Financial Health of an Institution: The Signals That Matter

Four separate questions sit inside the single question of whether an institution is healthy. Capital asks whether it can absorb losses. Asset quality asks whether the claims it holds are still good. Earnings ask whether it makes enough to replace what it loses. Liquidity asks whether it can pay what falls due today. An institution can pass three of the four, and the fourth still ends it.

Every one of those four signals arrives as a ratio, and a ratio is two decisions rather than one. Somebody chose what goes on top, and somebody chose what goes underneath. Two people can put the same numerator over two different denominators, produce two different numbers, and give both of them the same name. Almost all of the confusion in reading an institution comes from that second decision. The discipline that follows is small and it never lets up. A figure arrives with the base it was struck on, in the same sentence, and where the base is missing the figure is not computed at all.

What is actually being asked when somebody asks whether an institution is healthy?

Start by refusing the word. Healthy is one word for four questions, and the four do not answer each other. Can it absorb a loss that has already happened, are the claims it holds still good, does it earn enough to replace what it loses, and can it pay what falls due today are four separate readings taken on four separate things. The first is capital, the second is asset quality, the third is earnings and the fourth is liquidity. Nothing forces them to agree, and they routinely do not.

Here is the everyday version, and it is worth holding on to because the finance version is the same shape. A household has a salary coming in every month, no borrowings, and a flat it has finished paying for. On any sensible reading that household is in good condition. And on the fifth of the month the rent on the shop it runs falls due, the salary lands on the seventh, and there is nothing in the account on the fifth. Nothing about the flat helps on the fifth. The flat is capital, the salary is earnings, and the empty account on the fifth is liquidity, and the fourth question does not care what the other three said.

The fourth question is also the one that acts fastest. Capital erodes over quarters. Claims stop performing over months. Earnings sag over a year. The demand on an institution can change inside a morning, and the signal that looks the least interesting in a printed statement is the one that ends institutions.

What does capital indicate, and what is it measured against?

Capital is the owners' money inside the institution, and its job is structural rather than decorative: it stands between a loss and everybody else who has a claim. When something goes wrong, the owners absorb it first, and the depositor, the bondholder and the counterpartyThe other side of a deal, whose own promise has to be kept for the deal to work out as written. only reach the loss after the owners' money has gone. The ordering of who absorbs a loss first is the whole reason capital is watched.

Suvarna Commercial Bank Limited reports net worth of Rs 24,000 crore against total assets of Rs 2,40,000 crore. Divide one by the other and capital is 10.0 per cent of total assets. Saying the bank carries assets of 10.0 times its equity states the same fact. Both readings come from the same two figures and both name their base. Deposits of Rs 1,92,000 crore account for 80.0 per cent of the funding; what the rest of the funding consists of is not broken out in these figures.

Now the part a reader will expect and will not get. The capital ratio a regulator actually watches is not struck on total assets at all. The base it is struck on is built by risk weightingCounting a safer claim for less than a riskier one when adding up the base that a ratio is measured against, so that two books of the same size can produce two different bases., and a claim on a government does not count for the same amount as a claim on a borrower nobody has lent to before. The figures for Suvarna Commercial Bank Limited carry no such base. So no capital adequacy ratio can be computed from these figures. Dividing by total assets anyway would produce a number that looks like the regulatory one, sits in the same range, and answers a different question.

Capital Adequacy vs Solvency: which question does each one actually answer?

Capital adequacy and solvency get used as though they were the same word, and they are not even the same kind of thing. Solvency is a state, and capital adequacy is a measured ratio. Solvency asks whether the value of everything the institution holds exceeds everything it owes. If all of it were realised at the value recorded, every claim could be met and something would remain. Capital adequacy asks a harder and narrower question: is the capital enough for the risk this institution has taken, judged against a base somebody else defines and a level somebody else sets.

Three consequences follow, and each is usable immediately. An institution can be solvent and still fall short on the ratio. A positive total says nothing about how much risk sits behind it, and a book of risky claims needs more capital than the same rupees of safe ones. An institution can be solvent and still fail. Being able to meet claims eventually is not the same as meeting them on Tuesday. And the ratio is struck on a reporting date, so it is a photograph rather than a film: it describes the institution as it stood on one day, and says nothing about the day after.

The everyday version again. A flat worth more than the loan against it is solvency. Whether the bank would lend against that flat given what else the owner has borrowed is closer to adequacy. Whether this week's rent is in hand is neither, and it is the one that decides whether the owner sleeps.

What does asset quality indicate, and why does the denominator move?

Asset quality asks whether the claims the institution holds are still good. A lender's assets are promises other people made to it, and some of those promises stop being kept. The reading is built in three steps and the third one is where readers slip.

