Earlier this year, fears of an economic recession became widespread, largely driven by investor wariness of what is termed an “inversion”in the yield curve. A long-standing Wall Street rule of thumb holds that when the yields offered by long-term bonds give up the typical premium they hold over interest rates on shorter-term instruments, a recession is on the way. And that narrowing of the premium happened. The spread between the yields on 10-year treasury notes and 1-year treasury paper shrank during that time from 75 basis points (that is, seventy-five hundredths of one percentage point) to only 15 basis points (fifteen hundredths of one percentage point). But before simply accepting this common perception, it pays to have a look at how reliable this yield curve signal has been in the past. The results are mixed.
Inversions have given false signals on a number of occasions. In 1995 the yields on 10-year treasuries suddenly lost the premium they had over 1-year treasuries. That was six years before the 2001-02 recession, so we cannot count it as an early warning, especially because spreads widened again during the intervening period. In 1998, the yield curve inverted for a few months, and for some weeks the 10-year treasury yield fell below the 1-year yield. But then the “normal” upward sloping curve returned in 1999. A clear recession signal had to wait until 2000, when for a number of months the 10-year yield fell more than 50 basis points below the 1-year yield before the mild 2001-02 recession took hold.
The first hint of an inversion in this century came in early 2006, and some might consider it a very early warning of the Great Recession (as it is now remembered) of 2008-09. But that would be a stretch. For one thing, the yield curve by the middle of 2006 had returned to its normal upward slope. It then reverted to an inversion later in 2006 and early 2007, only to resume its upward slope later in 2007, and it continued to get steeper into 2008, when the 10-year yield stood fully 100 basis points above the 1-year yield. That can hardly be called consistent signaling. Indeed, using those statistics, observers relying on the yield curve signal would have been calling an “all clear” just before the 2008-09 great recession began. And, indeed, some did.
The yield curve offered far clearer signals during the last two decades of the twentieth century. Strong and persistent inversions predated the 1990-91 recession as well as recessions in 1980 and 1981-82. But if we look further back, the record again becomes spotty. Inversions in 1969 and 1973 did signal the recessions of respectively 1970-71 and 1974-75, but an inversion in 1966 saw no following recession. Even if one were to make the dubious claim that it was a very early indicator of the recession that did develop four years later, it still would not explain why the yield curve assumed its normal upward slope in those intervening four years.
While this uneven history may offer evidence that yield curve inversions provide a worthy signal, the record also suggests that the investor use such signals with care. Helpful here may be research by the Federal Reserve Bank (Fed) of St. Louis, which has looked back at some 60 years of economic cycles. The Bank found guidance in its research for interpreting the yield curve. It also uncovered other indicators to improve forecasting by offering a verification of the yield curve signal. Regarding the yield curve itself, the Bank found that inversions must persist if they are to give a clear signal, and that even when they do, a recession takes, on average, 10-18 months to develop. For verification, the Fed researchers recommend three checks:
First is the number of building permits for new houses. It can confirm a recessionary signal from yield curve inversions with the same average 10-18 month lead-time. In the current environment, these data (despite the first quarter’s overall economic strength) seem to suggest at least a tentative conformation of a recessionary signal. Permits for new construction rose a slight 0.6% from March to April, the most recent month for which data are available, but the April figure is 5.0% lower than April a year ago. However, permit requests are also up almost 3.8% from the lows of last August, which might indicate that the dip was less a cyclical sign than simply the inevitable fluctuation of building permits from month to month and quarter to quarter.
Second, the Fed’s economists identified employment in construction as another verifier of the yield curve’s recessionary signal. But at the moment, the reading is hardly recessionary. For April, the Labor Department reported that the U.S. employed 7,486,000 people in construction, up 3.5% from April 2018, and a steady, if slow, rise for 2019 so far. This picture, however, does not entirely contradict recessionary readings, because, on average, this indicator lags some months after the yield curve inverts before offering a confirmation of coming economic trouble. This same lag also is true with the third confirming indicator uncovered by the Fed analysts: manufacturing employment. The Labor Department indicates 12,838,000 people working in manufacturing in April, up about 1.6% from April 2018, and this too at a steady if slow pace so far in 2019. Here as well it is hard to make a call for a recession but from a strict reading of the averages, it is also impossible to dismiss a recession out of hand.
On these bases, one cannot dismiss the recent spate of recession forecasts. They do, however, seem premature. And even if –– despite the tentative evidence available so far –– the flattened yield curve really is giving a signal for recession, the averages indicate that the slide in real economic activity would only begin to emerge by the turn of the year at the earliest and would likely not gain force until summer 2020. Though only a fool would bet on such a forecast at this stage, there are enough straws in the wind to make it worthwhile to watch the indicators offered by the St. Louis Fed carefully for a confirming sign.