A bank ratio is rarely self-explanatory. A 1.00% return on assets can be strong, ordinary, or weak depending on the bank you put beside it. The same is true for efficiency, credit quality, capital, deposit mix, funding cost, growth, and valuation.
That is why peer groups matter. They are not a charting convenience; they are the frame that decides whether a metric becomes evidence or noise. A loose peer set can make a bank look better than it is, punish it for having a different business model, or hide the exact risk you meant to find.
The practical goal is simple: build a group that is broad enough to create context, narrow enough to stay economically comparable, and stable enough to reuse. If you use Banking Intelligence, that logic maps naturally into saved custom peer groups, but the method matters more than the tool.
Key Takeaways
- Start with the decision the peer group must support, not with a list of familiar bank names.
- Use asset size as the first filter, then test geography, loan mix, funding mix, and regulatory thresholds.
- Use wider peer sets for screening and tighter peer sets for valuation or board-level benchmarking.
- Do not mix banks that share a balance-sheet size but earn money in fundamentally different ways.
- Refresh peer groups on a schedule, not whenever the conclusion becomes inconvenient.
Build the Peer Group Backward From the Question
The first mistake is asking, which banks feel comparable? That invites name recognition and confirmation bias. The better question is: what decision does this comparison need to support?
A peer group for investor valuation should be tighter than a peer group for industry screening. A peer group for credit surveillance may need to emphasize loan mix and asset-quality history. A competitive-intelligence group may need branch footprint, deposit markets, and local economic exposure. A fintech prospecting group may care more about size, charter type, systems complexity, and product gaps than about valuation multiples.
Write the use case in one sentence before selecting any bank. If the sentence is vague, the peer group will be vague too.
A Practical Peer-Construction Method
- Define the target bank. Note assets, geography, charter, ownership, branch footprint, loan mix, deposit mix, funding model, and any unusual strategic feature.
- Set the use case. Screening, valuation, credit monitoring, competitive analysis, M&A, or sales segmentation all require different levels of precision.
- Filter by asset tier first. Size is not sufficient, but it keeps operating scale and regulatory context from drifting too far.
- Add market context. Geography matters when local credit cycles, deposit competition, real estate exposure, or branch economics matter.
- Match the earning model. Compare CRE lenders with CRE lenders, C&I banks with C&I banks, and retail-heavy institutions with other retail-heavy institutions where possible.
- Check funding quality. A core-deposit-funded bank and a wholesale-funded bank may have similar assets but very different margin risk.
- Remove obvious mismatches. Specialty lenders, trust-heavy banks, monoline models, and acquisition-distorted institutions may need separate treatment.
- Save the logic. A peer group should be reusable quarter after quarter, with changes documented.
Methodology Block
| Decision Point | Practical Rule |
|---|---|
| Target bank profile | Define assets, state/region, loan mix, deposit mix, funding mix, branch model, and any unusual concentration. |
| Core filters | Start with asset tier, then add geography, business model, funding profile, and credit-risk characteristics. |
| Peer-count range | Use 30-50+ banks for broad screening, 12-20 for operating benchmarks, and 8-15 for valuation or close comparable analysis. |
| Review metrics | ROA, ROE, NIM, efficiency ratio, equity-to-assets, loan growth, deposit growth, loan-to-deposit ratio, nonperformers, charge-offs, and funding-cost trend. |
| Refresh cadence | Review quarterly after Call Report updates; rebuild only when the bank changes strategy, crosses a key threshold, completes a deal, or exits the original peer logic. |
What Counts as a Real Peer?
Asset Size Is the First Cut, Not the Answer
Asset size is the cleanest starting point because scale affects staffing, systems cost, product breadth, regulatory burden, and management depth. A $500 million bank should not usually be benchmarked directly against a $50 billion regional bank.
But size alone can be deceptive. Two $5 billion banks may sit in the same asset band while one is a branch-heavy retail deposit franchise and the other is a commercial lender funded by larger, rate-sensitive accounts. Their ROA, efficiency ratio, deposit beta, and credit cycle will not mean the same thing.