Suvarna Commercial Bank Limited reports a non-performing advanceA loan whose payments have stopped arriving as agreed, counted as such from the point the rule maker says it stops being treated as performing. book of Rs 6,480 crore. Against gross advancesThe total amount lent out, before anything set aside against it has been taken off. of Rs 1,44,000 crore that is 4.50 per cent of gross advances. The bank has set money aside against those advances at provision coverageThe share of the advances that have stopped performing which the lender has already set money aside against. of 70.0 per cent, and the amount held is Rs 4,536 crore. Rs 1,944 crore of the bad book is therefore still uncovered. Net advancesGross advances after the amounts already set aside against them have been taken off. are Rs 1,44,000 crore less Rs 4,536 crore, or Rs 1,39,464 crore. Rs 1,944 crore over Rs 1,39,464 crore is 1.39 per cent of net advances.

The gross ratio and the net ratio are not struck on the same denominator, and a reader who puts both over the same number has got one of them wrong. The first sits over Rs 1,44,000 crore and the second over Rs 1,39,464 crore. Putting the net figure over gross advances instead gives 1.35 per cent. The error is small enough that nobody notices and large enough to matter when the same slip is repeated across a screen of institutions.

There is a second point hiding in the same arithmetic, and it is the more interesting one. The coverage of 70.0 per cent is a decision the lender made, so the net figure carries a judgement inside it while the gross figure does not. Gross is close to a count. Net is a count after somebody has stated how much of it they think is already dealt with. Both are worth reading, in that order, and the net one is never worth reading alone. The day a loan stops counting as a performing one, and the money parked against it from that day on, are both settled by the Reserve Bank of India at rbi.org.in.

What do earnings indicate, and which base is the margin struck on?

Earnings answer whether the institution generates enough to replace what it loses. A lender that loses a little every year and earns a lot every year is in a different position from one that loses the same amount and earns nothing, even though the losses are identical. Start with the largest line and watch the base do all the work.

Suvarna Commercial Bank Limited earned interest of Rs 18,600 crore and paid interest of Rs 11,160 crore, so net interest income is Rs 7,440 crore. Over earning assetsThe assets that actually produce interest, being the lending plus the market instruments, and not the branch building or the cash in the till. of Rs 2,04,000 crore, being advances of Rs 1,44,000 crore plus market instruments of Rs 60,000 crore, that is a margin of 3.65 per cent of earning assets. Over total assets of Rs 2,40,000 crore the very same Rs 7,440 crore is a margin of 3.10 per cent of total assets. The entire difference between 3.65 and 3.10 is the choice of denominator, and nothing else moved at all.

The bridge is one line: earning assets are 85.0 per cent of total assets, so the reading on total assets is 0.85 times the reading on earning assets. Worked on the exact figure the bridge is precise. Rs 7,440 crore over Rs 2,04,000 crore is 3.6471 per cent, and 0.85 of that is 3.10 per cent to the last place. Worked on the printed 3.65 the bridge gives 3.1025, and 3.1025 prints as 3.10 without being the same number. The gap between a figure computed from exact inputs and the same figure computed from printed ones is not a curiosity, and it returns below in a form that has confused careful readers.

Two more earnings readings, each with its base attached. Operating expenses of Rs 5,040 crore against total income of Rs 9,840 crore, being net interest income of Rs 7,440 crore plus other income of Rs 2,400 crore, is 51.22 per cent of total income. And profit after tax of Rs 2,250 crore over total assets of Rs 2,40,000 crore is a return of 0.94 per cent on assets, or 0.9375 per cent before it is rounded for printing. Hold on to that last figure. Why a bank ends up lending at a higher rate than it funds at, and how its mix of funding moves that gap, is worked in full separately.

What does liquidity indicate, and why can a solvent institution still fail?

Liquidity asks the narrowest question of the four, and it is about a day rather than a year: is the cash there when somebody asks for it. Not eventually. Today.

The reason this is a permanent condition rather than a mistake is structural: the promises an institution has made to the people who funded it come due sooner than the promises made to it by the people it funded. A deposit can be asked for now. A loan against a workshop comes back over seven years. Turning the second into the first is not a flaw in the arrangement, it is the arrangement, and it is precisely what the institution exists to do. Which parties perform that transformation, and on what, is mapped out separately.

Follow what that means and then sit with it. Every borrower can be paying exactly on schedule, every claim the institution holds can be perfectly good, and the institution can still be unable to meet what is asked of it this week. No amount of capital fixes that on the day. Capital is not a pile of cash sitting in a room; it is a residualWhat is left once everything owed has been taken off, rather than an amount anybody set aside in a particular place., and most of it has already been lent out. Ten shops in one mall share one driver, and if the mall shuts on Monday every one of them is short on Monday, however well each of them was trading on Friday.

