Development of a Methodology for the Supervisory Authority's Assessment of the Adequacy of the Amount of Expected Credit Losses Calculated by Commercial Banks

The article examines the development of an algorithm for assessing the adequacy of the expected credit losses (ECL), calculated by commercial banks from the point of view of the Central Bank of the Russian Federation as the supervisory authority of the banking system of Russia. For this purpose, the comparison between the modeling of potential losses for the loan portfolio of individual’s portfolio of homogeneous loans with the bank's actually created reserves for possible loan losses was made. There are identified separate independent components: PD, EAD, LGD, while ECL was modelled, each of which is predicted taking into account the characteristics of the indicator and the register of data provided to the Bank of Russia from credit organizations. The article concludes with recommendations on the preparation of the final report on the adequacy of the reservation by a commercial bank of the analyzed segment of the loan portfolio. The purpose of the study is to develop an algorithm for assessing the sufficiency of the value of the ECL calculated by commercial banks from the point of view of the Bank of Russia as the supervisory authority of the banking system of the Russian Federation. To achieve the goal, the following tasks were solved in the work: 1) The main components of the ECL are investigated; 2) Mathematical modeling of credit risk attributes is carried out taking into account the specifics of the Bank of Russia's activities; 3) The results of the model are interpreted from the point of view of the supervisory authority. Materials and Methods. Educational literature and scientific publications were reviewed for the analysis of ECL, which revealed theoretical approaches and practical aspects to the construction of PD, EAD, LGD models based on statistical algorithms and machine learning methods. Conclusions: the study of the main components of ECL and their impact on the final value of credit losses was conducted; a model for calculating ECL was made using various machine learning algorithms; was made the interpretation of the results of the ECL model from the point of view of practical application in the Bank of Russia.

Expected credit losses (ECL), the Bank of Russia, portfolio of homogeneous loans, Probability of default (PD), Exposure at default (EAD), Loss given default (LGD), reserves for possible loan losses