How to Analyse Pakistani Bank Stocks: Rates, NIMs and NPLs

Whatever you think you are buying when you buy the Pakistani market, you are mostly buying banks. Commercial banks carry the largest single sector weight in the KSE-100 — roughly a fifth of the index as of mid-2026 — and sit near the top of daily turnover most sessions. If you hold an index fund, a pension scheme or almost any diversified PSX portfolio, bank earnings are your earnings.

Yet most retail investors analyse banks with tools built for industrials — revenue growth, gross margins, inventory — and get nonsense out. A bank does not make anything. It is a spread business, and in Pakistan the spread is set by one institution: the State Bank.

Why banks dominate the index

The KSE-100 weights companies by free-float market capitalisation, and the big banks are exactly what that methodology rewards: large, widely held, liquid (how the index is built). But there is a deeper reason. Banking is one of the few Pakistani industries whose raw material — deposits — grows automatically with inflation and nominal GDP. Cement demand can stall for years; deposit growth almost never does, because rupees have to sit somewhere. Add the fact that the state cannot politically afford a large bank failure, and you get a sector that compounds through crises that flatten everyone else.

You cannot have a view on the PSX without a view on banks — the sector breakdown makes that plain.

The engine: deposits in, spread out

Strip away the branches and the apps and a bank does one thing. It gathers deposits, pays as little as it can for them, and places the money where it earns more — loans to companies and consumers, or government securities. The difference between what it earns on assets and what it pays for funding, scaled across the book, is net interest income (in Pakistani accounts, net markup income).

The ratio everyone quotes is NIM — net interest margin: net interest income divided by average earning assets. It is to a bank what gross margin is to a manufacturer, with one enormous difference: a manufacturer's margin depends on its own pricing and costs, while a Pakistani bank's NIM depends overwhelmingly on the SBP policy rate — a number its management does not set.

The policy rate cycle is the earnings cycle

One pattern explains most of what bank stocks have done over the past decade. When the SBP hikes, banks' asset yields reprice within months: treasury bills roll over at the new rate, most corporate loans float against KIBOR. Their funding does not reprice the same way, because a large slice of deposits — current accounts — costs nothing at any policy rate. So when the policy rate sat at 22% through 2023-24, the highest in the country's history, banks earned 22% on a mountain of assets funded substantially at zero, and the sector printed record profits.

Then the cycle turned. As the SBP eased from mid-2024, the same arithmetic ran in reverse: asset yields fell with every cut, while free deposits cannot get cheaper than free. NIMs compressed, and reported profits with them — at precisely the moment the wider economy was improving.

That inversion is the trap. A bank posting record earnings at a 22% policy rate is not a better business; it is a cyclical business at the top of its cycle, and paying peak multiples on peak NIMs is how people buy cyclical tops. With banks, the honest question is never "what did it earn?" but "where in the rate cycle did it earn it?"

The government-paper crutch

A textbook bank lends to businesses. Pakistani banks, to a remarkable degree, lend to the government. Chronic fiscal deficits make the state the biggest borrower in the country, and for a bank the trade is irresistible: government paper pays well, cannot default in rupees, and requires no capital held against credit risk. At points in the peak-rate years the sector's investments — overwhelmingly government securities — approached the size of its entire deposit base, while the advances-to-deposit ratio (ADR: loans as a share of deposits) sat near historic lows around 40%.

The government has periodically tried to tax this behaviour away. Finance measures in the 2020s levied extra tax on government-securities income at banks whose ADR fell below set thresholds — producing a strange spectacle: in late 2024, banks shoved loans out the door before year-end and some imposed fees to discourage large deposits, purely to drag their ADR over a tax line. The ADR-linked levy was later swapped for a higher flat corporate rate on banks; the treatment shifts with nearly every Finance Act, so verify the current rules against FBR notices before building them into a forecast.

The lesson: lending here responds to tax design and sovereign borrowing at least as much as to private credit demand. When you read "advances grew 20% in December", check the calendar before applauding.

