Top-down and Performance-based SME Probabilities of Default

Note available here.

This note summaries the results of an exercise to estimate probabilities of default (PDs) using the Top-Down approach (commonly employed by IRB banks) and an approach that we term “Performance Based” since it relies on the arrears status of loans as the basis for prediction of default.

The interest of this exercise stems from the fact that banks are currently considering, particularly in Europe, whether and how they might access the SEC-IRBA as an approach for calculating capital for securitisation exposures. To do this requires that they devise an approach to estimating PDs for securitisation pool loans in a way that satisfies regulators. Given pool PD estimates, they can calculate Basel on-balance sheet capital for the securitisation pool, the key input to the SEC-IRBA.

The European Banking Authority (EBA) has recently published Regulatory Technical Standards (RTS) clarifying how banks may implement the SEC-IRBA using an approach that already appears in the Basel framework and in the Capital Requirements Regulation (CRR) which translates Basel rules into European law. This approached, termed the Purchased Receivables Approach or PuRA allows banks to employ proxy data and to apply a “retail standards” approach even when loans are corporate.

In this note, we implement a “Top Down” PD estimation approach for Small and Medium Enterprise (SME) loan included in the securitisation pools of a single bank, Caixabank. “Top Down” is the term used in the Basel documents which corresponds to what in the CRR is labelled “retail standards”. The approach consists of calculating PDs for loans buckets into different categories. When banks implement this approach, categories are typically defined based on indicators such as region, vintage, bank internal rating and loan product type.

We compare the results of the Top Down estimation with an alternative that uses pool performance data. Performance data records what fraction of a loan pool fall into categories based on whether they are performing or are in different ranges of number of days in arrears. Such data are widely available because they are commonly included in securitisation investor reports and are often emphasised in ratings agency evaluations of pool credit quality. It is not clear from the EBA’s RTS on PuRA whether regulators will approve banks’ use of performance data in the estimation of PDs for SEC-IRBA implementation. This note helps to clarify that subject to careful implementation, performance-data-based modelling yields results comparable to Top Down modelling.