Concentration is the risk that ends banks
Institutions rarely fail because everything deteriorated at once. They fail because of one thing they had too much of, correlated in a way nobody had measured.
Look at how banks actually fail and a pattern shows up quickly. It is almost never a broad, even decline across a diversified book. It is one exposure: one property type, one borrower group, one town, one employer, one funding source. That exposure turned out to be much larger than the balance sheet made it look, because the pieces moved together.
That is the whole concept, and it is more slippery than it sounds. Concentration is not a size problem. It is a correlation problem wearing a size problem's clothes.
The thresholds are a screen, not a limit
The 2006 interagency guidance on commercial real estate concentrations gave supervisors two screening criteria that are now quoted in every board packet in the country: construction and land development at 100% or more of total risk-based capital, and total commercial real estate at 300% or more of capital combined with 50% growth over the preceding thirty-six months.
Two things about those numbers get lost.
The first is that they are not limits. Crossing them is not a violation and does not, by itself, mean the bank is doing anything wrong. It means the examiner will look harder at whether the institution's risk management is proportionate to the concentration it is carrying. A well-run bank above the threshold, with granular monitoring, stress testing on the specific portfolio and documented underwriting discipline, is in a considerably better position than a bank below it with none of that.
The second is that they are category tests, and category is a weak proxy for correlation. Total CRE at 250% of capital sounds comfortable next to the screen. If most of it is retail strip centers within thirty miles of each other, dependent on the same regional consumer economy, the category test has told you nothing useful about what happens when that economy turns.
Measure at the level where the risk actually correlates
The practical work is to ask what would have to go wrong, and then measure exposure to that, rather than to the reporting category.
By repayment source. Four separate loans to four separate borrowers, all of whose tenants are suppliers to the same manufacturer, are not four exposures. They are one exposure with four sets of documents. This is the single most common form of hidden concentration in community banking, and nothing in the standard category reporting will surface it.
By geography, at a real radius. County-level reporting is often too coarse to matter. If the collateral is clustered along one corridor or in one submarket, that is the unit of analysis.
By collateral type and vintage together. Loans underwritten in the same eighteen months, against the same property type, at the valuation levels of that moment, share an assumption. When that assumption is wrong it is wrong for all of them simultaneously.
By funding. Concentration is not only an asset-side idea. A deposit base weighted toward a small number of large uninsured relationships, or toward one sector, is a concentration with the same correlation logic and a much shorter fuse.
| Cut the book by | What actually correlates | What the category test misses |
|---|---|---|
| Repayment source | One employer, one tenant base, one supply chain | Four borrowers reading as four exposures |
| Geography, at a real radius | A corridor or submarket, not a county | Clustering inside a "diversified" county total |
| Collateral type and vintage | The valuation assumptions of one underwriting window | That the assumption is wrong for all of them at once |
| Funding | A handful of large uninsured relationships, or one sector | That concentration is not only an asset-side idea |
Four loans to four borrowers in one town, dependent on one employer, is one loan with four sets of documents.
Concentration as a choice
Community banks concentrate deliberately, and frequently that is the business. You lend against what you understand, in a market you know, and the knowledge is the edge. Nobody sensible is arguing for a synthetically diversified community bank; the diversified version would be a worse bank with a better-looking report.
So the distinction is not between concentrated and diversified institutions. It is between a concentration that was decided, sized against capital and monitored, and one that was arrived at, discovered later, and described afterwards.
Both look the same on the balance sheet. They differ only in whether anyone wrote down what would have to go wrong, and when.
← Back to all insights