A Regional Bank Is a Different Risk Machine
This series examines regional banks as a distinct part of the U.S. banking system, not as a smaller version of the largest banks. The central question is how a bank’s business model changes the way economic shocks reach earnings, loan losses, and capital. That question supports a practical judgment: when does a regional bank look resilient, and when does the same structure expose it to a concentrated or local problem?
The distinction matters because “regional bank” sounds like a size label, while the evidence points to an economic relationship. A bank takes in funding, makes loans, holds other assets, and earns income from those choices. Interest income from assets and interest expense on funding shape the margin; fees and trading can diversify revenue; credit losses reduce earnings; and capital absorbs the damage. If two groups allocate their balance sheets differently, the same recession, interest-rate move, property-market decline, or funding disturbance will not pass through them in the same way. A stress test that treats their responses as interchangeable can therefore produce a tidy answer to the wrong question.
The size label hides the economic subject
The Federal Reserve study at the center of this discussion defines regional banks as firms with at least $10 billion but less than $100 billion in consolidated assets throughout the sample period. It defines large banks as firms that exceeded $100 billion in assets during at least one quarter, using a “once large bank, always large bank” rule. Those definitions set the boundaries of the comparison, but they do not explain why the comparison is useful. The reason is that the two groups perform different versions of commercial banking.
The thresholds are also not a universal statement that every institution between those numbers behaves alike. They are the study’s way of forming a comparison group from a quarterly panel that generally runs from 1997 through 2023. A bank’s position in an asset band tells us something about scale, supervisory attention, and possible funding access, but it does not tell us whether its loans are fixed- or floating-rate, whether its depositors are stable, or whether its income depends on a particular property market. Those questions must be answered from the bank’s actual activities.
In the study’s 2019 fourth-quarter comparison, regional banks held deposits equal to 75.7 percent of assets and loans equal to 69.3 percent. The corresponding shares for large banks were 60.1 percent and 50.3 percent. Regional banks also had 41.6 percent of their loan books in commercial real estate, compared with 16.8 percent for large banks. Large banks, by contrast, had substantially more credit-card lending and more activities associated with capital markets, trading, investment banking, and transactional services. The appendix’s 2023 fourth-quarter figures show the same broad pattern after the pandemic and the 2023 banking stress: deposits and loans still occupied more of regional-bank assets, while CRE still made up a much larger share of their loan books.
These are not just differences in an accounting table. They describe how each group turns a balance sheet into revenue and risk. Regional banks depend more heavily on interest income from loans and on deposits as a funding base. Large banks draw relatively more on fee and trading income and wholesale or less core funding. In the earlier comparison, regional banks’ interest income was 4.29 percent of assets, versus 3.90 percent for large banks, while interest expense was 0.93 percent versus 1.25 percent. The numbers do not make one model universally safer. They show that the two models have different levers.
That is why “smaller” is an incomplete description. Size can affect supervision, funding access, and the scale of a loss, but it does not identify which assets are exposed, how quickly liabilities reprice, or which income stream absorbs a shock. A regional bank with a loan-heavy, deposit-funded balance sheet is not simply a reduced-size institution with the same operating logic as a globally diversified bank.
The transmission map is the point
A useful way to read a regional bank is to follow a shock through four linked questions. What funds the bank? What does it own or lend against? Where are those exposures located? And which income or capital measure moves first?
The first question matters because funding is not a neutral input. Deposits have different costs and different behavior. A stable, low-cost deposit base can support the spread between what a bank earns on assets and what it pays for funding. Wholesale funding or deposits that reprice quickly can make expenses respond faster when market rates change. The second question matters because an asset’s maturity, interest-rate sensitivity, collateral, and borrower determine how the shock appears: as a smaller margin, a fair-value loss, a delayed refinancing problem, or a realized credit loss. The third question determines whether national averages are informative or whether a local labor market and property cycle matter more. The fourth determines whether the first visible effect appears in pre-provision net revenue, net charge-offs, or capital.
The Federal Reserve’s evidence suggests that regional banks often use more low-cost deposits to fund longer-duration assets, including predominantly fixed-rate local commercial real estate loans. That structure can dampen the short-run pass-through of a change in short-term interest rates to both interest income and interest expense. When yields fall, a bank whose assets and funding costs adjust slowly may preserve a relatively stable margin. The same structure can create larger interest-rate exposure when yields rise, because the market value of longer-duration assets can fall before a loss is realized through a sale or a borrower default.
