Post-conference workshop 4, March 28, New York

WORKSHOP 4: From loss data to capital: Implementing a comprehensive operational risk capital estimation framework under the AMA-LDA

Major components, range of methodological practices, putting it all together, remaining challenges, and critical un(der)addressed issues

Led by: J.D. Opdyke, Quantitative Methods Group, Team Leader-Operational Risk Portfolio Analytics, GE CAPITAL

8:30 Registration and breakfast

9:00 Getting the goals right: more than half the battle
Objectives and overview of the framework, key issues re: operational risk loss data, unit-of-measure (UoM) definition

10:30 Morning coffee break

11:00 It's all about the data
Internal loss data, external loss data, scenario analysis data, and Key Risk Indicator data - statistical and econometric methods for combining, comparing, and utilizing to directly estimate capital in causal models

12:30 Lunch

1:30 Severity, frequency, and aggregate loss distributions
Defining relevant pools, testing and selecting, parameter estimation, convolution

3:00 Afternoon coffee break

3:30 Capital estimation, and allocation under imperfect dependence
Estimation methods (UoM level), hidden challenges and major un(der)addressed issues of extreme VaR, methods for estimating dependence across UoMs, methods of allocation under imperfect dependence

5:00 End of workshop

About this workshop

Estimating operational risk capital under Basel II's AMA-LDA framework requires the judicious application of multiple statistical (and sometimes econometric) models in a number of sequential steps. With raw loss data as our starting point, we identify the empirical, methodological, and regulatory challenges and constraints that make this a nontrivial exercise. This workshop gets us from there to here: from raw loss data (in all its forms) to defensible and methodologically sound capital estimates that are fully compliant with regulatory expectations, published guidance, and the relevant statistical academic literatures. The focus of the workshop is on practical application: it takes a solution-oriented cookbook approach that identifies and explains the range of reliable and tested methods widely utilized by industry, and often written by regulators. The scope is comprehensive, covering all components of the framework even though any one institution's framework will only make use of a subset of these, because the methods used in one section of the framework determine those that can be used for other components of the framework. This is often overlooked when researchers take a siloed approach to assessing these methods, so cognizance of how all the statistical pieces can and should fit together is a focus of the workshop. Another focus is on the use of econometric models to directly link key risk indicators (KRIs) to capital. This enables the operational risk modeler to make statistically causal statements to the operational risk manager, such as, "If you are able to decrease your system downtime by a standard deviation (X%), you can decrease your capital estimate by $25m, all else equal." A heretofore valid criticism of operational risk capital models is that they have not been tied directly to operational risk mitigation and management, but such econometric models make this connection and allow mitigation to directly and measurably and causally affect capital levels. Finally, previously unreported methodological pitfalls to some of the widely used methods are identified, and solutions proposed.

Who should attend:

  • Director-Operational Risk Modeling and Quantification
  • Senior Operational Risk Analytical Modeler
  • Managing Director-Operational Risk Capital Estimation
  • Senior Vice President-Operational Risk Regulatory and Economic Capital
  • EVP-Operational Risk Analytics, Regulatory Modeling and Capital Estimation
  • Senior Analyst and Modeler-Operational Risk
  • Economic and Regulatory Capital Modeler-Operational Risk
  • Senior Operational Risk Capital Modeler
Follow us
CPE Accredited