Which of the banks you depend on should you worry about this quarter, and why?
Regulators rate every US bank in confidence. Fintechs with a sponsor bank, treasurers with uninsured deposits and banks with counterparty exposure have only the public filings. Bank Health reads 26 years of them and scores every FDIC-insured bank on its chance of distress in the next twelve months.
Most banks are low risk. A thin edge carries most of the danger.
Risk bands that move with the economy
Each band means a level of risk measured on history, not a rank. So the High band swelled in the 2009–2011 crisis and is near its lowest now.
Bank-quarters per year in each band. Cut-offs were set using 2003–2019 predictions, each year scored by a model that had not seen it; 2020–2024 was held back as a test.
The twenty highest scores, and what drives them
| # | Bank | Probability of distress | Main reason | Driver |
|---|---|---|---|---|
| Loading… | ||||
Driver chip = the larger group push (credit vs earnings) in the model's explanation of the score; main reason = the largest single measure. All twenty are in the High band. Banks under three years old and US offices of foreign banks are not scored.
How to read a score
The probability
The model's estimate that the bank enters distress within twelve months: it fails, its bad loans pass 3%, its Tier 1 leverage falls below 5%, or it loses money two quarters running.
The band
A plain-language level set on history. Of banks in the High band, about 4 to 5 in 10 entered distress within a year; Elevated, about 1 in 5 to 1 in 4; Watch, about 1 in 15 to 1 in 10; Low, fewer than 3 in 100 (2003–2019 and 2020–2024).
The reasons
The three measures pushing the score up most, each beside the median of banks of the same size, so you can see how far the bank sits from normal.