No liquidity ratio can be computed here, for the same reason no capital adequacy ratio can. The amount a bank must hold against an assumed run, the run it must assume, and the stable funding it must carry against the book it holds are all set by the Reserve Bank of India at rbi.org.in, and these figures carry none of the bank's own numbers for any of them.

What does no single one of these numbers reveal?

Four refusals, and they are short on purpose. A ratio without its base is not a reading. The figure is a number with a name attached, and the name is doing work the arithmetic has not done. The base underneath has to be established before the figure on top can be accepted.

A figure struck on a reporting date is not a description of today. The bad advance book was Rs 6,480 crore on one date. The margin covers one period. Every reading here is a photograph, and the institution kept moving after the shutter closed.

A single observation is not a direction. One period for each institution, with no series behind it, cannot show improvement or deterioration. A direction needs at least two readings struck on the same base. The arithmetic on a single observation is exact and its reach is almost nothing, and that is an uncomfortable pair of facts to hold at once and worth holding anyway.

And a test applied to one institution and not to the one beside it is not a comparison. A margin struck on earning assets for one lender must be struck on earning assets for the other. A return on equity decomposed for one must be decomposed for both. A test belongs at every level or at none. A test applied unevenly produces a ranking of the analyst's own choices rather than of the institutions.

What happens when two institutions report the same headline number?

The two institutions in this comparison were built to collide.

Suvarna Commercial Bank Limited returns 9.38 per cent on equity. The bank reaches that figure as profit after tax of Rs 2,250 crore over net worth of Rs 24,000 crore, and the two limbs underneath are a return of 0.9375 per cent on assets multiplied by assets of 10.0 times equity. Rukmini Finance Limited returns 9.38 per cent on equity too. Its route there is profit after tax of Rs 337.50 crore over net worth of Rs 3,600 crore, and its limbs are a return of 1.875 per cent on assets multiplied by assets of 5.0 times equity. Both institutions land on exactly the same figure, and the equality was designed rather than discovered; an identical number arriving without explanation would otherwise read to a careful reader as a transcription error.

Now the trap inside the trap, and it is the reason the exact limbs are printed above. The reported limbs are rounded for printing: 0.94 and 1.88. Multiplying the printed limbs gives 9.40 on both institutions, not the 9.38 that is also printed, and a reader who checks the arithmetic concludes that they themselves have made a mistake. Nothing is wrong with any of the four figures. The rounded limbs simply cannot reproduce each other. The product only ties at the unrounded 0.9375 and 1.875, both of which give exactly 9.375 per cent. A decomposition is only checkable when the exact limb stands beside the reported one.

The collision demonstrates something worth more than the ratio itself. Half the return per rupee of assets and twice the assets per rupee of equity produce the same answer, so a reader ranking institutions on this one figure cannot tell these two apart, and they are not remotely the same business. One earns a thin margin on an enormous book funded largely by deposits. The other earns a wide one on a fifth of the size, funded in the market, with a fifth of the assets per rupee of equity behind it. Look at the funding line and the difference is unmissable: deposits are 80.0 per cent of the bank's assets and borrowings are 80.0 per cent of the finance company's, identical to the decimal, and yet one carries 10.0 times its equity and the other 5.0 times. The reconciliation is that equity is 10.0 per cent of assets at the bank and 20.0 per cent at the finance company, and the bank carries a further layer of liabilities that the finance company does not.

Neither arrangement is better than the other here. Higher leverage is not recklessness and a wider spread is not skill. Each institution carries one period, not a single failure and not one turn of a cycle. One period is not evidence about what either way of running money delivers over time, and only a full turn of a cycle could supply that evidence. The rule of practice that does follow is this: a return on equity is never quoted without the return on assets and the leverage that produced it, in the same place, for every institution being compared.

Who actually reads these four signals, and what do they do with them?

A credit committee at one lender, asked to open a line to another institution, reads them in a fixed order and stops at the first one that fails. Capital comes first, and it sets how much loss the counterparty absorbs before the line is touched. Asset quality comes next, gross before net, and the net figure carries the counterparty's own judgement about coverage. Earnings come after that. An institution that earns enough to replace what it loses can absorb a bad year without touching capital at all. Liquidity is listed last and bites first, and the line usually carries a limit on how much can be drawn on any one day.

An analyst covering the same two institutions does something narrower and more useful: they refuse to rank on any single figure and instead put the decomposition beside every headline. The habit that separates a usable comparison from a screen is writing the two limbs under every return on equity before the institutions are lined up at all. A household is doing a smaller version of the same thing when it asks not only what a deposit pays but who stands behind it, and where the cover on that deposit begins and ends. The cover is a question for the Deposit Insurance and Credit Guarantee Corporation at dicgc.org.in rather than for anybody selling anything.