Body
Conventional wisdom about the US Treasury yield curve goes roughly as follows: Long-run productivity growth appears likely to be low, and productivity growth and interest rates move largely together, so one should expect long rates to be low as well. And the fact that the Federal Reserve’s current policy of near zero rates has generated demand growth barely in excess of potential growth suggests that the United States is not far from these long-run low neutral rates already.
Such conventional wisdom should be questioned from a number of angles.
Take the statement about productivity growth. Yes, measured productivity growth has decreased, and seemingly not due to measurement error (Byrne et al. 2016, Syverson 2016). The best guess is that productivity growth will indeed remain lower than before the crisis, perhaps by 0.5 to 1 percent. But the uncertainty associated with this best guess is extremely large. From a statistical viewpoint, the correlation between (nonoverlapping) five-year average productivity growth rates is very low, equal to 0.1 for the United States since 1970. From a technological viewpoint, the tension between the current poor productivity numbers and the discussions about robots taking over human jobs is fascinating. So far, both the employment and productivity numbers clearly show that robots have not displaced labor on a large scale, and whatever jobs they have destroyed have been replaced by others. But when one listens to Silicon Valley, one cannot help but expect a substantial probability of a much larger role for robots and artificial intelligence in general, and by implication, much higher productivity gains. Bottom line: Expect lower productivity growth, but be ready to be surprised.
Take the statement about the link between productivity growth rates and interest rates. Many economists seem to believe that there should be, and that there is, a tight relation between the two. In fact, neither theory nor empirical evidence support the proposition. The theory that delivers the tight relation between the two is an example of how economists are sometimes thought-prisoners of their models. The model that yields this result is based on two assumptions: that people live forever, or act as if they lived forever, and that people are willing to defer consumption if the interest rate is higher. The first assumption is obviously false. The second has little if any empirical support. Short of these two assumptions, theory delivers a complex relation between the growth rate and the interest rate, with, at best, a loose relation between the two. And the empirical evidence shows no such relation. An exhaustive study by Hamilton and others (Hamilton et al 2016) of the comovements over the medium run between the two rates over more than a century and many countries finds no reliable relation between the two. In short, productivity growth may be lower in the future, but this has no obvious implication for the long-run real rate. The bottom line: In thinking about the yield curve, do not assume that lower productivity growth necessarily comes with lower rates.
Finally, take the proposition that low rates today suggest that we are already there. The argument is simple and, on the face of it, rather convincing: With roughly zero rates today, demand growth is only slightly higher than potential growth. Yet many of the brakes to demand that plausibly explained low demand growth in the past seven years are gone. Fiscal consolidation has given way to mild fiscal expansion. Banks are no longer deleveraging, and credit supply is abundant and cheap. Thus, if very low rates are needed to sustain demand today, why should it be different in the future?
The argument has one flaw however. Brakes from the crisis may indeed be largely gone. But another factor, the anticipation of a worse future, is now at play. As argued above, the best guess, while far from certain, is that productivity growth will be lower in the future than in the past. The slowdown in productivity, which initially could be explained by the crisis, looks likely to have a more permanent component. Forecasts of long-term growth, and the general commentary in newspapers, are gloomy. I believe that this bad news about the future largely explains the relative weakness of demand today. Put in more academic terms, bad news about the future supply side is leading to a Keynesian slowdown, or at least a weaker recovery today. (I have argued elsewhere that some past US recessions were indeed due to pessimistic expectations about the future [Blanchard 2016].)
It is useful to play with some numbers here. Suppose that you learn that your income over the next 30 years will rise at 4 percent rather than at 5 percent as you expected earlier (because income typically increases with age, individual income typically increases faster than aggregate income). This represents a roughly 20 percent decrease in the present value of your future earnings, and is likely to lead you to consume say 10 percent less. If this realization comes to you over a period of five years, you will decrease consumption by 2 percent each year relative to your income. Returning to aggregate implications, as consumers adjust their expectations the way you do, consumption growth will be weak. The same argument applies to investment. The lower the expected growth rate of profit, the lower the desired level of capital, and this in turn will lead to a period of low investment until the new lower level of capital is reached.
Returning to the yield curve, this has a clear implication: As consumers’ and firms’ expectations are adjusting to lower long-run growth, demand is lower than it would otherwise be. Despite low interest rates, demand growth is weak, and the interest rate needed to sustain it is low. As the adjustment of expectations comes to an end, however, demand may well pick up, and the Federal Reserve may find that it needs to tighten rates substantially to avoid overheating. The rate that eventually prevails may be substantially higher than the rate prevailing today, and the true yield curve may be steeper than markets currently assume.
To summarize: Low productivity growth is not necessarily a harbinger of low rates in the future. Neither are low rates today.
References
Blanchard, Olivier, ``Slow growth is a fact of life in the post-crisis world,’’ Financial Times, April 2016.
Byrne, David; John Fernald; and Marshall Reinsdorf, ``Does the United States have a productivity slowdown or a measurement problem?” Federal Reserve of San Francisco, Working Paper 2016-03, April 2016.
Hamilton, James; Eytan Harris; Jan Hatzius; and Kenneth West, “The equilibrium real funds rate: Past, present, and future,” mimeo, May 2016.
Syverson, Chad, ``Challenges to mismeasurement explanations of the U.S. productivity slowdown,” NBER Working Paper 21974, February 2016.
