A stronger margin, but less profit
In the third quarter of 2024, U.S. banks earned more from the difference between interest coming in and interest going out. The FDIC reported that the industry’s net interest margin rose seven to 3.23%, and increased $4.5 billion from the previous quarter. Yet total net income fell $6.2 billion, largely because earlier one-time gains did not repeat. One part of the banking business had improved while the bottom line deteriorated. That historical episode captures both the usefulness and the limits of net interest margin. [1]
What the percentage actually measures
, or NII, is interest earned on loans, securities and other earning assets minus interest expense on deposits and other funding. Net interest margin, or NIM, expresses that income relative to average earning assets. The denominator is the assets producing the interest stream, rather than total revenue, deposits or shareholders’ equity. Using an average connects income earned throughout a period with the balance supporting it during that period. A quarter-end balance can give a distorted answer when assets grow or shrink sharply near the reporting date. [2]
Here is a deliberately simplified hypothetical bank, with every amount invented for teaching. Its average earning assets are $100 million. During one quarter it earns $1.50 million of interest and pays $0.60 million, leaving $0.90 million of NII. Dividing $0.90 million by $100 million gives 0.90% for that quarter. Multiplying by four gives a 3.60% annualized NIM. Annualization puts the quarter on an annual-rate scale; it does not predict that the bank will earn the same amount for four consecutive quarters.
The illustration uses four equal quarters. If a calculation instead annualizes by days, a hypothetical 90-day period in a 365-day year uses 365 divided by 90. The same income and assets then produce 3.65%. A year-to-date numerator also needs a year-to-date average and the matching annualization factor; multiplying nine months of income by four would plainly overstate the rate. The period and the averaging convention belong with the percentage. [3]
Why the interest-rate spread is different
The familiar spread is the yield on earning assets minus the rate paid on interest-bearing liabilities. Those rates use different denominators. NIM instead divides the net dollar income by earning assets. Funding from noninterest-bearing deposits and equity helps explain why the two percentages differ. The Federal Reserve’s decomposition explicitly weights funding rates by the size of the relevant liabilities relative to earning assets. [3]
In our hypothetical bank, the $100 million of earning assets yields 6% a year. Assume $80 million of interest-bearing funding costs 3%, with another $10 million in noninterest-bearing deposits and $10 million in equity. The spread is 6% minus 3%, or 3 percentage points. Annual interest income is $6 million and expense is $2.4 million. NIM is therefore $3.6 million divided by $100 million, or 3.6%. The extra 0.6 percentage point is an arithmetic consequence of the funding mix, not a missing interest charge. This stripped-down balance sheet excludes premises and other non-earning assets.
Depositors can change the result without a loan changing
The same hypothetical bank now sees $10 million move from its noninterest-bearing accounts into accounts paying 3%. Total deposits and earning assets are unchanged, but annual interest expense rises $300,000. falls to $3.3 million and NIM to 3.3%. No customer has to leave the bank for its funding to become more expensive. The change is in where customers keep their money. This isolates account migration while holding every advertised interest rate constant.
Pricing changes are a separate mechanism. describes the change in a defined deposit rate divided by the change in a benchmark rate over a specified interval. A 0.40-percentage-point deposit-rate increase against a 1-percentage-point benchmark increase produces a hypothetical 40% beta. The measure needs its starting date, ending date and deposit population: interest-bearing deposits alone can tell a different story from all deposits. Federal Reserve economists’ April 2024 study found differences across banks and funding types, including slower deposit repricing than some other borrowing. A historical beta is a description of that experience, not a contractual promise about the next rate move. [4]
Two clocks are running after every rate move
Loans and funding do not reset together. A floating-rate loan may respond quickly to its benchmark, while a fixed-rate loan continues at the agreed coupon until repayment, sale or maturity. New lending gradually replaces old lending at current prices. On the other side, deposit competition and the renewal of term funding affect when costs change. Federal Reserve economists’ 2019 comparison of four tightening episodes found that margins did not respond identically in each cycle. Higher policy rates alone do not determine the outcome. [5]
A hypothetical $20 million floating-rate loan pool repricing upward by 1 percentage point adds $200,000 of annual interest income if balances and collection remain unchanged. If the bank’s $80 million interest-bearing funding reprices upward by 0.4 percentage point, expense rises $320,000. falls $120,000 even though loan yields rose. Reverse the relative speeds and the result can reverse. For falling rates, the same timing problem remains: assets can reset before funding becomes cheaper. There is no automatic one-for-one reversal of the previous cycle.
