Rating / Scoring expert speaks

Rating / Scoring expert speaks


The key question of traditional commercial banking is that who should be provided with a loan or any other credit facility? Theoretically the answer is quite obvious: anyone who will pay back the loan increased by the related fees, commissions and interest amounts that cover all the cost elements (refinancing of credit, operational and risk costs) plus a margin which makes the deal profitable for the Bank (i.e. generates an economic value added). The problem is that in advance the Banks do not know for sure whether the client would be able to pay according to the contractual terms. This uncertainty is one of the major source of credit risk (besides collateral and exposure risk).

To solve this ex-ante binary (dichotomous) classification problem the decision maker needs a risk classification tool. This tool is called credit scoring or rating. The scoring or rating category given to a potential client is strongly related to the Probability of Default (PD). The simplest decision rule is that if the customer’s PD is below a certain (profitability based) threshold than the Bank should classify him as a defaulter and reject its credit application. A more intelligent decision making rule – combined with the simple threshold based approval – is when Banks apply risk (credit quality and collateral) sensitive margins which cover the expected loss of the customer. PD is one of the most essential risk factor for sophisticated loan approval, risk-based pricing and credit portfolio modeling.

The determination of the rating category or the scoring value of a given customer could be based on expert judgment or statistical models. The most wide-spread statistical models are the linear regression, probit or logit models. The set of possible explanatory variables of the models strongly depend on the asset segment of the client (e.g. financial institution, sovereign, retail, corporate, etc.) and the length of the future time horizon of the risk assessment (point-in-time vs. through-the-cycle). The model development is based on historical databases including information on the default events as well (besides the obligor, transaction and macro-economy specific explanatory variables). The discriminatory power of the models should be continuously monitored and if necessary the existing model should be replaced with a new stronger one.

The solution of Loxon is a flexible tool for supporting the whole credit assessment and model management process (data handling, setting and maintenance of calculation algorithms). The tool could easily be integrated with other IT systems to integrate the rating/scoring calculation results (the assessment categories) into the everyday business decision making processes of the Bank.

About Loxon Products

    Email US
    Call US +36 1 789 0626