Validating the market risk measurement
Only quantitative validation can establish confidence in an internal risk model and the necessary acceptance within your institution as well as from auditors, regulators and rating agencies. The most important instrument for validating market risk figures is day-to-day backtesting. In an internal model backtesting also determines the amount of regulatory capital through the "add-on factor".
On the basis of our many years of experience in introducing internal models for market risk, we know the deciding factors for the quality of a risk model, and are familiar with validation procedures.
Backtesting
The quality of the model is monitored regularly by backtesting. The creation and implementation of an adequate backtesting process is demanding and, based on our experience, should be undertaken early in the process of introducing a market risk model. We have devised backtesting processes and have gathered practical experience in the following areas:
- clean P&L calculation (also referred to as no-action P&L or hypothetical P&L), which serves as the basis for backtesting
- backtesting processes that are mark-to-market compliant and particularly backtesting for partial VaR components
- taking into account ageing effects in backtesting, consistent with VaR calculation
- backtesting implementations for heterogeneous front-office system architectures
- reconciliation with drill-down capabilities for comparison of economic P&L and clean P&L
Validating the risk model
The validation process must constantly monitor all aspects that could significantly affect the model's forecast quality. This includes, in addition to monitoring the actual risk model, constant evaluation of the valuation models, market data and instrument data used in the model. The following topics give an outline of the scope of our recent projects in this field:
- performing portfolio analyses in order to check whether the selected risk factors adequately cover the market parameters that affect prices, e.g. based on sensitivity analyses or by comparing expected and realized P&L
- developing and implementing portfolio analysis procedures for the regular performance of adequacy analyses, including drill-down for sensitivities and stress tests
- validating the risk model for portfolios of complex financial instruments in which calibration processes have to be taken into account and stable sensitivities are often difficult to determine
- performing time series analyses in order to prove that selected stochastic models are appropriate for individual risk factors
- performing statistical analyses for validating correlation assumptions
Validating the valuation models
The quality of the market risk figures obtained from the model is highly dependent on the valuation models employed in the risk calculation. Thus, using appropriate valuation procedures for all financial instruments is a prerequisite for adequate market risk measurement. For validation of these procedures, we draw on the resources of our own valuation library as well as a team of financial engineering specialists.
Data quality
The best model can not adequately determine risk figures if its input data is of poor quality. Thus, ensuring high data quality is a crucial issue. We support you in checking your data sources and assessing the quality of data obtained.
We assist in selecting the appropriate process to validate, design and reliably implement your risk model from the large array of available methods and options.