Stress testing: CCAR and EU-wide
Banks have to comply with increasing challenges to meet stress testing requirements from the supervisors. In the US, banks have to deliver their inputs twice a year to the Fed and in Europe banks need to prepare for the 2018 EU-wide stress test.
The approaches to stress testing are very different from the established through-the-cycle (TTC) model approach of banks to feed the IRB regulatory capital calculations and are much more data intensive.
LoanPilot™ Stress Testing
OSIS offer different solution options for stress testing. There are three, different, broadly-recognised, types of stress testing: frequentist, deterministic and Bayesian which each offer something unique.
Stress testing is the process of finding the effect of a change to a portfolio (or any group of loans) due to certain events. There are various stress tests for credit risk, e.g. PD for individual counterparty/sector, LGD, exposure estimate, credit spread, portfolio capital e.g. concentrations, correlations, etc.
Organisations need to define their process for stress testing which should include objectives and uses of outputs, frequency that scenarios are reviewed, use of static and / or bespoke scenarios and review their strategies for adjusting portfolios (if required). The results of stress events can be seen through several results: change in expected loss or VaR (Value at Risk), expected shortfall, sensitivity and stressed level of PD/LGD, change in rating of portfolio.
All of these requirements means organisations need a solid technological infrastructure to support effective assessment and decision making.
Our modeling suit follows a Bayesian statistical approach and treats certain model parameters as uncertain. During calibration, the user can either “let the data talk” with minimal subjective input or use informed ‘priors’. LoanPilot™ provides a point-in-time (PIT) risk assessment where long term through-the-cycle (TTC) parameters can be specified as input. The key outputs are credit risk performance measures (e.g. PD, LGD, EAD, loan loss provisions and economic capital) over a multi-year horizon conditional upon macro economic scenarios. The user has access to all model details and, importantly, all data sets used for calibration and additional expert judgment (if applied).
All OSIS tools are available online and accessible through desktop or mobile devices.