Testing Measures for Degree of PIT-ness in PD models
(2024) EXTM10 20241Department of Economics
- Abstract
- This thesis explores the Point-in-Time (PIT) in probability of default (PD) models, focusing on evaluating the degree of PIT-ness. Unlike previous studies that developed new measures for PIT characteristics, this research compares existing measures and frameworks to identify unique challenges and opportunities. By analyzing and testing existing measures for 1-year PD models, the study offers new insights into the validation of PD models, which is essential for financial institutions like Swedbank. The research employs both historical default data and generated data to assess PIT measures. Using generated data allows for the creation of controlled scenarios with varying economic conditions, providing a detailed investigation into how PIT... (More)
- This thesis explores the Point-in-Time (PIT) in probability of default (PD) models, focusing on evaluating the degree of PIT-ness. Unlike previous studies that developed new measures for PIT characteristics, this research compares existing measures and frameworks to identify unique challenges and opportunities. By analyzing and testing existing measures for 1-year PD models, the study offers new insights into the validation of PD models, which is essential for financial institutions like Swedbank. The research employs both historical default data and generated data to assess PIT measures. Using generated data allows for the creation of controlled scenarios with varying economic conditions, providing a detailed investigation into how PIT measures respond to financial fluctuations. This approach addresses the limitations of relying solely on historical data, capturing the unpredictability of financial markets. Key findings highlight the limitations of existing measures, particularly those proposed by Carlehed and Petrov, the European Banking Authority (EBA), and the Prudential Regulation Authority (PRA). The study underscores the importance of precise estimation of correlation parameters and careful consideration of credit cycle forecasts. It suggests that models with lower degrees of PIT-ness are more stable and reliable, emphasizing the need for consistent application of PIT measures across different data aggregation levels. The primary objectives of this research are to provide a comprehensive overview of existing measures for assessing PIT-ness, generate market conditions to test these measures, and evaluate their effectiveness. The findings aim to enhance Swedbank’s validation of PD models, improve credit risk management practices, and ensure regulatory compliance, thereby contributing to the broader field of credit risk management. (Less)
Please use this url to cite or link to this publication:
http://lup.lub.lu.se/student-papers/record/9165435
- author
- Sandström, David LU and Zachau, Hobbe LU
- supervisor
- organization
- course
- EXTM10 20241
- year
- 2024
- type
- H2 - Master's Degree (Two Years)
- subject
- keywords
- Probability of default, Point-in-Time, Through-the-Cycle, Credit cycle, Actual default frequency, Default correlation, Rating grades
- language
- English
- id
- 9165435
- date added to LUP
- 2024-11-22 09:05:30
- date last changed
- 2024-11-22 09:05:30
@misc{9165435, abstract = {{This thesis explores the Point-in-Time (PIT) in probability of default (PD) models, focusing on evaluating the degree of PIT-ness. Unlike previous studies that developed new measures for PIT characteristics, this research compares existing measures and frameworks to identify unique challenges and opportunities. By analyzing and testing existing measures for 1-year PD models, the study offers new insights into the validation of PD models, which is essential for financial institutions like Swedbank. The research employs both historical default data and generated data to assess PIT measures. Using generated data allows for the creation of controlled scenarios with varying economic conditions, providing a detailed investigation into how PIT measures respond to financial fluctuations. This approach addresses the limitations of relying solely on historical data, capturing the unpredictability of financial markets. Key findings highlight the limitations of existing measures, particularly those proposed by Carlehed and Petrov, the European Banking Authority (EBA), and the Prudential Regulation Authority (PRA). The study underscores the importance of precise estimation of correlation parameters and careful consideration of credit cycle forecasts. It suggests that models with lower degrees of PIT-ness are more stable and reliable, emphasizing the need for consistent application of PIT measures across different data aggregation levels. The primary objectives of this research are to provide a comprehensive overview of existing measures for assessing PIT-ness, generate market conditions to test these measures, and evaluate their effectiveness. The findings aim to enhance Swedbank’s validation of PD models, improve credit risk management practices, and ensure regulatory compliance, thereby contributing to the broader field of credit risk management.}}, author = {{Sandström, David and Zachau, Hobbe}}, language = {{eng}}, note = {{Student Paper}}, title = {{Testing Measures for Degree of PIT-ness in PD models}}, year = {{2024}}, }