New date: 11-12 August
This two day introductory workshop provides a full overview of all credit scoring related concepts, from model development to validation and stress testing, and is ideal for newly hired credit risk personnel or credit risk professionals wishing to develop their Basel II credit risk modeling skills. Attendance will provide access to approaches designed to add practical value in both regulatory and business contexts. Attendees will gain practical insights into recent developments and emerging practice being employed within leading banks and other consumer lending institutions.
The key objectives are:
· Learn how to develop PD, LGD and EAD models
· Providing insights into model monitoring using both backtesting and benchmarking procedures
· Discuss the importance of qualitative model validation
· Discuss how to stress test credit risk models using both sensitivity and scenario analysis
· Provide new insights into dealing with low default portfolios
Next to the theoretical concepts and case studies, software demos on both Matlab and SAS for credit scoring will be given:
| Developing a Credit VaR System with MATLAB | During this presentation, it will be demonstrated how MATLAB can be used to develop a model to compute the credit VaR of a portfolio of bonds. Since a typical end user of such a system is not a MATLAB developer, the demonstration will also show how the MATLAB algorithms can be integrated within an Excel front end. This Excel application can then be shared with users that don't have access to MATLAB. The computation of the credit VaR requires a robust credit rating system and an algorithm to compute transition probabilities. It will be shown how the statistical tools can be used to develop these algorithms. |
| Using SAS and SAS Enterprise Miner for scorecard construction | In this demo, we will illustrate how to construct a scorecard using SAS and SAS Enterprise Miner. Using real-life application scoring data, we will illustrate how to do coarse classification, weights of evidence coding, scorecard construction and alignment. |
| Date: | New date: 11-12 August 2010 |
| Location: | Novotel Brussels Airport  |
| Leonardo Da Vincilaan 25 | |
| B-1831 Diegem | |
| Registration fee: |
Including lunch and VAT |
| 900 Euro for early registration (before August 1st) | |
| 1.200 Euro for late registration | |
| 400 Euro academic fee (for full-time students and academics) |
| Introduction to credit scoring | |
| prof. dr. ir. David Martens | |
| Lecturer and Risk Management Consultant | |
| MarFintel, Hogeschool Gent, K.U.Leuven
David is lecturer at Hogeschool Gent and K.U.Leuven, and obtained a Master degree in civil engineering and a Ph.D. in Applied Economics from K.U.Leuven in 2003 and 2008 respectively. He also obtained an MBA from Reims Management School in 2005. Since January 2010 he works as visiting researcher at New York University (Stern School of Business) on the topic of networked data for online advertizing. David has extensive expertience as data analysis and risk management consultant at major Belgian banks and a variety of retail companies. He regularly teaches on the topic of business intelligence and data mining to both master students and professionals. |
|
| prof. dr. Bart Baesens | |
| Professor Risk Management | |
| K.U.Leuven, University of Southampton
Prof. Dr. Bart Baesens is an associate professor at K.U.Leuven (Belgium), and a lecturer at the University of Southampton (United Kingdom). He has done extensive research on credit risk management, Basel II and predictive analytics. His findings have been published in well-known international journals (e.g. Management Science ,Machine Learning, IEEE Transactions on Neural Networks, IEEE Transactions on Knowledge and Data Engineering, …) and presented at international top conferences. He is also co-author of the book Credit Risk Management: Basic Concepts, published in 2008. He regularly tutors, advices and provides consulting support to international firms with respect to their credit risk management strategy. |
| Program · Recap of basic credit risk nomenclature o Credit scoring: application scoring, behavioural scoring, profit scoring o Bankruptcy prediction o Credit Ratings and Rating agencies · Basel I and Basel II o Basel I and Basel II regulation o PD versus LGD versus EAD o Merton/Vasicek model for calculating regulatory capital o A layered credit risk model architecture (data/scorecard/calibration) · Developing PD models for Basel II o Data preprocessing: sampling, outlier handling, missing values, data transformations, categorization, non-linear modeling o Reject inference o Classification techniques for PD modeling: logistic regression, decision trees o Input selection § Filters § Stepwise regression approaches o Measuring scorecard performance § Out-of-Sample/Out-of-time/Out-of-Universe validation § Gini coefficients, Area under ROC curve, Kolmogorov-Smirnov statistic, … o Scorecard implementation o Defining default ratings and PD calibration · Developing LGD and EAD models for Basel II o Defining LGD § Workout/market/implied market/implied historical LGD § Length of the workout period and incomplete workouts § Discount factor § Indirect costs § Loss drivers o Defining EAD § Credit conversion factors § Cohort/fixed time horizon/variable time horizon/momentum method § Exposure drivers o Modeling approaches § Linear/Beta regression § Regression trees § Mixture models § Two stage models o Performance metrics for LGD/EAD § R-squared, Gini coefficients, … o Calibration issues § defining ratings § economic downturn calibration · Validation, Backtesting and Stress testing o Quantitative validation: Backtesting versus Benchmarking o Qualitative validation: use test, model design, data quality, … o Backtesting PD, LGD and EAD models o Backtesting statistics (binomial, Vasicek, Chi-squared test, t-test, …) o Traffic light indicator approach o Backtesting action plans o Benchmarking (internal, external, champion-challenger, …) o Stress testing PD, LGD and EAD models o Sensitivity versus Scenario analysis · Low default portfolios (LDPs): Implementation and Validation o Definition of a low default portfolio o Mapping approaches o Sampling approaches o Likelihood methods · Demos from industry specific solution providers |