The bank's P&L: interest earned, NII, and what provisioning conceals

A bank's profit-and-loss account looks reassuringly small. A bank might carry ₹338,000 crore of assets and report ₹4,500 crore of profit - barely more than one percent of the balance sheet. illustrative That thinness is not weakness; it is the business. A bank does not earn a fat margin on a small amount of activity. It earns a wafer margin on an enormous pile of other people's money, and the whole art of reading its P&L is seeing where that margin comes from and which line decides how much of it survives to the bottom.

That line is provisions. A bank's reported profit is its pre-provision operating profit - the spread it earned, plus fees, minus running costs - less what it sets aside for loans that have gone or are going bad. The first part is hard to fake: it comes from visible spread, fee income and costs. The provision is a judgement - management's estimate of future losses - and it is the most discretionary number in the statement. Move it a little and reported profit moves a lot, in either direction, with no change in the actual business.

So this module reads the P&L for two things at once: how the margin is built, and how much the bottom line has been shaped by the provisioning judgement rather than by the business. There is no gross margin here to anchor on. Read a bank the way you read a manufacturer, and you will trust a number that was substantially chosen. The reported profit is the print the crowd reads; the pre-provision profit and the credit cost beside it are the signal, because they are the parts management cannot move without showing you.

Build the P&L one line at a time, from the spread up. Four ideas carry the whole reading.

Interest earned, less interest expended, is net interest income. The bank earns interest on its loans and investments and pays interest on its deposits and borrowings; the difference is the spread, and the bank's real top line. A bank with cheap funding - high CASA, a large share of low- or no-interest current and savings deposits - and disciplined lending earns a wide spread. One paying up for bulk deposits or chasing yield with risky loans earns a narrower or riskier one. Expressed against the earning assets, that spread is the net interest margin, which lets you compare a small bank with a large one.

Add other income, take out operating costs, and you have pre-provision operating profit. Fees, treasury and forex sit on top of the spread; operating expenses come off. What remains, before any provisioning, is PPOP - the earning power of the business before the most discretionary judgement is applied. Two banks with the same PPOP are earning the same operating profit, whatever their reported bottom lines say. This is the number to hold onto.

Then provisions land, and that is where the discretion lives. From PPOP the bank subtracts provisions for bad loans to reach profit before tax. This is the swing line. Provisioning is an estimate of future losses, chosen within a range: provide more, and reported profit falls though the book gets more honest; provide less, and profit rises though the losses merely wait. The credit cost - provisions as a percentage of the loan book - tells you how heavy the charge was this year, and comparing it against the trend in slippages (good loans turning bad) tells you whether the provisioning was adequate or convenient.

The discipline reduces to one rule: never read profit before tax without reading the PPOP above it and the credit cost beside it. PPOP tells you what the business earned; credit cost tells you what the bad loans took; and the gap between "PAT grew" and "PPOP grew" tells you how much of the reported growth was the lever rather than the bank.

Read the composite private bank's P&L in its fifth year, then watch what the provisioning line can do. illustrative

Interest earned is ₹22,600 crore; interest expended ₹12,800 crore; so net interest income - the spread - is ₹9,800 crore, a net interest margin of about 3.9% on earning assets. That margin is healthy because the funding is cheap: nearly half the deposits are CASA. Add other income of ₹3,600 crore, take out operating expenses of ₹6,000 crore, and pre-provision operating profit is ₹7,400 crore - the earning power of the business, before a single judgement about bad loans. Provisions of ₹1,400 crore then land, giving profit before tax of ₹6,000 crore and, after tax, ₹4,500 crore. Credit cost is a modest 0.75%.

Now the same bank's third year, the credit-stress year. PPOP that year was ₹5,200 crore - the business was still earning. But provisions jumped to ₹2,050 crore as bad loans rose, credit cost spiked to 1.35%, and profit after tax fell to ₹2,360 crore, almost half the good year. A reader watching only the bottom line sees a business in trouble. A reader watching PPOP sees a business still earning ₹5,200 crore of operating profit and taking its credit pain in one year - a very different, and more accurate, picture. That fall was the provisioning line doing its job, not the business collapsing. Misreading such a clean-up year as a bad year is the mirror image of the flattering trick, and just as common: over-provide once to clear the book and the profit drops precisely because the bank is getting healthier.

