Methodology

In this monitor, a non-performing loan is a loan that Bank Al-Maghrib classifies as a créance en souffrance in the statistics of Moroccan banks. The monitor does not apply a separate Serenor definition.

Bank Al-Maghrib publishes these loans in three categories: créances pré-douteuses (pre-doubtful), créances douteuses (doubtful) and créances compromises (compromised). The rules used for the years in this series are those of Circulaire n° 19/G/2002 of 23 December 2002, as amended on 9 December 2004. Circulaire n° 1/W/2025 was adopted on 5 December 2025 and applies from 1 January 2027. It does not change the figures stored here.

NPL stock is the outstanding amount of non-performing loans, in billions of Moroccan dirhams. Where Bank Al-Maghrib prints the figure in the annual supervision report, that printed figure is stored. It is not rebuilt from a more precise unpublished total.

The NPL ratio is Bank Al-Maghrib’s taux des créances en souffrance, also called the taux de sinistralité in the supervision report. The 2014 report’s statistical annex defines it as non-performing loans divided by total loans. The monitor stores the ratio Bank Al-Maghrib prints. It does not recalculate the ratio from a credit aggregate.

Provision coverage is the taux de couverture des créances en souffrance par les provisions printed by Bank Al-Maghrib. Provisions, where shown, are the specific provisions for non-performing loans printed in the same family of reports. General provisions for sensitive but still performing loans are not included in the coverage ratio.

Year-on-year growth is calculated by Serenor as (current NPL stock / previous NPL stock − 1) × 100, using the stocks in this dataset. It is labelled as calculated. Because the stored stocks are the rounded figures printed by Bank Al-Maghrib, the calculated rate can differ by about 0.1 percentage point from a growth rate printed in the narrative.

The share of a category is that category’s stock divided by the social-basis NPL stock, times 100. The change in the compromised share since 2022 is the difference between the two shares, in percentage points. It is not the percentage growth of the share.

The severity chart uses only years for which the supervision report prints a stock for all three categories on a social basis. Those years are 2019–2025. 2012–2018 are omitted. Where a report prints only a share, or only the annual change in a category, that figure is not turned into a stock. The classification text is Circulaire n° 19/G/2002: https://www.bkam.ma/content/download/499035/4963442/6%20-%20Circulaire%20n%C2%B019_G_2002.pdf

Créances pré-douteuses, article 5 of Circulaire n° 19/G/2002. They include amortising credits with an instalment unpaid 90 days after its due date, single-maturity credits unpaid 90 days after term, and leasing instalments unpaid 90 days after term. They also include credits that cannot be assessed for lack of information, and credits whose recovery is liable to be called into question because of the borrower’s capacity, events affecting managers or shareholders, disputes among shareholders, or difficulties in the borrower’s sector, even if no instalment is unpaid.

Créances douteuses, article 6. They include current-account debit balances that, for 180 days, have no real credit movements covering at least the interest and a significant part of the balance; amortising, single-maturity and leasing claims unpaid 180 days after term; claims on a counterparty in judicial reorganisation; and claims whose recovery is uncertain because the counterparty’s situation has deteriorated.

Créances compromises, article 7. They include the same families of claims at 360 days, and claims whose full or partial recovery is unlikely. The circular cites, among those cases, the loss of 75% of net assets for a public limited company or one third for another company form when the extraordinary meeting has not decided on continuation, legal proceedings to recover the claim, a judicial challenge of the claim, cessation of activity or judicial liquidation, and acceleration of the term or termination of a leasing contract. Article 8 classifies a credit repaid monthly as compromised once nine instalments are unpaid. Article 9 classifies a restructured claim as compromised if an instalment remains unpaid 180 days after its term.

Article 13 requires provisions of at least 20%, 50% and 100% of pre-doubtful, doubtful and compromised claims respectively, after reserved interest and the guarantees the circular allows. The coverage ratio in the monitor is the aggregate rate Bank Al-Maghrib prints. It is not these minima, and it is not a price.

In 2020 the three categories printed in that year’s report sum to MAD 79.7bn. The annual series uses 79.8, the figure restated on a later chart. The difference is 0.1 billion and is shown on the year. No category is adjusted to close it.

The public series is banks on a social basis (base sociale): the unconsolidated accounts of banks, as presented in the annual supervision report. Social basis is stored as solo. It is not treated as a consolidated figure.

Bank Al-Maghrib also publishes a consolidated series for banking groups, which includes foreign subsidiaries. At end-2025 that stock was MAD 141.7bn and the ratio was 8.8%, against MAD 102.3bn and 8.3% on a social basis. Consolidated observations are stored separately and are not drawn on the public chart.

Household and non-financial-company figures on this monitor cover banks and finance companies together, on a social basis, for 2021 to 2025. They are a wider population than the bank series and are not added to it.

The household series is non-performing loans held by banks and finance companies on households. The corporate series is non-performing loans held by the same institutions on non-financial companies. A year appears only when both stocks and both ratios are printed for that wording.

The window is 2021–2025. Earlier reports name consumer-credit companies rather than finance companies as a group, omit the institution from one of the two sentences, or print an approximate stock. Those years are omitted. An approximate bank-only stock, such as the 2025 phrase “près de 36 milliards” for household loans at banks, is not stored on this series.

The two stocks are not a split of the bank total. At end-2025 they are MAD 47.3bn for households and MAD 73.7bn for non-financial companies. The bank series for the same date is MAD 102.3bn. The difference is the scope, which includes finance companies, and it is not closed.

Where a later report restates a growth rate, the stock printed with the original year is kept. The 2021 household stock is 38.9. The 2023 household stock is 41.8. The 2024 corporate stock is 70.1. The change since 2021 is the difference in the printed risk rates, in percentage points. It is not a recovery rate.

The public series is annual and refers to 31 December. The monitor runs from 2012, the earliest year for which the stock and the ratio were read together from a supervision report, through 2025. Comparable social-basis figures before 2012 were not established in this release.

When a later supervision report restates an earlier year, the restated figure is stored and the original printing is kept in the observation note. The 2020 stock is 79.8 billion on the later charts; the 2020 report itself printed 79.7. The 2021 ratio is 8.6% on the later charts; the 2021 report printed 8.5%.

A missing provisions or coverage figure is left blank. It is not estimated, interpolated or replaced with zero. Coverage for 2021, and provisions for 2020, 2021 and 2022, were not read as precise printed stocks in this release.

The public annual series is taken from the Rapport annuel sur la supervision bancaire. That report states the social basis, the ratio and, in most years, coverage in one place.

Bank Al-Maghrib also publishes a monthly workbook, Ventilation du crédit bancaire par objet économique, with a line titled Créances en souffrance back to December 2001. That line is gross of provisions, in millions of dirhams. December values track the supervision-report stock closely but do not match it in every year. At December 2025 the workbook reads MAD 101.6bn against MAD 102.3bn in the 2025 supervision report. The two series are archived separately and are not spliced.

Observations were read from the official PDFs and recorded with the report, the printed page and the table or chart. Extraction method is manual. Nothing on the public page is fetched from Bank Al-Maghrib when the page loads.

A new supervision report is checked after Bank Al-Maghrib publishes it, usually in July for the preceding year. The monthly workbook can be fetched into a staging file with npm run data:bam:fetch. Staged figures are not published automatically, and the monthly series is not approved for the public chart.

If a fetch fails, the page continues to show the latest verified year. It does not replace that year with zero or with an unverified figure.