Many investors are raising cash due to what are perceived to be high market valuations, the lengths of the economic expansion and market rise and slowing growth in economies around the world. Some buy-and-holders are just accumulating dividends while others are doing that plus trimming stock holdings, buying bills and bonds or even going to all cash, waiting for the inevitable “reversion-to-the-mean” correction.
Nobody knows when that reversion will come. Nobody knows how deep the correction will be. Reversion to the mean is said by some to be “the most powerful force in finance.” If true, it therefore can’t be ignored. But, reversion to what mean?
That is an important question because as we accumulate cash while awaiting the reversion, sound planning dictates we have some sense of when it would be appropriate to get back into the stock market. A simple answer might be to wait until valuations are drawn down closer to the historical average, or mean. OK, but again, what is a meaningful historical average?
I recently read an article, “No Free Lunch: Return Is Determined by Valuation,” by John Mauldin (Mauldin Economics) and Ed Easterling (Crestmont Research), the topic of which is clearly stated in the title and one I agree with wholeheartedly for entire markets and individual stocks. You can find it here.
The gist of the article can be summed up thus:
The article led me to Crestmont’s website wherein I found the following interesting graph:
My eye immediately was drawn to the S&P 500 being net-net flat from 1900 to 1940, then it looks like the take-off angle of an F-35 being catapulted off an aircraft carrier, and continuing at that up-angle for 70 years plus. Why?
I theorize that, after the long flat period, the early 1940s saw the positive economic effects of the pre-war build-up and wartime manufacturing machine; followed by the post-war baby- and housing-booms; followed by the computer era, internet era and now AI era. These last three phases, beginning around 1950, caused faster and cheaper computation capability, increased communication depth, breadth and speed and the information from all of history being instantly available to everyone through the computer in their pocket.
But, should all these great inventions cause stocks to rise faster and valuations to increase higher over the last 70 years versus a longer time frame?
Let’s look at the graph again with a few calculations using the start of the computer era in 1950 as the base:
Notice how the S&P 500, for the 50 years from 1900 to 1950, rose at a CAGR of 2.1%/year and had a P/E 10 average valuation of 13.8. Whereas, for the almost 70 years from 1950 to 2019, the S&P rose at a CAGR almost 4 times faster at 7.6%/year and had a P/E 10 average valuation 44% higher, at 19.8. Did the computer, internet and AI eras dramatically change the definition of “normal” for stock market growth and valuation?
I think it’s possible as, to cite but one example, computer-driven productivity enhancements over the last 70 years have been astonishing, what with manufacturing robots, just-in-time supply chains and the middle-man being cut out of every transaction.
Is that 44% higher P/E 10 valuation really justified, or is it because stock ownership has become ubiquitous with 401Ks/403Bs/IRAs being owned by Americans who weren’t investors before the 1970s? Or could it be because stock transaction costs have decreased from $200/trade to $4.95? Or something else? I don’t know, but I’ll be interested to read opinions to the question. P/E 10 currently sits at 30.4, more than 100% higher than the 1900 to 1950 average of 13.8 and 50% higher than the 1950 to 2019 average of 19.8. Are stock markets currently overpriced, or are we at the beginning of a new age of valuations driven by productivity enhancements caused by 4th generation robots and AI? I don't know, but I wouldn't want to be a taxi or truck driver in the 2020s and beyond.
One would think stock prices, over the long term, would go up roughly in line with earnings, multiples-changes not considered. Well, earnings in 1900 were $16 for the S&P 500, earnings in 1950 were $29, up 1.8 times in those 50 years, or a CAGR of only 1.2%. Earnings in 2018 were $135, or up 4.7 times in the previous 69 years, almost doubling the CAGR to 2.26%. So, earnings grew almost twice as fast in the more modern period versus the 1900 to 1950 period. The S&P 500, on the other hand, rose at a CAGR of 2.1%/year (twice earnings growth) from 1900 to 1950, whereas from 1950 to 2019, the S&P rose at a CAGR almost 4 times faster than 1900-1950 at 7.6%/year and 3.4 times earnings growth.
So, the modern era saw earnings grow at twice the rate of the earlier era while the S&P grew 4 times faster. It had to be that the modern era demands a higher multiple. But, justifiably so?
S&P 500 P/E (not P/E 10) averaged 13.4 from 1900 to 1950, and 17.9 from 1950 to today. The modern era therefore is producing a 34% valuation premium over the older era.
Look again at the bottom section of the above graph, where it shows P/E 10 history. From 1900 to about 1995, P/E almost never went above 25 and then only barely and for 3 short intervals. But it’s been at or above 25 most of the time in the 25 years since.
When comparing how much higher than average current valuations are and considering to what level the S&P 500 might fall in the inevitable reversion-to-the-mean, whenever that occurs, it makes a significant difference in calculating the “mean” when using the time periods from 1900 or 1950, both to today.
Today’s P/E 10 valuation is 30.37. P/E 10 average from 1900 to today is 17.3. P/E 10 average from 1950 to today is 20.3. Will the S&P 500 revert 75% relative to its 1900-2019 overvaluation mean, or revert 50% relative to its 1950-to-today overvaluation mean? I would guess 50% is more correct. A 50% valuation correction to the 2,940 all-time S&P high means a drop to 1,960 or below, when the reversion completes its work and if 2,940 remains the all-time high for this cycle. If it dropped 75% by using 1900 to 2019 valuation means, it would land at or below 1,680.
Given that 70 years is itself a long time and that the pace of change is ever accelerating, I suggest we all stop using data from 1900, or before, for valuation comparison purposes. 1950 as the basis seems right to me because it captures the start of the computer era which spawned the internet era 30 years ago, and both of which are spawning the AI and algo-driven trading eras. Data from when horses and walking were the primary mode of transportation; few homes had phones, electricity or indoor plumbing; slide rules were the only thing better than pencils and mechanization was just getting off the ground seems illogical to use in comparing to today’s anything, much less stock market valuations.
I think the productivity enhancements we’ve so far seen are but the tip of the iceberg relative to what will be seen in decades to come. Those enhancements will drive down prices and drive up profits I’m sure, recessions aside. Will they further drive up valuations like we’ve seen in comparing the 2 eras above? I think yes. The bigger question is how will companies share the spoils? Will it create bigger income and wealth gaps or be used to help solve that problem?
Only time will tell.
In the meantime, I'll continue owning T-Bills paying 2.5% and start buying back in at S&P 500 = 2,200, or 28% below current price.
Stock prices, P/Es and P/E 10s are from multpl.com.
Disclosure: I/we have no positions in any stocks mentioned, and no plans to initiate any positions within the next 72 hours. I wrote this article myself, and it expresses my own opinions. I am not receiving compensation for it (other than from Seeking Alpha). I have no business relationship with any company whose stock is mentioned in this article.