The Fallacy of NIM to Drive Bank Performance

Bank Director’s 2026 RankingBanking study (HERE), sponsored by Crowe and compiled by Piper Sandler & Co. using S&P Global Market Intelligence data, ranks the 300 largest publicly traded U.S. banks on 2025 results. This year’s headline finding may look like a focus on net interest margin: the top 25 banks posted a median net interest margin (NIM) of 4.11%, versus 3.57% for the rest of the list, funded in part by cheaper deposits (1.52% average cost of deposits for the top 25 banks versus 1.94% for everyone else).

But the ranking’s methodology never scores NIM directly. Instead, RankingBanking ranks banks on four metrics: core return on average tangible common equity, core return on average assets, the tangible common equity ratio, and nonperforming assets to loans and OREO. NIM appears throughout the commentary as something that may be correlated with performance, but not as one of the criteria that determines it. Strong margins correlate with strong performers, but the scorecard rewards profitability, capital strength, and credit quality, not spread income on its own.

That is also why investors and bankers should prioritize (measure and improve) ROA and ROE rather than NIM. It is those two performance measures that reflect what NIM ignores: credit costs, operating efficiency, capital consumption, and fee income. A bank can carry an enviable NIM and still be a weak investment if that margin is funding an inefficient branch network or an undisciplined credit box. ROA shows how much of every asset dollar a bank actually keeps; ROE shows how well it compounds shareholders’ capital.

Our own bank illustrates the point. Southstate Bank ranked third among the 36 public banks above $50 billion in assets, behind only Columbia Banking System and East West Bancorp. Its 1.48% core ROAA was second-best in that peer group, and its 8.76% tangible common equity ratio placed it in the top quartile for capital. Much of that came from its 2025 integration of Independent Bank Group (cost saves and purchase-accounting accretion) layered on a genuinely low-cost deposit franchise: $56 billion in deposits across 1.4 million accounts averaging $40,000, at a 1.76% cost of deposits that analysts peg 10% to 20% below peers. SouthState won its ranking with disciplined M&A, diversified fee income, and cheap, granular deposits – not the widest loan spreads.

The NIM Mirage

Our own research shows the lack of connection between NIM and bank performance as measured by NIM. We continue to quote Charlie Munger’s line about how incentives explain performance: reward lenders for yield, and yield is what you get, but not profit.

The pattern holds statistically. Our review of five-year performance shows that community bank performance is not related to NIM. NIM and ROA show an R² near zero. Our tighter peer study of 16 community banks in the same state and market found the same thing up close: NIM explained just 12.5% of the variation in ROA, while efficiency ratio (R² of 0.911 for 5Ys and 0.7% for the quarter (below)) and fee income as a share of assets (R² of 0.760) explained most of it. Industry-wide, the long-run relationship between NIM and ROA is not just weak but slightly negative.

The starkest evidence comes from roughly 600 community banks acquired between 2020 and 2024, compared with banks that stayed independent. ROA and ROE were overwhelmingly the strongest predictors of which banks got bought (a one-tail Z-score of -3.69 on ROA, p = 0.0001). NIM predicted nothing: margins were statistically indistinguishable between the two groups (Z = -0.21, p = 0.42). The one place NIM did diverge was 2024, when acquired banks showed sharper compression – not from weaker loan yields, but from added asset duration that got caught by higher rates, exactly the interest-rate risk a margin-only scorecard misses when focused on NIM.

Together, these studies confirm what RankingBanking’s methodology implies: NIM measures the spread a bank earns today, not whether that spread survives credit losses, funding shocks, and the cost of running the franchise. Fee income, long-term customer stickiness, lower-cost and long-term deposit, efficiency, not loan yield, move ROA and ROE.

Two Inputs, One Efficient Market

NIM has only two moving parts: yield on assets and cost of funding those assets. If margin is the goal, widening the gap between the two is the job, but the two sides are not equally winnable.

