I recently came across an excellent article by Todd Tresidder discussing the many imperfections of the 4% withdrawal “rule” for retirement planning. In the article, one of the suggestions Todd makes is to incorporate market valuation levels (as measured by PE10) into retirement planning decisions.
Because using market valuation (most commonly measured as PE10) as an input in investing decisions is a topic that comes up fairly often, I thought it might be helpful to share my thoughts here on the blog.
What is PE10?
PE10 is calculated as the current market price of the S&P 500, divided by the average of the last ten years of inflation-adjusted earnings for the S&P 500.
Like regular P/E ratios, the idea is that it can be used as an indicator of whether the current price (of the S&P 500, in this case) is high or low relative to earnings. The purpose of using ten years of earnings rather than just one year is to eliminate the impact of meaningless short-term fluctuations in earnings.
Uses of PE10: Asset Allocation and Withdrawal Rates
The most common suggested use of PE10 is to use it to make asset allocation decisions. For example, researcher Wade Pfau wrote a fascinating paper showing that from 1871-2009, a market timing strategy using PE10 (i.e., moving to stocks when the market is at a low PE10 and moving to Treasury bonds when the market is at a high PE10) would have significantly outperformed a simple portfolio with a fixed 50/50 stock/bond allocation.
A second, related use of PE10 is to use it as an input when deciding how much you can safely spend from your portfolio per year in retirement. Again, Wade’s research on the topic is excellent. He shows that, historically, the higher the market’s value (relative to earnings) when you retire, the less you can safely spend from your portfolio per year.
Why I Don’t Use PE10
Despite the data showing the historical usefulness of PE10, I’m not comfortable using it for my own portfolio decisions.
As a general rule, the market does not like to be predictable. For the most part, once market inefficiencies (i.e., patterns that can be used to reliably outperform the market) become well known and easy to exploit, they tend to disappear.
It seems entirely likely to me that PE10’s predictive value is a market inefficiency like any other and that the primary reason it has existed for so long is that there was no way to exploit it. That is:
- Prior to computers, it would have been an enormous task to even calculate PE10, and
- Prior to the existence of no-load index funds (i.e., prior to 1977), there was no cheap, easy way to invest in the market as a whole. Moving in and out of stocks to capitalize on PE10’s predictive value would have meant buying or selling a large portfolio of individual stocks and paying transaction costs that are far higher than they are today. (Higher commissions, higher bid/ask spreads, and usually higher taxes.)
So rather than a data set of 140 years, we’re left with just 34 years (1977-2010). For an indicator that is only supposed to have useful predictive value over periods of 10+ years, 34 years isn’t a heck of a lot to go on.
Is Using PE10 a Terrible Idea?
Despite my personal lack of confidence in PE10 as a useful predictor, I think I’d place PE10-based decisions in the group of investment approaches that are at least reasonable — far better than, say, day-trading individual stocks.
For instance, if PE10 was at a historical high at a time when TIPS yields were also very high, I wouldn’t fault somebody for moving more of their portfolio to TIPS. Similarly, I think it would be reasonable to use a lower withdrawal rate if you retire at a time with unusually low TIPS yields and an unusually high PE10.
On the other hand, I’d be extremely reluctant to suggest either moving more of a portfolio to stocks or using a higher withdrawal rate from a retirement portfolio just because PE10 is low by historical standards.
Mike, thanks for the nice compliments. You’re really great at getting to the heart of complex issues and explaining them very clearly in just a short amount of space. I think you’ve made a fair assessment of the issue.
(I realize that this comment sounds like the generic spam comments I’m always getting, but I do sincerely mean it for you.)
Wade,
Likewise. Since encountering your work via the Bogleheads, I’ve had a great deal of respect for your research abilities and your willingness/eagerness to tackle actual problems/questions that real investors have.
Wade, you’ve done excellent research that has shed some light on how U.S. historical data may be optimistically skewed when it comes to sustainable withdrawal rates. I’d be curious to see the PE/10 methodology tested across global historical datasets. I vaguely remember DFA addressing PE10 as an indicator in this context but I don’t remember the specifics. Regardless, thanks Mike and Wade for your insights on this… very compelling!
Hi Mike,
Why PE/10? Why not PE/5 or PE/3 or PE/2? Do you know if anybody has looked at another averaging period besides 10 years to see if the results might be better?
Of course past performance doesn’t say anything about it’s predictive powers.
Hi Martin,
That’s a good question. (And I don’t know the answer.)
Personally, if I were considering relying on PE10, I’d actually want to check other similar periods for worse results rather than better. For example, if PE10 had predictive ability, but PE11 or PE9 didn’t, then I’d be more inclined to think PE10’s predictive ability was a fluke.
Martin, from Wade’s paper, “PE10 has become a widely accepted valuation measure, and as it would imply further data mining, there is no particular need to test whether other measures would produce even better results.” The fact that it’s so widely accepted is what gives me pause; “common knowledge” is rarely exploitable.
I had the exact same thoughts when I read Todd’s article. Trying to decide if the market is over or under valued seems a lot like market timing, which I believe to be fundamentally doomed to failure.
So often when I read your posts, it’s like you’re reading my mind. It’s like the friend of mine likes a particular movie reviewer because said reviewer agrees almost perfectly with my friend’s taste in movies.
Yes, you have a way of explaining things very concisely, and therefore your opinions/blogs are a pleasure to read.
The use of PE10 (or CAPE) has been touted as a very effective tool for “active” portfolio management and market timing, to counter the buy-hold-(and adjust) strategy which is deemed detrimental to economy as a whole.
I guess the jury is still out to decide which strategy (if either) is the winning one, but thanks for providing a few more counter-arguments to the debate on value-based investing.
@John About the case for Japanese savers, I did investigate the performance of PE10 for a past blog entry in March. Please search for “Japan PE10” on my blog.
For others interested in this subject, there is a great thread at the Bogleheads Forum that discussed the pros and cons of valuation-based investing for pages on end. The tread name is: “Valuation-based market timing with PE10 can improve returns?”