Can Bonds Tell Us Something About the Stock Market?

I was visiting with the head of fixed income at a large asset manager when I mentioned the subject of my last post, the deterioration of corporate credit in some sectors.  He noted that Baa bond spreads had been rising for a year before the stock market entered into its latest downturn. Then he raised a very interesting question, whether rising bond spreads could be an early warning signal for the equity market.  That gave me the idea for this piece. (Thank you for that insight and accompanying question.)  I explored using bond spreads as an early warning signal indicating equity corrections.  My results were mixed as you will see, but it gave me another idea.  That was to explore whether calmness in Baa corporate debt could indicate whether it might be a good time to increase equity allocations. That turned out better.

I thought that his conjecture was reasonable because, as he and I agreed, bond professionals are more sensitive to downside risk than equity managers. Managers of corporate bonds face asymmetric outcomes: If they are right, they get to collect their coupons and the bonds mature at par.  If they are wrong, the bonds plunge and they experience equity-like returns in the wrong direction. Thus, it makes sense that bond managers’ “spidey senses” might start tingling long before those of equity managers. Another important factor is that Baa bonds, the lowest rated bonds considered investment grade, tend to be issued by more cyclical companies. They should be more sensitive to the perception of changing financial conditions than their less cyclical counterparts.

R1000vBaa

Baa spreads (orange) were rising for a year before the equity market (blue) began selling off.

The great things about an idea like this are:

  1. You can test it if the data are available.
  2. You may be able to use it to help you invest if it works.

I tested the idea using Baa bond spreads (Moody’s Baa yields minus the US 10 year treasury) and the returns of the Russell 1000, a large cap index similar to the S&P500. I used weekly data going back to 1986, the longest run of weekly data I could get for the Baa spread. I guessed that if spreads moved significantly above where they had been for the previous few years then that might be a signal to underweight equities.  I defined a high spread to be one in the top 20% of the previous 3 years.  Then I set the model to cut equity exposure by 20% when the spread met my definition of high. A trial run of the strategy produced the following result:

R1vBaaHiStrat

It worked, but only because it caught the 2 big equity selloffs in the noughts, the tech wreck and the global financial crisis. There were more false positives, 10, than successful years, 5. For you numbers folks, the strategy produced 0.30% of excess return a year, tracking error of about 2% and an information ratio, excess return over the volatility needed to generate the excess return, of 0.15.  The information encompassed in high spreads may be useful, but it cannot be used automatically in this form. The type of crisis may matter.  The good years for the strategy came during US centric crises, while some of the years it did not work featured things like the Asia crisis and the European sovereign crisis. Maybe the corporate bond market does signal concern about financial problems, but a wider range of financial problems than those that will be reflected in US equity prices.

It then ocurred to me to test an opposite hypothesis, that low spreads might signal a good time to be overweight equities.   If that is true, then bond market complacency might be a proxy for clear weather in the financial world.  I tested whether overweighting equities when bond spreads lie in the bottom 20% of their 3 year range could lead to outperformance. And look Ma! That worked much better:

R1vBaaLoStrat

The low spread strategy produced positive results in 12 of the 16 years in which it was triggered. No one year dominated the returns as was the case with the high spread strategy. This strategy added an average 0.44% a year with just under 1% tracking error, resulting in an information ratio of 0.47. Typically, an equity fund manager who can sustain a 0.47 information ratio over a long time is in the 1st quartile.*

And if you are wondering whether 0.44% a year can make a difference: Behold the magic of compounding:

R1BaaLoStratCum

Investing $10,000 in the Russell 1000 in 1989 and holding it through early October 2015 would have produced $75,810.  Increasing your equity holdings by 20% when Baa spreads were in the bottom 20% of their 3 year range would have resulted in that same investment growing to $85,350. That is 12.6% or $9,540 more than you would have had without the tactical asset allocation strategy. Since the chart relects only the price change and not the dividends you would have earned, the effect of the strategy is understated.

It appears then that there is information in  Baa bond spreads that is worth considering when evaluating your tactical allocation to the US equity market. It just wasn’t the information that we thought we might find.

Disclaimer: Implementing such a strategy would be done only after considering any number of issues, trading costs not least among them.  This piece is simply an exploration of how to use data to explore possible relationships between financial and economic variables and how such relationships might be used in investing. The next time you hear someone on CNBC bloviating about how some relationship between X and Y always leads to a certain result, grab some data and test their proposition.  Let me know what you find.

  • A 0.47 information ratio would be likely to place a manager in the top quartile (best 25%), but that manager would almost certainly need to take on more tracking error (risk relative to the benchmark index) than 1%.
This entry was posted in Credit, Equities and tagged , . Bookmark the permalink.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s