Abstract
One of the most durable patterns in market behavior involves contagion—increases in correlation and volatility—during crash periods such as 2008. This condition can cause major problems for an investor when markets severely contract and anticipated diversification benefits vanish. To address contagion, the authors implement a machine-learning algorithm, trend filtering, to capture distinctive economic conditions. Over long horizons, they find that a multiregime simulation provides more accurate estimates of downside risk compared with traditional static portfolio models and can help in evaluating strategies for reducing the worst-case outcomes. The approach readily applies to nonprofit institutions that depend upon their endowment capital to fund liabilities and meet goals.
- © 2016 Pageant Media Ltd
Don’t have access? Register today to begin unrestricted access to our database of research.