Bank risk is not one number — it’s the interaction of three things: how much loss the bank can absorb, how many losses are coming, and whether it’s earning enough to outrun them. Here’s where each shows up in the data.
Pillar 1: capital (can it absorb losses?)
The capital ratio (equity ÷ assets) and the risk-based capital ratio measure the cushion. Thin capital — say a capital ratio under 6% — leaves little room for error. Most large US banks run 8–12%; the best-capitalised in our data include Deutsche Bank Trust Company Americas (24.8%) and Prosperity Bank (18.4%). See the safest-banks ranking.
Pillar 2: asset quality (how many losses are coming?)
| Metric | Risk signal |
|---|---|
| Nonperforming assets ÷ assets | Rising share = more problem loans |
| Texas Ratio | Problem assets approaching the cushion |
A jump in either is often the earliest warning, because bad loans erode profits first and capital later. But beware classification quirks — some past-due assets are government-guaranteed. Read the Texas Ratio explainer for the caveats.
Pillar 3: earnings (can it outrun the losses?)
Return on assets (ROA) around 1% is healthy for a large bank; negative ROA is a red flag. But high earnings can mask risk — rapid loan growth or concentration can boost profits right up until the cycle turns. The highest-ROA banks in our data (Sallie Mae at 4.31%, American Express at 3.74%) run specialised, high-margin models rather than typical retail banking. See most profitable by ROA.
The takeaway
No single metric is decisive. The clearest risk signal is when all three pillars point the same way — thin capital, rising problem assets and weak earnings together. That’s why BankGrade’s grade combines five signals. And remember: for insured depositors, FDIC insurance — not any ratio — is what protects the money.
Informational only — not financial advice or a solvency opinion. Source: FDIC BankFind Suite, Q1 2026.