Are interest rates the explanation for low vol’s strong performance?

I was interested to see last week a new research paper¹ that explores the connection between interest rates and the performance of low volatility stocks (low vol.) Among the paper’s findings are:

(a) that there is “a strong implicit exposure of low volatility portfolios to bonds,” and

(b) that “part of the outperformance of low volatility stocks can be explained by a premium for interest rate exposure.”

Low volatility stocks and interest rates

Low vol, of course, has long represented something of an anomaly because it’s had an extraordinarily good performance history relative to the broad stock market despite being by most reasonable measures a less risky thing to hold. The new paper by Joost Driessen, Ivo Kuiper and Robbert Beilo may shed some light on why.

Intuitively, the paper’s findings seem to make sense. The stock of highly leveraged corporations, for example, tends to more volatile, and leverage can be thought of as a short exposure to bond markets. For this and other reasons, the connection between stock volatility and bond markets is not surprising. And, given the remarkable fall in interest rates over the past 35 years, it’s reasonable to expect that this will have provided a tailwind for low vol’s performance.

Note, however, that the authors are careful to say that their analysis points to interest rates being only a partial explanation for the low volatility anomaly (they reckon somewhere between 20% and 80% “depending on the methodology chosen”.) And when we look just at the more recent years, the intuitive argument doesn’t seem quite as neat: bond markets haven’t been especially strong in the past five years, yet low volatility stocks have continued to do well. So this is certainly not “case closed” on the low vol anomaly.

And what, to me, is especially interesting here is the result that “the estimated premium for interest rate risk is much higher in the equity market compared to the bond market.” In fact, the authors’ models point to the reward for the (implicit) interest rate exposure in low vol stocks being 2 to 5 times larger than the reward for the (explicit) interest rate exposure in bonds.

On the surface, that’s a big enough difference that it might seem like the paper is maybe re–expressing part of the low volatility anomaly in a different form rather than really explaining it.

The limitations of factor models

Then again, there is a possible rationale for the differential in reward. When dealing with factor models we should never forget that there’s a big difference between (a) a factor that, on average over a certain period, has tended to behave like a thing and (b) the thing itself. This was the basis, for example, of my colleague Jim Gannon’s argument a few years ago that even if equities have non–zero duration (sensitivity to interest rates), they should not be counted as contributing to a pension liability–driven investing (LDI) strategy: “equity duration is not really the same duration we are talking about with liabilities.”²

In other words, just because an echo of bond market performance can be heard in low vol’s performance, that doesn’t mean that holding low vol is really the same thing as holding bonds. If there is an implicit exposure of low volatility portfolios to bonds, it’s nonetheless an indirect and inconsistent exposure. So the interest rate factor may well be rewarded differently in the two different contexts. Nonetheless, if Driessen and his co–authors are correct about the extent of the difference, that’s quite a remarkable finding.

An aside: defensive equity vs. low volatility

The topic of low volatility stocks is one I’ve touched on in this blog several times in the past, although it’s always been in the context of low vol as being a part of defensive equity. Defensive equity is a composite of quality factors (low leverage, stable earnings, return on equity) and low volatility. My colleague Dave Hintz (whose analysis several years ago led to the creation of the Russell Defensive indexes) explains why defensiveness tends to be a more robust measure of risk than merely looking at historical volatility alone: “Risk factors—such as the effect of leverage—don’t show up in the volatility numbers over every time period. So low volatility strategies that do not include balance sheet quality factors can be susceptible to producing negative returns when interest rates rise.”

I suspect that if the analysis described in the paper were to be carried out on defensive, rather than pure low vol, portfolios, then the conclusions are likely to be very similar—not just because low vol is a big part of defensive, but also because leverage, for example, is directly connected to the bond market. That remains to be tested, though.