Recently, the US yield curve has flattened dramatically. As a consequence, many market participants started to wonder whether this movement was pre-announcing a recession in the United States.
First, let us make a historical comparison.The slope of the 2y-30y segment hit 100 bps this week, what has not happened since November 2007. The last time that the slope of the 2y-30y segment hit 100 bps during a monetary tightening cycle was in April 2005. From this point, it took:
- 10 months for this segment of the curve to be completely flat (February 2006)
- 14 months for the unemployment rate to hit the bottom (June 2006)
- 14 months for the Fed to make the last fed funds hike of the cycle (June 2006)
- 2.5 years before the S&P500 hits the peak of this cycle (October 2007)
- 2.6 years before the US economy experiences a recession (December 2007)
Note that we would have obtained almost exactly the same dating with the slope of the 2y-10y segment and a threshold of 60 bps.
Second, we can doubt of the quality of the signal sent by the slope of the yield curve. For long, academic works have shown that the slope of the yield curve contains useful information for signaling future recessions. Rudebusch and Williams (2006) have even shown that the slope of the yield curve is more accurate in forecasting recession at horizons of one year than professional forecasters. However, Ang, Piazzesi & Wei (2006) have found that nominal short rate dominates the slope of the yield curve in forecasting GDP growth both in- and out-of-sample. Wright (2006) has confirmed these findings and has concluded that “forecasting recessions that use both the level of the federal funds rate and the term spread give better predictive performance, than models with the term spread alone.”
Having this in mind, we should highlight two points:
- The yield curve has flattened but fed funds remains at historically low levels: the fed funds target range is 1.00%/1.25% today while the terminal rate of the fed funds during the previous cycle was 5.25%.
- The yield curve has flattened as technical factors continue to weigh heavily on long-term interest rates. According to the Fed staff, the Fed’s quantitative easing programs imply that the 10y. term premium was 88 bps lower than where it should be in September 2017 (Bonis, Ihrig & Wei, 2017). Even if it is a purely theoretical exercise, we can think that the yield curve would be much steeper without the Fed’s extra- US Treasuries holdings. The San Francisco Fed’s economist Michael Bauer recently evoked recently a “new conundrum in the bond market” as “a lower ‘neutral’ interest rate, the risk of persistently low inflation, and fiscal and geopolitical uncertainty—may account for the yield curve flattening”.
Third, the flattening of the US yield curve has probably been too far too fast recently.The curve has flattened as markets started to revise up their fed funds expectations for 2018 from early-September. Consequently, the short-end of the curve has risen, more than the long-end. It is perfectly common to observe a flattening of the yield curve when the central bank raises its key rates and this is that will happen in 2018 but the recent move has been too marked. A basic macro model of the slope of the yield curve (using central bank rates, the output gap and the inflation gap) indicates that the curve should be steeper at this stage of the cycle. Despite our view that the US yield curve should flatten in 2018, it may be relevant to play the steepening in the short-run, especially for the 5y. – 30y. segment.
Ang, Piazzesi & Wei, 2006, “What does the yield curve tell us about GDP growth?”, Journal of Econometrics.
Bauer, 2017, “A new conundrum in the bond market?”, FRBSF Economic Letter.
Bonis, Ihrig & Wei, 2017, “Projected evolution of the SOMA portfolio and the 10-year Treasury Term Premium Effect”, FEDS Notes.
Rudebusch and Williams, 2006, “Forecasting Recessions: the Puzzle of the Enduring Power of the Yield Curve”, Journal of Business & Economic Statistics.