Home > XE Currency Blog > A perfect yield curve recession indicator - except for the most important times


XE Currency Blog

Topics7270 Posts7315
By New_Deal_democrat January 16, 2015 12:39 pm
  • XE Contributor
New_Deal_democrat's picture
New_Deal_democrat Posts: 547
A perfect yield curve recession indicator - except for the most important times

Sometimes I feel like the Sisyphus of economic analysis.  There is so much bad information out there, even from well-known and widely-followed sources that it is a nearly fruitless task to call it out.

But every now and then an indicator gets trotted out that is so egregiously incorrect that I can't help myself.  Such is the case with an article yesterday by Myles Udland at Business Insider, 


"One Of The Best Predictors Of Recessions Is Rapidly Approaching The Here-Comes-A-Recession Level."

"One indicator that has a perfect record of predicting recessions is creeping toward some uncomfortable levels: the yield curve.


"In an email Thursday, Deutsche Bank economist Torsten Slok circulated the following chart, showing that the yield curve was close to "inverting," meaning that the yield on shorter-dated paper would exceed that of longer-dated bonds."

There's just one little problem with using the 5 year vs. 10 year treasury yield as a "perfect indicator" of recessions:  it fails spectacularly.  Let's take a closer look.


First, here is the yield spread between 5 and 10 year treasuries over the last 10 years (red), plus for reference the S&P 500:



It's true that this yield spread did invert before the recession - for a grand total of 3 weeks in February and March 2006, by at most -0.05%.


So let's apply that to past periods.  Here's the period of the greatest bull stock market ever: 1982 to 2001:



The spreads between the 5 year and 10 year bonds hit 0.0% on a weekly basis in 1984 and 1994.  But beyond that, it also turned negative briefly by more than -0.05% in 1987 and 1998 - in each case nearly 3 years before the next recession. Not exactly "Here comes a recession," is it?

But the record is even worse when we go back to the 1960s and early 1970s, arguably the best of economic times ever for the US economy.  Here's the graph covering that period:



Once again, note that the spread between yields on the 5 and 10 year bonds inverted by more than -0.05% in 1971, at the very beginning of that economic expansion, and again 3 years before the recession.


But worst of all, the spread inverted at the beginning of 1966, only half-way through probably the best US expansion in the last 75 years!  And a full 4 years before the next recession.  And boy oh boy did it ever invert!


In other words, the indicator espoused by Slok and Udland has a perfect record - with the exception of missing half of the best expansion in living memory, and either failing to signal the worst recession in living memory (if we discount the -0.5% signal) or else including 3 other false positives throughout the 1970s, 80s, and 90s.


One final note:  even at its current levels, the yield spread between the 5 and 10 year treasuries is still among the widest of the last 60 years:



So much for "close to signalling a recession."

Paste link in email or IM