EBA staff paper — Systematic backtesting of probability of default models with regulatory data
Tag: S-2026-04-29-eba-pd-backtesting Type: paper (EBA staff paper) Author(s): European Banking Authority (staff) Date of source: 2026-04-29 (publication date on EBA publications page) Date ingested: 2026-06-01 Authority weight: medium — EBA staff paper; reflects staff analysis and methodology rather than binding Guidelines or an adopted EBA position. Raw file: S-2026-04-29-eba-pd-backtesting.md. External URL (PDF): https://www.eba.europa.eu/sites/default/files/2026-04/9917133d-9e43-4ceb-89ff-e3578f731c13/Staff%20paper%20-%20Systematic%20backtesting%20of%20probability%20of%20default%20models%20with%20regulatory%20data.pdf
What it claims
An EBA staff paper presenting a systematic approach to backtesting probability of default (PD) models using regulatory (supervisory) data. Based on the publication-page listing, the paper sets out a methodology for assessing the predictive performance of IRB PD models against realised default outcomes drawn from regulatory data holdings — i.e. a supervisory-data-driven benchmark for ongoing PD-model validation and performance monitoring. The full text of the paper was not retrieved this run; the substance here is taken from the document title and its listing on the EBA publications page.
Notable quotes
No verbatim quotes retrieved this run (the PDF itself was not fetched; only the EBA publications-page listing was captured).
What’s speculative vs. asserted
- Asserted (from the EBA publications page): the document exists, is an EBA staff paper, is titled “Systematic backtesting of probability of default models with regulatory data”, and is dated 29 April 2026.
- Speculative / not yet confirmed: the specific statistical methodology, datasets, scope (IRB vs other portfolios), and any concrete validation thresholds — these are inferred from the title and have not been confirmed against the paper’s text.
Topics this feeds
Open questions raised
- What specific backtesting methodology and tolerance thresholds the staff paper proposes.
- Whether the approach signals a future supervisory expectation (e.g. a Guideline) on PD-model backtesting, or remains staff-level analysis.
- How the regulatory-data backtesting approach interacts with firms’ existing IRB validation cycles.
Ingestion note
Identified via direct WebFetch of the EBA publications page (https://www.eba.europa.eu/publications-and-media/publications), where it was the most governance / model-risk relevant item not already in the wiki; the page’s newest entries at scan time (2026-06-01) were dated late April 2026. The PD-backtesting staff paper and the 29 April Supervisory Independence final report were already referenced on the EBA — European Banking Authority entity page without an S-tag — this page back-fills the citation for the staff paper. Confirm methodology and findings against the published PDF before relying on this page for client deliverables.