A Framework for Scenario Based Risk Management

The framework presented in this paper describes how a risk manager in a systematic and structured way can construct scenarios. It creates a natural platform where quantitative analysts, economists as well as top management within a large bank can discuss, quantify and implement scenarios. A key aspect of the framework is also that it links all components of integrated risk management: Sensitivity measures, Value at Risk calculations as well as stress testing. The core of the framework is a mathematical model of joint risk factor evolution through time and the concept of an information hypercube. The mathematical model consists of two components: models for the univariate marginal distributions of risk factors and a model of the codependence structure encoded in the notion of copulas. The aim of the mathematical model is to capture, as realistically as possible, the actual data generating process. That is, the unobserved true structure. The hypercube consists of a set of core risk factors and pre-defined future paths of these risk factors. Interaction in the scenario process appears at the hypercube level and through the application of the hypercube information on the mathematical model. Hence, reducing the degrees of freedom of the model