U.S. and German 2y10y Spreads Look Ready to Converge

The relative steepness between German and U.S. 2-year 10-year spreads is the widest (58 bps) since mid-2006. A swift streak of inflation surprise beats across the Eurozone have promoted a steeper German yield curve since October 2016.

We produce 50 business day forecasts (generalized additive model) for both spreads using data since 2013 from the following:

  • U.S. and Eurozone hard economic data (e.g. GDP and industrial production)
  • U.S. and Eurozone soft economic data (e.g. consumer confidence)
  • Inflation surprises
  • Swaption volatility (1Y x 1Y)
  • # of rate hikes expected over next 12 months by ECB and FOMC

Rate hike timing and soft economic data changes show the greatest significance to forecasting the U.S. yield curve. The German yield curve has been most impacted by soft and hard economic data.

Forecasts are shown below as darker lines with the German curve flattening from current of 117.1 to 105,9 bps and the U.S. curve steepening from 59.5 to 102.5 bps. We expect German flatteners become back in vogue near the steepest spreads seen in July 2017 at 122.4 bps. All in all, the German and U.S. yield curves appear ready to converge.

The bottom panel in the next chart shows the Eurozone (blue) and U.S. (orange) Citigroup Inflation Surprise Indices. The top panel provides a highlighted German 2-year 10-year spread based on when Eurozone inflation surprises are outpacing the U.S. or not.

Eurozone inflation beats have been waning in recent months. A break below zero by its index and potentially matching or falling below the U.S. have consistently been met with a flattening German yield curve (orange highlights).

Hard economic data like GDP and industrial production enjoyed a swift rebound into the end of 2017 for both the U.S. (orange) and Eurozone (blue). However both are butting up against past peaks, suggesting tempering hard data in the months ahead.

But, be aware the Eurozone’s hard data have shown a stronger propensity to persist at above average growth rates. Continued Eurozone improvement will only produce further hawkish commentary. Draghi et al are expected to announce an eventual end (September?) to bond purchases in June. Tailwinds from strong hard data would only make this decision easier for both the central bank and markets.

Conversely, a pull-back in U.S. hard data would rattle financial markets still hesitant to chase the lofty tightening expectations of the Federal Reserve. Any dent in tightening expectations has potential to lead to bull flattening by U.S. treasuries.

The seasonal outlook for Q1 economic data in the U.S. is quite negative, while the Eurozone has tended to see healthy improvements. Our model shown above puts a lot of weight on a negatively correlated relationship between the U.S. yield curve and soft economic data.

 

The U.S. treasury yield curve very much abides by cycles in economic data, specifically soft data. We provide one and three-year moving averages of soft data since 1977 in the bottom panel of the chart below. Be aware the axis has been reversed with lower index values indicating improving data.

The top panel shows the US 2-year 10-year spread with blue shading marking periods when soft data is improving as determined by moving average crosses. These periods of better data are dominated by swift flattening. Conversely, periods of worsening data (red shading) see intense steepening. The fact we have to slow down the soft data index with moving averages suggests its leads shifts in the yield curve.

A set-back in optimism has been a bellwether of a less hawkish Federal Reserve and a steeper U.S. treasury yield curve.

Lastly, we noted earlier this month an end to the negative correlation between inflation expectations and the VIX (S&P 500 implied volatility). A continued increasingly positive correlation would indicate a stock market becoming fearful of inflation. Additionally, past shifts in correlation between inflation expectations and the VIX led to swiftly steeper curve pre-crisis.

You may also like...