PD-LGD correlation study: Evidence from the Russian corporate bond market

Probability of Default Model to Estimate Ex Ante Credit Risk

“…There are models for Russian corporates, such as those presented by Dwyer et al (2010), Totmyanina (2014), Surzhko (2014), Karminskiy (2015, pp. 217-222), Ermolova and Penikas (2017), and Mogilat (2015.…”

“…Technical details for it are available in Korableva (2011). We could have chosen alternative models from Beaver (1966), Altman (1968), Ermolova and Penikas (2017), or Mogilat (2019). However, we wished to benchmark against this particular Moody's model due to its simplicity.…”

“…1) We use a novel dataset that allows us to introduce the prudentially compliant definition of default of 90 days past due used in the Basel II Internal-Ratings-Based (IRB) approach. Korableva (2011) and Ermolova and Penikas (2017) use the same definition of default. However, Korableva (2011) uses an outdated sample from 2002-2009, while we use recent data from 2017-2020.…”

“…However, Korableva (2011) uses an outdated sample from 2002-2009, while we use recent data from 2017-2020. Ermolova and Penikas (2017) study only a subsample of the largest borrowers. Mogilat (2019) covers the same data by the available firm identifiers (and definitely not by time coverage) as we do (only earlier sample), but she uses a non-compliant definition of default, as she looks at cases of bankruptcy only.…”

“…2) We confirm that using delinquency data helps to improve default predictions for Russian data compared to prudential reserve ratios. Neither Russian researchers, such as Korableva (2011), Ermolova and Penikas (2017), or Mogilat (2019), nor foreign researchers, such as Dell ' Ariccia et al (2017) or Paligorova and Santos (2017), have made such comparisons using Russian data.…”