,
12 tweets,
3 min read
Read on Twitter
Yield curve inversions in America and Britain are worrying some pundits. But do these inversions actually predict recessions? (A thread: 1/12) econ.st/2K5czHf
The yield curve is said to be “inverted” when the interest rate on long-term debt falls below that of short-term debt. (2/12) 
In America, each of the past seven recessions have been preceded by a period when one part of the yield curve—the difference between ten-year and three-month interest rates—has “inverted”. There has been just one false alarm in that time, too. (3/12)
Ten-year interest rates have been below three-month interest rates in America since the end of May. That has had many people worrying that a recession is on the horizon. (4/12)
And today, the interest rate on ten-year bonds fell below that of two-year debt in both America and Britain. That has spurned another flurry of articles warning of a looming recession. (5/12)
Yet our analysis finds that this latest inversion is not, in and of itself, a newsworthy event. There are three broad reasons for this. (6/12)
First: the existing literature suggests that the spread between three-month bills and ten-year bonds is the most informative for recession spotting. See: newyorkfed.org/research/curre… (7/12)
Second: yield-curve inversions only appear to be predictive in America. Our analysis finds that in 16 other rich countries (data as far back as 1960), 51 of 95 recessions were not preceded by inversions during the previous two years. And there were 63 false alarms, too. (8/12)
Third: squashing the data into a pair of binary variables—yield curve inverted 1/0 | recession 1/0 —is way too simplistic. Better to look at the yield-curve spread and GDP growth as continuous variables. (9/12)
In doing so we find that in 15 of 17 countries yield-curve spreads correlate with changes in GDP growth the next year. A 1%-point change in spread predicts a 0.55%-point change in growth. This chart demonstrates part of that relationship: growth slows after an inversion. (10/12)
What is more, forecasters do not appear to make use of the power of the yield curve. We were able to improve consensus GDP forecasts a little by including the 10-year ~ 3-month yield spread in those forecasts. (11/12)
In practice the yield curve may have lost some of its predictive power because of central bankers’ QE schemes. And they may well have caught on, too: the US Fed cut rates two weeks ago to assuage worries about a slowdown. Watch this space to see who's right! (12/12)