This is a trade-off, not a contradiction. A bank can report stable current income while holding assets whose economic value is more sensitive to rates. It can have strong capital and liquidity at the system level while a particular balance sheet has a difficult maturity or deposit profile. It can have a high CRE share without immediate delinquencies, while still facing a future refinancing wall. Reading the bank requires keeping these time horizons separate rather than asking whether the institution is simply “safe” or “unsafe.”
The asset side also concentrates the question. A high CRE share means that the bank’s outcome depends more on property types, borrowers, rents, vacancies, collateral values, and refinancing dates than a group with a more diversified loan book. Exposure is not the same as loss: a loan can be large without becoming delinquent, and a market can weaken without every property failing. But exposure determines which evidence deserves attention. A national average for consumer credit may be less informative than the refinancing profile of a regional bank’s office or multifamily portfolio.
Geography adds another layer. The median large bank operated in about 14 states in 2025, while the median regional bank operated in about four; roughly one-fifth of regional banks operated in only one state. That makes an individual institution more exposed to a local labor market, property cycle, or industry base. Yet the regional-bank sector itself was not confined to one local economy. The five largest states accounted for about 34 percent of regional-bank deposits, and the Federal Reserve reports a deposit concentration measure well below the conventional threshold that would signal a highly concentrated sector.
Both statements can be true. An individual regional bank may be locally concentrated while the sector’s deposits are spread widely enough that a problem in one place is unlikely, by itself, to become a system-wide event. Conversely, a broad macroeconomic shock can reach many banks at once even when no single bank dominates the national market. The level of analysis—one institution, the regional-bank group, or the banking system—changes what “concentration” means.
This map also explains the apparent paradox at the heart of regional-bank analysis. The same relationship-based model can be a stabilizer against one shock and a vulnerability under another. Lower sensitivity to aggregate variables does not mean lower exposure to every variable. A group average can look calm while a particular CRE segment, deposit base, or local economy is under pressure.
The first question is whether the lens fits
The Federal Reserve study tests the problem directly with the Forward-Looking Analysis of Risk Events, or FLARE. FLARE is a top-down, bank-level stress-testing model. It uses public financial-statement data and macroeconomic variables to project pre-provision net revenue, loan losses, and capital. Its equations estimate separate income and loss components using past values, market indicators, macroeconomic conditions, and bank-specific controls.
“Top-down” describes the level and direction of the exercise, not a claim that the model is simplistic. It begins with a specified path for broad economic and financial variables, estimates how bank-level outcomes have historically related to those variables, and carries the relationships forward into a scenario. A bottom-up exercise would start from more granular loans or securities and could ask a different question about individual exposures. Neither approach is automatically better; the right choice depends on whether the question concerns correlated system stress or a particular portfolio.
The important choice is not whether to use a model. It is which banks should determine the relationships inside it. In an all-bank model, large institutions represent most of the industry’s assets and can therefore dominate the statistical results. Applying those relationships to regional banks may impose the response pattern of a more diversified, capital-markets-oriented group on banks whose income and assets are organized differently. The study addresses that possibility by estimating one model on all banks and another on regional banks only.
This does not turn the regional model into a complete picture. Top-down analysis uses bank-level data and aggregate economic variables. It is designed to study correlated system stress, not to determine the solvency of one named bank from loan-level or security-level information. The regional-bank model can improve the fit between the modeled population and the question being asked, while still missing a local property cycle, a bank-specific funding run, or a hidden weakness in an individual portfolio.
The current supervisory picture makes that distinction more—not less—important. The OCC’s Spring 2026 risk report describes improved earnings, strong capital and liquidity, and manageable aggregate credit risk in the federal banking system. It also flags CRE refinancing, private-credit concentrations, and other risks that require monitoring. Those observations are useful context, but the report’s comparison of banks under $30 billion is not the same population as the Federal Reserve’s $10 billion-to-under-$100 billion regional-bank group, and system aggregates are not evidence that every regional bank has the same condition.
The question this series will keep returning to is therefore conditional rather than absolute: resilient against which shock, through which balance-sheet channel, at what level of aggregation, and with what evidence? To answer it, the next step is to inspect the balance sheet itself—how deposits, loans, CRE, revenue, duration, and geography fit together. Only then can a stress model’s coefficients or a current risk statistic be interpreted as a statement about a real regional bank rather than as a generic statement about bank size.