is the company's own contracted rate rather than any statement about what borrowing costs in the world.

Notice that the three rates are not ranked the way most people would guess. The secured term loan is the largest of the three and the one the lender can take security over, and it carries 7.80 per cent. The unsecured debentures, where the holder has no charge over anything, carry 8.50. The working capital facility, secured against receivables and inventory and renewable every year, carries the lowest of the three at 7.60. Security and tenure move a rate more than size does, and a company with three lenders has three separate answers when it is asked what it pays.

Getting one number out of three means weighting each rate by how much is drawn on it, never by counting the loans. Tranche 1 is half of the borrowing at 7.80 and contributes 3.9000 points. Tranche 2 is a third at 8.50 and contributes 2.8333. Tranche 3 is a sixth at 7.60 and contributes 1.2667. The three add to exactly 8.0000 per cent, the blended cost of debtThe average rate across every tranche, weighted by how much is drawn on each one. before tax. Working in crore reaches the same place: 300 times 7.80 is 2,340, 200 times 8.50 is 1,700, 100 times 7.60 is 760, the three sum to 4,800, and 4,800 over 600 is 8.00.

Averaging the three rates instead, as though each loan counted once, gives 7.9667 per cent. The gap between the two figures is not a rounding difference in disguise. The simple average answers a different question. The figure it gives is what the company would pay if it had borrowed the same amount on each tranche, and it did not.

Which of the borrowing figures does the mix actually rest on?

Gross Debt, and why the cash is not netted off first

Gross debtThe total amount borrowed, before deducting any cash the company happens to hold. is the total contracted amount outstanding, before any deduction. For this invented company it is Rs 6,00,00,00,000, being the three tranches added up. The company also holds Rs 1,20,00,00,000 of cash, so its net debtGross debt less cash and cash equivalents. is Rs 4,80,00,00,000. Both figures are correct and they answer different questions.

The mix is measured on gross debt for a reason that is easy to state and easy to forget. The lender's claim does not shrink because the borrower has money in the bank. The interest is charged on the whole Rs 6,00,00,00,000 whatever the cash balance happens to be on any given day, the covenants are written against the whole of it, and the instalment at the end of Year 5 falls due in full. Netting the cash off produces a number that describes how much of the borrowing could be repaid tomorrow if the company chose to. The repayment capacity is a useful thing to know. The weighting in a question about funding cost runs on the whole contracted amount instead.

Net debt earns its keep elsewhere. Against a year of trading profit it gives a sense of scale, and here net debt of Rs 4,80,00,00,000 against Year 0 operating profit before depreciation of Rs 2,88,00,00,000 is 1.67 times. Against the interest bill it gives a sense of comfort, and here operating profit of Rs 2,40,00,00,000 against interest of Rs 48,00,00,000 covers it exactly 5.00 times. Neither of those is the mix.

Should the mix be measured on the accounts or on the market?

The debt shareDebt divided by total capital, meaning debt plus equity. needs a denominator, and there are two candidates sitting on the same company at the same moment. The accounts of Sankalp Industrial Systems Limited carry equity of Rs 9,00,00,00,000, or Rs 45.00 a share. The market carries 20,00,00,000 shares at Rs 90.00, or Rs 18,00,00,00,000. Exactly twice the book amount, and that is arithmetic rather than a verdict on anything.

The two denominators give two answers. Gross debt over gross debt plus book equity is Rs 6,00,00,00,000 over Rs 15,00,00,00,000, or 40.0 per cent. Gross debt over gross debt plus market equity is Rs 6,00,00,00,000 over Rs 24,00,00,00,000, or 25.0 per cent. Fifteen percentage points apart, on one company, on one day, with nothing in dispute.

