Designing stress scenarios for portfolios

Abstract

Evaluating portfolio performance under different stress scenarios is an important risk management tool. Designing stress scenarios for portfolios can be complex as it involves determining potential market changes in different asset classes and risk factors in a coherent manner that are extreme but plausible. This paper describes an approach to design such stress scenarios where the portfolio performance depends on market changes in many risk factors and asset classes. The scenario design is customized for the portfolio and helps describe plausible market changes that would have the most adverse impact on the portfolio performance. The approach relies on historical data and derives the scenario based on market changes during historical periods that would have been the most stressful for the given portfolio. The proposed approach also allows one to adjust the level of severity and if desired incorporate any specific market conditions of concern (such as scenario design for increasing interest rate environment and/or certain level of unemployment rate, etc.). The main advantages of the proposed approach are (a) flexibility in scenario design with and without constraints on market conditions with adjustable levels of severities, (b) computational simplicity, (c) scalability to any number of market risk factors, (d) no need of prior assumptions on joint distribution of market risk factors, and (e) transparency of the results as they are developed from market changes during actual stressful historical periods.