Asset Quality: Gross NPA, Net NPA and What They Miss

Asset quality asks how much of what a lender is owed has stopped being paid on its terms. Two ratios report it as non-performing assets (NPA), and the two do not share a denominator. Suvarna Commercial Bank Limited's gross non-performing advances of Rs 6,480 crore are 4.50 per cent of gross advances, and its net non-performing advances of Rs 1,944 crore are 1.39 per cent of net advances, a smaller base.

Every rupee below belongs to Suvarna Commercial Bank Limited, a lender invented for teaching, and covers one reporting year. Five separate rule sets decide which advances land inside the two ratios, and every one of the five is revised from time to time. A requirement recalled from memory goes stale silently. The address at which each rule can be read is given below in place of the rule itself.

Provisions come off both sides of the net ratio. A provision writes down the carrying value of the advance book. So the bad loans fall by the provisions held against them, and the book falls by exactly the same amount. A reader who takes provisions off the numerator and leaves the denominator alone has taken half a step, and half a step lands 0.04 points away from the right answer with nothing anywhere to signal that something went wrong.

What does asset quality actually ask about a lender's book?

Picture a cloth shop on a lane where half the trade runs on trust. The shopkeeper keeps a credit book: fourteen names, each with an amount and a date it was promised by. Some names pay every month without being asked. Two have not paid since the monsoon and stopped taking calls. Asset quality is the single question of how much of that book has stopped moving on the terms it was written on, and nothing else. It is not asking whether the shopkeeper was right to extend credit to those two names, and it is not asking how much will eventually be recovered from them once a cousin intervenes.

Hold those three questions apart. A ratio answers exactly one of them. How much has stopped being paid is knowable today, from the books, as a number. How much will finally be lost is knowable only later, and is a forecast until then. Whether the lending was sound in the first place is a judgement that never becomes a number at all. Suvarna Commercial Bank Limited's non-performing ratios speak to the first of those three and are silent on the other two. The silence is not a weakness in the measure so much as the measure's actual size.

What is the gross non-performing ratio, and what is its denominator?

The gross non-performing ratio is gross non-performing advances divided by gross advances. Say the word gross twice, once about each side, and the definition is complete. Gross on the numerator means the bad loans counted before any provision is taken off them, and gross on the denominator means the whole advance book counted before any provision is taken off it. Both sides of the line therefore stand on the same basis. On Suvarna Commercial Bank Limited's own lines that is Rs 6,480 crore over Rs 1,44,000 crore, which divides to 4.50 per cent, on gross advances.

Two things about that fraction are worth slowing down for. The first is that the numerator counts amounts rather than borrowers: one large advance that stopped paying can outweigh two hundred small ones that are perfectly current, and the ratio does not say which of the two situations is in front of the reader. The second is that the numerator only contains what has been through classificationThe act of moving an advance out of the performing category and into a non-performing one, on rules that fix the point at which it happens.. An advance the lender has not yet moved across that line is sitting quietly in the denominator only, doing nothing to the ratio at all.

What is the net non-performing ratio, and why is its denominator different?

The net non-performing ratio is net non-performing advances divided by net advances, and each of those two words is doing work. Net non-performing advances are the gross bad loans less the provisions the lender already holds against them. Net advances are the gross advance book less the same provisions. Both sides move to the net basis together. The net ratio is therefore a different fraction, not the gross fraction with a smaller number on top.

Suvarna Commercial Bank Limited holds provisions of Rs 4,536 crore against gross non-performing advances of Rs 6,480 crore, provision coverageThe share of a lender's bad loans it already holds money aside against. Provision coverage is built as its own measure elsewhere and enters this division only as the amount subtracted. of 70.0 per cent. Take that Rs 4,536 crore off the bad loans and Rs 1,944 crore is left. Take the identical Rs 4,536 crore off the book and gross advances of Rs 1,44,000 crore become net advances of Rs 1,39,464 crore. The reason the same amount comes off both is that a provision is not a separate pot standing beside the balance sheet: it reduces the carrying valueThe amount an asset is shown at in the accounts after everything that has been taken off it, rather than the amount originally handed over. of the advances themselves.

Gross vs Net NPA: why do the two ratios never share a denominator?

Here is the sentence the whole argument turns on. A provision reduces what the advance book is carried at, so the rupees that come off the bad loans are the same rupees that come off the advances. The gross ratio and the net ratio are therefore two different fractions rather than one fraction with a smaller top, and a reader who moves only the numerator has built a third number that belongs to neither.

Watch what that does to the size of the answer. The same Rs 4,536 crore stands at 70.0 per cent against the bad loans of Suvarna Commercial Bank Limited and at only 3.15 per cent against its advance book. Subtract it from both. The numerator loses most of itself and the denominator barely notices, and that is exactly why 4.50 per cent on one basis becomes 1.39 per cent on the other. The gap is not a matter of presentation. The gap is the arithmetic of taking a large bite out of a small number and a small bite out of a large one.

How do both ratios come out of this lender's own lines?

Build them in order, and keep the word gross or net attached to every number as it is written down. Suvarna Commercial Bank Limited's gross advances are Rs 1,44,000 crore, and the gross non-performing advances sitting inside that book are Rs 6,480 crore, so the gross ratio is 4.50 per cent, on gross advances. Provisions held are Rs 4,536 crore. Net non-performing advances are Rs 6,480 crore less Rs 4,536 crore, or Rs 1,944 crore. Net advances are Rs 1,44,000 crore less the same Rs 4,536 crore, or Rs 1,39,464 crore. The net ratio divides Rs 1,944 crore by Rs 1,39,464 crore and prints 1.39 per cent, on net advances.

