Many accounts can still be one bet
A bank can lend to hundreds of unrelated legal entities without achieving meaningful diversification. A hotel, restaurant, shuttle operator and apartment owner may all rely on the same tourism market. A set of households may all work for one major employer. Different account numbers and industry labels can conceal a common source of repayment.
Concentration analysis asks which exposures may perform similarly under the same shock. It goes beyond identifying the largest borrower. It includes sectors, geography, collateral, sponsors, guarantors, distribution channels and indirect exposures. The goal is to identify the economic dependence before it becomes visible as simultaneous .
The OCC's current Concentrations of Credit booklet is version 3.0, dated September 2026. It explicitly discusses correlations across and within pools, including pools that do not independently meet a concentration definition. This article uses that current booklet rather than treating an older edition's description as the operative supervisory reference. [1]
A threshold is not a universal safe zone
The September 2026 booklet defines a concentration for its framework as aggregate direct, indirect or contingent obligations exceeding 25% of tier 1 capital plus the allowance for credit losses. It also explains that exposure may require committed or outstanding amounts depending on the circumstances. That supervisory definition is not a universal legal lending limit and does not make every exposure below it harmless. [1]
A small pool can be risky because of weak underwriting, rapid growth or a sharply concentrated repayment source. A larger pool can have meaningful internal diversity. The size threshold helps organize attention; it cannot replace analysis of what causes losses and how much the bank can absorb.
Likewise, compliance with a legal single-borrower limit does not prove portfolio diversification. A bank can stay within every individual limit while accumulating a large common-sector exposure across many borrowers. Legal compliance and economic risk measurement solve different problems.
Map the repayment engine
A useful exposure map starts with where the money ultimately comes from. A property loan may be secured by real estate but repaid from a tenant's business. A supplier's working-capital facility may depend on one manufacturer's orders. A consumer loan may be supported by wages from an employer that is also a major commercial borrower of the bank.
Consider a hypothetical regional bank with $30 million lent to a manufacturer, $20 million to its suppliers, $15 million in local commercial properties occupied by related businesses and $10 million in household loans to employees. The legal borrowers differ. A severe plant closure could affect all $75 million through different channels.
The exposure is not necessarily a guaranteed $75 million loss. Some borrowers may have other customers, collateral or income sources. The map identifies the population that warrants a common-shock analysis. Loss estimates then require separate assumptions for default probability, collateral recovery and timing.
Collateral can be correlated with the borrower
Collateral is most useful when it retains value in the circumstances that impair repayment. A lender financing local property may find that the same employment shock reduces both tenant cash flow and sale prices. A lender to commodity producers may see borrower income and equipment values weaken together.
This is a form of wrong-way exposure: the protection can deteriorate when it is most needed. It does not mean collateral has no value. It means recovery under a borrower-default scenario can differ from recovery at normal-market values.
A hypothetical loan of $1 million secured by property worth $1.3 million appears to have a substantial cushion. If a common shock reduces realizable value to $850,000 before selling costs, the original loan-to-value ratio no longer describes recovery. Senior claims and the time needed to sell also affect recovery.
Guarantees can concentrate rather than diversify
Several loans may have separate borrowers but the same guarantor. A portfolio can therefore appear dispersed while ultimately relying on one balance sheet. If the guarantor's assets are tied to the same sector, the protection may be weakest in the scenario affecting the borrowers.
The guarantee’s value depends on the guarantor's liquid resources, other obligations and enforceability of the promise. Adding the full guarantee value to every loan independently can overstate aggregate support. The same $10 million of available resources cannot repay ten separate $10 million guarantees in full at the same time.
The OCC's booklet includes common repayment sources and guarantors among concentration channels. It also points to indirect exposures through securities and protection providers. The broader lesson is that moving the risk into another contract does not necessarily move it to an independent economic source. [1]
Undrawn commitments can become drawn together
A credit line that is mostly unused in normal conditions may become a funding obligation during stress. Businesses facing the same disruption can draw at the same time. Concentration analysis based only on outstanding loans can therefore understate both credit exposure and needs.
Imagine ten hypothetical firms each with a $5 million committed line and $1 million currently drawn. The reported funded exposure is $10 million, while total commitments are $50 million. If all draw another $3 million during the same shock, funded exposure rises to $40 million just as credit quality weakens.
