Who knew that the subject of US Treasury bond yields could inspire such passion? When, in late June, I argued that they were likely to move considerably higher than the then-prevailing 3.75%, I attracted some vehement pushback. In a publication titled “Don’t Be a Dud,” analysts at Morgan Stanley insisted that the 10-year bond would experience a summer price rally and that the yield would ultimately settle into a longer-term range of 2% to 3%.
I’m sticking with my prediction. What’s more, I strongly suspect that the bond bull market that began in the early 1980s is over.
My forecast broke the 10-year Treasury yield down into three constituent estimates. First, there’s r*, the “neutral” short-term interest rate that the Federal Reserve would set if it wanted to neither hinder nor stimulate growth. I put this at 1%. Then, there’s the average long-term inflation rate: 2.5%. Finally, I estimated the term premium, the added yield that investors will require to compensate for the risks of longer-term lending: 1%. From there, the arithmetic was simple: 1% + 2.5% + 1% = a target yield of 4.5%.
The Morgan Stanley analysts, by contrast, drew from the experience of the past decade to forecast that r*, inflation and the term premium would all be lower, resulting in a lower overall yield.
Since then, the 10-year yield has risen significantly, to about 4.3%. But I’m not taking a victory lap. My evaluation focused on longer-term secular trends, while the past month’s increase has a lot to do with cyclical developments such as a stronger-than-expected economy. I certainly didn’t anticipate that yields would immediately shoot up.
That said, secular elements are also evident. First, the economy’s strength amid much higher interest rates suggests that the neutral rate is higher than previously believed. This is starting to seep into Fed officials’ forecasts: In the June Summary of Economic Projections, the central tendency for the long-term federal funds rate moved up slightly. I expect officials to keep revising their estimates of r* upwards, though this probably won’t be reflected in estimates based on econometric models, which are slow-moving and somewhat skewed by pandemic-period data.
