Best-practice stress-testing framework
Types of risk “Bolt from the blue”
Risk due to high-impact of extremely-low-probability “unknown unknowns”
Meteorite shower on earth
Decreasing probability, increasing impact of risk
▪ In 2007 only 1 analyst foresaw and factored in as a possible scenario oil prices rising above $80 per bbl in 2008; for others it was a complete unexpected shock . “New uncertainties” ▪ Increase in risk due to a downturn ▪ Risk due to low-probability “unknown unknowns”
▪ Chinese demand destruction leading to a collapse of basic materials prices
▪ Increase in risk due to a downturn.
▪ Risk due to low- probability high-impact “known unknowns”
▪ Decline in U.S. consumption levels impacting export-oriented industries in developing countries
Risk due to high- probability “known- unknowns”
Scope of stress testing
No analysis can capture this effect
Increased volatility of PD distribution for segments of portfolio not affected by stress
Scenario analysis comprising macro- economic, industry, financial market, regulatory/political scenarios
3. Capital allocation
While most of the banks we interviewed estimate economic capital, they typically keep the more conservative of economic or regulatory capital (usually the latter) and allocate capital to the business units based on this higher amount. Banks that choose to keep economic capital do so only when it is very similar to regulatory capital.
However, banks use different approaches to capital allocation. Banks that do not use allocated capital for pricing tend to allocate regulatory capital based on marginal contribution to economic capital.
Banks that drive pricing decisions on allocated capital are very sensitive to the mechanism of allocation. While a few banks only allocate economic capital based on marginal contribution, others use heuristics to drive business objectives or use models to allocate excess capital over economic capital.
4. Lessons from the crisis
What lessons can portfolio risk managers learn from the current crisis? The banks we interviewed said that the models themselves were not to blame, but rather management’s failure to believe the model outputs and take action. Most of the risk management teams felt that their bank leadership ignored the early warning signals triggered by the models because they did not tally with intuition. As the signals became far more intense, leadership began to