The correct conversion assumption depends on contract terms, cancellation rights and borrower behavior. It should not automatically count every commitment as fully funded in every scenario, but it should not assume unused capacity remains unused either. The stress needs to join funding demand and deterioration in repayment prospects.
A portfolio loss example
Assume a hypothetical $100 million sector pool. In a normal-case illustration, 2% defaults with 40% loss severity produce $800,000 of loss: $100 million multiplied by 2% multiplied by 40%. In a stress case, 12% defaults with 60% severity produce $7.2 million. The change reflects both more defaults and weaker recoveries.
These are invented scenario inputs, not a forecast or empirical default model. Their purpose is to show why common shocks are consequential. If the bank has $50 million of loss-absorbing equity, the stress loss equals 14.4% of that amount before earnings, taxes, allowances or other portfolio effects are considered.
A bank-wide scenario must then consider whether other supposedly different pools are exposed to the same driver. Simply summing independently calibrated average losses may miss tail dependence. Conversely, adding every worst-case pool loss without a coherent common scenario can overstate a plausible joint outcome. Consistency is more valuable than mechanical pessimism.
Diversification is not just adding categories
A bank can buy loans in a new geography and still retain the same dependence if the borrowers sell to the same customers or rely on the same funding market. A new asset class may be exposed to the same interest-rate or refinancing shock as the existing portfolio. Labels are a useful starting taxonomy, not proof of independence.
A more persuasive diversification claim explains why repayment sources react differently under relevant stresses. It may also acknowledge where correlations can rise. Historical low correlation measured during stable conditions is weak evidence for behavior during a regime change.
The current OCC booklet specifically recognizes that correlations can change over time. The original portfolio design therefore does not provide permanent evidence of safety; groupings and assumptions can become outdated. [1]
Limits need a response attached
A limit that is never approached may be too loose to influence behavior. A limit routinely breached without consequence may be little more than a reporting convention. The purpose of limits and early-warning thresholds is to trigger decisions while the bank still has options.
Possible responses include slowing new originations, tightening selected terms, obtaining more independent support, selling participations, hedging where effective or adding capital. Each has a cost and a time horizon. A plan to sell loans during stress may be unrealistic if many lenders are trying to reduce the same exposure.
The OCC booklet describes meaningful thresholds, limits, sublimits and mitigation actions in relation to a bank's size, complexity and risk profile. It does not replace judgment with one prescribed percentage for every sector. The governance question is who can approve an exception, what evidence is required and how the bank returns within its risk appetite. [1]
Data quality determines what can be seen
A concentration report is only as good as the links in its data. Missing sponsor identities, outdated industry codes, incomplete guarantor records and inconsistent collateral classifications can fragment one exposure into apparently independent pieces. Purchased loans can introduce further gaps if the buyer receives only summary fields.
A useful system allows an exposure to belong to multiple analytical pools without double counting it in the total portfolio. A property loan can be part of a geographic, tenant-industry and sponsor concentration simultaneously. The total funded amount remains one loan; the overlapping views answer different risk questions.
Reconciliation to underlying accounts and explanations for large changes determine how informative reported concentrations are. An unexplained fall after a data conversion is not evidence that risk disappeared. Reclassification, portfolio growth and genuine repayment need separate treatment.
Concentration is not automatically a bad strategy
Specialization can create knowledge, distribution advantages and better underwriting. A community bank may reasonably know its local economy better than distant sectors. A specialist lender may understand collateral and borrower behavior more deeply than a generalist. Diversifying into an unfamiliar market can introduce risks of its own.
The tradeoff is that expertise does not eliminate common shocks. A specialized business needs capital, , pricing and controls consistent with the tail exposure it accepts. A strong strategy explains why the expected return compensates for that dependence and what the bank will do if the central assumption fails.
The central test is whether the bank understands the common source of repayment behind its many accounts. Concentration management turns that understanding into limits, coherent stress tests and actionable decisions. Counting borrowers is easy; identifying when they may all need help at once is the real work.
Sources
- OCC, Concentrations of Credit, version 3.0, September 2026; actual current booklet checked October 4, 2026Official source · PDFBack to text: ↑1↑2↑3↑4↑5↑6
- OCC, current Concentrations of Credit indexOfficial source