The current credit crisis has shown, however, that overreliance on historical data and analytical models may not provide sufficient data to analyze very high-impact events and might actually lead to overconfidence. In addition, the process of building models has a tendency to focus developers on preconceived risk constellations and the impact of specific events, whether singularly or in combinations determined through correlation matrices.
Scenario analysis and stress tests based on consideration of shock events and their possible repercussions can provide useful information to management and regulators on a company’s resiliency through a chain of events, as well as support the consideration of a firm’s operations as an integral part of a wider financial system. By careful selection, construction and analysis of scenarios unfolding over a period of time, a more holistic picture of the firm’s risk position can be created. Additionally, because such scenarios have at their heart a story-line, the communication process with key stakeholders is less abstract than discussions focused on distributions, tails and other mathematical constructs.
Finally, we also discuss how scenario analysis and stress testing can be used to define a company’s risk appetite, which is at the core of a well-embedded ERM framework. The theoretical approach discussed will be supported through the presentation of the construction and analysis of an event-chain scenario deriving from recent global financial developments.
There is no shortage of literature on the (in)ability of human beings to assess risk properly. Collectively we have short-term memories along with a disinclination to forego short-term gains when we perceive risks to be distant or unlikely. The literature of how people view risk depending on context, group size and numerous other factors is extensive. Quantitative models have proven to be extremely useful in helping us quantify risks, understand observed phenomena, explore the sources and impacts of financial risk, and develop tools and methods for managing risks. At their best, models remove a great deal of bias and subjectivity from risk analysis as well as give us a measurement tool.
