sexta-feira, julho 14, 2017


"Companies often maintain a list of the main risks that managers believe they face, which they report as their “risk register” in annual reports. These include discrete operational events, such as major industrial accidents, cyberattacks, or employee malfeasance. If they take the next step to quantify those risks, many simply turn to that list and model them, often for the first time, onto their financial outlook. That’s a good start, as it gives managers some insight into how sensitive the company’s financial health is to changes around individual risks, which many companies don’t do. But measuring individual risks discretely does little to illuminate a more complex landscape of interrelated risks that often move together in the real world. That requires the further step of coherently clustering risks together into scenarios.
Scenarios are more appropriate because they help managers consider the effects of a variety of severe but plausible scenarios without being farfetched. They can also accommodate interaction effects among sensitivities.
We frequently encounter companies willing to model broader, everyday market variables, such as GDP or inflation, or more specific variables, such as the rate of formation of new companies. But we seldom find companies willing to model more extreme variables (see sidebar, “Stress testing for an energy utility company”) or one-off events, such as a cyberattack or a natural disaster. The data to measure the effects of the former are fairly easy to come by, some argue, while reliable data on the latter are not. Others believe that their employees would sufficiently rally together to counter such events."
Trechos retirados de "Stress testing for nonfinancial companies"

Sem comentários: