The Shiller cyclically adjusted (for inflation) price-to-earnings ratio—referred to as the CAPE 10 because it averages the last 10 years’ earnings and adjusts them for inflation—is a metric used by many to determine whether the market is undervalued, fairly valued or overvalued. Employing a 10-year average for earnings, instead of the most current 12-month earnings, was first suggested by legendary value investors Benjamin Graham and David Dodd.
In their classic 1934 book “Security Analysis,” Graham and Dodd noted that traditionally reported price-to-earnings ratios can vary considerably because earnings are strongly influenced by the business cycle. To control for the cyclical effects, Graham and Dodd recommended dividing price by a multiyear average of earnings and suggested periods of five, seven or 10 years.
Then, in a 1988 paper, economists John Campbell and future Nobel Prize-winner Robert Shiller, using a 10-year average, concluded that a long-term average does provide information in terms of future returns. This gave further credibility to the concept, and led to the popular use of the CAPE 10.
A Changing Horizon
However, as Graham and Dodd noted, there’s really nothing special about using the 10-year average. Other time horizons also provide information on future returns. With that in mind, today we’ll analyze how changing the horizon can impact our view of the market’s valuation.
We’ll begin by looking at the current level of the CAPE 10. As of April 13, it was 26.3. This compares to a long-term (136-year) average of 16.7. The differential between the CAPE 10 today and its historical average has led many observers to conclude the market is overvalued and headed for a sharp decline. Jeremy Grantham and John Hussman have been among market gurus who, for the last four years or so, have been warning about an impending debacle as valuations eventually revert to their historical mean.
However, the market looks less overvalued if we change the horizon. Why would we consider a different horizon? First, as mentioned earlier, there is nothing magical about using 10 years to calculate the earnings average. Graham and Dodd even suggested using a five- or seven-year period. More importantly, in 2008, the earnings of the S&P 500 temporarily collapsed as a result of the financial crisis.
Note that in the following analysis, I’ve used the operating earnings of the S&P 500 as shown on NYU professor Aswath Damodaran’s website. The Shiller CAPE 10, however, uses “as reported,” or GAAP (generally accepted accounting principles), earnings. This distinction is important because operating earnings are generally higher, especially during recessions. With the Great Recession causing S&P 500 earnings to not recover to their 2007 level until 2010, we will look at CAPE ratios using earnings beginning in 2010. Thanks to the folks at AQR, we can examine both the CAPE 6 and CAPE 5 ratios using operating earnings as our measure.
CAPE Based On Operating Earnings
As of April 13, the current CAPE 6 was 18.7. Its average since 1960 was 15.5. That puts the CAPE 6 about 21% above its average over the past 56 years. Observe that when using GAAP earnings, the CAPE 6 is also lower than the current CAPE 10 of 26.3. It’s now at 22.7, or roughly 19% above its mean since 1960 of 19. The CAPE 5 was 18.5, again, as of April 13. Its average since 1960 was 15.7, placing it approximately 18% above its average. Using GAAP earnings, the current CAPE 5 would be 22.1, also about 18% above its mean of 18.1 since 1960.
While this still leaves the market looking somewhat highly priced compared with its historical averages, it no longer looks dramatically overvalued. That said, before you draw any conclusions, we need to consider the issues related to Shiller’s use of a 136-year historical mean.
The discussion that follows should highlight why I chose to look at the mean since 1960 instead of since 1880. In addition, some of the issues raised are based on changes made in just the last 20 years. When adjustments for them are made, the current high valuation (19% above the CAPE 6 mean since 1960 and 18% above the CAPE 5 mean) could disappear. Keep this in mind as you read the arguments.
Problems With Using The 136-Year Mean
In finance, it’s generally best to look at the longest data series available, thereby minimizing the risk of data mining. However, there are several reasons why using a 136-year average for the CAPE 10 will lead to a false conclusion that the market is overvalued.
The Shiller CAPE 10’s historical mean is about 16.7, with the dataset for the full period going all the way back to 1880. The data includes economic eras in which the world looked very different to investors than it does today.
Consider just two examples. For a significant part of the period, there was neither a Federal Reserve to dampen economic volatility nor an SEC to protect investor interests. The presence of both organizations has helped to make the world a safer place for investors, justifying a lower equity risk premium and thus higher valuations. In addition, we haven’t experienced another Great Depression, and there haven’t been any worldwide wars since 1945.
Another reason for the CAPE 10 rising over time is that the U.S. has become a much wealthier country since 1880. This matters because, as wealth increases, capital becomes less scarce. All else equal, less scarce assets should become less expensive.