The mix is measured at market here, and the reason is that the two sides of the ratio are not symmetrical. A lender's claim is a contract at a stated amount, so what the accounts say about it is close to what it actually is. A shareholder's claim has no stated amount at all: it is worth whatever somebody will pay for it, and because this invented company is listed, somebody is paying Rs 90.00. Using market weightsWeights worked out on the market value of equity and debt rather than on the amounts in the accounts. means the equity side of the ratio is measured the same way the equity holder measures it.

There is a fair objection, and it is worth saying rather than hiding. The market price moves and the accounts do not, so the mix measured this way moves on days when the company has done nothing at all. The objection is true, and the movement is the price of the other property. A market weight reflects what a claim is worth today rather than what it once cost. Where a company is not listed there is no observed price to use, and estimating the weight is a different subject covered separately.

Does the mix change what the company is worth at all?

Capital structure was built as a subject to answer this question, and the first serious answer was no. Franco Modigliani and Merton Miller, writing in the American Economic Review in 1958, argued that under a stated set of conditions the way a business is funded cannot change what a business is worth. Their reasoning is worth carrying around because it is simple. The cash a business produces comes from what it does, not from how it was paid for. Splitting a claim on that cash into two pieces and handing them to two different people changes who holds what; it does not change how much cash there is to hold.

Their irrelevance resultThe finding that with no taxes and no distress the funding mix cannot change what a company is worth. rests on four conditions. There are no taxes. There are no costs of financial distress, and everybody borrows at one rate that does not change with how much they borrow. Managers and investors know the same things about the business. And what the company invests in does not change according to how the investment is funded.

The operating side of this invented company shows why the argument bites. Free cash flow to the firm in Year 1 is Rs 98,00,00,000. The figure comes from valves being made and sold and serviced. The funding sits on the other side of the balance sheet from the machines, so not one rupee of that cash moves when the funding changes. So if the mix is going to change the value of the company, it can only do so through the rate the cash is discounted at, or through the tax paid on the way. Nothing else is available to it.

Which of those conditions does this invented company break?

Three of them. The fourth is kept so that the arithmetic can be read at all. A condition is a place to look rather than a thing to memorise, so going through them one at a time repays the effort.

No taxes fails immediately. Sankalp Industrial Systems Limited assumes an effective tax rate of 25.0 per cent. The rate is the company's own assumption and not a statement about any tax system anywhere. Interest is deducted before that rate is applied and a dividend is not, so a rupee of interest and a rupee of dividend do not cost the same. The rupee arithmetic of the deduction follows below.

No distress costs and a single borrowing rate fails twice over. Sankalp borrows on security, with a charge over specific assets on two of the three tranches, and its lenders reprice as the borrowing grows. There is also a real event sitting in the schedule: the whole of Tranche 1, Rs 3,00,00,00,000, falls due in one instalment at the end of Year 5. A company with that in its diary behaves differently from a company without it, and behaving differently is exactly what a distress cost is.

Symmetric information fails the way it fails in every company. The people running a valve business know more about the order book, the warranty claims and the quality of the aftermarket revenue than the people pricing its shares do. The information gap is not a scandal but the ordinary condition of a listed company, and it has a consequence that shows up further on, in the order the company reaches for funding.

The fourth condition holds, and it holds because the forecast was built to keep it. Sankalp is assumed to reinvest the same Rs 1,00,00,00,000 of net new capital in every one of the five forecast years whatever its funding looks like. A fixed investment programme is what makes the whole exercise readable. If the investment moved with the funding, there would be no telling whether a change in value came from the mix or from the machines.

What is the interest deduction actually worth, in rupees?

The slogan version of this idea has done a great deal of damage, so say it as arithmetic instead. Interest of Rs 48,00,00,000 is deducted from the profit the tax is charged on. At an assumed effective rate of 25.0 per cent, that deduction reduces the tax bill by Rs 12,00,00,000. The reduction in tax is called a tax shieldThe reduction in a tax bill caused by interest being deducted from taxable profit., and Rs 12,00,00,000 is the whole of it for this year at this mix.

