Felix Salmon smackdown watch, earnings-yield edition
The economic historian James Macdonald emails with a very different version of my earnings-yield chart. Instead of going back to 1985, he’s found annual data all the way back to 1920. And instead of using Treasury bonds — a sui generis asset class with its own dynamics surrounding flight to quality and the like — he’s uses the yield on BAA-rated bonds instead. The result is fascinating:
It is not reasonable to relate earnings yields to T-bonds. They are not comparable investments. The proper comparison is with the corporate bond yields. My preference is not even AAA, since that is quite rarefied, but rather BAA – definitely investment grade but more reflective of the typical corporation and more susceptible to market jitters.
The historical pattern seems to me to be as follows:
- a twenty-year period from 1920-1940 when bond and earnings yields were more or less comparable
- a twenty-year period from 1940 to 1960 when earnings yields were significantly higher than bond yields
- a twenty year period from 1960 to 1980 when the two yields were more or less comparable
- a twenty-five year period (?) from 1980 to 2005 when earnings yields were significantly lower than bond yields. (I put a question mark, because there is also a major divergence of yields in 2009. However, I take that as a temporary aberration because of the collapse of S&P earnings during the Great Recession thanks to the losses of the financial sector).
Now the two yields are more or less the same. Are we supposed to take that as a signal to sell bonds and buy shares?
I am not at all convinced, at least on the basis of the historical evidence. We may be entering a long period when the two yields track each other. Or we may be entering a period (perhaps to be expected after a major financial disaster) when investors demand a premium yield from shares compared to bonds. In any case, the period from 1985 onwards can scarcely be taken as the historical norm, since, as we know from other statistics, such as long-term PE ratios, shares have been overvalued for almost the whole of this period.
My feeling is that a long-term period when investors demand a premium yield from shares compared to bonds is precisely the kind of period that a long-term investor wants to be putting her money into shares rather than bonds. Again, there are market timing issues here — shares might fall rather than rise for much of that period, especially during times when the premium is rising rather than falling. But if you just go back to basics and look at securities as giving you ownership of an income stream, then it seems perfectly sensible to me to pick the larger stream, given the choice.
If there’s a big risk to this strategy, it seems to me, the risk is overleveraged companies. If companies get in over their head and find themselves unable to make their interest payments, then their equity can be wiped out, and ownership transferred to bondholders. But in this particular economy, I don’t see that as a big risk — certainly not for BAA-rated companies. We saw during the financial crisis that nearly all non-financial public companies had been decidedly conservative with respect to the big debt binge known as the Great Moderation; if you were looking for massive corporate leverage, you needed to look to private equity deals.
And what if we’re just entering another period akin to the 60s and 70s, where the two yields track each other while rising? Again, I think I’d rather be in stocks, if only because they’ll do much better than bonds if and when inflation kicks in.
Finally, there’s the possibility I raised in my original chart — that stock prices are uncommonly cheap relatively to bond prices right now. Which certainly seems to be the case if you think that history began in 1985. If stocks revert to the post-1985 norm of yielding less than bonds, then moving from bonds to stocks at these levels would probably make a fair amount of sense on a mark-to-market basis. As a long-term investment, though, stocks might look a bit less attractive: you’d be less reliant on earnings and more reliant on capital gains to make up the difference.
What really surprises me about Macdonald’s chart, however, is nothing to do with the difference between stock yields and corporate bond yields, but rather the way in which nominal corporate bond yields move very slowly and smoothly. There does seem to be something reassuringly safe and reliable in that orange line. So if your priority is preservation of nominal capital, rather than maximizing your long-term net wealth, there’s something pretty attractive about corporate bonds, too.
Update: I missed this on Friday, but Jake from EconomPic Data has his own version of the chart, using T-bonds but going back even further than Macdonald does. I love this:
What happened in 1970 or thereabouts that this correlation suddenly became so strong, after barely existing before that? Could it possibly be the arrival of Modern Monetary Theory and Modigliani-Miller?