Changing FASB Rules
Another reason the Shiller CAPE 10’s full-period mean may be an inappropriate benchmark is because accounting rules have changed, impacting how earnings (and thus price-to-earnings, or P/E, ratios) are determined. In 2001, the Financial Accounting Standards Board (FASB) changed the rules regarding how goodwill is written off.
As a post on the blog Philosophical Economics explained: “In the old days, GAAP required goodwill amounts to be amortized—deducted from earnings as an incremental non-cash expense—over a forty year period. But in 2001, the standard changed. FAS 142 was introduced, which eliminated the amortization of goodwill entirely. Instead of amortizing the goodwill on their balance sheets over a multi-decade period, companies are now required to annually test it for impairment. In plain English, this means that they have to examine, on an annual basis, any corporate assets that they’ve acquired, and make sure that those assets are still reasonably worth the prices paid. If they conclude that the assets are not worth the prices paid, then they have to write down their goodwill. The requirement for annual impairment testing doesn’t just apply to goodwill, it applies to all intangible assets, and, per FAS 144 (issued a couple months later), all long-lived assets.”
While FAS 142 may have introduced a more accurate accounting method, it also created an inconsistency in earnings measurements. Present values end up looking much more expensive relative to past values than they actually are. And the difference is quite dramatic. Adjusting for the accounting change would put the CAPE 10 about 4 points lower.
Another reason not to rely on the long-term historical mean of the Shiller CAPE 10 as a yardstick is that far fewer companies pay dividends now than in the past. For example, in their 2001 study, “Disappearing Dividends: Changing Firm Characteristics or Lower Propensity to Pay?”, Eugene Fama and Kenneth French found that the firms paying cash dividends fell from 67% in 1978 to 21% in 1999. This has resulted in the dividend payout ratio on the S&P 500 dropping from an average of 52% from 1954-1995 to just 34% from 1995-2015.
In theory, higher retention of earnings should result in faster growth of earnings as firms reinvest that retained capital. That has been the case for this particular period; from 1954-1995, the growth rate in real earnings per share (EPS) averaged 1.72%, and from 1995-2015, it averaged 4.9%.
As the post on Philosophical Economics explained, to make comparisons between present and past values of the Shiller CAPE 10, any differences in payout ratios must be normalized. The adjustment between the 52% payout ratio (the average from 1954-1995) and the 34% payout ratio (the average from 1995-2015) corresponds to approximately a one-point difference on the Shiller CAPE 10.
The Liquidity-Risk Premium
And there’s yet another reason why the long-term mean of the CAPE 10 might be misleading. Investors demand a premium for taking liquidity risk (less-liquid investments tend to outperform more liquid investments).
All else equal, investors prefer greater liquidity. Thus, they demand a risk premium to hold less-liquid assets. Over time, the cost of liquidity, in the form of bid/offer spreads, has decreased. There are several reasons for this, including the decimalization of stock prices and the provision of additional liquidity by high-frequency traders. In addition, commissions have collapsed in price.
But we aren’t done quite yet. Another important factor is that the presence of financial instruments that allow investors to buy and sell illiquid assets indirectly (such as index funds and ETFs) work to lower the sensitivity of returns to liquidity. These instruments enable investors to hold illiquid stocks indirectly with very low transaction costs, reducing the sensitivity of returns to liquidity. With these innovations in markets, all else equal, we should see a fall in the equity risk premium demanded by investors, and thus higher valuations.
If we make an adjustment from a CAPE 10 to a CAPE 6 or CAPE 5, and we use a still very long period of 56 years and operating earnings, we find the current valuation of the market is less than 20% above its mean. That’s a dramatically lower figure than the 57% difference between the current CAPE 10 and its long-term average.
What’s more, even the 20% difference doesn’t consider adjustments for any of the issues we raised, including the changes in accounting rules, the reduction in the tendency to pay dividends and the dramatic fall in transactions costs.
The bottom line is that once those adjustments are considered, there’s a case to be made that the market no longer looks overvalued. However, that doesn’t mean expected returns aren’t lower than the historical average. If we use the CAPE 6 of 18.7, that results in an earnings yield of 5.3%.
However, since earnings grow over time, we still have to account for an average lag of three years. I suggest using a real earnings growth forecast of 2%. Thus, we need to multiply 5.3% by 1.06 (1 + [3 x 2%]), producing an adjusted earnings yield of 5.7%. And that would be the forecast for future real returns for the S&P 500.
If we use the CAPE 5 of 18.5, we get an earnings yield of 5.4% and multiply that by 1.05 (1 + [2.5 x 2%]), which produces a slightly higher real expected return of 5.7%. In either case, that remains well below the historical 7% real return, but not as bad as the forecast of 4.2% you get from using the CAPE 10 figure of 26.3.
Larry Swedroe is the director of research for The BAM Alliance, a community of more than 140 independent registered investment advisors throughout the country.