The traditional asset allocation model “fails to address the short-term risks—those unknown unknowns—that can happen very suddenly,” he explained. Looking at the financial crisis, for example, shows that correlations between asset classes can change very quickly and very dramatically, he added.
So how are plan sponsors dealing with it? “Since the global financial crisis, there’s been much more focus on buying insurance. But the issue with insurance is that it’s quite expensive,” noted Khasgiwale—particularly if it’s ongoing.
Historical scenario testing also has limitations, he added, which is why forward-looking stress testing is important.
Khasgiwale advised plan sponsors to consider three aspects:
Scenario Construction—”The whole point of forward-looking stress testing is to think of extremely unlikely but plausible events,” he said, adding that it’s important to incorporate investment opinions from experts such as economists, not just risk and quant teams.
Multi-variate model for tail-risk analysis—Investment managers can incorporate a Monte Carlo simulation, giving them more data points to work with.
Entropy pooling—This means tilting the data to reinforce the data points of most relevance and fade out those that are less meaningful, Khasgiwale explained.
Examples of forward-looking scenarios might include an abrupt end to quantitative easing, the EU moving apart, a commodity shock, or a crisis in China.
Running forward-looking stress tests with such scenarios can provide useful information for both plan sponsors and investment professionals. “This type of output can really help define how diversified your portfolio really is,” he added, noting that it also helps portfolio managers to understand if the diversifiers in the portfolio are actually acting that way.
Khasgiwale offered three final takeaways for pension plan sponsors:
- Add forward-looking stress testing to your suite of risk tools;
- Review existing processes; and
- Consider incorporating absolute return strategies.
Alyssa Hodder is editor, Benefits Canada.