The Cohort Earnings Divergence — Q1 2026

The Split

The pattern across the asset-size ladder is close to monotonic: the larger the credit union, the higher the return on assets, with the smallest tier sitting in a category of its own.

The under-$100M tier returned 0.13% on assets. The next tier up, $100M–$500M, returned 0.75% — more than five times as much — and the curve keeps climbing to 1.04% for the 24 institutions above $10 billion. A 91 basis point gap separates the smallest cohort from the largest. That is not a rounding difference in earnings; it is the difference between a credit union that can fund its own growth and one that cannot.

It Isn't Margin, and It Isn't Credit

The intuitive explanations both fail on contact with the data.

The smallest credit unions are not being squeezed on margin. Their annualized NIM is 3.73% — second only to the largest tier's 3.79%, and well above the $1B–$10B cohort's 3.26%. Small institutions tend to hold simple, relationship-priced loan books and low-cost core deposits, and in a quarter where funding costs eased, that mix produced one of the widest spreads in the system.

They are not being buried by credit losses either. The under-$100M tier charged off 0.44% of loans on an annualized basis — the lowest of any cohort, and barely a third of the 1.37% the over-$10B tier absorbed. Their 60+ day delinquency, at 0.93%, sits below the largest tier's 1.12%. Conservative underwriting and member familiarity show up exactly where you would expect them: in the loss columns.

So a cohort with a top-quartile margin and the cleanest credit book in the system earns almost nothing. The money is being made on the asset side and lost somewhere else.

It's Scale

Walk the income statement down from the margin and only one large line remains: operating expense. A 3.73% margin and a 0.44% loss rate leave ample room for a healthy return — unless fixed operating costs consume nearly all of it. For the smallest tier, they do.

The arithmetic is unforgiving for a sub-scale institution. Core systems, examinations, audit, compliance, insurance, and a minimum viable staff cost roughly the same whether a credit union holds $40 million in assets or $400 million. Spread across a small balance sheet, those fixed costs swallow the margin; spread across a large one, they leave most of it behind as profit. That is why the ROAA curve rises so cleanly with size — it is a fixed-cost-absorption curve more than a credit or pricing curve.

The harder part is the trend. The under-$100M tier's loans contracted 1.1% in the quarter, extending a run of negative loan growth that the larger tiers have not shared — the over-$10B cohort grew loans 4.5% in the same quarter. Loan growth is how a small institution earns its way to scale: more earning assets to spread the same fixed costs over. A shrinking loan book does the reverse. It raises the per-dollar cost of running the institution at exactly the moment those costs need to be diluted. The margin is healthy, but there is less and less of a balance sheet to apply it to.

The Release Valve Is Consolidation

This is not a system-stability problem. The under-$100M tier holds a small share of total assets; if every one of these credit unions earned zero, the system ROAA would barely move. The aggregate is genuinely healthy, and nothing here threatens that.

It is a structural-attrition problem, and the data already shows the resolution underway. The system held 4,250 federally insured credit unions at quarter-end, down 161 from a year earlier — a roughly 3.6% annual decline, almost all of it concentrated in the smallest tiers. When a sub-scale credit union cannot earn its way to scale and cannot grow into it, the remaining path is to merge into an institution that already has it. The earnings divergence is the pressure; the shrinking charter count is the release.

What that means for boards of small institutions is concrete, not abstract. A 0.13% return with a clean book and a strong margin is not a credit-quality failure or a pricing failure to be fixed with a better loan product. It is a scale problem, and scale is the one thing organic effort rarely solves at this size. The decision in front of those boards is whether to find scale through partnership on their own terms while the margin is still strong, or to wait until a weaker quarter makes the same decision for them.

Connecting the Lines

This breakdown develops Signal 3 (Record Margins, and a Cohort Earning Nothing) from the Q1 2026 System Signals, and it reframes the loan-growth weakness noted in the Q1 2026 Ledger. The soft 0.5% system loan growth is not evenly soft — it is the large tiers still growing while the smallest tier shrinks, and that divergence is the same force visible in the earnings split.

The seasonal credit relief of Q1 (Signal 1) buys the smallest tier some room. It does not change the math: a margin this good and losses this low should produce far more than 0.13%, and the gap is scale that organic growth is no longer supplying.