Friday, August 12, 2011

Is The Earnings-Yield Divergence Unprecedented?

I'm a big fan of Felix Salmon (you will notice he is on my blogroll), but he dropped the ball in this post outlining the "unprecedented" divergence between the earnings yield of the S&P 500 and the ten year Treasury.:
After I wrote my post on Monday about the huge divergence in yields between stocks and bonds, I wondered just how historically unprecedented this divergence was. And now, with the help of this fabulous chart (many thanks to Nick Rizzo, Dan Burns, and Stephen Culp), it’s pretty easy to see: we’re at levels which match those at the height of the financial crisis, and which are otherwise historically utterly unprecedented.
Indeed, from 1985 through about 2002, it was just as common for the S&P earnings yield to be lower than the Treasury yield as it was for the yields to be the other way around. The two tracked each other, and the spread between them almost never moved beyond 2 percentage points either way.
Unprecedented? No...

As I outlined in a post a bit more than a year ago, the relationship between earnings yield and Treasuries is a new phenomenon (on a relative basis). If you looked past 1985 (i.e. the time in Felix' research), you would see a strong relationship going back just another 15 years or so. Before that... nothing for another 100 years.



I have no issue with an investor using this indicator to prove there is value in equities. BUT, they must believe something drastically changed around 1970 and that change remains. Otherwise, "history" indicates the strong relationship over the last 40 years may be what was unprecedented.

18 comments:

  1. Is it possible that the rise of arbitrage trading -- using computerized algorithms and real-time data -- is the reason your pre-1975 charts are not an accurate reflection of modern markets?

    ReplyDelete
  2. I think arb trading wasn't really in full scale until the late 1980's / early 1990's, but I could be wrong. What I do know is that back in the day is was thought that it was an impossibility that stocks could earn less than Treasuries because they were more risky. Apparently investors assuming that was a temporary phenomenon blew themselves up in the 1950's.

    Some other thoughts of what made equities more attractive:

    * Demographics (i.e. baby boomers - younger folk have a higher demand for equities)
    * The growth of retail investing (i.e good marketing)
    * Inflation concerns beginning in the 1970's

    ReplyDelete
  3. I should have clarified... if you're looking for a rise of the tight coupling seen starting in the 1970's then algorithms based on Markowitz's work is probably the reason. MPT didn't become popular until the 70s, but once it did, it created what you're seeing in the chart.

    MPT and Markowitz's work were then used by arbitrageurs (who also used Black & Scholes' work) to price everything 'efficiently'. Of course, the efficiency of it only made sure that option prices or that portfolio's hewed to the standards set forth by mathemeticians.

    ReplyDelete
  4. Got it... that makes a lot of sense.

    ReplyDelete
  5. So... if Markowitz and CAPM are debunked, the relationship could in theory break down. Interesting....

    ReplyDelete
  6. Debunked? I dunno, maybe.

    Taleb seems to think so. He thinks they were never bunked in the first place and has said that MPT and the like have made the system inherently more unstable.

    ReplyDelete
  7. NIck R -- Kyoto, JapanAugust 13, 2011 at 8:47 AM

    It's not just arbitrage trading that drives the modern relationship. Corporations are far more likely to buy back stock today as well as engage in LBOs, MBOs, private equity spinoffs, junk bond financing, etc. This is the "real arb" that prevents the equity/bond yield differential from blowing out in liquidity trap conditions.

    This corporate behavior was largely absent pre-1970. Interesting, this real arb is far less common in Japan, and Japan's spread is far more reminiscent of the 1930s.

    ReplyDelete
  8. In other words, if a corporation can borrow at 5% and get a return of 6% on their stock buyback... do it.

    ReplyDelete
  9. No, but Walmart is borrowing long term debt at 3% and buying back its stock with an earnings yield of 10%.

    GE, Radio Corp of America, and US Steel could have done the same thing in 1936, but they were unaware of the advantage.

    ReplyDelete
  10. I bet he means unprecedented since the US dollar became a fiat currency. I am having trouble finding it, but it would be interesting to see the trading volumes over that historical range. Do they start to rise shortly before the US went off the gold standard?

    ReplyDelete
  11. All sorts of reasons, good ones already put forward, and here are some more:

    a) removal of gold convertibility
    b) rise of pension & insurance funds who require stable returns, but higher than the absolute terms the risk-free market can offer (particularly now with ZIRP)

    As for MPT debunked, it's provided increasingly better returns in this decade & the last as bonds have smoothed out the mammoth equity volatility.


    -M

    ReplyDelete
  12. Oh, BTW, the link you give to the Felix Salmon post you're commenting on is broken. Presumably you meant this.

    ReplyDelete
  13. Whatever the explanation for the recent convergence, using the term "historically utterly unprecedented" to refer to the last 40 years is a very serious error for a professional journalist. Felix dropped his pants on this one.

    ReplyDelete
  14. "Unprecedented" is too strong a word for Felix. But OTOH one shouldn't dismiss a 40-year relationship too glibly.

    Lots of things happened circa 1970. For one that comes to mind right away, Nixon closed the gold window. That had a signifcant permanent impact on interest rate behavior.

    ReplyDelete
  15. The historical Spread Earnings Yield - 10y Treasuries stands at 2.84. Now the 10y stands at 2.1739. So, a normal Earnings yield should be NOW 5.01 (or a Price Earnings of 20). Which should put S&P at 1623. Instead S&P 500 trades at 1191.60 NOW. All this means stocks are undervalued by roughly 27% as we speak.

    ReplyDelete
  16. Three things:

    * Earnings are not guaranteed (they may grow, they may fall)
    * A P/E of 20 is very high by historical standards
    * This assumes the 10 year yield stays at current levels (forward markets are it to rise)

    ReplyDelete