The lever cuts both ways. Suppose the bank had wanted that third year to look good. With PPOP of ₹5,200 crore it could have provided ₹1,000 crore instead of ₹2,050 crore, reported a far higher profit, and left half its credit losses unrecognised. Nothing about the business would have changed - only the judgement. The instrument below makes that physical.

Set the yield on advances and the CASA share - which fixes the cost of funds and therefore the spread - and watch net interest income, then PPOP, appear. Then move the credit-cost slider. Notice that PPOP does not budge - the business earned what it earned - while profit before tax swings sharply. That gap is the provisioning lever in one picture: the number most people read is the number management has most discretion over, and the cleaner one is sitting one line above it.

The spread-and-provisions structure is specific to lenders, and it is worth seeing exactly which parts have no equivalent elsewhere - because "financial company" is not one template but several.

The inversion to hold is that a bank's entire profit engine - a spread as the revenue, a provisioning line as the biggest judgement - simply does not exist for a manufacturer, which earns a gross margin and treats interest as a cost. And the asset-manager row makes the finer point: it earns fees with almost no credit risk, so it has no provisioning line to relax. Two companies can both sit in the financial sector and be read nothing alike.

Pre-provision operating profit strips out the provisioning discretion, but it cannot tell you whether the spread that produced it is safe or borrowed from risk. A bank can show a wide, stable PPOP by lending aggressively to risky borrowers at high rates - the margin looks great right up until those loans go bad and the credit cost catches up. This is why two identical net interest margins are not equally good: the same 3.5% spread can be earned on cheap deposits and prime loans, or squeezed out of expensive funding and risky lending. NIM sizes the spread; it does not tell you how safely it was earned, and a high margin from risky lending is a warning dressed as a strength - the yield is high precisely because the risk is. To judge durability you look at what the bank lends to and at what yield, which lives in the loan-book disclosures, not the P&L.

Nor does the credit-cost line tell you whether the provisioning is adequate, only what was taken. A low credit cost is reassuring if bad loans are genuinely low and falling, and alarming if slippages are rising while the bank provides less anyway. The number on its own is neutral; its meaning comes from the trend it sits against - slippages, the restructured book, the coverage ratio from the balance sheet. Reading credit cost without those cross-checks is how a deliberately light provision passes as strength.

And the P&L cannot catch a bank that avoids recognition rather than under-providing against recognised loans. Provisioning discretion is the visible lever; the invisible one is not classifying a loan as bad in the first place - evergreening it, restructuring it, keeping it standard. A bank doing that shows a clean credit cost and a healthy PPOP because the losses never entered the numbers. The P&L reads the loans the bank has admitted are troubled; the ones it quietly carries at par are a balance-sheet and disclosure question - and they are where the largest failures hide.

When reported profit jumps, a sharp analyst checks whether the business grew or the provisioning line was relaxed. The question sounds like this: "PAT grew 18% but PPOP was roughly flat and credit cost came down 60 basis points. How much of the growth is operating, and is the lower provisioning a genuine improvement in asset quality or a normalisation you can't repeat?" It separates the business from the lever.

A candid answer concedes the source and shows the asset quality behind it:

"Honest answer: most of the PAT growth this year is lower credit cost, not PPOP. Operating profit was up about 3%; the rest is provisioning normalising from last year's stressed 1.35% back to 0.75% - genuine recoveries and lower slippage, our gross NPA fell and coverage rose to 78%. But that's a normalisation, not a repeatable engine. Next year we'd guide credit cost around 0.75% and PAT tracking PPOP in low double digits. Don't extrapolate this year's headline."

It hands you the operating number, proves the improvement was real (coverage up, not down), and warns against extrapolation. Contrast the evasion:

"We're very pleased with our 18% profit growth, which reflects the strength of our franchise and disciplined execution. Asset quality continues to improve, our provisioning is prudent and adequate, and we remain confident in delivering healthy, sustainable growth going forward."

It claims the 18% as franchise strength while never mentioning that PPOP was flat and the growth was provisioning. "Prudent and adequate" is asserted, not shown against coverage or slippage. "Sustainable" implies the 18% repeats when it plainly does not. The follow-up that settles it - "excluding the change in credit cost, what was PAT growth, and what credit cost are you guiding next year?" - forces the headline back to the operating number. If "18% growth, strong franchise" is allowed to stand, investors capitalise a one-off release as if it were earning power. The silence is the tell: either PPOP growth is embarrassingly low, or the trend behind the lower credit cost would not survive the question.