Larry Summers recently raised the possibility of the US economy being in danger of facing a long period of subpar growth. The fear is at least somewhat reflected in several long-run growth forecasts. Table 1 shows the most recent estimates of steady state real GDP growth from the FOMC Economic Projections, Blue Chip Economic Indicators and the Congressional Budget Office (CBO). All three are 2.3-2.4 percent, more than a full percentage point below the 1955-2005 average.
|Real T Bill|
|Real Fed Funds|
Over the period from 1955-2005 the 3-month Treasury Bill rate deflated by PCE inflation has averaged 1.7 percent, while the federal funds rate deflated by PCE inflation has averaged 2.2 percent. So real short rates have averaged around 2 percent. But the dismal growth outlook that is envisioned should lower equilibrium real rates. Table 1 shows forecasts for the long-run value of real short rates from the FOMC, Blue Chip and CBO. The CBO and FOMC forecasts are both exactly 2 percent. These forecasts paint a picture for real GDP growth that is quite unlike the last 50 years, but with no effect on real rates. We believe that this growth outlook should imply a materially lower equilibrium level of real short rates. In partial defense of the FOMC forecast it should be noted that they are forecasting the real federal funds rate which ought to be a little higher than the real T Bill yield. Also, four of the respondents to the FOMC projections predict a long-run value of the real funds rate of 1.5 percent, which is at least a nod in the right direction. Still the median and modal FOMC forecast is for a real funds rate of 2 percent, which is very close to its historical mean, despite a growth outlook that is far from the historical mean. The Blue Chip forecast is 1.6 percent, which at least slightly more consistent with the growth outlook than the CBO projection.
Table 1 also shows predictions of the steady-state term premium defined as the spread between long-run forecasts of ten-year over three-month yields. The CBO and Blue Chip term premium forecasts are 1.3 and 1.1 percent, respectively (the FOMC does not predict this variable). Empirical term premium estimates have been trending down over the last twenty years, and are generally now around zero. There is a potential economic story for this. In the 1980s and 1990s, inflation was the risk that worried investors most. Inflation erodes the value of a long-term nominal bond, making these bonds risky assets. But more recently, it is financial instability and recession that worries investors more. US Treasury bonds will perform excellently in this environment, and are thus a hedge against the risks that keep investors up at night. So the term premium should be lower, or indeed negative. Robert Shiller has compiled data on one-year and long-term bond yields going back to 1871. On average from 1871 to 1960, the long-term yield was 40 basis points below the one-year yield. Outside of the Great Inflation and its aftermath, there doesn’t seem to have been much of a term premium.
Now if healthy growth resumes, 1.3 percentage points might be a reasonable calibration for the term premium. But conditional on the real GDP outlook in Table 1, investors are going to be more worried about weakness and demand deficiency than an overheating economy. If one believes this outlook for activity, the term premium should be much lower—around zero, or perhaps even negative.
The CBO forecasts are incoherent. They call for a subpar outlook for growth with little consequence for the configuration of interest rates. If this unhappy growth outlook turns out to be correct, we think that the new normal for interest rates will be much lower. The same problem arises, though to a lesser extent, with the FOMC and Blue Chip forecasts.
Implications for Budget Forecasts
What happens to budget forecasts if we pencil in both a shift down for the average real short rate and a shrinking of the average term premium? The next 5 years look somewhat better. And the crisis of the next 25 years all but disappears. CBO assumes that the real interest rate on debt is 30 basis points below the real ten-year yield. For 2018 and beyond, for CBO, this is around 3%.
If we keep CBOs real growth outlook, but imagine 1.25% as the steady-state real short rate, and compress the ten-year term premium to 50 basis points, then the real interest rate falls below 1.5%. Keeping the primary deficit unchanged from the most recent CBO extended outlook, we calculate that in 2038 (25 years hence) the ratio of debt held by the public to GDP will be 76%. This is little changed from its current level, and dramatically lower than in the CBO projection.