Criteria for the Effectiveness of Shadow Rating Models in Assessing the Creditworthiness of Low-Default Borrowers

Abstract

Assessment of borrowers' creditworthiness is the most important process affecting the activities of a modern commercial bank. Creditworthiness assessment processes occur both at the stage of decision-making to issue a credit product and during the process of regular creditworthiness assessment for the purposes of reserving and calculating economic capital. This is the reason why the commercial bank needs to develop and maintain the effective models of credit rating estimation, which are able to determine the borrower's solvency accurately and steadily by predicting its probability of default. This examines with the problem of determining the criteria for the effectiveness of shadow rating models used to estimate the probability of default of low-default segments of bank lending. Shadow rating models can be used both for business purposes and for regulatory purposes. Depending on the goal set, a number of problems specific to this class of models arise at each stage of shadow rating model development, which form the basis for the definition of performance criteria: correct specification of data samples, harmonization of rating agencies' assessments, correct choice of calculation algorithm, satisfaction of quantitative validation criteria and validity of expert corrections. Compliance with these criteria, taking into account the established objective, allows us to conclude on the effectiveness of the obtained model.