When to reduce the volatility of an equity portfolio

Having worked as a money manager with mandates that allowed me the freedom to move at will between cash, bonds and equities, while still being measured against the world equity index, my key asset allocation decision was always when to reduce the portfolio’s volatility below that of the benchmark—and which tool to use in pursuit of this goal.

This problem is something of a holy grail for money managers and anybody who thinks they have found a permanent solution should be dispatched with due haste back into a research position. That said, it is as a researcher that I offer up a suggestion to this old problem. I will start with a few statements that are easily verifiable by observation: (i) in an inflationary period, government bonds and equities have a positive correlation, (ii) when interest rates decline, the stock market rises. Not so in a deflationary period: when interest rates decline, the stock market declines as the implication is that the economy is falling out of bed.

Hence, a long-dated bond is a good hedge against a deflationary bust as has been shown in almost every bear market since 1987. This, of course, requires the long-bond not to be stupidly overvalued as is now the case in Europe, with 30-year bunds yielding 1.30%. Remarkably, this is just below the yield offered on 30-year Japanese government bonds. In fact, long-dated bonds in Japan and Germany offer risk with no return—never a good bet. This means that it will be hard to protect against deflation risk in Japanese or German equity portfolios using their respective long-bonds.

In the US things are different as 30-year treasuries yield 3%, putting them in the middle of the range of my valuation model. Should a deflationary shock hit (then) yields could quickly fall to German or Japanese levels, implying a capital gain of up to 50%, which, if enough protection is in place, should offset a big fall in equities. So the idea that 3% longbond yields are “too low” to defend against a deflationary bust is just plain wrong. It happens to be a view of those who have been expecting US growth to rebound strongly and for inflation to come back.

I conclude that if an equity portfolio is to be protected from a deflationary bust, the only effective tool left on the rack is the 30-year US treasury. The next question is how this tool should be used. One option is to keep a constant 50:50 allocation in the style of late 19th century institutional investors which, in fact, defaulted to a 60:40 mix favoring bonds as the overall climate was unambiguously deflationary. For those minded to time their moves, a more dynamic system is needed.

What is clear is that a balanced portfolio generates superior returns than a long-only equivalent comprised of just bonds or equities, and offers much lower volatility (incidentally bonds beat the hell out of equities, pretty much going back to 1982). And to make this portfolio as usable as possible I have eschewed my preferred 30-year US zerocoupon bond in favour of the Merrill Lynch government bond index, with duration exceeding ten years. This, I believe, is a “buyable” proposition.

Those more inclined to time the market may choose to apply a bond hedge only when the danger of a deflationary bust appears. As such, my “new” US recession indicator may be a useful tool. In the chart above, shaded areas denote periods when the recession indicator is below 0 (currently it is at -5). History suggests that performance could have been significantly improved by keeping a 100% weighting in equities during periods when the indicator is positive and then shifting to a 50:50 position when it turns negative.

Since the beginning of last summer the indicator has been in negative territory and hence on the basis of historical experience, this should be a period when a hedge is warranted. Such reasoning runs contrary to the generally bullish commentary surrounding the US economy’s trajectory, but I am deeply skeptical of these arguments and would tend to have greater faith in a tool that has consistently delivered results.