When Bank Health Claims Don't Add Up: NPLs and Capital Adequacy 2026
Introduction
On August 6–7, 2026, the U.S. Government Accountability Office (GAO) reported that DOGE’s celebrated claim of $110 billion in federal savings was “vastly inflated and riddled with errors.” A claimed headline number — repeated for months — did not survive independent verification. The episode is a reminder that in any system where someone reports their own numbers, the reported figure and the verified figure can diverge dramatically.
Banking is the most important such system in the world economy. Banks report their own nonperforming loan ratios and capital buffers, and those self-reported figures drive everything from regulator stress tests to investor confidence. When the numbers don’t add up, the consequences run far beyond a correction: they are the story of every banking crisis. So which economies’ bank health claims deserve the closest scrutiny? Two metrics, read together, tell the story.
The two numbers that cross-check each other
Nonperforming loans (NPLs) are the share of a bank’s loan book that is past due or unlikely to be repaid — the most direct measure of whether loans are actually being repaid. Across 144 economies, the World Bank’s data shows a range from 0.2% in South Korea to 37.4% in Ukraine.
Capital adequacy (bank capital to assets) is the buffer shareholders have put up to absorb losses — the cushion that keeps a bank solvent when loans go bad. It ranges from 47.9% of assets in the Dominican Republic to negative territory in Equatorial Guinea (-6.4%).
Neither number is meaningful alone. High capital with high NPLs means the buffer may be doing real work. Low capital with high NPLs is the dangerous combination — the bank health claim that doesn’t add up. Our explainer what bank capital adequacy means and what nonperforming loans mean for bank stability cover each metric in depth.
Where loan books look cleanest
The lowest NPL ratios in the world — the countries whose banks claim the healthiest loan books:
| Country | NPLs (% of gross loans) | Year | Global Rank (144) |
|---|---|---|---|
| South Korea | 0.2% | 2022 | 1st |
| Norway | 0.4% | 2023 | 3rd |
| Sweden | 0.4% | 2023 | 4th |
| Canada | 0.5% | 2023 | 6th |
| Australia | 0.7% | 2022 | 7th |
| Switzerland | 0.8% | 2023 | 8th |
| United States | 0.8% | 2023 | 9th |
| Denmark | 0.9% | 2023 | 11th |
These are also, with few exceptions, economies with high financial depth — see our ranking of where bank borrowing runs deepest. Clean loan books are easier to maintain in wealthy, diversified economies with strong legal enforcement of contracts.
Where the claims strain credulity
The other end of the table shows economies where reported loan quality is far weaker — and where, in several cases, capital buffers are simultaneously thin:
| Country | NPLs (% of gross loans) | Capital (% of assets) | Year |
|---|---|---|---|
| Ukraine | 37.4% | 5.0% | 2023 |
| Chad | 31.5% | -1.0% | 2023 |
| Equatorial Guinea | 31.1% | -6.4% | 2023 |
| San Marino | 21.0% | 5.9% | 2023 |
| Ghana | 20.6% | 5.2% | 2023 |
| Algeria | 20.3% | 8.4% | 2022 |
| Angola | 15.0% | 10.0% | 2021 |
In Ukraine, Chad, and Equatorial Guinea, the combination is unambiguous: over a third of loans nonperforming and capital buffers near or below zero. In Ghana, NPLs above 20% with capital at 5.2% leaves a thin margin — the bank health claim here is much harder to believe than South Korea’s. Countries recovering from conflict or commodity shocks dominate this list, which is why we examined how conflict affects financial systems and where bank loans go bad.
Capital adequacy: the buffer that can absorb (or fail to)
Capital adequacy tells you how much of the shock a banking system can absorb. The strongest buffers:
| Country | Capital (% of assets) | Global Rank (143) |
|---|---|---|
| Dominican Republic | 47.9% | 1st |
| Tajikistan | 20.9% | 2nd |
| Maldives | 19.1% | 3rd |
| Argentina | 18.3% | 4th |
| Uganda | 15.5% | 6th |
| Saudi Arabia | 13.7% | 10th |
And the thinnest — where a loan-book surprise has the least to fall back on:
| Country | Capital (% of assets) | Global Rank (143) |
|---|---|---|
| Denmark | 4.9% | 137th |
| Pakistan | 4.8% | 138th |
| Canada | 4.7% | 139th |
| Japan | 4.1% | 140th |
| Chad | -1.0% | 142nd |
| Equatorial Guinea | -6.4% | 143rd |
The striking thing about the bottom of this table is how many of these are wealthy, stable banking systems. Japan and Denmark run thin capital ratios because their assets are high-quality and their funding stable — the claim is credible. Chad and Equatorial Guinea run thin capital ratios with a third of loans nonperforming — the claim is not. One ratio, two very different stories. That is exactly the point the GAO made about DOGE’s savings: a single reported number carries no information until you know how it was produced.
Analysis: the credibility test
The DOGE episode and the NPL/capital data share a structure. In both cases, the reporting party has an incentive to make the number look good, and the only discipline is independent verification:
- Cross-check the metrics. A bank can window-dress one ratio, but it is much harder to make NPLs, capital, and liquid reserves all look healthy at once when reality is deteriorating. Our article how to read a bank’s health walks through the triangulation.
- Check the year. NPL data ranges from 2021 to 2023 across countries — a 2021 reading from Angola is not comparable to a 2023 reading from the United States. As with DOGE’s claims, timeliness is part of credibility.
- Check the context. A 4.7% capital ratio in Canada is a different statement than a 4.8% ratio in Pakistan, because the underlying asset quality differs. Ratios only make sense with the loan book attached.
For the broader picture on what bank capital protects against, see what bank capital adequacy ratios mean, and for liquidity as a second buffer, bank liquidity reserves explained.
Context
The GAO finding lands in a year when bank supervision is under scrutiny everywhere. Regulators in advanced economies have spent a decade stress-testing capital; emerging markets are still building the data infrastructure to do the same. The countries where the NPL and capital numbers disagree most — Chad, Equatorial Guinea, Ukraine — are precisely the ones where external oversight is weakest and the need for independent verification greatest. That is not a coincidence: like DOGE’s receipts, the numbers are only as good as the audit behind them.
Sources & method
Nonperforming loans (% of total gross loans, World Bank WDI indicator FB.AST.NPER.ZS) and bank capital to assets (FB.BNK.CAPA.ZS) cover 144 and 143 economies respectively; latest data year varies by country (2021–2023). Ranks are among economies with available data; rank 1 = lowest NPLs / highest capital. Where a country lacks one of the two metrics, the cell is marked ”—”. See our methodology page and the reading statistics responsibly guide for how we handle differing data years and country coverage.