Concentration Risk and Granularity Adjustments

Standard portfolio credit models like the Basel IRB formula assume a bank's loan book is infinitely granular — made up of so many small, roughly equal exposures that no single default can meaningfully move the whole portfolio's loss. Real books rarely look like that: a handful of large corporate loans, or a cluster of loans all tied to one industry or region, can dominate the risk even in a book with thousands of line items. Concentration risk is the extra danger this creates, and a granularity adjustment is the correction added on top of the base model to account for it.

There are two flavors of concentration to watch for: name concentration (a few borrowers making up an outsized share of total exposure) and sector concentration (many small borrowers that are all correlated because they share the same industry or region, so they tend to default together in a downturn).

Worked example. Two banks each hold $1 billion in loans. Bank A has 1,000 loans of $1 million each, spread across unrelated industries. Bank B has the same $1 billion, but 200 of that is a single $200 million loan to one borrower. The base IRB formula, which assumes granularity, would assign both banks a similar capital charge. A granularity adjustment recognizes that Bank B's single $200 million exposure defaulting wipes out a fifth of the portfolio in one event, something Bank A's book simply cannot experience, and adds a capital add-on proportional to how concentrated the largest exposures are.

Regulators historically imposed single-obligor limits (a cap on exposure to any one borrower as a percentage of capital) as a blunter, simpler alternative to computing a precise granularity adjustment for every portfolio.