Dividend Yield Predicts Market Returns

Research on Safe Withdrawal Rates

Moderator: hocus2004

Post Reply
Mike
*** Veteran
Posts: 278
Joined: Sun Jul 06, 2003 4:00 am

Dividend Yield Predicts Market Returns

Post by Mike »

JWR1945
***** Legend
Posts: 1697
Joined: Tue Nov 26, 2002 3:59 am
Location: Crestview, Florida

Post by JWR1945 »

This is an excellent article for those with technical backgrounds. For those without such training, I offer these remarks.

A straightforward analysis of dividend yield D/P, book-to-market value BV/P and earnings yield E/P a long, long time ago showed that all three made great predictors of future stock returns.

The next stage was a series of refinements in the statistics that consistently lowered the confidence levels associated with such predictions. The cause is called small sample bias. Eventually, almost all of the claims of predictive value were assigned to this statistical artifact and almost all of the original conclusions were rejected outright.

During this interval it was fashionable to claim that nothing in the market was predictable.

This latest study not only reinstates the original conclusions, it strengthens them. There is considerable value in stock market predictions based on dividend yield D/P, Book Value to Market BV/P and earnings yield E/P. The newest statistical refinements, at least in terms of dividend yield, actually increase the confidence levels associated with predictions beyond those originally claimed and they are much, much higher than the interim tests had indicated. In terms of predictive merit, the best indicator is dividend yield, followed by Book Value to Market and finally earnings yield.

The earnings yield in this case uses a single year of earnings. It is almost certain that using the ten year average of earnings E10 (in P/E10) would do a better job.

How can this happen? It is necessary to make a series of assumptions whenever one makes a statistical analysis. Most of them are benign, but sometimes they only seem to be benign. In this case, the analysis during the intermediate period treated the randomness of dividend yields and predictions made from dividend yields almost independently. There are, in fact, important constraints on predictions when dividend yields remain stable. They are so important as to cause the radical changes indicated.

There remain a series of assumptions in the new analysis that are only rough approximations, appropriate only because they are needed to make calculations possible.

I think that the emphasis of researchers has had a great deal of influence in terms of the history. These days, researchers are looking for powerful statistical tests. They are no longer satisfied when they fail to detect something. Previously, if someone said that he did not see anything, that was considered a valuable finding. People did not even ask what it would have taken for something to be noticed.

Have fun.

John R.
Mike
*** Veteran
Posts: 278
Joined: Sun Jul 06, 2003 4:00 am

Post by Mike »

Thanks for the summary John.
Post Reply