PD and LGD Estimation

The IRB Approach in the Basel II/III capital accord gives the bank higher flexibility in terms of determination of credit risk capital requirements. If applied appropriately, the IRB approach leads to optimized capital requirements and thus to higher Return On Equity.

  1. Derive appropriate segmentation of the portfolio into risk classes.
  2. Data mining within each segment, as well as between related segments.
  3. Drafting of risk management methodologies under the Basel IRB framework.
  4. Calculation of risk weights using the new methodologies on sample, representative assets from the bank's portfolio.
  5. Selection of appropriate risk models with a view towards lower and more robust credit risk capital requirements.
The following table summarizes the deliverables during the IRB implementation
Deliverable Methodology
Portfolio segmentation Application of qualitative and quantitative criteria. Strong emphasis on the data availablity and the data quality.
Rating / scoring system Data mining activities like for example:
  • Bayesian analysis for missing or incomplete data.
  • Identification, analysis and smoothing of data outliers.
  • Calculation of various statistical performance measures to assure homogenuity within a rating class.
  • Benchmarking of results to established industry ratios.
Probability of Default (PD) Estimation One or a combination of methods like:
  • past due / roll rate analysis (e.g. for retail and SME loans)
  • rating migrations analysis (e.g. for bonds)
  • credit scoring model (e.g. for corporate financing)
Loss Given Default (LGD) Estimation One or a combination of methods like:
  • collateral assessment (e.g. for mortgages)
  • macroeconomic regression (e.g. for business loans)
  • recovery cash flow analysis (e.g. for retail loans)
  • prices of defaulted debt (e.g. for bond-like instruments)
Calculations of credit risk weights As early as possible during the project, we will evaluate several alternative models with respect to their:
  • Statistical validity (goodness-of-fit, stability, standard error), AND
  • Impact on capital requirements (through risk weights).
The model selection thus will be based on both theoretical and practical considerations.
Quality Assurance of Implementation RiSK-VBox includes a set of best-practice templates for IRB implementation. These templates are adapted to the specific situation of the bank.