NPLs: the cost that arrives late

A loan roughly 90 days overdue becomes non-performing. NPLs hurt earnings through provisions — charges booked against profit to cover expected losses. The NPL ratio (non-performing loans as a share of gross advances) tells you how much of the book has gone bad; the coverage ratio (provisions held against those NPLs) tells you how much of the damage is already absorbed. Coverage above 100% means the bad loans are fully provided for, with a buffer.

Provisions are where lending sins surface, years late. A bank can grow advances aggressively through the good years and hand the profits back in one bad cycle; expected-loss accounting pulls some of the pain forward, but credit losses still lag credit growth.

The Pakistani twist follows from the crutch above: because so much of the balance sheet is sovereign paper, loan books are cleaner than the economy would suggest. A bank that barely lends to the private sector barely has NPLs — a low NPL ratio may be evidence of not lending, not of skill.

CASA: the quality of the raw material

Two banks with identical deposit totals can be very different businesses. What separates them is CASA — the share of deposits sitting in current and savings accounts. Current accounts pay nothing. Savings accounts at conventional banks are bound by an SBP minimum rate linked to the policy rate. Fixed deposits cost close to market, so a high-CASA bank funds itself far more cheaply than one renting deposits at term rates.

For years Islamic banks were exempt from the minimum savings rate that binds conventional banks — a structural funding advantage that helps explain MEBL's extraordinary profitability, alongside genuine depositor demand for Shariah-compliant banking. The SBP has since moved to narrow that gap, so check current rules before assuming the edge persists at full strength.

Deposit franchise is the closest thing banking has to a moat. Rate cycles give NIMs and take them away; a customer base that keeps free money with you is yours.

Reading a bank's accounts

Bank statements will disorient you if you learned on industrials. There is no revenue line, no cost of goods, no inventory. The income statement runs: markup earned, markup expensed, net markup income, then non-markup income (fees, FX, capital gains on those government bonds), then provisions and admin costs. The balance sheet is not a footnote to the business — it is the business. The general literacy in reading PSX financial statements still applies, but the ratios shift: P/E misleads at cycle turns because the E is cyclical, and book value becomes the steadier anchor.

The cheat sheet

Valuation: P/B with ROE as the anchor

The workhorse for banks is price-to-book read against return on equity. Book value is what shareholders own, ROE is what the bank earns on it, and a bank persistently earning more than investors require deserves a premium to book; one earning less, a discount.

The catch in Pakistan is the hurdle. When the risk-free rate has spent most of recent memory in double digits, peaking at 22%, the return equity investors demand is punishing. A bank earning a 20% ROE — spectacular by global standards — may only just clear it. That is a large part of why Pakistani banks so often trade below book, and why "below book, therefore cheap" fails as an argument. The discount is deserved when ROE sits below the hurdle, coverage is thin, or the deposit base is expensive. It is undeserved when the market is capitalising trough-cycle NIMs as if they were permanent — which is exactly when patient money gets interested.

The listed names span the framework rather than rank it: HBL is the largest by assets, UBL carries a heavy international and remittance footprint, MCB and BAHL have historically been the high-CASA, high-ROE franchises, and MEBL runs the Islamic model at a premium P/B the market has usually judged its ROE to deserve. Put any two side by side on the compare tool and the differences in ROE, payout and valuation are immediate. Treat them as worked examples, not picks.

What to watch each results season

Most results-day commentary fixates on the profit figure — the least interesting line on the page. Instead:

1. NIM direction against the rate cycle. Compression while the SBP eases is arithmetic, not failure; expansion while rates fall means the bank is winning on mix.
2. Deposit growth and CASA mix. The franchise measure. A profit beat built on expensive fixed deposits is borrowed, not earned.
3. The provisioning line, in both directions. Charges reveal credit pain; large reversals can flatter a weak operating quarter.
4. Fee and other non-markup income. The only earnings stream the policy rate does not own.
5. Capital adequacy and payout. Dividends are why most people hold these stocks; capital headroom makes them sustainable — check current minimums against SBP circulars, not memory.

And one closing discipline. When a Pakistani bank reports record profits, check the policy rate before applauding. When it reports a lean year, check whether the free deposits still arrived. The first tells you where the cycle is; the second tells you whether the franchise is intact. Only the second is worth paying a premium for.