May

11

PhilAAA rated bonds are the best corporate bonds available. BAA represent a somewhat riskier class of such bonds. So, using St. Louis Fed data monthly (series AAA and BAA) it might be instructive to look at the yield spread between the two bonds versus the monthly returns of the Dow Jones average. The yield spread was calculated as BAA - AAA yield from 1928 to the present. The spread is currently near historically high levels and thus seems especially relevant now. One way to think of the yield spread is as a measure of fear in the bond market or as a measure of the current availability of credit to riskier corporations.

The following correlations are for the yield spread lagged n months ago. Zero lag is the concurrent relationship. To be significant at the 5% level, two tailed with n =1000 the correlations should be above 6.2%.
Lag    correlation
12     0.054331609
11     0.060955747
10     0.05292927
9      0.039778526
8      0.035184767
7      0.01231181
6      0.000377968
5      0.004196862
4      0.008836528
3      0.027029827
2      0.043577867
1      0.030773782
0     -0.03036919

But what is interesting is that they are all positive at the various lags. In other words if there is a credit crunch indicated by a larger yield spread, then stocks subsequently go up. This is not what one would have expected.

Suppose one were able to obtain a newspaper for the yield spread over the next 12 months (in the future). Well the correlations between the Dow and the future yield spreads were as follows:

Lag   Correlation
1    -0.112798121
2    -0.126495119
3    -0.124710641
4    -0.096857777
5    -0.107535598
6    -0.112776958
7    -0.117648043
8    -0.10364356
9    -0.117856956
10    -0.134941986
11    -0.137242692
12    -0.148479036

They are all negative and statistically significant. Again not what was expected.

Dr. McDonnell is the author of Optimal Portfolio Modeling, Wiley, 2008

Charles Pennington asks:

Is this summary correct:

1) High (or rising?) yield spread predicts excess positive returns for stocks in the future, though not statistically significant and 2) Rising stocks predicts narrower than average yield spreads in the future?

If so, then why would you find item 2 to be "not what was expected"? I would have expected that rising stock prices would lead to lower yield spreads, as you observe.

Also, it sounds like you've used "levels" rather than "moves" for the yield spread. I would think that the yield spread is something that moves around on a fairly slow time scale, and that there would be a lot of serial correlation in its level from month to month. That would explain why the second set of 12 numbers are all about the same.

Phil McDonnell responds:

Your first and second points are correct.

It has long been held as common wisdom on Wall St. that yield spreads indicate credit conditions, and that widening spreads predict stock market problems. In this case stocks seem to predict yield spreads. As always you make a good point. I did use levels in the yield spread only, percent changes in the Dow. It is probably worth looking at the changes as well.

Dr. McDonnell is the author of Optimal Portfolio Modeling, Wiley, 2008

Bill Rafter adds:

This number was down to a -349 and has now risen to -259. Above -270 it took out the post-LEH, post-TARP periods, meaning that the credit problems associated with those events have been rendered old-news by recent behavior. It's nice corroborative information but not actionable, in my opinion.

More important than this is that the "dumb" money is still short, and still shorting the rally.

Dr. Rafter is President of Mathematical Investment Decisions, a quantitative research consultancy


Comments

Name

Email

Website

Speak your mind

Archives

Resources & Links

Search