At-risk banks jumped 57% in a single quarter

At-Risk Banks (Score < 90)
120
→
189
+57.5%
FDIC Problem Proxy (Score < 81)
47
→
87
+85.1%
Average Bank Score
96.3
→
96.1
−0.2 pts
Avg Failure Prob (1Q)
1.27%
→
1.71%
+0.44pp
Left-tail concentration: the median bank is fine; the vulnerable are deteriorating fast

The system-wide mean score barely moved (−0.2 pts), but the 1st-percentile score dropped 11 points (79.96 → 69.00) while the median fell just 0.08 pts. This is not a broad system-wide deterioration, it is a sharp, concentrated deterioration in the most vulnerable segment of the industry. 86 banks crossed into at-risk territory (Score < 90) for the first time, while only 14 improved enough to exit.

The primary drivers are capital compression (median Tier 1 ratio −1.04 pp) and sharply slowing asset growth (−3.04 pp). Paradoxically, earnings metrics actually improved quarter-over-quarter, suggesting banks are earning more but facing balance-sheet and capitalization pressures regardless.

Risk tier counts by quarter

At-Risk Tier Breakdown
Banks with Score < 90 — by severity tier
Q2 2026 Q3 2026
Score Distribution (Full Population)
Bank Score buckets — lower score = higher risk
Q2 2026 Q3 2026
Risk Tier Score Range Q2 Count Q2 Share Q3 Count Q3 Share Change % Change
HIGH RISK TIERS
Extreme Risk 0–20 4 0.09% 6 0.14% +2 +50%
Very High Risk 20–40 5 0.12% 8 0.19% +3 +60%
High Risk 40–60 9 0.21% 16 0.37% +7 +78%
Elevated Risk 60–80 26 0.61% 52 1.22% +26 +100%
MODERATE RISK
Moderate Risk 80–90 76 1.77% 107 2.51% +31 +41%
LOW RISK
Low Risk 90–100 4,176 97.2% 4,078 95.6% −98 −2.3%
Total Banks — 4,296 100% 4,267 100% −29 charters
System-wide median metrics — all banks

The paradox: earnings improved while capital eroded

Earnings metrics (ROA, ROE, NIM) all moved in the right direction between Q2 and Q3, with NIM expanding significantly (+0.88 pp). Yet capital ratios fell, reserve coverage declined, and asset growth decelerated sharply. This combination of profitable but capital-constrained and slowing growth is the signature of an industry under balance-sheet pressure even as the income statement looks decent.

The model responds primarily to capital and growth signals, which explains why rising NIM didn't offset the risk increase.

Metric (Median, All Banks) Q2 2026 Q3 2026 Change Direction
CAPITAL ADEQUACY
Tier 1 Capital Ratio 11.64% 10.60% −1.04 pp ↓ Worse
ASSET QUALITY
Non-Performing Loans / Total Loans 0.432% 0.448% +0.016 pp ↓ Worse
Loan Loss Reserves / Total Loans 1.199% 1.065% −0.134 pp ↓ Worse
MANAGEMENT / EFFICIENCY
Efficiency Ratio 67.22% 66.59% −0.63 pp ↑ Better
EARNINGS
Return on Assets (ROA) 0.608% 0.660% +0.052 pp ↑ Better
Return on Equity (ROE) 6.70% 7.42% +0.72 pp ↑ Better
Net Interest Margin (NIM) 3.07% 3.95% +0.88 pp ↑ Better
LIQUIDITY
Loans / Deposits (Loan-to-Deposit Ratio) 70.26% 68.12% −2.14 pp ↑ Better
SENSITIVITY / GROWTH
Asset Growth (YoY) 5.02% 1.98% −3.04 pp ↓ Worse
Loan Growth (YoY) 3.65% 2.42% −1.23 pp ↓ Worse
Deposit Growth (YoY) 3.18% 1.76% −1.42 pp ↓ Worse
Core Deposit Growth (YoY) 2.91% 1.54% −1.37 pp ↓ Worse
Biggest movers between Q2 and Q3

Largest Score Declines
Banks that deteriorated most Q2→Q3
1.
Bank of Vici
OK
−75.8 pts
2.
Eastern National Bank
FL
−46.1 pts
3.
Talbot State Bank
GA
−44.3 pts
4.
First Bank of Utica
NE
−41.2 pts
5.
Inland Western Bank
UT
−38.6 pts
Largest Score Improvements
Banks that improved most Q2→Q3 (exited at-risk)
1.
Stride Bank National Association
OK
+16.5 pts
2.
Community First Bankshares
ND
+14.2 pts
3.
First National Community Bank
PA
+12.8 pts
4.
Riverton Savings Bank
WY
+11.3 pts
5.
Farmers & Merchants State Bank
MN
+9.7 pts
Flow analysis: 86 banks newly at-risk, 14 improved out

