Loss Given Default (LGD) modelling

For a large Asian Bank


  • The Client is one of the universal banks.
  • It has strong customer base in the region.
  • The Client wanted to identify the estimated potential credit losses, so we calculated loan’s projected profitability.
  • Under IFRS 9, banks are required to recognize credit losses at all times based on reasonable information, considering past, current and future events.
  • Models for expected credit loss (ECL) were to be developed using the past information and performance of existing and former clients.


  • Identified data sources, resolved issues, cleaned and generated the base dataset for the model development.
  • We identified the optimal workout period for recoverable loans with precision and accuracy.
  • We used K-Means clustering, an unsupervised machine learning algorithm, to form cluster centroids and group the data points by minimizing their distances from the centroids.
  • Obtained LGD for specific customer segments using either data-driven clustering or business logic.


  • Accurately projected future LGD with macroeconomic adjustments.
  • Calculated LGD in ECL calculations and loss mitigation strategies for more reliable estimates.