Discerning the Value of Market Dividend Yields

Blogging is often a cooperative venture, so this piece begins with thanks to three people:

• James Picerno of The Capital Spectator, who gave me the idea for this piece in his article, What, If Anything, Can Dividend Yield Tell Us?
• Eddy Elfenbein, who saw a question that I posted on Bespoke’s website, and sent me a link to
• Dr. Robert Shiller’s website, where I got the data for this post.

When I read the piece at The Capital Spectator, my response was “Huh, neat article, wonder how it would look with a larger data set?”  Given Eddy’s help, I had that data set, and so I got to work.  Here is the main result:

The results at The Capital Spectator went from 1995 to 2003. My results go from 1871 to 2003. His results give a tight relationship, while mine indicate a loose relationship.  His R-squared was far  higher than my 7%.  Why?

In aggregate, many relationships in finance are tenuous.  Do interest rates mean-revert?  Yes, but the tendency is weak.  In this case, high dividend yields foreshadow high five-year total returns, but that tendency is weak.

In the graph above I tried to highlight the eras for different alignments of dividend yields and future five-year returns.  Depending on the era, the relationship of dividend yield to future returns differed.  In the long run, there is a weak positive relationship between dividend yields and total returns, but in the short run, many other factors predominate.

So what does this tell us?

• Use larger data sets when possible.
• Realize that many relationships in finance are not stable.  Indeed, that is a strength of Capitalism.  It adjusts to changing conditions, and is not stable.
• When dividend yields are high the market is attractive.  Of course, factor in how high bond and cash yields are at the time.
• Beware relying on intermediate-term relationships in quantitative finance.  They last for less than a decade.
• Beware trusting correlation coefficients calculated over short intervals.
• In finance, we know less than we think, so we should be cautious in our conclusions.
• The best forecasts come when we are at extreme values of the system.  In the middle, everything is a muddle.

I am a firm believer in dividends. My portfolio has an above average dividend yield.  In general, high dividend yields pay off in investing, subject to credit quality.  But, the payoff varies over time; a heavy reliance on the dividend yield of the market as a sole indicator is not advised.

Comments

[Chris B:]

Yes, dividends are nice, but I have to question the sincerity of the banks who are taking government bailout money and yet not eliminating their dividends. Can GE and GS really afford to hand out dividends at a time when they are paying Warren Buffet 10% interest to borrow equivalent cash? Is doing so really maximizing shareholder value? Can the shareholders earn 10% elsewhere risk free?

GM's effective borrowing cost is approaching 30%, but yet they still borrow that money to pay a dividend. It's like some companies know they are bankrupt and just want to accumulate as much debt as possible and pay as many dividends as possible before it inevitably happens.

2008 Oct 30 01:17 PM

[Dividends A...:]

Thanks for the informative article David.

2008 Oct 30 09:16 PM

[THofler:]

Chris B: GM hasn't paid a dividend in many months (they've been zeroed out for some time) and do you consider what GS pays a dividend? The payout ratio for GS is now 8% & I think it was more like 5% back when they had good earnings.

2008 Oct 30 11:40 PM