How to Read a Bank's Earnings: NIM, Provisions, Efficiency, and the Stock

Bank earnings look intimidating because the income statement is structured differently from an industrial. Revenue isn't "sales" — it's net interest income plus fees. Cost of goods doesn't exist — instead you have interest expense and loan losses. Once you know which four or five numbers actually matter, a 60-page 10-Q becomes a 10-minute read.

Here is the framework I use, in roughly the order the market cares about them.

Net interest margin (NIM): the core profit engine

NIM is the spread between what a bank earns on its assets (loans, securities) and what it pays for its funding (deposits, borrowings), expressed as a percentage of earning assets. A NIM of 3.0% on $400B of earning assets generates $12B of net interest income per year. That's usually 60-75% of total revenue at a traditional bank.

Three things to check when NIM moves:

Asset yields. Did loan yields rise with rates, or is the bank stuck holding low-coupon securities from 2021? Banks that loaded up on long-duration Treasuries at 1.5% yields are the ones that faced stress in 2023.
Deposit costs (the "deposit beta"). When the Fed hikes 100 bps, how much of that does the bank pass through to depositors? A bank with sticky checking accounts (high non-interest-bearing mix) has a low beta and a wider NIM in rising-rate regimes. A bank dependent on brokered CDs has a high beta and gets squeezed.
Mix shift. Are deposits migrating from checking into money-market accounts? That alone can compress NIM 20-40 bps without the Fed doing anything.

A "good" NIM depends on the model. Large universal banks run 2.0-2.5%. Regional commercial banks target 3.0-3.5%. Card-heavy banks can run above 6%. Compare a bank to its own history and to direct peers, not to the sector average.

Provisions and credit quality: the cycle line

Provision for credit losses (PCL) is the charge the bank takes each quarter to top up reserves against expected future losses. Under CECL accounting, banks have to reserve for lifetime expected losses on day one, so provisions are forward-looking and macro-sensitive — they jump when unemployment forecasts rise, even before any actual defaults.

What to actually look at:

Net charge-offs (NCOs): what's actually going bad right now, expressed as an annualized percent of loans. Card NCOs of 3-4% are normal; 5%+ is stress. Commercial NCOs above 50 bps deserve attention.
Non-performing loans and 30/90-day delinquencies: the leading indicators. Delinquencies tick up before charge-offs do.
Allowance coverage ratio: total allowance divided by total loans. If it's falling while delinquencies rise, the bank may be under-reserving.
Office CRE exposure: a source of stress in recent cycles for regionals.

Provisions are the single most volatile line in a bank's P&L. A bank can go from a $500M release (income) to a $2B build (expense) in two quarters. That is why bank earnings are so cyclical.

Efficiency ratio and operating leverage

The efficiency ratio is non-interest expense divided by total revenue. Lower is better. It's the bank version of an operating margin, inverted.

Rough benchmarks:

Best-in-class large banks: mid-50s.
Solid regionals: high 50s to low 60s.
Struggling or sub-scale banks: 65%+.
Trust banks and asset-light models: can run lower 60s but with very different revenue mix.

What you really want is positive operating leverage — revenue growing faster than expenses. When management talks about "operating leverage," they mean efficiency ratio is improving. When efficiency is deteriorating despite revenue growth, costs are running away (tech spend, comp, regulatory).

Capital, buybacks, and what actually moves the stock

This is the part retail readers most often miss. On print day, the stock rarely moves on reported EPS. It moves on:

NIM guidance for next quarter — usually given as a range or directional comment on the call.
Net interest income (NII) dollar guidance for the full year — banks restated this constantly in recent years.
Provision outlook — any hint of credit normalization or stress.
CET1 ratio and buyback capacity. CET1 is the core capital ratio. A bank running CET1 of 13% with a 10% regulatory minimum has roughly 300 bps of excess capital it can return. When that buffer grows, buybacks accelerate, and the stock re-rates.

The sequence matters: NII guide first, credit second, capital return third. Fee income (investment banking, wealth, cards) is the swing factor that can surprise either way. Share buybacks deserve scrutiny on funding source and price discipline — a bank returning capital at 0.8x book is not the same as one buying at 1.2x book.

What to watch next

Pull up the next bank earnings release on your watchlist and find these four numbers in this order: NIM, NCOs, efficiency ratio, CET1. Skip everything else on the first pass.
Compare NIM and efficiency to the same bank one and four quarters ago — not to peers. Trend beats level.
Listen specifically for forward NII guidance on the call. That's where the stock reaction comes from.
Track the spread between delinquencies and charge-offs across the sector each quarter — it's the cleanest early-warning signal for the credit cycle.