Distracted by the U.S. Federal Reserve’s oscillating taper plans, bond investors risk losing sight of the big picture. To those invested for the long haul, the precise contours of the road immediately ahead are barely relevant. What matters is the ultimate destination: A return to a normal level of bond yields, with the fatter coupons that come with it.
But just what constitutes “normal” for a financial instrument that has long plumbed the depths, and was never particularly stationary to begin with? A plausible answer comes from the combined wisdom of four estimation techniques.
The first method focuses on the historical record. The second takes its cue from the Fed’s own long-term expectations. The third capitalizes on a connection between economic growth and the level of bond yields. The fourth uncovers the bond market’s own thinking on this subject.
We start with what history can tell us about “normal” bond yields. Since the start of the Industrial Revolution (1750), the median long-term interest rate in the advanced world is a tame 4.13 per cent. We can drill even deeper by sacrificing a bit of history in exchange for better data: Since 1870, the median U.S. 10-year government bond yield is just 3.92 per cent.
The second model takes its cue from central bankers. The Fed helpfully estimates a neutral federal funds rate (its benchmark policy rate) at just shy of 4 per cent. To telescope out the yield curve, we must affix a term premium on top; an extra 0.75 to one percentage point compensates investors for the additional uncertainty attached to a longer-dated bond.
However, a definitional problem remains. We have arrived at a central bank-inspired neutral bond yield. But we seek “normal,” not “neutral.” What’s the difference? Central banks tend to behave asymmetrically, dipping much further below their neutral rate than they venture above. With this in mind, we adjust this model’s projection to a range of 4 to 5 per cent.
The third model is near and dear to any economist’s heart. It observes that there is a historic connection between nominal economic growth and the level of bond yields. This is unsurprising. When the economy is strong, central banks increase rates, and so bond yields rise. At the micro level, if yields are materially lower than GDP growth, investors prefer to invest elsewhere in the economy. If yields are much higher, borrowers cannot earn a sufficient return on their undertakings to justify the loan.
By evaluating the likely trajectory for demographics, productivity and inflation, we arrive at nominal GDP growth forecast of 4 to 4.25 per cent in a decade’s time. By extension, this also represents the third model’s “normal” yield estimate.
The fourth and final technique throws what we know about economics, monetary policy and the history of bond yields out the window – relying instead on the sagacity of market expectations. The forward curve informs us that the U.S. 10-year yield in a decade’s time should average 4.20 to 4.45 per cent.
The four models are now in hand, but the work isn’t quite done. A sprinkling of special considerations must now be sifted into the mix. Arguing in favour of a tilt toward higher yields, the U.S. public debt load has soared. However, we are reluctant to embrace this notion too readily. The relationship between public debt and bond yields is notoriously loose, and the U.S. Treasury arguably enjoys a special dispensation to borrow without serious fear of reprisal, due to its reserve status.
Pushing in the opposite direction, history demonstrates that it is not unusual for policy makers to subtly repress interest rates in an era of high public debt. Furthermore, despite all of the U.S. Treasury debt floating about, there may actually be a shortage of safe assets; a shrunken pool of highly rated borrowers has more than met its match in de-risking banks and emerging market savers who still lack “risk-free” domestic investments. Meanwhile, an aging population does more than just slow economic growth. Risk tolerance also declines with age, increasing the thirst for bonds. Netted out, these motley effects should depress yields by up to 50 basis points.
Averaging across the four models and applying this final flourish, the “normal” U.S. 10-year yield looks to be just shy of 4 per cent. This is significantly higher than today, but not monumentally so. In fact, it contends that the much-feared bond bear market is already almost half done.
Does this estimate feel low? If so, an anchoring bias may be to blame. Investors are naturally inclined to define “normal” based on what they have personally experienced. But, empirically, the 1970s through 1990s were actually an era of abnormally high rates.
To be fair, the rhythms of the business cycle and the mercurial nature of the bond market itself argue that yields will regularly deviate above “normal” for lengthy periods of time. But they will also have long bouts below it, as demonstrated by current conditions.
Eric Lascelles is chief economist at RBC Global Asset Management. This article is based on a longer report he recently published on the topic, available here.