86 banks crossed into at-risk territory (Score < 90) between Q2 and Q3, primarily entering the Moderate (80–90) bucket. The score bucket 85–90 alone absorbed 73 newly deteriorating banks (+24 net). These are borderline cases. That is, these banks not in deep distress, but the movement is meaningful enough to flag. Only 14 banks improved sufficiently to exit at-risk status, yielding a net increase of 69 (the total at-risk count rose by 69 from 120 to 189).

The worst deteriorators (Extreme, Very High tiers) represent distinct idiosyncratic situations (i.e., institution-specific shocks rather than a systemic pattern). The systemic story is concentrated in the Elevated tier doubling (26→52) and the Moderate tier growing by 41%.

What Q3 2026 may signal for the banking sector

Capital compression + growth deceleration is the core story

The 1.04 pp decline in median Tier 1 capital ratios in a single quarter is striking. Banks can lose capital ratios through three mechanisms: (1) loan losses exceeding reserves, (2) dividend/buyback distribution exceeding earnings, or (3) rapid balance sheet expansion diluting ratios before retained earnings catch up. With NPL ratios rising modestly and earnings improving, mechanism (3) is the most likely culprit, even as growth itself is slowing. This may reflect prior-quarter expansion still being absorbed.

The reserve coverage decline (1.199% → 1.065%) is a secondary concern. While coverage ratios above 1% remain adequate, a 13 bp decline suggests banks are releasing reserves faster than NPLs warrant, possibly in response to earnings pressure. This leaves less buffer for the inevitable loan deterioration cycle.

Elevated tier doubling is the leading indicator to watch

The Elevated tier (Score 60–80) doubled from 26 to 52 banks in one quarter. Historically, Elevated-tier banks face meaningfully higher actual failure rates over the subsequent 4–6 quarters than the broader population. This tier represents the pipeline for potential FDIC problem-bank designation. Monitor Q4 2026 closely: if the Elevated tier continues growing or if any current Elevated-tier banks slip into Very High/Extreme, the FDIC problem bank count will likely rise significantly in 2027.

The geographic and size distribution of newly at-risk banks would be worth looking into if the deterioration is concentrated in specific regions (e.g., commercial real estate-heavy markets) or size bands (community banks below $1B).

NIM expansion is a genuine positive but model-subordinate

The +0.88 pp NIM expansion is the most significant single-metric improvement this quarter and reflects the lagged repricing benefit of the 2022–23 rate cycle finally flowing through bank income statements. This is real earnings power that improves banks' ability to build capital organically going forward. However, the underlying PennAnalytics models weight capital and growth features heavily. NIM alone is insufficient to offset capital ratio deterioration in the model's risk calculus. The model is correctly prioritizing balance-sheet integrity over income-statement optics.

At model retraining (Q4 2026 data, ~February 2027), we'll consider whether NIM's predictive weight should be revisited given the current rate environment.

⚠︎ Note on FDIC Failures — Q3 2026 Counts Understate True Stress

The tier counts above reflect only banks that filed FFIEC Call Reports as of September 30, 2026. Banks that failed during Q3 exit the reporting population before the quarter-end cutoff and are therefore absent from the Q3 CSV entirely — they are not counted in any tier, including Extreme Risk. Four FDIC-insured institutions failed during Q3 2026:

Bank State Closed Last Scored Last Score Last Tier
Kentland Federal S&L IN 10-Jul-26 Q1 2026 18.09 Extreme Risk
Small Business Bank KS 17-Jul-26 Q1 2026 16.05 Extreme Risk
Tioga-Franklin Savings Bank PA 21-Aug-26 Q2 2026 3.70 Extreme Risk
Nano Banc CA 25-Sep-26 Q2 2026 35.32 Very High Risk

Three of the four carried Extreme Risk scores well below 20 in their final scored quarter. Nano Banc deteriorated rapidly from 64.71 (High Risk, Q1) to 35.32 (Very High Risk, Q2) before closing in September. The Q3 at-risk counts — and especially the Extreme Risk count of 6 — should therefore be read as a floor. Source: FDIC Failed Bank List, cross-referenced against FFIEC Call Report cert numbers.