Run it the long way as a check. If the company borrowed nothing, operating profit of Rs 2,40,00,00,000 would be taxed at 25.0 per cent for a bill of Rs 60,00,00,000. As it is funded, interest of Rs 48,00,00,000 comes off first, leaving Rs 1,92,00,00,000 to be taxed for a bill of Rs 48,00,00,000. The difference is Rs 12,00,00,000, the same answer the short route gave. Two routes agreeing to the rupee is the sign that the idea has been understood rather than repeated.

The interest deduction is why the borrowing costs less after tax than before. The blended rate is 8.00 per cent pre-tax; at the assumed 25.0 per cent it is 6.00 per cent after tax, exactly. Against a cost of equityThe return shareholders require for holding the claim that gets paid last. of 14.00 per cent, settled before this subject and restated here rather than rebuilt, the borrowed money looks like the bargain of the two.

Two warnings travel with that sentence and neither is optional. The 25.0 per cent is this invented company's own assumed effective rate, not a rate any tax authority anywhere has set. And whether interest is deductible at all, whether there is any limit on how much of it is deductible, and what happens to a company whose borrowing is large relative to its capital are all questions of law that change, and the current text of that law is what settles them.

If borrowing is the cheaper source, why is it not free?

Because the price of the cheaper source is not paid in the rate. The price is paid in the promise attached to the borrowing, and the promise is not conditional on the year going well.

Sankalp Industrial Systems Limited owes Rs 48,00,00,000 of interest this year. Operating profit is Rs 2,40,00,00,000, so the interest is covered exactly 5.00 times over and nobody has to think about it. Now suppose two customers slow their orders and operating profit falls by two thirds. The interest bill is still Rs 48,00,00,000, to the rupee, on the same dates. The cover has dropped to about 1.67 times, the covenants written into the two secured tranches start to matter, and the company begins to make decisions it would not otherwise make: deferring the capital spending, stretching the payables, going to the lender before the lender comes to it.

None of those decisions appears in an interest rate, and all of them cost money. The cost of those decisions is the second half of the reason the mix is not irrelevant. A spreadsheet is worst at showing this half. The cost arrives as things that never happened rather than as a line item.

There is also the shape of the promise, not just its size. Tranche 3 is renewed every year, so it has to be agreed with a lender again in every one of the five forecast years. Tranche 1 falls due in one instalment, in full, at the end of Year 5. A company with an even repayment profile and a company with a single large date in the diary can carry the same debt share and be in quite different positions, and what a company does about a date like that is covered separately, further into this subject.

Why does the rate rise as a company borrows more?

Because the lender is pricing a claim whose safety depends on how much else has been promised ahead of it, and each new rupee of borrowing makes every earlier rupee slightly less comfortable. So the rate is not one number a company borrows at forever. The rate is a schedule, and a company moves along it.

Sankalp has an invented cost of debt schedule contracted with an invented lender. Its shape is the whole counterweight to the tax deduction, so it is worth reading carefully. At a 25 per cent debt share the pre-tax rate is 8.00 per cent. At 30 it is 8.25, at 35 it is 8.60, at 40 it is 9.00, at 45 it is 9.75, at 50 it is 10.75 and at 60 it is 13.00.

Two things about that schedule deserve saying out loud. The first is that the schedule carries almost no information at the low end and almost all of it at the high end. Between a 10 and a 25 per cent debt share the rate moves between 7.75 and 8.00 per cent, close enough to flat that nothing can be built on it. The first point of the schedule, at a zero debt share, is notional in any case: at a zero debt share there is no borrowing, so 8.00 per cent there is the rate the company would pay on its first rupee rather than a rate it pays on anything.

The second is that the steepening is not gentle. Going from a 25 to a 40 per cent debt share costs a point of rate. Going from 40 to 60 costs four. Whatever else is true, the last slice of borrowing on this schedule is priced nothing like the first.

So what does moving the mix do to the blended rate?