Geography Matters When the Market Drives the Metric
Geography is most useful when the metric is tied to local conditions. CRE exposure, deposit pricing, branch density, labor cost, mortgage demand, and local economic concentration all vary by market. A national peer group can be useful for scale, but a state or regional group may explain performance better.
The test is whether local market conditions plausibly affect the metric you are interpreting. If yes, use geography. If no, do not force it.
Business Model Is Where Peer Groups Usually Break
The most common bad peer set is made of banks that look similar in total assets and completely different in economics. A securities-heavy balance sheet, a specialty finance portfolio, a trust or wealth engine, a heavy CRE mix, or an unusually digital deposit base can all change what normal performance looks like.
This is also where original analysis matters. Do not just filter by labels. Look at where earnings come from, how loans are distributed, how deposits are gathered, and what risk the bank is paid to carry. If those mechanics differ too much, the peer comparison will explain the peer set more than the bank.
Funding Mix Changes the Meaning of Profitability
A high ROA produced with stable low-cost deposits is not the same as a high ROA produced with brokered, wholesale, or rate-sensitive funding. The first may signal franchise strength. The second may signal spread leverage that needs a stress case.
For banks in the same size band, funding mix is often the difference between a fair comparison and a misleading one. At minimum, review non-core funding, uninsured deposit exposure, loan-to-deposit ratio, and deposit cost trend before accepting the group.
Regulatory Thresholds Are Comparability Boundaries
Peer groups should not casually straddle thresholds where the operating model changes. The $10 billion level matters because the CFPB supervises banks, thrifts, and credit unions above that asset level, along with their affiliates.[3] That can affect compliance cost and consumer-finance oversight.
The $100 billion level is a different kind of breakpoint. Federal Reserve supervisory stress-test rules and capital-planning requirements apply to U.S. BHCs, covered SLHCs, and IHCs with $100 billion or more in assets.[4] That does not mean every smaller bank is free of stress-testing or capital-planning expectations, but it does mean a $95 billion bank and a $110 billion bank may be operating under different public supervisory frameworks.
Use UBPR as a Baseline, Not the Finish Line
The FFIEC Uniform Bank Performance Report is still a useful reference point, but the commercial-bank peer-group framework should not be described from memory. The FFIEC announced changes to UBPR commercial-bank peer group population definitions in February 2026, with implementation on or shortly after February 26, 2026.[1]
The key update: for commercial banks with assets up to $300 million, the FFIEC removed office count and metropolitan/non-metropolitan headquarters location from the peer-group criteria. The revised approach uses asset-size groupings for those smaller-bank groups, aligning them more closely with the asset-size logic already used for institutions above $300 million.[2]
That change is a useful lesson for private analysts. Official peer groups are valuable because they are consistent, maintained, and transparent. Custom peer groups are valuable because they can answer a sharper question. Use UBPR to understand the official baseline; use custom peers to test the specific business model, geography, and risk profile you actually care about.
How Many Banks Should Be in the Group?
The right count depends on the job. The contradiction to avoid is using one number for every purpose.
| Use Case | Better Range | Why It Works |
|---|---|---|
| Industry screening | 30-50+ banks | Wide enough to surface outliers and avoid building the answer into the sample. |
| Operating benchmark | 12-20 banks | Enough observations for medians and quartiles, while still allowing business-model checks. |
| Valuation or close comparables | 8-15 banks | Tight enough for narrative discipline, but not so small that one odd quarter drives the result. |
| Local competition review | 5-10 banks | Useful for branch-market or deposit-market questions, but should be labeled as local context, not a full benchmark universe. |
When no clean set exists, say so. A thin peer group is not automatically wrong, but it should be labeled as judgment-heavy. In those cases, show the narrow group beside a wider screening universe so readers can separate local comparability from population context.
Worked Example: A $4B Texas CRE Lender
Suppose the target is a $4 billion Texas-based commercial bank with a CRE-heavy loan book and mostly core deposit funding. A disciplined peer build might look like this:
- Asset tier: Start around $3 billion to $6 billion, staying inside the broader $3 billion to $10 billion UBPR size band while keeping operating scale tighter.
