What the Cost to Income Ratio Measures, and What It Hides

The cost to income ratio sets a financial institution's operating expenses over its total income for the same period. Operating expenses are what it costs to run the place, counted before provisions and before tax. Total income is net interest income plus other income, already net of what the institution paid for its own money. Both of those definitions have to travel with the number.

Two lines, one division, four seconds of arithmetic. And yet more analysts have gone wrong on this ratio than on almost any other measure a lender publishes. Both of its lines can be assembled more than one way out of the same set of accounts, and each way gives a different answer for the same institution in the same year. The ratio therefore carries almost no information until the two lines that produced it are named. Naming those two lines takes far longer than the division does, and the division gets a single sentence each time.

What is the cost to income ratio, before anything is built?

Operating expenses over total income, both struck over the same period. At Suvarna Commercial Bank Limited for the stated year, Rs 5,040 crore sits over Rs 9,840 crore. The division gives 51.22 per cent, and the whole measure is that division.

Now look at what that sentence carried which a bare formula does not. The sentence carried a definition of the top line, a definition of the bottom line, and a period that both were struck over. Every disagreement anybody has ever had about this ratio has been a disagreement about one of those three, and never about the division. Two institutions arguing about whose figure is better are usually not arguing about their spending at all. The argument, unstated, is about what counts as an operating expense and what counts as income.

What belongs in the numerator, and what does not?

Operating expenses are what it costs to run the institution. The people, the premises, the systems, the auditors, the electricity, the paper. Somebody had to be paid, or something had to be kept switched on, so that the institution could go on doing what it does. The cost of that sits here.

Three specific figures get pulled into this line by people who are being careful rather than careless, and none of the three belongs. Interest expended, provisions and tax are each a cost of the money, of lending gone wrong and of profit earned. Each one changes what the ratio measures the moment it goes in. Take them in turn.

Interest expended of Rs 11,160 crore is the cost of the money itself. Suvarna Commercial Bank Limited did not spend that on running anything. The bank paid that amount to the people whose money it was using, and the amount has already been taken out further up. The subtraction is exactly why the denominator is a net figure. Putting it back in the numerator would charge the institution twice for the same rupee.

Provisions of Rs 1,800 crore are the cost of lending going wrong. Provisions are a real cost and somebody should certainly be reading them, but they are not a cost of operating. The provision charge moves when borrowers stop paying. How carefully the institution manages its own spending has almost nothing to do with it. Mixing the two gives a figure that swings with credit quality while everybody goes on describing it as spending discipline.

Tax of Rs 750 crore is neither of those and is not an operating expense either. Tax is a charge on profit that has already been earned. An institution cannot run itself more cheaply by paying less tax, so a measure of how cheaply it runs itself has no business containing it.

Why is the denominator net rather than gross?

Total income is net interest income of Rs 7,440 crore plus other income of Rs 2,400 crore. The sum is Rs 9,840 crore. The net interest income figure is interest earned of Rs 18,600 crore less interest expended of Rs 11,160 crore, and the subtraction is the whole point of the word net.

Picture a tea stall for a moment. The stall takes Rs 4,000/- across the counter in a day and pays Rs 2,500/- of that to the milk and leaf supplier. Nobody sensible would say the stall has Rs 4,000/- to work with. The stall has Rs 1,500/-, and out of that Rs 1,500/- comes the rent, the boy who washes the glasses and the electricity. A lender is in the same position. Interest earned is what came across the counter; interest expended is the supplier's bill; net interest income is what is actually left to run the place on. A denominator built on interest earned would credit a lender for money that went straight back out. So the ratio is struck on income kept rather than income received.

Work the error out rather than warning about it. The size of the error is the argument. Interest earned of Rs 18,600 crore plus other income of Rs 2,400 crore is Rs 21,000 crore. The same Rs 5,040 crore of operating expenses over that gross figure is 24.00 per cent. The institution has not changed, the year has not changed, and its cost of running itself has just been reported as less than half of what it is.

How does one institution produce three ratios in one year?

Take the same Rs 5,040 crore and divide it three ways. Over total income of Rs 9,840 crore it is 51.22 per cent. Over net interest income of Rs 7,440 crore alone it is 67.74 per cent. Over interest earned plus other income of Rs 21,000 crore it is 24.00 per cent. One institution, one year, one numerator, and three answers spread across forty-four percentage points.

