Yesterday I noted that the gap between a benchmark long-term interest rate -- the yield on 10-year Treasury notes -- and the federal funds rate had, over the past year, fallen from about 3.7 percentage points to about 0.80 percentage points. In asking whether we should be worried about the shrinking spread between long-term and short-term interest rates, I linked to a post from Jim Hamilton at Econbrowser, who suggests the answer is yes, we ought to be worried:
... as I commented earlier, if we see longer term yields drop back down, that could put us in a position of an inverted yield curve, which historically has been a fairly ominous development.
We are, of course, a distance from an inversion of the yield curve -- a situation in which short-term rates actually rise above long-term rates -- and over the past several days the 10-year Treasury yield has been moving northward. Whether that trend will persist is unknown, of course, as is the issue of whether the Fed will actually continue along with it's merry measured pace of funds rate increases (although yesterday's speech by Richmond Fed president Jeff Lacker -- duly reported by Mark Thoma -- indicates that at least one FOMC participant thinks they will.)
But we can still ask this question: How far is too far? That is, what does history tell us about the association of future GDP growth and a particular level of the difference between long-term and short-term interest rates?
The following set of pictures help me think about that question. The first picture shows the distribution of the 10-year-Treasury/federal-funds-rate yield spread, focusing on times when GDP growth over the ensuing four quarters was "normal". I define "normal" GDP growth as the range from about 0.9% to about 5.4% (that is, plus-or-minus one standard deviation from the mean over the period from 1971-2003.)
The picture clearly indicates that it has not been that rare for periods of quite reasonable GDP growth to be preceded by spreads as low as, or lower than, the current level. Even outright inversions are sometimes associated with "normal" GDP growth, although those have been relatively unusual events.
On the other hand, we have for sure entered the zone that has, in the past, been associated with sagging economic growth. Here's the distribution of long-term/short-term spreads when the ensuing four-quarter growth rate of GDP was less than 0.9%:
Pretty clear picture, that. Spread over 100 basis points? No problem (if history repeats itself). Less than that? You have a few things to think about.
Just for the sake of completeness, here's the distribution of the spread when the rate of GDP growth over the subsequent four quarters exceeded 5.4% (which is technically referred to as "gangbusters"):
That's sort of an interesting graph. Rapid GDP growth has not generally been preceded by a particularly large yield spread, and a leading spread as low as one percent -- about where we are now -- has not been uncommon. There's even the odd inversion.
Where does that leave us? To paraphrase the old saw, it looks like low spreads between short-term and long-term interest rates predicted twelve of the last eight slowdowns (or something like that). These distributions mask, of course, the specific circumstances surrounding each particular realization of the long-term/short-term gap, which is just another way of saying that all else is not always equal. Consider this, from Hamilton's post:
... it appears that the Fed may be trying to replicate the "soft landing" of 1995, hoping to retrace the U-shape of that episode with the same pattern now, albeit at a lower level of interest rates. If one thinks of the long-run downward trend in nominal interest rates since 1982 as being dominated by a reduction in the long-run inflation rate, replicating the U-shape of 1991-1995 during 2001-2005, but at a lower level, might seem like an ideal strategy.
You can see the U-shape Hamilton is referring to in this picture (the shaded areas are NBER-dated recessions):
It sure looks like the present situation is similar to the 1991-1995 episode, but Jim is not convinced:
There's just one problem with that interpretation: the trough in interest rates in 2003 came not from a continuing reduction in inflation but rather from an excessive creation of liquidity on the part of the Fed. The inflation rate went up between 2001 and 2002, even as the Fed drove interest rates much lower, and inflation has continued to creep up since.
Well, maybe. Here's what CPI inflation and inflation expectations -- measured by four-quarter-ahead Blue Chip forecasts -- looked like over the 1991-1995 period. (I've obtained the Blue Chip forecasts from an incomplete collection of paper reports, so some of the data is missing.)
While it's true that the realized rate of inflation was drifting down in 1993, it is not at all obvious that this was the case with inflation expectations. Furthermore, over the course of 1994, inflation expectations took a decided tick upwards, as did realized price-level growth. It was arguably the rate increases of 1994 that brought both back into check in 1995.
Real short-term interest rates were in fact quite low in the 1992-1993 time period, just as they were when the FOMC commenced with the current string of rate increases in June 2004. Back in 1993, we spoke of "headwinds", and it was clear at the time that the federal funds rate was being held at rates that were explicitly meant to be stimulative. Long-term interest rates began to rise sharply in the latter part of 1993, prior to any moves by the FOMC, meaning that policy was becoming increasingly stimulative near the end of the year. It is not too hard to believe that the deterioration of inflation and inflation expectations in the latter part of 1994 were at least in part due to this extended period of prior accommodation.
All of this is to say that it is not clear, to me anyway, that fears of too much liquidity creation should be any greater today than they were back then. Nor is it obvious that we are any closer to a large mistake in excessive tightening. So if you liked the 1991-1995 episode, take heart.