Tuesday, December 19, 2017

The bond yield conundrum revisited: narrowing corporate spreads vs. a flattening Treasury yield curve


 - by New Deal democrat

Two introductory notes: first of all, next week is the last week of the year including the Christmas holiday, and there will be almost no economic data.  So, light posting, probably including a final update of my "Five graphs for 2017" as well as marking my 2017 forecast to market over at XE.com.

Secondly, in the coming weeks, I anticipate having much to say about the bond market, as I have done a great deal of examining its signals in the background. This is because, while a yield curve inversion has always been bad news, in times low very low interest rates, like the 1930s through mid-1950s, recessions including at least one very bad one (1938) have occurred without an inversion ever occurring.

Turning to my focus today, in my most recent "Weekly Indicators" column, I noted that the crosscurrents in bond yields. BAA  corporate bond yields, but not AAA rated yields, made a new 12 month low, and were just 0.02% above their 50 year lows from July 2016 -- a very bullish sign.  Meanwhile longer term US Treasury bonds have been meandering generally sideways for the last year. This has driven the spread between Treasuries and BAA corporate bonds to a new expansion low.

All of this is against a backdrop of a tightening, but not inverted, yield curve.

This is a very curious set of circumstances, so I went looking to see if it was unique, or something that had happened before. The bottom line is that it has not been unique, although the iterations, over nearly 100 years of data (!) have not been uniform.

On a broader scale, there have been a number of bond market markers that don't include an inverted yield curve, that have typically foreshadowed an economic downturn. Those will be explored in coming weeks.

But on the narrow issue, let me begin by comparing the spread between the more sensitive BAA corporate bonds and 10 year Treasuries, going back 30+ years (blue), and comparing that with the Fed funds rate (red):



Here is the same graph for the entirety of the 1980s:


In general the spread between corporates and Treasuries has declined as the economy strengthens, and risen as it decelerates or weakens. Meanwhile, at some point if the economy is strong enough, the Fed begins a tightening  cycle. In each of the last 4 expansions, the bottom in the yield spread has occurred during the period that the Fed is tightening: 1989, 1994, 1999 (secondarily), early 2005, and now.

When spreads turned higher from these lows, that was a clear sign that the Fed had weakened the economy -- although, in the case of 1994, there was a "soft landing" in 1995 followed by a boom.

Here are breakout graphs for each of the last 4 expansions, comparing BAA yields (blue) , AAA
yields (red), and Treasury yields (green).








Note that the current situation is just like that of early 2005. Then, as now, BAA but not AAA bonds made a new low, in the face of flattish Treasury yields (this was part of Greenspan's "conundrum").
In the two earlier expansions of the 1980s and 1990s, the narrowing of spreads was accomplished via a period of flat Treasury yields (1986, early 1998) with declining corporate bond yields.

I strongly suspect that the current move is due to a belief that corporate profits are about to improve substantially in the wake of the GOP tax bill, particularly aiding less creditworthy companies. I also suspect that this may be a "buy the rumor, sell the news" type of move, where spreads do not further decline significantly, once the Act actually takes effect.

In other words, the decline in BAA yields, and the according decline in spreads is an unequivocal good sign for the present and the short term future as long as one year out. It is only a harbinger of longer term trouble if BAA yields, and spreads, begin to turn back up.

But does this hold up upon examination of earlier economic cycles? As I pointed out at the outset, we have decent data going back all the way to the 1920s.  So I'll look at that next.