The three readings are not three approximations of one true figure, and treating them that way is the error underneath most of the confusion this ratio causes. Each division is exact. Each one answers a different question. Running the place costs how much per rupee of income the institution kept? Fifty-one and a bit. How much per rupee of interest margin alone, ignoring everything the institution earns from fees and other work? Sixty-seven and three quarters. How much per rupee that came across the counter before the funding bill was paid? Twenty-four. All three are true statements, and a figure quoted without its denominator has answered none of them.

Why is the operating margin already in hand?

Every rupee of total income at Suvarna Commercial Bank Limited either went out as an operating expense or stayed behind as operating profit. There is no third thing it could have become. So if Rs 5,040 crore is 51.22 per cent of Rs 9,840 crore, then Rs 4,800 crore of operating profit is the remaining 48.78 per cent of the same Rs 9,840 crore, and the two shares close at exactly a hundred.

A cost to income ratio and an operating margin are one bar read from opposite ends, so anybody who has computed one has already computed the other. This sounds like a small point and it saves a surprising amount of confusion. Readers routinely treat the two as separate measures that happen to be related, hunt for both in a set of accounts, and then worry when the second one they find does not match the first they computed. There is nothing to match. Subtract from a hundred and stop.

How does the whole build read on one institution?

The whole thing set down in one place shows which rows a published ratio draws on. The lines above the rule are the ones this measure uses. The lines below the rule sit in the same set of accounts and are outside the measure entirely.

Divide the numerator by the denominator and the stated year at Suvarna Commercial Bank Limited comes out at 51.22 per cent. Struck on net interest income by itself the same expenses read 67.74 per cent, and struck on interest earned plus other income they read 24.00 per cent. The three readings sit 43.74 percentage points apart at their extremes and 16.52 points apart between the two that a careful reader is most likely to meet, and every one of the three is arithmetically correct.

Can two published cost to income ratios be set against each other?

The ratio does most of its damage here, so define both sides fully before either is contrasted with the other.

Side one. Suvarna Commercial Bank Limited publishes 51.22 per cent for the stated year, struck as Rs 5,040 crore of operating expenses over total income of Rs 9,840 crore, where total income is net interest income of Rs 7,440 crore plus other income of Rs 2,400 crore.

Side two. Rukmini Finance Limited, for the same stated year, has operating expenses of Rs 540 crore and net interest income of Rs 1,386 crore, being interest earned of Rs 2,610 crore less interest paid of Rs 1,224 crore. Rs 540 crore over Rs 1,386 crore is 38.96 per cent. And one absence decides the rest: this record carries no other income line for Rukmini Finance Limited at all, so the only denominator that can honestly be built for it is net interest income, and its 38.96 per cent is therefore struck on a different base from the bank's published figure.

Take the absence seriously rather than reading it as a zero somebody measured. If Rukmini Finance Limited does earn something that is not interest, and this record simply does not carry it, then any such amount would only enlarge the denominator and push the ratio down. So 38.96 per cent is a ceiling on what its total-income-based ratio could be, not a measurement of it. A ceiling is a weaker claim than a reader wants, and it is the only claim the record supports.

Now do the comparison properly. Rebuild the bank on the same base the finance company is stuck with. Rs 5,040 crore over net interest income of Rs 7,440 crore is 67.74 per cent, against 38.96 per cent. On the published pair the two sat 12.26 points apart. Rebuilt like for likeReworked so that two figures are struck on the same base and over the same period. Nothing can be set against anything else until that has been done. they sit 28.78 points apart. The gap more than doubled, and the direction did not change.

Here is the harder point, and it is the one a reader almost never reaches alone. Neither of those gaps is a measure of how well either institution is run. Most of what moved between them is a single line: Suvarna Commercial Bank Limited has Rs 2,400 crore of other income and Rukmini Finance Limited has nothing recorded against that line at all. Earning Rs 2,400 crore that is not interest is work. Earning it takes people, systems and premises, and the cost of all three is sitting in the very expense line the ratio is dividing. So a lender that does more kinds of work will tend to show both a larger denominator and a larger numerator, and the net effect on the ratio depends entirely on which grew faster.

The rebuilt gap of 28.78 points mostly records that these two lenders are in different lines of business. The gap is not a verdict on spending discipline. Neither lender can be called the better run of the two. The record here holds one year, no downturn and no second period for either of them, and nothing in it could settle that question even if the question were asked properly.

