Content

This introductory course provides the context of regulatory requirements with regard to the estimation of the PD risk parameter and treats the main challenges for developing and validating PD models.

The participants will learn to think critically about the customary building blocks of PD estimation: identifying data requirements, selecting risk drivers, building PD scorecards, PD quantification, as well as model performance testing.

Learning Goals

• understand each of the main building blocks of PD model development
• identify important regulatory requirements and critical decisions taken during PD model development
• grasp the concepts rating and calibration philosophy and implications for the modelling approach.
• be familiar with common approaches to build a PD ranking model and spot common pitfalls
• gain a basic understanding of the Margin of Conservatism for PD
• learn the key considerations and approaches to calibrate PD models
• understand the backtesting approaches for PD models and related regulatory requirements.

Target Audience

• IRB model developers with little or no prior experience in the field
• Specialist supervisors wishing to gain insight in internal model development

Prerequisites

The material will be taught in English.
Participants are advised to come equipped with basic understanding of:

• (Credit) risk management
• Statistics and probability theory

Some experience in programming with python or a similar statistical computing language is recommended for the case studies.

Schedule

The module will be taught over2 sessions of each 4 hours. The track schedule will be planned in coordination with the client based on the selection of modules.

Example case studies

For this module, examples of case studies are

• Compare and identify material weaknesses of alternative proposed PD estimation approaches
• Present and challenge arguments whether or not a proposed treatment of PD calibration is compliant with regulation
• Work out an outline of a minimum viable approach for estimating PD values
• Compare and understand different treatments of PD calibration based on prepared code samples and realistic data sets

Please Contact us for details and options.