- Geography: Use Texas plus nearby states such as Oklahoma, Louisiana, New Mexico, and Arkansas if the goal is regional CRE-cycle context.
- Loan mix: Keep banks with CRE concentration near the target; exclude pure C&I shops, consumer-heavy lenders, and banks dominated by specialty portfolios.
- Funding mix: Exclude banks whose funding model depends heavily on brokered, listing-service, or high-cost wholesale funding if the target is core funded.
- Final group: Aim for roughly 8-15 close peers for valuation, or 12-20 if the main purpose is operating benchmarking.
The insight is not that Texas banks only belong with Texas banks. It is that the peer group should preserve the economic drivers that make the target bank what it is: scale, CRE cycle, deposit base, and funding behavior. A same-state, same-size filter is a starting screen, not an analytical conclusion.
Common Mistakes That Create False Signals
Starting With Famous Names
Recognizable banks are not always comparable banks. Familiarity often smuggles a narrative into the analysis before the data has a chance to disagree.
Letting the Event Define the Peer Set
Post-event peer groups can help explain what happened, but they are dangerous for measuring whether the target was an outlier beforehand. Build wide first to see the distribution, then build narrow to understand the mechanism.
Mixing Business Models
If the group contains a retail deposit franchise, a CRE specialist, a trust-heavy bank, and a specialty lender, the median may be mathematically correct and analytically useless.
Changing Peers When the Answer Is Awkward
A saved peer group disciplines the analysis. If a bank underperforms, the first move should be to understand why, not to redefine the comparison until the underperformance disappears.
Ignoring Survivorship and Acquisition Noise
Mergers, restructuring charges, balance-sheet repositioning, and major strategy shifts can distort peer metrics. Keep those banks if they are truly comparable, but flag the distortion instead of treating the median as clean.
Turn the Peer Group Into a Quarterly Workflow
A peer group has the most value when it becomes part of a recurring review. The workflow does not need to be complicated. It does need to be consistent.
- Keep a saved peer definition and a short explanation of why each filter exists.
- Run the same metrics each quarter before adding new one-off questions.
- Compare the target to median, upper quartile, lower quartile, and named outliers.
- Separate structural differences from temporary quarter-specific noise.
- Document additions and removals so the peer group does not drift silently.
Banking Intelligence supports that operating rhythm through institution search, saved custom peer groups, the compare workflow, and export tools for downstream modeling, memos, and screening.
FAQ
Should I use UBPR peer groups or custom peer groups?
Use both for different jobs. UBPR gives you a consistent official baseline. Custom peer groups let you sharpen the comparison around the target bank’s actual business model, geography, funding mix, and risk profile.
How often should a bank peer group be refreshed?
Review it quarterly after new regulatory data is available. Do not rebuild it every quarter. Change it when the target bank or a peer materially changes through an acquisition, strategic shift, threshold crossing, or business-model break.
What if no clean peers exist?
Say that directly. Use a two-layer approach: a wider universe for distribution context and a narrower judgment-based set for qualitative comparison. The lack of perfect peers is itself useful information about the bank.
When is a small peer group acceptable?
A small group can be acceptable for local competition or a highly specialized business model, but it should not carry more precision than it deserves. Use medians carefully, name the limitations, and show a broader reference set when possible.
Product data note: Banking Intelligence local U.S. Call Report-backed context in the app currently runs through 2025-12-31.
Sources
- FFIEC announcement – February 13, 2026 announcement of changes to UBPR commercial-bank peer group definitions.
- FFIEC UBPR Peer Group Memorandum – Details on the February 26, 2026 UBPR peer-group changes and revised asset-size logic.
- CFPB supervised institutions – CFPB supervisory authority over banks, thrifts, and credit unions with assets over $10 billion.
- Federal Reserve 2026 stress test scenarios – Current stress-test framework and $100 billion applicability language for covered firms.