What sits below the line where the ratio stops?

The cost to income ratio ends at operating profit. The stopping point is not a weakness anybody built in by accident. Stopping there is what the measure is for. But it does mean that everything under that line is outside its view, however carefully the two lines above it were assembled.

At Suvarna Commercial Bank Limited, provisions of Rs 1,800 crore sit below the line, and after them a profit before tax of Rs 3,000 crore and a tax charge of Rs 750 crore. At Rukmini Finance Limited, credit costs of Rs 396 crore sit below the line. A ratio that stops at a line cannot be asked a question about anything under that line, and the answer it gives when asked anyway is the answer to a different question.

Now a second comparison, and this one needs handling more carefully than the first. Per rupee of assets, Suvarna Commercial Bank Limited spends Rs 5,040 crore against total assets of Rs 2,40,000 crore. The bank's spending is 2.10 per cent of its assets. Rukmini Finance Limited spends Rs 540 crore against assets under management of Rs 18,000 crore, and its spending is 3.00 per cent of that book. Below the line, the bank's provisions of Rs 1,800 crore are 0.75 per cent of its total assets, and the finance company's credit costs of Rs 396 crore are 2.20 per cent of its assets under management.

The two lenders are not standing on the same base, so the denominators come before the gap. The bank's two readings are struck on a balance sheet total. The finance company's two are struck on the book it runs, named here as assets under management. A book run is not the same object as a balance sheet total. Four flows, two different bases, and not one of them a stock that could be set beside another. So the honest statement here is about direction rather than size: at these two lenders the larger difference sits below the line rather than above it, and the cost to income ratio was built to see only what is above it.

What can the ratio never show, however carefully it is built?

Suppose everything set out above has been done correctly. Both lines were taken off the accounts directly, interest, provisions and tax were kept out of the numerator, the denominator was built net, the base was written beside the answer, and the other institution was rebuilt on the same base before comparing. There is still one question the number cannot answer, and no amount of care will make it answer it.

The ratio cannot say whether the spending bought anything. The ratio counts what was spent as a share of income and has no term at all for what came back. The missing term is a limit of the measure, not a caution about using it. Two institutions can spend an identical share of income and one of them has put it into a branch network in towns where it actually lends, while the other has put it into an application almost nobody opens. Same numerator behaviour, same denominator, same printed figure. A reader who calls the lower figure the more efficient one has quietly swapped a question about spending for a question about results, and the ratio never held the second one.

How does somebody use this in an actual working week?

An analyst covering lenders meets this ratio several times a week, almost always in a summary table where the base is not printed beside the number. The working habit is short and it is the same every time. The two lines come off the face of the accounts rather than from a ratio somebody else struck. The two lines are divided. The base goes next to the answer in the note, in words. A reader coming back to it in four months then knows what it measures. Then the answer is subtracted from a hundred. Subtracting gives the operating margin, and it costs nothing.

A lender's own finance team reads it from the inside and uses it differently. For them the interesting movement is not the level but what the two lines did separately: whether the expense line grew because more work is being done or because the same work got dearer, and whether income grew alongside it. The question needs a second period. A single stated year per lender is all this record holds, with nothing before it and nothing after it, so no movement of any kind can be read from it. How a cost ratio behaves when income and costs grow apart is covered separately.

A household version makes the discipline stick. When a shopkeeper says the shop spends thirty per cent on running costs, the first question is thirty per cent of what. Of everything that came through the till, including what was paid to the wholesaler? Or of what was left after the wholesaler was paid? The two answers can differ by a factor of two on the same shop in the same month, exactly as they do at Suvarna Commercial Bank Limited between 24.00 per cent and 51.22 per cent. Operating leverageThe way an institution's profit moves faster than its income when a large part of its costs does not move with volume. Operating leverage is covered separately. and everything to do with how those two lines move over time belong elsewhere; the base is the thing to fix first.

One small thing before anybody goes hunting for round numbers. Suppose the question were what Rukmini Finance Limited would have to spend to print exactly 40.00 per cent on its net interest income of Rs 1,386 crore. In whole crore, there is no such figure. Rs 554 crore prints 39.97 per cent and Rs 555 crore prints 40.04 per cent, and nothing between them is a whole crore. A ratio quoted to two decimal places is finer than the amounts that produced it. Remember that before reading meaning into somebody's second decimal.