I have seen many “Average P/E’s” published over the years and there has been a fairly wide variance between those I have seen, so I began to look into them more deeply. Some of the reasons I found for this variance included: the index used, the sample time period, the time basis of the sample P/E (i.e. yearly, quarterly or monthly) and any adjustments that were made to the data.
Many of those that I have found the methodology for omitted several years of data that they claimed would have skewed the data due to the high P/E’s at the time, yet didn’t throw out any of the extreme lows during these time periods, only the highs. These included the highest average P/E’s I have seen for the Dow Jones Industrial Average.
It can be argued that deleting the extreme highs “skewed the data,” especially if it is supposed to be an average P/E. The data would be more creditable if it had offset these deletions with an equal number of extreme lows, but in either case it is an adjusted average P/E, not the average P/E over the time period.
There are also average P/E’s floating around that use Robert Shiller’s data as a basis. I find Shiller’s data very interesting, but he uses an S&P Composite in his data that goes back further than the first S&P index. This data is subjective, it is not the actual data and although I think it is very useful, I would not use it for anything but study as my research leads me to believe there are several flaws in this data due to these subjective accounts.
I found several that claim to be the average of the S&P 500 since 1945 or before. I tend to distrust anything that claims to be the average P/E of the S&P 500 that starts before 1957. The S&P 500 did not exist before March 4, 1957, any numbers they used prior to 1957 were likely estimates based on the S&P 90 (I had mistakenly reported this as the S&P 16 in some previous articles).
The P/E’s prior to those of the actual S&P 500 are likely too low. Growth companies tend to trade at higher P/E’s, the companies that were in the early indexes were the blue chips of the day, and undoubtedly traded at lower P/E’s.
For example, right now the lowest 90 P/E companies in the S&P 500 have a combined un-weighted TTM P/E of 9.24. The highest 90 P/E companies in the S&P 500 have a combined un-weighted TTM P/E of 33.52. Adjusting the earnings of 500 companies to fit either set would certainly skew the data, wouldn’t it?
Something I found interesting about the two sets of 90 stocks used above. The lowest P/E stocks have a total of over twice the earnings of the highest 90 P/E stocks in the index, but the total price of the highest 90 is nearly twice of that of the lowest P/E set. Why is that?
There are some exceptions in both sets of course, but many of the lower P/E stocks are what would be considered the Blue Chips of the index. Older larger companies with relatively stable earnings and that often pay good dividends, but that probably aren’t going to grow as fast as smaller companies anymore. Many of the highest P/E stocks in the index are what would be considered the growth stocks of the index. Many pay little or no dividend, but have much higher growth potentials than the blue chips. These stocks tend to trade with higher P/E’s and they are pulling the overall index P/E higher with them.
Look at the P/E of the Russell 2000, it is over 30. Most in this index are considered growth companies. The expansion of the S&P index to 500 companies brought a great deal of these growth companies into the index, and it resulted in a higher P/E. Comparing the modern S&P 500 to a long term average that has been adjusted to those with lower P/E’s is not particle.
Another problem with using the S&P 90 is it consisted of 50 Industrials, 20 Railroads and 20 Utilities, not a very diverse index and nothing near the S&P 500 of today. There are 60 Industrials (all three of the index’s remaining Railroads are now part of this sector) and 31 Utilities in the current S&P 500, these sectors make up only 18.2% of the S&P 500. They are also currently trading with a lower un-weighted TTM P/E than the full index, 15.97 as compared to 16.93, and a lower un-weighted forward P/E 13.81 as compared to 14.12. Although I have not completed my work on this, it appears to be the norm.
Using the historical data of the Dow Jones as comparison data gives similar problems; it began as only Railroad stocks in the hay day of their existence. It also increased the number of components over time, and diversified these companies greatly.
There are many other reasons that comparing past P/E’s to the present can be misleading. These include changes to tax codes, GAAP practices, and legislation like Sarbanes/Oxley. Based on constant earnings, most of these changes have caused the reported P/E to increase fairly dramatically over the years, making historical P/E comparisons apples to oranges.
I will continue to look into the historical average PE’s, but it is quite possible the “Average P/E” we have used as a basis for so long is nowhere near “Average” but adjusted much lower due to those deciding years with high P/E’s should not be included in this average or through the use of estimated data that does not reflect the S&P 500 or Dow properly, since it is basing this estimate on companies with historically lower P/E’s. It has also not been adjusted higher for the changes made to the earnings that companies can report over time.
I use comparisons to historical P/E averages in my articles, and for my investments, but it is important to realize the comparison data is probably lower than it should be.
Many of these sources of information were used in this article.
Have a great day trading,
Disclosure: I am currently about 89% invested long in stocks in my trading accounts.
This article is intended to provoke thought about investment possibilities. Acting on the information provided is at your own risk. You are urged to do your own research, and where appropriate, seek professional investment advice before acting on any information contained in these articles.