Securities and hedges alter the path
A fixed-rate bond purchased years earlier keeps its contractual coupon when market rates rise. Its market value can fall at the same time. Holding it, selling it and replacing it create different combinations of current income, realized results and future yield. The 2010 interagency interest-rate-risk advisory therefore treats earnings and economic value as distinct perspectives. A healthy current margin does not establish that a securities portfolio could be sold without a loss. [6]
Hedges add another layer. In a hypothetical receive-floating, pay-fixed interest-rate swap, an increase in the floating rate increases that leg’s receipts, potentially offsetting exposure elsewhere. The swap can also cost money when rates move the other way. Its accounting treatment and relationship to the hedged position matter to where effects appear in reported results. Wells Fargo’s 2025 average-balance table expressly says its asset and liability rates include hedging effects. A loan coupon alone therefore cannot reconstruct the whole reported margin. [7]
The missing expenses still matter
Credit-loss provisions are not subtracted in the basic NIM numerator. They are separate from interest expense, as Wells Fargo’s 2025 financial review illustrates. Staff costs, technology, other operating expenses and taxes also sit outside NIM. A higher margin can coexist with a weaker final profit. [7]
Suppose our original hypothetical bank still earns $900,000 of quarterly , but its quarterly credit-loss provision rises from $100,000 to $400,000. NIM remains 3.60% under the same annualization and asset assumptions. Pre-tax profit falls $300,000, all else equal. A loan book paying higher interest can bring greater credit exposure; NIM by itself does not price that tradeoff. If a troubled loan also stops contributing interest income, NIM can fall as well.
What makes two published margins comparable
Definitions can change the numerator. Wells Fargo reported 2025 GAAP of $47.484 billion, while its taxable-equivalent table showed $47.787 billion and a 2.64% margin. The $303 million adjustment translated tax-exempt income onto a taxable-equivalent basis. It was not extra cash received. A comparison mixing adjusted and unadjusted numbers would blur that distinction. [7]
The reporting entity matters too. A bank subsidiary’s accounts and a consolidated holding company’s accounts describe different boundaries. The Federal Reserve’s FR Y-9C covers the consolidated holding-company organization, potentially including parent, bank and nonbank businesses. Combining one entity’s interest income with another’s assets creates a ratio with no coherent meaning. The name on the filing is part of the measurement, not an administrative detail. [8]
Finally, current income cannot summarize all the future exposures. The interagency advisory identifies deposit behavior, loan prepayments and other assumptions as important to rate-risk analysis. Read alongside its funding mix, credit results and reporting definitions, NIM explains how a bank’s interest-earning business worked during a period. It is a useful chapter in the earnings story, rather than a complete verdict on the institution. [6]
Sources
- FDIC: third-quarter 2024 banking results; December 12, 2024Official sourceBack to text: ↑
- Federal Reserve: Banking System Conditions; November 2024Official sourceBack to text: ↑
- Federal Reserve economists: Why Are Net Interest Margins of Large Banks So Compressed?; October 5, 2015Official releaseBack to text: ↑1↑2
- Federal Reserve economists: Is This Time Different?; April 12, 2024Official sourceBack to text: ↑
- Federal Reserve economists: monetary policy and bank margins across four tightening episodes; April 19, 2019Official sourceBack to text: ↑
- Federal Reserve: interagency interest-rate-risk advisory; January 6, 2010Official sourceBack to text: ↑1↑2
- Wells Fargo & Company: 2025 annual financial review, Tables 3–4Filing / reportBack to text: ↑1↑2↑3
- Federal Reserve: FR Y-9C reporting form and consolidated-company scope; checked October 6, 2026Official sourceBack to text: ↑