DataBrighten

    Bank nonperforming loans to total gross loans (%)

    2,357 × 6rows × columns: 154 countries, 2000 to 2025

    Real data: World Bank, World Development Indicators, indicator FB.AST.NPER.ZS, retrieved 19 November 2026. Values are as published; we only reshaped the file.

    textnumberdateyes/no
    Rows2,357country-year values
    Countries154with at least one value
    Years2000 – 202526 years with data
    LicenceCC BY 4.0credit the World Bank

    What it measures

    The indicator measures the proportion of a deposit taker’s loan portfolio that is impaired or at risk of default. It is calculated as the ratio of non-performing loans (NPLs) to total gross loans, where NPLs are defined as loans that are past due by 90 days or more or are otherwise considered unlikely to be repaid in full without the realization of collateral. Both non-performing loans and total gross loans should be reported at their gross book value, without deducting for loan-loss provisions or collateral. This indicator provides a key measure of asset quality and potential credit risk in the banking system.

    Limits and exceptions (World Bank note)

    Reporting countries compile the data using different methodologies, which may also vary for different points in time for the same country. Users are advised to consult the accompanying metadata on the IMF FSI website (data.imf.org) to conduct more meaningful cross-country comparisons or to assess the evolution of the indicator for any of the countries.

    The numbers in 2025

    In 2025, 90 countries have a value. The highest is Ghana (18.92); the lowest is Sweden (0.426). The median country value is 2.56.

    Top 10 countries, 2025

    %

    Source: World Bank, FB.AST.NPER.ZS, retrieved 19 November 2026. Computed from bank-nonperforming-loans-to-total-gross-loans-percent.csv.

    Highest, 2025

    #CountryValue
    1Ghana18.92
    2St. Kitts and Nevis15.83
    3San Marino14.8
    4Ukraine13.92
    5Tonga13.47
    6Kyrgyz Republic10.06
    7Sri Lanka9.87
    8Dominica8.92
    9St. Lucia8.51
    10Nigeria8.12

    Lowest, 2025

    #CountryValue
    1Sweden0.426
    2Norway0.449
    3Canada0.706
    4Switzerland0.774
    5Lithuania0.794
    6United Kingdom0.948
    7United States0.96
    8Australia1
    9Nicaragua1.04
    10Saudi Arabia1.04
    How many countries report, by decade
    DecadeMost countries in one year
    2000s78
    2010s148
    2020s145

    First rows

    bank-nonperforming-loans-to-total-gross-loans-percent.csv

    countrytextiso3textregiontextincome_grouptextyearnumbervaluenumber
    AfghanistanAFGMiddle East, North Africa, Afghanistan & PakistanLow income201049.9013229069412
    AfghanistanAFGMiddle East, North Africa, Afghanistan & PakistanLow income20114.6968178252385
    AfghanistanAFGMiddle East, North Africa, Afghanistan & PakistanLow income20124.95890140377645
    AfghanistanAFGMiddle East, North Africa, Afghanistan & PakistanLow income20134.85228704838131
    AfghanistanAFGMiddle East, North Africa, Afghanistan & PakistanLow income20147.7810908959605
    AfghanistanAFGMiddle East, North Africa, Afghanistan & PakistanLow income201512.0536179767696
    AfghanistanAFGMiddle East, North Africa, Afghanistan & PakistanLow income201611.0743789647307
    AfghanistanAFGMiddle East, North Africa, Afghanistan & PakistanLow income201712.2018301907814

    Data dictionary

    ColumnTypeMeaning
    countrytextCountry or economy, World Bank name
    iso3textISO 3166-1 alpha-3 code
    regiontextWorld Bank region
    income_grouptextWorld Bank income group
    yearnumberCalendar year of the value
    valuenumberBank nonperforming loans to total gross loans (%), exactly as the World Bank API returns it (not rounded)

    Only country-years with a value are in the long file; blank cells in the wide file mean no value was published.

    Practice questions

    answers computed from the file

    Which country has the highest value in 2025?

    Ghana, with 18.92.

    Excel (on the long CSV: E = year, F = value)

    =FILTER(A2:A100000,(E2:E100000=2025)*(F2:F100000=MAXIFS(F2:F100000,E2:E100000,2025)))

    SQL (table wb = the CSV)

    SELECT country, value FROM wb WHERE year = 2025
      AND value = (SELECT MAX(value) FROM wb WHERE year = 2025);

    pandas (df = pd.read_csv("bank-nonperforming-loans-to-total-gross-loans-percent.csv"))

    y = df[df.year == 2025]
    y[y.value == y.value.max()][["country", "value"]]

    File: bank-nonperforming-loans-to-total-gross-loans-percent.csv

    How many countries report a value for 2025?

    90 of the 217 economies in the file's country list.

    Excel (on the long CSV: E = year, F = value)

    =COUNTIF(E2:E100000,2025)

    SQL (table wb = the CSV)

    SELECT COUNT(*) FROM wb WHERE year = 2025;

    pandas (df = pd.read_csv("bank-nonperforming-loans-to-total-gross-loans-percent.csv"))

    (df.year == 2025).sum()

    File: bank-nonperforming-loans-to-total-gross-loans-percent.csv

    What is the median across countries in 2025?

    2.56 (median of the 90 country values; not a World Bank aggregate).

    Excel (on the long CSV: E = year, F = value)

    =MEDIAN(FILTER(F2:F100000,E2:E100000=2025))

    SQL (table wb = the CSV)

    -- SQLite has no MEDIAN(); this picks the middle value(s)
    SELECT AVG(value) FROM (SELECT value FROM wb WHERE year = 2025 ORDER BY value
      LIMIT 2 - (SELECT COUNT(*) FROM wb WHERE year = 2025) % 2
      OFFSET (SELECT (COUNT(*) - 1) / 2 FROM wb WHERE year = 2025));

    pandas (df = pd.read_csv("bank-nonperforming-loans-to-total-gross-loans-percent.csv"))

    df[df.year == 2025]["value"].median()

    File: bank-nonperforming-loans-to-total-gross-loans-percent.csv

    Which country rose the most between 2015 and 2025?

    Sri Lanka: from 3.24 to 9.87, a rise of 6.63 (out of 79 countries with both years).

    Excel (on the long CSV: E = year, F = value)

    Use the wide file (bank-nonperforming-loans-to-total-gross-loans-percent-wide.csv): add a column = [2025] - [2015], then sort it largest to smallest.

    SQL (table wb = the CSV)

    SELECT a.country, b.value - a.value AS rise FROM wb a JOIN wb b ON a.iso3 = b.iso3
    WHERE a.year = 2015 AND b.year = 2025 ORDER BY rise DESC LIMIT 1;

    pandas (df = pd.read_csv("bank-nonperforming-loans-to-total-gross-loans-percent.csv"))

    w = df.pivot(index="country", columns="year", values="value")
    (w[2025] - w[2015]).dropna().nlargest(1)

    File: bank-nonperforming-loans-to-total-gross-loans-percent.csv

    Which year has the most countries reporting?

    2018, with 148 countries. (Later years are still filling in as countries report.)

    Excel (on the long CSV: E = year, F = value)

    Make a pivot table: year in Rows, Count of value in Values, then sort by the count.

    SQL (table wb = the CSV)

    SELECT year, COUNT(*) AS n FROM wb GROUP BY year ORDER BY n DESC, year DESC LIMIT 1;

    pandas (df = pd.read_csv("bank-nonperforming-loans-to-total-gross-loans-percent.csv"))

    df.groupby("year").size().sort_values(ascending=False).head(1)

    File: bank-nonperforming-loans-to-total-gross-loans-percent.csv