This advice seems to contradict the data. After all, there is a positive relationship between equity valuations–whether measured against earnings or dividends–and future returns. The higher the dividend yield in one year the stronger the expected returns over the following five years.
After all, more than a century of data show this to be true. Sure, the statistical significance could be stronger, but the general tendency is clear.
And if you had bought U.S. equities when price-to-dividend ratios were below 14, you’d have generated annual total returns of more than 10% a year during the following ten years, three times higher than had you bought when the price-to-dividend ratio was above 35.
This seems a pretty good investment rule–buy when cheap; sell when dear.
Except it’s not.
The problem with this data analysis, argues Professor Dimson, is that it is done with hindsight. Using the full century and more of data, we know how things turned out during the whole of that period. We know what was cheap and what was dear.
The LBS researchers ran the same analysis, buy low sell high, over the whole of the past century, but using only data prior to the point in time they were looking at (a rolling analysis). In effect, they were looking at history as if at each point they were in the same position as contemporaneous investors, with only knowledge of past data and not what was to come.
In that case, timing the markets based on mean reversion in all cases causes investors to underperform those who merely stay in equities. Trying to time the market by selling high and buying low leads to worse returns than methods like dollar cost averaging or just buy and hold.
-from WSJ
Categories