Two of those rows are shaded because they are correct and two are shaded because they are not, and the arithmetic that produced all four is identical in kind. The difficulty sits in one image: nothing about performing a division reveals whether the right pair of numbers was picked up to divide.

What are the two ways a reader gets that division wrong?

A pair can be mismatched in exactly two ways, and both are worth naming. Error one puts the net numerator over the gross denominator: divide Rs 1,944 crore by Rs 1,44,000 crore and 1.35 per cent comes out. Error two puts the gross numerator over the net denominator: divide Rs 6,480 crore by Rs 1,39,464 crore and 4.65 per cent comes out. The first understates the correct 1.39 per cent by 0.04 points and the second overstates the correct 4.50 per cent by 0.15 points.

The understating error is far the more dangerous of the two, and the reason is that its gap is small enough to look like a rounding difference. Nothing about 1.35 per cent announces itself as wrong. The wrong figure sits in the range a reader expects, survives being copied into a note, and by the time it is quoted in a third place nobody remembers which pair of numbers produced it. The overstating error is the same mistake wearing a warning label: 4.65 per cent against a widely quoted 4.50 per cent invites somebody to check, and checking finds it.

One habit prevents both errors. Gross or net is said about the numerator and about the denominator, out loud, in the same sentence, every single time. Not once at the top of a note and then abbreviated afterwards. The abbreviation is where the two pairs get mixed up. The full form costs four extra words and it is the only defence that still works on a tired reader.

What do both ratios leave out entirely?

Three kinds of trouble sit outside both figures, and any one of them can be larger than the difference between the two ratios just separated. The first is advances that are showing strain but have not been classified: a borrower paying late, or paying from a source that will not last, still counts as performing until the classification rules move it. The second is advances that have been restructuredGiven changed repayment terms after the borrower struggled to pay. Rules set by the Reserve Bank of India decide how the advance must be reported afterwards., which may sit outside the non-performing figure depending on rules set by the Reserve Bank of India. The third is advances that have been written off and have left the gross figure altogether.

So the two ratios describe the part of the book that has already been classified and reported, and say nothing whatever about the part that has not been. That is worth holding on to without turning it into cynicism. A lender reporting 4.50 per cent gross is stating something real and checkable. The lender is simply not stating the size of what is queued behind that number, and no disclosure sizes the three absences, so they can be named but not sized.

What does a write-off do to both ratios?

Go back to the cloth shop for a moment. Chasing two of the names has cost more than they owe. The shopkeeper takes a pen and rules a line through both in the credit book. The book now shows fourteen names minus two, and the share of it that has stopped moving has fallen. Not one rupee has arrived, and in most cases the two names still owe every paisa of it. Ruling the line changed the record, and it did not change the world.

Now the arithmetic, sharper than the story. A write-off removes the advance from the bad loan figure and from the advance book together, so the gross ratio falls for certain. The net ratio moves or holds still depending on how much provision was already held against that particular advance, and the case worth knowing is the extreme one. If the advance was fully provided, the provision leaves with it: net bad loans and net advances are both unchanged, and the net ratio does not move at all.

Suvarna Commercial Bank Limited reports no write-off figure, so an amount is assumed purely to watch the arithmetic move. The assumed amount is Rs 500 crore, fully provided, and that Rs 500 crore appears nowhere in this lender's own lines. Gross bad loans go to Rs 5,980 crore and gross advances to Rs 1,43,500 crore, so the gross ratio falls from 4.50 per cent to 4.17 per cent, on gross advances. Provisions held fall to Rs 4,036 crore. Net bad loans stay at Rs 1,944 crore and net advances stay at Rs 1,39,464 crore, so the net ratio stays at 1.39 per cent, on net advances, to the second decimal and beyond it.

If the advance had been only partly provided, both ratios would have fallen. The unprovided part comes off the numerator and the denominator together, and taking the same amount off both ends of a fraction smaller than one pulls it down. Either way the direction of travel is downward and the cash collected is nil. Two lenders with identical books and different write-off habits will report different ratios. So the figure to ask for before believing any movement in a lender's non-performing ratios is the write-offs for the period. What has to be disclosed about write-offs is settled by the Reserve Bank of India.

What can a reader actually conclude from one year of these figures?

A level, and nothing else. Suvarna Commercial Bank Limited's book stood at 4.50 per cent gross non-performing advances on gross advances and 1.39 per cent net non-performing advances on net advances at one date, on stated bases, with coverage of 70.0 per cent sitting between them. Both readings are real and both can be quoted. A direction needs a second observation, and one year of figures carries only the first, so no statement about direction can rest on it.

Three things would have to be present before improving or worsening could be said out loud, and one year of figures holds none of the three: an earlier year to compare against, a write-off figure to see what left the book rather than got repaid, and a restructured book to see what was moved rather than cured. Writing that down is a result in its own right and not an absence of one. A reader who produces a direction out of a single year has not found a second observation but invented one, and the sentence reads exactly the same either way.

What is worth keeping about asset quality?

Four sentences, and everything above hangs off them.

1. The two ratios do not share a denominator. Gross non-performing advances go over gross advances, net non-performing advances go over net advances, and the provisions come off both sides at once.
2. Say gross or net about the numerator and about the denominator, in the same sentence. That one habit removes both misdivisions, the 1.35 per cent that reads as rounding and the 4.65 per cent that overstates.
3. What sits outside both ratios can be larger than the gap between them. Strain not yet classified, restructured advances and written-off advances are all outside, and neither ratio sizes any of the three.
4. A write-off improves the reported picture without collecting anything. Ask for the write-offs for the period before believing any movement, and remember that a fully provided write-off leaves the net ratio exactly where it was.