Every financial crisis is described as unprecedented. In practice, the mechanism is usually familiar. Concentrations that appear reasonable in isolation reveal themselves as a single, correlated exposure once conditions change. The Savings and Loan crisis; Texas banks during oil busts; mortgage lenders and monoline insurer failures during the Global Financial Crisis; and more recently Silicon Valley Bank, First Republic, Signature Bank and Credit Suisse all failed for different proximate reasons. Structurally, they shared the same blind spot: reliance on static views of risk inside systems that evolve continuously.
Concentration risk has historically been managed as a static snapshot problem. Where are we exposed today? However, the important questions are dynamic in nature. How are our concentrations changing over time, and how do they correlate and behave under stress scenarios we haven't yet experienced? Institutions that consider those questions will be better positioned when conditions turn.
From Snapshots to Systems
Most concentration frameworks sort exposures into buckets: sectors, geographies or borrower types measured against fixed limits. This approach often unintentionally assumes risk exposures added near capacity are similar to those added earlier, correlations are stable and tomorrow will resemble yesterday.
Balance sheets do not behave that way. Each new origination changes portfolio sensitivity. Borrowers share suppliers, customers, funding sources and information channels. Depositors respond together. Operational dependencies deepen gradually and often invisibly. Concentration is not something an institution simply has; it is something it becomes.
Managing concentration risk dynamically means understanding how marginal decisions change the behavior of the system as a whole.
Pricing the Next Dollar
Loan pricing models typically use a static cost of capital assumption and do not dynamically adjust for portfolio-level concentration effects. Individual loans may be appropriately priced on their own merit while still being underpriced when consideration is given to reflect the marginal cost of deepening an existing concentration in a given sector, geography, borrower type, etc. As concentration grows, the balance sheet becomes ever more sensitive to the loss factors correlated with the underlying concentration.
Other industries handle scarcity more directly. Airlines sell early seats on a flight cheaply to stimulate demand; prices rise as capacity fills. The seats do not change, but their scarcity does. The last seat on a full flight is expensive because it consumes the final unit of available capacity.
Loan pricing tends to ignore this logic. Static limits treat the first dollar and the hundredth dollar in a sector as equivalent, even though marginal risk rises non‑linearly. The tenth large multi-family loan in a tight geography carries very different portfolio implications than the first.
What is often overlooked is marginal risk contribution: how much total portfolio risk would decrease if a given exposure were removed. That is the quantity that matters for concentration pricing, and it must be measured before it can be priced.
The challenge ahead is shifting from limit-based dashboards to contribution-based views, while also explaining to boards and examiners why that distinction matters. Done well, dynamic pricing aligns origination incentives with risk capacity without constant policy conflict and produces valuable market signals when borrowers refuse concentration premiums.
When Sector Labels Fail
Many concentration policies rely on industry classifications developed for a different economy. NAICS and SIC codes describe what borrowers do, not what risks they are exposed to or bring to a lender.
A bank may limit "retail" exposure to a modest share of the portfolio and still accumulate significant positions in logistics real estate, consumer credit, payment processing and e‑commerce lending. Separately, each sector passes concentration policy limits independently. Together, they represent a single economic bet on consumer discretionary spending and platform-driven disruption.
This is the difference between nominal and economic diversification. Industries with different labels can behave almost identically under stress.
The common blind spot is assuming that counting categories produces diversification, while what really matters is shared sensitivity to economic drivers.
Forward-thinking leaders are moving toward factor-based views of exposure: sensitivity to consumer spending, interest rates, supply chains, capital expenditure cycles or trade dependence; and updating limits as correlation structures evolve rather than only during annual policy reviews.
Liability-Side Concentration
Concentration risk is often treated as a credit problem, yet the same forces operate on the liability side of the balance sheet.
Silicon Valley Bank made this visible, but it was not unique. SVB combined a homogeneous depositor base with highly rate-sensitive assets. First Republic exhibited a related dynamic through a concentration of rate-sensitive, high-net-worth clients. Signature Bank showed how depositors tied together by sector and information flow can move collectively despite retail classification.
These failures were not fundamentally about deposit size or asset quality. They reflected an inability to measure behavioral correlation among depositors.
Retail deposits have historically been modeled as slow-moving. In homogeneous communities defined by industry, employer, geography or social network, that assumption breaks down. Digital channels and real-time information compress reaction times in ways older models do not capture.
The challenge ahead is building depositor concentration metrics by industry, geography, account type, channel dependency and network overlap, using the same rigor applied to loan books. Regulatory expectations are already moving in this direction.
The Overlooked Risk: Operations
While institutions have refined asset-side concentration frameworks, operational reliance on a small number of vendors for critical infrastructure has grown indirectly through mergers and acquisitions.
Losing money in a concentrated loan portfolio is very different from losing access to customer interfaces, payment rails or core data. Over time, the vendor you approved years ago may no longer be the one you depend on today. Platform consolidation, outsourcing layers and private-equity ownership can materially change vendor risk profiles without changing the vendor's name. Often, the only signal is a revised contractual addendum.
Concentration without explicit limits is still concentration.
Many institutions do not regularly test backup providers, inventory service level agreement performance across critical vendors, or ensure core data is backed up in a safe-harbor environment capable of being restored independently through a crisis.
Artificial intelligence may introduce needed competition and alternatives, or it may deepen fragility through opaque dependencies and new failure modes. Either outcome increases the need for systematic operational concentration analysis.
Why Limits Break in Stress
During financial crises, correlations rise sharply, often converging toward one. Exposures that appear diversified in calm markets behave as a single concentrated challenge precisely when diversification is most needed.
This dynamic contributed to monoline insurer failures in 2008, wholesale funding collapses at Continental Illinois and counterparty concentration problems at Credit Suisse. Correlation is not a characteristic of the portfolio; it is a property of the environment.
A fixed sector cap implicitly assumes stable correlations. When conditions change, the limit remains unchanged, while the economic reality does not.
The practical response is to measure concentration under both long-run and stressed correlation assumptions and focus on the gap between them. That gap represents hidden concentration.
What Dynamic Management Requires
Moving from philosophy to practice does not require novel mathematics. Capital markets have used these concepts for decades.
It does require:
- Exposure data at sufficient granularity across assets, liabilities and operations;
- Correlation and behavioral models being refreshed more frequently than policy cycles; and
- Risk signals that are integrated into front-office tools, not confined to reports.
Measurement alone is not enough. Governance, culture and vendor discipline must advance alongside analytics. Dynamic frameworks sharpen judgment rather than replace it.
What is new is bringing this discipline into credit unions and community banks to ensure the industry is better prepared for an evolving and uncertain future.
The Quiet Risk
Concentration always looks reasonable until conditions change. Institutions rarely fail because risks were unknown; they fail because frameworks assumed static behavior in dynamic systems.
The next cycle will expose concentrations that were never measured: shared dependencies, behavioral linkages and marginal risks that never made it into pricing or limits.
Institutions that pair strategic conviction with measurement rigor will not only survive that moment, they will enter it with options.
Ben Schexnayder is Director of Enterprise Risk Management for the Dallas, Texas-based financial advisory firm ALM First.
Ben Chilanga is Associate, Balance Sheet Strategy for ALM First.
© Arc, All Rights Reserved. Request academic re-use from www.copyright.com. All other uses, submit a request to TMSalesOperations@arc-network.com. For more information visit Asset & Logo Licensing.