The Failure of Risk Models
And why average correlations aren't what they seem.
BY Caroline Cakebread | February 22, 2018
Investors are putting risk at the centre of the asset allocation decision – and that’s good news. The trouble is, the risk models they’re using aren’t up to the task.
So says Sebastien Page, head of global multi-asset at T. Rowe Price. Despite shortcomings with the current models, he believes there are solutions investors can use to avoid the ups and downs that come with market volatility.
On average, Page explains, Canadian defined benefit pension plans in the Pension Investment Association of Canada (PIAC) universe have 35% of their portfolios allocated to equities.
But when it comes to risk, equities account for 85% – far beyond the allocation. That’s because most portfolios also have equity-like exposure in other areas such as real estate, private equity and hedge funds.
Not only is the equity risk factor the most volatile, says Page, volatility changes over time, something most risk models don’t account for. “The problem with this approach to risk is that it relies on volatilities and correlations that are assumed to be stable. But risk is not stable through time.”
To illustrate his point, Page shows how correlations between stocks and credit spreads and between stocks and treasuries can change depending on the business cycle. The variation can be significant.
During an expansionary period, for example, the correlation between stocks and credit spreads is 37% while the correlation between stocks and treasuries is zero. Change up the environment and it’s a different story.
During times of recession, the correlation between stocks and credit spreads shoots up to 58% while correlations between stocks and treasuries plunges to -34%.
“Clearly correlations aren’t stable through time,” he concludes, adding that investors who rely on average correlation to determine risk aren’t diversifying at all. “It’s like having your head in the freezer and your feet in the oven. Your body temperature might be fine on average, but your chances of survival are low because of exposure to extremes.”
The failure of risk models becomes the most evident during tail events, where correlations rise drastically. So which type of diversification should investors rely on for portfolio construction?
While investors traditionally rely on bonds to diversify equity risk, that can be very challenging during regime changes. Will volatility be driven by inflation and interest rate shocks? If so, bonds might not be good diversifiers.
“But if we continue to have no inflation risk, business cycle variables will dominate and bonds will hold their traditional role as diversifiers of equity risk,” explains Page.
Ultimately, investors need to look at strategies that stabilize portfolio volatility (so-called “managed volatility” strategies) and control exposure to loss – those who fail to do that expose themselves to tail risk.
“Diversification may disappear when we need it most,” concludes Page, adding that even the stock-bond correlation can be highly unstable. Ultimately, investors need to manage volatility first and foremost to best work around weaknesses in prevailing risk models.