EBA · 2019_4495 Rejected question

Use of the last available data for risk quantification sample and out-of-time validation sample

Regulation
Regulation (EU) No 575/2013 (CRR)
Article
175, 179, para. 4,1
Topic
Credit risk
Submitted by
Credit institution
Submitted
2019-01-31

Question

Given the requirements of articles 175(4)(b) and 179(1)(a) , in case of a model development, should the last available one-year snapshot be used for risk quantification purposes (i.e. for the computation of the Long-run average default rate) or be set aside for validation tests?

Background

Articles 179(1)(a) requires that “an institution's own estimates of the risk parameters PD, LGD, conversion factor and EL shall incorporate all relevant data, information and methods” on the other hand article 175(4)(b) requires out-of-time and out-of-sample performance tests for validating the model. While in the risk differentiation phase the presence of an out-of-time validation sample can be guaranteed, in the context of initial validation, it is not clear if the last available snapshot should be used for calibration purposes (i.e. should be part of both long run average default rate computation sample and calibration sample) or if it should be set aside for the out-of-time performance test. Moreover it is not clear whether a partial overlap of the validation and calibration sample would satisfy the requirement of article 175(4)(b).
No answer published yet.

Original source: European Banking Authority, Q&A ID 2019_4495

This Q&A is published by European Banking Authority and is non-binding. It does not constitute legal advice. Updated weekly from official ESA sources.

Similar Q&As

More Q&As on this topic

📋 Track EU financial regulation continuously

Forseti monitors EU financial regulation and delivers personalised alerts anchored to verified official sources.

14-day free trial. No credit card required.