Workshop on Credit Risk Analytics with SAS Applications

Part I: Data Pre-processing and Credit Scoring

• Corporate & retail data
• Outlier detection
  
• Standardisation & categorisation
• Default definitions & roll rate analysis
• Credit scoring techniques

---

Part II: Probabilities of default (PD) - Part I

• Discrete time models (Logit & Probit)
  
• Maximum likelihood estimation
  
• Non-monotone relationships
  
• Calibration & discrimination
• Through-the-cycle & point-in-time

---

Part III: Low default portfolios

• Varying sample windows
  
• Undersampling & oversampling
  
• SMOTE analysis
  
• Adjusting posterior probabilities
  
• Cost sensitive learning
  
• Mapping to agency ratings
  
• Confidence level approach
  
• LGD & EAD for low default portfolios
  

---

Part IV: Probabilities of default (PD) - Part II

• Estimation of rating migration probabilities
  
• Fitting & forecasting
  
• Formation of rating classes
  
• Stress testing
  
• Continuous time hazard models
  

---

Part V: Default correlations and credit portfolio risk

• Basel asset correlations
  
• Empirical correlations
  
• Analytical, numerical & simulation based solutions for
  
• Value-at-Risk
  
• Method of Moments estimation
  
• Maximum Likelihood estimation
  

---

Part VI: Loss give default (LGD) and recovery rates

• Computing observed LGD
  
• LGD discount rates
  
• LGD modeling
  
• Marginal modeling (linear, fractional Logit & beta regressions)
  
• PD-LGD models (Tobit, Heckman, Censored beta)
  
• Random effect models
  
  

---

Part VII: Exposure at default (EAD) and adverse selection

• Conversion measures
  
• Credit lines & flexible repayment schedules
  
• PD-EAD models & adverse selection
  
• Payoff probabilities
  

---

Part VIII: Model validation

• Validation framework
  
• Backtesting PDs
  
• Brier scores & Hosmer-Lemeshow
  
• Binomial tests with correlations
  
• ROC curves, AUROC & accuracy ratio
  
• Backtesting LGD and EAD
  

---

RETURN