Interpretation of the term "all observed defaults" in Article 500 of Regulation (EU) No 575/2013 as amended

Both Article 181(1)(a) and Article 500(1)(c) of Regulation (EU) No 575/2013 as amended contain the expression “all observed defaults”. Is that expression meant to be interpreted in the same way in the context of both articles to include both completed and incomplete recovery processes? It is our understanding of Article 181(1)(a) that no observed defaults should be excluded. This understanding is further supported by paragraph 163 of EBA/GL/2017/16, which states that for the purpose of LGD quantification, “institutions should not exclude any defaults in the historical observation period that fall within the scope of application of the LGD model”. Should this view also be extended to the calculation of the threshold under Article 500(1)(c) of Regulation (EU) No 575/2013 as amended, and can it be in particular be deduced that neither completed nor incomplete recovery processes can be excluded?

One of the conditions for the application of Article 500(1) of Regulation (EU) No 575/2013 is that “the cumulative amount of defaulted exposures disposed of since the first date for disposals in accordance with the plan referred to in point (a) has surpassed 20 % of the cumulative amount of all observed defaults as of the date of the first disposal referred to in points (a) and (b).” The question relates to the interpretation of “all observed defaults”. Some readers of Article 500 consider that institutions should be able to apply the article as soon as 20% of all open defaults are covered by the disposal. They are of the view that completed recovery processes can be excluded from the calculation. In our view, this interpretation is not correct: - All observed defaults, in the context of Article 181(1)(a) of Regulation (EU) No 575/2013, refers to all completed and incomplete recovery cases. Given that Article 500 further specifies Article 181(1)(a), the same term, used in two closely connected articles, should have the same meanings. - The interpretation of “all observed defaults” to include all completed and incomplete workouts is more logical and in line with the legislative intent. The intention seems to have been to ensure that the LGD adjustments is only applied in cases in which the adjustment is likely to have a significant impact on LGD estimation since at least 20% of exposures in the reference data set are affected. - From the point of view of LGD estimation, the interpretation of all observed defaults as only including all incomplete workouts makes less sense. Imagine an institution has a large dataset of completed recovery processes, but very few incomplete recovery processes. Then the disposal of a few incomplete recovery processes could be considered a massive disposal even though it would have a big impact on neither the balance sheet of the bank nor the reference data set used for LGD estimation.