1. User Guide
  2. Segmentation Run

9 Segmentation Run

Segmentation Run of OFS HM is one of the foremost runs that must be executed during the process of computing various parameters required under IFRS 9. However, the use of segments (execution of the Segmentation run) is not mandatory. If you do not want to use features that require segments (for example, Historical Transition Matrix, PD Model, and so on).

Segmentation is the process of grouping accounts into homogeneous clusters using various dimensions as parameters. Segments or Portfolios help in simplifying the process of treating or evaluating accounts by treating accounts belonging to one homogeneous group in the same manner.

The Segmentation Run of HM can be used to create such Segments or Portfolios, wherein every account is mapped to a specific Segment. The first step in the Segmentation process is to create a Segment Type and the master list of Segments that you want to map each of the accounts to, using the AMHM window s. The standard product has a seeded Segment Type - IFRS9 - ECL and corresponding Segments. You can choose to use the existing Segments or create a new Segment type and corresponding child Segments.

After the master list of Segments is available, the second step is to configure Rules that assign a Segment to an account taking into consideration various dimensions, using the flexible (configurable) Run Rules Framework. The standard product considers two dimensions to assign a Segment to an account. These are Product Type and Customer Type. As required, more Dimensions can be considered by reconfiguring the dataset and the Rule, such as Industry, Region, Branch, Country, and so on.

The output of HM’s Segmentation Run is stored in a specific table where for any given date only one set of Account-Segment mapping, the final one, is stored.

This mapping is then considered for further processing in both the OFSAA's Hedge Management and Loan Loss Forecasting and Provisioning applications.

Segmentation is mandatory for the functioning of the following features:

  • Historical Transition Matrix
  • Historical Loss Rates
  • Inbuilt PD Model
  • ECL Computation using Roll Rate Methodology