How Capital Structure Affects Cost of Capital

The weighted average cost of capitalThe blended cost of all the funding, weighted by how much of each is used. is the two component costs weighted by how much of each is used. For this invented company that is 0.75 times 14.00 plus 0.25 times 6.00, or 10.50 plus 1.50, or exactly 12.00 per cent. The 12.00 per cent is settled before this subject and is restated here rather than rebuilt. The eleven inputs behind it are covered separately.

The problem has two dimensions, so drawing it as area is the clearest way to see what moving the mix does. Width is how much of the capital sits in each source. Height is what that source costs. Area is what it contributes. Move the divider to the right and two things happen at once, in opposite directions.

The first force helps and the second hurts, and which one wins depends entirely on where along the schedule the company is standing. Force one is the weight: more of the width sits under the shorter block, and shifting width from a 14.00 per cent block to a 6.00 per cent block pulls the average down. Force two is the heights: as the borrowing grows the lender reprices, so the short block gets taller, and the shareholders are now standing behind a bigger fixed claim, so the tall block gets taller too.

Near the current mix the schedule is nearly flat and the weight is doing all the work, so the first force is winning. Push far enough and the second force takes over. The weight can only shift so far while both heights keep climbing. Which means the blended rate does not fall forever and does not rise forever; it turns somewhere in between. Where it turns for this invented company, what the table of it looks like row by row, and what it is worth in rupees are covered separately, further into this subject. An interior point exists, and that is what matters here.

The two questions want different treatments. Defining the mix and measuring it is one question. Where value peaks on one invented schedule is a computation with its own assumptions, its own flatness and its own warnings, and a number carried away from it without those warnings attached would mislead.

In what order does a company reach for its funding, and who says so?

Stewart Myers and Nicholas Majluf, writing in the Journal of Financial Economics in 1984, described an order that companies are observed to follow: internal cash first, then borrowing, then new shares last. The order is called the pecking orderThe observed order in which companies reach for funding: internal cash first, then borrowing, then new shares., and its interest is that it is not a preference. The order falls out of the information gap.

The argument runs like this. Managers know more about the value of the business than the people buying its shares do. If managers issue shares when they think those shares are worth more than the price, they hand value to the new holders and away from the existing ones. So they tend not to. Investors, who can work this out too, read an equity issue as a signal about what the managers think, and price it accordingly. A lender is repaid a fixed amount either way, so internal cash carries no such signal at all and borrowing carries very little.

Watch this invented company do it without ever announcing that it is doing it. Sankalp puts Rs 1,00,00,00,000 of net new capital to work in every one of the five forecast years. Rs 25,00,00,000 of that comes from new borrowing, drawn on the working capital facility. The other Rs 75,00,00,000 comes from cash the business itself generated. No shares are issued in any of the five years, and the funding it does use is three quarters internal and one quarter borrowed, exactly the order Myers and Majluf described.

What decides the mix once the arithmetic has had its say?

Less than most people expect, and that is the honest answer. The schedule that generates the answer is nearly flat across a wide stretch of the range, and every input in it is an estimate, so the arithmetic narrows the question and then goes quiet. The rest of the gap is filled by something other than a formula.

Four things do most of the filling. The first is how steady the operating profit is. A fixed claim is easy to carry against a steady stream and hard against a jumpy one. The second is whether the tax position can actually absorb the deduction, since a deduction against a profit that is not there is worth nothing. The third is what the lenders will actually lend and on what terms, and a negotiation is not a curve. The fourth is what is already in the diary. A company with a large instalment falling due keeps more room than a company without one.

No arithmetic on a mix establishes that the mix is right, safe or suitable for a particular company. Rightness is a conclusion about a specific situation, and a ratio cannot reach one. The arithmetic supports something narrower and more useful: what is being mixed, how to measure it without self-deception, what each source costs, and which of the four conditions the mix would have to be irrelevant under is failing.