In addition to the above, other qualitative factors that are showing early signs of warning are considered. The ECL calculation ECLs are computed as unbiased, probability weighted amounts which are determined by evaluating a range of reasonably possible outcomes, the time value of money, and considering all reasonable and supportable information, including that which is forward-looking. It also incorporates how defaulted loans are expected to be recovered, including the probability that the loans will cure and the value of collateral or the amount that might be received for selling the asset. The maximum period for which the credit losses are determined is the contractual life of a financial instrument, unless the Company has the legal right to call it earlier. The ECL is the product of PD, LGD and EAD. These parameters are derived and are adjusted to reflect forward-looking information as described below: The probability of default (‘PD’) refers to the likelihood that a borrower will default over a particular time horizon. The PD of an obligor is a fundamental risk parameter in credit risk analysis and depends on obligor-specific characteristics as well as on macroeconomic risk factors. The PD models are derived using logistic regression and Vasicek approach to model monthly default rates. As part of the modelling phase, the variables having the most significant predictive default power were identified using the information value statistics. Variables were shortlisted based on their significance in predictive default and possible combinations were assessed using multivariate analysis to achieve the best-fit model. The performance of the final models was assessed to test the fit of the estimated PD curves against the historical default rate. The PD term structures have been updated in 2023 based on latest available macroeconomic information and any data from external sources as required in the PD model framework. For the different segments, different features including macroeconomic variables have been chosen for inclusion in the logit models based on their statistical significance in explaining defaults as well as intuitiveness of the coefficients. By definition, loss given default (‘LGD’) refers to the magnitude of the likely loss on a given facility in the event of default. It takes into account the loss of principal, interest foregone and workout expenses. The LGD estimation for Cim has been updated with additional data for the recent period. The LGD estimates include the following key elements within the methodology: (i) Cure Rate, (ii) Recovery Rate, (iii) Discounting Rate, (iv) Administration Cost. The models were derived using the logistic regression technique and yielded to statistically significant estimates. Where historical data was insufficient for modelling, Basel estimates of LGD for unsecured exposures were applied. Risk Management Report Continued MUR MUR 74 CIM FINANCE ANNUAL REPORT
RkJQdWJsaXNoZXIy MzQ3MjQ5