01 · Asset quality
Non-performing loans, households and companies
There is no print for 2024 Q4. The series is discontinuous around that quarter, so the line is broken rather than interpolated. Comparisons that cross the gap compare two different definitions; we show the break instead of smoothing it.
02 · Institutions
The defaulted book, bank by bank
—Bar is the latest print, the tick is where the series opened in 2025 Q1, so a long bar with a distant tick is deterioration inside the window. Click a row to load it into the focus card.
The series starts in 2025 Q1 and is not interpolated backwards. Before that date IF.Data publishes a regulatory grading (AA–H), which is not a 90-day measure, so no earlier print is shown. Two institutions re-scope their book inside the window — BRB (3.1% → 15.5%) and CloudWalk (0.9% → 14.2%) at 2026 Q1 — and PicPay's 2025 Q4 zero print is treated as a filing gap, not as zero risk. Acquirers (Stone, PagSeguro) sit near zero structurally: their book is merchant receivables with recourse.
This ratio is computed from the SCR filing (document 3040), not the Cosif balance sheet, so it will not tie to BCB's published system NPL rate — the circle above and this panel are two different scopes, and neither is used to correct the other.
03 · Cover
Provisions against the defaulted book
—Horizontal axis is the defaulted share of the credit book (NPL90); vertical axis is the provision stock divided by the defaulted stock, on a log scale so thin and thick books share one frame. Bubble area is the credit book. Points below the full-cover rule carry a red outline. The ratio is built from two filings — see the note below.
Provisions are implied, not filed as a balance in this extract. They are reconstructed as the reported coverage ratio (provisions ÷ credit) × the credit book, while the defaulted stock comes from the SCR active portfolio in report 130. The two sources do not share a denominator, so this multiple measures provisioning intensity; it is not a filed coverage ratio and will not tie to a bank's own disclosure. Provisions ÷ credit and NPL90 are printed separately for that reason.
Below 1.0× means the provision stock is smaller than the defaulted balance — not that the bank is uncovered: write-offs, collateral and guarantees sit outside this comparison. An institution with no defaulted print or no credit book is left out rather than counted at zero, so the filer count in the header changes with the quarter.
04 · Cost
What risk costs, and what is set aside
Both series are all-universe, credit-weighted aggregates of reported ratios. They are not re-aggregated income statements.
05 · Mix
The retail book, by modality
Shares are as published in the IF.Data modality table; they are not a closed 100% breakdown of the portfolio.
Each tile is the same series across thirteen quarters, scaled to its own range.
| Institution | NPL90 | Cost of risk | Coverage ÷ credit | Cover of NPL | ROE | Credit | Share of credit | NPL90 · 6Q | Cost of risk · 13Q |
|---|
Ranked by cost of risk, the twelve-month provision charge over the book. NPL90 is the stock of defaulted credit from IF.Data report 130 — a balance, not a flow. A high charge against a low stock is provisioning ahead of the cycle, or a book that seasons quickly; the two columns answer different questions. “Cover of NPL” is the same implied provision stock divided by that defaulted balance.