Consider a test that stumps most bankers. You are offered two identical commercial loans. The loans are the same size, maturity, market, and cross-sell potential, and both are originated by different banks to different borrowers at the same time in similar markets. Loan A prices at SOFR plus 1.50%; loan B at SOFR plus 3.50% with the only difference between the two loans is credit quality. Which do you take? Most bankers reflexively pick B for the extra 200 basis points, reasoning that without more information, the higher-paying loan is the better economic bet. The better answer is A.

The reasoning is behavioral and reflects fundamentals of finance and game theory. In a perfectly efficient credit market, risk and return would move together in a straight line, and no bank could consistently out-earn another on a risk-adjusted basis. In practice, mispricing clusters unevenly: low-risk credits price close to fair value, but higher-risk credits get mispriced far more often because credit losses are a distant, uncertain cost while yield books immediately, because production is what gets incentivized while shareholders absorb the eventual losses, and because competitors will always underwrite the risky deal a disciplined bank walks away from. The net effect is that bankers systematically underprice the riskiest loans, which is exactly why loan yield do not behave like an efficient market for a given risk level. However, loan yields behave less efficiently as the risk increases.

Deposit costs behave differently. RankingBanking shows a 42-basis-point gap in average cost of deposits between the top 25 banks and the rest of the list (1.52% versus 1.94%), and that gap reflects franchise quality, not the level of market rates. As an example in the article, Bank First Corp. funds roughly 30% of its balance sheet with noninterest-bearing demand deposits by treating the primary checking relationship as, in its president’s words, the “linchpin” of every customer relationship. That kind of deposit base is built, not bought off a rate sheet, which is exactly why cost of funding, not asset yield, is where a community bank can build a durable NIM advantage.

What Actually Widens the Gap

If asset yield is close to a zero-sum game and deposit cost is the franchise value driver, priorities follow directly. Long-term, stable, low-cost deposit relationships matter more than a wide spread on any single loan. City Holding Co.’s CEO treats deposits as the bank’s “primary business, not something we do to fund our loans.”  Cullen/Frost Bankers goes further: bankers get no credit for a new relationship unless it includes the checking account. This is a discipline it credits for some of the lowest deposit costs in its market. Commerce Bancshares, this year’s top-ranked bank overall, built its 160-year franchise the same way: a low-cost deposit base, roughly 40% fee income, and a credit culture that will “forgive you for almost anything but bad debts.”

The same logic runs through the loan book. Larger, longer relationships amortize acquisition and servicing costs better than small, short ones; expanding relationships beat static ones; and conservative underwriting avoids the mispricing described above. Often the safest, most relationship-rich borrowers command the tightest pricing, not the widest (as loan A did in the earlier example). Collectively, sticky deposits, durable relationships, right-sized credits, and disciplined underwriting are what separate banks that compound ROA and ROE over a cycle from banks that show a healthy NIM right up until the cycle turns.

Positioning for What Comes Next

The current rate backdrop makes this discipline more urgent. The Fed held its target range at 3.50%–3.75% at its July 2026 meeting, over the dissent of three regional presidents who wanted a hike, after annual inflation reached 4.2% in May – the highest reading in more than three years, driven in part by an oil-price spike tied to the U.S.-Iran conflict. New Fed Chair Kevin Warsh has said the committee has “no tolerance for persistently elevated inflation,” and markets that had priced cuts are now pricing multiple hikes in 2027.

Banks that ran into trouble last cycle were disproportionately the ones that extended asset duration to lock in yield and increase NIM, then watched margins compress when rates moved against them. Bankers should resist doing that again. Favor floating-rate and shorter-duration loans and securities, ladder maturities instead of reaching for a few extra basis points of fixed yield and keeping enough flexibility to reprice assets upward if the next move is a hike.

Pair that posture with continued investment in core, low-cost, long-tenured deposits, so that whichever way rates move, funding costs lag the market rather than lead it. The goal is not winning next quarter’s NIM print but compounding ROA and ROE through the cycle, which the evidence says is the only scoreboard that predicts which banks are still standing five years from now.