In 2015, an average Italian bank might report that roughly 16 to 18% of its loans were going bad. That same year, a typical large US bank reported closer to 1.5%. Same business (lending money and getting paid back), a tenfold difference in visible rot.
That gap is not an accident, and reading it correctly is one of the most useful skills in banking analysis. It comes down to three metrics: the NPL ratio, the coverage ratio, and the cost of risk. Learn to calculate and interpret these three, and you can size up the health of almost any bank's loan book in minutes.
NPL stands for Non-Performing Loan: a loan where the borrower has stopped paying, usually defined as more than 90 days past due, or where the bank judges repayment unlikely.
The NPL ratio measures how much of the loan book has gone bad:
NPL ratio = Non-performing loans / Total gross loans"Gross" means before deducting any provisions (money the bank has set aside for losses). We use gross loans so the ratio is not artificially flattered.
Imagine Banca Esempio has a loan book of 100 billion euros. Of that, 12 billion is non-performing.
NPL ratio = 12bn / 100bn = 12%That is high. For context, benchmarks worth memorizing (all approximate, and they move with the cycle):
So a 12% NPL ratio would flag Banca Esempio as a problem bank by 2026 standards, though it would have looked normal in Italy a decade earlier.
An NPL ratio tells you how many loans went bad. It does not tell you whether the bank is prepared for the losses. That is what the coverage ratio does.
When a loan looks doubtful, the bank books a provision: an accounting charge that sets aside money against the expected loss. The stock of these set-asides is the loan loss allowance (or loan loss reserves).
Coverage ratio = Loan loss allowance / Non-performing loansBack to Banca Esempio: 12 billion in NPLs, and suppose it has set aside 6 billion in allowances.
Coverage ratio = 6bn / 12bn = 50%A 50% coverage ratio means the bank has already recognized losses on half the face value of its bad loans. The interpretation:
European banks have typically targeted coverage in the 40 to 60% range in recent years. The exact "right" number depends on collateral: a mortgage book backed by real estate can justify lower coverage than an unsecured credit-card book, because the bank can seize and sell the house.
The NPL ratio and coverage ratio are snapshots (stocks). The cost of risk is a flow: how much new loss the bank is booking this year, relative to its lending.
Cost of risk = Loan loss provisions (this period) / Total loansIt is usually quoted in basis points (bps), where 100 bps = 1%.
If Banca Esempio books 800 million euros of new provisions this year against its 100 billion loan book:
Cost of risk = 800m / 100bn = 0.8% = 80 bpsBenchmarks to anchor on:
Cost of risk is the metric that hits the income statement directly. Higher provisions mean lower profit, which is why analysts watch it every quarter.
Vérification des acquis
1. Why does the NPL ratio use gross loans (before provisions) in the denominator rather than net loans?
2. Two banks report identical 3% NPL ratios in the same year. What does this alone tell you about their relative loan-book health?
3. A borrower who has stopped paying and is more than 90 days past due is classified as non-performing primarily because:
4. Select ALL correct answers about why one country's banks may show a much higher aggregate NPL ratio than another's in the same year.
Sélectionnez toutes les réponses correctes.
5. Select ALL correct answers about how the three asset-quality metrics relate to one another.
Sélectionnez toutes les réponses correctes.
Now the payoff: why the same business produced a tenfold gap. Four drivers, all worth knowing.
This is the biggest one. In the US, a bank can foreclose and clear a bad loan relatively quickly. In Italy, recovering collateral through the courts historically took several years (estimates of 4 to 8 years were commonly cited). A loan stays classified as non-performing until it is resolved. Slow courts mean NPLs pile up on the books instead of flowing off. The stock stays high even if new defaults are similar.
US banks routinely sell or write off bad loans fast. For years, Europe lacked a liquid market to offload NPL portfolios. Italian banks were stuck holding them. The market did eventually develop: large NPL portfolio sales and securitizations, helped by Italy's GACS scheme (a state guarantee on senior tranches of securitized bad loans), drove the ratio down sharply in the late 2010s.
The euro area suffered two downturns back to back: the 2008 to 2009 financial crisis, then the 2011 to 2013 sovereign debt crisis. Italy's economy barely grew for a decade. More borrower defaults, fewer recoveries.
The European Central Bank (ECB) took over supervision of the largest euro area banks in 2014 under the Single Supervisory Mechanism (SSM). It pressured banks to recognize bad loans honestly and set NPL reduction targets. The ECB's guidance to banks on NPLs accelerated the fall. What looks like a natural decline was partly regulators forcing the issue.
Never read one in isolation. A worked comparison:
|------|-----------|----------|------|
| Bank A | 2% | 65% | Clean book, conservatively provisioned. Healthy. |
| Bank B | 2% | 30% | Clean-looking, but under-reserved. Watch for surprises. |
| Bank C | 10% | 60% | Ugly book, but losses largely recognized. Known problem. |
| Bank D | 10% | 30% | Ugly book AND under-reserved. The danger zone. |
Bank D is where crises start. High bad loans plus thin coverage means large unrecognized losses lurking. When they surface, capital gets wiped out fast.
This is exactly the profile many Italian and Greek banks showed in the early 2010s, and why the cleanup took a decade and, in some cases, state intervention.
You will increasingly see NPE (Non-Performing Exposure) instead of NPL in European reporting. It is a broader EBA-defined measure that includes not just loans but other exposures such as debt securities and off-balance-sheet commitments. For most analysis the intuition is identical: it is the share of credit that has gone bad. Just know that NPE ratios can differ slightly from old NPL ratios for the same bank.