Expensive High Yield

I’ve seen a number of articles recently arguing that high yield bonds are still cheap. Today I began an investigation to analyze this claim.

Here’s my bias: at the first investment shop I worked in, the high yield manager told me that there is a nominal yield for high yield bonds which reflects the risk.  It doesn’t matter where Treasury yields are, high yield bonds don’t care.  As a result, when people in the media, or writing blogs those argue that high yield is cheap because yield spreads are wide, it is time to disregard then when Treasury yields are artificially low, because of government interference.  (Financial Repression)

High yield bonds do care about credit conditions.  High yield bonds do care about the stock market.  From all of my research, high yield bonds are highly sensitive to credit conditions, particularly those of its industry.  They are also sensitive to the stock market.  After all, if the high yield bonds are doing badly, the stock is doing worse.

And here’s the rub: high yield bonds do not react to yields on Treasuries, except negatively, because when Treasuries rally hard, times are not good, and high yield bonds do poorly, with yields rising.

Here’s a graph to show how yields have done over the last 15 years for various corporate bond ratings.

My data this evening comes from the Federal Reserve Bank of St. Louis’ website FRED.  Been using it for 20 years, it is one of the best economic data repositories on the web.  Even used it during the bulletin board era, pre-web.

Merill Lynch has recently provided many of its bond yield indexes to FRED.  Previously, all that was there were two long yield series from Moody’s.

Now, if the concept of yield spreads is valid, when I do regressions of treasury yields on corporate index yields, I should see tight correlations of the yields versus Treasuries, and beta coefficients near one.  Here’s what I obtained:

AAA-CCC refer to ratings categories.  HYM is High Yield Master II, which is an average of high yield bond yields, and is usually very close to single-B yields, no surprise.

As you will note, spreads work reasonably to poorly for investment grade bonds.  The yields on investment grade bonds do not fall as much as yields on Treasury bonds do.  The yields on high yield bonds are barely affected when Treasury yields fall.  Look at the R-squareds on the regressions versus Treasuries only, high yield bonds do not have any economically significant relation ship to Treasuries alone.

Thus, it doesn’t make sense to talk about high yield bonds in terms of spreads over Treasuries.  High yield bonds react more to lending conditions, and derivatively, how well the stock market is doing.

But if we introduce credit spreads into the analysis, everything changes, and R-squareds skyrocket.

To me, BBB bonds are the touchstone for credit conditions.  Why?  They are on the edge of investment-grade creditworthiness.  They are also a large part of the corporate bond market.  When their yields rise or fall, it is a sign that financing rates for corporations are changing.

So, when I did regressions including BBB yields in addition to 5-year Treasury yields, guess what?

  • Junk yields were highly geared to BBB yields.
  • When Treasury yields fall, junk yields rise, and vice-versa.
  • These relationships are in general more statistically significant than those of high investment grade corporates versus Treasuries.

So what does this prove?

  • Yield spreads over Treasuries are not a good way to define value in bonds, and particularly not junk bonds.
  • Better to analyze high yield bonds versus BBB bond yields, and consider Treasury yields as a negative factor when rates are low.

So, is high yield cheap or dear at present?

Whether I look at the Merrill High Yield Master 2, BBs, or Bs, junk bonds look expensive.  CCCs look a little cheap.  The yields on the High Yield Master 2 look about 0.8% expensive in terms of yield (that’s the residual in the above graph).  I will be lightening credit bond/loan positions in the near term.  Of course this is just my opinion, so do your own due diligence.

And, please realize that movements in the stock market may swamp my observations.  If the stock market runs, high yield can run further… but there will be an eventual snap-back.   The bond market is bigger than the stock market, eventually the stock market reacts to bond market realities.

1 Comment

  • ljoneill says:

    David, good and interesting analysis. This is another “keeper” to keep in the toolkit for analyzing various opportunities.

    Using the residual line in the graph, what seems clear to me on first blush is that there is a lot of fuzziness around the midline. When you look at the HY Master II returns over the past 15 years, here are the years with double digit returns (which should qualify as a solid return!): 1997 (13.27%), 2003 (28.15%), 2004 (10.87%), 2006 (11.77%), 2009 (57.51%), and 2010 (15.19%). Conversely, the BAD years were really limited to 2000 (-5.12%) and of course 2008 (-26.39%). The really good 1997 year wouldn’t have been predicted by the residual, since it was already looking marginally expensive at the start of the year. The residuals starting looking VERY cheap in late 2000, which was early but set you up for the great run of 2003-2006. But yet the residual crossed the midline again in mid-late 2004, so that was very early. I guess the point is that maybe you should only implement a major tilt in a balanced portfolio when the residual has gotten to really significant levels. And then use momentum in asset class returns to capture the run until it reaches too far the other direction. The squishy middle should probably just be ignored, it seems to me. Maybe you could put a 1-2x standard deviation test on it so you only put a major tilt on at a z-score of 1.5 or higher…or something like that. Nonetheless, great analysis.

4 Trackbacks