The Interchange You Can't See

Where Interchange Lives — and Why You Can't Measure It

The 5300 Call Report carries no interchange line. Per NCUA's own account instructions, card-swipe income is reported inside Account IS0020 — "Other Operating Income (Includes unconsolidated CUSO Income)" — the residual non-interest bucket for "income other than" the named categories. The instruction explicitly lumps interchange together with NCUSIF dividends, interest on certain purchased participations, and unconsolidated CUSO income. There is no sub-line, no memorandum item, no way to pull interchange back out.

That makes IS0020 an upper bound on interchange, not a measurement of it. In 2025 the system reported 14.8 billion dollars of Other Operating Income — but an unknown share of that is CUSO profit and insurance-fund dividends that have nothing to do with cards. A separate line, Account 131 — Fee Income (10.0 billion in 2025), captures fees charged to members (overdraft, ATM, account fees). It is a sibling revenue stream, not interchange: interchange is paid by merchants and card networks, member fees are paid by members. Together the two lines are the system's 24.8 billion dollars of non-interest income, with interchange embedded somewhere inside the larger half and not separable from it.

The Trend: Non-Interest Income Is Compressing

Measured against assets — the only way to compare across a consolidating system — non-interest income has been in steady decline for four years. The interchange-bearing bucket carries the entire move.

The decline is almost entirely in the interchange-bearing bucket: Other Operating Income fell from 79.0 to 60.7 basis points of assets, while member fee income held near 43 bps. The sharpest move was a single step — IS0020 dropped 15 bps from 2021 to 2022 — that warrants its own footnote: 2021 was the first year under the new account definition and an unusually rich one for CUSO income, so part of that fall is a definitional and base-effect artifact rather than a collapse in card revenue. But the post-2022 drift is real and steady: in dollar terms IS0020 has grown, just slower than the 18% asset growth underneath it. Credit unions are earning more non-interest income and a thinner slice of it per dollar of balance sheet.

Two Revenue Models, Split by Size

The system-wide average conceals two different businesses. Sorting the 2025 data by asset size shows that small and large credit unions earn their non-interest income in opposite ways.

Two patterns run in opposite directions. Fee income falls as size rises — from 53.5 bps at the smallest credit unions to 29.0 bps at the largest — because small institutions lean hardest on overdraft and account fees, the revenue most exposed to current regulatory scrutiny. The interchange-bearing bucket does the reverse, and the share of non-interest income coming from IS0020 climbs monotonically with size, from 39% to 66%. The largest credit unions run interchange-and-CUSO machines; the smallest run on member fees.

The mid-tiers are the income sweet spot in absolute terms — the $500M–$1B cohort earns the most non-interest income per dollar of assets (116.9 bps), drawing roughly evenly from both sources. This is the same mid-tier advantage visible in the cohort earnings divergence: big enough to run a card program, small enough to still collect member fees.

Why Debit, Not Credit, Is the Engine

If interchange can't be measured directly, can it be inferred — backed out by correlating IS0020 with card activity? We tested it, and the answer is instructive: no, but the attempt reveals which card business actually matters.

The naive correlation looks strong — IS0020 tracks credit-card balances at 0.93 across credit unions. But that is an illusion of scale: IS0020 correlates with total assets at 0.98, because large institutions have more of every line item. Strip size out — compare income-per-asset to card-balances-per-asset — and the relationship collapses to roughly 0.09. Credit-card balances explain less than 1% of the cross-institution variation in Other Operating Income intensity.

Adding a debit proxy is what exposes the real driver. In a size-controlled model, the number of share-draft (checking) accounts — a stand-in for debit transaction volume — carries roughly seven times the explanatory weight of credit-card balances, with a near-unit elasticity. In plain terms: a credit union's checking franchise predicts its Other Operating Income far better than its credit-card book does. That is exactly what you would expect if debit interchange, driven by everyday swipe volume, dominates credit interchange at credit unions — and it is why a credit-card-balance proxy was always going to fail. The Call Report records card balances, never card spend, and interchange is a tax on spend.

What the Preemption Rule Changes

The June interim final rule preempts state interchange statutes — such as Illinois's law excluding tax and gratuity from the interchange base — for federal credit unions, paralleling the OCC's parallel move for national banks. The relief is real, but the data says it lands unevenly. It is worth most to the institutions whose non-interest income is most interchange-weighted — the over-1-billion-dollar cohorts, where two-thirds of non-interest income flows through the IS0020 bucket. For the 2,514 credit unions under 100 million dollars, which draw 61% of their thinner non-interest income from member fees rather than interchange, the bigger regulatory variable is the opposite end of the fee debate: the ongoing scrutiny of overdraft and account fees that make up the bulk of their Account 131 revenue.

That is the throughline. Interchange deregulation strengthens the revenue model of the large, card-heavy credit unions; fee scrutiny pressures the model of the small, member-fee-dependent ones. The same policy season pushes the two cohorts in opposite directions — widening, at the margin, the structural divide the size-cohort analysis has been tracking.

FINASENSE Assessment

The honest headline is a caveat: the Call Report cannot tell you how much interchange credit unions earn, and any figure that claims to is reverse-engineered from data that does not contain the answer. What it can tell you is structural, and three findings hold up. First, non-interest income is compressing — from 1.22% to 1.02% of assets in four years — concentrated in the interchange-bearing bucket. Second, the system runs two revenue models that diverge cleanly by size: member fees for the small, interchange-and-CUSO income for the large. Third, to the extent interchange drives that larger bucket, debit is the engine, not credit — which means the franchise to watch is the checking account, not the credit card.

For an analytics product, the discipline is the value: we report Other Operating Income and Fee Income as what they are, flag interchange as embedded-but-unmeasured, and resist the temptation to publish a precise interchange number the regulator's own data cannot support. The same restraint applies to reciprocal deposits — another concept the policy debate treats as measurable that the Call Report does not yet isolate.