Many investors made that move away from duration in 2014 and 2015, and many of them probably regret it now. Over that period, flows proliferated into bond strategies that either seek to eliminate interest rate risk entirely or have the ability to outright short duration (and many did). However, many of these strategies have been hurt not only by the negative carry and the principal loss from falling rates in 2014 and during the summer of 2015, but also by the spread widening of volatile credit spread exposures.
This raises an important question: In a low-yield environment, does having a “structural” absence of duration in a bond portfolio, or even an outright short duration position, make a good long-term investment strategy? We believe the answer is an emphatic no. Over time, bond investors are typically compensated to be long duration – as long as yield curves are upward-sloping, which has nearly always been the case historically. One also should not forget one of the primary reasons to invest in bonds to begin with – duration provides a valuable anchor and diversifier to risk assets, such as equity and credit. Eliminating duration from a portfolio can also eliminate these valuable potential benefits of fixed income.
Then just how much duration is appropriate?
In active bond portfolios, tactical moves into lower duration or short duration positions may be warranted in certain situations when managers have a firm conviction that differs from the longer-term strategic view. However, a structural zero-duration (or negative duration) position is not advisable or warranted in most cases. This naturally leads to the question: How much duration should investors hold in their portfolios in the current low yield environment – especially given concerns about the prospect for higher rates? Of course, to even begin to address this question, we would need to know an investor’s objectives, risk tolerance, time horizon, exposure to risk assets and many other kinds of information.
It may be helpful, however, to provide more broad context about the potential impact of rising interest rates on a bond portfolio over time. The magnitude
and time over which interest rates rise are critical components in the equation. Is it a sharp, overnight spike in rates? Steady over time? How long will
the rising rate period last? And how high will rates climb?
As a rule of thumb, the sooner the pain of the rate spike is over, the sooner the investor tends to be able to capitalize on the subsequent higher
level of carry.
The mirror exercise
Thanks to bond math, we could model hypothetical bond portfolio returns under any of the above scenarios, to any degree. But no one can see into the future
with certainty. So, rather than making any formulaic assumptions, what if – as a purely hypothetical exercise – we were to instead analyze potential bond
returns in an interest rate environment that followed a path that was the mirror image of the previous 15 years? In other words, what if the
forward path for rates retraced what we experienced historically up to the end of 2014, in reverse? (See Figure 1.) For example, beginning at the 2.17%
yield as of year-end 2014, in 2015 we would see a gradual rise in rates to close the year at 3.03% (where we were at year-end 2013), then a “reverse taper
tantrum” in 2016 (reflecting May 2013), then a spike upward in 2018 (reflecting the European peripheral crisis in 2011), etc. This analysis may be easy to
conceptualize, as it simply retraces what markets and investors experienced in the past. And is it possible this is a hypothetical picture of the gradually
rising interest rate path that many bond bears envision in the future?
To be clear, this is a purely illustrative exercise on the impact of duration in a portfolio. The scenario implied in this “mirrored path” analysis is not PIMCO’s base case for U.S. interest rates going forward – in fact, it is quite far from it. Rather, PIMCO’s outlook is for a modest uptick in U.S. growth over the cyclical horizon, which will lead to the first Federal Reserve tightening cycle in over 10 years. The timing and magnitude of hikes is likely to be muted, with the Fed eventually ending the tightening cycle at a lower level than it has historically. This would leave the U.S. with a neutral policy rate closer to 2% in nominal terms than to the 4% neutral policy rate that prevailed before the crisis, if the Fed doesn’t overshoot neutral as it has in past rate hike cycles.
By contrast, under the mirrored path scenario, interest rates (and, specifically, the 10-year Treasury yield shown in this illustration) would trend significantly higher over the 15-year study, rising from the 2.17% level at the end of December 2014 to 6.44% by the end of 2029 – a rise of 427 basis points. Such a rise in yields would seem to spell disaster for bond investors, right? And surely if one were to invest in bonds over the period, one would want to keep duration positioning low, right?
Before we share the results, one quick reminder: Generally, when predicting future bond market returns, a (second) rule of thumb has always been to start with a bond’s (or a bond index’s) initial yield, which may explain a significant portion of the bond market’s return. The historical data generally supports this theory: A regression of monthly data since 1973 on the Barclays U.S. Aggregate Index, for example, suggests this coefficient is as high as 78% – so initial yield is indeed highly meaningful. Would such a rule of thumb apply here, where rates rise so significantly?
Using the mirror-image path of 10-year U.S. Treasury rates, we ran a simulation to model annualized total returns for constant maturity two-year, five-year and 10-year U.S. Treasuries. The results, shown in Figure 2, may come as a surprise. Thanks to the magic of compounding yields, not only do bonds fare well in the rising rate environment illustrated in this scenario, they actually outperform initial yield indicators. Particularly notable is that the 10-year Treasury – the bond with the highest interest rate sensitivity of the three – turns out to be the highest outperformer at the end of the 15-year simulation and also during significant stretches of the entire period.
As for the influence of initial yield in this scenario, Figure 3 depicts the 2014 year-end yields for each of the bonds, along with the actual trailing annualized 15-year return and the forward-looking mirrored path simulation return.
In this case, not only are initial yields a fair indicator of future returns, but in this simulated path of significantly higher rates over time, the elevated carry provides a net positive to future returns, outweighing the impact of principal losses from higher interest rates over the long term.
While 2%–2.5% hypothetical returns are not terribly compelling, keep in mind that these are U.S. Treasuries. Investors in broader, actively managed core bond strategies could target additional returns from credit spread and alpha strategies. And the potential to generate consistent returns and with less volatility over the long term than many other asset classes, together with the ability to serve as a hedge during economic downturns, remain key reasons to invest in bonds.
In light of this hypothetical look in the mirror, where duration exposure holds its own by offering long-term returns even in a higher-than-expected rising rate scenario, investors may decide that having some duration in a bond portfolio is not quite as scary as the bears seem to think.