of the private loan book was non-performing in Aug-2026 verif·Aug 2026↗
it was 2.65% in Nov-2023
Pillar statusAT RISKTrend▼ worsening
The latest figure does not meet the promise yet, and is moving away from it.
What was promised
That disinflation and the credit recovery hold without the financial system piling up non-performing loans.
Where it stands today
Non-performing ratio of credit to the private sector: 7.75% in August 2026, against 7.73% in July. The BCRA itself sums it up: «it remained without significant changes over the last four months». Inside, the two segments moved in opposite directions: households fell to 12.81% (12.94% in July) and companies rose to 3.80% (3.62% in July), their second month in a row of increases and, according to the BCRA, with a «generalized» rise «in loans to all economic activities». The fall for households is not all improvement: the non-performing balance fell partly because a group of banks moved uncollectible loans to off-balance-sheet accounts, and the BCRA says so in those words. By household line, cards fell (12.22%), personal loans (17.04%) and car loans (8.17%) rose, and mortgages remain the lowest, 1.65%, practically unchanged. Provision coverage fell to 87% of the non-performing book (87.6% in July). The estimated probability of default fell for the third month in a row, to 2.26%. The long arc is what matters and is reported as it is: 7.75% today against 2.65% in November 2023 (the inheritance) and 5.56% in December 2025. verif·Aug 2026↗
Delinquency rising while credit ACCELERATES → the ratio is moving on the seasoning of new lending, not on deterioration of the old book → financial deepening is still real and the number does not contradict it. R3
Delinquency still rising once credit SLOWS → it is no longer seasoning, it is deterioration → SME credit gets pricier and rationed → the satellite supplier loses precisely the lever it was going to scale with, without profits of its own. R3
An official, monthly, stable-URL series that contradicts the optimism → any claim of ours about credit quality becomes checkable → publishing the uncomfortable figure next to the favorable one is what makes the favorable one credible. R6
What we watchThat non-performing loans keep rising while credit slows: then it would no longer be the maturing of a new book but the deterioration of the old one. ⚠️ The warning this watch had written was met («if companies keep rising two months in a row, it is rewritten»): companies went from 3.50% in June to 3.62% in July and 3.80% in August, and the BCRA describes August's rise as «generalized in loans to all economic activities». Meanwhile households fell from 12.94% to 12.81%, but partly because banks moved uncollectible loans off the balance sheet, not only because people pay better. ⇒ The target of the watch changes: it is no longer whether household deterioration jumps to companies, but whether the rise in companies continues. If companies rise again in September, the deterioration has stopped being a consumer phenomenon. It is checked in the «Calidad de Cartera (por líneas)» sheet of the monthly annex to the Informe sobre Bancos.
Policies and rules related to this pillar13
We do not attribute this indicator to a single rule. Here we gather the policies and rules we track because they can affect the conditions related to this pillar, each with its status and its source.
These markets belong to sectors whose activity also shows up in relation to this pillar. We show them to connect the aggregate indicator with concrete markets for goods and services that we track in the large projects. Their presence here does not mean this indicator is the cause of that demand.
The pillar shows the economic context. For a company we run a different cross-check: which projects buy what you make, who buys, when that purchase appears and what you would need to prepare to get in.
BCRA's report on banks: the non-performing portfolio comes out with a ~2-month lag. ⚠️ SINCE 2026-08-30 THIS CLOCK COVERS MORE THAN THE HEADLINE RATIO: the indicator also publishes the breakdown by lending line from the «Portfolio Quality (by lines)» sheet of the same annex (households 12.80% against companies 3.48%, May-2026), i.e. eight further figures that age at the same pace. When refreshing, BOTH surfaces have to move: dato_actual.valor AND watchlist, which is also rendered.
verif primary source · prob primary source pending · unconf a source said it · estim our own calculation · thesis our reading · the date belongs to the datum, at the precision its source allows