A FORWARD-LOOKING MACROECONOMIC SCALAR FOR IFRS 9 PD IN DEVELOPING ECONOMIES

Abstract

With the introduction of an anticipated credit loss (ECL) framework to account for impairments, IFRS 9 regulates the reporting of financial assets and liabilities. At every reporting date, the ECL model updates recognized ECLs by taking into account historical data, present circumstances, and anticipated information. Usually, ECL is divided into three parts: exposure at default (EAD), loss given default (LGD), and probability of default (PD). In order to improve accuracy, this research suggests a way for creating a macroeconomic scalar to modify PD in the ECL model by including forward-looking data. Datasets from Cameroon and Nigeria are utilized to demonstrate the suggested methodology, and different modeling approaches are employed for validation. Five steps make up the methodology: (1) planning and research, (2) data preparation, (3) model construction, (4) macroeconomic scalar computation, and (5) model validation. Regression, feed forward neural networks, random forest, gradient boosting, and GLM (Logit, Probit) are some of the methods used. For the Nigerian and Cameroonian datasets, the macroeconomic scalar that was constructed successfully adjusted PD within the ECL model. Every modeling strategy offered insightful information, proving the scalar's capacity to enhance ECL forecasts. A reliable technique for adding forward-looking data to the ECL model is to incorporate a macroeconomic scalar. This improves PD accuracy and can take into account the uncertainty, volatility, and sparse data that are characteristics of developing countries.In this study, a scalable method for modifying PD in ECL models with macroeconomic data is presented, with a focus on developing nations.