What's interesting today is that a projected 3.6% annual total return for the coming decade, compared with a "normal" 10-year return of 10% annually, implies roughly the same level of overvaluation as indicated by discounted dividends - putting fair value roughly 40% below present levels. On that note, I was struck by Alan Abelson's latest piece in Barron's, where he offered:
"The latest calculation by Andrew Smithers, the smart Brit who runs the eponymous London-based investment firm Smithers & Co., is that U.S. equities are more than 70% overpriced, according to q, his favorite yardstick and essentially a measure based on replacement value.
"Just to put you at ease, we haven't quite lost our minds, nor Andrew his. The market, rest assured, isn't about to vanish into the void. And Andrew is quick to point out that by his reckoning, stocks are well below their valuation peaks of 1929 and 1999, but more or less even-steven with the highs of 1906, 1937 and 1968.
In the chart below - courtesy of Doug Short - the historical norm for Q is 0.70, which is nearly 40% below the recent Q ratio of 1.12. Equivalently, the recent level is nearly 70% above the historical norm.
Abelson continues, "For all his wariness for the long pull, he doesn't see share prices suffering a steep fall so long as the Federal Reserve keeps pumping liquidity into the system and Washington stalls on meaningful deficit reduction. Frankly, although we greatly esteem Andrew's perspicacity, we aren't so sure he's right. Not least because so many market mavens now share his view, which suggests to us, as the old Street clichƩ has it, it probably has already been discounted in the latest bump up in equities."
With advisory bullishness back to 2007 extremes and equity put/call ratios at similarly extreme levels, I have to agree with Alan on that point.
So that is where valuations stand. I recognize that the conclusions seem implausible. I would not be inclined to share this data if it didn't have a strong historical record. The first criticism of these valuation implications is undoubtedly that they imply P/E ratios on forward operating earnings that seem far "too low." On this note, it's important to recognize that profit margins are currently about 50% above the historical norm. Moreover, while a multiple of 15 may be appropriate for trailing net earnings on normalized profit margins, it is a wholly inappropriate multiple to apply to forward operating earnings on elevated margins.
It is one thing to factor that reality into valuations - if margins can remain 50% above the norm for a full decade before contracting, it's easy to show that stocks should be valued about 15% higher than otherwise. But it is entirely another thing to assume that profit margins will remain 50% above historical norms forever, ignoring every bit of historical evidence that they revert to the norm over time. Analysts who blindly apply a multiple that "feels right" to next year's projected operating earnings are implicitly assuming that margins will remain permanently elevated at record levels. The common practice of blindly applying an arbitrary multiple to the coming year's projected earnings is an error that reflects profound misunderstanding of how securities are priced.
The second criticism of course, is that 10% may not be an appropriate discount rate, given 30-year Treasury yields at 4.5%. On this, I'll make two observations. The first is that in post-war data, the projected long-term total return for stocks has averaged about 4.25% more than long-term Treasury bonds. So if one believes that long-rates will remain at 4.5% forever, it's probably appropriate to bring the discount rate down, which would make stocks less overvalued, but highly vulnerable to any interest rate surprise. The second observation is that the S&P 500 has historically carried an average duration of about 30 years (if you work through some calculus, the duration of stocks turns out to be roughly equal to the price/dividend ratio), so the appropriate benchmark would normally be a 30-year zero coupon bond. Presently, the S&P 500 has a duration upward of 53 years. In order to price it properly, you can't simply refer to the current, depressed 10-year yield on a coupon-bearing Treasury. You have to think of the rate of return investors will demand 5 years, 10 years, 20 years, 30 years, and even 40 years from today. A 10-year Treasury has a duration of roughly 7 years. There's neither a theoretical nor historical basis (if you actually test it, which most people don't) for using the 10-year Treasury as a benchmark return for equities.
We focus on valuation models where the deviation from fair value is strongly correlated with subsequent market returns over the following 5-10 years. We hear a lot of "valuation" calls from analysts who seem to believe it is unnecessary to subject their models to historical tests. But investors should demand no less than a 7th grade math teacher does - "Please show your work." We're certainly open to alternative models that have a testable historical record. We're not interested in making a bullish case or a bearish case - our objective is to estimate prospective returns accurately. From where we stand, the evidence is presently not encouraging.
To say that fair value is far below present levels does not imply that the market will revert quickly to that level - only that long-term total returns are likely to be tepid as prices grow slower than fundamentals for an extended period. Moreover, to say that stocks may not revert quickly to our estimates of fair value means that we will need to have some willingness to accept market risk even in periods when stocks are still overvalued from a longer-term perspective. As I've noted in recent weeks, we've broadened the range of Market Climates we define in a way that we believe is robust, and will allow us to accept moderate exposures to market fluctuations more frequently than we have in recent years. In short, while we are not enthusiastic at all about market valuations, we've also improved our ability to play the hand that the market deals us as we move forward.
